(RILEM Bookseries 9) Ida NÃ Slund, Helena Johnsson (Auth.), Simon Aicher, H.-W. Reinhardt, Harald Garrecht (Eds.) - Materials and PDF

RILEM Bookseries
Simon Aicher
H.-W. Reinhardt
Harald Garrecht Editors
Materials and Joints

in Timber Structures
Recent Developments of Technology
Materials and Joints in Timber Structures
RILEM BOOKSERIES
Volume 9
RILEM, The International Union of Laboratories and Experts in Construction

Materials, Systems and Structures, founded in 1947, is a non-governmental scien-
tific association whose goal is to contribute to progress in the construction sciences,
techniques and industries, essentially by means of the communication it fosters be-
tween research and practice. RILEMs focus is on construction materials and their
use in building and civil engineering structures, covering all phases of the building
process from manufacture to use and recycling of materials. More information on
RILEM and its previous publications can be found on www.RILEM.net.
For further volumes:

http://www.springer.com/series/8781
Simon Aicher H.-W. Reinhardt
Harald Garrecht
Editors
Materials and Joints

in Timber Structures
Recent Developments of Technology
ABC
Editors
Simon Aicher Harald Garrecht
University of Stuttgart University of Stuttgart
Stuttgart Stuttgart
Germany Germany
H.-W. Reinhardt
University of Stuttgart
Stuttgart
Germany
ISSN 2211-0844 ISSN 2211-0852 (electronic)

ISBN 978-94-007-7810-8 ISBN 978-94-007-7811-5 (eBook)
DOI 10.1007/978-94-007-7811-5
Springer Dordrecht Heidelberg New York London
Library of Congress Control Number: 2013949137
c RILEM 2014
No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by
any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written
permission from the Publisher, with the exception of any material supplied specifically for the purpose
of being entered and executed on a computer system, for exclusive use by the purchaser of the work.
Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)
Preface
The use of timber in structures represents one of the most promising approaches
to meeting the urgent needs for sustainability, environmental friendliness and CO2
emission reduction in building technology. Due to the worlds fast growing popu-
lation, the achievement of the mentioned aims is increasingly understood as one of
the primary keys to ensure future life on earth.
Timber construction has experienced considerable progress within recent years.
Advancement can be seen within the both equally important interacting construc-
tion components - materials and joints. With regard to materials, it is undeniable
that cross-laminated timber (cross-lam) has widened the range of possibilities of
timber constructions the most. Its re-invention of plywood, now based on sawn
boards instead of peeled veneers, is the basis for multi-storey buildings of up to 9
storeys and for realistic perspectives of up to 30 storeys. Though itself rigid, cross-
lam in combination with energy dissipative joints is well suited for buildings in
zones with high seismic activity. The recent production of glulam and LVL made
of high strength hardwoods such as beech, oak, chestnut and several tropical hard-
woods, further broadens architectural horizons and meets the demands of forestry
towards sustainable and soil/climate apt tree cultivation. The spectrum of glulam
possibilities has also been expanded substantially due to so-called block gluing,
whereby several glulams are glued to form massive cross-sections which may, for
instance, be ideally used in bridges. An additional renewable, wood-like material,
bamboo, is also considerably extending the possibilities and efficiency of wooden
materials. Processed and densified bamboo strands enable the production of plates
and beams with strength properties similar or even higher than those of steel. Low
density wood fibre boards, representing a highly ecological insulation material, are
now simultaneously used for structural bracing purposes as well. Timber-concrete
composites with different connection technologies are increasingly gaining trac-
tion in buildings, most notably as cost-effective floor elements, due to good proven
performances regarding strength, stiffness, vibration and response to fire. Innova-
tive, glued, lightweight, composite deck-element constructions based on solid wood
and panel materials are also now manufactured industrially with lengths of over
30 meters.
VI Preface
Regarding mechanical joints, self-tapping screws produced at lengths of up to

1.5 m have substantially changed jointing and reinforcement technology. The use
of the screws as primarily tension or compression transferring devices at an angle
to the fibre direction represents a new highly efficient jointing and reinforcement
technology. Contrary to dowel type fasteners which act by embedment and hereby
inherently induce splitting of the wood, self-tapping screws overcome the problem
of tension perpendicular to the grain, hereby increasing load capacity of the joint
and changing the failure mode from brittle to ductile. Further, self-tapping screws
used in combination with specially formed i.a. dovetailed metal plates enable archi-
tecturally esteemed, completely hidden, end-grain connections.
The diversity of structural adhesives (now comprising phenolic resorcinol,
melamine urea, one-component polyurethanes, emulsion polymer isocyanates and
epoxies) with properties tailored to specific timber products and production pro-
cesses has increased considerably within the last years. In the case of joints with
glued-in rods, recently approved and reliable adhesive systems have been introduced
into timber construction technology and now enable new jointing solutions, i.a. for
wide-span grid-like spherical domes. Further, glued-in, perforated steel plates allow
for stiff and strong connections, i.a. in timber-concrete composite structures. Re-
cently, a new glued jointing technology for glulam beams which enables the trans-
fer of forces without any reduction of the full cross-sectional capacity has also been
developed.
The present Conference follows the earlier RILEM conference Joints in Timber
Structures held in Stuttgart from the 12th to 14th of September, 2001. The objec-
tive of the Conference is to bring together world leading experts in the mentioned
fields of timber materials, joints and construction to present the latest state of tech-
nology and to identify future research needs to conserve the pace of progress. The
main subjects of the conference are: cross-laminated timber, glued laminated timber,
timber based facades, modified wood, i.a. acetylated and thermally treated wood,
adhesives, self-tapping screws, glued and mechanical joints, seismic and cyclic be-
haviour of joints and structures, durability issues, surface treatment, timber-concrete
compounds and bamboo-based materials.
The financial support from the sponsoring companies is gratefully acknowledged.
Sincere gratitude is owed to the staff of the Division of Timber Structures of the Ma-
terials Testing Institutute (MPA), Otto-Graf Institute, University of Stuttgart, par-
ticularly to Maximilian Henning, Nikolai von Ruckteschell and Zachary Christian
for all of their organizational efforts. We also greatly appreciate the effort and en-
gagement from the editorial staff of Springer in the compilation of the Proceedings,
particularly Ms. Anneke Pot.
Simon Aicher
H.-W. Reinhardt
Harald Garrecht
Sponsors
The Conference was kindly supported by the following enterprises:

VIII Sponsors
Contents
Part I: Structures
Horizontal Displacements in Medium-Rise Timber Buildings:
Basic FE Modeling in Serviceability Limit State . . . . . . . . . . . . . . . . . . . . . 3
Ida Nslund, Helena Johnsson
Improving the Moment Resistance of a Concealed Timber Post Base
Joint . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Jrme Humbert, Sang-Joon Lee, Joo-Saeng Park, Moon-Jae Park
The Multifunctional TES-Faade Joint . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Stefan Loebus, Stephan Ott, Stefan Winter
Green-Glued Products for Structural Applications . . . . . . . . . . . . . . . . . . . 45
Erik Serrano, Jan Oscarsson, Magdalena Sterley, Bertil Enquist
Experimental Analysis of a Post-tensioned Timber Connection . . . . . . . . 57
Flavio Wanninger, Andrea Frangi
Risk Based Investigations of Partly Failed or Damaged Timber
Constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Gerhard Fink, Jochen Kohler
Naturally Grown Round Wood Ideas for an Engineering Design . . . . . . 77
Matthias Frese, Hans Joachim Bla
Recycling and End-of-Life Scenarios for Timber Structures . . . . . . . . . . . 89
Annette Hafner, Stephan Ott, Stefan Winter
Advancements for the Structural Application of Fibre-Reinforced
Moulded Wooden Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Jrg Wehsener, Tom-Egmont Werner, Jens Hartig, Peer Haller
X Contents
Sole Plate Fixing Details for Modern Methods of Timber

Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Jesus M. Menendez, Kenneth Leitch, Robert Hairstans
Thin-Walled Timber Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Benoit P. Gilbert, Steven B. Hancock, Henri Bailleres
Recommendations for the Design of Complex Indeterminate Timber
Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Andrew Lawrence
Novel Lightweight Timber Composite Element: Web Design in Shear
and Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Simon Aicher, C. Stritzke
New Timber Bridges: Inventive Design by Block-Gluing . . . . . . . . . . . . . . 149
Frank Miebach, Dominik Niewerth
Part II: Mechanical Connections

Steel-to-Timber Joints with Very High Strength Steel Dowels Using
Spruce, Beech and Azob . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
Jan-Willem van de Kuilen, Carmen Sandhaas, Hans Joachim Bla
Wood Load-Carrying Capacity of Timber Connections: An Extended
Application for Nails and Screws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Pouyan Zarnani, Pierre Quenneville
Ductility in Timber Structures: Investigations on Over-Strength
Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Frank Brhl, Jrg Schnzlin, Ulrike Kuhlmann
An Experimental Study on Bearing Strength in Compression
for Bolted Joint of Plywood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Akiko Ohtsuka, Sumiya Takahashi, Takumi Ito, Wataru Kambe
Investigations Concerning the Force Distribution along Axially
Loaded Self-tapping Screws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
A. Ringhofer, G. Schickhofer
Experimental Analysis on the Structural Behaviour of Connections
with LVL Made of Beech Wood . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
Peter Kobel, Ren Steiger, Andrea Frangi
The Embedment Failure of European Beech Compared to Spruce
Wood and Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Steffen Franke, Nolie Magnire
Contents XI
Modelling of Non-metallic Timber Connections at Elevated

Temperatures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Daniel Brandon, Martin P. Ansell, Richard Harris, Pete Walker,
Julie Bregulla
Analysis of the Brittle Failure and Design of Connections Loaded
Perpendicular to Grain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Bettina Franke, Pierre Quenneville
Structural Performance and Advantages of DVW Reinforced Moment
Transmitting Timber Joints with Steel Plate Connectors and Tube
Fasteners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Daniel Brandon, Adriaan Leijten
Fully Threaded Self-tapping Screws Subjected to Combined Axial
and Lateral Loading with Different Load to Grain Angles . . . . . . . . . . . . 265
Robert Jockwer, Ren Steiger, Andrea Frangi
Alternative Approach to Avoid Brittle Failure in Dowelled
Connections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
Daniela Wrzesniak, Massimo Fragiacomo, Andr Jorissen
Resistance and Failure Modes of Axially Loaded Groups of Screws . . . . . 289

U. Mahlknecht, R. Brandner, A. Ringhofer, G. Schickhofer
A Method to Determine the Plastic Bending Angle of Dowel-Type
Fasteners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301
Michael Steilner, Hans Joachim Bla
Low-Damage Design Using a Gravity Rocking Moment Connection . . . . 307
Mamoon Jamil, Pierre Quenneville, Charles Clifton
Part III: Glued Joints and Adhesives

Finger Jointing of Freshly Sawn Norway Spruce Side Boards
A Comparative Study of Fracture Properties of Joints Glued with
Phenol-Resorcinol and One-Component Polyurethane Adhesive . . . . . . . 325
Magdalena Sterley, Erik Serrano, Bertil Enquist, Joanna Hornatowska
Pressure Distribution in Block Glue Lines Analyzed by Theory
of Beams on Elastic Foundation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341
Gordian Stapf, Simon Aicher
EPI for Glued Laminated Timber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Kristin Grstad, Ronny Bredesen
Bonding of Various Wood Species Studies about Their Applicability
in Glued Laminated Timber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
Y. Jiang, J. Schaffrath, M. Knorz, Stefan Winter
XII Contents
Fatigue Performance of Adhesive Connections for Wooden Wind

Towers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
Leander Bathon, Oliver Bletz-Mhldorfer, Jens Schmidt,
Friedemann Diehl
Multifunctional Wood-Adhesives for Structural Health Monitoring
Purposes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
Christoph Winkler, Ulrich Schwarz
Assessment of the Glue-Line Quality in Glued Laminated Timber
Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
Bettina Franke, Florian Scharmacher, Andreas Mller
Review of Recent Research Activities on One-Component
PUR-Adhesives for Engineered Wood Products . . . . . . . . . . . . . . . . . . . . . 405
Christian Lehringer, Joseph Gabriel
Part IV: Timber and Concrete/Cement/Polymer

Composites
Development of a High-Performance Hybrid System Made
of Composites and Timber (High-Tech Timber Beam ) . . . . . . . . . . . . . . 423
Markus Jahreis, Martin Kstner, Wolfram Hdicke, Karl Rautenstrauch
Experimental Study of the Composite Timber-Concrete SBB
Connection under Monotonic and Reversed-Cyclic Loadings . . . . . . . . . . 433
Manuel Manthey, Quang Huy Nguyen, Hugues Somja, Jrme Duchne,
Mohammed Hjiaj
The Predictive Model for Stiffness of Inclined Screws as Shear
Connection in Timber-Concrete Composite Floor . . . . . . . . . . . . . . . . . . . . 443
F. Moshiri, R. Shrestha, K. Crews
Shear Performance of Wood-Concrete Composite with Different

Anchorage Length of Steel Rebar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 455
Sang-Joon Lee, Jrme Humbert, Kwang-Mo Kim, Joo-Saeng Park,
Moon-Jae Park
Development of Prefabricated Timber-Concrete Composite Floors . . . . . 463
Petr Kuklk, Pavel Nechanick, Anna Kuklkov
Wood-Based Construction for Multi-story Buildings: Application
of Cement Bonded Wood Composites as Structural Element . . . . . . . . . . 471
Alireza Fadai, Michael Fuchs, Wolfgang Winter
Rehabilitation, Upgrading and Repair of Historic Timber Structures
with Polymer Concrete and FRP-Reinforcement . . . . . . . . . . . . . . . . . . . . 485
Markus Jahreis, Karl Rautenstrauch
Contents XIII
Fatigue Performance of Single Span Wood-Concrete-Composite

Bridges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493
Leander Bathon, Oliver Bletz-Mhldorfer
Hybrid Wall-Slabs for Multi-storey Buildings: Made of Timber
with a Directly Applied Mineral Cover Layer . . . . . . . . . . . . . . . . . . . . . . . 499
Christian Dorn, Alexander Stief, Markus Jahreis, Karl Rautenstrauch
An Innovative Prefabricated Timber-Concrete Composite System . . . . . . 507
Roberto Crocetti, Tiziano Sartori, Roberto Tomasi, Jos L.F. Cabo
Part V: Cyclic, Seismic Behaviour

A Component Model for Cyclic Behaviour of Wooden Structures . . . . . . 519
Giovanni Rinaldin, Massimo Fragiacomo
Overview of a Project to Quantify Seismic Performance Factors
for Cross Laminated Timber Structures in the United States . . . . . . . . . . 531
M. Omar Amini, John W. van de Lindt, Shiling Pei, Douglas Rammer,
Phil Line, Marjan Popovski
Force Modification Factors for CLT Structures for NBCC . . . . . . . . . . . . 543
Marjan Popovski, Shiling Pei, John W. van de Lindt, E. Karacabeyli
Experimental Testing of a Portal Frame Connection Using Glued-In
Steel Rods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 555
James Walker and Robert Xiao
Part VI: Hardwood, Modified Wood and Bamboo

Bending Strength and Stiffness of Glulam Beams Made of Thermally
Modified Beech Timber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 569
Robert Widmann, Wilfried Beikircher, Jos L.F. Cabo, Ren Steiger
Structural Veneer Based Composite Products from Hardwood
Thinning Part I: Background and Manufacturing . . . . . . . . . . . . . . . . . . 577
Ian D. Underhill, Benoit P. Gilbert, Henri Bailleres, Robbie L. McGavin,
Dale Patterson
Glue Laminated Bamboo (GluBam) for Structural Applications . . . . . . . 589
Y. Xiao, B. Shan, R.Z. Yang, Z. Li, J. Chen
Glulam Composed of Glued Laminated Veneer Lumber Made
of Beech Wood: Superior Performance in Compression Loading . . . . . . . 603
Gerhard Dill-Langer, Simon Aicher
XIV Contents
Structural Performance of Accoya Wood under Service Class 3

Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615
Julian Marcroft, Ferry Bongers, Fernando Perez Perez, John Alexander,
Ian Harrison
Structural Veneer Based Composite Products from Hardwood
Thinning Part II: Testing of Hollow Utility Poles . . . . . . . . . . . . . . . . . . . 629
Benoit P. Gilbert, Ian D. Underhill, Henri Bailleres, Robbie L. McGavin
Glulam from European White Oak: Finger Joint Influence
on Bending Size Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 641
Simon Aicher, Gordian Stapf
Non-homogeneous Thermal Properties of Bamboo . . . . . . . . . . . . . . . . . . . 657
Puxi Huang, Wen-Shao Chang, Andy Shea, Martin P. Ansell,
Mike Lawrence
Part VII: Cross-Laminated Timber

Tapered Beams Made of Cross Laminated Timber . . . . . . . . . . . . . . . . . . . 667
Marcus Flaig, Hans Joachim Bla
Influence of the Connection Modelling on the Seismic Behaviour
of Crosslam Timber Buildings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 677
I. Sustersic, B. Dujic, Massimo Fragiacomo
Behaviour of Cross-Laminated Timber Panels under Cyclic Loads . . . . . 689
Igor Gavric, Massimo Fragiacomo, Marjan Popovski, Ario Ceccotti
CLT Plates under Concentrated Loading Experimental
Identification of Crack Modes and Corresponding Failure
Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 703
Georg Hochreiner, Josef Fssl, Josef Eberhardsteiner, Simon Aicher
Seismic Strengthening of Existing Concrete and Masonry Buildings
with Crosslam Timber Panels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 713
I. Sustersic, B. Dujic
In-Plane Stiffness of Traditional Timber Floors Strengthened
with CLT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725
Jorge M. Branco, Milos Kekeliak, Paulo B. Loureno
Propose Alternative Design Criteria for Dowel Type Joint With CLT . . . 739
Shoichi Nakashima, Akihisa Kitamori, Takuro Mori, Kohei Komatsu
Contents XV
Part VIII: Properties and Testing of Wood

Length Effects on Tensile Strength in Timber Members With
and Without Joints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751
R. Brandner, G. Schickhofer
New Perspectives in Machine Strength Grading: Or How to Identify
a Top Rupture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 761
Julia K. Denzler, Andreas Weidenhiller
Aspects of the Difference between the Local and Global Modulus
of Elasticity of Structural (hardwood) Timber . . . . . . . . . . . . . . . . . . . . . . 773
G.J.P. Ravenshorst, P.A. de Vries, Jan-Willem van de Kuilen
Part IX: Glulam

A Study of Australian Glulam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 787
H.R. Milner, Con Y. Adam
Improving Strength of Glulam Laminations of Norway Spruce
Side Boards by Removal of Weak Sections Using Optimized
Finger Jointing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 801
Jan Oscarsson, Anders Olsson, Bertil Enquist
Double Span Continuous Glulam Slabs Strengthened with GFRP . . . . . . 813
Jorge M. Branco, Marco P. Jorge, Jos Sena-Cruz
Simplified Design of Glued Laminated Timber Girders
for the Torsional Moment Caused by Stability Effects . . . . . . . . . . . . . . . . 823
R. Hofmann, Ulrike Kuhlmann
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 831
Keyword Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 835

Part I
Structures
Horizontal Displacements in Medium-Rise
Timber Buildings: Basic FE Modeling in
Serviceability Limit State
Ida Nslund and Helena Johnsson
Division of Structural and Construction Engineering, Lule University of Technology,

SE-971 87 LULE Sweden
ida.naslund@ltu.se
Abstract. Higher and larger timber buildings are built today. The building is ex-
posed to static lateral wind loads that cause displacements that might lead to dis-
comfort and non-function of the building. To determine the size and behavior of
horizontal displacements, two timber systems has been studied, a light frame sys-
tem with shear walls and a post and beam system with diagonal bracing. The sta-
bilizing wall segments have been analyzed with a FE model and subjected to static
lateral wind load and vertical dead load in serviceability limit state. Both plywood
and particle board were used as sheet materials. To reduce flanking transmission
Sylomer is applied in the light frame system. The total lateral displacement va-
ries between 4 to 125 mm in the light frame system and 2 to 5 mm in the post and
beam system. Removing the Sylomer damping material from the light-frame
system would decrease the lateral displacements with 1.5-3 times, which needs to
be further investigated.
Keywords: Stabilizing element, finite element model, lateral displacement, tim-

ber-frame, post and beam, serviceability limit state.
1 Introduction
Higher and larger timber buildings are constructed as progression in technology
moves forward. Serviceability aspects such as displacements, movements, and
deformation become more important when designing higher buildings. When ex-
posed to horizontal action e.g. from wind, a medium-rise building can experience
large lateral displacements due to the low mass density of wood. There are differ-
ent ways to stabilize medium-rise timber buildings against static lateral wind load.
In light-framed systems shear walls are used for stabilization and in beam and post
buildings different principles are used e.g. diagonal bracing, moment resisting
connections, and wall diaphragms. Shear walls take up load in their own plane and
work through diaphragm action in roof and floors. The diagonal bracing can be
wooden and work in both compression and tension or a steel chord which works
only in tension.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 3
DOI: 10.1007/978-94-007-7811-5_1, RILEM 2014
4 I. Nslund and H. Johnsson
To transfer load between structural elements connections are used. How the
connection acts between floors and foundation is important for the whole struc-
ture. Depending on the design the strength and stiffness will vary in the connec-
tion. It must be designed so unexpected slip is avoided. Slip is also generated in
materials used for vibration damping in connections. There is no limit established
for static displacements in structures in Eurocode 5 (2009), merely on single
building elements as beams. Clients are responsible for setting up displacement
limits in tendering. In an early version of Eurocode 5 (2009) a maximum of hori-
zontal displacement was set to H/300 were H are the height, but this has been
removed. The Recommendation in the German design code is H/500 (Kllsner &
Girhammar, 2008).
Depending on the building system and stabilization method, lateral displace-
ments will vary. To qualify as a stabilizing wall segment it needs to be continuous
from the ground to the top of the house. A building consists of several stabilizing
walls that must resist lateral load, but in this study one wall is studied.
Figure 1 and Figure 2 present stabilizing wall segment for a shear wall system
and a diagonal bracing system respectively. Several models to predict the capacity
of a shear wall subjected to static lateral load has been proposed, both analytical
and numerical e.g. Kllsner & Girhammar (2009); Falk & Itani (1989); Kllsner
(1984); Foschi (1977). Basic FE modeling is made for two different cases, a light-
frame system and a beam and post system. The light frame model consists of
studs, sheets, sheathing-to-frame nails, anchorage and Sylomer to reduce flank-
ing transmission. Sheets will be either plywood or particle board. The post and
beam structure is stabilized by wooden diagonals connected by bolts.
The aim is to study the total static horizontal displacement in stabilizing walls
in a light-frame system and a post and beam system in serviceability limit state.
Second order theory is not taken into account in this study. The frequent load
combination was chosen to represent serviceability limit state.
1.1 Case Study Houses

The case study houses are medium-rise apartment buildings with a timber frame.
They are 3-5 storeys, which correspond to a height of 11 to 18 meter. The length
of the houses varies between 24 to 62 meters and the width between 9 to 36
meters. The number of stabilizing walls in each house varies between 1 to 9 walls
with a length of 2.4 to 9.6 meter. No openings and walls shorter than 1.2 meter
counts as a stabilizing segment because of high loads at the end of the walls
(Kllsner & Girhammar, 2008). For the beam and post system, tr8 is used as an
example (Tlustochowicz, 2011). The wind is applied according to Eurocode 1
(2009) and distributed through rigid floor diaphragms to the stabilizing element;
see Figure 1 and Figure 2.
Horizontal Displacements in Medium-Rise Timber Buildings 5
Dead load
Lateral displacement
Wind load
Fig. 1 Shear wall segment, see figure 3 for detail A
Dead load
Lateral displacement
Wind load
Fig. 2 Post and beam wall segment, see figure 3 for detail B
2 Materials and Method
2.1 Light-Frame System

The timber-framed wall consists of timber of strength class C24 (SS-EN 338,
2009) and two different sheathing materials; plywood and particle board (EN
12369-2, 2011; SS-EN 312). In Table 1 material properties are presented. The
cross-section of the timber element was set to 45x120 mm2 and the thicknesses of
the sheets were 9 mm and 12 mm respectively for plywood and particle board.
The width and height of one wall segment is 1.2x3.0 m2. To minimize flanking
transmission between floors a damping material, Sylomer is applied under each
stud, between the upper and lower sills. Sylomer is a polyurethane (PUR) elas-
tomer material that transfers compression loads and small shear loads. Different
stiffness of Sylomer is used for different compression loads. Due to production
ease, the same stiffness is often used on at least two floors. Material properties for
Sylomer are assumed according to the product sheet from the supplier and are
presented in Table 2. Anchorage devices are applied between each floor to resist
the tensile up-lift, 15 nails are applied. The stiffness of the anchorage is calculated
according to Eurocode 5 (2009) and is presented in Table 3.
Table 1 Material properties
Timber Glulam Plywood

Particleboard
C24 GL28c P30
Elastic modulus
11000 12600 1050 2000
[N/mm2]
Shear modulus
Not used Not used 500 960
[N/mm2]
Poissons ratio Not used Not used 0.05 0. 4
Density [kg/m3] 420 380 Not used Not used
Table 2 Stiffness of Sylomer
SR 850 SR 1200
Kx
304 342
[N/mm]
Ky
2736 3952
[N/mm]
Sheets are connected to the frame by nails. The stiffness of the sheathing-to-
timber joints are calculated with data collected from U. A. Girhammar, Bovim, &
Kllsner (2004). In U. A. Girhammar, Bovim, & Kllsner (2004) tension tests on
sheathing-to-timber joints is presented for three different sheet materials; hard-
board, particle board and plywood. The tests were performed both perpendicular
and parallel to the grain of the frame member. Data from the test were analyzed
here-in and the slip modulus calculated according to SS-EN 26 891. In the finite
element model the sheathing-to-timber joints are modeled by two springs in each
corner, x- and y-direction in Figure 3. The fastener force distribution varies along
the framing member and the resulting stiffness of the joints is not the number of
fasteners times the stiffness (Kllsner & Girhammar, 2009). A numerical calibra-
tion was performed to arrive a stiffness value for the corner springs to simulate
nailing along the perimeter of the sheathing. The distance between fasteners was
set at 200 mm with fasteners on the perimeter studs only. The numerical analysis
was performed in Matlab R2012b with properties that are described in the para-
graph finite element model. The stiffness was determined by comparing lateral
displacements for the two models and adjusting the stiffness of the corner springs
until it coincided with the stiffness of the model with springs around the perimeter,
Table 3.
Table 3 Stiffness of connections
Type Plywood Particleboard Anchorage Bolts

Kx
2564 3566 Not used Not used
[N/mm]
Ky
1318 2437 26093 Not used
[N/mm]
K [kNm] Not used Not used Not used 106
2.2 Post and Beam System

In the post and beam system glulam GL28c (SS-EN 1194, 1999) was used with a
cross-section of 360x160 mm2. In Table 1 material properties are listed. To
connect the diagonal bracing four bolts were applied. The rotational stiffness was
calculated according to Eurocode 5 (2009) and presented in Table 3. The post and
beam system has no Sylomer and no anchorage devices.
2.3 Load
Wind load was calculated according to Eurocode 1, (2009) with a basic wind
velocity of 24 m/s (Stockholm, Sweden) and a pressure factor of 1.1. The wind is
exponentially distributed over the height of the building. The frequent load com-
bination was used, which represents a reversible state with temporary damage e.g.
temporary deflection of a beam due to a short high loads (Eurocode 0, 2009).
Dead loads are applied on the building: 0.3kN/m2 for the walls and 0.6kN/m2 for
the slabs where the ceiling is included.
2.4 Finite Element Model

For the calculation Matlab R2012b with a finite element toolbox, CALFEM (Ola
Dahlblom et al., 1986) was used. The toolbox is based on (Niels & Saabye Otto-
sen, 1992). The light-frame model consists of three basic elements: frame mem-
ber, a sheathing member and fasteners. The frame members are modeled as beam
elements with three degrees of freedom in each node. A plane stress
isotropic element is used for the sheathing with two degrees of freedom in each
node (four node element). Two springs represent fasteners between the frame
member and the sheet which are connected in the translation direction only to the
corner. In Figure 3 a) all connections are displayed. Since the studs are pinned
towards the top and bottom rail an additional degree of freedom (rotation) was
added in those nodes that contain frame members, since the studs and top/bottom
rail will not have the same rotation. The top/bottom rail will behave as a conti-
nuous beam. To model the Sylomer between each floor, single spring pairs were
used in the translation direction. All floors are connected to each other with hold-
downs which only can take tension load and are placed at the end of the wall.
Hold-downs are represented by a spring in the model in the vertical direction. The
bottom rail at the first floor is hindered to move in any translational direction, it is
rigid to the foundation.
Y Y
X X
Beam
Tie-downs
Diagonal
Sylomer Pinned
Bolts
Pinned Column
Sheating-to-frame joints Frame Sheet
a b
Fig. 3 Detail A in the shear-wall system and detail B in the beam-and-post system
The post and beam system consists of columns and beams that are modeled as
beam elements, Figure 3b. Diagonals are also modeled as beam elements and are
connected with bolts to the columns. The bolted connection is represented by rota-
tional springs. Diagonals can take both tension and compression. Columns are pin
connected to the beams (additional degree of freedom) and the beams are modeled
as continuous. The bottommost columns are hindered to move in any translational
direction and are rigid to the foundation
All springs have two nodes where each node has one degree of freedom. For
springs, different types are available see e.g. Judd & Fonseca (2005). In this study
all springs behave linearly due to the serviceability limit state.
The wind load is distributed to each beam element node. Load distribution are
set to 36 m and distributed to 5 walls in the light-frame model and distributed to 2
walls in the post and beam system. In the light-frame model the wind load is di-
vided in two since walls often consist of sheets on both side of the frame. Dead
load is applied to all beam element nodes.
3 Result
For the system with shear walls as stabilizing elements, the number of shear walls
has been varied using two different sheathings. The number of floors in the build-
ing has also been varied. For the post and beam system, the number of floors and
a b
c d
e f
Fig. 4 Building height-top corner horizontal displacement curves for the light frame system,
a) - c) 3-5 storeys plywood, d) - f) 3-5 storeys particle board
the number of wall segments was varied. Figure 4 a)-c) present top corner
horizontal displacements (Figure 1) for 3-5 storeys stabilized with shear walls of
plywood. Referring to Figures 4b) and c): using two or three wall segments pro-
duces a non-linear response. This is due to bending of the stabilizing segments and
changing stiffness of Sylomer higher up in the building. In Fig. 4a) the response
is almost linear since the building is low. The results for the particle board in Fig-
ure 4 d)-f) show the same overall behavior, but the response is stiffer overall. In
the light frame system the displacements vary between 6 125 mm for plywood
and 4 102 mm for particle board. If removing the Sylomer from the FE simula-
tion, the horizontal displacements are 1.5-3 times lower i.e. maximum lateral
displacement amounts to 42 mm in the light-frame system
Figure 5 presents the post and beam system where the top corner displacements
(Figure 2) vary little with the number of wall segments and the behavior is linear.
The displacement requirements H/300 and H/500 are fulfilled easily. The post and
beam system shows a displacement between 2-5 mm.
a b
c
Fig. 5 Building height-top corner horizontal displacement curves for the post and beam
system a)-c) 3-5 storeys
4 Conclusion and Discussion
This study provides an overview of the possible horizontal displacement range for
medium-rise timber building stabilized by either a light frame system or a post and
beam system. To stabilize the light-frame system, 5-8 wall segments of 1.2 m
length is a minimum if considering the frequent load combination. When taking
into account any eccentrical loading, this number may increase. In the beam and
post system, horizontal displacements are much smaller. For several of the wall
configurations in the light-frame system, the horizontal displacement fall short of
the recommendations, (H/300 or H/500) which shows that it is important to check
displacements in higher buildings. The analysis here-in was performed with the
frequent load combination, which represents a load value that is exceeded 1% of
the life-time of the structure. Since the light-frame model considers pinned
connections where the real behavior is semi-rigid, the model presented here is
somewhat weaker than the actual case. Sylomer seems to affect the building
negatively when looking at horizontal displacements, further investigations are
needed.
Further research is to; model these semi-rigid connections, study the effect of
Sylomer in detail and also produce a 3D model that considers the surrounding
structures.
Acknowledgements. The authors wish to acknowledge the financial support from the cen-
tre for Lean Wood Engineering sponsored by VINNOVA in Sweden and SWECO struc-
tures in Sweden. The authors also wish to thank Bertil Nslund for advice in programming.
References
EN 12369-2, Wood-based panels characteristic values for structural design part 2. Ply-
wood (2011)
Eurocode 0, SS-EN 1990: Eurocode 0 basis of structural design. SIS, Stockholm (2009)
Eurocode 1, SS-EN 1991-1-4 (2005): Eurocode 1: Actions on structures - part 1-4. General
actions - wind actions. SIS, Stockholm (2009)
Eurocode 5, SS-EN 1995-1-1(2004): Eurocode 5: Design of timber structures - part 1-1.
General - common rules and rules for buildings. SIS, Stockholm (2009)
Falk, R.H., Itani, R.Y.: Finite-element modeling of wood diaphragms. Journal of Structural
Engineering 115(3), 543559 (1989)
Foschi, R.: Analysis of wood diaphragms and trusses - 1 diaphragms. Canadian Journal of
Civil Engineering 4(3), 345352 (1977)
Girhammar, U.A., Bovim, N.I., Kllsner, B.: Characteristics of sheathing-to-timber joints in
wood shear walls. In: 8th World Conference on Timber Engineering, Lathi, Findland
(2004)
Judd, J.P., Fonseca, F.S.: Analytical model for sheathing-to-framing connections in wood
shear walls and diaphragms. Journal of Structural Engineering-ASCE 131(2), 345352
(2005)
Kllsner, B.: Panels as wind-bracing elements in timber-framed walls, Stockholm, Sweden

(1984)
Kallsner, B., Girhammar, U.: Analysis of fully anchored light-frame timber shear walls-
elastic model. Materials and Structures 42(3), 301320 (2009)
Kllsner, B., Girhammar, U.A.: Horizontal stabilising of light frame timber structures.
Plastic design of wood-framed shear walls. SP Technical Research. Institute of Sweden,
2008:47 (2008) (in Swedish)
Niels, S.O., Saabye Ottosen, N.: Introduction to the finite element method. Prentice Hall,
New York (1992)
Dahlblom, O., Peterson, A., Petersson, H., Dahlblom, O., Peterson, A., Petersson, H.:
CALFEM - a program for computer: aided learning of the finite element method. Engi-
neering Computations: International Journal for Computer: Aided Engineering and
Software 3(2), 155160 (1986)
SS-EN 1194, Timber structures - glued laminated timber - strength classes and determina-
tion of characteristic values (1999)
SS-EN 26 891, Timber structures joints made with mechanical fasteners general prin-
ciples for the determination of strength and deformations characteristics (1991)
SS-EN 312, Particleboards - specifications (2010)
SS-EN 338, Structural timber strength classes (2009)
Tlustochowicz, G.: Stabilising system for multi-storey beam and post timber buildings.
Doctoral thesis, Lule Univeristy of Technology (2011)
Improving the Moment Resistance
of a Concealed Timber Post Base Joint
Jrme Humbert*, Sang-Joon Lee, Joo-Saeng Park, and Moon-Jae Park
Korea Forest Research Institute (KFRI), 57 Hoegiro, Dongdaemun-gu, Seoul, Korea

jerome.humbert@gmail.com
1 Introduction
The Korean traditional house, Hanok, is a post-and-beam timber structure with a
large tiled roof designed to keep the house cool in summer (Fig. 1). The solid
wood members are connected together without metallic connectors, using complex
woodwood joints, while the posts are fitted inside cornerstones. Associated with
the stabilizing effect of the heavy roof, this provides a basic lateral resistance to
the house, which historically proved effective against the few earthquakes which
occurred in this area of low seismicity. Nowadays, the Hanok house is still fairly
popular among Koreans thanks to its cultural, aesthetic, and eco-friendly aspects.
However, some concerns include e.g. a limited thermal insulation, the presence of
woodwood connectors with brittle behavior, and a design not suited to high-
rising structures. In a country where over one-third of the population lives in the
metropolitan area around the capital city of Seoul, the lack of space implies
favoring multi-story buildings over single-story ones. Moreover, the recent
seismic events in the region raised concerns on the seismic resistance of buildings.
Additionally, the complexity and diversity of traditional timber joints require
skillful artisans and are not well-suited to industrialized processes. Finally, in an
eco-friendly approach, the structural use of domestic wood species implies using
engineered wood products to overcome limited mechanical properties.
In this context, the Korea Forest Research Institute (KFRI) developed a new
modernized design of multi-story timber house based on the visual identity of
Hanok and aiming at addressing the mentioned issues. A modular hybrid design
was proposed, coupling an outer post-and-beam structure with inner lateral-
resistant elements 66. Using engineered timber members and metallic connectors,
the outer post-and-beam structure provides the necessary resistance for vertical
loads while ensuring a safer design. Indeed the semi-rigid metallic connectors
provide a source of increased ductility and energy dissipation, and prevent the
unsafe brittle fracture of wood. Moreover, the use of glued laminated (glulam)
timber members associated with factory pre-cutting together provide reduced
mechanical and geometrical variabilities, and thus better-quality constructions
with easier assembly than what is obtained with on-site works using naturally
highly-variable solid wood elements.
*
Corresponding author.
DOI: 10.1007/978-94-007-7811-5_2, RILEM 2014
14 J. Humbert et aal.
Fig. 1 A traditional Korean timber house Hanok composed of an apparent post-and-beaam

structure and a heavy tiled ro
oof. (Source: Wikipedia)
Concerning the lateraal resistance, the post-and-beam structure is known tto

intrinsically display limitted performances. Therefore, additional lateral-resistannt
elements are used. Severral technologies were considered, including the heavilyy-
studied light-frame shearrwalls with sheathed OSB panels, which served as a
reference design 6, and thhe structural insulation panels (SIP), which provides botth
a lateral resistance and an
n increased thermal insulation 6.
One concern regarding g the development of this hybrid timber house design waas
to maintain the visual ideentity of the Hanok. In particular, replacing woodwoood
connections with metalliic connectors in the apparent post-and-beam structurre
called for the use of co oncealed metallic connectors. Therefore new metalllic
connectors were designed d, based on existing commercial models adapted to suuit
the modular hybrid design n developed.
In this study, we investigate the moment resistance of a concealed post basse
joint connector under monotonic
m and reversed cyclic loading. Although thhe
primary use of the post-and-beam structure is to undergo vertical loadings, thhe
enclosing configuration of o the outer glulam members around the inner lateraal-
resistant structure has an effect
e on the overall mechanical behavior of the structurre
under lateral loading 6. Moreover,
M although fairly limited the lateral resistance oof
the outer post-and-beam structure
s alone is not completely negligible as compareed
to those of the inner light--frame shearwall. Finally, assessing the rotational rigiditty
and other moment-related d mechanical properties of the joint is of interest both foor
structural design and num merical modeling.
Results of bending testts were already reported in 6.
In the following, we present
p in a first step experimental tests under reverseed
cyclic loading, and a stu udy of the moment-resisting properties of the post basse
joint. After concluding onn the limited performance of the connector, we propose iin
a second step some mod difications of the connector to improve its ductility annd
energy dissipation. A new w series of experimental tests is then conducted to validaate
Improving the Moment Resistance of a Concealed Timber Post Base Joint 115
these modifications. In paarallel, a refined 3D finite element (FE) model is derived,

and results from numericaal simulations are compared with experimental ones.
Keywords: post and beam

m structure, cyclic loading, metallic connections, ductility..
2 Experimental Tests
T
2.1 Materials and Methods

M
A picture of the metallicc connector is presented on Fig. 2 along with its maiin
dimensions. A schematic diagram of a quarter of the lower part of the connector is
presented in Fig. 3. In thee upper part, it is composed of four vertical metal wingss
arranged to form a cross, and connected to the timber member using four metalllic
pins of d = 16 mm diam meter. The pins are grouped in two pairs arrangeed
perpendicularly at 20 mm m and 50 mm from the top of the connector. This ensurees
that a sufficient end distaance is maintained when connecting the timber membeer.
For reference, the Europeean design codes 6 prescribe a minimum timber membeer
end distance of max(7d, 80 mm) = 112 mm. In these tests we comply with thhis
requirement.
The lower part of the connector
c is composed of two 170 mm 170 mm paralllel
horizontal metal plates, with
w a 30 mm gap between them leaving enough space foor
the bolt nut and washer. TheT upper plate features four 30 mm holes to insert thhe
bolt nuts. The lower platee has four 20 mm holes for the anchor bolts. The platees
are connected together wiith small vertical metallic segments located both directlly
under the wings and alon ng one of the half-borders of the connector, leaving thhe
other half-border open forr the insertion of a wrench. All metallic plates composinng
Fig. 2 Metallic connector

Fig. 3 Diagram of the lower section of the connector showing the position and length of thhe
welds (thicknesses not to sccale). (1) Full-length weld between the upper face of the uppper
plate and the vertical wings.. (2) Short weld between the lower face of the upper plate annd
the vertical segments.
the connector have a thickkness of 5 mm. The welding of the metallic parts togetheer
is presented in the diagram
m on Fig. 3. In the upper part, the wings are welded botth
together and to the upper horizontal plate of the base along their full length (welld
(1) on Fig. 3). This is cllearly visible on Fig. 2. In the lower part however, thhe
welds between the horizontal plates and the vertical segments do not cover the fuull
length of the pieces (weld (2) on Fig. 3). This difference has a strong impact oon
the mechanical behavior, as explained below in the experimental results section.
The post-and-beam structure of the hybrid timber house design is composed oof
Japanese larch (Larix kaeempferi) glulam timber members with a cross-section oof
180 mm 180 mm, fix xed at their base to the studied metallic connectorrs.
Therefore, in this study we
w perform pseudo-static reversed cyclic moment tests oon
the joint using timber sppecimens with the same wood species and cross-sectioon
dimensions. The specimeens of timber posts are, depending on the tests, eitheer
800 mm or 1000 mm talll, and are horizontally loaded at their top with a 100 kkN
oil jack, effectively appllying a moment load to the joint. The reversed cycllic
loading (Fig. 4) is based on the ISO 16670 protocol 6, and is composed of cyclees
of increasing amplitude, namely
n 1, 5, 10, 20, 40, and 60 mm, followed by a finnal
monotonic ramp up to 80 0 mm which usually leads to the failure of the joint. Thhe
loading speed is 1 mm/ss, which allows keeping the test duration within a 330
minute span while avoidin ng possible dynamic effects.
The internal load cell of the oil jack and an external laser displacement sensoor
provide a measurement of o the rotation and moment in the joint. Because of thhe
experimental setup and the
t effect of rotation, for large displacements the actuual
horizontal displacement as measured by the laser sensor at constant height (aas
shown on Fig. 4) differss slightly from the displacement of the actuator itsellf.
Therefore, in order to compute the rotation angle we use the displacement valuue
measured by the externaal laser sensor rather than the internal measure of thhe
actuator stroke. The momment is derived directly from the load cell and the height at
which the load is applied. In order to monitor a possible differential displacemennt
between the connector an nd the post after damaging started, an additional linear
variable differential traansformer (LVDT) displacement sensor provides a
measurement of the verticcal displacement of the base of the post. The base of thhe
connector is affixed to the metallic testing frame with four bolts of 20 mm m
diameter representing the anchor bolts usually used with post base connectors.
Experimental tests are performed at least two times per configuration (see otheer
configurations below) in order
o to minimize experimental discrepancies.
Fig. 4 Reversed cyclic displaacement loading as recorded by the external laser sensor
2.2 Experimental Results

R
The reversed cyclic loadin ng applied to the joint always leads to its failure througgh
the damaging of the meetallic connector. In the initial configuration presenteed
above, the glulam membeer is not damaged, nor are the metallic pins. The failurre
occurs only in the connector base. The connector first begins to plasticize; namelly,
the upper plate begins to o bend. However, this phenomenon is brief, as the loaad
quickly reaches the resistance limit of the welds, leading to their tearing. Morre
specifically, the short welds
w binding the upper horizontal metal plate of thhe
connector with the verticcal elements underneath, labeled weld-(2) on Fig. 3,
experience the earliest faiilure as shown on Fig. 5. Indeed, their length is limited tto
a mere 20~30 mm, and d therefore they constitute the weakest point of thhe
connector under this tractiion-inducing moment loading.
Fig. 5 Failure of short-lengtth welds on the lower face of the upper horizontal plate of thhe
connector base
Fig. 6 presents the evo

olution of the moment M in the connector function of thhe
rotation angle . The joint
j displays a semi-rigid non-linear behavior witth
hysteresis loops typical of timber joints with metal connectors. The momennt
resistance is 11 kN.m in n the positive direction and 8 kN.m in the negativve
direction, and the initiall elastic rotational stiffness is approximately equal tto
460 kN.m/rad. We can no ote an asymmetry in this behavior, which is due in part tto
the loading history: the co
onnector is first loaded on the positive side, and then oon
the negative side. The dammaging of the connector when loaded on the positive sidde
lowers its overall resisttance, leading to a lower maximum moment wheen
subsequently loaded on thet negative side. In this particular test, an experimenttal

error also adds up to the lo
oading to increase this asymmetry.
However, in the preseent case we can clearly see two different parts in thhe
evolution. Initially, the evolution
e displays a high rotational stiffness. Yet, in a
second step after the tearring of the welds the rotational stiffness is much loweer,
and so is the resistance. Despite
D preventing the brittle fracture of wood in favor oof
the plasticizing of the metallic
m connector, this behavior limits the ductility annd
energy dissipation of the joint.
j
ment M in the joint function of the rotation angle

Fig. 6 Evolution of the mom
Fig. 7 Bilinear model (EEEP curve) for the original configuration, displaying a low ductility
In order to quantify th he ductility, a possible approach is to use the Equivalennt

Energy Elastic-Plastic (EE EEP) method from 6. The EEEP curve corresponding tto
the test result presented in
n Fig. 6 is shown on Fig. 7. It is obtained using the secannt
at 10% and 40% of the peeak moment Mmax as the initial elastic rotational stiffnesss.
The ultimate moment Mu is computed to equal the energies. The ductility m is theen
defined as the ratio of u over
o v. In this case, we obtain a ductility value m = 1.46,
which is close to 1.0 (elasstic fragile behavior) and indicates a low ductility for thhe
joint.
Fig. 8 FE mesh of the specim

men
Concerning the uplift, the measured values show no significant relative

displacement between the metallic connector and the timber member. This is
consistent with post-test observations which do not show any damage in the pins
or the timber member.
In conclusion, the failure of the joint occurs in the metallic connector as it could
be expected for a use under reversed cyclic loading. However, considering an
application in a structure undergoing a seismic loading, the ductility and energy
dissipation of this joint remain low because of the failure in the welds rather than
in the pins, as usually found in pinned timber connectors.
3 Numerical Study
In order to complement the experimental results, and as a preparation for future
modelings of a complete timber house, we perform a numerical analysis using a
refined Finite Element (FE) model.
3.1 Numerical Model

The FE model of the joint is composed of a mesh of approximately 105 second-
order volume elements (20-node hexahedra and 10-node tetrahedra) with a size
ranging from 50 mm in areas of lesser interest down to 1 mm in areas of
concentrated stresses, i.e. typically around pin holes and weldings (Fig. 8). We
perform static nonlinear simulations using the FE analysis software ANSYS. The
nonlinearities arise only from the presence of contact between the wood and the
metal. The base of the connector is fixed, and a monotonic load is applied on the
upper part of the wood member on an area of 100 mm height (0.018 m2) roughly
equal to the size of the loading metal plates used in the experimental tests, in the X
direction as shown on Fig. 8.
The material parameters of the FE model are summarized in Table 1. The wood
is modeled using a linear elastic orthotropic material with a density w of
520 kg/m3, a longitudinal elasticity modulus Ew,L of 14.2 GPa (along the column
long side, vertically), and a tangential modulus of elasticity Ew,T of 1 GPa. Those
parameters correspond to a larch species as reported in 6. The steel of the
connector and the pins is modeled with a linear elastic isotropic material with an
elasticity modulus Es of 210 GPa and a density s of 7850 kg/m3.
Table 1 Material properties of the FE model
Material Parameter Value

Density w = 520 kg/m3
Wood Elasticity modulus (L) Ew,L = 14.2 GPa
Elasticity modulus (T) Ew,T = 1 GPa
Density s = 7850 g/m3
Steel
Elasticity modulus Es = 210 GPa
3.2 Numerical Sim

mulation Results
The numerical results shoow a concentration of high stresses around the pins and iin
the lower part of the metaallic connector, that is in the regions of mesh refinemennt
(Fig. 9). The maximum sttress in the connector is reached at the base, between thhe
horizontal upper plate and
a the vertical wing above it. Qualitatively, this is
consistent with the observ
vations of the experimental tests since this is precisely thhe
location of the failure.
Fig. 9 Stress in the metallic parts

p of the joint
Quantitatively, because these numerical results are computed using materials

with a linear elastic behavior, there is no strain or stress limitation during the
simulation. Therefore, additional criteria must be verified a posteriori to ensure
physical consistency. The selected criteria are summarized in Table 2. For the wood,
the embedding strength of wood fh is taken equal to 35 MPa for a load parallel to
wood grain 6, and the tensile strength ft is taken equal to 3 MPa for a load
perpendicular to wood grain 6. The ultimate strength of the steel constituting the
connector itself and the pins is taken equal to s,u = 460 MPa. As seen on Fig. 9,
this limit is reached for a load of approximately 10 kN. This value is consistent
with the resistance of the connector observed in the experimental tests.
Table 2 Physical criteria for FE model post-processing
Material Parameter Value

Embedding strength fh = 35 MPa
Wood
Tensile strength ft = 3 MPa
Steel Ultimate strength s,u = 460 MPa
4 Improving the Joint

Based on those results, we can draw two main conclusions. First, the failure of the
joint occurs in the metallic connector rather than in the timber member. This is
generally considered safer because the sudden brittle fracture of wood is replaced
by the much slower plasticizing of steel, allowing for a visual hinting about an
imminent mechanical failure. Conversely, despite the failure occurring in the
metallic connector, the experimental tests underlined the relatively low ductility
and energy dissipation of the joint. Therefore, when considering even a moderate
seismic event the energy released during the earthquake cannot be absorbed in the
connectors and need to be dissipated elsewhere, possibly leading to the failure of
structural elements and the collapse of the structure.
Based on these conclusions, we propose improving the joint to enable a greater
ductility and larger energy dissipation. Because in the original configuration the
failure occurs in the weldings in the base of the connector, we consider two
possible modifications. First, for the sake of simplicity, we simply bolt the
connector directly from the upper plate, allowing the metallic wings to be more
directly connected to the ground (Fig. 10). With this setup, we need to cut the end
of the timber member for practical reasons. Alternatively, a second possibility is to
reinforce the welds by extending them to cover the full length of the wings (see
Fig. 5). Finally, a mixed setup is also considered using both modifications.
In the following, we present the experimental results of tests conducted using
those alternative connector configuration, all test parameters being otherwise
equal to those presented earlier for the original joint configuration. We also
present results on a mixed configuration using both approaches at the same time.
Fig. 10 Plasticizing of the connector

c when using the modified setup with a cut-end timbber
member. The failure still occcurs in the short weldings of the connector, but is delayed bby
the deformation of the upper plate. Full-length weldings at the metal wing base do nnot
experience any damage.
4.1 Cut-End Config

guration
The momentrotation ev volution in the joint for the cut-end configuration is
presented in Fig. 11. As itt was the case for the original configuration, the failure oof
the joint occurs in the weak
w welds. Yet the resistance of the connector (peaak
moment) is lower than the one in the original configuration. The reason is thhe
absence of contact betweeen the lower end of the timber member which was cuut
Fig. 11 Evolution of the moment

m M function of the rotation angle in the cut-ennd
configuration
and the upper plate of th he connector. In this configuration, the rotation is onlly
limited by the metal wing gs. However, the failure occurs for a much larger rotation,
over 0.1 rad. This, assocciated with the shape of the hysteresis loops, togetheer
suggest a larger ductility and
a energy dissipation.
The EEEP curve correesponding to the test of Fig. 11 is presented in Fig. 12.
In this case, we obtain an a ultimate moment Mu = 7.5 kN.m lower than the onne
nfiguration. However, the ductility factor is larger, with a
obtained in the initial con
value of m = 2.64 indicating a more ductile behavior as compared to the originnal
configuration (m = 1.46). The
T direct bolting of the upper plate therefore delayed thhe
failure of the welds and allowed
a a greater plasticizing of the metallic connector, aas
shown on Fig. 10, while th he absence of contact lowered the moment resistance.
Fig. 12 EEEP curve (bilinearr model) for the cut-end configuration. Mu = 7.5 kN.m
4.2 Full-Length Welds

W
In a second step, we consider a configuration with full-length welds only, i.e.
without a cut-end timbeer specimen. The momentrotation evolution in thhis
configuration is presented d in Fig. 13. In this case, the resistance of the joint is
much higher than in th he previous configurations, with a maximum momennt
Mmax = 16.5 kN/m, that is an increase of roughly 50% in resistance. The associateed
EEEP curve is presented in Fig. 14. We obtain a ductility ratio m = 2.36 which is
intermediate between the initial configuration and the cut-end configuration. Thhe
behavior is still ductile, and
a some energy dissipation occurs. The failure occurs iin
the welds again, but wood d fracture is also visible directly above the metal wings.
ment M function of the rotation angle with full-length welds

Fig. 13 Evolution of the mom
Fig. 14 EEEP curve (bilinear model) for the full-length weld configuration. Mu = 155.9
kN.m.
4.3 Mixed Cut-End

d and Full-Length Welds Configuration
Lastly, we perform exp perimental tests on a mixed configuration using botth
previous approaches at thhe same time: the timber specimen is cut at its lower end,
the upper plate of the con
nnector is directly bolted to the ground, and the welds arre
extended to cover the full length of the metal elements.
Fig. 15 presents the momentrotation

m evolution in the joint. The evolution is
similar to the one obtaained in the cut-end configuration, with a resistancce
Mmax = 10.7 kN/m slightlly higher. Yet the resistance remains much lower thaan
when using full-length welds alone because of the absence of contact. Indeed w we
can observe in the experim
mental tests some fracture directly above the metal winggs,
indicating that the momeent in the area of the pins is larger. The EEEP curve is
presented in Fig. 16 and gives
g a ductility factor m = 2.70, which is slightly higher
than when bolting the low
wer plate to the ground.
Fig. 15 Evolution of the moment M function of the rotation angle with mixeed
configuration
Fig. 16 EEEP curve (bilinearr model) for the mixed configuration. Mu = 9.5 kN.m.
28 J. Humbert et al.
Table 3 Comparison of the performance of the joint in the tested configurations
EEEP Ductility
Joint Strength EEEP limit
stiffness factor
configuration Mmax [kN.m] Mu [kN.m]
[kN.m/rad] m [-]
Initial (I) 10.9 9.0 458 1.46
Cut-end (C) 8.8 7.5 214 2.64
Full-length 16.5 15.9 351 2.36
welds (W)
Mixed (C+W) 10.7 9.5 250 2.70
5 Conclusions
We presented in this paper some experimental and numerical results on the
moment resistance of a concealed timber post base joint aimed at replacing in a
modern design introduced lately the woodwood joints used in the traditional
Korean Hanok timber house.
Experimental and numerical results show that the original configuration of the
joint offers a limited moment resistance and a low ductility and energy dissipation.
In an attempt to mitigate those limitations without undergoing major changes in
the connector, three new configurations are proposed and investigated. Motivated
by the wish to prevent the early failure in welds, a first approach consists in
directly bolting the connector's upper plate to lower the stress on the weak welds.
Alternatively, another approach focuses on increasing the strength of these welds
by extending their length to the full width of the metal wings. Finally, a third
configuration investigates the effect of those two approaches combined.
Experimental results on those three alternative configurations show that cutting
the end of the timber member to allow direct bolting of the upper plate also have
the effect of reducing the moment strength of the joint. This is because there is no
more contact at the lower end, and therefore the timber specimen is only
constrained by its contact on the wings. Unfortunately, a gap is generally present
between the connector and the timber member to allow for an easier setup. This
gap enables some free rotation, and favors the splitting of wood, overall producing
a larger ductility factor but lowering the joint resistance.
On the other hand, increasing the strength of the welds by extending their
length has a positive effect on the strength, with a gain of over 50% as compared
to the initial configuration, while also improving the ductility of the joint, albeit in
a more moderate way as the previous configuration.
The mixed approach benefits from both alternatives but unfortunately also
shares their weakness. The resulting strength is about the same as the original
configuration, while the behavior is much more ductile. Yet wood splitting is
increased, and the concealing of the connector somewhat decreased. Moreover
cutting the timber member generates a potential concern regarding the vertical
load.
In conclusion, we believe that reinforcing the welds is the best option among
the presented ones. The resulting joint exhibits a larger moment strength and a
much more ductile behavior than in the original configuration. As a result, we
believe that this connector is fitted for use in earthquake-resistant structures,
provided that the timber frame is associated with suited lateral-resistant structural
elements, and that its tensile resistance is found to be sufficient to sustain the uplift
forces usually produced by a horizontal loading. This last point is currently being
investigated experimentally.
Acknowledgment. The authors thanks M. Whi Lim Son for his help during the experi-
mental tests.
Disclaimer
This work is a continuation of the study presented in 6. Therefore, some initial results
already published are reported here for the sake of consistency, along with some corrections
resulting from the analysis of new tests. The second part of this work presents new
configurations and is completely original.
References
[1] Hwang, K., et al.: Shear performance of post and beam construction by pre-cut pro-
cess. Journal of the Korean Wood Science and Technology (Mokchae
Konghak) 35(6), 112 (2007)
[2] Park, M.-J., et al.: Shear performance of hybrid post and beam wall system with
structural insulation panel infill. In: Proceedings of the 11th WCTE, paper 456 (2010)
[3] Humbert, J., et al.: Cyclic Behavior of Timber Column Concealed Base Joint. Journal of
the Korean Wood Science and Technology (Mokchae Konghak) 41(2), 123133 (2013)
[4] Hwang, K., et al.: Resistance performance of cross-shaped metallic joint for use in
multi-story timber structures. In: Proceedings of the Korean Society of Wood Science
and Technology (KSWST) Annual Meeting, paper D-12, Gwangju, Korea, April 4,
pp. 6465 (2011) (in Korean)
[5] International Organization for Standardization, ISO 16670: Timber Structures
Joints made with mechanical fasteners Quasi-static reversed-cyclic test method,
Geneva, Switzerland (2003)
[6] European Committe for Standardization (CEN), Eurocode 5: Design of timber struc-
tures Part 1-1: General Common rules and rules for buildings, EN 1995-1-
1:2004, Brussels, Belgium (2004)
[7] Architectural Institute of Japan, New estimation method for shear wall performance.
Standard for Structural Design of Timber Structure, 104 p. (2002) (in Japanese)
[8] Yamada, M., et al.: Seismic Performance Evaluation of Japanese Wooden Frames. In:
Proceedings of the 13th WCEE, paper 753, Vancouver, British Columbia, Canada,
August 1-6 (2004)
[9] Kretschmann, D.: Wood Handbook, Chapter 05: Mechanical Properties of Wood.
Technical Report FPL- GTR-190 (2010)
[10] Sawata, K., Yasumura, M.: Determination of embedding strength of wood for dowel-
type fasteners. J. Wood Sci. 48, 138146 (2002)
The Multifunctional TES-Faade Joint
Stefan Loebus, Stephan Ott, and Stefan Winter
TU Mnchen, Chair of Timber Structures and Building Construction, Germany

{loebus,ott,winter}@tum.de
Short Summary. In comparison to a demolition of existing buildings with severe

technical deficits, usually the retrofitting of buildings is more effective in order to
prepare them for low energy consumption and new necessities as communication
and media connection or HVACinstallation (Heat, Ventilation and
AirConditioning). Prefabricated retrofit solutions are developed throughout
Europe to enable higher levels of industrialization in building envelope
modernization and hence additionally improvements in energy efficiency. Five
years of experience and a reasonable number of demonstrations done with
timberbased element system (TES) faades show tendencies for best-practice
building construction.
This paper focuses on the jointing between single faade elements and the
connection of those elements to the existing building. Being a crucial construction
detail within the TESfaade, the joint area shall meet various requirements and
challenges, from load bearing over hygro-thermal to fire safety functionality. The
results of in-depth construction detailing lay out the requirements and principles of
the TES joint.
Keywords: Energy efficiency, refurbishment, faade construction, timber

construction, prefabrication, fire safety, building envelope.
1 Introduction
Prefabrication a well-known concept in new timber construction still needs the

implementation of proper detailing considering the different circumstances and
assembly methods in refurbishment in order to apply it for retrofit projects. TES
Energy Faade and the follow-up smartTES, two international research consortia,
work on this idea since 2008 [1]. Since then the TES faade elements for energy
efficient retrofit of the building envelope with prefabricated timber based elements
are demonstrated in several projects. A first milestone was the definition of an
integrated work process starting from planning, digital survey, prefabrication and
assembly [2]. A key feature was the development of horizontal and vertical joints
of faade elements for certain assembly routines and structural properties. Further
findings are dealing with fire safety, ecologic properties, thermal and moisture
behaviour in specific climate conditions. This article shows the development of
DOI: 10.1007/978-94-007-7811-5_3, RILEM 2014
32 S. Loebus, S. Ott, and S. Winter
principles of detailing timber framework in refurbishment; analogy to existing

systematics in timber construction, the requirements, and generic solutions.
2 Retrofit with Prefabricated Faades
Modern, prefabricated timber frame construction is highly developed as a building

system for new construction. There is also no limit for small series or so-called
pilot prefabrication [3][2]. Because of the flexibility in the production planning
and manufacturing, custom timber framed components can be produced without
overhead [4]. This fact makes timber framed faade elements very interesting for
the refurbishment of existing buildings, because individual elements are required
for almost each building and project. Numerous pilot projects and demonstrators
in the context of research projects have shown that the transferability of the
principles of the timber framed building on a conceptual level is possible and
useful [5][6][7]. But at the detail level numerous new developments and
adjustments are necessary to meet the requirements of a reliable and economically
feasible design.
The accurate measurement of the faade is a prerequisite for precise
prefabrication and mounting Geometry data have to be combined with a thorough
survey of the existing faade construction, assessing materials, detailing, load
bearing capacity and if necessary building services und further in depth
information from the existing substance [2].
Fig. 1 Vertical section of Fig. 2 TES horizontal joint Vertical

typical semi-balloon frame
construction [9]
3 Timber Framed Construction in Refurbishment

Systematics of New-Built Timber Framework and Analogy
The three principles of timber frame construction are the platform frame
construction, balloon frame, and the semi or virtual balloon-frame [8].
The Multifunctional TES-Faade Joint 33
The virtual balloon-frame construction is the most suitable type as an analogy

for a TES EnergyFacade element. The elements have storey-high size and are
horizontally oriented; they can be produced in original upright position, put on the
truck and handled on site the same way without changing orientation. All elements
are stacked on top of each other and the dominant joint is the horizontal one. This
has the advantage of a continuous faade layer or building skin, without
interruptions and unbroken functional layers for air tightness, insulation or fire
safety. In analogy to the ceiling support edge beam a so called coupling beam is
introduced as a connection between the existing ceiling and the timber-framed
wall. The coupling beam is adjusted exactly in height and depth. It has the
additional advantage to assist horizontal alignment of TES elements. In contrast to
new building in virtual retrofitting balloon-frame, no vertical forces are transferred
to the coupling beam. The vertical loads are transferred only through the timber
framed wall. The horizontal loads have to be induced into the existing ceiling
plate. Hygrothermal exposure and wood protection of such a detailing solution can
be solved in analogy to new building. The fire safety must be examined in depth
because facades have a certain risk in the spread of fire and fire safety has always
observed in relation with the boundary conditions of the existing building.
Structure
Hygrothermal
Economy
behaviour
Existing
building
Ecology Fire safety
Assembly Building services
Fig. 3 Parameters of a TES faade system
4 Definition of Requirements
The number of parameters affecting the TES element and its details is large. Most
influential are apart from structural issues, planning building physics of timber
construction, correct production of multilayered elements and assembly on
existing exterior walls. The main parameters about strengths and fixtures are
derived from the load-bearing structure of the elements. As a result, they are at the
same time in conjunction with the materiality, manufacture, structural
arrangement, aka. tectonics, and the assembly process. The other group of
influences originates from building physics. They affect TES elements on the
main inner and outer protection functions and therefore define the requirements
for layering and details. In addition to the production, they are in interaction to the
assembly, economy, ecology and technical building equipment and must be linked
to the static requirements. These interactions take place not only within the TES
system, but and above all, they are in interaction with the existing building. Some
protective functions of the existing exterior wall are partially replaced by the TES,
such as the thermal protection; they are fulfilled primarily by the TES envelope.
5 Limitations of the Analogy
The development of the retrofit construction system starts from the main structural
influences and the appropriate way of tectonics of parts. The focus is on the area
of the horizontal element joint. It is the connection of the existing exterior walls
with the existing ceiling. The horizontal joint is the most relevant detail to solve
from the technical and economic perspective, because it is the most frequent joint
of a virtual balloon-frame. A 1950ies multi-story house with 8 apartments on four
floors has a total length of basement and roof eaves joint line between circa 80 and
130 m. All other horizontal joints between the floors have a total length of
approximately 240 and 390 m. The complete vertical joint length is around 65 to
75 m. It matches a ratio between vertical and horizontal joints from 1:5 to 1:7. The
high deviation in this ratio clearly shows how important detailing of the horizontal
joint has to be done. It influences assembly effort and costs to a large extent. The
technical requirements and the effort in this area of the element have to be
adjusted very carefully to avoid high costs or unforeseeable risk. The developed
solutions have to be in line with the condition of efficiency. A robust solution of
but- jointing two elements will reduce the risk of the construction system for
failures. The detailed analysis of the construction principles on the applicability
and specialties of the construction process with large-format element in
refurbishment results in clear advantages for the balloon-frame principle. The
requirements of the continuity of the insulation layer and the uninterrupted surface
with defined axis of fixing, as alignment, and mounting aid for the faade
elements are established as significant characteristics.
A further invention is the adaption layer between the uneven surface of the
existing exterior wall and the plane back side of the TES element. The gap should
be at least fifty millimetres wide for further filling with blown-in insulation
material, preferably cellulose fibre. The process of blowing thermal insulation
material in the gap has the advantage of adapting the thickness, where it is needed.
Around penetrations of the exterior wall and the elements like windows, the gap
should be stuffed with non-combustible mineral wool.
Fig. 4 Development of platform frame, (a) balloon-, and semi balloon-frame (b-c) to TES
(d) in analogy to (c)
6 Results in Timber Framework for Refurbishment

Development of the Horizontal Construction Joint
Principles of Timber Framing for Refurbishment Existing
Buildings
As in classic timber frame construction the solid wood studs are positioned in a
span of 625 mm and butt-jointed perpendicular to sill-beam and wall-plate. On the
element inside an OSB is applied and on the outside a gypsum fiber board. The
cavities are filled with thermal insulation, e.g. cellulose fiber. Usually a timber-
frame element measures about 12,0 x 3,2 m, limited by production facilities and
transportation. [5]
Fig. 5 Mounting a TES-Element

While the TES-facade is built with a tolerance of +/-2 mm, the existing exterior
wall deviates +/-40 mm in depth. This unevenness is leveled by the coupling
beam, which is anchored into the existing wall and runs around the entire building
horizontally on each floor level. Thus, while mounting, its possible to align the
faade elements to the coupling beam and fix it with self-drilling timber screws.
The assembly of the elements happens horizontally row by row. The design of the
element joint was inspired by the simple joining method of a tongue-and-groove
connection and enhances the common butt-joint known from classic timber-frame
constructions. The tongue consists of the upper elements sill-beam. The groove
consists of wall-plate and the overlapping gypsum fiber board of the lower
element. That way the upper element is fitting precisely into the lower element,
without any larger alignment work necessary and still guaranteeing a force-fitting
connection. [5] The vertical connection is a butt-joint of two members without any
further connection.
Fig. 6 TES horizontal joint Vertical Fig. 7 TES horizontal joint Horizontal
All facade components such as the window generally are integrated in the TES
element, see Figure 7. This has the advantage of better manufacturing control and
simplified construction site operations. The new window can be installed in
different configurations as a replacement for the existing one. It can be integrated
only as a supplementary window in the TES, which enhances the existing window
and improves its thermal and acoustic properties. This solution is an important
option for upgrading existing components of good quality, speeds up the
construction process significantly and reduces disturbances of the residents.
The variety of applicable cladding materials is analog to new timber-frame
buildings; however, retrofitting a larger multi-storey building, the options are
strongly limited by fire safety requirements. In this paper the construction is given
with a rear ventilated wooden faade.
7 Structural Joint
Essentially there are two main loads on the TES-faade that have to be taken in to
account constructing the loadbearing system: Self-weight and wind-load.
Fig. 8 Force conduction
The self-weight of the given construction is with about 80 kg/m (varying by

thickness, thermal insulation material and cladding) comparably high. This is due
to the extended timber loadbearing structure, the required fire safety boards, and a
dense cellulose fibre thermal insulation. The vertical force caused by self-weight
conducts all the way through the timber frame into a foundation system at the foot
of the faade. (For faade foot construction solutions see [1]) As the vertical force
accumulates over the building height, the horizontal joint between the elements is
built as full form fit contact connector. (Fig. 8 No. 1) The horizontal forces from
the wind-load are induced shortest way into the next ceiling. The connection is
built with anchor bolts and self-tapping screws. By placing coupling beam and
element joint on same height with the ceiling, eccentricities can be minimized and
heights of cross-sections optimized with the minimum edge distances of the bolts
and screws as one leading factor. Hence the sill-beam of the top element and the
wall-plate of the bottom element need to be built in one level. (Fig. 8 No. 2)
8 Hygrothermal Behaviour of TES Elements
The avoidance of thermal bridges is the first principle in detail development as no

high heat-conductive material from the hot to the cold side goes through, so that
conduction of heat is minimised. Thus, the impact of material-related thermal
bridges can be largely reduced by the use of wood or wood-based materials in the
joint area. A flexible variation of the insulation layer thickness can be achieved by
the adaptation of the depth of studs.
Regarding diffusion safety of the TES layer, the back panelling of the element
has to be vapour retardant. Diffusion open (to the outside) cross section of the
element with a low 0,3 sd 4m on the outside of the element and an sd at least
six times higher on the inside of the element is recommended. [10]
Air and wind tightness of timber framed elements are one of the most important
measures to stop convection of warm and moist air to the cold side of wooden
parts. A proper airtight construction also improves fire safety and sound protection
by overcoming leakages.
The horizontal and vertical TES joints have to be sealed for airtightness reasons
to ensure the moisture performance and avoid heat conductivity through wet wood
or insulation materials in the joint area.
Fig. 9 Hygrothermal detail
Wind tightness is provided by a wind barrier as an outer panelling or membrane

of the closed TES element. The wind barrier has to be sealed at connection by
tapes or clamped joints.
Besides construction material has to be dry, and elements are protected during
construction by closed cavities and heavy weathering to minimize moisture intake.
Diffusion open layering supports the drying capacity of elements. Cavities
between TES and existing walls have to be avoided to reduce the risk of
uncontrolled convection. An appropriate measure is the filling of cavities with soft
and dense insulation material during the assembly process. There is a good
experience with blown-in cellulose fibre that has a sorption quality that can buffer
the moisture from the existing exterior walls and reduces the moisture load on the
wooden parts of TES. The existing wall and the window connection is very often
permeable, therefore the coupling beam is horizontally sealed to inhibit vertical
convection.
The elements and element joints are made air-tight with sealing stripes
positioned between the elements and all paneling joints; thereby it is important
that the inner sealing stripe is connected air-tightly with the OSB by sealing tape.
(Fig. 9).
9 Wood Protection and Durability
The built-in wood and wood-based materials are sufficiently protected through
compliance with the previously mentioned hygrothermal requirements and
moisture management in closed timber frame components. Thus the TES elements
fulfill the requirements of use class 0. [12]
10 Fire Safety
Normally, the existing structure already fulfills the fire safety objectives. However
the condition of the existing building has to be verified by a fire safety survey. [1]
The TES-faade layer has to fulfill the requirements of the relevant building
classes. In building classes 4 and 5, in which most of the potential TES-retrofitting
objects are classified, or buildings 3 storeys, non-loadbearing exterior walls
have to be at least fire-retardant (Fig. 10 No. 1) and difficult combustible cladding
and insulation material of at least class C- or B-s2,d0 is required. Timber or other
combustible faade materials are possible but need special construction and
additional approval. Where combustible material and ventilated gaps are used, the
cladding must have a storey-wise separation either by suitable fire stops or with a
floor wise horizontal separating construction. In this construction a metal sheet
serves as fire stop. It is disposed slightly above the joint area in order not to
obstruct the screwing of the sill-beam to the horizontal coupling beam. Cladding
and inner TES-Element is separated by a gypsum board to inhibit an ignition
mainly of combustible thermal insulation in case of a burning cladding. Closing
Fig. 10 Fire safety detail

the gypsum board layer at the element joint, requires a minimum of on-site work,
the connection is backed with an additional stripe of gypsum board and
horizontally with another below the wall-plate. Thus the joining of the elements
does not require any additional gluing of the boards. (Fig. 10 No. 2) [1][14][15]
The component requirements for the ceiling have to be taken in account especially
if the existing wall is removed or quite thin. The spread of fire and smoke into the
next storey in detour over the TES-faade has to be prevented. If there does not
exist an exterior wall, by definition, the TES-joint becomes part of the existing
ceiling. (Fig. 10 No. 3)
11 Mounting
An important difference to new building is the accessibility of elements rear side

while mounting, because one side is blocked by the existing wall. The higher the
prefabrication level the less work needs to be done on site, but covered connection
points will be harder to access this way. Especially if a visible cladding has been
prefabricated, necessary joining works must leave the visible side without
damage.
Pilot projects have shown that by beveling the plane between upper sill-beam
and lower wall-plate and additionally building a tongue-and-groove connection
with the front gypsum board, the mounting of the elements is simplified a lot. The
tongue-and-groove connection has another advantage of enabling a screwing of
both elements to the coupling beam in one horizontal level and thereby minimizes
the compulsory gap in the cladding.
Before the prefabricated TES-Elements are mounted to the building, its
necessary to set the foot foundation, building up surrounding scaffolding and
remove existing windows, if no elevating platform is used.
Fig. 11 Mounting process

The mounting process is divided in four steps (Fig. 11):

(1) Fixing the coupling beam to the wall: For a fast mounting process and a
clean plane stop for the elements the coupling beam is positioned by using a laser-
assisted positioning system. The fixation points were taken during site measuring
in the survey phase. With the help of computer-aided design a leveled plane is
calculated onto the uneven surface of the existing building. The gap between
existing wall and coupling beam is filled with swelling mortar.
(2) Placing the lower element, butting the wall-plate against the coupling beam,
and fixing the connection with screws.
(3) Filling the adaption layer by blowing cellulose fiber through an opening in
the coupling beam.
(4) Insertion of the upper element into the lower element, screwing the sill-
beam to the coupling beam, and closing the wind-barrier.
12 Principles of Timber Framed Jointing in Retrofit
For the structural connection of the TES-Faade to an existing building, the

construction of the existing ceiling needs to be examined closely regarding
stability and positioning. The pull-out strength of anchoring into the floor slab
needs to be determined. In some cases there is a thin layer of masonry in front of
the ceiling edge, this has an influence on the anchorage length. The self-weight of
the faade needs be carried by the existing foundation. For the foundation system
of the TES-Faade various solutions are available [1] and should be chosen
according to the existing load-bearing capabilities.
The moisture management has to be dealt with like in new building. The claims
for hygrothermal protection are airtightness of timber framed elements back-
panelling and horizontal air stops in the vertical gap zone at each storey.
Convection and airborne sound might find also horizontal ways to neighbouring
units through leakages around window openings in exterior wall. Faults in
assembly may occur but can be improved from the inside afterwards.
By adding the TES-Faade as wall part not only the fire safety of the wall as
such has to be taken into account, but also the fire and smoke spread through gaps
between new TES-Faade and existing structure.
13 Conclusion
Generally spoken the interface of the existing building envelope and the new
timber-framed wrap around has to be planned very carefully. The building survey
has to examine all existent types of exterior walls and junctions of ceilings with
exterior walls. These are evident for structural jointing and critical routes for air,
moisture, sound and fire propagation. In the early planning process it is
prerequisite for risk mitigation. Timber framed elements for faade retrofit have
no high risks as shown in the proposed robust solutions.
14 Outlook
The timber-frame-system enables a self-bearing faade. Additionally loadbearing

components like balconies, spatial extensions, or roof-top extensions can be built
and integrated in the same system. Also HVAC-Systems can be included into
TES. Research to potential topology and detailing is ongoing.
Acknowledgements. TES EnergyFacade, run from 2008 to 2009, and still running
SmartTES, from 10/2011 to 09/2013 are funded in the European ERA-Net Woodwisdom-
Net by the German federal ministry for research BMBF, represented by PTJ.
References
[1] TES ENERGYFACADE: TES EnergyFaadeprefabricated timber based building
system for improving the energy efficiency of the building envelope, Period from
01/2008 to 12/2009, funded by BMBF, represented by PTJ,
http://www.tesenergyfacade.com
[2] Larsen, K.E., Lattke, F., Ott, S., Winter, S.: Surveying and digital workflow in energy
performance retrofit projects using prefabricated elements. Automation in
Construction 20(8), 9991011 (2011), doi:10.1016/j.autcon.2011.04.001
[3] Davidson, C.H.: The challenge of organizational design for manufactured construction.
Construction Innovation. Information, Process, Management 9(1), 4257 (2009),
doi: 10.1108/147141709109315342009
[4] Jensen, P., Olofsson, T., Johnsson, H.: Configuration through the parameterization of
building components. Automation in Construction 23, 18 (2012),
doi:10.1016/j.autcon.2011.11.016
[5] Ott, S., Loebus, S., Winter, S.: Vorgefertigte Holzfassadenelemente in der
energetischen Modernisierung. Bautechnik 90(1), 2633 (2013), doi:10.1002/
bate.201330024
[6] Lattke, F., Ott, S., Winter, S.: TES EnergyFaade - Innovative vorgefertigte
Fassadenelemente aus Holz. Holzbau (dnq) 3 (2011)
[7] Lattke, F., Ott, S., Winter, S.: TES EnergyFacade Vorfertigung bei der
energetischen Sanierung. In: Bautechnik Innovative Fassadentechnik, vol. 88(9).
Ernst & Sohn Verlag, Berlin
[8] Kolb, J.: Holzbau mit System Tragkonstruktion und Schichtaufbau der Bauteile.
Birkhuser, Basel (2007)
[9] Hubweber, et al.: Holzrahmenbau. Holzbau Handbuch. In: Holzabsatzfonds (ed.)
Informationsdienst Holz. Reihe 1, T. 1, Folge 1, Bonn (2008) ISSN 0466-2114
[10] DIN 68800-2:2012-02 Wood Preservation. Part 2: Preventive constructional
measures in buildings. Beuth Verlag, Berlin (2012)
[11] DIN 68800-1:2011-10 Wood Preservation. Part 1: General. Beuth Verlag, Berlin
(2011)
[12] EN 335:2006-10: Durability of wood and wood-based products Definition of use
classes. XXX Verlag, City (2006)
[13] E2REBUILD: Industrialised energy efficient retrofitting of residential buildings in

cold climates, Period from 01/2011 to 06/2014, funded by the EU within the Seventh
Framework Program (FP7), http://www.e2rebuild.eu
[14] HTO Research Report HTO TP2 (High-Tech-Offensive Bavaria, Subproject 2),
Brandsicherheit im mehrgeschossigem Holzbau, TU Mnchen (2009)
[15] HTO Research Report HTO TP 11 (High-Tech-Offensive Bavaria, Subproject 11),
Mechanismen der Brandweiterleitung bei Gebuden in Holzbauweise, TU Mnchen
(2009)
Green-Glued Products for Structural
Applications
Erik Serrano1,2, Jan Oscarsson2, Magdalena Sterley2, and Bertil Enquist1

1
Linnaeus University, Department of Building and Energy Technology, Vxj, Sweden
erik.serrano@lnu.se
2
SP Technical Research Institute of Sweden, Bors
Abstract. The results from bending tests on 107 laminated, green-glued, beams
manufactured from Norway spruce side boards are presented. The beams were
made by face gluing 21-25 mm thick boards using a commercial one-component
moisture curing polyurethane adhesive. In addition to the bending test results,
results from shape stability measurements after climatic cycling and bond line
strength and durability test results are also presented. The results from the bending
tests show that, by applying very simple grading rules, it is possible to obtain
beams with high bending strength (with a 5%-percentile characteristic value of
40,1 MPa) and substantial stiffness (mean value of 14360 MPa). Also the shape
stability of the beams and the strength and the durability of the interlaminar bonds
were found to be satisfactory.
Keywords: green gluing, glulam, bond line shear strength, durability, adhesive
bonds.
1 Introduction
For a typical south-Swedish saw mill, approximately 30% of the produced volume
consists of so-called side boards (here we focus on saw mills producing spruce
timber, Picea abies). Such side boards are of rather narrow dimensions, 18-25 mm
in thickness, and are not of major interest for structural applications. At the same
time, the huge number of narrow-dimension boards to be dealt with increases
considerably the costs for internal logistics at the saw mills.
The side boards are cut from the logs at a relatively large distance from the pith
and it is well known (see e.g. Steffen et al. 1997) that basic mechanical properties,
such as the modulus of elasticity along grain, are positively correlated to the dis-
tance from pith. Thus, the narrow-dimension side boards are of great interest if it
would be possible to use them in structural applications.
With the above in mind, an obvious approach would be to produce large-
dimension products from the side boards, similar to how traditional glued
laminated timber (glulam) production is done. However, in traditional glulam
production, the laminations have to be dried to approximately 12% moisture con-
tent (MC), and, in addition, laminations have to be planed within a short period of
DOI: 10.1007/978-94-007-7811-5_4, RILEM 2014
46 E. Serrano et al.
time prior to gluing. If kiln-dried thin side boards are used, such planing would
cause a substantial waste. If instead the laminations could be face-glued already in
the wet state, with a minimum of planing (or no planing at all) much would be
gained. This is the basic idea behind the research presented here and thus the ap-
proach has been to investigate the mechanical properties of green-glued laminated
products, produced from freshly sawn spruce side boards.
The research presented herein relates to three research projects running at Lin-
naeus University from 2006 to 2013. Only the main results in terms of mechanical
behaviour (stiffness and strength) from beam bending tests are reported in detail
here. In addition, investigations on the shape stability of beams exposed to varying
climates, on the durability and strength of adhesive bond lines, on the fracture
mechanical behaviour of bond lines and on the performance of green-glued finger
joints have been performed. Only some results from these investigations are
briefly mentioned here, but further details can be found in e.g. Serrano et al.
(2010), Serrano et al. (2011) and Sterley (2012).
2 Materials and Methods
2.1 Material Selection and Manufacturing Methods
2.1.1 Wood Raw Material, Grading and Adhesives

All wood material used is Norway spruce (Picea abies). The material was taken
from ordinary production at the saw mills, and was not graded before delivery to
the production facilities were the laminated beams were manufactured.
In two out of a total of three test series, the laminations were used as is, i.e. in
the saw falling quality delivered to the production facility, although with some
rejects due to vane. In the third series an effort was done to produce cross sections
with high quality outer laminations. In that test series it was decided to use two
grading criteria in order to classify the boards into inner laminations, outer lamina-
tions and rejects. First, a visual grading was performed, in which maximum knot
size was checked. For outer laminations, the maximum wide face knot size was set
to 25 mm. No criterion on maximum knot size was set for inner laminations. Fol-
lowing the visual grading, a hand-held grader (Timber Grader MTG, Brookhuis
Brookhuis Micro-Electronics) was used in order to determine the dynamic
modulus of elasticity (MOE) along the board direction of each lamination. For
outer laminations the MOE-criterion was 12 000 MPa, boards with MOE-values
between 7 000 and 12 000 MPa were used as inner laminations and boards with
MOE less than 7 000 MPa were rejected. Thus, laminations with MOE > 12 000
MPa, but with maximum knot size being > 25 mm were rejected.
In all tests, the laminations were delivered at an MC close to the fibre saturation
point, or considerably above that. The same adhesive was used in all test series a
commercial one-component, moisture curing polyurethane. The adhesive is not
specifically designed for green gluing, but is in fact an approved structural wood
Green-Glued Products for Structural Applications 47
adhesive for load bearing applications according to the European standardisation

and approval system when used in traditional dry gluing.
2.1.2 Beam Lay-Ups and Manufacturing

The original test plans included a total of 128 beams in three main series with
different variations. The variations included the annual ring orientation of the
laminations, grading of the laminations, and various climatic cycling schemes.
The various beam test series are denoted series I-III, with letters a-d denoting the
variations within each series.
Lamination lay-ups that were discussed in the planning phase are shown in
Fig. 1. Note that beam lay-ups B and C rely on the laminations being split to their
final width before gluing the laminations together to form the beam cross-section.
Beam lay-ups A and D can be manufactured by splitting the beam after gluing the
laminations together. Splitting the beam after gluing has the advantage of mini-
mising waste, since only two sides have to be planed and since the waste from the
saw cut in general is less than the waste from planing.
It was decided, based on the results from simple finite element analyses, to in-
vestigate orientations type A, B and D in the various test series. Lay-ups A and D
were chosen since these are the most practical in production and lay-up B was
chosen since the finite element analyses showed that orientation B would lead to
less distortion of the cross-section.
Two different approaches in manufacturing of the beams from the laminations
were thus used. In series I (comprising lay-up A) and in series II (comprising lay-
up D) the beams were manufactured with 120 mm wide laminations and the
beams were split into two halves of approximately 60 mm width after curing.
Thus, each such pair of beam halves comprises two beams with a good match of
the lamination properties.
In test series III annual ring orientations B and D were used, orientation D be-
ing used for reference to test series II. By comparing the distortion between orien-
tations B and D for pairs of beams with matched laminations the variability could
be kept at a minimum. The matching procedure was realised by first splitting
120 mm wide boards into two halves, one such half being used as a type D-
lamination in a type D-beam, the other half being used as a type B-lamination at
the same position in the corresponding type B-beam. Consequently, test series III
included pairs of beams, each pair being one type B-orientation beam and one type
D-orientation beam. Each lamination in a type B-beam was thus matched as close
as possible with the lamination at the corresponding position of the type D-beam.
Before gluing the laminations in series III, the laminations were also graded, as
described below. This meant that not all laminations could be used, since there
was a constraint that both halves should be placed at the same position in corre-
sponding beams. Thus, if one half was graded (based on either stiffness or knot
size) as being an outer lamination and the other as an inner lamination, both halves
had to be rejected. The somewhat complicated procedure used in series III was
used in order to minimise variability, and thereby simplify the evaluation of the
results. In production, clearly this approach cannot be used, for practical reasons.
Fig. 1 Lamination lay-ups used in the various test series
All beams were manufactured with full length laminations, thus no finger joints
were used. The beam gluing was done in semi-industrial scale beam presses. A
new pressing equipment was developed for use in series II and III due to insuffi-
cient performance of the equipment used in series I. The nominal clamping
pressure was around 0,5 MPa in series I and 0,9 1,0 MPa in series II and III. The
press time varied slightly between the various series, but was in general 45 - 60
minutes. The nominal spread rates used in applying the adhesive was 250 g/m2 in
series I and 200 g/m2 in series II and III.
After curing, and in relevant cases after splitting of the beam into two halves,
the beams were kiln dried together with structural timber of similar dimensions
(thickness). Finally, planing to target dimensions was done after kiln drying. All
test specimens manufactured for the three series are specified in Table 1. After
delivery to the laboratory, the beams to be tested without climatic cycling were in
all cases stored under controlled climate (20C/50%RH or 20C/65%RH) for at
least 3 months before testing. The beams to be exposed to climatic cycling were
stored for 3-7 months in the various climates.
Table 1 Beam series and parameter variation in manufacturing and climatic conditioning.
Laminations marked with * were split to half their width (giving two matched lamination
halves) prior to assembly of the beam.
Test Number Lay-up Laminations Grading of Beams Gluing Climates,

series of beams thicknesswidth laminations Widthdepth condi- RH (%).
length (mm3) (mm2) tions
Ia 15 A 211204900 No 50300 Dry 501
Ib 12 A 211204900 No 50300 Green 501
Ic 13 A 211204900 No 50300 Green 50-90-501
IIa 16 A 251205400 No 50300 Green 652
IIb 16 A 251205400 No 50300 Green 65-35-652
IIc 16 D 251205400 No 50300 Green 652
IId 16 D 251205400 No 50300 Green 65-35-652
IIIa 5 B 21,51205400* Yes 50300 Green 652
IIIb 7 B 21,51205400* Yes 50300 Green 65-35-90-652
IIIc 5 D 21,51205400* Yes 50300 Green 652
IIId 7 D 21,51205400* Yes 50300 Green 65- 35-90-652
1
Beams were to be tested after conditioning at climate 20C/50%RH.
2
Beams were to be tested after conditioning at climate 20C/65%RH.
2.2 Beam Bending Tests

The bending stiffness and the bending strength of the beams were evaluated. The
tests were all performed in 4-point bending, as is prescribed in EN 408. Due to the
limited length of the beams in series 1, the standard 18 times beam depth for the
total span was not possible to use. Instead, approximately 15,3 times the beam
depth was used in series I and 16,7 times the beam depth was used in series II and
III. The minimum requirement according to the standard is 16 times the beam
depth. The distance between the loads was, however, always 6 times the nominal
beam depth. The test set-up used is depicted schematically in Fig. 2 for the case of
16 times the beam depth.
Two different types of deflection measurements were used, denoted v and w in
the figure. The local deformation, v, includes the deformation due to bending only,
since there is no shear force in the mid span. The deformation denoted w includes
deformation due to shear but also due to local effect at supports. Therefore, the
most relevant stiffness to report is the one calculated using the deformation v. This
stiffness is also known as local modulus of elasticity (local MOE). All tests were
performed in displacement control with the loading speed set such that failure was
reached within 5-10 minutes.
Prior to testing to failure, the stiffness was determined by calculating the de-
formation between two load levels within the elastic domain. All test results in
terms of strength have been evaluated in relation to EN 14358, and EN 14080,
assuming a lognormal distribution of bending strength with unknown standard
deviation. The characteristic values are determined as 5-percentiles at 75% confi-
dence level. The size correction factor according to EN 1194 has been used, and
such corrected strength values are thus representative for a 600 mm deep and 150
mm wide beam. Since the beams in fact were approximately 300 mm in depth, the
actual strength of the beams in testing was approximately 13% higher than the
reported values.
Fig. 2 Test set-up used in beam bending tests
2.3 Shape Stability, Bond Line Shear Strength and Durability

The shape of the beams was measured at delivery to the University and after com-
pleting the different climatic cycles. In those measurements a special rig making it
possible to measure twist, cup, bow, crook and cross-sectional distortion (meaning
deviation from 90 angle at the corners). Here, twist, bow and crook were meas-
ured over a length of 3,72 m while cup was measured over a 270 mm length in the
height direction of the cross section.
For assessment of the adhesive bonds in the beams, shear strength and wood
failure percentage (WFP) were determined according to the European standards
EN 392 and EN 386, respectively. The delamination of the bonds according to
EN 391 was also measured. In a first test series, specimens for shear and for de-
lamination tests were taken from four beams from series II (two with a annual ring
orientation of type A and two of type D) that had been tested in bending (without
climatic cycling). In total 48+48 bonds from the type A beams and 24+24 bonds
from the type D beams were tested for shear strength and delamination, respec-
tively. A second test series was performed by taking additional specimens from
two type A that were manufactured especially for the shear and delamination tests
and thus had not been tested in bending or been exposed to climatic cycling. These
beams had two different spread rates: 150 and 200 g/m2. Here in total 24+24
bonds for each spread rate were tested for shear strength and delamination, respec-
tively. A third test series was realized by cutting specimens from beams that had
been tested in bending after being exposed to climatic cycling. These specimens
were used only for delamination tests.
3 Results
3.1 Beam Bending Tests

The results from all tests performed are summarised in Table 2 and Table 3. The
characteristic strength values of Table 3 correspond to the 5%-percentile estimates
calculated according to the provisions of EN 14358. Size-effect corrected values
are calculated according to EN 1194. Bending MOE values are based on local
MOE, cf. Fig. 2. Note that in manufacturing the test specimens according to the
schedule in Table 1, there was at one occasion a malfunction of the pressing
equipment, and thus three of the planned tests of Series III could not be per-
formed. In addition, cracks running from the end of the beams had developed
during kiln drying in three specimens, such that at testing of these beams failure
Table 2 Bending strength, fm, MOE, density and MC. kh is the size factor according to EN
1194. Density and MC (only measured in one series) refer to beam mass/beam volume at
the time of testing, after conditioning in 20C/50%RH or 20C/65%RH, cf. Table 1.
Series N fm fm kh MOE Density MC
Mean COV Mean Mean COV Mean Mean

[MPa] [%] [MPa] [MPa] [%] [kg/m3] [%]
Ia 15 43,9 9,1 38,8 14210 4.2 495 -
Ib 12 44,8 9,3 39,6 13860 6.2 498 -
Ic 12 42,6 12,6 37,6 13750 5.6 504 -
IIa 16 45,5 10,0 40,2 13680 4.7 495 ~14-16
IIb 16 45,5 14,8 40,1 13800 6.6 492 ~13-14
IIc 16 46,8 12,1 41,3 14030 5.5 507 ~14-16
IId 16 50,4 7,1 44,4 14510 6.8 504 ~13-14
IIIa 2 48,7 - 43,0 13880 - 500 -
IIIb 7 52,5 5,7 46,3 14690 3.8 495 -
IIIc 3 51,8 - 45,7 13830 - 493 -
IIId 7 49,8 5,1 44,0 14380 4.6 493 -
was due to shear along that same pre-existing crack. Such specimens have been
excluded from the evaluation. Thus, out of the planned 128 beams, only 122 are
evaluated.
Table 3 gives the results statistics for various combinations of beam lay-ups (cf.
Table 1). Most of the variations studied seem to have a relatively small influence
on average bending strength, except combination D (graded laminations) which
gives a higher strength, and also a lower COV, resulting in a considerably higher
characteristic bending strength. The MOE variation was in all cases small, both
within the test series and between test series.
Table 3. Test results, combinations of test series. kh is the size factor according to EN 1194
Series N fm fm kh MOE
Mean COV Char Mean Char Mean COV
[MPa] [%] [MPa] [MPa] [MPa] [MPa] [%]
A: Series Ib+Ic+II+III 107 47,0 11,2 36,9 41,5 32,6 14020 5.9
B: Series II+III 83 48,0 10,7 37,9 42,3 33,5 14080 5.9
C: Series II 64 47,1 11,5 36,5 41,5 32,2 14000 6.3
D: Series III 19 51,0 5,2 45,4 45,0 40,1 14360 4.3
3.2 Shape Stability, Bond Line Durability and Strength

An example of results from the shape stability measurements are given in Fig. 3,
showing the twist of the beams (measured over a length of 3,72 m) of Series III
after the various climatic conditioning schemes.
Fig. 3 Measured twist after various climatic conditioning schemes for series III
Results from the bond line durability and shear strength tests are presented in
Table 4. The results fulfil on average the strength versus wood failure percentage
requirements of EN 386. Individual specimens do however, for one of the test
series, not fulfil the requirements: Using only 150 g/m2 adhesive spread rate, two
individual tests (out of 24 tests) failed to comply with the strength requirements of
EN 386 (for structures in service class 3). For the delamination tests on bonds that
had not been subjected to climatic variations a 3,0-5,2 % delamination was
obtained. On average this meets the requirements and the spread rate 200 g/m2
appears to be enough to obtain both sufficient mechanical strength and to obtain
adequate resistance to delamination (note that some of the delamination tests were
done on specimens taken from tested beams). The test series showing highest
delamination values (see Table 4) are tests performed on specimens subjected both
to bending tests and climatic cycling.
Table 4 Results from tests of adhesive bonds. Standard deviations in parenthesis.
Lay-up and specimen type Mean shear Mean wood Delamination

strength failure (%) (%)
(MPa)
A. (200 g/m2, after bending test, no climatic cycling) 9,3 (1,4) 94 (12) 3,4
D. (200 g/m2, after bending test, no climatic cycling) 10,3 (1,4) 91 (13) 5,1
2
A(150 g/m , no bending test, no climatic cycling) 9,5 (1.9) 86 (13) 5,2
A (200 g/m2 no bending test, no climatic cycling) 9,9 (1.2) 93 (10) 3,3
A (200 g/m2 after bending test, climatic cycling) - - 19 (9)
Min 1 Max 34
D (200 g/m2 after bending test, climatic cycling) - - 17 (7)
Min 3 Max 26
Requirements for average according to EN386 6 Min. 90
8 Min. 72 Max 5
11 Min. 45
4 Discussion and Concluding Remarks
With the approach presented here it was possible to obtain laminated beams with
strength corresponding to GL36h (according to EN 1194) and almost matching the
MOE requirement (14360/14700=97,7%) of that same glulam grade. Using very
simple visual grading rules and MOE-measurements based on axial excitation, it
was possible to obtain a characteristic beam bending strength of 40,1 MPa.
In terms of shape stability (distortion of cross section and shape changes due to
moisture variations) it was found that the influence of lamination lay-up is only of
minor importance. In general, distortions (cup, twist, bow and crook) were very
small. The interlaminar bond line strength, achieved with reasonable amounts of
adhesive spread rate (200 g/m2) and moderate clamping pressure (0,5-1,0 MPa)
fulfil the requirements of the EN standards for glued laminated timber to be used
in service class 2 (on average the strength requirements for service class 3 are also
fulfilled). For some of the delamination tests series performed the requirements of
the EN-standards were also fulfilled, but for other test series this was not the case.
Since the specimens used in some of the delamination tests had been subjected to
climatic changes already before the delamination test it is difficult to assess
whether they would have passed the requirements if the tests had been performed
on specimens cut from virgin beams which is assumed to be the case in the
standard. However, for the only series performed with a normal adhesive spread
rate and virgin material specimens, the requirements were fulfilled.
On-going research (see e.g. Sterley 2012) has shown that finger-joints made in
the green state, show a good potential in terms of creating close-contact and
homogeneous bonds with high strength. It is believed that including such finger-
joints in combination with defect elimination in the laminations even better beam
performance can be expected.
In order for the timber industry to be able to take full advantage of the green
gluing technology in timber engineering applications, there is a need for standardi-
sation and/or approval procedures. Todays EN standards do not include the pos-
sibility of producing e.g. CE-marked green glued laminated beams. A CUAP
procedure for the type of product presented here is under development, with the
aim of having this approved during 2013.
Acknowledgments. This research was possible thanks to the financial support from The
Knowledge foundation, Linnaeus University and CBBT (Centre for building and living
with wood). The authors gratefully acknowledge this support.
References
EN 386:2003. Glued laminated timber Performance requirements and minimum produc-
tion requirements
EN 391:2003. Glued laminated timber Delamination test of glue lines
EN 392:1995. Glued laminated timber Shear test of glue lines
EN 1194:1999. Glued laminated timber Strength classes and determination of characteris-
tic values
EN 14080:2005. Timber structures Glued laminated timber Requirements
EN 14358:2006. Timber structures Calculation of characteristic 5-percentile values and
acceptance criteria for a sample
Anon: THE SWEDISH FOREST INDUSTRIES Facts and figures 2011. Swedish Forest
Industry Federation (2011), http://www.forestindustries.se/ (accessed
May 14, 2013)
Serrano, E., Blixt, J., Enquist, B., Kllsner, B., Oscarsson, J., Petersson, H., Sterley, M.:
Wet glued laminated beams using side boards of Norway spruce, Linnaeus University,
School of Engineering, Report No 5 (2011) ISBN: 978-91-86491-79-6
Serrano, E., Oscarsson, J., Enquist, B., Sterley, M., Petersson, H., Kllsner, B.: Green-glued
laminated beams High performance and added value. In: Proc. of World Conference
on Timber Engineering, WCTE 2010, Riva del Garda, Italy (2010)
Steffen, A., Johansson, C.-J., Wormuth, E.-W.: Study of the relationship between flatwise
and edgewise moduli of elasticity of sawn timber as a means to improve mechanical
strength grading technology. Holz als Roh und Werkstoff 55, 245253 (1997)
Sterley, M.: Characterisation of green-glued wood adhesive bonds. Dissertation, Linnaeus
University press (2012) ISBN: 978-91-86983-57-4
Experimental Analysis of a Post-tensioned
Timber Connection
Flavio Wanninger and Andrea Frangi
Chair of Structural Engineering, ETH Zurich, Zurich, Switzerland

{wanninger,frangi}@ibk.baug.ethz.ch
Abstract. The moment-rotation-behaviour of a post-tensioned beam-column timber

joint has been analysed extensively with a series of static bending tests. The timber
joint was loaded at the end of the beams in order to apply a moment to the connec-
tion. The tests were conducted with various forces in the tendon, from 300 kN up to
700 kN. The bending tests were performed with a controlled load level, so that no
failure perpendicular to the grain in the column occurred.
The maximally allowable vertical load to be applied was estimated using a simple
spring model. A final bending test was conducted in order to study the failure mode
of the post-tensioned timber beam-column joint. The vertical load on the beams was
increased until the tendon-elongation got so high that the test had to be stopped for
safety reasons.
This paper presents the main results of bending tests on the post-tensioned timber
joint. Attention will be given to the structural behaviour and to the influence of the
applied post-tensioning force on the connection stiffness. The experimental results
will be compared to the results obtained using a simplified analytical calculation
model.
Keywords: beam column connection, post-tensioning, LVL, rocking behaviour.
1 Introduction
In the past decades precast concrete frames were developed using tendons to
connect the columns and the beams [1, 2]. These systems showed favourable seis-
mic behaviour, being able to avoid residual deformations after an earthquake (self-
centering systems). Furthermore a model, the monolithic beam analogy, was
developed to describe the connection behaviour [3].
A similar system for timber was introduced in New Zealand at the University
of Canterbury [4, 5]. A timber frame made of laminated veneer lumber was post-
tensioned, resulting in a good structural behaviour. Design proposals were published
[6, 7] and first buildings were constructed [8].
Post-tensioned timber joints are also being studied at the Institute of Structural
Engineering at the ETH in Zurich. A post-tensioned beam-column timber joint has
DOI: 10.1007/978-94-007-7811-5_5, c RILEM 2014
58 F. Wanninger and A. Frangi
been developed using glued laminated timber with local strengthening of the joint
using hardwood (fraxinus). No steel elements are required for the moment-resisting
timber joint exept from a single straight tendon, which is placed in the middle of the
beam. The developed post-tensioned beam-column timber joint is characterised by
a high degree of pre-fabrication and easy assemblage on site.
Within the framework of a doctoral thesis the post-tensioned timber frame will
be further developed to also withstand significant horizontal loading. A simplified
analytical spring based calculation model is being developed in order to predict the
structural behaviour of a post-tensioned timber connection.
2 Specimen and Test Setup

The test specimen consists of two beams and a column made of glulam. The glulam
beams are made of spruce GL24h [9] except for the three bottom lamellae, which
are made of ash D40 [9]. The column is a hybrid element as well, made of spruce
and ash. The hardwood is used in the connection area between the column and the
beam (darker areas in figure 1(a)), where high stresses perpendicular to the grain
occur.
A straight tendon is inserted in the middle of the beams and attached at each
side of the beams. A 50 mm thick steel plate is used at the end of the beams for the
load transmission from the tendon into the beams. The moment-resistant connection
does not require further steel elements. The shear force between beam and column
is transferred via friction and through the small supports under the beams, which
were produced by cutting a notch into the column.
All the tests were performed at the ETH Zurich on a strong floor. A rigid steel
frame was built for the tests (figure 1(b)). The frame consists of two columns, which
are fastened to the strong floor with high-strength pre-stressed bolts. The columns
are further connected to a strong wall with two horizontal beams. The test specimen
is attached to the columns with stiff steel profiles. The load F on the beams is applied
by two cylinders, which are connected to the beams at a distance of 1.24 m from
1.62 0.36 1.62
F F
0.6
(a) (b)
Fig. 1 1(a): Test specimen (dimensions in m), 1(b): Test set up with steel frame
Experimental Analysis of a Post-tensioned Timber Connection 59
the interface column/beams (see figure 1(a)). The cylinders allow applying a load
of max. 300 kN on each side of the specimen. The cylinders are connected to a
hydraulic pump, but can be controlled separately, so that several load cases can be
investigated. It is for example possible to apply the load only on one beam, while
the other one is unloaded. One beam can also be loaded to a certain constant force
while the second beam is loaded to a different force. The weight of one beam with
two cylinders is 480 kg, which has to be accounted for in the analysis.
3 Experimental Analysis
3.1 Structural Behaviour of the Connection
While prestressing the tendon, the beams are being pressed against the column.
This leads to an initial compression at the interface, which has to be measured. A
mean initial value for the compression was estimated using LVDTs, as shown in
figure 2(a), where the measurements show a uniform horizontal displacement of
approximately -0.4 mm at the load level F=0 kN. This deformation is the initial
compression after stressing the tendon (and before the beams are loaded).
If the vertical load is applied on the beams, the interface starts rotating which
leads to decompression and to an opening of a gap. Figure 2(b) qualitatively shows
the rotation of the beams and the recorded horizontal displacements from the LVDTs
during a test for different loads F (figure 2(a)).
From the horizontal displacement measured at the interface, a rotation can be
calculated (slope of the lines in figure 2(a)). All results herein will be presented as a
function of the rotation in the connection. Since this parameter is crucial, inclinome-
ters where placed on top of the beams, which measured the inclination and therefore
the rotation. The values were used to verify the rotations measured with the LVDTs.
T
11.31
600
500
Position [mm]
400
300
0 kN
200 20kN
40kN
100 60 kN
80 kN
0
1 0.5 0 0.5 1 1.5 2 2.5 3
Displacement [mm]
(a) (b)
Fig. 2 2(a): Horizontal displacement of the beam-column interface. Position = Height of the
beam measured from the bottom. Load levels = load F on the beam, 2(b): Definition of the
rotation.
From the measured displacements at the interface it is also possible to calculate

the height of the compressive zone, the second key variable to describe the structural
behaviour of the connection.
The third value, calculated using the acquired measurements, is the maximal
compressive stress at the beam-column-interface. It is assumed that the stress distri-
bution is elastic, no plastic behaviour has been taken into account during the exper-
imental analysis.
The fourth value needed to describe the structural behaviour of the connection is
the tendon force, which is measured during the tests with a load cell.
3.2 Test without Tendon Elongation

Figure 3 shows the main results for a test conducted with a post-tensioning force of
550 kN. The structural behaviour of the post-tensioned beam-column timber joint is
characterised by 4 different stages:
Before decompression: The beam-column timber joint is subjected to compres-
sion over its total beam height (see Figure 3, x- -diagram), i.e. the compressive
zone x = 600 mm. The stiffness of the beam-column timber joint corresponds
to the slope of the moment-rotation-curve (M- ), which has the highest value
in stage 1. The two connections (left and right) behave identically.
After decompression: The gap starts opening between the column and the beam
and thus the compressive zone gets smaller. The reduction of the compressive
zone leads to a decrease of the stiffness of the joint. The moment of decom-
pression, when the gap opening occurs, is estimated to be 55 kNm. It is also
noticeable that the two connections (left and right interface) behave differently;
the right connection is stiffer than the left one. The beam-column connections
interact with each other, i.e. when the left beam is being pulled down, the right
beam wants to move upwards and vice versa. It may be possible, that the left
beam was pulled down slightly before the right one, making the right connec-
tion stiffer. This effect was very distinctive for small tendon forces and became
smaller as the tendon force increased.
Tendon elongation: When the gap opening reaches the position of the tendon in
the middle of the beam, the tendon starts to be elongated. However, in this test
this phenomenon is hardly visible; only at the end of the test, at a rotation of
5 mrad, a small increase in the tendon force can be noticed. This is confirmed by
the height of the compressive zone at the left interface, which became smaller
than 300 mm, i.e. the gap opening has reached the position of the tendon. This
effect can be seen more clearly in the test described in the next section.
Failure: Failure can not be observed during this test. There are no residual de-
formations as can also be seen in figure 3, where the moment-rotation curve
always goes back to its origin after unloading the specimen.
During the test in figure 3 the specimen remains purely elastic. The non-linear
behaviour is due to the gap opening and not due to plastic behaviour of timber. No
residual deformations could be measured at the end of the test.
120 700
100 600
500
80
M [kNm]
P [kN]
400
60
300
40 left left
200
right right
20 model 100 model
0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
[] 3 [] 3
x 10 x 10
600 10
500 8
inf [N/mm2]
400
6
x [mm]
300
4
200 left left
right 2 right
100 model model
0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
[] 3 [] 3
x 10 x 10
Fig. 3 Results of the test without tendon elongation. Key: : rotation, M: moment, P: tendon
force, x: height compressive zone, in f : maximal compressive stress at the interface.
3.3 Test with Tendon Elongation

Figure 4 shows the main results of a final test conducted with a post-tensioning
force of 550 kN. The vertical load on the beams was increased as much as possible
in order to gain information about the failure behaviour of the connection.
The connection shows a hysteretic behaviour in the moment-rotation-curve (dif-
ferent loading and unloading paths, see figure 4). Since only minor plastic deforma-
tions occurred during the test (the residual deformation were negligible) nearly no
energy was dissipated from the system. The hysteretic behaviour is mainly due to
loss of the post-tensioning force during the test (leakage in the hydraulic system).
The tendon force was 550 kN at the beginning of the test, and 510 kN at the end.
The tendon force (P) starts to increase at a rotation of 6 mrad. The gap open-
ing reaches the position of the tendon, which is therefore elongated as the load is
increased.
The compressive zone is constant at a value of 600 mm up to a rotation of
1.2 mrad, at which point it starts to decrease, indicating the moment of decom-
pression. The height of the compressive zone decreases to 150 mm. This means that
the gap has a total height of 450 mm and is 150 mm under the position of the tendon,
leading to a remarkable elongation of the tendon, which results in a tendon force of
750 kN.
200 800
150 600
M [kNm]
P [kN]
100 400
50 test 200 test

model model
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
[] []
600 25
test
500 model 20
400 inf [N/mm2] 15
x [mm]
300
10
200
5 test
100 model
0 0
0 0.01 0.02 0.03 0.04 0.05 0 0.01 0.02 0.03 0.04 0.05
[] []
Fig. 4 Results of the test with tendon elongation. Key: : rotation, M: moment, P: tendon
force, x: height compressive zone, in f : maximal compressive stress at the interface
The maximal compressive stress, calculated based on elastic behaviour, reaches

values over 20 MPa and is therefore not in the elastic range of the material any more.
Some measuring devices were used beyond their measuring range at large ro-
tations. This can be seen for the compressive zone and the maximal compressive
stress, whose values are incorrect after a rotation of 20 mrad. The tendon force and
the rotation were measured correctly at large rotations, since the force was measured
directly with a load cell and the rotation was estimated with the inclinometers.
3.4 Discussion of the Test Results

The rocking behaviour of the post-tensioned timber connection was studied with a
series of static bending tests. The joint is characterised by an initial (higher) stiff-
ness up to the point where a gap opening occurs (decompression point). After the
decompression point, the connection shows a non-linear structural behaviour due to
a decreasing contact zone between the beams and the column.
If the load is increased, the gap opening reaches the position of the tendon and the
tendon starts to be elongated while the tendon force increases. At some point, the
column will start to yield due to the ductile embedment failure of timber subjected
to compression perpendicular to the grain.
Thus, one important mechanical property governing the design is the timber em-
bedment strength perpendicular to the grain. However, during the tests only minor
plastic deformation of the timber column subjected to compression perpendicular
to the grain occurred. The residual deformations were small and the damage at the
beam column interface negligible. The test had to be stopped due to the increase
in tendon force. However, it has to be considered, that the tendon was very short
leading to a fast increase in the tendon force.
The applied tendon force governs the structural behaviour of the beam-column
connection. The tests showed that the initial stiffness (stiffness up to the moment of
decompression) increases as the tendon force increases (figure 5 left). However, the
relationship between initial stiffness and tendon force is non-linear and the tendon
force cannot be increased infinitely; an upper limit exists due to the limited strength
of the column.
4
x 10
5 250
4 200
KI [kNm/rad]
M [kNm]
3 150
2 100
1 tests 50 tendon elongation

curve fitting no elongation
0 0
0 200 400 600 800 0 0.01 0.02 0.03 0.04
P [kN] []
Fig. 5 Initial stiffness as a function of the tendon force (left), moment-rotation behaviour
with and without considering tendon elongation (from analytical model)
After decompression, when the gap has reached the position of the tendon, the
latter begins to elongate which leads to an increase in stiffness as opposed to the
case without tendon elongation (figure 5, right). Based on this observation it would
be more favourable to change the position of the tendon when designing frames with
post-tensioned timber connections for gravity loads, i.e. the tendon should be placed
in the upper part of the beam, so that the elongation of the tendon already begins at
a lower load level.
4 Analytical Model
4.1 Without Tendon Elongation
In order to describe the structural behaviour of the beam-column interface, an an-
alytical model is being developed. The elastic model simulates the column with
springs, which are only able to bear compressive forces. This allows a gap to open
as soon as decompression is reached. The beam is modelled as a rigid body, since it
is much stiffer than the column which is loaded perpendicular to the grain.
Fig. 6 Model before decompression (left) and after decompression (right)
The equilibrium of forces and bending moments leads to equations for the mo-
ment, the rotation and the compressive zone. The resulting compressive stresses are
calculated based on a linear elastic theory (see table 1).
Table 1 Equations for calculating the compressive zone, rotation, moment and stresses be-
fore decompression (left column) and after decompression (right column)
before decompression after decompression

h M
x=h (1) x = 3 (2)
2 P0
1 2 1 1 1
= (3) = (4)
c x c x

b h3 h 2 P0
M = c = c IB (5) M = P0 (6)
12 2 9 bc
2 P0
P0 M P M 1 = (8)
1,2 = = 0 + bh2 (7)
A W bh 3 b h2 PM0
6
Key: b: width of the beam, h: height of the beam, P0 : initial tendon force, M: bending moment,
: rotation, : stresses, c: spring constant, x: height of compressive zone
The only parameter needed for the model is a spring constant c. It is suggested
to calculate the spring constant by placing a uniform load on the interface. The
spring constant is then defined as the ratio between the compressive stress and the
displacement due to the load, which only depends on the geometry and the modulus
of elasticity of timber perpendicular to the grain.
4.2 With Tendon Elongation

The tendon is positioned in the middle of the beam. If the gap opens and reaches the
tendon, the latter is elongated and its force increased. Both values can be calculated
by using geometric expressions and Hookes law (equation 9 and 10).
L p = (d x) (9)
(d x)
P = P0 + A p E p (10)
Lp
Key: L p : tendon elongation, L p : length tendon, A p : cross section area tendon, E p : modulus
of elasticity tendon, P0 : initial tendon force, P: tendon force, d: position of the tendon, x:
height compressive zone, : rotation
Since the tendon force has become a function of the rotation and the height of
the compressive zone, an iterative procedure is necessary to obtain a solution for the
case with tendon elongation.
5 Comparison Tests Model and Conclusions

In figures 3 and 4 the prediction made by the spring model is plotted amongst the test
data. The spring stiffness was calculated with a modulus of elasticity perpendicular
to the grain of 860 MPa, as recommended by the EN 338 [9]. This value leads to
accurate estimations for the initial decompression of the beams.
The model fits the test data for the test without tendon elongation well. The initial
stiffness up to the moment of decompression is slightly underestimated. However,
the moment of decompression is predicted well as is the height of the compressive
zone and therefore the moment when tendon elongation starts. The stresses also
seem to be modelled reasonably well.
For the case where tendon elongation occurs, the differences between the model
and the test become noticeable. The model takes the elongation of the tendon into
account. Therefore an increase in the tendon force is visible in the test as well as
in the model. However, the increase according to the model is higher than during
the test. This leads to a model which describes a stiffer structural behaviour of the
beam-column connection than the actual behaviour of the connection, as can be seen
in the moment-rotation diagram.
One possible reason for the difference between model and tests is that the model
does not take the plastic deformations of timber subjected to compression perpen-
dicular to the grain into account. However, embedment failure was observed during
the test (a residual deformation of 2 mm was measured by the column at the in-
terface under the beams, which corresponds to an additional rotation of 1.6 mrad).
A second possible reason is that the tendon may have some space to move, since the
hole in the specimen is larger than the tendon itself.
The results of the extensive experimental analysis show the favourable structural
behaviour of the developed post-tensioned beam-column timber joint using glued
laminated timber with local strengthening of the joint with hardwood. Further ex-
perimental and numerical analysis are planned in order to study and optimise the
structural behaviour of the beam-column connection and to establish economic and
reliable design rules. A first building will be built at ETH in autumn 2013 using the
developed post-tensioned timber frame construction and will be monitored during
construction and later use.
Acknowledgements. The authors gratefully acknowledge the financial support by the

Swiss Commission for Technology and Innovation (CTI) and the industrial support of
Haring & Co.AG.
References
1. Priestley, N.M.J.: Overview of PRESSS Research Program (1991)
2. Pampanin, S., Priestley, N.M.J., Sritharan, S.: Analytical Modelling of the Seismic Be-
haviour of Precast Concrete Frames Designed with Ductile Connections. Journal of Earth-
quake Engineeringl (2001)
3. Palermo, A., Calvi, G.M., Castellani, A.: The Use of Controlled Rocking in the Seismic
Design of Bridges. PhD thesis, Politecnico di Milanol (2004)
4. Palermo, A., Pampanin, S., Buchanan, A.: Experimental Investigations on LVL Seismic
Resistant Wall and Frame Subassemblies (2006)
5. Buchanan, A., Deam, B., Fragiacomo, M., Pampanin, S., Palermo, A.: Multi-Storey Pre-
stressed Timber Buildings in New Zealand. Structural Engineering International (2008)
6. Newcombe, M.P.: Seismic Design of Multistorey Post-Tensioned Timber Buidlings
(2008)
7. Newcombe, M.P.: Multistorey Timber Buildings Seismic Design Guide (2010)
8. Buchanan, A., Palermo, A., Carradine, D., Pampanin, S.: Post-tensioned timber frame
buildings. The Structural Engineer (2011)
9. CEN. DIN EN 338 - Structural Timber - Strength classes; German version. European
Committee for Standardization (2009)
Risk Based Investigations of Partly Failed or
Damaged Timber Constructions
Gerhard Fink1 and Jochen Kohler2

1 Institute of Structural Engineering, ETH Zurich, Wolfgang-Pauli-Strasse 15,
8093 Zurich, Switzerland
fink@ibk.baug.ethz.ch
2 Department of Structural Engineering, NTNU Trondheim, Rich. Birkelandsvei 1A,
7491 Trondheim, Norway
jochen.kohler@ntnu.no
Abstract. In this paper a framework for updating the load and resistance proper-
ties of a partly failed timber construction based on different kind of information is
presented. That includes information available, such as the resistance of the failed
member(s) or the results of destructive and non-destructive measurements. In ac-
cordance to the type of information different updating methods are presented. The
application of the framework is exemplary illustrated on the case study.
Keywords: failures of structures, risk analysis, duration of load, damage,

deterioration.
1 Introduction
In the last decade a significant amount of failures in timber constructions has been
reported and analysed [1, 2]. Fortunately the majority of those constructions show
only partly failure. Depending on spatial extension of the failure (e.g. a single com-
ponent, a single connection, a group of components, etc.) the construction has to be
repaired or in the worst case rebuild, to guaranty a safe use.
In particular cases the failure occurs only within one structural member while
the remaining structural members are apparently unaffected but could be damaged
also due to additional loading caused by load redistribution. In such cases the cor-
responding engineer has to decide for different repair alternatives. Possible alter-
natives could be (a) exchange of the failed member, (b) reinforcement of the all
members or (c) complete renovation of the entire construction.
To find the optimal decision it is essential to know the extend of the damage,
the load bearing capacity of the remaining structural elements as well as the load
at the time of failure or better the entire load history of the construction. Here it
has to be considered that the load bearing capacity of the remaining structural ele-
ments is composed of the load bearing capacity at the time of construction and the
deterioration during utilization.
DOI: 10.1007/978-94-007-7811-5_6, c RILEM 2014
68 G. Fink and J. Kohler
It is well known that a failure within one structural member occur if the effective
stresses through the applied load exceed the strength of the member. Therefore both
a huge realization of the load and a small realization of the resistance is essential.
A huge realization of the load can be a result of an extreme event (e.g. extreme
snow load) or a deviation from the planed conditions (e.g. green roof: excessive
fill) or a combination of both. On the other hand a low realization of the resistance
might have the following two reasons - or their combination: (a) the basic population
of the structural components do not fulfil the requirements which are given in the
corresponding requirements; e.g. the actual bending capacity fm,g,k a is smaller than
t
the target load bearing capacity fm,g,k . (b) the basic population of the structural
components fulfil the requirements and the failed member is a very low realization
of the population. Depending on the existing reasons different repair alternatives
should be chosen.
In general it is difficult to estimate the actual material properties of existing struc-
tural components. Under the consideration of information available, such as the age
of the building, the history of load (e.g. amount and duration of load) or the amount
of the damage a first estimation of the material properties can be made. However
in the majority of the cases the gained knowledge might be not sufficient for an ap-
propriate estimation of the material properties to make a final decision. Additional
to the information at hand, several destructive or non-destructive measurements can
be performed to enhance the estimation. It is obvious that an increasing amount of
additional measurements lead to an increase of the quality of the estimation.
In the present paper a framework will be presented on which the material prop-
erties can be estimated or rather the estimated material properties can be updated
based on the gained information (load history and/or test results). Following the
application of the framework will be shown in a case study.
2 Updating
A convenient way to combine all kind of information is by using Bayes updating.
The given information (so-called prior information) will be updated with the re-
sults of inspections of measurements. It seems likely that the planed conditions,
such as the target material properties can be used as prior information. For updating
the prior it has to be differentiated between so-called equality and inequality in-
formation. Equality information corresponds to measured variables and inequality
information denotes the information obtained by a measurement that some variable
is greater than or less than some predefined limit. Further it can be differentiated
between direct and indirect information; i.e. direct measurements of the quantity of
interest and the measurement of some indicator of the quantity respectively. Using
that aspects the information can be subdivided into 4 types; c.f. Table 1.
The procedure of the updating process is depending on the type of information,
in particular between equality and inequality information. Both procedures will be
described in the following.
Risk Based Investigations of Partly Failed or Damaged Timber Constructions 69
Table 1 Types of information
Information Description Example
Type A Equality Direct Bending capacity of the failed member

Type B Inequality Direct Minimum bending capacity of the non-failed members
Type C Equality Indirect Stiffness measurement on the non-failed members
Type D Inequality Indirect Status inspection
2.1 Updating Using Equality Information

The inspected parameter (e.g. resistance of the structural members) is represented
by the variable X with the probability distribution function FX (x). The parameters
= (1 , 2 , ..., n )T of the distribution function are not known with accuracy; i.e.
they are product of engineering knowledge, physical understanding or earlier ob-
servations of the quantity. The parameters are in general expressed as random
variables itself specified by the so-called prior density function f (). The uncer-
tain parameters can be updated based on new observations of realizations of the
variable X, x = (x1 , x2 , ..., xn )T . The general scheme for updating of the parameters
= (1 , 2 , ..., n )T is:
f ()L(|x) n
f (|x) = with L(|x) fX| (xi |) (1)
f ()L(|x)d i=1
where f (|x) is the posterior distribution function of the parameters , L( |x) is
the likelihood function, representing the knowledge gained by the observations and
n is the number of observations. Based on f (|x) it is possible to calculate the
predictive distribution according to:

f (x) = fX (x| ) f (|x)d (2)
Having a normal or log-normal distributed variable with a known standard devia-

tion the approach can be simplified. In the following the analytical solution for the
predictive distribution of the strength R of structural components are described. R
can be assumed log-normal distributed. Thus Z = log(R) is a normal distributed
variable, with the parameters = (z , z )T . The standard deviation z is assumed
to be known and the mean value is a random variable z N( , ). Having direct
or indirect observations the variable Z can be updated as following:
Direct information: Assuming the variable Z will be updated with n observations x.
Then the posterior distribution of the mean is normal distributed z N( , ).
Following also the predictive distribution of the variable Z is normal distributed
Z N( , ); c.f. [3, 4].

nx + n z z2
= = n = n = n +n = 2 + z2 (3)
n n 2
Indirect information: The variable Z can be also updated with n indirect observations
of an indicator . The interrelation between the estimated and the actual strength
might be expressed through log(r) = log( ) + ; where is a normal distributed
error term N(0, ). Thus Eq. (3) can be extended with the expected mean
strength log( ) and its variability / n. The predictive distribution of the variable
Z is normal distributed Z N( , ).

nlog( ) + n z2 + 2 /n
= = (4)
n n
2.2 Updating Using Inequality Information

The prior information can also be updated having inequality information. Assuming
a member will be proof loaded up to specific stresses level l without failure. Thus
the resistance can of the structural members can be represented by the variable X
following a truncated distribution function. In Faber et al. 2000 [5] the following
approach is proposed; here FR (r) is the distribution function of the resistance:
FR (r) FR (l )

FR (r) = r l (5)
1 FR (l )
3 Collecting Information
The first and most likely step for the estimation of the remaining reliability of the
non-damaged components is the collection of information available. That includes
information about the strength class of the investigated members, the load bearing
capacity of the failed member, the minimum load bearing capacity of the non-failed
members, the design load and a detailed investigation of the applied load. Here
permanent loads (especially for green roofs) and live loads should be investigated.
Updating the Load: Based on the investigations the prior information of the perma-
nent load (based on the design) can be updated (information Type A). An example
therefore could be the density of the fill. There the target density can be updated
with measured densities. Significant differences between the target and the mea-
sured density, indicates that the build in material might be wrong. In that case more
detailed investigations should be performed and if it became more accurate that the
wrong material is built in, the prior information has to be neglected.
Load Bearing Capacity of the Failed Member: Based on the updated load the ac-
tual stresses l at the time of damage can be calculated. It obvious that the ac-
tual stresses are consistent with the load bearing capacity of the failed member
ri = l (information Type A). However, if uniform loading is assumed, and all other
members have not failed, thus l is the lower limit of their load bearing capacity
r j l , j = {1, 2, ..., n\i} (information Type B). This information can be used to
update the prior information of the strength properties (strength class). At this point
it has to be mentioned that based on the strength properties of a single structural
component it is not possible to identify if the the basic population of the structural
components do or do not fulfil the requirements which are given in the code. The
failed member can be a very low realization of the population. To analyse this further
investigations are essential.
Inspection Methods: In general the results, identified through the procedure de-
scribed in Section 2, might not be sufficient to make a final decision. Thus addi-
tional destructive and non-destructive measurement can be performed to enhance
the estimation. In the following a few selected methods are discussed. Thereby it is
assumed that the failed structural member is a glued laminated timber beam (GLT).
In Table 2 an overview of the selected methods is given.
Bending tests (GLT): The only way to gain information Type A is by testing one (or
more) of the structural components. Therefore the components have to be removed
from the building and tested into a lab. It is obvious that this method is time and
money consuming. However the gained information might be very efficient for fur-
ther decisions. It has to be considered that the adjacent components might be already
partially damaged through the additional load.
Tensile tests (Lamella / FJ): From the destroyed GLT beam the lamellas can be
cut out and tested under tension. There the tensile strength and the tensile stiffness
of the lamellas and the finger joint connections (FJ) are estimated. The specimens
should be taken from low loaded areas of the GLT to exclude lamellas that are
already deteriorated. Assuming that all GLT are produced out of the same source
material and under similar conditions (e.g. same factory) the material properties of
the beams can be estimated. Therefore models from the literature (e.g. [6]) can taken
into account.
Proof loading: The construction could be loaded again up to certain (higher) load
level. If the construction do not fail an additional information Type B is collected
(Otherwise information Type A).
MOE measurements: An indirect way to estimate the bending capacity is by using
the correlation to the bending modulus of elasticity (MOE). Therefore different non-
destructive methods exists; e.g. ultrasonic runtime or Eigenfrequency. Alternative
the MOE can also measured during a proof loading test (elastic deformation).
Visual inspections: The GLT members can investigated in detail with the focus on
the identification of possible indication of deterioration. However it is difficult to
quantify the outcomes. In best they can be used to estimate an upper value of the
load bearing capacity.
In Fig 1 a possible outcome of the updating process is illustrated. Based on the
density distributions the probability of failure and thus the reliability of the mem-
bers as well as the reliability of the structural system can be calculated. Following
Table 2 Compilation of selected DT and NDT inspection methods
Inspection method DT / NDT Information Expected Efficiency

type costs
Bending tests (GLT) DT Type A ***** *****
Tensile tests (Lamella / FJ) DT Type C *** ****
Proof loading NDT (DT) Type B (A) **** **
MOE measurements NDT Type C * *
Visual inspections NDT Type D * *
Fig. 1 Schematic illustra-

tion of a possible outcome S R actual
of the updating process target
decisions concerning the further procedure, such as repair alternatives or more de-
tailed investigations, can be made.
4 Duration of Load Load History

As mentioned before it has to be considered that the load bearing capacity of the
remaining structural elements at the investigated time ti is composed of the load
bearing capacity at the time of construction t0 and the strength deterioration during
utilization. A typical realisation of R(t) and S(t) is illustrated in Fig 2.
In timber engineering strength deterioration during utilization is of particular in-
terest in the case of huge load [7]. According to the literature load of 60% of the
maximal load bearing capacity R(t) leads to strength reduction. Here it has to be
considered that R(t) is a function of the time, thus the probability of reaching the
critical load level is increasing with time. However deterioration during utilization
is a complex topic itself and will not disused in detail within this paper.
5 Case Study
In order to illustrate the developed methodologies a case study is made. Thereby a
risk analysis will be made for a partly damaged timber-hall located in Zurich (Year
of construction: 2000). The main structural elements are n = 100 single supported
beams (b h = 200 1600mm; GL28h) with a length of 20m each. The elements
of the secondary structure are single supported beams, located between the main
structural elements. A troughed sheet together with a green roof filling constitutes
Fig. 2 Typical realisation of R, S

the load and resistance over
Realisation of R
the time, adapted from [8]
Realisation of S
Table 3 Load and resistance
Load / Resistance Char. value Distribution function

Permanent load gk = 14kN/m Normal distribution g N(14, 0.5)
Snow load gs,k = 8kN/m Gumbel distribution gs G(3, 1.5)
Bending capacity fm,k = 28MPa Log-normal distribution f m L(36.2, 5.43)
Assuming COV=0.15
the exterior shield of the roof structure. In Table 3 the applied load and resistance
are summarized.
The following failure scenario was observed: One GLT member of the structural
system failed in bending and the adjacent members showed no visible damage (Year
of damage: 2012). The load at the time of failure is investigated limitedly after the
damage and is thus assumed to be known: g + q = 20 + 4 = 24kN/m (78% of design
load).
In a first step the actual load has to be compared with the designed load. It is
obvious that the measured permanent load is significantly above the the target value
(which might be a results of the wrong fill material), whereas the snow load has not
reached a critical magnitude. For the following investigation it is assumed that the
fill will be exchanged g N(14, 0.5).
Correspondingly fm can be updated. Concerning the relatively huge load over the
entire live time of the building a constant reduction of bending strength of 20% is
assumed. Given this assumption the probability that 1 of 100 member fails under
the allied stresses m 14MPa is rather small Pf 8.1 105. This indicates that
the assumed bending bending capacity might overestimate the real bending capac-
ity; i.e. the mean bending capacity is significantly lower. Thus in addition to the
variability of fm through constant standard deviation (z COV = 0.15) a standard
deviation of the uncertain mean of = 0.10 is assumed. Following the information
of the bending capacity can be updated using information available (m = 14MPa)
and information based on additional inspections. In this particular case study the
following inspections are assumed: Bending tests on 3 GLT beams. NDT strength
estimation on 5 GLT beams (measurement uncertainty = 5MPa). Two different
Table 4 Compilation of the estimated material properties and the corresponding relability
Description Information fm fm,k Pf

n fm [MPa] [MPa] [MPa] [-] [-]
1 14
No test 22.9 17.4 3.79 7.60 105
99 14
4 {14, 18, 22, 17}
Case A 96 14 20.2 15.8 3.45 2.85 104
20 = 20
4 {14, 29, 24, 28}
Case B 96 14 25.1 19.4 4.20 1.31 105
20 = 26
structural reliability of a single component
0.14 1
0.12
0.8
Probability density
0.10
Probability
0.08 0.6
Prior
0.06 0.4 No tests
Case A
0.04
0.2 Case B
0.02
0 0
0 20 40 60 0 20 40 60
Bending strength [MPa] Bending strength [MPa]
Fig. 3 Estimated bending strength
scenarios of test result are assumed (Case A, B; c.f. Table 4). Based on information
available and the experimental investigations the corresponding strength properties
are estimated using Eq. (3-5). For practical reasons primary the equality informa-
tion is used for updating. Using the updated resistance the reliability of the structural
components expressed through the reliability index can be calculated.
In Tab 4 a compilation of the results is given. The estimated strength properties
of the remaining GLT beams are significantly different for the two assumed inspec-
tion scenarios. In Case A the estimated strength properties are significantly below
the required values fm = 20.2MPa, fm,k = 15.8MPa. This indicated that the fail-
ure occurs as a result of inadequate material properties (wrong strength grade) or a
significant amount of strength deterioration as a result of the high permanent load.
However the results clearly show that the remaining structure does not fulfil the re-
quirements. A reinforcement of the members might be the optimal solution. In Case
B the fm = 25.1MPa, fm,k = 19.4MPa which is slightly above the design values.
Even in this case reinforcement might be a solution. Alternative an exchange of the
roof structure (to reduce the permanent load) might be adequate. However in both
scenarios the experimental investigation leads to an improvement of the estimation
which could be very useful for the responsible engineer. At this point it has to be
mentioned that the here presented method is sensible to the chosen prior informa-
tion and the real strength deterioration through duration of load. For a practical
application both should be investigated in detail.
6 Conclusions and Outlook

In this study a framework for updating the load and resistance properties of a
partly failed timber construction based on different kind of information is presented.
Thereby it is primary focused on constructions where only a marginal part of the
construction failed (1 of 100 structural members). It is presented how information
available, like the load bearing capacity of the failed member or the minimum load
bearing capacity of the non-failed members can be used to estimate the load bearing
capacity of the remaining structural component. Further it is shown how information
of different types of destructive and non-destructive measurements can implemented
in the estimation. The application of the framework is illustrated on the case study.
The here presented model can be optimized having detailed knowledge about the
load history and its influence on the strength deterioration. This should be taken into
account in detail within a further study.
Acknowledgements. Part of this work was supported by COST Action FP1101 Assess-
ment, Reinforcement and Monitoring of Timber Structures.
References
1. Fruhwald, E., Serrano, E., Toratti, T., Emilsson, A., Thelandersson, S.: Design of safe
timber structures - How can we learn from structural failures in concrete, steel and timber?
Technical report, Lund University (2007)
2. Bla, H.J., Frese, M.: Schadensanalyse von Hallentragwerken aus Holz. Karlsruher
Berichte zum Ingenieurholzbau 16. KIT Scientific Publishing (2010)
3. Rackwitz, R.: Predictive distribution of strength under control. Materiaux et Construc-
tion 16(4), 259267 (1983)
4. Faber, M.H.: Statistics and Probability Theory: In Pursuit of Engineering Decision Sup-
port, vol. 18. Springer (2012)
5. Faber, M.H., Val, D.V., Stewart, M.G.: Proof load testing for bridge assessment and up-
grading. Engineering Structures 22(12), 16771689 (2000)
6. Bla, H.J., Frese, M., Glos, P., Denzler, J.K., Linenmann, P., Ranta-Maunus, A.: Zu-
verlassigkeit von Fichten-Brettschichtholz mit modifiziertem Aufbau, vol. 11. KIT Sci-
entific Publishing (2008)
7. Srensen, J.D., Svensson, S., Stang, B.D.: Reliability-based calibration of load duration
factors for timber structures. Structural Safety 27(2), 153169 (2005)
8. Melchers, R.E.: Structural reliability analysis and prediction. John Wiley & Son Ltd.
(1999)
Naturally Grown Round Wood
Ideas for an Engineering Design
Matthias Frese and Hans Joachim Bla
Karlsruhe Institute of Technology (KIT),

Timber Structures and Building Construction,
R.-Baumeister-Platz 1,
76131 Karlsruhe, Germany
{matthias.frese,blass}@kit.edu
Abstract. The hardwood species European oak, European chestnut and black
locust are hardly used for engineering structures in their naturally grown shape,
although such a use is generally possible. From forest thinnings these hardwoods
are available as low-cost material to a certain extent and their durability and re-
markable strength can fulfil the demands of robust timber structures. However, the
irregular shape of naturally grown logs and the subsequent complexity of connec-
tions inhibit engineering applications. Thus, there is a gap between the potential of
an available low-cost material and the possibility of a value-added application. In
order to bridge this gap the present contribution aims first at giving basic ideas to
make naturally grown logs more calculable and second at presenting a modular
screw connection. Hence, compression tests on naturally grown logs, screw with-
drawal tests and tests on screw connections were performed. Based on the experi-
mental results, strength models for the compression capacity and the withdrawal
resistance of screws were developed and a high strength and ductile connection
type was designed. The findings of the work may widen the possibilities in the
engineering design of timber structures made of naturally grown logs.
Keywords: round wood, logs, strength model, oak, chestnut, steel connectors.
1 Introduction
In the past, logs in their naturally grown shape played always a certain role in
heavily loaded structures and branches of the industry where material costs rather
than architectural appearance are the main aspect. Bridges built by the Romans
were founded on timber poles and more or less debarked round wood was used to
support mining adits and tunnels under construction.
A brief literature review about the structural use of naturally grown logs (NGL)
and of machined cylindrical logs (MCL) revealed various activities in the two past
decades. Broader overviews on the topic are given in Ranta-Maunus et al. (1999),
Wolfe (2000) and Stern (2001). Specific ideas how to connect NGL/MCL to each
DOI: 10.1007/978-94-007-7811-5_7, RILEM 2014
78 M. Frese and H.J. Bla
other are presented in Eckelman and Senft (1995), Lusambo and Wills (2002),
Eckelman (2004), Eckelman et al. (2007), Shim et al. (2009), Gorman et al. (2012)
and Brito and Junior (2012). The works of Wolfe and Moseley (2000), Wolfe and
Murphy (2005) and Green et al. (2008) report, among others, results of bending
and compression tests that were performed with NGL. In addition to the first three
references above, case studies about the use of NGL/MCL in buildings and struc-
tures are described in Burton et al. (1998), Cantrell et al. (2004) and Yeh and Lin
(2007). Round wood projects were erected in Doncaster (Clark R 2000) and Dor-
set (Architectural Association Inc. 2009, cf. Burton et al. 1998) in the UK.
Trinkert (2008/10/12) presents three different round wood tower structures.
The mentioned works show in particular the potential of softwood species in
naturally grown shape for various load carrying structures. On this base, the pre-
sent work aims at adding some new aspects. From thinnings in German forests
small-diameter European oak, European chestnut and black locust are available as
low-cost material to a certain extent. Due to their natural durability and hardwood
strength these species could be used in weathered space frame timber structures
(fig. 1) that need an engineering design. The problems of the present work do
therefore address the following key aspects in relation with green timber: 1. the
compression capacity of NGL and its prediction, 2. the withdrawal capacity of
screws and a corresponding strength model and 3. an idea how to join NGL mem-
bers with regard to a modular connection as well as a high capacity and pro-
nounced ductility. The present work is a continuation with wider coverage of a
prior study by the authors (Bla et al. 2012).
Fig. 1 Towers, pylons and bridges: visions of timber structures with durable hardwoods
2 Material and Methods
For the research project, 32 European chestnut trees (castanea sativa) were har-
vested in autumn 2010 and 23 European oak trees (quercus robur) as well as 16
Naturally Grown Round Wood Ideas for an Engineering Design 79
black locust trees (robinia pseudoacacia) were harvested in early spring 2011. The
stands were in Southern Germany; the one of the chestnuts was Oberkirch (Black
Forest) and the one of the oaks and the black locusts was close to the Rhine not far
away from Freiburg. The breast height diameter including bark was between 20
and 40 cm. The trees were then divided into single logs. Occasionally up to three
logs at least 5 m in originating length came from one tree. In doing so, 54
chestnut, 30 oak and 21 black locust logs were realised, in total 105 pieces. Imme-
diately after dividing the trees, the logs were X-rayed to measure the natural three-
dimensional shape of the surface between the sapwood and the bark. With the data
obtained the corresponding surface shape was numerically described to determine
the geometric properties in a later step. Prior to testing the logs in January/Febru-
ary 2012 they were stored outside without any weather protection. It was intended
to test the timber in green condition. Thus, no special measures regarding drying
of the logs were undertaken. All tests described hereafter were performed as short
term tests. Bark and bast were not removed.
Fig. 2 Preparation of the specimens for the buckling and compression tests; position of the
tree discs for the moisture and density distribution
The 105 logs were tested twice: A longer section of each log was used for a
buckling and a shorter one for a compression test. Fig. 2 shows the corresponding
division of the logs. Section 1 with a constant log length of 4350 mm and a con-
stant buckling length (sk) of 4810 mm was used for determining the compression
capacity (Fc1,max); section 2 was planned as a control for determining the compres-
sive strength (fc2). Compared to the log length, the buckling length was longer due
to mounted hinges at both ends, cf. fig. 3c. Before testing section 1, the section
modulus (W1), the cross sectional area (A1) and the maximum eccentricity (eef,max)
were calculated using the numerical surface data. W1 and A1 are mean values
representative of the middle third of the buckling length; eef,max represents the
maximum eccentricity in the middle third. These terms are exemplified in fig. 3a.
The section modulus, the cross sectional area and the maximum eccentricity do
only refer to the present wood substance without the bark. The compression mem-
bers were compressed (fig. 3b) far beyond the compression capacity with a
displacement of at least 30 mm. The compressive strength of the shorter section 2
was determined with the compression capacity (Fc2,max) and its mean cross sec-
tional area (A2). Finally, two tree discs were sawed off as stated in fig. 2 of each
log. These specimens were used to measure the moisture content (u) and the wet
density (u). In doing so, in each tree disc two outer cubes (A) were planned to
represent the outer annual rings and one inner cube (B) to represent the inner
annual rings.
Fig. 4a shows the test scheme to determine the screw withdrawal resistance
(Rax). Fully threaded screws made of stainless steel were used. The nominal di-
ameter was 8 mm. The screws were inserted into holes that were predrilled with a
diameter of 5 mm. The tensile capacity of the screws lies between 16.3 and
16.8 kN. The penetration depths (ef) were 40 to 60 mm for black locust, 60 mm
for oak and 60 to 100 mm for chestnut. Different orientations in radial und tangen-
tial direction were used. After each test the wet densities and the corresponding
moisture contents were examined. Fig. 4b exemplifies a typical failure mode
observed at the surface of a chestnut specimen.
The screw connection described in Fig. 5a was newly developed. It is shown in
an early development stage. The connection principle should include the use of
inclined screws, ensuring high stiffness, and steel connectors, easy to mount.
Regarding later realisation in timber structures as space frame structures, the
preparation of the members should be performed by computerised numerically
controlled woodworking machines. The final aim is to produce modular members.
The main features of the connection are: chestnut logs were machined to a cross-
shaped specimen. Two times four steel angles were arranged in the re-entrant
edges. Specific washers were inserted in appropriate holes in the angles to ensure
proper load transmission in the inclined screws ( = 45) connecting the timber
and the two legs of the angles (fig. 5b). The screw type was the same that was
used in the withdrawal tests. It was intended to arrange the screws with tangential
orientation to avoid screw positions in probable radial V-cracks which are typical
for round wood containing the pith. The connection was designed so that the
screws are fully embedded in the durable heartwood. The testing load was applied
at both ends at the midpoint of the four angles. It is, therefore, more or less equally
distributed between the angles. Three specimens identical in construction were
tested in tension up to a displacement of at least 15 mm in one of both connections
(either top or bottom). After the tests the moisture content and the wet density of
the wood that directly surrounds the inserted screws was determined. These
values are later used to link the results of the withdrawal tests to the results of the
connection tests.
a) b) c)
Fig. 3 Geometric log data (a), buckling test (b) and equipment for the hinge joint of the
buckling tests, the bearing shell was mounted at the end of the logs (c)
a) b)
Fig. 4 Withdrawal test with fully threaded screws (a) and typical failure (b)
a) b)
Fig. 5 Tension test (a) and connection detail with inclined screws (b)
3 Results
Fig. 6a shows load displacement curves of buckling tests. Three curves are plotted
for each species. These curves are representative of log members with the highest,
medium and lowest compression capacity. Independently of the species compres-
sion capacity is reached between 10 and 20 mm vertical displacement. The geo-
metric, mechanical and physical data of the log sections are compiled in table 1.
Since the buckling length was constant for all tested log sections, the oaks and
chestnuts feature the highest cross sectional areas and the lowest mean eccentrici-
ties and do therefore show the highest compression capacities.
Table 1 Statistics of the geometric, mechanical and physical data
N mean std min max

European oak
A1 in cm 30 411 101 220 610
eef,max in mm 30 69.5 30.1 21.3 136
Fc1,max in kN 30 377 228 110 1070
fc2 in N/mm 30 30.3 3.32 24.2 36.4
u (A) in % 120 61.7 9.41 24.9 78.6
u (B) in % 60 63.4 8.60 30.5 78.4
u (A) in kg/m 120 939 58.1 704 1070
u (B) in kg/m 60 962 52.1 789 1050
European chestnut
A1 in cm 54 427 136 207 803
eef,max in mm 54 46.9 24.4 11.5 131
Fc1,max in kN 54 509 256 126 1170
fc2 in N/mm 54 27.3 3.32 20.5 33.9
u (A) in % 216 82.7 15.8 49.5 140
u (B) in % 108 77.2 14.1 52.9 130
u (A) in kg/m 216 847 80.2 622 1022
u (B) in kg/m 108 791 66.2 606 935
Black locust
A1 in cm 18 367 70.2 259 470
eef,max in mm 18 106 38.6 41.7 171
Fc1,max in kN 21 297 118 123 571
fc2 in N/mm 21 38.9 5.10 30.2 50.1
u (A) in % 84 40.5 6.22 21.8 59.1
u (B) in % 42 38.7 5.27 26.3 47.3
u (A) in kg/m 84 851 62.0 745 1080
u (B) in kg/m 42 798 61.6 683 912
The mean compressive strengths that were determined with the control sections
2 are 27, 30 and 39 N/mm for chestnut, oak and black locust, respectively. The
stated mean moisture contents and mean wet densities are calculated on the basis
of the two tree discs containing four outer (location A) and two inner cubes (loca-
tion B). Corresponding moisture and density differences between two discs be-
longing to one log (cf. fig. 2) were low. The comparison between the locations A
and B shows moderate moisture and density differences. Considering all cubes at
the locations A and B, the mean moisture contents (mean wet densities) are ap-
proximately 40 (830), 60 (950) and 80 % (830 kg/m) for black locust, oak and
chestnut. The relation between the compression capacity and the cross sectional
area is shown in fig. 6b. Black locust and oak show lower capacities for the same
cross sectional area compared to chestnut. This is most likely due to the more
pronounced deviation from the ideal cylindrical shape of the first two species.
Regarding black locust, the influence of the compressive strength on the compres-
sive capacity seems to be low.
1.2 1.2
____ Oak
........Chestnut
_ _ _B. locust
Fc1,max [MN]
0.8 0.8
Fc1 [MN]
0.4 0.4
Oak
Chestnut
B. locust
0.0 0.0
0 5 10 15 20 100 300 500 700 900
Displacement [mm] A1 [cm]
a) b)
Fig. 6 Load displacement curves of the buckling tests (a) and compression capacity and
cross sectional area (b)
Fig. 7a shows the relation between the withdrawal resistance and the penetration
depth. The mean moisture contents (mean wet densities) are 46 (854), 43 (659)
and 30 % (842 kg/m) for the oak, chestnut and black locust specimens. Black
locust does therefore show the highest withdrawal resistance and chestnut the
lowest for a given penetration depth. The performance of oak is similar to that of
black locust. The withdrawal resistance is limited by the maximum tensile capac-
ity of 16.8 kN of the screws used.
Fig. 7b shows the load displacement curves for the three connection tests. For
each test the tension force is plotted over the displacement of the failed connection
side. Each connection showed one side that survived and one side that failed. The
capacities range from 302 to 347 kN related to a displacement of approximately
5 mm. Beyond this displacement the load slip behaviour was ductile. For the ulti-
mate displacement of 20 mm each connection showed a capacity of approximately
70 % of the maximum.
18.0 400
15.5 300
Rax [kN]
Ft [kN]
13.0 200
10.5 100 Test

Oak 1
Chestnut 2
B. locust 3
8.0 0
30 40 50 60 70 80 90 100 110 0 5 10 15 20
lef [mm] Delta on the failure side [mm]
a) b)
Fig. 7 Withdrawal resistance of screws and penetration depth (a) and load displacement
curves of the connection tests (b)
4 Strength Models and Validation of the Connection Tests
Based on the data of the buckling and withdrawal tests two strength models were
found. For that, a nonlinear and a linear regression analysis, respectively, were
performed. Here, the specified models are a scientific base to reproduce the capac-
ity of compression members and the withdrawal capacity of screws. In both
models, r is the coefficient of determination, e is the error of the prediction and N
denotes both the number of observations used and a normal distribution with mean
value and the standard deviation (in brackets), respectively. The models can be
applied independently of the three examined hardwood species.
Equation (1), a nonlinear model, describes a static equilibrium between the
maximum external compression force and the resulting internal bending and
compression stress. Therefore, the equation features linear bending-compression
interaction. It enables a capacity prediction denoted as Fc1,max,p. The prediction
necessitates the given eccentricity, the section modulus and the cross sectional
area as independent variables. The unknown model parameters, the ideal bending
(fm,i) and ideal compressive strength (fc,i), were estimated by the regression analy-
sis. The specified values are 33.5 N/mm for the ideal bending and 24.3 N/mm for
the ideal compressive strength. It is not a contradiction, that the ideal compressive
strength is lower than the ideal bending strength. From technological point of view
one would expect a higher bending strength in comparison with the compressive
strength particularly for green timber. As the ideal compressive strength compen-
sates for further capacity reducing influences that are not covered by the static
equilibrium, it is less than the mean compressive strength of fc2 = 27.3 N/mm
which is the mean compressive strength of chestnut and the lowest among the
three species, respectively (table 1). The agreement between the experimental and
predicted capacities is shown in fig. 8a. The plot shows that an individual error
distribution is indicated in particular for black locust. Normal distributions of er-

rors that depend on the three species are therefore given in term (2). Their unit is
Newton. As a result, equation (1) provides a means to calculate a characteristic
compression capacity for compression members. For that, individual limits for the
eccentricity and for the minimum cross sectional areas have to be found. For the
three hardwood species different error distributions apply. Calculated capacities
are associated with a maximum buckling length of 4810 mm. Consequently, de-
viations from this have to be adequately considered in practical applications.
e 1
Fc1,max,p = 1/ ef,max + + e
f W
m,i 1 f c,i A1 (1)
N = 102 r = 0.92
2
e:N ( 1.3 10 ;69 10 3 3
)
Oak: N = 30 e:N ( 25 103 ;62 103 )
Chestnut: N = 54 e:N ( 4.0 103 ;73 103 ) (2)
B.locust: N = 18 e:N ( 46 103 ;43 103 )
Equation (3) states a linear model. It is suitable to calculate a prediction for the
withdrawal resistance (Rax,p in Newton) of screws with a nominal diameter of 8
mm. The prediction depends on penetration depth (in mm), wet density (in kg/m)
and moisture content (as ratio). The agreement between the experimental and pre-
dicted values is shown in fig. 8b. As opposed to fig. 8a the individual errors are
well balanced around the bisection line. The model and its errors do therefore
apply without major restrictions to the three hardwood species.
ln(Rax,p ) = 6.91 + 0.0305 ef + 0.00178 u 1.33 u 2 0.000118 ef 2 + e

(3)
N = 78 r 2 = 0.81 e : N(0;0.0730)
1.2 9.8
9.6
Fc1,max [MN]
0.8
ln(Rax) [-]
9.4
0.4
9.2
Oak Oak
Chestnut Chestnut
B. locust B. locust
0.0 9.0
0.0 0.4 0.8 1.2 9.0 9.2 9.4 9.6 9.8
Fc1,max,p [MN] ln(Rax,p) [-]
a) b)
Fig. 8 Experiment over prediction: compression capacity (a) and withdrawal resistance (b)
Estimated values of the withdrawal capacity (Rax,p) were compared to the

experimental capacities (Ft,max) of the three connection tensile tests. For that, equa-
tion (3) was used. Inserting the penetration depth of 80 mm, individual wet densi-
ties and moisture contents (present in the cross-shaped specimens, cf. fig. 5a)
result in individual withdrawal capacities (table 2, column 5). A total of 28 screws
and its inclination of 45, regarding the static equilibrium, were considered to
calculate the estimated tension capacity Ft,p (column 7). The experimental capaci-
ties Ft,max (column 8) amount to 109 % of the calculated ones. The difference can
be explained among others by friction in the wood-steel interface and a screw
angle different from 90. As in general the influence of screw angles between 90
and 45, regarding the fibre deviation, on the withdrawal capacity is low, the fric-
tion coefficient that is effective in the wood-steel interface amounts to not more
than 0.09.
Table 2 Validation of the connection tests
Test ef u u Rax,p ncos Ft,p Ft,max Ft,max/Ft,p

mm kg/m - N - N N -
1 80 651 0.303 15200 280.707 300,000 326,000 1.09
2 80 787 0.480 16100 280.707 318,000 347,000 1.09
3 80 588 0.263 14000 280.707 277,000 302,000 1.09
5 Conclusions
The results obtained confirm that naturally grown logs (NGL) of the durable
hardwood species European oak, European chestnut and black locust are suitable
to be used in engineered structures that are exposed to the weather. The tests on
NGL show high compression capacities that mainly depend on the diameter and
the eccentricity. The tested connection prototype for NGL involving inclined
screws provides a high load transmission and a pronounced ductility. This is a
necessary criterion for NGL to be used e.g. in space frame structures. All the ex-
periments were performed with NGL having moisture contents above the fibre
saturation point. Consequently, the results apply to green timber and regarding
structural use, drying measures are not necessary.
More research should address the grading of NGL and the determination of
suitable limits for deviations from the ideal cylindrical shape. Screw connections
with economic steel connectors have to be enhanced. Open questions concern a
probable debarking and the long term behaviour of the less durable sapwood.
Characteristics of the axial stiffness have to be found and the effect of load
duration on the deflections is still not known.
Acknowledgements. This work was funded by Baden-Wrttembergs Ministry for Rural

Area and Consumer Protection and by the European Union. Parts of the support are means
of the European Regional Development Fund (ERDF) with the aim of Competitiveness and
Employment. The authors would like to thank their project partners University of Freiburg
Institute of Forest Utilization and Work Science, HECO-Schrauben GmbH & Co. KG
Schramberg and Forstliche Versuchs- und Forschungsanstalt Baden-Wrttemberg for their
cooperation.
References
Ranta-Maunus, A. (ed.): Round small-diameter timber for construction. VTT publications
383, Technical Research Centre of Finland, Espoo (1999)
Wolfe, R.: Research challenges for structural use of small-diameter round timbers. Forest
Products Journal 50(2), 2129 (2000)
Stern, E.G.: Construction with small-diameter roundwood. Forest Products Journal 51(4),
7182 (2001)
Eckelman, C.A., Senft, J.F.: Truss system for developing countries using small diameter
roundwood and dowel nut connection. Forest Products Journal 45(10), 7780 (1995)
Lusambo, E., Wills, B.M.D.: The strength of wire-connected round timber joints.
Biosystems Engineering 82, 339350 (2002)
Eckelman, C.A.: Exploratory study of high-strength, low-cost through-bolt with cross-pipe
and nut connections for square and roundwood timber frame construction. Forest Prod-
ucts Journal 54(12), 2937 (2004)
Eckelman, C.A., Haviarova, E., Erdil, Y.: Exploratory study of small timber trusses con-
structed with through-bolt and cross-pipe heel connectors. Forest Products Jour-
nal 57(3), 3947 (2007)
Shim, K.B., Wolfe, R.W., Begel, M.: Nailed mortised-plate connections for small-diameter
round timber. Wood and Fiber Science 41, 313321 (2009)
Gorman, T.M., Kretschmann, D.E., Begel, M., et al.: Assessing the capacity of three types
of round-wood connections. In: Quenneville, P. (ed.) Proceedings of the 12th World
Conference on Timber Engineering, Auckland, July 15-19. Curran Associates Inc., Red
Hook (2012)
Brito, L.D., Junior, C.C.: Types of connections for structural elements roundwood used in
Brazil. In: Quenneville, P. (ed.) Proceedings of the 12th World Conference on Timber
Engineering, Auckland, July 15-19. Curran Associates, Inc., Red Hook (2012)
Wolfe, R., Moseley, C.: Small-diameter log evaluation for value-added structural applica-
tions. Forest Products Journal 50(10), 4858 (2000)
Wolfe, R., Murphy, J.: Strength of small-diameter round and tapered bending members.
Forest Products Journal 55(3), 5055 (2005)
Green, D.W., Gorman, T.M., Evans, J.W., et al.: Grading and properties of small-diameter
Douglas-fir and ponderosa pine tapered logs. Forest Products Journal 58(11), 3341
(2008)
Burton, R., Dickson, M., Harris, R.: The use of roundwood thinnings in buildings a case
study. Building Research & Information 26(2), 7693 (1998)
Cantrell, R., Paun, D., LeVan-Green, S.: An empirical analysis of an innovative application
for an underutilized resource: small-diameter roundwood in recreational buildings. For-
est Products Journal 54(9), 2835 (2004)
Mc, Y., Yl, L.: Use of small thinning logs in a round-wood trussed bridge. Forest Products
Journal 57(3), 3438 (2007)
Clark, R.: The Solar Canopy and sustainable energy promotion. In: Sayigh, A.A.M. (ed.)
Proceedings of the 6th World Renewable Energy Congress, Brighton, July 1-7. Elsevier
Science Ltd., Oxford (2000)
Architectural Association Inc., Hooke Park, Buildings & Projects (2009),
http://www.aaschool.ac.uk/AALIFE/HOOKEPARK/
hookebuildings.php (accessed April 24, 2013)
Trinkert, A.: Der Turmbau zu Warstein. Bauen mit Holz 110(12), 1013 (2008)
Trinkert, A.: Ein Blick bis zu den Alpen. Bauen mit Holz 112(4), 1417 (2010)
Trinkert, A.: Mit Blick ber das Wasser. Bauen mit Holz 114(7/8), 2023 (2012)
Bla, H.J., Frese, M., Brchert, F., et al.: Computertomographische Untersuchungen und
Druckversuche an Robinienrundholz. Report No. KIT-SR 7610, KIT Scientific Publish-
ing, Karlsruhe (2012)
Recycling and End-of-Life Scenarios for Timber
Structures
Annette Hafner, Stephan Ott, and Stefan Winter
Chair of Timber Structures and Building Construction,

Technische Universitt Mnchen, Germany
{a.hafner,ott}@tum.de
Abstract. In consideration of sustainable buildings, closing life cycle loops be-

comes more and more important. Up to now reuse and recycling is taken rarely
into account in building processes. With rising consumption of wood for energetic
use recycling of material becomes more important.
Up to now there are various studies in EU market ([1], [2], [3]), which quantify
the usage of wood in market shares. Explicit calculations on recycling of wooden
material in the building sector have not yet been done. In general the demand for
reclaimed wood products in the building sector will rise due to the fact that the
preferred option has to be the reuse and the recycling of reclaimed wood. The
thermal use of wood is the last option in the cascade of use. On this option
the refinement of reclaimed wood for innovative products as well as the broaden-
ing and enhancement of the cascades of reuse and recycling is strongly needed for
the timber construction industry. Long-term and resource efficient use of wood
from premium quality (like laminated wood, plywood, timber frame construction)
is necessary to ensure sustainable construction with wood. In the process of plan-
ning new wooden construction the dismantling and reuse / recycling of the prod-
ucts has to be considered too.
In this paper outcomes of the woodwisdom-net research project ECO2 wood
in carbon efficient construction as well as calculations on wood consumption of
wide-span timber structures and investigated case studies on a very detailed level
are brought together to show the state of art and theories to improve resource effi-
cient usage of wood. Aim is a realistic estimation of theoretical scenarios for end
of life and their influence on planning processes as well as the influence on life
cycle assessment according to EN 15978. In another approach the total demolition
of an old wooden house in the Alps was evaluated. It is a typical example for a
long-used construction with numerous repair intervals, changes, and additions.
This leads to a wide variety of fractions and often to a contamination of wood
from preservatives. The fractions of the demolished house mainly consist of small
bits and pieces dedicated to different recycling options than wide span structures.
The different waste wood fractions in strength, scale, and size will tolerate certain
processing options with an emerging range of recycling products.
A better management of its renewable resources supports the material supply of
the wood sector to ensure a long-term availability of solid wood products at
DOI: 10.1007/978-94-007-7811-5_8, RILEM 2014
90 A. Hafner, S. Ott, and S. Winter
reasonable prices. This will allow preservation and also gain market shares now
and in the future.
Keywords: Life cycle, end of life, reuse, recycling, timber structures, ECO2-
project.
1 Introduction
With a growing importance of wood as significant biomass component of the re-
newable energy supply, there might be a shortage in the availability of wood at
reasonable prices in the future [4]. The European Commission therefore proposed
to increase the efficiency in the production and the use of wood [5]. A better man-
agement of its renewable resources helps the wood sector to ensure a long-term
availability of solid wood products. Solid and glued wood products should have
the chance of a second, similar product life instead of recycling without preserva-
tion of solid wood properties. Today's most important challenge is to activate the
potential of the wood sector to provide construction products fulfilling these
requirements.
For sustainable buildings closing life cycle loops becomes more and more
important. Up to now reuse and recycling of existing buildings are not examined
systematically. If at all reuse and recycling is rarely taken into account for new
building processes. The advantage of wood as a carbon storage can be obtained
through reuse and recycling of the material. Only after all options of recycling are
exploited the additional advantage of energy recovery should be used. This has a
positive effect, which will lead to improved lifecycle assessment (LCA) indicators
for wood based products. This paper shows the state of art and theories to improve
resource efficient usage of wood.
As shown in [6] the consumption of wood for material use will rise slowly in
the future, whereas the consumption of wood for energetic use will rise dramati-
cally. In Germany wood for energetic use already exceeds the material use in
2012, five to six years earlier as actual calculations from Mantau have predicted in
2010. This fact causes a rising share of fresh wood thermally used, and a shortage
and in long run rising cost for raw material supply of the timber construction sec-
tor. The use of good quality, recovered wood not only as energy source reduces
this economic pressure. Old timber structures can provide the necessary dimen-
sions and quantities of wood.
Therefore various research was done estimating the amount of material for re-
cycling in wooden buildings and wide-span timber structures. In addition related
outcomes of finished woodwisdom-net project ECO2 wood in carbon efficient
construction are included in this paper.
2 General Framework on Recycling of Wood

The European Union has set an objective to develop itself as a recycling society,
where waste generation is avoided, and waste generated is utilised as a resource.
Recycling and End-of-Life Scenarios for Timber Structures 91
The latest waste directive from 2008 [7] contains an article for reuse and recycling
of materials. Among other things, it requires that the member countries have to
proceed with necessary actions to recycle materials and products. To fulfil the
normative requirements, the industries and R&D are in charge to develop products
that can be easily recycled. In the wood product sector, the waste hierarchy is
largely underdeveloped, so far. A lot of wood products that could be utilised in
secondary product life cycle are burned for energy recovery or are down-cycled
that they lose material properties of solid wood. For realistic recycling scenarios,
the knowledge about the sources, the quantity and quality of reclaimed wood is
necessary.
There are various studies in EU market, which quantify the usage of wood in
market shares [1], [2], [3]. Up to now no exact evaluation is confirmed on how
much recycled wood exists in our building stock and how high the potential of
reuse or recycling is from this material source. There is already a small amount
from recovered wood used in the production of particle boards or wood fibre prod-
ucts but it is a small share. Explicit calculations on quantities of recovered wooden
material in building sector have not yet been done.
Beside precise evaluation about volumes of wooden material for reuse purpose,
reuse and recycling faces other obstacles. Wooden material can only be reused if it
is not contaminated with harmful substances. Timber has been treated with poi-
sonous chemicals since the beginning of the 20th century, to reduce the risk of
mould or to impregnate it against insects. This fact decreases the possibility of
reuse and even recycling of reclaimed wood nowadays. For reuse purposes the
contaminated wood has to be identified and its amount has to be quantified. Addi-
tionally the type of chemical treatment has to be analysed for bio-hazardous sub-
stances and human-hazard toxicity of treated timber construction. These quantities
have to be sorted out and treated according to the rules for hazardous wastes [8].
According to German law the term reclaimed or recovered wood (Altholz) means
used wood from production and end user, as far as it is covered by the German
life-cycle Resource Management Act [9].
3 Related Outcomes of ECO2-Project: Wood in

Carbon-Efficient Construction
The European woodwisdom research project ECO2 wood in carbon efficient

construction focused on creating holistic understanding of carbon efficiency in the
full life-cycle of a wooden building and defining technical potential and obstacles
for the use of wood in carbon efficient construction. Aim was a clear minimization
of carbon emissions during production, construction and in the full life-cycle.
Different case studies on life cycle of wooden products and buildings were exam-
ined and energy and carbon balance examined.
3.1 General Framework on Lifecycle Assessment in Buildings

According to the standard EN 15978: sustainability of construction works As-
sessment of environmental performance of buildings Calculation method the
lifecycle of a building is divided up in different modules from A to D. Fig. 2
shows the division in phases and the input of residues in lifecycle. In that context
all scenarios for recycling and in general end of life of a product are integrated in
module C. It includes deconstruction / demolition (C1), transport to the products
waste processing (C2), waste processing for reuse, recovery or recycling, recovery
and/or disposal (C3), disposal (C4).
To show possible benefits and loads of materials beyond the product system
boundary, an additional module D is introduced. This means that in module D the
recycling potential, the persistence of mineral building products, avoided impacts
of thermal energy recovery, embedded renewable energy or carbon stored in the
product can be shown. Up to now end-of-life scenarios for wooden products
nearly only consist in incineration and energy recovery.
Fig. 1 Residues from wood production in different phases of lifecycle
The treatment of stored carbon in life cycle analysis is a crucial factor which is
not handled equally in LCA calculations up to now. Trees embed carbon in the
material while growing in the forest. Several LCA calculations therefore regard
wood as GHG neutral and dont calculate them at all. This is only the case if
the wood is not leaving the forest system and if these forests are not being har-
vested. Wood from forest is counted with negative greenhouse gas emissions
when it enters module A due to the carbon stored in the product. The biogenic
carbon emissions directly attributed to a wood based product result either from the
use of biomass energy in production phase (module A) or from combustion of the
product after end-of-life (module C). These emissions are equal to the amount of
carbon sequestered in the growing tree which provides the biomass for the wood
or the energy used. Prerequisite for that is always that sustainable forestry is ap-
plied. LCA calculations done according to the standards of EN15804 and
EN15978 do not give instructions for the handling of wood and sequestration of
carbon. But according to these standards the carbon and primary energy have to be
accounted separately in the different modules. This requires that carbon balance is
shown divided up in the modules. Hence wooden materials become a negative
value in module A1 and a positive value in C4. Energy gains and the carbon stored
in the product (if it is reused or recycled) have to be shown in module D. Carbon

storage gets prolonged by recycling or reusing wooden material at end-of-life for a
second life-cycle.
3.2 Characteristics of Wood in End-of-Life of LCA

While buildings are seen as a whole in the use phase, for end-of-life it comes
down to the specific construction and the materials they are made of. Building
components can be decomposed into different layers to get a deeper understanding
of their impact at end of life. The layers of the building have different exposures,
durability and therefore a different life span. The disassembly allows the identifi-
cation of required service life of building parts and has to be considered in main-
tenance, inspection, end-of-life scenarios. In modern (timber) buildings different
layers are also common to fulfil a wide variety of technical requirements.
The principles of faade layering are divided into its structural and functional
layers. The primary construction needs to outlast the whole life span, wooden
primary construction has high mass and can store a high amount of carbon over
this period. The other layers e.g. cladding will be replaced many times and there-
fore are relevant in terms of recycling potential and burdens / benefits at EoL
stage. This can be used as design methodology for the improvement of environ-
mental performance. Through the interdependency of use- and EoL-scenarios, the
different layers can separately be optimized more easily and then be designed for
reuse. The jointing between layers and the frequency of renewal are additional
criteria for the end-of-life phase apart from the material impact. [11]
Because of a wide spread of unplanned demolition reuse of construction parts is
problematic. For optimal reuse the available construction parts have to be classi-
fied at least for different end-of-life scenarios. Construction wood and engineered
wood products can be dismantled easier, because they are used in dedicated layers,
are often used in modular assemblies and are light-weight. All these properties
allow professional waste streaming and management. After the planned disman-
tling process of removable parts, the rough demolition of monolithic portions can
take place. Such fractions may contain smaller amounts of wooden bits and pieces
which are reserved for recycling or thermal use.
4 Case Study on an Existing Wooden Building for

Calculation of Quantities
Aim of this study was a realistic assessment of the material fractions in existing
old wooden buildings. Basis for the examinations were an overall inventory analy-
sis and documentation of existing construction with classification in qualities and
quantities, see [12]. Used methods were: research with existing planning material,
survey on site and examination of material samples for hazardous substances.
The amount of wood contained in a historic cottage in the Alps, dedicated to

demolition, was evaluated. It is an example for construction with a long life cycle
with numerous repair intervals, changes, building extensions, and a conglomerate
of different materials.
Thereby reusing and recycling is not possible without a careful separation of
the materials, which are non-toxic and toxic. Via pyrolysis and use of spectros-
copy harmful substances like polycyclic aromatic hydrocarbon (PAK) and Penta-
chlorphenol (PCP) in coatings were detected. Result of the analysis showed, that
the main part of the timber construction could be reused as further components.
Only parts of the faade cladding and materials included in the construction phase
of the 1960s contained hazardous substances. Most material from the original
cottage was free of harmful substances and was also in good and dry condition.
6.7% of wooden material contained harmful substances, mainly the cladding
which was exposed to painting in the last decades of the 20th century.
Although the cottage is an example from wooden buildings material results
show the following. Analysis of material fractions in volume shows a high per-
centage of massive material in the cottage. The share of wooden material reaches a
share of 38% in volume with very simple wooden constructions. The main mass is
concrete / stone 63%. Complete analysis of construction is shown in Fig. 4.
Fig. 2 Construction material different fractions. Comparison of volume [12]
To compare results to calculations done in chapter 5 a combined value of m

wood per m gross floor area is introduced. It is 5.1 m/m or as a reciprocal value
0.20 m wood per m gross floor area. This value shows the amount of used wood
in comparison to the m of gross floor area.
5 Analysis of Wide-Span Timber Structures

The problems of the reuse of contaminated wood shift the focus on younger build-
ings without chemical treatment of the timber construction. High quality engi-
neered wood products like glulam, solid wood panels or CLT are in centre of
further investigation.
Structures with wood in every grade and in every size can be found throughout
entire Germany. TUM has a large database of with detailed information about
wide span timber structures. The presence of comprehensive data such as the
number of wood structures, their type and span, gives precise figures about the
wood volume stored. The collection of data is limited to roof structures made of
wood, as they are relatively easy to evaluate, compared to other wooden buildings.
The available beam dimensions and types of wooden roof structures are well
suited for reuse because they are serialized, modular and similar in dimension and
grade. For evaluating 116 hall structures, as well as other large-scale structures, a
wide range of use type, size and age, is recorded in tabular form. The evaluation is
adapted as a whole as well as on the individual structure type.
The classification of the parameters starts with the desired result, the amount of
used wood. To estimate the amount of used wood, span, beam depth, truss spacing
and truss number as input variables are necessary in addition to the truss form and
design. To describe the use of wood in a structure accurate, it is necessary to ex-
plain it according to its shape-defining parameters. This is the consumption of
wood in structures in relation to the footprint of the structures.

=
This ratio m/m allows a comparison of trusses with each other. The relation-
ship is suitable for a general assessment of the truss, since it contains all necessary
parameters; with the counter, namely the footprint, which includes span, truss
number and spacing, as well as the denominator, consumption of wood. The out-
put size will give the covered area and standardizes it according to the quantity
wood used. Parameters truss number and distance is secondary. There is no influ-
ence on wood consumption coefficient due to correlation between area and wood
volume, compare Fig. 3. The representation of wood consumption in relation to
the floor space for each type is calculated as an own indicator from the averages of
the demand for wood, compare Fig. 7. The nominal size of this factor increases
the resource savings. The inverse value defines the amount of exploitable recycla-
ble material per square meter floor space.
Table 1 Average results of lattice versus solid cross section trusses
Floor area Span S Amount n Distance e Wood

[m] [m] [-] [m] [m]
Solid-CS 1415 25 9 6 91
Lattice 836 19 12 5 41
The type of the saddle roof truss is the most used with over 40 observations in
the database. According to Table 1 the solid cross section has advantages in dis-
tance and number of trusses although these numbers do not influence the wood
volume. Fig. 3 shows the results of saddle roof truss with a lattice construction
versus a solid cross section. The lattice truss obviously has the lower wood con-
sumption but that definitely proofs an assumption that can be made easily. The
distribution of spans of lattice truss in Fig. 3 shows a concentration around 15
meters while the solid truss distributes over full bandwidth. A reason might be the
production effort of lattice truss is lower in short span while long spans are more
effective in solid construction due to continuous lamination technology.
450
Voll-QS
400
FW
350
Poly. (Voll-QS)
volume wood [m]
300 R = 0.74
Power (FW)
250
200
150
R = 0.67
100
50
0
0 5 10 15 20 25 30 35 40 45
Span [m]
Fig. 3 Saddle roof truss framework (FW) compared to solid cross section (Voll-QS) in
correlation with span
Fig. 4 Comparison of wide-span roof structures shows the covered floor space in relation to
1 m3 of used wood
The evaluation could be completed with the statement, that the most used
trusses have the lowest covered floor space in relation to 1 m of used wood and
therefore have the highest wood consumption. The scattering of values around the
average of 20 m/m and the plateau at this point indicates that regardless of truss
type the evaluation shows evenly distributed results without outliers. Results of
Fig. 4 normalized to the span of the three-hinged arch truss offers a variation of
8.4 to 13.6 m/m of wood consumption for the other types of trusses. The types
with the highest deviation are parallel trusses and three-hinged arch trusses.
Higher wood consumption of parallel truss can be explained with the inefficient
structural shape compared to for example saddle roof truss. The three-hinged arch
has the same problem of a continuous cross section although there is an advantage
that the normal force follows the arch shape. The type load bearing structure and
therefore the bending moment also have an influence on consumption of wood.
6 Conclusions
For future planning the reuse and recycling of buildings has to be integrated quite
early in the planning process to be able to reuse the materials in the best way.
Small scale drawings also show how components are built in and how jointing is
done. This information is useful for dismantling.
With existing building documentation fictive mass calculation could be pro-
duced which then can be updated by partial surveys. In future planners contracts
have to be enlarged to include also reuse of material in early planning stages. For
recycling of existing buildings exact calculation of actual mass and volume as well
as building condition is necessary. Therefore drawings (as built) help to calculate
overall quantities. They need to be verified on-site with survey. On-site survey
focuses on quality of materials. Only on-site reliable information on the qualities
of wood and usage of harmful substances can be gained. These analyses are im-
portant to choose material for recycling without introducing critical substances in
the recycling process. For reuse the classification system has to be improved to the
same level as the reclaimed wood classes regarding the contamination. This could
be difficult, because the way of reuse is unclear and needs further development. In
general conclusion can be drawn that the Alpine cottage is an example for a sim-
ple building with a minimized construction which allows a high level of reuse. A
comprehensive evaluation but also a heterogeneous result is achieved with data
from wide span timber structures. In difference to the historic building the amount
of wood per square meter is significantly lower and the reuse options for similar
and modular parts are different. The existing data in literature about characteristic
span of truss types is mostly correct [13]. The average amount of 0.05 m/m of
wood in these structures which equals about 21 kg/m of softwood. Although there
are losses due to the removal of construction joints and areas with dowels or nails,
there is still a reasonable amount of quality engineered wood which can be
reclaimed for a second use phase.
Cascaded use of wood can affect the LCA positively, as carbon storage is
prolonged. The material choice for wooden products will be influenced more posi-
tively, if the used wooden material is not burned immediately, but used as materi-
als. Therefore, there is a need of a recycling proof and elimination concept for a
future use of reclaimed wood of high quality.
Acknowledgments. This paper was based on research done in woodwisdom project ECO2,
bachelor thesis of Mathias Woznik, Florian Hofbauer and on-site survey with students in
summer 2012.
References
[1] Lang, A.: Charakterisierung des Altholzaufkommens in Deutschland. Rechtliche
Rahmenbedingungen Mengenpotenzial Materialkennwerte. Mitteilungen der
Bundesforschungsanstalt fr Forst- und Holzwirtschaft, Kommissionsverlag
Buchhandlung Max Wiedebusch Hamburg, vol. 215 (2004)
[2] Weimar, H.: Empirische Erhebungen im Rohstoffmarkt am Beispiel der neuen
Sektoren Altholz und Grofeuerungsanlagen. Sozialwissenschaftliche Schriften zur
forst- und Holzwirtschaft, vol. 9. Internationaler Verlag der Wissenschaften Frankfurt
am Main (2009)
[3] Mantau, U.: Rohstoffknappheit und Holzmarkt. In: Waldeigentum: Dimensionen und
Perspektiven. Springer, Heidelberg (2010)
[4] Holzzentralblatt, Leinfelden-Echterdingen (November 2012)
[5] Green Paper. On Forest Protection and Information in the EU: Preparing forests for
climate change SEC, 163 final (2010)
[6] Mantau, U., et al.: EUwood - Real potential for changes in growth and use of EU for-
ests. Final report. Hamburg/Germany, 160 p. (June 2010),
http://ec.europa.eu/energy/renewables/studies/doc/
bioenergy/euwood_final_report.pdf
[7] Directive 2008/98/EC of the European Parliament and of the Council of 19 November
2008 on waste and repealing certain Directives. Official Journal L 312, 00030030 (2008),
http://eurlex.europa.eu/LexUriServ/LexUriServ.do?uri=
OJ:L:2008:312:0003:01:EN:HTML (July 16, 2012)
[8] Verordnung ber Anforderungen an die Verwertung und Beseitigung von Altholz
Altholzverordnung AltholzV (February 24, 2012)
[9] Ordinance on the Management of Waste Wood, Bundesministerium fr Umwelt (Oc-
tober 25, 2012), http://www.bmu.de/files/pdfs/allgemein/
application/pdf/wastewood_ordinance.pdf
[10] Hafner, A., et al.: Rohstoffverwertung in Nutzungskaskaden. Holzbau die neue
Quadriga, Wolnzach 3, 4145 (2012)
[11] ECO2-book, http://www.eco2wood.com (forthcoming, Summer 2013)
[12] Hafner, A., et al.: Hllentalangerhtte case study for end of life reuse and recycling
methodologies. In: Proceedings of Sustainable Building Conference sb 2013 Munich,
Munich, pp. 14571466 (2013)
[13] Natterer, J., et al.: Holzbau Atlas Zwei, 1st edn. Birkhuser (2001) ISBN 978-
3764362300
Advancements for the Structural Application
of Fiber-Reinforced Moulded Wooden Tubes
Jrg Wehsener1, Tom-Egmont Werner2, Jens Hartig1, and Peer Haller1
1
Technische Universitt Dresden, Institute of Steel and Timber Structures,
01062 Dresden, Germany
{joerg.wehsener,jens.hartig,peer.haller}@tu-dresden.de
2
STM Montage GmbH, Cossener Strae 2, 09328 Lunzenau, Germany
STM-Montage@t-online.de
Abstract. Thermo-hygro-mechanical processes can be used to densify and to form

wood. This is applied to produce wooden tubes of structural size by shaping of
boards, which were densified previously transverse to the grain. Profiles with
hollow cross sections, like tubes, have several advantages compared to those with
compact cross sections. Due to the larger moments of inertia, a higher load-
bearing capacity for bending and buckling with a given amount of material can be
reached. Furthermore, the low thickness of the tube walls allows employing small-
sized tree sections, which are currently used only energetically or for fibre produc-
tion. Thus, production of these tubes might increase the added value of the forests.
In this contribution, a brief overview over the load-bearing behaviour of circular
moulded wooden tubes exposed to axial compression, bending and torsion is
given. Moreover, one of the first practical applications of the tubes, the tower of a
small-sized wind power plant is presented.
Keywords: moulded wooden tubes, densified wood, axial compression, poles,

wind power plant.
1 Introduction
Thermo-hygro-mechanical processes can be used to densify and to form wood

(Sandberg et al. 2013). This ability is applied to produce wooden tubes of struc-
tural size by shaping of plates, which were densified previously transverse to the
grain (Haller 2004, Haller et al. 2006).
The process steps are shown schematically in Fig. 1. At first, planks are densi-
fied up to about 50 % transverse to the grain in a heating press at temperatures of
about 120 C to 140 C. This process takes advantage of the softening of the lig-
nin leading to an almost damage-free plastic deformation of the wood cell walls.
After cooling down to ambient temperature and resetting of the lignin (< 80 C),
the densified planks are cut into slats with a thickness according to the desired
DOI: 10.1007/978-94-007-7811-5_9, RILEM 2014
100 J. Wehsener et al.
wall thickness of the tube. The slats are then glued to boards with the densification
direction of the wood oriented parallel to the plane direction. Subsequently, the
boards are steam-treated with water steam temperatures of about 100 C resulting
in plastic deformability of the wood due to re-softening of the lignin. Thus, a po-
tential recovery strain of about 100 % can be reached allowing for reshaping the
plane board to a tube. Therefore, suitable moulds and compression facilities are
necessary. After cooling down and adhering the open joint, the tube is ready for
application. In a further optional processing step, a reinforcement layer, e. g.
of fibre reinforced polymers, can be applied on the outer surface of the tube for
increasing durability, load-bearing capacity and ductility.
Fig. 1 Processes for producing moulded wooden tubes (Haller et al. 2013)
The tubes have several advantages compared to structural elements with com-
pact cross sections. Due to the larger moments of inertia, a higher load-bearing
capacity for bending and buckling with a given amount of material can be reached.
Furthermore, the low thickness of the tube walls allows employing small-
sized tree sections, which are currently used only energetically or for fibre
production. Thus, production of these tubes might increase the added value of the
forests.
Since the first ideas for producing moulded wooden tubes, about ten years ago
(Haller 2004, 2007), a considerable number of investigations was carried out to
develop applicable processes, at first, and to determine the load-bearing behaviour
and ecologic impact, afterwards. The investigations regarding the load-bearing
behaviour are briefly summarised in this contribution. More detailed information
on the ecologic impact of moulded wood compared to other building materials can
be found in (Manthey et al. 2010, Haller et al. 2011).
Moulded wooden tubes have various fields of application, for instance struc-
tural elements like columns and transport elements like industrial water lines
(Putzger et al. 2012, Haller and Nendel 2013). A first application, which was re-
cently taken in service, is the pole of a wind power plant with a height of 9 m. The
respective construction is described in here.
Advancements for the Structural Application 101
2 Performance of Moulded Wooden Tubes
2.1 Load-Bearing Behaviour

The load-bearing behaviour of moulded wooden tubes was investigated for differ-
ent types of loading with and without additional fibre reinforcement on the outer
surface. A quite detailed summary can be found in (Haller et al. 2011, 2013).
Thus, in this contribution only a very condensed overview is given.
Each of the tested specimens had a diameter of about 27 cm and a wall thick-
ness of about 2 cm. The specimens were made of spruce. Tests were performed
regarding axial compression, torsion and bending. Except for the case of axial
compression, additional fibre reinforcement on the outer surface was crucial to
compensate the low tensile strength transverse to the grain and shear strength typi-
cal for wood and to obtain a reasonable load-bearing capacity for building pur-
poses. The fibre reinforcement comprising of glass or carbon fibres and polymeric
resins was applied by a winding technique.
2.1.1 Axial Compression

The most extensive investigations (Heiduschke and Haller 2009, 2010, Haller et
al. 2011, 2013) were performed concerning compression in axial direction, which
is the most important case of loading for column-type applications. The tested
columns had a length of 2.5 m. The mean ultimate load of the unreinforced col-
umns (8 samples) was 564 kN, which corresponds to a ultimate compressive stress
of 42 N/mm and is close to the nominal compressive strength of spruce according
to DIN 68364 (fc=45 N/mm). The material behaviour is linear elastic. The failure
occurs in a brittle manner due to exceeding the tensile strength transverse to the
grain by the circumferential stresses leading to longitudinal splitting of the tube
wall.
With an outer fibre reinforcement, the low tensile strength of the wood trans-
verse to the grain can be effectively compensated. Thus, the reinforcement serves
as a confinement. With the confining fibres, increases of more than 50 % of the
ultimate load and more ductile failure modes were observed. Only a relatively low
amount of reinforcement in the magnitude of a few hundred g/m of glass or car-
bon fibres are necessary to achieve the increase in the load-bearing capacity. The
fibre orientation played a less important role concerning the performance as long
as the confining function was ensured.
2.1.2 Torsion
Wood has a relatively low resistance against torsion loading because of its low
shear strength. However, it is known that a fibre reinforcement with +/-45 % ori-
entation of the fibres can effectively increase the respective load-bearing capacity.
As an example, a tube was reinforced with 4 layers of carbon fibre reinforcement

with inclination angles of 7/-7/45/-45 (Haller et al. 2011, 2013). The amount
of fibres in each layer was 300 N/mm. The compound tube resisted a torsion
moment of about 34 kNm and failed at a principal strain of about 4 , which
corresponds to a maximum twist of 3.5/m, due to shear failure.
2.1.3 Bending
It is known that ring shaped profiles are poorly suited for bending loads. Neverthe-
less, bending almost always occur in structural applications. Consequently, a car-
bon fibre reinforced tube was exposed to a four point bending test (Haller et al.
2011, 2013) to determine the load-bearing behaviour. Two layers of carbon fibres
oriented in +/-45 direction and a fibre amount of 300 N/mm in each layer were
applied on the surface of the tube. The tube consisted of two pieces, which were
connected at midspan by finger jointing. The span of the beam was 4.5 m. The
loading positions had a distance of 1 m.
The ultimate load was reached at 63 kN. While linear elastic behaviour was ob-
served up to a load of 60 kN, gradual degradation occurred while increasing the
load until failure, which resulted from local buckling at the position of load appli-
cation. This led to a delamination of the reinforcement. However, a rupture of the
reinforcement in the tension zone was not observed after exceeding the ultimate
load.
3 Application Pole of a Wind Power Plant
Moulded wooden tubes were applied to construct a small-sized wind power plant,
which is in service since January 2013, cf. Fig. 2. The pole of the wind
power plant, which has a height of 9 m, is constructed as a bundle of three tubes.
Each tube consists of six smaller tubes of 1.5 m length due to the limitations of the
production facility. At the joints of the tubes ring-shaped internal wooden cores
with a thickness of 3 cm and a length of 40 cm are arranged, cf. Fig. 3a,c. More-
over, each two tubes are reinforced with a wounded reinforcement of polyester
resin impregnated glass fibres with a fibre weight per unit area of about
1000 g/m. The wounding was performed with a woven fabric and a fibre orienta-
tion of 0/90 related to the longitudinal axis of the tube. For improving the
surface and the durability, a glass fibre fleece and a weather-resistant topcoat were
applied.
These 3 m long tube pieces are connected by steel brackets at heights of 3 m
and 6 m, which also connects the tube bundle, cf. Fig. 3d. The fastening between
the steel and wood parts is performed by pre-drilled commercially available
screws (diameter 5 mm).
Fig. 2 Wind power plant with moulded wooden tubes in service
On top, a commercially available Darrieus wind turbine is arranged, which has

a power of 1 kW. It is attached to the tubes with another steel part connecting also
the tubes at this position. For fixation of the cantilever type construction at the
base, a sleeve foundation of steel is used, which is again attached to the tubes by
means of screws, cf. Fig. 3b,c.
Moreover, for reasons of redundancy and reducing lateral deflections at the
top of the tower, steel rods are attached to the steel brackets in longitudinal
direction of the tower to participate in bearing tensile loads in the case of bending,
cf. Fig 2.
a) b)
c) d)
Fig. 3 Construction details of the pole of the wind power plant a) Connected tubes (length
3m) before application of fibre reinforcement; b) Detail base construction; c) View into the
hollow tube (thickness 2 cm) with stiffening ring (thickness 3 cm); d) Detail steel clamp to
connect the 3m tube pieces
4 Static Verification
4.1 Material Properties

The static verification of the column consisting of three wooden tubes, which are
connected in a pointwise manner by steel clamps is a challenging task. One of the
uncertainties arise from the mechanical properties of the densified and subse-
quently recovered wood. It is not possible to rely on standards or design codes
because such does simply not exist for densified and recovered wood, hitherto.
Thus, the mechanical properties have to be estimated based on available ex-
perimental results. It is known that densified wood has increased stiffness and
strength compared to the base material because the void space given be the cell
lumen is reduced (Vorreiter 1949, Kollmann 1955, Haller & Wehsener 2004).
Therefore, stiffness and strength increase towards respective values of the cell
walls, which are the upper limits that can be reached. However, it is also known
that the brittleness increases because of a beginning thermal degradation (Vorre-
iter 1949, Kollmann 1955, Haller & Wehsener 2004).
During recovery of the densification the void space increases again and, conse-
quently, also stiffness and strength of the wood reduce. Because the crimping of
the cell walls and thermal transformations will result in a certain degradation, it
has to be assumed that stiffness and strength values will be reduced compared to
the initial material before densification (Wienhaus 1999). However, the experi-
mental investigations regarding the axial compression of the tubes, mentioned
previously, revealed that the reduction of strength is only low.
For the static verifications, it might be a convenient approach to use material
values provided by Eurocode 5 (DIN EN 1995) and choose a less grading class.
4.2 Static System

Another challenge arises from the construction. It is difficult to estimate the actual
shear stiffness between the tubes provided by the steel pieces attached by screws.
For reasons of simplicity, it is assumed for the static verification that the steel
brackets do not apply shear rigidity but only distribute the loads uniformly to the
three tubes. Thus, an increase of the moment of inertia is not taken into account
due to the shear connection but the tubes act like three parallel cantilevers with
equal loading.
The steel rods, which are attached to bear some of the tensile loads in case of
bending, are also neglected for reason of simplicity and are only considered as a
means for increasing redundancy of the structural system.
Moreover, it is assumed that tensile stresses at the tube joints are carried by the
FRP layers or the steel brackets.
4.3 Static Verifications

Following these assumptions, static verifications were performed. Besides the
dead loads of the wooden tubes including reinforcement layer and stiffening rings
(3.6 kN), the steel fasteners (0.75 kN) and the rotor (0.55 kN), wind loads were
taken into account. The wind loads were estimated with 0.8 kN/m according to
Eurocode 1-4 (DIN EN 1991-1-4). According to the datasheet, a maximum
bending moment of the rotor at the attaching position to the tubes of 1.2 kNm was
considered. The maximum torsion moment of 90 Nm caused by the rotor accord-

ing to its datasheet is neglected.
Following the assumption of equal load sharing between the tubes, each tube is,
thus, loaded at the fixing position at the base with a normal force of 2.1 kN, a
transverse force of 2.2 kN and a bending moment of 3.8 kNm. These are already
design values considering a partial safety factor of 1.35 for the dead loads and 1.5
for the wind loads according to Eurocode 5.
For estimating the material design properties, it was assumed that the moulded
wood has at least properties according GL 24h according to DIN EN 1194. This is
a conservative approach, because the applied wood was free of knots. The design
compressive/bending and shear strengths are, thus, fc,0,g,d = fm,g,d = 11.1 N/mm and
fv,g,d = 1.2 N/mm, respectively.
The verifications in the ultimate limit state were performed according to Euro-
code 5. The verification against failure of the wood at the base due to combined
bending and compression (Eurocode 5 Section 6.2.4) led to a ratio between al-
lowable and actual stresses of about 0.3, which is considerably smaller than 1 and
a conservative result. For shear failure of the wood at the base (Eurocode 5 Sec-
tion 6.1.7) the respective ratio has a value of about 0.15. The ratio of allowable
and actual stresses for the case of buckling due to bending and compression ac-
cording to (Eurocode 5 Section 6.3.2) is about 0.4.
As already mentioned, each tube consists of 6 tube pieces, which are connected
either by the fibre reinforced composite on the surface or by the steel clamps and
predrilled screws. Both connections have to be verified. It is sufficient to verify
that the appearing tensile stresses are transferred by the connections. The com-
pressive stresses are assumed to be transferred by contact pressure at the tube
cross sections. The screw connections were designed corresponding to the rec-
ommendations in (Eurocode 5 Section 8.7.1). The screw distance in circumfer-
ential direction is 4 cm. At the base, six and at the clamps in the middle of the
tubes three staggered rows of screws with a distance 1.8 cm between each row are
arranged, cf. Figs. 3b,d.
The amount of glass fibres in the composite was determined in an estimating
manner by experience such that the composite will not fail at the joints due to the
mentioned types of loading.
The verification of the deformations in the serviceability limit state is difficult
to perform in an exact manner due to the combination of different materials and
the compliance of the connections. According to (Eurocode 5 Section 7.2), the
tolerable deflections for cantilevers are for the final state (t=) in the range of
L/175 = 4.0 cm up to L/125 = 5.6 cm. Neglecting steel connections, stiffening
cores, fibre reinforcement and compliance of the fixation at the base and assuming
a mean modulus of elasticity in the final state of E0,mean,inf = 7250 N/mm (E0,mean =
11600 N/mm), the calculated deflection at the top of the cantilever is about 7 cm.
This is only about 50 % more than the allowed value. Taking into account the
fibre reinforcement, the shear connections between the tubes and the additional
steel rods will supposedly reduce the occurring deflections to the allowed value.
5 Summary and Conclusions
In this contribution, the concept of producing moulded wood profiles was briefly
reviewed and recent advancements regarding the determination of the load-
bearing behaviour as well as one of the first practical applications were addressed.
The completion of the aforementioned first wind power plant is an important step
towards placing moulded wooden tubes on the market of structural elements. An
important challenge for the future is the production of longer moulded wooden
tubes in order to reduce the effort for jointing. Moreover, it would be advanta-
geous to have tubes with larger diameters and wall thicknesses to increase the size
of the wind power plant. For a broader application of moulded wood profiles in
the building industry, a standardisation will be necessary.
Acknowledgements. The authors gratefully acknowledge the financial support from Ger-
man Federal Ministry of Economics and Technology and AIF Projekt GmbH for ZIM re-
search project "Pressholzhohlprofil fr Windkraftanlagen" under grant numbers
KF2132401WZ8 und KF2132601WZ8. Furthermore, the authors also want to thank Mr.
Thomas Hndel for producing the tubes.
References
DIN 68364:2003-05, Properties of wood species - Density, modulus of elasticity and
strength. Deutsches Institut fr Normung, Berlin (2003)
DIN EN 1194:1999-05, Timber structures Glued laminated timber Strength classes and
determination of characteristic values. Deutsches Institut fr Normung, Berlin (1999)
DIN EN 1995:2010-12, Eurocode 5: Design of timber structures. Deutsches Institut fr
Normung, Berlin (2010)
DIN EN 1991-1-4:2005-07, Eurocode 1: Actions on structures - Part 1-4: General actions,
Wind actions. Deutsches Institut fr Normung, Berlin (2005)
DIN EN 1995:2010-12, Eurocode 5: Design of timber structures. Deutsches Institut fr
Normung, Berlin (2010)
Haller, P.: Vom Baum zum Bau oder die Quadratur des Kreises. Wissenschaftliche
Zeitschrift der Technischen Universitt Dresden 53(1-2), 100104 (2004)
Haller, P.: Concepts for textile reinforcements for timber structures. Mater. Struct. 40(1),
107118 (2007), doi:10.1617/s11527-006-9153-5
Haller, P., Nendel, K.: Holz von der Rolle - Formholzrohre leiten heie Solen unter Druck.
Deutsches Ingenieurblatt 2013(5), 2831 (2013)
Haller, P., Wehsener, J.: Festigkeitsuntersuchungen an Fichtenpressholz (FPH). Eur. J.
Wood Wood Prod. 62(6), 452454 (2004), doi:10.1007/s00107-004-0516-8
Haller, P., et al.: Hochleistungsholztragwerke HHT Entwicklung von hochbelastbaren
Verbundbauweisen im Holzbau mit faserverstrkten Kunststoffen, technischen Textilien
und Formpressholz: Abschlussbericht zum BMBF-Forschungsvorhaben 0330722A-C.
Technische Universitt Dresden, Dresden (2011)
Haller, P., Putzger, R., Wehsener, J., Hartig, J.: Formholzrohre Stand der Forschung und
Anwendungen. Bautechnik 90(1), 3441 (2013), doi:10.1002/bate.201200065
Haller, P., Wehsener, J., Ziegler, S.: Wood Profile and Method for the Production of the
Same. European Patent 1390181 B1 (2006)
Heiduschke, A., Haller, P.: Zum Tragverhalten gewickelter Formholzrohre unter axialem
Druck. Bauingenieur 84(6), 262269 (2009)
Heiduschke, A., Haller, P.: Fiber-Reinforced Plastic-Confined Wood Profiles Under Axial
Compression. Struct. Eng. Int. 20(3), 246253 (2010), doi:10.2749/101686610792016772
Kollmann, F.: Technologie des Holzes und der Holzwerkstoffe, vol. 2. Springer, Berlin
(1955)
Manthey, C., Guenther, E., Heiduschke, A., Haller, P., Heistermann, T., Veljkovic, M.,
Hajek, P.: Structural, economic and environmental performance of fibre reinforced
wood profiles vs. solutions made of steel and concrete. In: Braganca, L., et al. (eds.)
Sustainability of Constructions - Integrated Approach to Life-time Structural Engineer-
ing. European Science Foundation - COST Action C25, Timisoara, pp. 275289 (2009)
Putzger, R., Eckardt, R., Eichhorn, S., Nendel, K., Haller, P.: Investigation on fibre rein-
forced moulded wood pipes for the conduction of brine at high temperature and pres-
sure. In: COST Action FP0904 Workshop, Book of Abstracts Current and Future
Trends of Thermo-Hydro-Mechanical Modification of Wood Opportunities for New
Markets, pp. 174176. Universit de Lorraine, Nancy (2012)
Sandberg, D., Haller, P., Navi, P.: Thermo-hydro and thermo-hydro-mechanical wood
processing: an opportunity for future environmentally friendly wood products. Wood
Mater. Sci. Eng. 8(1), 125 (2013), doi:10.1080/17480272.2012.751935
Vorreiter, L.: Holztechnologisches Handbuch, vol. 2. Verlag Georg Fromme, Wien (1949)
Wienhaus, O.: Modifizierung des Holzes durch eine milde Pyrolyse abgeleitet aus den
allgemeinen Prinzipien der Thermolyse des Holzes. Wissenschaftliche Zeitschrift der
Technischen Universitt Dresden 48(2), 1722 (1999)
Sole Plate Fixing Details for Modern Methods
of Timber Construction
Jesus M. Menendez, Kenneth Leitch, and Robert Hairstans
Institute for Sustainable Construction, Edinburgh Napier University, Edinburgh, UK

{j.menendez,k.leitch,r.hairstans}@napier.ac.uk
Abstract. In order to resist lateral loads, modern methods of timber construction

are reliant on the in-plane shear strength of the walls orientated parallel to the
applied action. In closed panel systems, the shear stresses are transferred to the
foundations by the sole plate through the sheathing board, which is usually me-
chanically jointed to the timber frame. Since closed panels are delivered to site as
single units, access to the internal bottom rail is rather restricted and novel effi-
cient solutions to secure the panel to the substrate are required. Sole plate fixing
components for open and closed panel systems were tested in isolation and com-
bination in order to validate a simplistic version of the weakest link theory. As a
result, findings were embedded into a software database with a direct link to a
previously developed sole plate and racking design application. This integrated
process facilitates the structural optimization of the sole plate detail.
Keywords: timber to concrete, shear wall, racking, closed panel system, MMC.
1 Introduction
The housing sector in the United Kingdom (UK), regardless of the current finan-
cial scenario, is experiencing an upward trend in housing prices mainly caused by
the long term imbalance between housing demand and number of homes built [1].
In order to mitigate this trend, the UK government has encouraged the use of
Modern Methods of Construction (MMC) to produce a higher quantity and quality
of houses [2] with a clear strategy for efficiency and elimination of waste [3].
Timber presents ideal properties to be manufactured offsite under lean princi-
ples and with a considerable high level of finishing detail. Furthermore, timber
frame closed panel systems also benefit from excellent carbon footprint, low
thermal conductivity, high strength-to-weight ratio and ease of construction. As a
result, offsite timber frame closed panel systems can be considered as a cost-
effective MMC for low energy buildings [4].
These panels are used in general as load bearing shear walls, roof and floor cas-
settes for low-rise platform construction. Nevertheless, closed panel shear walls,
contrary to open panel construction, must deal with a relatively complex connec-
tion system to the foundations or substrate. This collection of connections is
DOI: 10.1007/978-94-007-7811-5_10, RILEM 2014
110 J.M. Menendez, K. Leitch, and R. Hairstans
commonly referred as the sole plate base fixing detail (SPFD). On site access to
the SPFD is rather difficult in prefabricated closed panel units. The solution to this
problem is often found in the design of structurally convoluted sole plate geome-
tries and the inclusion of additional shear planes. As a consequence, the SPFD for
closed timber frame wall panels presents a potential decrease of strength and stiff-
ness in comparison with standard open panel construction. Furthermore, additional
design caution must be taken to comply with the minimum end and edge connec-
tion distances, particularly, if slanted fasteners are used.
This study describes the research work carried out by the Centre for Offsite
Construction and Innovative Structures (COCIS) at Edinburgh Napier University
on the optimisation of sole plate base fixing designs for closed panel timber frame
walls. The study comprises an empirical sole plate test procedure, an analytical
design methodology and the development of structural software applications for
SPFD and shear walls.
2 Experimental Procedure
This section describes a series of laboratory tests undertaken to determine the

structural behaviour of the sole plate components in isolation as a methodology to
assess its combined performance when using in service. The findings of the ex-
periment provide a dual purpose in this research study; assessing the analytical
method, further explained in section 3 and; as a validated direct input for design
specification in a software application via a database as detailed in section 4.
2.1 Sole Plate Base Fixing Components

Several materials and fasteners form this connection. Fig. 1 illustrates the different
sole plate components for open and closed panel standard construction types. The
standard connection between the bottom rail and the stud is disregarded as it is
considered part of the timber frame structure [5]. The SPFD for the closed panel
system includes a plywood packer to provide a level surface for the bottom rail to
be secured.
In order to investigate the performance of the connections shown in Fig. 1, the
following sole plate sub-connections were identified and isolated:
a) c) Type: Timber to concrete:

The study of this connection for both open and closed panel systems in-
cludes 7.5x100mm Express nail type fastener. Substrate material is dense
aggregate block of 7N/mm2 compressive strength (DAB07).
b) Type: Timber to timber:
Apart from joining the wall framing members, this connection is also
found at the base of the open panel and it is critical in terms of transfer-
ring the racking forces from the wall to the foundation. Fasteners tested
include 3.0x90mm smooth wire nail.
Sole Plate Fixing Details forr Modern Methods of Timber Construction 1111
wood to timber:
d) Type: Timber to plyw
This non-standarrd double shear connection comprises of 45x70mm tim m-
ber batten to the sole plate packer through 18mm plywood bby
4.4x115mm self--tapping screw.
e) Type: Timber to OSBB to timber:
Again, this non-sstandard double shear connection horizontally secures thhe
interlock timber to the closed panel rail by 4.4x115mm self tappinng
screw.
Fig. 1 Detail of the two sole plate details studied
2.2 Test Methodolo

ogy
Table 1 summarises the relevant
r European Standards and bespoke methods useed
in this experiment. Note that
t for fastener determination of yield moment, observeed
deformations did not com mply with BS EN 409:1993 recommendations [6]. As a
result, the double plastic hinge deformation model developed by Coste [7], annd
based on the work of Jorrissen and Blass [8], was adopted instead. Furthermorre,
embedment tests resulted d in bending of the fastener. BS EN 383 standard invalli-
dates embedment test resu ults if bending of the fastener occurs [9].
The current standard foor the design of racking walls in the UK, PD6693-1 [100],
allows the SPFD for restrraint provision against overturning moments. Thereforre,
fastener withdrawal tests were undertaken in order to assess its mechanical
property.
Table 1 Description of the tests undertaken in this study
Test Method Equipment Applied load

Lateral load capacity BS EN1380 100kN SCHENK BS EN26891
Tensile strength BS EN ISO898-1 30kN Lloyd R30k 1mm/min
Yield moment BS EN409 30kN Lloyd R30k 1mm/min
Embedment BS EN383 30kN Lloyd R30k BS EN26891
Pull through BS EN1383 100kN SCHENK 2mm/min
Withdrawal BS EN1382 100kN SCHENK 2mm/min
The overall strength performance of the SPFD is determined by the capacity of

the weakest sub-connection as a revised model of the weakest link theory. On the
other hand, the overall stiffness of the SPFD is defined by the accumulative dis-
placement occurred at each shear plane [5].
2.3 Test Results

The results of the isolated connection tests are given in Table 2. Due to the ductile
nature of the connections, the ultimate strength is based upon the measured force
resistance at 15mm of displacement.
Table 2 Strength and stiffness results of the isolated sole plate components
Connection Ultimate strength Slip modulus(1)

fmax (N) Kser (N/mm)
a) c) 4087 1594
b) 3159 1500(2)
d) 5727 1297
e) 11470 1542
(1)
Slip modulus taken as linear stiffness between 0.1-0.4Fmax
(2)
Slip modulus in accordance with BS EN 26891
Fig. 2 shows the connection types a) b) and type d) being tested with their asso-
ciated load slip curve results.
In addition to the lateral shear resistance for each connection type, the experi-
ment also includes testing of fasteners for tensile strength, yield moment and axial
withdrawal capacity. On the other hand, nominal wire and root diameter were
considered to analytically determine the mechanical properties of nails and screws
respectively. Test results, as mean values, are shown in Table 3.
Sole Plate Fixing Details for Modern Methods of Timber Construction 113
a) b)
d)
Fig. 2 Test execution and displacement results of isolated sole plate components
Table 3 Mechanical properties for sole plate fasteners
Fastener type Tensile Yield moment With- Pull-

strength My (Nmm) drawal through
fu (N/mm2) strength(1) strength(1)
(N/mm2) (N/mm2)
3.0x90mm 894 5126 6.4 50.1
SWN
4.4x115mm 1079 21562 18.2 30.5
STS
7.5x100mm 1290 44845 3.58(2) 50.0
EXN
(1)
Values given refer to the performance of the fastener when installed in C16
timber
(2)
Withdrawal strength given when installed in DAB 7N
In order to confirm the performance of the SPFD in isolation with the overall
performance of the complete detail, full sole plate fixing open and closed panel
details were also tested. These tests were performed according to a heavily modi-
fied version of the BS EN 1380 test set up so as to replicate the shear load being
transferred from the wall panel to the substrate [5]. Fig. 3 illustrates the SPFD test
set up and the load slip curves for the open (OP) and closed panel (CP) systems.
OP
CP
Fig. 3 Test execution and displacement results for full open and closed panel sole plate
details
Furthermore, Table 4 presents the strength and stiffness test results of the full
SPFD. The maximum strength value, fmax, is determined when 15mm displacement
of the bottom rail relative to the substrate occurs. Similarly, stiffness, Kser, is based
upon the displacement of the bottom rail at 40% of its ultimate strength.
Table 4 Strength and stiffness results for both open and closed panel SPFD
Sole plate Ultimate strength Slip modulus

fmax (N) Kser (N/mm)
OP 2841 1364
CP 4036 745
2.4 Sole Plate Overall Performance

Test results for the open and closed panel SPFD contained in Table 2 and Table 4
corroborate that the stiffness of the sole plate connection equals the sum of the
displacements of each individual connection [5].
On the other hand, the overall strength resistance of the SPFD is dictated by the
weakest connection with no provision for load sharing enhancing factors. Table 5
compares the test results of the connection detail in combination with the isolated
components. Hence, the proposed isolation-combination methodology is validated.
Table 5 Isolation and combination test results for open and closed panel sole plate details
Connection Isolation (N) Combination Ratio

a) b) c) d) e) (N)
OP 4087 3159 2841 0.90
CP 4087 5727 11470 4036 0.99
This validation allows for the optimisation of the SPFD in MMC by designing
effective shear planes of similar strength. However, the spacing of the fasteners
for each component along the runner also influences the design as the strength
capacity of the sole plate is given in load per meter run (N/mm).
3 Design Tool for Optimisation

In order to achieve the advantages of timber as structural material for MMC, inno-
vative products and processes need to be further exploited by the architectural,
engineering and construction (AEC) sector. Osterrieder and Richter [11] con-
cluded that the timber industry should be provided with viable design aided tools
in order to facilitate the knowledge transfer from research to the final market and
that these tools should enable data sharing instead of data exchange.
COCIS, in collaboration with industrial partners, is developing a platform to
streamline the release of research findings on innovative products and methodolo-
gies to the AEC industry [12]. The software for structural engineering Tedds, from
CSC (UK) Ltd, is used to both develop an application to design timber frame
walls and to populate a database with the SPFD test results for direct use in the
Fig. 4 Tedds racking application optimisation flow chart

116 J.M. Menendez, K. Leitch, and R. Hairstanns
Tedds application. The Tedds

T designing optimisation procedure for the rackinng
application is shown in Fiig. 4.
The development of prrofessional Tedds applications is driven by the sequentiial
completion of a series of pre-established tasks. Firstly, the user interface is deveel-
oped in C++ through the add-on application Tedds Interface Designer. Alongsidde
with the interface, detailed active drawing sketches of the wall panel, design notees
and references are produ uced. Then, according to the relevant code of practicce,
restrictive parameter valuues are defined in the interface so the design falls withiin
the scope and assumption ns of the method. Fig. 5 illustrates a snapshot of the rackk-
ing application interface.
Fig. 5 User interface for Ted

dds Racking App
Once the interface is operative, a flow chart with logical expressions of the caal-
culation methodology is programed
p in C++ through another auxiliary applicationn:
Tedds Calc Designer (Fiig. 6). This process will create the output report of thhe
calculation in a word padd fashion. The user can select a summary or a full outpuut
calculation.
In the meantime, the results
r achieved in the SPFD tests are transferred into a
Tedds database by the app plication Tedds DataList Designer. Variables are defineed
to describe the geometry, properties and components of the SPFD. The name annd
units of these variables are consistent with the Tedds applications in order to
ensure a correct informatiion data sharing.
After the application is completed and tested, developmental files are packed
up into one single executable file and distributed to the Tedds users who will in-
stall the new calculation directly on the standard Tedds program.
Fig. 6 Calculation flow chart for Tedds Racking App
The user begins the calculation by introducing the design input parameters. In-
stantaneously, the outcome of these parameters is displayed in the interface. As a
result, the user can accordingly adjust relevant variables, such as member density,
member thickness or spacing of fasteners, in order to optimise the overall strength
and stiffness of the SPFD and the racking wall panel. Once the user accepts the
final design, the software will report the calculation ready for verification and
approval.
4 Discussion and Conclusions
In comparison to on-site open panel construction, sole plate fixing details in tim-
ber MMC systems, such as racking wall panels, are critical as accessibility is
restrained resulting in the introduction of more and complex shear planes in the
design.
A series of laboratory tests for two different SPFD, open and closed panel, were
carried out in order to verify the hypothesis of the isolation combination meth-
odology where the overall sole plate resistance is given by the strength of the
weakest component. The stiffness of the sole plate fixing results in the addition of
the individual displacement of the components which is directly proportional to
the size of the fastener and the density of members. As a result, the research find-
ings can facilitate the optimisation of the sole plate fixing detail by enhancing the
limiting structural parameters of the critical component part.
Furthermore, findings on structural timber research can rapidly be disseminated
by means of structural software applications in Tedds (CSC UK Ltd). Tedds has a
potential outreach of more than 15,000 engineers in the UK. Moreover, the output
of this research is released in the Tedds Racking Design application via direct
database input. The final application is installed in the standard Tedds library of
calculations after being beta tested.
Acknowledgments. The authors would like to appreciate the financial support provided by
the Engineering and Physical Science Research Council (EPSRC). Software licenses and
technical support offered by CSC (UK) Ltd is also acknowledged.
References
[1] Barker, K.: Review of Housing supply - Delivering stability: securing our future
future housing needs. HMSO, London (2004) ISBN 1-84532-010-7
[2] Egan, J.: Rethinking construction. Department of the Environment, Transport and
the Regions, London (1998)
[3] Office C. Government Construction Strategy. Cabinet Office, London (2011)
[4] Hairstans, R.: Off-Site and Modern Methods of Timber Construction: A Sustainable
Approach. TRADA Technology Limited (2010)
[5] Leitch K. The development of a hybrid racking panel. Edinburgh Napier University,
Edinburgh (2013)
[6] BSI. BS EN 409:1993. Timber structures - Test methods - Determination of the
yield moment of dowel type fasteners - Nails. BSI (1993)
[7] Coste, G.: The assessment and applications of a new connector type for use in struc-
tural timber systems. In: University EN, ed. United Kingdom (2010)
[8] Jorissen, A.J.M., Blass, H.J.: The fastener yield strength in bending. Savonlinna
(1998)
[9] BSI. BS EN 383:2007. Timber Structures - Test methods - Determination of em-
bedment strength and foundation values for dowel type fasteners. British Standard
Institution, London (2007)
[10] BSI. PD6693-1. Complementary Information to Eurocode 5 Design of timber struc-
tures Part 1 General Common rules and rules for buildings, p. 66. British Standard
Institution, London (2012)
[11] Osterrieder, P., Richter, S., Fischer, M.: A product data model for design and fabri-
cation of timber buildings. In: WCTE 2004 Proceedings, Lahti (2004)
[12] Menendez, J., Hairstans, R., Leitch, K., Turnbull, D.: A Structural Engineering Plat-
form for Timber Modern Methods of Construction. In: International Conference on
Innovation in Architecture, Engineering and Construction, Sao Paulo, August 14-18
(2012)
Thin-Walled Timber Structures
Benoit P. Gilbert1, Steven B. Hancock1, and Henri Bailleres2

1
Griffith School of Engineering, Griffith University, Australia
b.gilbert@griffith.edu.au
2
Salisbury Research Centre, Department of Agriculture, Fisheries and Forestry,
Queensland Government, Australia
Abstract. Due to their efficiency, lightweight, ease of erection and low cost, steel
and aluminium thin-walled structures have become very popular in the construc-
tion industry over the past few decades. Applications include roof and wall sys-
tems (purlins and girts), storage racks, and composite concrete and steel slabs. The
effectiveness of these structures lies in the cross-sectional shape of the profiles
which enhances their strength by controlling the three fundamental buckling
modes: local, distortional, and global. However, despite the attractiveness of these
structures, steel and aluminium are greenhouse gas intensive materials and do not
produce sustainable structural products. This paper presents an investigation per-
formed at the Griffith School of Engineering, Griffith University, which shows
manufacturing these types of profiles in timber is possible. Short composite thin-
walled timber Cee-sections (500 mm long) were fabricated by gluing together thin
softwood (Araucaria cunninghamii) veneers (1 mm thick). Two types of Cee-
sections were considered, one with a web stiffener to increase the local buckling
capacity of the profile and one without. The profiles were tested in compression
and the test results are presented and discussed in the paper in terms of structural
behaviour and performance. Further research directions are proposed in order to
provide efficient and lightweight sustainable structural products to the timber
industry.
Keywords: thin walled timber C-sections, veneers, axial compression, column

tests, post-buckling behaviour, Araucaria Cunninghamii.
1 Introduction
Cold-formed steel and aluminium thin-walled profiles are commonly used in the
construction industry due to their efficiency, lightweight, ease of erection and low
cost [1]. The worldwide market for these products is significant and estimated at
more than two billions dollars. Typical industrial and civil engineering applica-
tions include roof and wall systems (purlins and girts), steel storage racks and
composite concrete and steel slabs. Typical cold-formed steel profiles are shown
in Figure 1.
DOI: 10.1007/978-94-007-7811-5_11, RILEM 2014
120 B.P. Gilbert, S.B. Hancock, and H. Bailleres
Due to the nature of the manufacturing process, consisting of bending a thin

sheet of metal to a desired cross-sectional shape, open profiles such as Cee- or
Zee-cross-sections are generally used. Three fundamental buckling modes,
referred to as local, distortional and global (see Figure 2), can occur for these pro-
files and usually govern the compressive and bending strength. The section
capacity for local buckling is typical enhanced by adding intermediate stiffeners to
walls having significantly large width-to-thickness ratios (see Figure 1), while the
section capacity for distortional buckling can be enhanced by changing the cross-
sectional shape, mainly by adding lip stiffeners.
Fig. 1 Examples of cold-formed steel profiles, with and without intermediate stiffeners
(a) (b) (c)
Fig. 2 Fundamental buckling modes for thin-walled open cross-sections in compression, (a)
local, (b) distortional and (c) global
However, despite the attractiveness of these types of profiles, steel and alumin-
ium are greenhouse gas intensive materials and do not produce sustainable struc-
tural products [2]. Manufacturing profiles similar to the ones shown in Figure 1 in
timber is possible, but to the authors best knowledge, has not yet been investigat-
ed in the published literature. This paper presents the first step of a study aiming at
developing extra light and structurally sound timber profiles for applications in-
cluding amongst others (i) emergency shelters which can be rapidly assembled
and disassembled, (ii) purlins for major timber buildings, as a substitute to steel
purlins currently used and (iii) wall stubs.
Thin-Walled Timber Structures 121
Short composite thin-walled timber Cee-sections (500 mm long) were fabri-

cated by gluing together thin softwood (Hoop pine - Araucaria cunninghamii)
veneers. Two types of Cee-sections were considered, namely one with a web stiff-
ener to increase the local buckling capacity of the profile and one without, and this
paper introduces the steps involved in the manufacturing process. The sections
were tested in compression and the test results are presented and discussed in the
paper in terms of structural behaviour and performance. Further research direc-
tions are also proposed.
2 Investigated Sections
2.1 General
Three sets of two 210 mm deep 105 mm wide Cee-sections (with 50 mm lip
stiffeners) were manufactured and investigated. The two cross-sections in a set are
of different type. Type A cross-section had no intermediate stiffeners, as shown in
Figure 3 (a), and the cross-section was designed so local buckling of the web
would govern the strength of the profile. On the contrary, a web stiffener was
added to the second type of cross-section (Type B), as shown in Figure 3 (b), so
the local buckling capacity of the cross-section would theoretically be higher than
the maximum compressive strength of the material. The nominal section proper-
ties of the cross-sections are given in Table 1.
(a) (b)
Fig. 3 Cee-sections manufactured and tested, (a) Type A and (b) Type B
Each cross-section was composed of 5 layers of nominal 1 mm thick Hoop pine

rotary sliced veneers. The grain of the three inner layers was orientated along the
member longitudinal axis, while the grain of the two outer layers was perpendicu-
lar to the inner layers and orientated in the member transverse direction (see insert
in Figure 3). In this configuration, the three inner layers would mainly resist the
compressive load while also providing resistance against bending of the walls in
the longitudinal direction, and the two outer layers would mainly provide re-
sistance against bending of the walls in the transverse direction, therefore enhanc-
ing the local buckling capacity of the profile.
Table 1 Nominal section properties
Type Area Imajor axis Iminor axis Warping J

(mm2) (mm4) (mm4) (mm6) (mm4)
A 2375 1.58107 3.92106 4.971010 1.98104
B 2466 1.58107 4.64106 5.031010 2.06104
2.2 Manufacturing Process

Veneers were delivered in sheets of 1.2 m 1.3 m and sheets with a minimum
number of or with no natural defects (knots, resin veins, etc...) were selected. Dif-
ferent veneer sheets were used for each layer constituting a profile. Yet, all cross-
sections in a set were manufactured from the same sheets, glued in the exact same
order, allowing comparison between the two types of cross-sections in a set.
Each cross-section was manufactured as:
Step 1: Sheets were cut to size and veneers were soaked in water for 48 hours.
Step 2: The veneers were laid flat on a bench, with the grain in the appropriate
orientation (see section 2.1), heated using a steamer and bent around a jig
to form the cross-section.
Step 3: Each flat side of the cross-section was clamped to the jig, as seen in Fig-
ure 4 for Type B cross-section.
Step 4: The jig was placed in an oven at 40C for about 12 hours, till the bends
were dried and hold their shapes.
Step 5: The veneers were unclamped and removed from the jig, further left to dry
in the oven for about 4 hours, and stored in an air-conditioned room until
their moisture content reached equilibrium.
Step 6: The veneers were glued around the jig at ambient temperature with Res-
orcinol formaldehyde structural adhesive. Similar to Step 3, each flat side
of the cross-section was clamped to the jig. Rubber sheets were inserted
between the veneers and the clamping plates to uniform the applied pres-
sure. The glue was left to set for a minimum of 48 hours before unclamp-
ing the cross-section.
Step 7: The excess wood in the lip stiffeners were cut to form the final cross-
sections.
To simplify the manufacturing process and obtain more reliable products, espe-
cially for the bends, a vacuum press is planned to be used in the future. Figure 5
shows a photo of the final products.
Fig. 4 Type B Cee-section during Fig. 5 Final Cee-sections

manufacturing Step 3
To determine the mechanical properties of each set of cross-sections, flat panels

were also manufactured using the same sheets as their associated cross-sections
glued in the exact same order and orientation. Before testing, cross-sections and
flat panels were left in an air-conditioned room until their moisture content
reached equilibrium.
3 Material Testing
3.1 MOE Parallel to Grain

The Modulus of Elasticity (MOE) parallel to the grain of the veneers used to
manufacture the cross-sections was estimated using a non-destructive resonance
method [3]. A 300 to 500 mm long 30 mm wide sample of each veneer was
simply supported on rubber bands and impacted with a hammer. The natural fre-
quency of the tested sample was recorded using a microphone and analysed using
the software BING (Beam Identification by Nondestructive Grading) [4]. Table 2
gives the measured MOE parallel to the grain of each veneer used.
Table 2 Modulus of Elasticity for each veneer used
MOE parallel to the grain (MPa)

Set Layer 1 (Inside) Layer 2 Layer 3 Layer 4 Layer 5 (Outside)
1 12969 13279 11955 12471 14648
2 20399 19317 16811 19164 15113
3 17187 16186 16725 18776 14401
3.2 MOR Parallel to the Longitudinal Axis

The flat panels of each set of cross-sections were cut in four and reglued to form
20 mm nominal thick panels. Nominal 100 mm long 46 mm wide samples were
then tested in compression parallel to the longitudinal axis of the cross-sections, in
a 500 kN capacity MTS testing machine, following the method proposed in the
Australian and New-Zealand Standard AS/NZS 4357.2 [5]. The apparent Modulus
of Rupture MORapp is calculated as,
Pmax
MOR app = (1)
bd
where Pmax is the maximum applied load, and b and d are the measured width and
depth of the sample, respectively. Table 3 gives the average apparent MOR for
each set of cross-sections.
Table 3 Average apparent Modulus of Rupture for each set of cross-section
Set Number of tests Average MORapp (MPa) CoV

1 2 29.1 0.014
2 3 36.3 0.034
3 3 34.8 0.023
4 Stub-Column Tests
4.1 Test Set-Up

The profiles were tested in compression in a 500 kN capacity MTS testing ma-
chine in which the lower platen was fixed, while the upper platen was mounted on
a half sphere bearing which could rotate so as to provide full contact between the
platen and the specimen, as shown in Figure 6. The tests were performed in dis-
placement controlled at a stroke rate of 0.6 mm/min. Load was applied through the
theoretical centroidal axis of the cross-sections.
This test arrangement, i.e. consisting of testing short columns of lengths typi-
cally less than twenty times their least radius of gyration, aims at determining the
effect of local buckling on the column performance and allows the determination
of the column strength (or section capacity for local buckling) [6].
For Type A cross-sections only, three Linear Variable Displacement Transduc-
ers (LVDT) recorded the horizontal displacements of the centreline of the web at
heights of 175 mm (LVDT 2), 250 mm (mid-height LVDT 1) and 325 mm
(LVDT 3), and one LVDT recorded the horizontal displacement of the centreline
of the north flange at mid-height (LVDT 4), as shown in Figure 6.
(a) (b)
Fig. 6 Stub-column test set-up (shown for Type A cross-section), elevations: (a) flange
view, (b) web view
4.2 Test Results

Table 4 gives the recorded section capacity Ns of each cross-section (defined as
the maximum applied load) and the associated squash load Nsquash, representing the
upper bound capacity of the cross-section, defined as,
N squash = MOR app A (2)
where MORapp is given in Table 3 and A is the nominal cross-sectional area given
in Table 1.
Table 4 Stub-column test results
Type A Type B
Set Ns (kN) Nsquash (kN) Ns/Nsquash Ns (kN) Nsquash (kN) Ns/Nsquash
1 41.3 69.1 0.60 54.7 71.8 0.76
2 47.5 86.2 0.55 63.5 89.5 0.71
3 49.8 82.7 0.60 61.3 85.8 0.71
Average 46.2 79.3 0.58 59.8 82.4 0.73
It can be seen from Table 4 that the addition of the web stiffener in Type B
cross-section resulted in an increased average section capacity of 29% for an in-
creased cross-sectional area of 3.8%. Type A and B cross-sections reached an
average of 58% and 73% of their upper bound capacity Nsquash, respectively.
Figure 7 and Figure 8 plots the recorded displacements of the four LVDTs and
shows the failure mode, respectively, for the Type A cross-section of the third set.
It can be noticed in Figure 7 and Figure 8 a large post-buckling behaviour, similar
to the one encountered in steel structures.
Fig. 7 Test results for set n3, Type A cross-section Fig. 8 Failure mode for set n3,
Type A cross-section
4.3 Comparison with Cold-Formed Steel Structures

Using typical densities of 550 kg/m3 for hoop pine [7] and 7,850 kg/m3 for steel,
the cross-sectional area required for a steel profile to equal the linear weight of the
studied cross-sections is in the range of 166 (Type A cross-section) to 173 mm
(Type B cross-section). This range of cross-sectional area would lead to a maxi-
mum section capacity Ns of 74.7 kN to 77.9 kN for 450 MPa graded steel, i.e.
about 39% and 22% greater than the average observed capacities of cross-sections
Type A and B, respectively. Yet, these steel profiles would likely have slender
wall elements leading to low local buckling capacities and therefore reduced col-
umn strength.
For instance, the C7510 cold-formed steel Cee-section (75 mm deep and 1.0
mm wall thickness) from the Australian manufacturer Lysaght [8] has a nominal
cross-sectional area of 162 mm2 and a stub-column capacity of 46.2 kN [9] for
450 MPa graded steel. This capacity is equal and 29% less than the average ob-
served capacities for cross-sections Type A and B respectively. Moreover, the
bending stiffness EI about the major axis of bending of the C7510 is equal to
2.881010 N.mm2 (E = 200,000 MPa and I = 1.44105 mm4), while the bending
stiffness of Type A cross-section is equal to 1.231011 N.mm2 (E = 13,000 MPa at
12% moisture content [7] and I = 9.45106 mm4 when only considering the three
inner layers resisting bending), i.e. 4.3 times greater than the C7510 steel profile.
Similarly, Yap and Hancock [10] performed stub-column tests on 128 mm deep
and 0.42 mm wall thickness complex high strength cold-formed steel open pro-
files. The nominal yield stress of the profiles was equal to 550 MPa, the cross-
sectional area to 158 mm2 and the average tested stub-column capacity to 45.7 kN,
i.e. 1% and 31% less than the average observed capacities for cross-sections Type
A and B respectively. The profiles tested in [10] had a bending stiffness about the
strong axis of bending equal to 4.741010 N.mm2 (I = 2.37105 mm4), i.e. about
2.6 times less than the studied timber profiles.
5 Future Studies
By advancing the structural and mechanical knowledge on thin-walled timber

structures, the final aim of this project is to eventually develop design rules, so
engineers can safely use the proposed products. Current researches involve devel-
oping a Finite Element model to reproduce the test results presented in this paper.
Moreover, the ultimate strength and post-buckling behaviour of thin-walled struc-
tures is significantly influenced by geometric imperfections [11] and understand-
ing the influence of the moisture content on these imperfections, and therefore on
the capacity of the cross-section, is important. Additional proposed future research
also includes (i) developing joining details to connect manufactured lengths of
cross-sections, (ii) testing and understanding the structural behaviour of long col-
umns and beams, (iii) extend the Finite Strip Method [12] currently used to deter-
mine the buckling curves of thin-walled isotropic and orthotropic materials to
composite materials, and (iv) extend the Direct Strength Method [13] currently
used to design cold-formed steel profiles to thin-walled timber profiles.
6 Conclusion
This paper presented an experimental investigation on thin-walled timber struc-

tures. 210 mm deep, 105 mm wide and 5 mm wall thickness composite timber
Cee-sections were fabricated by gluing together thin softwood (Hoop pine - Arau-
caria cunninghamii) veneers. Two types of cross-section, with and without inter-
mediate web stiffener, have been investigated. The manufacturing process is
detailed in the paper. Stub-column tests were performed and results show that (i)
manufacturing structurally sound thin-walled timber structures is possible, (ii) the
investigated cross-sections have significant post-buckling behaviour and (ii) are
able to compete with cold-formed steel products with similar or greater weight to
compressive capacity ratio and significantly higher bending stiffness.
References
[1] Hancock, G.J.: Design of cold-formed steel structures (to AS/NZ 4600:2007), 4th
edn. Australian Steel Institute. North Sydney, Australia (2007)
[2] Yan, H., Shen, Q., Fan, L.C., Wang, Y., Zhang, L.: Greenhouse gas emissions in
building construction: A case study of One Peking in Hong Kong. Building and En-
vironment 45, 949955 (2010)
[3] Brancheriau, L., Bailleres, H.: Natural vibration analysis of clear wooden beams: a
theoretical review. Wood Science and Technology 36, 347365 (2002)
[4] CIRAD, BING (Beam Identification by Nondestructive Grading) software,
http://ur-bois-tropicaux.cirad.fr/produits/bing/usage (ac-
cessed on April 23, 2013)
[5] AS/NZS 4357.2, Structural laminated veneer lumber, Part 2: Determination of struc-
tural properties - Test methods, Standards Australia, Sydney, Australia (2006)
[6] Galambos, T.V.: Guide to stability design criteria for metal structures. Wiley and
Sons, New York (1998)
[7] Kingston, R.S.T.: The mechanical properties of Queensland grown hoop pine,
Proj.TM.9-7, Final report, Commonwealth of Australia - Council for Scientific and
Industrial Research - Division of Forest Products, South Melbourne, Australia
(1947)
[8] BlueScope Steel Limited, LYSAGHT Zed & Cee Purlins and Girts,
http://www.lysaght.com/go/product/lysaght-zed-and-cee-
purlins-and-girts (accessed on April 23, 2013)
[9] AS/NZS 4600, Cold-formed steel structures, Standards Australia, Sydney, Australia
(2005)
[10] Yap, D.C.Y., Hancock, G.J.: Experimental Study of Complex High-Strength Cold-
Formed Cross-Shaped Steel Section. ASCE Journal of Structural Engineering 134,
13221333 (2008)
[11] Schafer, B.W., Li, Z., Moen, C.D.: Computational modeling of cold-formed steel.
Thin-Walled Structures 48, 752762 (2010)
[12] Cheung, Y.K.: Finite Strip Method in structural analysis. Pergamon Press, Inc., New
York (1976)
[13] Schafer, B.W.: Designing cold-formed steel using the direct strength method. In:
LaBoule, R.A., Yu, W.W. (eds.) Proceedings of the 18th International Specialty
Conference on Cold-Formed Steel Structures, Orlando, Florida, pp. 475490 (2006)
Recommendations for the Design of Complex
Indeterminate Timber Structures
Andrew Lawrence
Arup, MA (Cantab), CEng, MICE, MIStructE, London, United Kingdom

andrew.lawrence@arup.com
Abstract. Complex indeterminate structures with multiple loadpaths are now in-
creasingly common in timber. However, due to connection fit-up and the material
variability, one loadpath might be stiffer and attract forces which are perhaps two
or three times the level predicted by a simple elastic analysis. To achieve overall
ductile behavior is important to ensure that the connections are sufficiently ductile
to protect the brittle timber members from damage. This paper examines whether
the rules in current design codes are adequate for the design of such structures.
Keywords: indeterminate structures, grillage type structures, load paths, brittle

damage, ductile damage.
1 Introduction
Timber has traditionally been used in one-way spanning structures such as arches
and trusses. However, advances in analysis and fabrication techniques over the
past 5-10 years, have enabled the construction of ever more complex two-way
spanning indeterminate timber grillages. Such structures, including the Metz
Pompidou, the Metropol Parasol (see Figure below) and the Canary Wharf Station
Roof, in which the author has been involved, and are now very much part of the
modern architectural aesthetic.
Timber structures have traditionally been designed using a single elastic analy-
sis. However, for complex indeterminate timber structures with multiple loadpaths
this will no longer necessarily be a safe design method. For example, if there are
several close supports, then due to connection fit-up and material variability the
load in some members near the supports could be significantly higher than pre-
dicted by the simple analysis. As the loads attempt to redistribute, damage to the
brittle timber elements could result.
Elastic analysis is typically used for ductile materials such as steel and (under-
reinforced) concrete, which can accommodate a degree of load redistribution.
However, an elastic analysis will only be applicable for a brittle material such as
timber, if the overall structure has been designed to behave in a ductile manner.
DOI: 10.1007/978-94-007-7811-5_12, RILEM 2014
130 A. Lawrence
This paper examines whether the rules in current timber design codes are ade-
quate to ensure overall ductile behaviour. There are two obvious alternatives to
such an approach:
a) Making the structure determinate this would reduce the overall robust-
ness as it would remove the possibility of alternative loadpaths in the
case of an accidental design situation;
b) Using a sensitivity analysis to consider the effect of varying connection
stiffness except for very simple buildings, this will generally lead to
structures which are expensive to design and construct, because of the
need to design for multiple loadpaths.
While these approaches will sometimes be useful, they will not be discussed
further in the current paper. Ensuring ductile behaviour is believed to offer several
advantages:
- it should lead to a more efficient design (because it avoids the need to
design for multiple alternative loadpaths);
- it will be more robust in the event of accidental damage;
- the large deflections which would occur before failure would provide a
visible warning of any potential problems.
2 Ensuring Ductile Behaviour of the Overall Structure
Timber is an inherently brittle material. This is especially true in shear and in ten-
sion perpendicular to grain (because of the weak bond between the fibres) and also
Recommendations for the Design of Complex Indeterminate Timber Structures 131
in tension parallel to grain (because of the effect of natural defects such as knots).
Therefore the easiest way to ensure overall ductile behaviour of the structure is:
- to design the connections to fail before the (brittle) timber members,
thereby limiting the global load which the members can attract;
- to ensure the connections fail in a ductile manner;
- to ensure that the connections have adequate ductility to allow the loads
within the structure to redistribute, thereby compensating for uncertainty
over the initial loadpath.
This is similar to the Capacity Based Design referred to in seismic codes.
Each of these three requirements will now be discussed in turn.
3 Ensuring Ductile Behaviour of the Connections
There are three common types of timber connection:

- bearing connections (including traditional joints);
- dowel type fasteners;
- bonded rods.
To avoid damage to the connected member, it is important to ensure that the
connection fails in a ductile manner under the various applied loads (axial, bend-
ing, shear and, if applicable, torsion). It is assumed that the member is able to
resist any directly applied external loads; the issue is the extent to which it can
support the overall structure beyond and therefore it is necessary to limit the load
which can be applied to the member by the connection. Each of the three types of
connection will now be examined in detail.
3.1 Bearing Connections

These rely mainly on compression parallel and perpendicular to grain. Such fail-
ures are generally ductile. However, notched ends are susceptible to a brittle split-
ting failure in tension perpendicular to grain. This is recognised in design codes
which limit the shear on notched ends to prevent splitting. However, in complex
structures, where the load on those notched ends might be two to three times the
predicted value, there is still a risk of splitting. The risk of splitting therefore needs
to be prevented, which is easily done by introducing cross-grain reinforcement,
such as screws or cross-veneers.
3.2 Dowel Type Metal Fasteners in Shear

These include nails, screws, dowels and bolts. Formulae for the potential failure
modes were originally developed by Johansen [4] and are now included in modern
132 A. Lawrence
design codes. The formulae rely on either crushing of the wood, yielding of the
fastener or a combination of both (Eurocode 5, Figure 8.2). Several measures can
be taken to encourage ductile behaviour:
- Johansen failure modes which include yielding of the fastener will obvi-
ously be more ductile. The fasteners should therefore be so proportioned
(relative to the thickness of the wooden parts) to ensure that these modes
govern;
- Johansen assumes that the full crushing (i.e. embedment) strength of the
timber can be developed and that premature brittle failure does not occur
due to splitting ahead of the fastener, tensile failure of the residual timber
section or plug shear failure. To avoid splitting, codes impose minimum
spacings, end and edge distances. Codes also require additional checks
to be made to ensure that brittle tensile, block or plug shear failure does
not occur;
- Since the rules to prevent splitting are empirical and could be compro-
mised by, for example splits due to drying shrinkage, it is recommended
that screws are incorporated as cross grain reinforcement. For a small ex-
tra cost these will give a more reliable behaviour and also allow more
ductility and distortion before failure. Guidance on their design will
hopefully be included in future revisions of Eurocode 5;
- Where several fasteners in-line are loaded parallel to grain, testing shows
that the load distribution between the fasteners is uneven (partly due to
fit-up) and that a strength reduction factor needs to be applied (the neff
factor in Eurocode 5) to prevent premature splitting which would then
prevent the redistribution of forces within the connection. Testing [9, 10]
shows that cross-grain reinforcement screws can again be used to ensure
more reliable behaviour and to enable the full crushing strength of the
timber to be mobilized;
- Testing [6] also shows that multiple bolted connections can have less
ductility, because the oversized holes may mean that not all the bolts are
in contact with the timber, again leading to an uneven load distribution. It
is therefore recommended that dowelled connections which have a more
reliable behavior are used in preference to bolts;
- Smaller fasteners will obviously be more ductile than large fasteners and
should therefore be used in preference, especially if a large amount of
ductility and rotation is required to redistribute the forces within the
structure. It is interesting to note that Eurocode 8 limits fasteners to only
12mm diameter. However, as well as achieving ductility this will be to
ensure good energy dissipation and therefore is considered to be an ex-
cessive limit for non-seismic areas.
Recommendations for the Design of Complex Indeterminate Timber Structures 133
3.3 Bonded Rods

Limited research to date [11] suggests that to ensure ductility:
- Mild steel rods should be used, rather than high strength steel;
- The rods should have a sufficient unbonded length that is necked down to
allow for elongation without damaging the glue bond. The proportions
can be varied to control the degree of ductility required;
- Cross-grain reinforcement screws should be used to prevent splitting of
multiple bonded rods.
4 Ensuring That the (Brittle) Timber Members Are Stronger

Than the (Ductile) Connections
As noted, to ensure that the overall structure behaves in a ductile manner, it is not
sufficient just to make the connections ductile; it must also be ensured that the
(brittle) timber members are stronger than the (ductile) connections.
Fortunately, connections are generally the weak points of timber structures and
therefore this criterion will usually be satisfied automatically. However, checks
should always be made to ensure that this is the case. In making this check, it
should be remembered that the connection strengths derived from design codes are
conservatively low (to account for the variability of the timber and the approxima-
tions of the Johansen formulae [4]) and therefore the timber members must exceed
the code-derived connection strength multiplied by a suitable overstrength
factor.
Limited testing to date [7] on dowel type fasteners suggests that this factor is
about 1.6. If there is any uncertainty about the members being stronger than the
connections, then the connections may need to be tested; reference [7] that sug-
gests testing on at least 10 similar connections to determine the overstrength factor
(based on the 95th percentile of connection strength).
5 Ensuring That the Connections Have Adequate Ductility

In some cases it will be necessary to undertake a non-linear analysis with elastic-
perfectly plastic springs to confirm that the connections have adequate ductility
(for example to ensure that they can accommodate sufficient rotation) to allow
redistribution of the forces within the structure. More research and testing is
needed to help predict how much deformation a connection can accommodate
before failure.
6 Summary
For complex indeterminate timber grillage-type structures with multiple loadpaths
it is recommended that the following conditions are met:
134 A. Lawrence
(1) The connections should be ductile. For connections with dowel type fasteners,
this will be the case if the fasteners are so proportioned to ensure that one of the
Johansen [4] failure modes governs which incorporate yielding of the fasteners.
Dowels should also be used in preference to bolts. To improve the level of ductili-
ty smaller fasteners and reinforcement perpendicular to grain can be used; the
latter will ensure more reliable and predictable behaviour. Checks should also be
made to ensure that the dowels yield (with an adequate factor of safety) before the
onset of brittle plug shear or tensile failure.
(2) Members should be stronger than the connections.
(3) In some cases it will be necessary to undertake a non-linear analysis, incorpo-
rating elastic-perfectly plastic springs to represent the ductility of the connections,
to demonstrate that the connections have adequate ductility, possibly backed up by
testing to help understand the degree of deformation which the connections can
accommodate.
Acknowledgements. The author is grateful to colleagues at Arup for their assistance and
also to the detailed research which has been undertaken at Eindhoven, Karlsruhe, Stuttgart,
UBC and Zurich which has been referred to in this paper. There is still much work to be
undertaken in the field and it is hoped that this paper may help guide the areas for future
research.
References
[1] EN 1995-1-1:2008, Eurocode 5 Design of timber structures part 1-1: General
common rules and rules for buildings
[2] EN 1998-1:2004, Eurocode 8 - Design of structures for earthquake resistance part
1: General rules, seismic actions and rules for buildings
[3] SIA 265. Swiss code for timber structures
[4] Johansen, K.W.: Theory of Timber Connections. IABSE 9, 249262 (1949)
[5] Mischler: Dowelled timber connections with high efficiency. In: Proceedings of the
First International RILEM Symposium on Timber Engineering (1999)
[6] Leijten, A.J.M.: Requirements for moment connections in statically indeterminate
timber structures. Engineering Structures 33, 30273032 (2011)
[7] Jorissen, A., Fragiacomo, M.: General notes on ductility in timber structures. Engi-
neering Structures 33, 29872997 (2011)
[8] Bruhl, F., Kuhlmann, U., Jorissen, A.: Consideration of plasticity within the design
of timber structures due to connection ductility. Engineering Structures 33, 3007
3017 (2011)
[9] Blass, H.J., Schadle, P.: Ductility aspects of reinforced and non-reinforced timber
joints. Engineering Structures 33, 30183026 (2011)
[10] Lam, F., Wrede, M.S., Yao, C.C., Gu, J.J.: Moment resistance of bolted timber con-
nections with perpendicular to grain reinforcements. In: Proceedings 10th WCTE
Miyazaki, Japan
[11] Tlustochowicz, G., Fragiacomo, M., Johnsson, H.: Provisions for ductile behaviour
of timber-steel connections with multiple glued-in rods. ASCE Journal of Structural
Engineering (September 13, 2012)
Novel Lightweight Timber Composite Element:
Web Design in Shear and Compression
Simon Aicher and C. Stritzke
Materials Testing Institute, University of Stuttgart, Department of Timber Constructions,

Pfaffenwaldring 4b, 70569 Stuttgart, Germany
Simon.Aicher@mpa.uni-stuttgart.de
Abstract. The paper examines the build-up and some aspects of the mechanical
behavior of a novel lightweight timber composite called the Keel-web element.
The element is a double-skinned composite similar to a multiple box beam ele-
ment. The flanges consist of finger jointed lumber chords arranged and glued in
parallel. The name-giving specific characteristic of the element consists of the
multiple S-shaped webs made of plywood or OSB, resembling ship keels, glued in
between the flanges. The element, which can be produced in a fully automatized
process, as straight or cambered, with lengths of up to 35 m, has recently obtained
a German technical building approval. The build-up will first be examined, and
afterwards, the engineering design approach for the shear force and compression
capacity, partly following Eurocode 5, is shown.
In order to analyse in greater detail the stability and nonlinear bending of the
pre-curved webs at end supports, the load deflection behavior of the webs at an
increasing support force is studied by 2nd and 3rd order beam column and plate
theory. Additionally, the bending stresses introduced from the manufacturing pro-
cess of the S-shaped webs which are subject to relaxation, must also be consid-
ered. With increasing loads, the out of plane web displacements are restrained by
the adjacent webs, leading to a reduction of the free web height. Solving the dif-
ferential equation for a fixed-end beam, valid results for small deformations can
be calculated, while a nonlinear finite element simulation is performed in order to
consider large deformations and contact boundaries. It is shown that the stress-
state of a 3D shell model with contact simulation of the webs provides a good
estimate of the experimental load capacities.
Keywords: Lightweight timber element, Keel-web element, glued composite

structure, S-shaped webs, Buckling, Nonlinear bending, Support design, Shear
capacity, Plywood, OSB.
1 Introduction
Wood as a construction material can be considered as lightweight due to its rela-

tively high strength-to-weight ratio. In comparison to other common building
DOI: 10.1007/978-94-007-7811-5_13, RILEM 2014
136 S. Aicher and C. Stritzke
materials, the tearing length of a softwood member loaded in tension parallel to

the fiber direction is large. Depending on the strength class, defect state and cross
sectional size, the tearing length is between 1.5 to 3 times larger as compared to
usual construction steel. Additionally, the compression strength of wood is within
the same range as that of concrete. However, as wood is increasingly being used
due to its undisputed ecological benefits (for example CO2-storage) as well as its
sustainability, it is approaching the limits of sustainable forestry. The result is a
steady price increase for wood products, which is expected to rise significantly in
the coming years. The higher costs already have had an impact on the competi-
tiveness of massive wooden building products such as cross-laminated timber
(CLT). Hence, a growing future demand consists in the development of material
resource preserving lightweight structures.
Recently, the palette of wooden lightweight structural elements was expanded
with a new innovative glued element type, known as the Keel-web element [1].
Based on extensive trials, testing and analysis, the construction principle has re-
cently obtained a German Technical building approval [2]. The elements are al-
ready being produced on an industrial scale and are increasingly being used in
building structures.
The elements have a very high load capacity and stiffness and are ideally suited
for long-span roof structures as well as ceilings. To get high stiffnesses in all spa-
tial directions, the webs are bent in the manufacturing process to the characteristic
S-shape. This particularity leads to interesting structural mechanical behaviour,
especially of the web structure. One of the major interests concerns the buckling
behaviour and nonlinear bending in the end support areas, which is the focus of
this paper.
2 Build-Up of Keel-Web Elements
The Keel-web element is a double-skinned wooden composite element with thick

skins made of finger jointed softwood lumber chords glued in parallel at the nar-
row edges. The lumber cross-sections are within the size range of (thickness x
width) 40 mm x 70 mm to 90 mm x 175 mm, depending upon the specific build-
up of the element, with a height between 228 mm and 800 mm (see Table 1). Be-
tween two adjacent lumber cords, which shall have a minimum strength class of
C24, the ends of the webs are glued-in (see Figs. 1 and 2). The webs consist of
characteristically S-shaped bent thin wooden based panels, being either three-
layered plywood acc. to EN 636 [3] (suitable for damp areas) with a thickness of
4.8 mm or OSB/3 material acc. to EN 300 [4] with panel thicknesses of 8, 10 or
12 mm. Webs made of plywood are confined to element depths of 228 mm to
380 mm, whereas OSB/3 webs are used for elements with depths of 485 mm to
800 mm.
Novel Lightweight Timber Composite Element 137
Fig. 1 View of Keel-web elements with webs made of plywood (Height: 380 mm)
The bending strength class of the plywood material must be at least class
F60/10 and at most class F70/15. The flexural modulus of elasticity (MOE) class
must be at least class E100/5 and at most class E40/20. Both panel materials shall
be CE-marked on the basis of EN 13986 [5]. The web and flange dimensions can
be designed according to the applied loading, within the size and geometry re-
strictions specified in the German technical Approval Z.9.1-831 [2].
In the layout of the element, special consideration shall be given to the bending
stresses induced in the web during production to achieve the typical S-shape, ob-
tained by a relative horizontal offset of the tension and compression flange chords
by half of the width, /2, of the lumber cross-sections. The maximum web stress
to be considered is then (where: , = width and depth of flange (see Fig. 2)
and , , = MOE in bending of panel material perpendicular to span):
, , =3 , , /(2 ) . (1)
Taking into account a high degree of stress relaxation, the bending stress in the
webs at the connection to the flanges is limited in [2] to
, , , < 0.8 , , for plywood webs and (2a)

, , , / < 1.0 , , for OSB/3 webs. (2b)
In the manufacturing process performed by Kulmer Bau GesmbH & CoKG, Aus-
tria, the webs are bent into the S-shape and bonded with the lumber chords in a
single operation, thus forming elements with standard widths of 1.2 m. At the
present time, the maximum element length is up to 35 m and can optionally be
produced pre-cambered.
Table 1 gives an overview of typical product ranges allowable given the
provisions in [2].
Table 1 Cross-sectional size and geometry ranges of Keel-web elements according to [2]
web-material Element height [mm] Dimensions of flanges [mm]

(thickness )
228 70-90 40-50
Plywood
280 70-112 40-57
(4.8 mm)
380 70-112 40-67
485 70-120 40-67
560 70-140 57-80
OSB/3 615 70-155 57-80
(8, 10, or 12 mm) 730 70-140 67-80
785 90-165 80-90
800 90-175 80-90
3 Simplified Design Rules
3.1 General
For a sufficiently accurate simplified engineering design in the case of static load-
ings, a splitting of the cross-sectional structure into equal I- or half box beams (see
Fig. 2) can be performed [2]. Similar to I- or box beams, the following stresses
must be checked [2; 3]:
- / , / , bending stresses on the top and bottom of the cross sectional
area
- , / compression / tensile stresses at the centroid of the cross sectional ar-
ea of flanges
- , shear stresses in the glue lines between flanges; and webs, and in
web-web glue lines, respectively
- , shear stresses at the centroid of the cross sectional area of the webs,
considering buckling (see below)
3.2 Shear Force Design

Due to the comparatively slender webs, it is of upmost importance to ensure that
buckling of the web is excluded. The design verification implemented in [2] con-
siders the web stability in an implicit manner through use of a material-specific
rule for the geometry dependent effective shear strength , , for the web. The
design procedure follows, apart from material related aspects, the shear force ca-
pacity design procedure specified in Eurocode 5 [6] for glued I- and box beams
with straight thin webs. The latter is briefly outlined below to demonstrate the
analogy of standard design of the shear capacity design to that of the Keel-web
Fig. 2 Simplification of the cross-sectional structure of Keel-web elements[1] for design in

[2] a) unit cell of real structure; b) I-beam approximation; c) half box beam (for shear
capacity design)
elements with the S-shaped webs. Eurocode 5 specifies that unless a detailed
buckling analysis is performed, and provided that the web geometry suffices
/ 70 (see Fig. 2c), the subsequent design procedure should be followed:
, , , , (3)
where
design value of acting shear

, ,
force of each web
design shear force re-
, , = , , , (4)
sistance of each web
0.5 , + ,
, = 1+ effective web cross-section (5)

or in case of equal flange thicknesses = , /
, = + = , effective web cross-section (6)
and
, , = , , / 35 (7)
35
, , = , , = , , 35 / 70 . (8)
/
The design strength values and the characteristic values are related as usual via
, , , = , , , / . (9)
where is the strength modification factor for influence of load duration and
moisture content; and is the partial strength (safety) factor. The above equa-
tions are slightly modified from those given in [6] by the introduction of , ,
and in order to enable the calculation of the S-shape in a manner similar to that
for a standard I-shape. The equations show that EC5 does not consider any buck-
ling phenomenon up to a web aspect ratio of / = 35 whereby the design (d)
or characteristic (k) shear force capacity is fully determined by the effective cross-
sectional web area and the characteristic in-plane shear strength value , , of the
web panel material. For larger aspect ratios, the buckling effects caused by shear,
in-plane compression close to the support and from in-plane bending are consid-
ered together as an effective shear strength value , , whereby the reduction
factor depends exclusively on the aspect ratio / .
In the derivation of the design equations for the technical approval for the Keel-
web element, for sake of simplicity the same approach as specified in EC5 for
glued I- and box beams for computing shear force capacity per web on the basis of
an effective shear strength value , , was pursued. Based on a best fit to the
5 %-quantile values of the shear force capacities of full scale element tests with
several different geometry and size configurations (see Fig. 4), the material specif-
ic effective shear strength expressions analogous to Eqs. ( 7) and ( 8 ) were
derived and are given as follows (graphs of , , acc. to Eqs. ( 10 ) to ( 12 ) are
shown in Fig. 4):
, , , = , , , , , , / , = , , / , (10a,b)
In case of plywood ( = 4.8 mm)
=1 / < 30 (11a)
= 0.1124 + 772( / ) 30 / 61 (11b)
and for OSB/3 ( = 8, 10 and 12 mm)
=1 / < 46 (12a)
= 0.0133 + 2144( / ) 46 / 64. (12b)
The characteristic nominal in-plane shear strength values for the plywood and the
OSB/3 material were derived in the tests for [2] as:
, , , = 7.5N/mm and , , / , = 4N/mm . (13)

Fig. 3 Test setup for shear capacity tests of Keel-web elements
8 f v,0,plywood,k
effective shear strength
approval [2]
7
EC5 [3]
fv,eff,k [ N/mm ]
5
f v,0,OSB/3,k
4
3
approval [2]
2 EC5 [3]
0 10 30 40 50 60 70
web geometry ratio (hw / bw )
Fig. 4 Effective shear strength dependent on web geometry ratio / and material type
4 Buckling and Bending Due to Support Normal Forces
4.1 General and Eigenvalue Results

It has been previously revealed in [7] that the vertical reaction force at the end
support locations leads to a complex interaction of normal force, bending moment
and shear force loading of the S-shaped web. Super-imposed to the section forces
from the vertical reaction force is the bending moment ( = 1 /12)
[ ] = 6 , , ( 2 )/ (14)
introduced during the production process due to the horizontal web displacement
= /2. The normal force per unit length of the web in span direction
[ ]= , [ ] imposed at the bottom side of the web by the support force
142 S. Aicher and C. Stritzkke
(Fig. 5) decreases in an approximately

a linear manner towards the upper unloadeed
edge of the element simillar to a classic I-beam, see Fig. 6 and [7]. The comprees-
sion stress distribution , ,90 thus depends considerably on the length of the supp-
port and the excess length
h. The pure compression strength verification of the weeb
can be performed in the case of sufficient excess length /4 analogous tto
EC 5 [6] as
, [0] ,
, , = = = (15)

where n = number of webs; V= total shear force per element and
= +2 . (166)
Flanges Web
11000 5500
370 370
[N/mm]
370 1000
690 700
690 700
69 200
0.35 0.0422
[-]
0.35 0.0155
0.015 0.35
Fig. 5 Geometry notations att end support and material parameters used for stress analysis
a) b)
Fig. 6 Compressions stress distribution in the web of I-beams with straight web at ennd
supports (see Fig. 5) at V = 1 kN (dimensions as for KSE380, see below), calculated wiith
FEM a) with excess length; b)
b without excess length
As opposed to a straight web, an additional bending moment [ ] results

from the linearly diminishing normal force [ ] along the S-shaped web (see Fig.
7b). The maximum value at the support edge of the varying moment distribution
along depth of the web has been determined in [7] numerically as
[0] = (1.5 10 ) (17)
where
= ( /(2 )) (18)
is the averaged linearized inclination of the web.

For a straight, two-sided clamped beam-type column subjected to a constantly
distributed tangential compression line load q, an analytical expression for the
critical buckling load
= = 8.2445 , , / (19)
has been derived in [8]. For a two-sided clamped S-shaped beam subjected to a
uniformly distributed tangential axial line load, no analytical solution exists. Finite
element computations performed with S-shaped beams revealed as for the case of
a linearly variable tangential compression load [7] a rather small decrease of the
critical load (1st eigenvalue) as compared to the straight (reference) beam on the
order of approximately 2 6 %. This decrease is within practically relevant rota-
tion angles in the range of 10 to maximally 20. Fig. 8 shows the critical beam
buckling load for an S-shaped web with dimensions tested at element type
KSE380 ( = 294 mm, = 4.8 mm, = 108 mm and , = 1000 MPa);
further i.a. the result according to Eq. (21) is given.
Fig. 7 Section forces of S-shaped web a) [ ] from production; b) [ ] and [ ] from

support force N = V
4.2 Higher Order Theory Bending

It is obvious that for the pre-curved web, the eigenvalues of a classical buckling
problem are rather arbitrary numbers. Nevertheless, they may however serve as a
verification of the results of 2nd order theory computations which must converge to
the eigenvalue result [9]. It can be easily seen that the pre-curved web starts to bend
stably under the action of an applied increasing normal force [ ], leading to higher
order deflection and hence to a stress/ strength problem for ultimate capacity.
Some features will be highlighted considering first a beam situation followed
by plate solutions. The governing differential equation is
[ ] + (1 / )( [ ]+ [ ]) = 0 (20)
where
= , , ; = ; [ ]= (3 2 ) / and = /2 (21)
represents the bending curve or shape function of the initial S-shape of the web at
zero normal force. [ ] is the additional lateral deflection resulting from the
linearly distributed normal force, which when integrated along the total web
height , delivers the vertical reaction force , [0].
Employing the boundary conditions [0, ] = 0 and [0, ] = 0 one ob-
tains the relationship of load vs. mid-web deflection according to 2nd order theory
as shown in Fig. 8 for a specific element configuration (Note: The results pre-
sented in Fig. 8 refer to the geometry and stiffness configurations of element type
KSE380 where = 380 , , / = 43 , = 108 , = 294 m
and = 4.8 ). It is evident from the graph that the large deflections of the
2nd order theory solution render the results to be simply of theoretical interest and
are practically irrelevant.
In a further step, the problem of large deflections can be handled as a 3rd order
theory problem resulting in a coupled partial differential equation system where
deflection and longitudinal displacements are coupled. The beam solution ob-
tained from finite element analysis, is shown in Fig. 8, revealing the typical 3rd
order theory effect of stiffening of the structure at large deflections. It can be seen
that the load deflection curve now reaches the characteristic test load and the cal-
culated shear capacity according to Eqs. (10a) and (12). Nevertheless, the pure
3rd order theory approach is again irrelevant for determining realistic failure loads
(after conversion to stresses) as the horizontal mid-depth displacement of the web
is by far too large as compared to any realistic situation or test result. (Note: The
almost exact agreement between characteristic test load and calculated shear ca-
pacity shown in Fig. 8 results from the above explained derivation of the effective
shear strength).
In a further attempt to obtain the true load capacity, the authors have considered
the experimentally witnessed and mechanically plausible fact that two adjacent
webs move closer together in the horizontal direction at increasing vertical loads
until making contact along a fictive vertical line/plane. This issue is schematically
illustrated in Fig. 7b. The empirically proven phenomenon was implemented in
the FE-modelling by a contact boundary which prevents any horizontal displace-
ment beyond the vertical line at y = 0, i.e. at the symmetry line of the keel formed
by adjacent webs.
5,0
load [kN]
4,5
char. test load Vu,k
4,0
calculated shear capacity
3,5 3rd order theory
contact boundary
3,0 beam modell 3rd order theory
Pcrit,straight beam modell
2,5
Pcrit,curved
2,0
3rd order theory 2nd order theory
1,5 contact boundary beam modell
3D modell
1,0
1st order theory
0,5 beam modell element type:
KSE380
0,0
0 10 20 30 40 50 60 70 80 90
displacement vN[hw/2] [mm]
Fig. 8 Curves of vertical load vs. horizontal displacement at mid-height of web acc. to
beam and 3D shell analysis for different boundary conditions
It is apparent that the contact between the webs from moving closer together
reduces the free height of the S-shaped web. Fig. 8 reveals for a beam solution and
for a full 3D shell model that the vertical boundary contact leads to a significant
reduction of the horizontal displacements and hence to a stiffening and stress re-
duction of the webs at increasing loads. Fig. 9 shows the changes in the moment
distribution along the web height for increasing loads, hereby revealing the differ-
ences which result from the discussed individual theoretical approaches.
In order to derive the load capacities associated with the nonlinear web
bending, the interacting stresses from normal force and bending moment have to
be considered. Load capacities were computed on the basis of a simple linear
interaction equation
, ( ) , ( ) , ( )
(1 ) + + 1 (22)
, , , , , ,
which accounts for the following three contributions: i) the outerfiber bending
stress , [ ] resulting from the initial production process, ii) , [ ] being the
outerfiber bending stress resulting from the bending moment [ ] induced by
support force [ ] and iii) the axial compressive stress , [ ]. In addition, the
stresses , [ ] from the initial bending process are not considered to the full
elastic initial degree as a strong relaxation of the stresses can be assumed, which
has been verified by ongoing experimental investigations. The reduction of the
initial bending stress , [ ] by a relaxation factor in Eq. (22) can be
assumed realistically in the range of about 20 40 % . Fig. 10 shows the evalua-
tion of the stress interaction equation for increasing vertical load further depend-
ing parametrically on the relaxation factor . It can be seen that the strength
normalized stresses deliver a good prediction of the load capacity, especially when
the relaxation factor is assumed in the range of about 0.4.
60000
30000 moment [Nmm]

0
-30000 3rd order theory considering contact boundary conditions

vertical load = 3000 N
-60000 30000
15000
-15000
15000 -30000
7500
2nd order theory
0
3rd order theory
-7500
-15000 15000
moment [Nmm]
element typ: KSE380 7500
-7500
vertical load = 0 N
-15000
0 hw/2 hw
Fig. 9 Moment distribution along web height while support force increases
0 19 horizontal displacement f/hw [%] const. 19

element type:
result of interaction Eq. (24) [-]
KSE380
1.0
0.75
char. test load Vu,k

krelax
0.5 0.0
0.4
0.25
1.0
0
0
-1000 0 1000 2000 3000 4000
zero vertical load [N]
Fig. 10 Stress interaction result vs. vertical load
5 Conclusions
Keel-web elements are innovative, strong and stiff lightweight composites made
of edge-glued solid timber members and thin S-shaped webs of either plywood or
OSB. Contrary to an easy verification of the bending stresses, the design of shear
and compression force capacity at the support as determined by the curved webs
poses a demanding mechanical problem. At present, the shear force verification is
based on an effective shear strength, derived from extensive element tests. It is
further assumed that the compression force design at the supports can be per-
formed as specified in Eurocode 5 for glued I- and box beams where buckling is
not considered explicitly.
The paper revealed that the present design of the new element type smears
several effects of interacting section forces which contribute to the nonlinear
bending of the webs. It is shown that all influences could well be separated and
that it is possible to determine the load capacity by linear interaction of the rele-
vant stresses. Several major issues remain to be solved. It is important to quantify
the relaxation behavior of the high bending stresses induced in the manufacturing
process of the S-shaped webs. Further, simplified equations have to be derived
from the 3D Finite Element solutions to enable a transparent calculation of the
different stress components dependent on curvature and edge aspect ratio of the
web material.
References
[1] Austrian patent: AT 501521: Trgerartiges/aus Einzelteilen zusammengesetztes
Bauelement sowie Verfahren zur Herstellung des Bauelements. Holder of patent:
Kielsteg GmbH, Graz, Austria
[2] DIBt: German Technical Approval Z-9.1-831Kielstegelement, Holder of approval:

Kielsteg GmbH, Graz, Austria (2013)
[3] EN 636: Plywood - Specifications (2012)
[4] EN 300: Oriented Strand Boards (OSB) Definitions, classification and specifications
(2006)
[5] EN 13986: Wood-based panels for use in construction. Characteristics, evaluation of
conformity and marking (2006)
[6] DIN EN 1995-1-1, Eurocode 5: Design of timber structures Part 1-1: General
Common rules and rules for buildings; German version EN 1995-1-1:2004 + AC:2006
+ A1:2008 (2010)
[7] Stritzke, C., Aicher, S.: Traglastverhalten von innovativen geklebten Holzleichtbau
Elementen. In: Kuhlmann, U., Brhl, F. (eds.) Holzbau Forschung + Praxis, pp. 115
124. University of Stuttgart (2012)
[8] Hauger, W.: Die Knicklasten elastischer Stbe unter gleichmig verteilten und linear
vernderlichen, tangentialen Druckkrften. Ingenieurarchiv 35, 221229 (1966)
[9] Petersen, C.: Stahlbau, 4th edn. Springer (2013)
New Timber Bridges: Inventive Design by
Block-Gluing
Frank Miebach and Dominik Niewerth
Ingenieurbro Miebach, D-Lohmar, Germany

info@ib-miebach.de
Summary. The technology of gluing in wood construction has evolved

considerably in recent years. This has been shown especially in timber bridges that
Glulam is now the main building material. A further development is the so-called
block gluing, which provides a good basis for supporting structures. Numerous
bridges, especially in central Europe, appeal by unique design and monolithic and
solid construction. These structures base on two main developments: block
lamination of glulam and the composite of timber and concrete to one structural
system.
Keywords: Block-Gluing, Glulam, Timber-Concrete-Composite.
1 Introduction
1.1 History in Europe after Industrialization

The history of bridge construction has always been connected to the technical
state of the art. It started with applications of natural material as stones and wood.
During the industrialization construction materials changed to steel and afterwards
to concrete and high developed wood constructions got displaced by concrete or
steel structures. Due to a mind change to ecological matters there is a rediscovery
of wooden structures in the last decade. And the characteristics of wood light but
strong, easy to process push this natural material into focus of architectural
interests.
1.2 Retarding in Evolution

Timber constructions have their own industrialization, but with some retardation.
In 1906 a German Otto Hetzer patented the gluing of wooden beams. The modern
glue-lamination was born. Glulam is a structural timber product consisting of
number of layers glued together to a certain cross section. Additionally it is
possible to block laminate a number of glulam beams together to achieve bigger
cross sections.
DOI: 10.1007/978-94-007-7811-5_14, RILEM 2014
150 F. Miebach and D. Niewerth
This invention provides the possibility to produce wooden beams with huge
dimension in a constant quality: length of 50m and more, width and heights of 3m
and more. The gluing process even offers the possibility to produce curved
elements or even twisted elements. This allows thinking and constructing in free
forms.
Since 40 50 years glulam is used in halls not just for the roof but also for
the posts and reaches a respectable market share by now. Wooden bridges are on
the rise since almost 15 20 years.
There is a diversity of reasons for using of timber in bridges. For example,
prefabricated large-sized parts are easy to handle and assemble because of their
lower weight - a well-known advantage for economical constructions.
The durability of the construction often is discussed seriously because latest
planning often does not consider historic knowledge as constructive wooden
protection.
The development of gluing technology including block lamination is a type of
the retarded timber industrialization.
2 Block Laminated Structures
Block lamination generates massive timber cross section to achieve loadbearing

structures for large spans with a thin side view. Block lamination comprises three
main steps: limitless gluing of lamellas through finger jointing, gluing layers of
lamellas to a glulam beam and finally gluing numbers of beams to a massive
block.
A certain form like a curve is created in step 2 lamination of layers. The

curve is always vertical to the longitudinal axis because the layers only are
flexible according to the thickness. Therefore smaller radiuses require thinner
lamellas.
According to the required form a block laminated beam may have a curvature
in top view or in side view. The difference in production is
a) Horizontal block lamination (curvature in side view)
New Timber Bridges: Inventive Design by Block-Gluing 151
or b) vertical block lamination (curvature in top view).
Even twisted forms are possible. Twisted beams are curved around two axes.
The first curvature is generated in step 2 lamination of layers as usual.
Afterward a horizontal band saw divides the curved beams into thin layers. These
thin layers are sufficient flexible to be curved again in a second direction during
block lamination.
A part from curved gluing, through block lamination stepped beams are feasible
both in cross section and in longitudinal section. There for glulam beams with
different height or length are block-laminated to one block beam.
3 Timber-Concrete-Composite Bridges
The increasing role of ecology and sustainability of building materials more and
more influences construction industry and leads to a rethink on site of the builders.
The symbiosis between wood and concrete has the best conditions to meet all
resulting claims. Constructions of massive wood beams statically connected to a
concrete slab on the upper side ensuring optimum utilization of the material
specifications of both materials. In this case the wooden cross-section is
considered to take tensile forces and the concrete slab takes the pressure forces.
Special connectors ensure the interaction of timber beams and concrete slab to get
a more effective loading capacity and serviceability.
Advantages in comparison to conventional timber bridges:
- higher load capacity with lower height of construction
- good structural wood protection through cantilevered concrete slab on the top
side
- optimal load spreading of point loads by the concrete slab
- better cross bracing
- use of proven details in connections to the concrete
New Timber Bridges: Inventive Design by Block-Gluing 153
Advantages in comparison to conventional concrete bridges:

- lower weight of the superstructure and thus more efficient structure
- fast and efficient installation with high degree of prefabrication without
extensive formwork
- cost savings in foundation and the abutment
- improved energy balance and eco-balance, sustainability through CO reduction
3.1 HBV-Shear Connector

The HBV-shear connector by TiComTec (Haibach) is an expanded metal part that
is glued perpendicularly into the wooden structure. The dimension and number of
connectors is determined to the static needs. The concrete slab usually has a
thickness of at least 20 cm and has several functions: road decking, carrier plate
for the dispersal of transverse loads and constructional wood protection.
3.2 Head Bolts

As known from steel-concrete bond structure in this alternative the connecting
parts between wood and concrete are steel bars with welded head bolts. The bars
fit exactly into milled kerfs and are fixed with screws. To achieve an efficient
utilization, the axial distances of the dowel bars correspond to the traffic load and
transverse force caused by the traffic load.
3.3 Kerfs and Glued in Reinforcement Bars

The shear connection in this type of timber concrete composite is implemented by
a combination of kerfs and glued-in steel bars. The kerfs are milled in transverse
direction and transfer shear forces (from longitudinal direction) into compressive
forces to the vertical sides of the kerfs. To avoid a lifting of the concrete additional
fasteners or connectors are considered (e.g. glued in reinforcement bars).
4 Conclusions
A changed awareness on ecological matters is the proper basis for timber bridges.
Timber is the one and only material that saves and stores CO2 permanently. And
the technical possibilities are still growing. For example, researches on modern
glues and gluing methods show in near future glues will be more efficient
(temperature-resistant and the level of bonding pressure during the gluing process
will be halved from 0,4 N/mm2 to 0,2 N/mm2). Against this background the
combination of block lamination and timber-concrete-composite is the beginning
of the latest development: The comeback of timber bridges as adequate road
bridges.
References
Simon, A.: Analyse zum Trag- und Verformungsverhalten von Straenbrcken in Holz-
Beton-Verbundbauweise. PhD. thesis, Bauhaus - Universitt Weimar (2008)
Gerold, M.: Holzbrcken am Weg. Eigenverlag, Karlsruhe (2001)
Miebach, F.: Holzbrckenbau 3.0 als Weiterentwicklung. Brckenbau (March 2012)
Dietrich, J.R.: Faszination Brcken. Verlag Georg D. W. Callwey, Mnchen (1998)
Part II
Mechanical Connections
Steel-to-Timber Joints with Very High Strength
Steel Dowels Using Spruce, Beech and Azobe
Jan-Willem van de Kuilen1 , Carmen Sandhaas2 , and Hans Joachim Bla3

1 Holzforschung Munchen, Winzererstr. 45, 80797 Munchen
Delft University of Technology, Timber Structures and Wood Technology,
Stevinweg 1, 2628 CN Delft
vandekuilen@wzw.tum.de
2 Karlsruhe Institute of Technology, Timber Structures and Building Construction,
R.-Baumeister-Platz 1, 76131 Karlsruhe
Delft University of Technology, Timber Structures and Wood Technology,
Stevinweg 1, 2628 CN Delft
sandhaas@kit.edu
3 Karlsruhe Institute of Technology, Timber Structures and Building Construction,
blass@kit.edu
Abstract. Present-day very high strength steel (vhss) grades show high plastic
deformation capacity whilst reaching tension strengths of up to 1400 MPa. These
properties open new application fields in timber engineering. Replacing mild steel
dowels in timber joints with vhss dowels should lead to higher load carrying capaci-
ties or to leaner joints (thinner dowels and smaller cross sections) without losing the
desired joint failure mode with one or two plastic hinges per shear plane. Extensive
test series on double-shear timber joints with slotted-in steel plates have been car-
ried out using 12 mm and 24 mm dowels. The chosen timber species were spruce,
beech and azobe. One, three and five dowels in a row were tested and the used dowel
steel grades were high strength steel (hss) with a mean tension strength of 590 MPa
and vhss with a mean tension strength of 1390 MPa. The test outcomes have shown
that joints with vhss dowels reach a higher load carrying capacity than the same
joints using hss dowels, but are still able to develop plastic hinges. No correlation
between density, load carrying capacity and stiffness within one wood species could
be found. The effective number of fasteners showed a trend to be lower for the joints
with vhss dowels and at the same time, is dependent on the used wood species as
generally, ductile species such as beech show large deformations and subsequently
high load carrying capacities if one dowel is used.
Keywords: steel-to-timber joints, high strength steel dowels, spruce, beech, azobe,
plastic hinges, failure modes.
DOI: 10.1007/978-94-007-7811-5_15, RILEM 2014
158 J.-W. van de Kuilen, C. Sandhaas, and H.J. Bla
1 Introduction
In timber joints with dowel-type fasteners, the preferred failure mode is combined
failure of yielding of the fastener (one or two plastic hinges per shear plane) and
embedment of the timber. Apart from the geometry, the load carrying capacity is
therefore defined by the embedment strength of the timber and the yield moment of
the dowel-type fastener.
The utilisation of very high strength steels (vhss) with subsequent higher yield
moments is promising in order to optimise joints and get high performance joints.
Thinner dowels and thinner member sections should lead to the same load carrying
capacity as have thicker mild steel dowels with bigger member cross sections - as
long as failure mode 2 or 3 is reached (one or two plastic hinges per shear plane).
Another option is that fewer dowels could be used to obtain the same performance.
As for the plastic deformation capacity of the joints, very high strength steels avail-
able nowadays are able to ensure a good overall ductile behaviour without brittle
dowel failure modes.
This is especially valid and even more advantageous for high density timber. The
embedment strength of species with high densities is higher. Therefore, less member
thickness would be needed when using vhss dowels in comparison to softwood.
2 Review
The database of test results on timber joints with dowel-type fasteners, both dowels
and bolts, is huge [5, 8, 12, 16]. Most of the research was focussed on softwood and
only some research was done with hardwood [5, 16]. In previous research, hss or
vhss dowels were mainly used to ensure brittle failure modes in joints [14] and not
to investigate possible higher load carrying capacities.
Since 2006, first tests were undertaken at TU Delft to investigate the applicability
of hss dowels in double-shear timber-to-timber joints [6]. Azobe and spruce mem-
bers were assembled with steel dowels of grade S690 and diameters of d = 8, 16, 24,
30 mm. Further tests were carried out on double-shear timber joints with slotted-in
steel plates [7, 11, 15]. In [11, 15], tests were carried out with spruce and 8 mm
hss dowels of grade 12.9 (i.e. ultimate strength of 1200 MPa and yield strength of
1080 MPa) with 1, 3 and 5 dowels in a row. In [7], tests undertaken on double-shear
timber joints with slotted-in steel plates with tropical hardwoods and hss and mild
steel dowels. The dowel diameter was 8 mm and 1, 3 and 5 dowels in a row were
used.
In [9, 10] all above-mentioned research was analysed and results and conclusions
were summarised. The main conclusion with regard to the practicability of joints
with vhss fasteners was that the failure modes could be predicted with the Johansen
equations and plastic hinges could develop in the vhss dowels.
Timber Joints with Very High Strength Steel Dowels 159

Double-shear timber joints with slotted-in thick steel plates and two different dowel
steel grades were tested. Three different wood species were used, two different
dowel diameters of d = 12, 24 mm and 1, 3 and 5 dowels in a row. Per series, 5
specimens were tested which led to a total of 180 tests.
Test Setup. The tests were carried out according to EN 26891 [4]. The loading
scheme was load-controlled up to 70% of the estimated maximum force and after-
wards displacement-controlled up to failure. Four LVDTs, one on front and rear of
each timber member, were used to record the relative slip between timber members
and steel plate.
Steel Properties. Analoguously to [6, 7, 11, 15], the joints were designed with
the Johansen equations using mean material properties. The yield moment of the
dowels was calculated with Equation (1) from mechanics and not with the empirical
Equation (2) from EC5 [3].
d3
My = fy , (1)
6
My = 0.3 fu d 2.6 , (2)
where d = dowel diameter, fu = ultimate strength, fy = yield strength
The empirical Equation (2) derived by Blass et al. [1] is based on the assumption
that only at a bending angle of 45, the plastic capacity of the dowel is fully de-
veloped which is never the case for timber joints unless the fasteners are very thin.
Furthermore, Equation (2) was derived for steel grades with a ratio between yield
and ultimate strength of about 0.65 which is not the case for hss and vhss steel dow-
els where this ratio is 0.95 (see Table 1). Especially for large diameter fasteners,
Equation (2) is punishing [10].
Tension tests have been carried out on the used steel dowels to establish their
mechanical properties, see Table 1. The used steel dowels could be assigned to the
grades high strength steel with an ultimate strength between 500 and 700 MPa and
very high strength steel with an ultimate strength between 800 and 1100 MPa.
Wood Properties. The embedment strength parallel-to-grain fh,0 on the other hand
was calculated according to [2] as the embedment strength equation in EC5 was
Table 1 Results of tension tests on steel dowels
Steel grade n fy [MPa] fu [MPa] strain at failure [%]
hss 9 563 590 11.3

vhss 12 1311 1389 9.4
Table 2 End and edge distance, spacing, d = dowel diameter
Loaded end a3,t 7d

Spacing a1 5d
Edge distance a4 3d
considered to yield too conservative results. Equation (3) resulted from a regression
analysis of embedment test results on hardwoods:
fh,0 = 0.102 (1 0.01d) mean (3)

where mean is the mean wood density and d is the dowel diameter.
The chosen timber species were spruce (Picea abies) with a mean density of
mean = 445 kg/m3, beech (Fagus sylvatica) with mean = 715 kg/m3, and azobe
(Lophira alata) with mean = 1120 kg/m3 . Spruce and beech were stored at a 20/65
climate chamber whereas azobe was stored in a wetter climate chamber with 20C
and 85% relative humidity. The final mean wood moisture content was 12% for
spruce, 9% for beech and 21% for azobe.
Design. The member thickness was designed with the above-described assump-
tions for yield moment and embedment strength in such a way that the failure modes
of the joints with hss dowels lay exactly between failure mode 2 and 3 with 1 or 2
plastic hinges per shear plane respectively.
The end and edge distances listed in Table 2 were the same for all specimens and
corresponded to the minimum distances specified in EC5. It should be noted that the
fastener spacing a1 is smaller than the end distance a3,t .
All further information on used materials and methods can be taken from [13].
4 Experimental Results
The test results in terms of moisture content, density, maximum load and Kser (ac-
cording to [4]), all at time of test, are given in Table 3. The increase in load carrying
capacity when using vhss steel dowels is ranging between 10% for the beech joints
with three 12 mm dowels and 69% for the azobe joints with one 24 mm dowel. This
increase in load carrying capacity is expected to be even higher if mild steel dowels
would have been compared with vhss dowels. The wood quality of the spruce joints
with three 24 mm vhss dowels was very low which may explain their low load car-
rying capacity due to their premature splitting. (All results and descriptive statistics
can be found in [13].)
Fig. 1 shows exemplarily the load-slip graphs for the beech joints with 12 mm
vhss dowels. Plastic deformation behaviour is observable with mean deformations
at maximum load of 16 mm for the joints with one dowel (COV 14.5%), 7 mm for
the joints with three dowels (COV 34.9%) and 5 mm for the joints with five dowels
Fig. 1 Load-slip graphs for

beech joints with 12 mm
300 five dowels
vhss dowels
three dowels
250 one dowel
200
Load [kN]
150
100
50
0
0 5 10 15 20
Displacement [mm]
(COV 17.4%). Failure mode 2 with one plastic hinge per shear plane was observed
in all tests shown in Fig. 1 and as expected, the possible maximum deformation
decreases with increasing number of fasteners.
Effective Number of Fasteners. In Fig. 1, the influence of an effective number of
fasteners can be seen as the load carrying capacity of joints with three (170 kN) and
five (300 kN) dowels did not correspond to multiples of the capacity of a joint with
one fastener (70 kN).
In EC5, the effective number of fasteners ne f is calculated according to Equation
(4):
n
ne f = min a1 (4)
n0.9 4 13d
where n is the number of fasteners, a1 is the fastener spacing and d is the dowel
diameter.
In EC5, the effective number of fasteners is only depending on the fastener diam-
eter and the spacing, other possible influence factors such as slenderness or steel
grade/density are not taken into acount. For the tested specimens with dowel diam-
eters d of 12 mm and 24 mm and a spacing of 7d, the effective numbers of fasteners
according to Equation (4) are resulting to 2.12 for joints with three dowels and 3.35
for joints with five dowels.
If the EC5 predictions are compared with the effective number of fasteners re-
sulting from the tests as given in Table 4, ne f from tests are generally higher. If
comparing ne f for joints with vhss and hss dowels, it can be observed that espe-
cially for the timbers with higher densities, the effective number for the joints with
vhss dowels is lower than ne f for the joints with hss dowels.
Dependency on Density. Figs. 2 and 3 show the dependency of the load carrying
capacity and the stiffness Kser on the density of the timber members where a uniform
Table 3 Joint test results and ratio of Fmax with vhss dowels over Fmax with hss dowels
Number of Steel Moisture content Density Fmax [kN] Kser [kN/mm]

dowels grade [%] [kg/m3 ]
mean COV mean COV mean COV mean COV
Spruce, 12 mm
1 hss 11.9 2.4 456 3.6 32 8.4 21 34.5
vhss 12.1 1.6 457 5.8 44 3.4 20 32.6
3 hss 12.7 1.6 462 7.0 90 9.4 51 20.9
vhss 12.4 2.4 485 2.7 115 6.3 51 8.4
5 hss 12.5 1.9 460 5.6 136 9.3 52 12.3
vhss 12.6 1.5 478 1.9 169 15.7 68 6.5
24 mm
1 hss 12.2 1.9 424 10.8 110 12.2 126 27.5
vhss 12.0 1.8 413 4.2 165 4.2 118 13.8
3 hss 11.9 1.1 418 3.7 324 5.1 127 22.3
vhss 11.8 1.7 410 3.5 297 18.0 122 6.6
5 hss 12.4 1.7 436 4.4 482 9.3 137 6.1
vhss 12.5 2.3 439 4.7 570 9.4 247 44.0
Beech, 12 mm
1 hss 8.6 3.4 720 1.6 59 3.6 36 40.6
vhss 8.7 3.1 722 2.5 70 6.0 49 44.2
3 hss 8.5 0.7 730 2.2 157 4.6 61 28.8
vhss 8.7 1.4 732 4.8 173 4.9 67 8.4
5 hss 8.8 12.6 718 3.1 240 2.5 110 4.2
vhss 8.5 1.0 703 2.3 294 2.5 129 13.3
24 mm
1 hss 8.7 8.0 718 2.2 208 3.6 114 20.4
vhss 8.5 2.0 710 1.0 300 2.4 157 7.7
3 hss 8.4 0.8 711 1.7 510 5.0 202 17.2
vhss 8.3 1.3 724 0.9 659 5.1 205 2.9
5 hss 9.8 2.9 701 1.0 823 5.3 171 7.2
vhss 9.4 4.5 695 2.3 927 3.5 192 6.3
Azobe, 12 mm
1 hss 21.4 1.4 1115 2.3 57 6.5 45 29.5
vhss 22.0 1.4 1115 2.3 72 2.7 53 12.4
3 hss 21.8 0.9 1117 1.4 128 7.6 39 6.7
vhss 22.9 3.1 1131 4.0 183 7.9 64 11.2
5 hss 21.7 0.8 1115 1.6 233 14.1 132 6.0
vhss 17.2 2.4 1130 3.9 281 13.2 158 22.1
24 mm
1 hss 21.5 5.9 1122 3.1 181 10.6 103 15.1
vhss 21.5 3.1 1154 1.9 306 6.4 201 17.5
3 hss 22.6 10.7 1104 2.3 539 7.2 186 33.3
vhss 21.3 6.6 1116 1.9 706 5.5 246 13.8
5 hss 22.5 6.9 1118 2.4 867 11.0 321 16.8
vhss 15.9 15.7 1101 1.3 1128 16.0 374 40.9
F(vhss)/F(hss) Spruce Beech Azobe

12 mm 24 mm 12 mm 24 mm 12 mm 24 mm
one dowel 1.38 1.50 1.18 1.44 1.26 1.69
three dowels 1.29 0.92 1.10 1.29 1.43 1.31
five dowels 1.25 1.18 1.22 1.13 1.21 1.30
Table 4 Effective number of fasteners ne f
Spruce Beech Azobe

hss vhss hss vhss hss vhss
diameter 12 24 12 24 12 24 12 24 12 24 12 24
three
dowels 2.80 2.94 2.62 1.80a 2.65 2.45 2.48 2.20 2.23 2.98 2.53 2.31
five dowels 4.24 4.38 3.84 3.45 4.04 3.96 4.20 3.09 4.06 4.80 3.90 3.69
a Spruce joints with three 24 mm vhss dowels suffered from premature splitting due to very
low wood quality
200 Beech Azob Number

Spruce dowels
300 300 1
Fmax/dowel [kN]
Fmax/dowel [kN]
Fmax/dowel [kN]
150 3
5
200 200 Steel grade
100
hss
24mm dowels 24mm dowels 24mm dowels mild
12mm dowels 100 12mm dowels 100 12mm dowels
50
0 0 0
380 400 420 440 460 480 500 675 700 725 750 775 1075 1100 1125 1150 1175 1200
Density [kg/m3] Density [kg/m3] Density [kg/m3]
Fig. 2 Density versus maximum load per dowel
250 Spruce 250 Beech 250 Azob Number dowel

1
measured Kser per dowel
3
200 200 200 5
[kN/mm]
[kN/mm]
[kN/mm]
150 150 150 Diameter

12
24
100 100 100
50 50 50
0 0 0
380 400 420 440 460 480 500 675 700 725 750 775 1075 1100 1125 1150 1175 1200
Density [kg/m3] Density [kg/m3] Density [kg/m3]
Fig. 3 Density versus Kser per dowel and two shear planes
load and stiffness distribution between the single dowels was assumed. Within one
wood species, no correlation between density and Fmax and Kser could be found.
5 Conclusions
A comprehensive joint test series including the wood species spruce, beech and
azobe has been carried out using two different steel grades, vhss and hss. The ex-
perimental results are summarised below:
In timber joints, hss dowels can be replaced by vhss dowels. Joints with vhss
reached higher loads than the same joints with hss dowels.
No strong correlation could be found between density of the wood and load car-
rying capacity of the joints using one wood species.
The failure modes can be predicted with the Johansen equations. This is valid
also for different steel grades and wood species.
Ductile failure modes with one or two plastic hinges per shear plane were pos-
sible also when using vhss dowels as modern vhss steel grades possess enough
deformation capacity.
Compared to hss dowels, the failure modes of the joints with vhss dowels were
shifted towards failure modes with less plastic hinges. Furthermore, the bigger
the dowel diameters, the more a shift towards modes with no plastic hinges or
one plastic hinge per shear plane could be observed.
The observed effective number ne f for more dowels in a row showed a trend to
be lower for the joints with vhss dowels. The difference between ne f for hss and
vhss dowels was smaller for beech. This may be due to the higher deformation
capability of beech and hence higher ultimate loads in the joints with one dowel.
On wood species level, the stiffness Kser is not strongly related to the density
of the used wood species and only weakly to the dowel diameter. An effective
number ne f could be used for Kser .
Acknowledgements. Thanks to the companies Pollmeier Massivholz GmbH&Co KG for

donating the beech wood, De Groot Vroomshoop for sponsoring the spruce glulam and to
Groot Lemmer for sponsoring and manufacturing the azobe specimens.
References
[1] Blass, H.J., Bienhaus, A., Kramer, V.: Effective bending capacity of dowel-type fasten-
ers. In: CIB-W18 Meeting 33, Paper 33-7-5, Delft, The Netherlands (2000)
[2] Ehlbeck, J., Werner, H.: Softwood and hardwood embedding strength for dowel-type
fasteners. In: CIB-W18 Meeting 25, Paper 25-7-2, Ahus, Sweden (1992)
[3] EN1995-1-1: Eurocode 5. Design of timber structures - Part 1-1: General - Common
rules and rules for buildings (2004)
[4] EN26891: Timber structures - Joints made with mechanical fasteners - General princi-
ples for the determination of strength and deformation characteristics (ISO 6891) (1991)
[5] Gehri, E., Fontana, M.: Betrachtungen zum Tragverhalten von Passbolzen in Holz-Holz-
Verbindungen. Tech. Rep. Int. Bericht 83-1, Institut fur Baustatik und Stahlbau, ETH
Zurich, Switzerland (1983)
[6] Hieralal, R.: Application of high strength steel in steel pin connections and double-shear
timber joints. Master thesis, Faculty of Civil Engineering, University of Technology
Delft, The Netherlands (2006)
[7] Islamaj, L.: Application of high strength steel dowels in steel-to-timber double-shear
joints with slotted-in steel plate. Master thesis, Faculty of Civil Engineering, University
of Technology Delft, The Netherlands (2009)
[8] Jorissen, A.: Double-shear timber connections with dowel-type fasteners. Dissertation,
University of Technology Delft, The Netherlands (1998)
[9] Van de Kuilen, J.W.G.: Verbindungsmittel aus hochfesten Stahlen - Stabdubelver-

bindungen. In: Tagungsband Ingenieurholzbau - Karlsruher Tage: Forschung fur die
Praxis, Karlsruhe, Germany (2009)
[10] Van de Kuilen, J.W.G., De Vries, P.A.: Timber joints with high strength steel dowels.
In: 10th World Conference of Timber Engineering WCTE, Miyazaki, Japan (2008)
[11] Langedijk, A.: Houtverbindingen met hoge sterkte staal. Bachelor thesis, Faculty of
Civil Engineering, University of Technology Delft, The Netherlands (2007)
[12] Mohammad, M., Quenneville, J.H.P.: Behaviour of wood-steel-wood bolted glulam
connections. In: CIB-W18 Meeting 32, Paper 32-7-1, Graz, Austria (1999)
[13] Sandhaas, C.: Mechanical behaviour of timber joints with slotted-in steel plates. Dis-
sertation, University of Technology Delft, The Netherlands (2012)
[14] Schmid, M.: Anwendung der Bruchmechanik auf Verbindungen mit Holz. Dissertation,
University of Karlsruhe, Germany (2002)
[15] Van Groesen, J., Kranenburg, M.: Houtverbindingen met hoge sterkte staal. Bachelor
thesis, Faculty of Civil Engineering, University of Technology Delft, The Netherlands
(2007)
[16] Werner, H.: Untersuchungen von Holz-Verbindungen mit stiftformigen Verbindungs-
mitteln unter Berucksichtigung streuender Einflussgrossen. Dissertation, University of
Karlsruhe, Germany (1993)
Wood Load-Carrying Capacity of Timber
Connections: An Extended Application for Nails
and Screws
Pouyan Zarnani and Pierre Quenneville
Department of Civil and Environmental Engineering, Faculty of Engineering,

University of Auckland
pzar004@aucklanduni.ac.nz, p.quenneville@auckland.ac.nz
1 Motivation
The wood engineering community has dedicated a significant amount of effort

over the last decades to establish a reliable predictive model for the load-carrying
capacity of timber connections under wood failure mechanisms. Test results from
various sources (Foschi and Longworth 1975; Johnsson 2003; Quenneville and
Mohammad 2000; Stahl et al. 2004; Zarnani and Quenneville 2012a) demonstrate
that for multi-fastener connections, failure of wood can be the dominant mode.
In existing wood strength prediction models for parallel to grain failure in
timber connections using dowel-type fasteners, different methods consider the
minimum, maximum or the summation of the tensile and shear capacities of the
failed wood block planes. This results in disagreements between the experimental
values and the predictions. It is postulated that these methods are not appropriate
since the stiffness in the wood blocks adjacent to the tensile and shear planes
differs and this leads to uneven load distribution amongst the resisting planes
(Johnsson 2004; Zarnani and Quenneville 2012a).
The present study focuses on the nailed connections. A closed-form analytical
method to determine the load-carrying capacity of wood under parallel-to-grain
loading in small dowel-type connections in timber products is thus proposed. The
proposed stiffness-based model has already been verified in brittle and mixed
failure modes of timber rivet connections (Zarnani and Quenneville 2013b).
Keywords: nailed connections, rivet connections, brittle, ductile and mixed failure
modes, elastic spring model, ultimate capacity, design model.
2 Stiffness-Based Predictive Model
The proposed analysis for wood strength is best explained using the analogy of a
linear elastic spring system in which the applied load transfers from the wood
DOI: 10.1007/978-94-007-7811-5_16, RILEM 2014
168 P. Zarnani and P. Quenneville
member to the failure planes in conformity with the relative stiffness ratio of each
resisting adjacent volume to the individual failure plane (Fig. 1). By predicting
these volumes stiffness, one can derive the portion of the connection load that is
channelled to each resisting plane and from the resistance of each failure planes,
one can determine which failure plane triggers the connection failure.
Fig. 1 Proposed elastic spring model
The difference in the loads channelled to the tensile and shear planes is a
function of the modulus of elasticity and modulus of rigidity, the volume of wood
surrounding each of the failure planes (bottom, end and edge distances-dz, da and
de) and also the connection geometry (Fig. 2). For details regarding the
determination of stiffness of the resisting planes, refer to Zarnani and Quenneville
(2013b).
Fig. 2 Simplified analytical model
By predicting the stiffness of the wood surrounding each of the failure planes
(Kh, Kb and Kl), one can predict the proportion of the total connection load applied
to each plane, Ri = K i K . By further establishing the resistance of each of the
failure planes as a function of a strength criterion, one can verify which of the
Wood Load-Carrying Capacity of Timber Connections 169
failure planes governs the resistance of the entire connection. It should be asserted
that the strength of the shear planes cannot be higher than the tensile capacity of
the adjacent wood volume where the load is channelled to these resisting planes
(Fig. 3). If the attracted load by the resisting shear planes be larger than the tensile
capacity of in-contact wood volume, then the wood block torn out from the
member would be as wide as or as deep as the member corresponding to wood
failure mode (b) and (c) (Fig. 4).
Fig. 3 Loads acting on the wood volume adjacent to the shear resisting planes: (a) bottom
block, (b) lateral blocks
Thus, the wood load carrying capacity of the connection (Eq. 1) is the load
which results in the earlier failure of one of the resisting planes due to being
overloaded and equals to the minimum of Pwh, Pwb and Pwl. It is important to note
that the connection resistance given by Eq. 1 is a summation of the critical plane
failure load plus the load carried by the other planes.
Kb Kl
Pwh = f t ,m Ath (1 + + ) , Mode (a)
Kh Kh
K h Kl f v,m C ab Asb , Mode (a)

Pw = np . min Pwb = (1 + + ). min (1)
Kb Kb f t ,m X l d z , Mode (c)
K h Kb f C A , Mode (a)
Pwl = (1 + + ). min v,m al sl
Kl Kl 2 f t ,m t ef d e , Mode (b)
In Eq. 1, ft,m and fv,m are the wood mean strength in tension and shear along the
grain (MPa). Ath, Asb and Asl are the areas of the head, bottom and lateral resisting
planes with respect to the wood effective thickness, tef, subjected to tension and
shear stresses. Also, Cab and Cal are the ratios of the average to maximum stresses
on the bottom and lateral shear planes respectively (Zarnani and Quenneville
2013b) and Xl is the joint width. For a one-sided joint having only one plate, np=1,
the member thickness value, b, to be used to determine dz = b/2-tef is twice the
thickness of the wood.
Grain
Mode (a) Mode (b) Mode (c)
Fig. 4 Different possible failure modes of wood block tear-out
It should be noted that when one plane fails, then the entire connection load
transfers to the remaining planes in accordance with their relative stiffness ratios.
It could be possible that the occurrence of the first failure of one plane does not
correspond with the maximum load of the connection. This is more susceptible in
case of either small edge or bottom distances accompanying large shear resisting
areas which leads to wood failure mode (b) and (c) respectively (Fig. 4).
Therefore, Eq. 1 needs to be checked again for the remaining planes by defining
no value for the terms related to the failed planes.
In the case of fasteners which are inserted into predrilled holes, the area
corresponding to the cutting diameter is to be subtracted from the resisting plane
surfaces. This affects the strength of the tensile and shear resisting planes and not
their stiffnesses.
3 Design Procedure
The design of timber joints using small dowel-type fasteners such as rivets, nails
and screws is governed by either the brittle, mixed or ductile failure mode of the
joint. The occurrence zone of these potential failure modes is illustrated on a
typical load-displacement curve of a timber joint (Fig. 5).
In the brittle zone, the fasteners deflection is in the elastic range, therefore, the
effective wood thickness for the joint corresponds to the elastic deformation of the
fasteners, tef,e. In this failure zone, the wood capacity of the connection, Pw,tefe, is
less than the fastener yielding resistance, Pr,yld. It should be noted that the Pr,yld is
not an ultimate failure but constitute a boundary. As the yield point is reached, the
effective wood thickness reduces if the yield mode is not Mode I. This reduction
in effective wood thickness, tef,y, leads to the generation of a new connection
failure mode. If the wood capacity of the new connection, Pw,tefy, cannot resist the
Fig. 5 Occurrence zone of potential failure modes of timber joints
fastener yielding load, a sudden failure with slight deflection on the fasteners
which is called mixed failure mode occurs. Even if Pw,tefy > Pr,yld, the mixed failure
mode can happen as the deflection of the connection progresses if Pw,tefy is lower
than the connection ultimate ductile strength, Pr,ult. If the wood strength based on
tef,y is greater than Pr,ult, the ductile failure governs and there is no wood rupture.
By following the described mechanism for the potential failure modes, the
connection ultimate capacity, Pc,ult, can be predicted as follows (Zarnani and
Quenneville 2013a);
Pw,tefe if Pw,tefe < Pr,yld (Brittle mode)

Pr,yld if Pw,tefy < Pr,yld Pw,tefe (Mixed mode)
Pc,ult = Pw,tefy if Pr,yld Pw,tefy Pr,ult (Mixed mode) (3)
Pr,ult if Pr,ult < Pw,tefy (Ductile mode)
If a designer wants to rely only on the yield limit state as the connection
maximum capacity, therefore, the above design procedure can be simplified to
Pc,ult = min (Pw,tefe, Pr,yld).
4 Material and Experimental Method
4.1 BNG, MNG and DNG Test Series on LVL

The laboratory tests (Fig. 6) involved specimens with nail configurations designed to
observe the three possible modes of failure and to evaluate the effect of geometry on
connection strength. Specimens were manufactured from 90*200*1200mm and
90*240*1200mm New Zealand Radiata Pine LVL grade 11. The tests series
involved 8 groups with 3 replicates for each configuration. Two sizes of nail were
used; nails with a shank length of 30mm and a diameter of 3.2mm (Lp=22mm) and
nails with a shank length of 33mm and a diameter of 3.8mm, (Lp=25mm). The
geometric parameters of connections (Fig. 7) tested varied from 8 to 10 for number
of rows (nR) and 5 to 10 for number of columns (nC) as shown in Table 1. Sp of 32,
38 and 72mm (Sp 10d) and Sq of 8 and 10mm (Sq 5d) were adopted conforming
with NZS3603 (Standards New Zealand 1993). The holes in the nail plate along the
columns were staggered by half of the column spacing (Sp) to increase the amount of
nails for the failure block size while still conforming to the codes minimum spacing
perpendicular to grain. Edge distance, de varied from 55 to 92mm and end distance
from 50 to 100mm (da >12d). The specimens had nailed plates on both faces of
timber, resulting in a symmetric connection. The nail plates were 8 mm thick of 300
grade steel with predrilled 3.3mm and 3.9mm holes to ensure the head of the nail
was rotationally fixed. Nail holes on the wood were not pre-drilled. In six of the
groups, the characteristics were specified in order to prompt wood failure and
maximize the amount of observations on the brittle and mixed failure mechanism.
All specimens were conditioned to 20C and 65% relative humidity to attain a
target 12% equilibrium moisture condition (EMC). The wood had an average
density of 607 kg/m3 with a coefficient of variation of 2% at the time of the tests
and an average moisture content of 13.2%. The wood mean tensile and shear
strengths (ft,m, fv,m) used in the analysis were determined from the characteristic
values provided by the manufacturer. A coefficient of variation of COV=15% and
COV=10% for the wood tensile and shear strengths respectively was used to back
calculate the average values. The estimated ft,m and fv,m were 39.8 MPa and 7.2
MPa correspondingly.
Fig. 6 Typical specimen Fig. 7 Definition of connection

in testing apparatus geometry variables
4.2 REC, GRP, NOR, SPR and TEN Test Series on Glulam
The REC, GRP, NOR, SPR and TEN test series presented in Table 1 were
conducted by Johnsson (2003). A similar test setup was used as shown in Fig. 7.
The specimens cross sections were 90*225 to 360 mm from glulam of strength
class GL28c (produced from Norway Spruce). In the RECX series, the thickness
of the specimen varied between 66, 78, 90, and 140 mm with 5 replicates in each
group. The GRP series had nails placed in two groups with a gap in between equal
to 75, 150 and 300 mm for GRPS, GRPL and GRPX respectively. All the joints
consisted of one steel plate on one side. The nail used had a diameter of 4.0 mm
and penetration depth of Lp=40 mm. Nails were inserted in pre-drilled holes for all
the test series.
The wood had an average density of 454 kg/m3 and an average moisture
content of 11.2% (Johnsson 2004). The wood mean tensile strength reported in
Johnsson (2004) and the mean shear strength found by Crocetti et al (2010) for the
same timber class of GL28c were used as inputs to the proposed model. The ft,m
and fv,m were 28.4 MPa and 4.9 MPa respectively. For the stiffness properties,
based on data available in the literature, an average ratio of modulus of rigidity to
modulus of elasticity (G/E) is considered equal to 0.045 and 0.069 for LVL and
glulam respectively in order to make the planes stiffness equations independent
of G and E values.
5 Results and Discussion
5.1 Test Observations

The load slip curve for each specimen tested in Auckland was plotted and the
ultimate load and the types of failure were recorded. The peak loads ranged from
180 kN to 324 kN. When brittle or mixed failure occurred, the connection did not
develop the full ductile capacity of the fasteners and the ultimate capacity of the
connection was dictated by plug shear failure initiation. Specimens that failed
brittlely experienced displacements of approximately 1.5mm. Specimens for
which a mixed failure was observed suffered slightly more slip, experiencing
minor plastic deformation of the fasteners before plug shear failure occurred at a
slip of 1.5-2.5mm. Brittle and mixed failure occurred in tests groups with a large
number of tightly spaced nails (Fig. 8). When a brittle plug shear failure occurred,
a block of wood contained within the nail group was pulled away from either one
side or both sides of the specimens. In the case of ductile failures, the connections
were able to reach the full capacity of the nails. Before the ultimate capacity was
reached, the connections experienced large deformations, far beyond yield
(defined by the 5%-offset method) and design serviceability allowances.
The specimens were named based on the observed modes of failure. BNG,
MNG and DNG stand for nail group configurations with brittle, mixed and ductile
modes of failure respectively.
Fig. 8 Specimen and failed block exhibiting the brittle/mixed modes of wood failure
Table 1 Configuration of the nailed joints on LVL and glulam
Spacing of
Nail diameter
No. of rows rows
Test and penetration Member Member End
and columns and columns thickness width distance
groups* (mm)
(mm)
b (mm) W (mm) da (mm)
nR nC Sp Sq d Lp
BNG1-L 10 10 32 8 3.2 22 90 200 50
MNG2-L 10 10 32 8 3.2 22 90 200 50
MNG3-L 8 10 32 8 3.2 22 90 200 50
MNG4-L 8 10 32 8 3.2 22 90 240 50
MNG5-L 8 10 32 8 3.2 22 90 200 100
DNG6-L 8 6 32 8 3.2 22 90 200 50
MNG7-L 10 8 38 10 3.8 25 90 200 50
DNG8-L 10 5 72 10 3.8 25 90 200 50
RECS-G 9 7 28 7 4.0 40 90 225 66
RECL-G 19 8 28 7 4.0 40 90 265 80
RECX0-G 19 15 28 7 4.0 40 66 265 60
RECX1-G 19 15 28 7 4.0 40 78 265 60
RECX2-G 19 15 28 7 4.0 40 90 265 60
RECX4-G 19 15 28 7 4.0 40 140 265 60
GRPS-G 19 10 28 7 4.0 40 90 265 56
GRPL-G 19 13 28 7 4.0 40 90 265 56
GRPX-G 19 18 28 7 4.0 40 90 265 56
NORS-G 7 5 40 10 4.0 40 90 225 60
NORL-G 13 6 40 10 4.0 40 90 265 60
NORX-G 15 11 40 10 4.0 40 90 280 60
SPR-G 19 8 40 7 4.0 40 90 265 56
TENS-G 13 2 28 7 4.0 40 90 225 60
TENL-G 33 2 28 7 4.0 40 90 360 60
*
L and G stand for LVL and glulam respectively. In LVL groups, the joints are double-
sided (np=2) and in glulam one-sided.
5.2 Wood Effective Thickness

The thickness of the wood block pulled out of the brittle and mixed failure mode
specimens was measured. The thickness was compared to the predicted effective
thickness, tef (Table 2). For the brittle failure mode, the elastic deformation of the
nail modelled as a beam on an elasto-plastic foundation was assumed to determine
the tef,e (Zarnani and Quenneville 2012c) and in the case of mixed failure, the tef,y
was predicted based on the bearing length corresponding to the governing yielding
mode of the nail (Zarnani and Quenneville 2013b).
Table 2 Experimental results on wood effective thickness compared to predictions
Nail tef,e (mm) tef,y (mm)

Penetration
diameter, for brittle failure mode for mixed failure mode
length, Lp
d Average Average
(mm) predicted predicted
(mm) observed observed
3.2 22 0.87Lp = 19.1 17.5 16.4 14.6
3.4 25 0.84Lp = 21.0 - 18.7 19.9
4.0 40 0.75Lp = 30.0 27.5* 15.1 -
*
Average of 25-30 mm, based on the tests conducted by Johnsson (2003).
The average measured block depth was within 2mm of the predicted effective
thickness for both the brittle and mixed failure modes. For the mixed failure
modes, there was a visible deformation of the nails, in the form of a plastic hinge
formed at the plate (Fig. 9). This plastic hinge is consistent with the predicted
ductile yield failure mode IIIm using Johansens yield theory (1949) which is the
foundation for the European Yield Model (EYM) prediction formulas in Eurocode
5 (2004). As shown in Table 2, a good agreement can be found between the
predictions and observations for the effective wood thickness.
Fig. 9 Wood mixed failure showing deformed nails

5.3 Validation of the Proposed Stiffness-Based Model

Results for the current LVL groups tested and tests on glulam reported in
Johnsson (2003) are listed in Table 3 along with the predominant modes of failure
observed. The connection ultimate capacities were calculated using the proposed
stiffness-based model and the algorithm presented to determine the connection
failure mode. As shown in Table 3, for the BNG test group, the estimated wood
strength corresponding to the nail elastic deformation, Pw,tefe, was lower than the
nail yielding resistance, Pr,yld. For this configuration, the connection ultimate
capacity is predicted as Pc,ult=Pw,tefe with a brittle failure mode which is consistent
with the observation. However, in the MNG test groups, the predicted wood
strength for tef,e is higher than the nails yielding strength. The strength of the
connection is thus checked for the possible mixed or ductile modes of failure. In
these test series a mixed mode failure occurred which can be explained by the
wood strength corresponding to tef,y being weaker than the nails ultimate strength
(Pw,tefy < Pr,ult). In the MNG2 test group, since Pw,tefy is greater than Pr,yld, thus the
Table 3 Nailed connection ultimate strength prediction using the proposed analysis
compared to experimental results on LVL and glulam
Connection
Wood strength Nail Wood strength Nail
Ultimate strength
corresponding yielding corresponding ultimate Failure mode
Test Pc,ult (kN)
to nail elastic strength to nail strength
groups* Ratio
deformation Pr,yld yielding mode Pr,ult
Pw,tefe (kN) (kN) Pw,tefy (kN)* (kN)* Predicted Observed of pred. Pred. Obser.
(COV%) to
obser.
BNG1-L 248 271 N/A N/A 248 298 (9%) 0.83 BRT BRT
MNG2-L 262 244 250 338 250 263 (3%) 0.95 MIX MIX
MNG3-L 230 216 205 N/A 216 231 (4%) 0.94 MIX MIX
MNG4-L 230 216 205 N/A 216 228 (1%) 0.95 MIX MIX
MNG5-L 214 216 N/A N/A 214 263 (7%) 0.81 BRT MIX
DNG6-L 222 130 192 180 180 180 (5%) - DUC DUC
MNG7-L 305 272 293 440 293 324 (5%) 0.90 MIX MIX
DNG8-L 292 170 283 275 275 275 (4%) - DUC DUC
RECS-G 73 136 N/A N/A 73 88 (9%) 0.83 BRT BRT
RECL-G 138 329 N/A N/A 138 162 (5%) 0.85 BRT BRT
RECX0-G 199 634 N/A N/A 199 250 (9%) 0.80 BRT BRT
GRPS-G 168 329 N/A N/A 168 181 (7%) 0.93 BRT BRT
GRPL-G 184 329 N/A N/A 184 217 (5%) 0.85 BRT BRT
GRPX-G 218 329 N/A N/A 218 229 (8%) 0.95 BRT BRT
NORS-G 88 81 77 N/A 81 97 (5%) 0.84 MIX BRT
NORL-G 164 166 N/A N/A 164 178 (6%) 0.92 BRT BRT
NORX-G 229 361 N/A N/A 229 293 (5%) 0.78 BRT BRT
SPR-G 187 329 N/A N/A 187 253 (3%) 0.74 BRT BRT
TENS-G 46 48 N/A N/A 46 59 (4%) 0.78 BRT BRT
TENL-G 123 152 N/A N/A 123 136 (11%) 0.90 BRT BRT
*
Not applicable (N/A) for all the test groups based on the presented design algorithm.

BRT, MIX and DUC stand for brittle, mixed and ductile failure modes, correspondingly.
connection ultimate cap pacity is determined as Pc,ult=Pw,tefy. However, foor

the MNG3 and MNG4 test series where Pw,tefy < Pr,yld, Pc,ult is predicted as Pr,yld. IIn
the case of the DNG test groups,
g the wood strength based on tef,y is greater than thhe
connection ultimate ductile strength, Pr,ult, therefore, the ductile failure governns
(Pc,ult=Pr,ult) and there wass no wood rupture.
For the REC, GRP, NO OR, SPR and TEN test series (Johnsson 2003), since thhe
nails were inserted into predrilled holes, the area corresponding to the cuttinng
diameter (0.8d) was subtrracted from the resisting plane surfaces. This affected thhe
strength of the tensile and d shear resisting planes and not their stiffnesses. For thhe
GRP series in which the nails
n were placed in two groups with a gap in between, it
was assumed that the gap g is filled with the nails by the same pattern (thhis
assumption was only app plied for defining the wood capacity). As shown in Tabble
3, the predicted wood strength
s corresponding to the nail elastic deformation,
Pw,tefe, was lower than th he nails yielding resistance, Pr,yld (except for the NO OR
group). Therefore, Pc,ult=P= w,tefe and the observed thickness of the brittle faileed
blocks is comparable to the t bearing length based on the nail elastic deformatioon
(Table 2). The yield and ultimate capacities of the nails with 3.2 and 3.8 mm iin
diameter were derived basedb on the conducted tests and based on the valuees
reported in Johnsson (20 003) for the tests with nails of 4.0mm diameter. Thhe
yielding strength used in the analysis were 1.4, 1.7 and 2.3 kN for the 3.2, 3.8 annd
4.0 mm nails diameter resspectively and for the ultimate resistance, 1.9, 2.8 and 3.2
kN correspondingly. For all the experimental groups with the brittle/mixeed
fashion, the failure modess of the wood block tear-out was predicted to be mode (a)
(Fig. 4) which was in linee with the test observations.
Fig. 10 Comparison between

n predictions and test data in brittle/mixed failure modes
As shown in Table 2 and 3, there is very good conformity between the

predictions and observations for the thickness of the failed block, the governing
failure mode, and the strength of the connection. Fig. 10 shows the strength
predictions of the experimental groups compared to the tests results. One can note
that the proposed analysis results in precise predictions with a correlation
coefficient (r2) of 0.92, a mean absolute error (MAE) of 16.6% and a standard
deviation (STDEV) of 9.4%. It is assumed that this method of calculation could
also be applicable to screwed connections.
6 Conclusions
The closed form predictive model for determining the wood capacity of rivet
connections in glulam and LVL, developed by Zarnani and Quenneville (2013b),
can be applied to nailed connections. The method takes into account the strength
of the failure planes and the stiffness of the adjacent wood which distributes the
member load to these planes. Also, a design algorithm is presented which allows
the designer to calculate the resistances associated with the predictions of the
different possible brittle, ductile and mixed failure modes. Results of nailed joint
tests on LVL and the test data available from Johnsson (2003) on glulam confirm
the validity of this new method and show that it can be used as a design provision
for the wood load-carrying capacity prediction in timber connections. The
proposed method can be extended to other small dowel type fastener such as
screws to predict accurately the connection ultimate capacity and its failure mode.
Acknowledgements. The authors wish to thank the New Zealand Structural Timber
Innovation Company (STIC) for funding this research work. Likewise the authors would
like to express their gratitude to The University of Auckland undergraduate students
Alexander Jessep and Andrew McQueen, who did the experimental work as part of their
final year project.
References
Crocetti, R., Gustafsson, P., Danielsson, H., Emilsson, A., Ormarsson, S.: Experimental and
numerical investigation on the shear strength of glulam. In: Proc., International Council
for Research and Innovation in Building and Construction, CIB-W18, New Zealand,
paper 43-12-5 (2010)
European Committee for Standardization (CEN). Eurocode 5-Design of timber structures.
EN 1995-1-1:2004, Brussels, Belgium (2004)
Foschi, R.O., Longworth, J.: Analysis and design of griplam nailed connections. J. Struct.
Div. ASCE 101(12), 25372555 (1975)
Johansen, K.W.: Theory of timber connections. Publications of International Association
for Bridge and Structural Engineering 9, 249262 (1949)
Johnsson, H.: Plug shear failure in nailed timber connections: experimental studies. In:
Proc., International Council for Research and Innovation in Building and Construction,
CIB-W18, Colorado, paper 36-7-2 (2003)
Johnsson, H.: Plug shear failure in nailed timber connections. Doctoral dissertation,
Department of Civil and Environmental Engineering, Lulea University of Technology,
Sweden (2004)
Quenneville, P., Mohammad, M.: On the failure modes and strength of steel-wood-steel
bolted timber connections loaded parallel-to-grain. Can. J. Civ. Eng. 27(4), 761773
(2000)
Stahl, D.C., Wolfe, R.W., Begel, M.: Improved analysis of timber rivet connections. J.
Struct. Eng. ASCE 130(8), 12721279 (2004)
Standards New Zealand. NZS 3603:1993 Timber Structures Standard, Wellington, New
Zealand (1993)
Zarnani, P., Quenneville, P.: A stiffness-based analytical model for wood strength in timber
connections loaded parallel to grain: Riveted joint capacity in brittle and mixed failure
modes. In: Proc., International Council for Research and Innovation in Building and
Construction, CIB-W18, Sweden, paper 45-7-1 (2012a)
Zarnani, P., Quenneville, P.: Reliable yield model for strength prediction of timber rivet
connection under ductile failure. In: Proc., 12th World Conference on Timber Eng.,
Auckland, New Zealand (2012b)
Zarnani, P., Quenneville, P.: Wood effective thickness in brittle and mixed failure modes of
timber rivet connections. In: Proc., 12th World Conference on Timber Eng., Auckland,
New Zealand (2012c)
Zarnani, P., Quenneville, P.: Design procedure to determine the capacity of timber
connections under potential brittle, mixed and ductile failure modes. In: Proc.,
International Council for Research and Innovation in Building and Construction, CIB-
W18, Vancouver (2013a)
Zarnani, P., Quenneville, P.: Wood block tear-out resistance and failure modes of timber
rivet connections - A stiffness-based approach. Journal of Structural Engineering, ASCE
(in press, 2013b)
Ductility in Timber Structures: Investigations
on Over-Strength Factors
Frank Bruhl1 , Jorg Schanzlin2, and Ulrike Kuhlmann1

1 Universitat Stuttgart, Institute of Structural Design, Pfaffenwaldring 7,
70569 Stuttgart, Germany
{Frank.Bruehl,U.Kuhlmann}@ke.uni-stuttgart.de
2 Konstruktionsgruppe Bauen AG, Bahnhofplatz 1, 87435 Kempten, Germany
Schaenzlin@kb-ke.de
Abstract. This paper presents a study on the implementation of connection ductil-

ity in timber structures. Regardless for which purpose ductility in timber structures
is needed, it is necessary to avoid a brittle failure of the timber element before the
ductile element is in the stage of yielding. An over-strength factor is introduced
to consider the required distance of the load-bearing resistance of the beam ele-
ment from the introduced bending moment initiated by the load-carrying capacity
of the fasteners. Hence, a Monte-Carlo simulation was conducted, focusing partic-
ularly on the scattering of the material properties to determine a reliability index of
a joint loaded in bending. Since the reliability index is based on the application of
the ductility, a range of over-strength factors is given for different reliability indices.
The Monte-Carlo simulation is based on the mean material properties of the exper-
imental specimens. The experiments are explained and the non-linear behaviour is
displayed not only for connections loaded in tension but also for joints loaded in
bending.
Keywords: dowel type connections, reinforcement, over-strength factor, reliability,

Monte-Carlo simulation.
1 Introduction
The question of ductility of structural elements has attracted more and more interest
in recent years. The demand for ductile connections is widespread. The moment-
rotation behaviour of a connection is important in seismic design [1], in the con-
sideration of the redistribution of internal forces [2, 5.1(3)] or if robustness [3] is
demanded. However, no information about the non-linear behavior of fasteners is
given [2]. Therefore it is necessary to gain knowledge about the ductile behaviour
of connections and joints in timber structures.
Several types of connections behave in a ductile manner; in particular dowel type
fasteners generally show a significant ductile behaviour [4] if timber is not at risk
of splitting within the connection [5, 6]. By the introduction of simplified nonlinear
load-slip relations of different types of fasteners, it is possible to integrate the ductile
DOI: 10.1007/978-94-007-7811-5_17, c RILEM 2014
182 F. Bruhl, J. Schanzlin, and U. Kuhlmann
behaviour for practical applications as already included in EN 1995-1-1 [2, 5.1(3)].

The inherent material properties of timber only allow the formation of a plastic hinge
within the joint. Therefore it is indispensable to introduce an over-strength factor,
which ensures that the joint is in the state of yielding before brittle failure occurs.
2 Ductile Behaviour of Dowel-Type Connections

2.1 Material
The width of all timber members was 180 mm, with an associated timber grade
of GL24h. A homogenous lamination was chosen to ensure that all dowels were
embedded in the same timber grade. The mean value of density in the connection
experiments was 443.5 kg/m3 and 444.4 kg/m3 for the joint test setup.
Previous investigations on dowel-type connections showed that the tensile strength
is usually higher than the target value of the ordered steel grade [7, 8]. Within these
investigations, the dowels had an ordered steel grade of S235 and a diameter of 12
mm. The tensile strength was tested at the beginning of the test series to ensure that
all of the dowels belonged to the same lot. Again, an enhanced value of 581 N/mm2
was ascertained, which is outside the range of 360-510 N/mm2 given in [9].
All of the experiments conducted were displacement-controlled following EN
26891 [10]. Failure was defined as a rupture of the specimens or as a decrease of the
load to 80% of the maximum load [11].
2.2 Connection Tests

The connection tests were conducted on three different dowel arrangements with a
slotted plate connection. All of the specimens were reinforced with fully
threaded screws and tested under a tension
loading. The examined dowel group was set
at the bottom of the specimen; a steel cover
was attached on the top of the specimens with
fully threaded screws (see Figure 1). There-
fore it was ensured that only one connection
was tested until the defined failure occurred,
regardless of the single-load-carrying capac-
ity. The dowel arrangements varied from a
wide alignment of 24 dowels to a rather
stretched arrangement of 52 dowels. The
first number indicates the number of dowels
parallel to the grain and the second the num-
ber of dowels perpendicular to the grain. Fig-
ure 2 shows the mean test results, based on
four experiments conducted for the different
Fig. 1 Connection test setup dowel arrangements. All of the experiments
Ductility in Timber Structures: Investigations on Over-Strength Factors 183
showed a constant ductile behaviour. Hence the ductility was evaluated based on the
definition Df = uf /uy [12]. A ductility ratio of 9 to 9.5 could be achieved. The con-
nections are therefore classified as highly ductile [8]. The displacement at yielding
(uy ) was determined by the regulations of [11, 13], and uf describes the displace-
ment at failure. If the classification is not only based on the relative value Df , but
also on the absolute value Dfy = uf -uy , a plastic displacement of more than 20 mm
could be achieved [14].
In order to compare the different dowel arrangements, the applied load was nor-
malized to the bearing resistance calculated based on the initial material properties
of one fastener [15]. It is shown that the load-carrying capacity is almost identical
for one fastener within each arrangement. Therefore it can be confirmed that within
reinforced connections the effective number of fasteners is equal to the installed
number [6, 2]. Furthermore, the initial stiffness shows a good accordance regardless
of the fastener arrangement.
2.3 Simplified Mechanical Model

No information on the non-linear load-displacement behaviour is given in the cur-
rent standards [2, 13]. In order to achieve practical applicability, it is necessary to
introduce a simplified method. The introduced approach is based on known meth-
ods associated with the load-bearing resistance and the stiffness known from the
standard [2]. The simplified method describes the load slip behaviour as a tri-linear
approach which follows the procedure of EN 1993-1-8 [16].
Thus the first part of the triangular graph is characterized by the initial stiffness
(Kser ) and the load-carrying capacity (Fv,Rk ) of the fastener based on Equation (9).
The first describing point is found by:
1,4
1,2
F / Fv, Rk [15]
1,0
0,8
0,6
0,4
Mean 24
Mean 33
0,2 Mean 52
0
0 5 10 15 20 25 30 35
Displacement [mm]
Fig. 2 Mean values of the load-displacement behaviour in the experiments conducted (

12mm)
F1 2
Fv,Rk
2 u1 = = 3 1,5 (2)
F1 = F v,Rk (1) K1 k
d
3 23
The stiffness of the second part is given by one third of the initial stiffness. Hence
the second point is calculated as:
F Fv,Rk
u2 = u1 + = u1 +
1
3 K1 K1
F2 = F v,Rk (3)
2
Fv,Rk Fv,Rk 5
Fv,Rk
= 3
+ = 3
(4)
K1 K1 K1
The third part is characterized by an infinitesimal stiffness.
Figure 3 shows the comparison of the gained load-displacement behaviour based
on the simplified model and the mean value obtained in all experiments on speci-
mens with a diameter of 12 mm. Two different proposals are shown: on the one hand
the simplified behaviour based on the initial material properties, and on the other the
material properties according to the standard. Within the standardized determination
a steel grade of S355 (fu,k =510 N/mm2 ) was chosen and a mean value of the density
of g,mean =420 kg/m3 [17] was used to calculate the bearing resistance.
The load-bearing resistance found experimentally shows a good conformity with
the tri-linear approach based on the initial properties. Since the input values of the
standardized determination are lower than the actual properties, the load-bearing re-
sistance underestimates the experimental value. On the other hand, the initial stiff-
ness given by the characteristic density g,k based on [17] confirms the experimental
studies.
The displacement at failure (uf ) is determined as the 2% percentile based on all
experiments conducted on specimens with a diameter of 12 mm. Assuming a log-
1,2
1,0
K3 =0
0,8
F / Fv,Rk [15]
F1 K2 = 31 K1
0,6
0,4
Mean 12mm
0,2 Approach based on [15]
K1 Approach based on experimental data
0
0 5 10 15 20 25 30
u1 u2 uu
Displacement [mm]
Fig. 3 Comparison of the load-displacement curve of the tri-linear approach with the mean
value of the tension component obtained experimentally ( 12mm)
1,2
1,0
0,8
M / Mmax, 52
0,6
0,4
Mean 24
Mean 33
0,2 Mean 52
0
0 20 40 60 80 100 120 140 160
Rotation [mrad]
Fig. 4 Mean values of the moment-rotation behaviour observed experimentally ( 12mm)
(normalized to the maximum moment of dowel arrangement, 52)
normal distribution, a 2% percentile value of 25.4 mm is determined with a mean of

32.1 mm and a coefficient of variation of 11%.
2.4 Joint Tests

Besides the pure connection tests, experiments on joints were conducted to prove
the rotational capacity of moment-resistant joints. The experiments were performed
as four-point bending tests with a joint acting in the middle of the test setup. The
bending moment is split into a compression force and a tension force (see Figure
4). The previously described connections are acting in the tension zone, whereas
the compression force is transferred via a defined compression zone (steel block 65
mm180 mm).
Figure 4 shows the mean values determined on the basis of three experiments
on specimens with a dowel diameter of 12 mm. The given rotation is associated
with either side of the joint. A dowel arrangement of 52 dowels consists of the
highest bending moment capacity due to the larger inner lever arm and the largest
number of fasteners. Furthermore it can be seen that the decrease in the plastic
level is more pronounced for a stretched arrangement (52) compared to a rather
wide arrangements (24). The decrease in the moment capacity is associated with
a geometrical effect, which is described in [18].
3 Introduction of an Over-Strength Factor

3.1 General
The previous chapter has shown that the variation of the material properties has a
significant influence on the load-bearing resistance of the dowel-type fastener (see
brittle element ductile element brittle element

Rcs Rjoint Rcs
Rjoint kcs Rcs
Fig. 5 Series of different structural elements [19]
Figure 3). Regardless of the application of ductility in timber structures, it is impor-

tant to ensure that the plastification of the ductile element takes place before a brittle
element fails. The capacity design method was developed by Paulay and Priestley
[19] with regard to the earthquake safety of reinforced concrete structures. The gen-
eral idea is illustrated in Figure 5. An over-strength factor kcs is introduced (comp.
Equation (5)), which ensures that no brittle failure occurs with a certain probability.
M joint
kcs kcs < 1.0 (5)
Mcs
The current version of the Swiss timber code [13, 4.6.3.1] already implies an over-
strength factor. Hence the brittle element must consist of a 20% higher bearing re-
sistance as the ductile element.
3.2 Determination of the Over-Strength Factor

The over-strength factor is determined based on a Monte-Carlo simulation. The in-
vestigations are conducted following the geometrical and material properties of the
experiments loaded in bending (comp. Section 2). The reliability index is deter-
mined with a C++ program based on 108 calculations following the equation given
in EN 1990 [20] assuming a normal distribution of the limit state function:
g
= (6)
g
where g represents the mean value of the limit state function and g the corre-
sponding standard deviation. The limit state function is given by [20]:
g = RE (7)
R represents the resistance and E the effect on the system.

Within this consideration the load-carrying capacity of the dowel-type fasteners
is set as an effect on the brittle element. The limit state function therefore becomes:
g = Wnet fm XM cs n Fv,Rk e = Mcs XM cs M joint (8)
with:
Table 1 Input variables based on [22] and test results [24]
tensile timber bending model

diameter width height
strength density strength uncertainty
fu [MPa] [mm] [kg/m3 ] fm [MPa] [mm] [mm] [-]
log-normal normal log-normal log-normal
distribution normal normal normal
[21] [22] [22] [21]
581
180
320

2x4 mean value 441.7 33.8
test
12
1


COV 0.04 [21] 0.001 0.1 [22] 0.15 [22] 0.0025 0.0015 0.1 [21]
Wnet : net section modulus fm : bending stress

XM : model uncertainty cs : variable to determine kcs
n : number of fasteners e : inner lever arm (see also [18])

Rk,1 = fh,i,k ti d [2]

Fv,Rk = min Rk,2 = fh,i,k ti d 4My,k
2 + f dt 2 1 [2] (9)

h,i,k i
R = 2 2 M f
k,3 y,k h,i,k d [15]
Table 1 shows the main input variables of the Monte-Carlo simulations. The material
properties of the lamellae used for the test beams were recorded during the manu-
facturing process. Therefore it was possible to gain knowledge about the density and
the tension strength of the fabricated lamellae within the different specimens. The
bending strength of the beams was determined based on the tension strength of the
boards, following the equation given in [23].
Certain material properties of timber are showing a dependence on each other
[22]. According to [22] a correlation coefficient of 0.6 was introduced to take into
account the dependency of the bending strength on the density within the beam
element.
3.3 Results of the Monte-Carlo Simulation

Figure 6(a) shows the reliability index obtained by the Monte-Carlo simulation for
different input values of cs . The load-carrying capacity of the fasteners has a direct
influence on the reliability index. The decrease of the reliability index for a dowel
arrangement of 52 with a larger inner lever arm and a larger number of fasteners
is higher compared with a dowel arrangement of 24 dowels with a smaller lever
arm.
A hidden over-strength factor is inevitably integrated within the design of joints.
This is the result of the differences in the design of the bearing resistance of the
8
24 8
7 33 (kcs ) = 7.65 7.65 kcs
52 7
24
6
6 33
5 52
5
4

3
3
2 2
1 1
0 0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1,6 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1,6
cs kcs
(a) Factor cs for different reliability indexes (b) Normalized factor kcs for different relia-
of various dowel arrangements bility indexes
Fig. 6 Over-strength factors cs /kcs depending on the the reliability index based on prop-
erties given in Table 1
cross-section (Mcs,design ) and the design of the load-carrying capacity of the joint
(Mjoint,design ).
An over-strength factor between 0.5 and 0.75 was considered in the design of the
experimental joint. Therefore the normalized over-strength factor kcs can be found
by:
M joint,design
kcs = cs (10)
Mcs,design
The design values Mjoint,design and Mcs,design are determined based on the mean in-
put values of the Monte-Carlo simulation (comp. Table 1). Figure 6(b) shows the
normalized over-strength factors kcs of the different reliability indexes for the
considered joints [14]. The obtained reliability line can be expressed by:
(kcs ) = 7.65 7.65 kcs (11)
The verification of the conducted reliability line shows a reliability index of zero
for a kcs value of one. A reliability index of zero gives a failure probability (Pf ) of
0.5. This means that the system is in an indifferent condition. This is a reasonable
number since both the effect and the resistance are of the same magnitude.

Reinforced dowel-type connections show a significant ductile behaviour not only
when loaded in tension but also when loaded in bending. A simplified model is
given to describe the non-linear behaviour of connections based on known parame-
ters. Regardless for which purpose ductility is needed, it is important to ensure that
the ductile behaviour is activated before a brittle member fails. A reliability line is
determined based on a Monte-Carlo simulation, which gives certain over-strength
factors. Thereby an over-strength factor can be chosen.
Further investigations of the reliability line need to be conducted to consolidate

the results.
Acknowledgements. This research work (16184N) coordinated by International Associa-

tion for Technical Issues related to Wood (iVTH e.V.), is supported in the program of Indus-
triellen Gemeinschaftsforschung (IGF) and financed by the German Federation of Industrial
Research Association (AiF) [24]. Furthermore we would like to thank the Deutschen Institut
fur Bautechnik (DIBt) for its support [25].
References
[1] EN 1998-1:2004 + AC:2009: Eurocode 8: Design of struictures for earthquake
resistance- Part1: General rules, seismic actions and rules for buildings. European Com-
mittee for Standardization (CEN), Brussels (2004)
[2] EN 1995-1-1: Eurocode 5: Design of timber structures - Part 1-1: General - Common
rules and rules for buildings. European Committee for Standardization (CEN), Brussels
(2004)
[3] EN 1991-1-7: Eurocode 1: Actions on structures - Part 1-7: General actions - Acciden-
tal actions. European Committee for Standardization (CEN), Brussels (2006)
[4] Bruhl, F., Kuhlmann, U., Jorissen, A.: Consideration of plasticity within the design of
timber structures due to connection ductility. Structural Engineer 33, S.3007S.3017
(2011)
[5] Jorissen, A.: Double shear timber connections with dowel type fasteners. Delft Univer-
sity of Technology, The Netherlands, Dissertation (1998)
[6] Bejtka, I.: Verstarkungen von Bauteilen aus Holz mit Vollgewindeschrauben. Univer-
sitat Karlsruhe, Lehrstuhl fur Ingenieurholzbau und Baukonstruktionen, Dissertation
(2005)
[7] Schickhofer, G., Augustin, M., Jeitler, G.: Einfuhrung in die Verbindungstechnik
mit Stabdubeln, Schrauben und eingeklebten Stahlstangen. In: 6. Grazer Holzbau-
Fachtagung (2007)
[8] Smith, I., Asiz, A., Snow, M., Chui, I.H.: Possible Canadian/ISO approach to deriv-
ing design values from test data. In: Proceedings of the Meeting No. 39 of Working
Commission W18 - Timber Structures, CIB, Florence, Italy (August 2006)
[9] EN 10025-2: Hot rolled products of structural steels. Technical delivery conditions for
- non-alloy structural steels. European Committee for Standardization (CEN), Bruxelles
[10] EN 26891: Timber structures; Joints made with mechanical fasteners; General princi-
ples for the determination of strength and deformation characteristics. European Com-
mittee for Standardization (CEN), Brussels (1991)
[11] EN 12512: Timber structures - Test methods - Cyclic testing of joints made with me-
chanical fasteners. European Committee for Standardization (CEN), Bruxelles
[12] Munoz, W., Mohammad, M., Salenikovich, A., Quenneville, P.: Need for a harmo-
nized approach for calculations of ductility of timber assemblies. In: Proceedings of the
Meeting No. 41 of Working Commission W18 - Timber Structures, CIB, St. Andrews,
Canada (August 2008)
[13] SIA 265:2012: Holzbau. Schweizerischer Ingenieur- und Architektenverein (2012)
[14] Bruhl, F.: Ductility in timber structures - possibilities and requirements. Institut fur
Konstruktion und Entwurf, Universitat Stuttgart, Dissertation (in preparation)
[15] DIN EN 1995-1-1/NA: National Annex - Design of timber structures - Part 1-1: Gen-
eral rules and rules for buildings. DIN-Deutsches Institut fur Normung e.V. (2010)
[16] EN 1993-1-8: Eurocode 3: Design of steel structures- Part 1-8: Design of joints. Euro-
pean Committee for Standardization (CEN), Brussels (2005)
[17] E DIN EN 14080: Holzbauwerke - Brettschichtholz und Balkenschichtholz - An-
forderungen; deutsche Fassung prEN 14080:2011. DIN-Deutsches Institut fur Normung
e.V. (DIN) (2011)
[18] Bruhl, F., Kuhlmann, U.: Requirements on Ductility in Timber Structures. In: Proceed-
ings of the meeting No. 45 of Working Commission W18 - Timber Structures, CIB,
Vaxjo, Schweden (paper No 45-7-5) (August 2012)
[19] Paulay, T., Priestley, M.J.N.: Seismic design of reinforced concrete and masonary build-
ings. John Wiley & Sons, Inc. (1992)
[20] EN 1990: Eurocode: Basis of structural design. European Committee for Standardiza-
tion (CEN), Brussels (2002)
[21] Kohler, J.: Reliability of timber structures, Swiss Federal Institute of Technology,
Zurich, Dissertation (April 2006)
[22] J OINT C OMMITTEE ON S TRUCTURAL S AFETY (JCSS): Probabilistic Model Code.
Part 3: Resistance Models (3.5 Properties of timber) (2006)
[23] EN 14080: Timber structures - Glued laminated timber and glued solid timber - Re-
quirements. European Committee for Standardization (CEN), Brussels (2011)
[24] Kuhlmann, U., Bruhl, F.: Vorteilhafte Bemessung von Holztragwerken durch duk-
tile, plastische Anschlusse / Institut fur Konstruktion und Entwurf. 2012.
Forschungsvorhaben im Auftrag der iVTH, gefordert duch die AiF, Forschungsbericht
AiF 16184 N (2012)
[25] Kuhlmann, U., Bruhl, F.: Robuste Holztragwerke durch duktile Anschlusse mit
stiftformigen Verbindungsmitteln/Institut fur Konstruktion und Entwurf. 2011.
Forschungsvorhaben im Auftrag des Deutschen Instituts fur Bautechnik DIBt (2011)
An Experimental Study on Bearing Strength
in Compression for Bolted Joint of Plywood
Akiko Ohtsuka1, Sumiya Takahashi1, Takumi Ito1, and Wataru Kambe2
1
Dept. of Archi., Fac. of Eng., Tokyo Univ. of Sci.,
6-3-1, Niijyuku, Katsushika-ku, Tokyo, Japan
ohtsuka.akiko@gmail.com
2
Dept. of Arch. and Env. Design, Kanto Gakuin Univ., Dr. Eng.
1-50-1, Mutsuura-higashi, Kanazawa-ku, Yokohama-city, Kanagawa, Japan
Abstract. In recent years, in the field of wood industry, especially in Japan, it is

required that planed raw materials switch from import materials to domestic
lumber and development of new usage suitable for domestic lumber is strongly
demanded. To promote and increase the demand of domestic cedars, the various
types of products or building members by use of plywood has been suggested in
Japan. We suggested the combined structure which sandwiched a steel material in
plywood. And also, to investigate the structural resistant mechanism and
performance under compression, the loading test of the sandwiched members has
been conducted. Furthermore, we assume that the combined structural system
would apply to sheet lightweight section steel structures, and the examination
about the stiffening effect with the system is done, too. In addition, it is thought
that it is necessary to establish enough clearance between the hole of plywood and
bolt because of the accuracy of finishing of materials and precision of
constructions. On the other hand, it is desirable that the clearance becomes small
as much as possible, because the excessive clearance causes a drop of the structure
unity of the combined material and invites the paths for thermal air environmental
and humidity. Then, the moderate clearance around the joint would be inevitable,
it is necessary to clarify the influence on the stress transfer mechanism and
modification quality by the bearing pressure of the circumference of a bolt joint,
and the above. In this research, the two types of filling up method for the
clearance are studied; the first is a wet construction method by use of filler, and
the other is a dry construction method with steel hardware. Herein, to develop and
propose the above filling up methods, the structural resistant performance is not
only evaluated, the difficulties of production, the costs and the human body effects
are estimated. To select the suited filler materials, the construction examination
and material testing were carried out. Form this selection, an urethane, an epoxy
and a cement were picked up. And the bearing pressure experiment tests with the
parameter of materials, the size of clearance, i.e. the thickness of filler, and the
procuring period were conducted. The next, the steel hardware is designed with
consideration the absorbing the gap resulted from construction and production
DOI: 10.1007/978-94-007-7811-5_18, RILEM 2014
192 A. Ohtsuka et al.
errors. And the bearing pressure experiment study is conducted. From the
experimental result, the relations of rigidity and strength with the parameters of
fillers are clarified. And also, from the comparison of the results of filler and the
ones of steel hardware, it can be said the steel hardware becomes the one of the
most suitable methods.
Keywords: Plywood, Joint, Resistant mechanism.
1 Introduction
Nowadays there is a movement that is transferred to domestic materials from
import one. It is require that the way to use of domestic timber. Especially the
ways of using domestic cedar are positive suggested, in order to plan promotion of
utilization of domestic Japan cedar materials. [1] We proposed the composite
member which sandwiched the steel component with plywood, as shown in Fig. 1,
and the experimental examination about the structural resistant mechanism and
performance under compression were performed. [2]
This panel system is that a steel component is put with plywood, a bolt is
inserted in the hole prepared beforehand, and it joins by hand-clapping for
attaining simplification and the increase in efficiency of construction,
management, etc.. In this time, if it takes into consideration that processing
accuracy of the bolt hole of each component and the accuracy of execution when it
is erected, it is thought that it is necessary to prepare sufficient clearance to the
design intention value of a bolt position, as shown in Fig. 2. On the other hand, he
clearance becomes small as much as possible, because the excessive clearance
causes a drop of the structure unity of the combined material and invites the paths
for thermal air environmental and humidity. In this research, it aims at clarifying
the following two points about the clearance of the circumference of the joints.-1)
Influence which it has on the structure performance and the dynamic mechanism
of a sandwich panel component.2) Influence which it has on the stress transfer and
modification quality by bearing strength in compression for bolted joint of
plywood. Although we are examining 1) separately, in this time it is an
examination experimental about the influence of the clearance given here to.
plywood
bolt-hole
Design objective(Dashed line
Bolt
plywood
hole of plywood
bolt
steel frame
Clearance
Fig. 1 Proposed composite member [1] Fig. 2 Clearance around the hole
An Experimental Study on Bearing Strength in Compression 193
In this paper, as the filling method for clearance, two kinds of methods, the wet
construction method by one filler and a dry type construction method with a steel
hardware, are proposed. And the bearing pressure experiment is conducted to
examine about structural resistant mechanism and performance under
compression.
2 Outline of Clearance and Filling Method
We propose two methods for filling up with clearance, the wet construction
method by use of filler, and the dry type construction method which uses ones of
steel hardware.
2.1 The Wet Construction Method by Use of Filler

It is desirable that a clearance is filled up with a suitable method, since the
excessive clearance of the circumference of a joint may lead to the problem of
reducing the one of steel materials and plywood in sandwich panel structure. For
example, the crevice resulting from the crack may affect safety, usability,
durability, etc. by corrosion etc. Therefore, various kinds of fillers are developed
and put in practical use. Then, these fillers are investigated, and when filled up
with the circumference of the bolt joints of a sandwich panel, the optimal filler is
selected from various viewpoints. The basis of selection of fillers and its process
are as being shown below.
1). Set Up Basis of Selection from a Viewpoint of Safety, Construction, and

Environment, and Select Eight Sorts from 100 Sorts of Materials.
2). Carry Out Construction Examination about Material Selected by 1, and Select
Five Sorts from
3). Carry Out Material Testing about Material Selected by 2), Material
characteristics are evaluated.
2.1.1 The Basis of Selection of Material
The basis of selection is uniquely set up and selected from a viewpoint of safety,
construction, and environment to 100 sorts of materials. The details of a basis of
selection are as follows.
Safety The standard of non-formaldehyde is cleared.
Construction Processes, such as becoming hot at the time of construction, are
not required.
Environment It can construct without using a special tool.
We select also as that of the basis of selection, and decide to carry out a
construction examination to four sorts of epoxy, urethane, emulsion, silicone, and
eight sorts of materials of cement.
2.1.2 The Method and Result of a Construction Examination

Using t = 24 mm of thickness of plywood, a bolt is set to M16-F10T. 17, 19, 23
and 26mm hole are opened in the center of plywood, and a 16-mm bolt is inserted.
It is not filled up about 1 mm of clearance between a point hole and a bolt, but the
clearance 3, 5 and 10 mm are filled up with the filler using the fillers selected
2.1.1. The specimens are cured in the temperature-controlled room (temperature of
20 , 60% of humidity). It is considered as curing period 1, 3, 7, and 28 days, and
prepared one body each. A result is shown in Table. 1.
Table 1 Construction test result
working The days to protect fillers

filler Construction Appearance
life 1 3 7
Epoxy A
Urethane
Emulsion
Epoxy B
Epoxy C
Epoxy D
Silicon
Cement
: good : Defect : Mean value
2.1.3 Method and Result of a Construction Examination
From the experimental result of 2.1.2, we experiment monotonous compression
loading about two sorts of epoxy, urethane, emulsion, and cement In the
temperature-controlled room (temperature of 20 , 60% of humidity), it is cured
in each filler. The curing period was made into 1, 3, 7, 14, and 28 days, and is
prepared every one body each. About the specimen, it should be based on JIS-
K7181 except cement, and it should be based on JIS-A1132 about cement.
The Young's modulus in a material-testing result is shown in Table 2. The
following four points can be said as a feature.
1). Young's Modulus of Epoxy A, Urethane, and Emulsion is near 0.
2). Although Urethane changes greatly, after unloading it is highly rich in
elasticity.
3). Epoxy B shows stable Young's Modulus after the days 7th.
4). As for cement-fulled, the increase in Young's modulus is checked in
connection with days in care-of-health between 3 and 14 days.
Except epoxy, cement, the Young's modulus which have distinguished Young's
modulus were hardly able to be checked based on the above experimental result.
But the characteristic of each material, it was filled up with the filler using the
urethane which is rich in elasticity and returns to the form near a basis after
modification. As a result, its decided to do the next examination.
93
Table 2 Young's modulus in the material 16 15 15 16
.5 6 .5 17 .5 6 .5
-testing result of a filler [N/mm2]
The days to protect fillers

Filler Steel pipe
1 3 7 28 6 t= 6mm
= 60 mm
Steel plate (SS400)
1. Epoxy A t = 6 mm
2. Urethane 7.75 7.59

3. Emulsion Hole of the bolt
= 17 mm
4 Epoxy B 109.1 103.2
5. Cement 1622 902 4464 4483

4
2
6
6
6
Fig. 3 The steel hardware
Table 3 The material-testing result of hardware
Young's Yield Tensile Surrender

Cross-sectional Fracture
modulus strength strength distortion
shape growth [%]
[N/mm2] [N/mm2] [N/mm2] []
Monotonous 2.07105 288 443 38 2112
Circular steel pipe 2.17105 334 512 21 2479
2.2 Wet Construction Method by Use of Steel Hardware

By the dry type construction method using steel hardware, it is expected that the
ease of carrying out, the adjustment of the pressure area under compression are
improved by comparison with the fillers. Then, the method of filling up clearance
with this research with hardware as shown in Fig. 4 is devised. This hardware is
preparing several kinds of positions of the point hole of an inside steel plate, and
has intention of absorbing an error. The steel plate of the outside and an inner side
is set to SS400(from Japanese Industrial Standards :JIS grade), and the steel pipe
is set to STKM13A. The mechanical properties are shown in Table 3.
3 Specification of a Specimen
The bearing pressure experiment with an assumption on the mechanical condition

around the bolt joint by reference of BS EN3833 to is conducted.

Using t= 24 mm of thick of plywood, a bolt is set to M16-F10T. 17, 19, 23 and
26mm hole are opened in the center of plywood shown in Fig5, and 16-mm bolt is
inserted. It is not filled up about 1 mm of clearance between a point hole and a
bolt, but the clearance 3,5 and 10 mm are filled up with the filler. The specimens
are cured in the temperature-controlled room (temperature of 20 , 60% of
humidity). It is considered as curing period 1, 3, 7, and 28 days, and prepared one
body each.
3.2 Dry Construction Method by Use of Steel Hardware

As previous mentioned in 3.1, thick material plywood (t= 24 mm) is used, and a
bolt is set to M16-F10T(from JIS grade. The Scale and layout of test specimens
are shown in Fig4 and 5. The specimen has as show Fig5, the 63mm-hole opened
for steel hardware.
200 24
100 200 24
224 100
112 112
42 Steel Hardware
24
0
0
08 3
06
1 08
Plywood
=17mm,19mm
26mm
Fig. 4 Specimen figue(filling) unit:mm Fig. 5 Specimen figure(Steel Hardware) unit:mm
Load Elastic line

2
P-Max
Pc3
Rigidity 2
Pc2
Pc1
Displacement
Rigidity 1
backlash
Fig. 6 The outline of a bearing pressure experimental result

4 Results of Bearing Pressure Examination
4.1 Load--Relations
The outline of the load displacement relation to the experimental result of this
paper is shown in Fig. 6.

In an experimental result, load displacement relations are shown in Fig.7and it is
presented by a type of filler and curing days. While it is not concerned with curing
days in the case of a urethane-filled and rigidity hardly changed, in the case of the
epoxy -filled or the cement -filled, when care-of-health days progressed, it is
observed that rigidity becomes large. From the Fig.7, the difference of rigidity is
confined whether the test specimen has a filled or not. About the urethane-filled,
in case of urethane, the rigidity is almost same concerning with curing days. On
the other hands, in case of epoxy-filled and cement-filled, the rigidity becomes
large as the case of curing days.
From the results of urethane-filled, the rigidity shows very small, however, the
it is almost same as curing days is progressed of 3-day-curing.
From the result of epoxy-filled after 7-day-curing, it is revealed that the rigidity
becomes large compared with plywood. Between 7 to 28 curing-day, the rigidity is
shown as almost same.
Cement filled it is observed that the rigidity of cement-filled after 3, 7 and 28
curing-days shows almost same with its of plywood only.
16 16 16
14 14 14
12 12 12
Load[kN]
Load[kN]
Load[kN]
10 10 10
8 8 8
6 6 6
4 4 4
2 2 2
0 0 0
0 5 10 15 20 25 0 5 10 15 20 25 0 5 10 15 20 25
Diplacement[mm] Diplacement[mm] Diplacement[mm]
(a)Urethane -filled (b)epoxy -filled (c)cement -filled

explanatory notesa legend :1day :3days :7days :28days no filling
Fig. 7 The load-displacement relation
4.3 In Case of Hardware

The result of load-displacement relation in case of steel-hardware-used is shown
in Fig 8. And the results issue of plywood and cement-filled of 28-curing-day.are
compared in Fig.8, too. From the comparison of Fig.8, proof stress and the rigidity
in case of steel hardware-used are presented about 2-times compared with other
specimens.
35
30
25
Steel Hardware (for one body)
Load[kN]
20
15 No Filling(17mm)
10
5 Cement System 28days(19mm)
0
-5 0 5 10 15 20 25
Diplacement[mm]
Fig. 8 The load-displacement relation of using steel hardware
5 Analytical Examination about Rigidity
From the experimental result of the preceding chapter, it is observed that the
difference of strength, rigidity and behavior are presented as s parameter with the
kind of filler, curing days, etc. That is, it is supposed that the material-
characteristics value of the filler influence to bearing pressure performance. Then,
as shown in Fig. 10, a model is assumed as a series -filled of a filler and plywood.
And an analytical examination is performed about rigidity herein. Rigid body
assumption of the bolt shall be carried out, and it shall be given by the following
formula as a series system of a filler and plywood. That is, the bearing pressure
rigidity K shall be given by rigid K2 of rigid K1 and the filler of plywood by the
following formula.
1
K= (1)
1 K1 + 1 K 2
The rigidity K1 of plywood shall be set to rigid K1 when sinking into plywood
based on "woody structural design standard" 4 in Japanese, shall be devoted fiber
rectangular cross direction k90 with the fibrous direction k0 of the following
formula, and shall be given by rigid average value K1=Ac/(k0+k90)2.
A fibrous direction caves in and it is rigidity: k E1

0 = (2-1)
31.6 + 10.9 d
The fiber rectangular cross direction k

k90 = 0
3.4 (2-2)
caves in and it is rigidity:
Next, the rigidity K2 of a filler is obtained from the following formula.

0

h 2

h 2

( t
h
)
P = 2 E2 t R0 + cos2 + G2 t R0 + sin2 d = E2 + G2 R0 +
h h h
2

t h
K2 = (E 2 + G 2 ) R0 + (3)
h 2
Analytical results by use of Formula (1)-(3) are shown in Fig.9. And also test
results are compared in Fig.9. also.
Rigidity
Rigidity
Rigidity
Days Days Days
(a)Urethane system (b)epoxy -filled (c)cement -filled

Introductory notes: experimental resultSolid line( ), analysis resultDotted line (---)
The size of the hole:19,21,26
Fig. 9 Comparison of the experimental result and analysis result
6 Comparison of Experiment and Analysis
1) In cases of epoxy-filled and a cement-filled, the rigidity becomes large with the
high Youngs modules of the material.
Moreover, in the case which uses the filler with high Youngs modules, rigidity
and proof stress become large, with large clearance.
Of course, the large clearance will expand the contact area between the filler
and plywood. So then the rigidity becomes large if filler with high Youngs
modules is used.
Moreover, when hardware is used, it is guessed that it is easy to secure proof
stress and rigidity also with a repeated load.
2) Since rigidity and proof stress were high as compared with the case which
uses a filler and transfer of the strength to plywood was also large when hardware
was used, it turned out that it excels in the bearing pressure performance.
Moreover, when hardware is used, it is guessed that it is easy to secure proof
stress and rigidity also with a repeated load.
3) From an analysis result, it is thought that the bearing pressure mechanism of
a joint can be expressed with the series -filled of a bolt, a filler, and plywood.
Moreover, a simple analysis model can estimate bearing pressure rigidity in
general in the case which uses the filler.
7 Conclusion
In this research, the method to be filled up using a filler and steel-hardware is

proposed against the clearance around between the plywood and steel members.
The bearing pressure experiment is conducted as parameter with a kind of filler,
curing days, and so on. From the experiment results, the strength, rigidity and
behavior are investigated. Di the influence of the material characteristic of filler
and curing-day are compared and.
From the experiment, the structural characteristic and construction in a wet
construction method and a dry type construction method have been grasped.
Moreover, a analytical model for rigidity is proposed. And from the
comparison, it shows good agreement with test results.
References
1. Ito, T., Kambe, W., Kondo, S., Takahasi, S.: An experimental study on compression
resistant mechanism of sandwiched panel of structural playwood and steel member.
Journal of Structural and Construction Engineering, Journal of Architecture and
Building Science, No. 18-40, 941946 (2011) (in Japanese)
2. BS EN383: Determination of embedment strength and foundation values for dowel type
fasteners (2007)
3. Architectural Inst. of Japan: Quality of wood structure design standard, commentary -
permissible stress degree, permission proof stress design law, p. 233 (December 2006)
Investigations Concerning the Force
Distribution along Axially Loaded Self-tapping
Screws
A. Ringhofer and G. Schickhofer
Institute of Timber Engineering and Wood Technology,

Graz University of Technology, Austria
{andreas.ringhofer,gerhard.schickhofer}@tugraz.at
www.lignum.at
Abstract. Self-tapping screws, as simple fasteners with a high load carrying po-
tential if stressed axially, are frequently applied in timber engineering as tensile
joints in wide span GLT truss systems or as reinforcements against stresses
perpendicular to grain. In fact, force distribution along axially loaded screws has a
very important influence on the joint behaviour. Some models based on Volk-
ersens theory combined with fundamentals of linear elastic fracture mechanics
already exist for glued-in rods or lag screws.
This paper provides a measuring technique estimating the force distribution
based on the determined elongation of the threaded part over the inserted length
by several strain gauges while the composite timber-screw is stressed axially.
Therefore, 16 withdrawal push-pull tests were carried out in solid timber vary-
ing the slenderness , given as the ratio lef/d, from 5 to 20 and with angles of screw
axis to grain direction of 0, 45 and 90. The results are used to verify existing
models of comparable configurations which are further adapted to self-tapping
screws.
Beside the fundamental knowledge of the withdrawal behaviour, also structural
analysis of tensile joints with self-tapping screws can be improved considering the
location of the stress centre and its impact on eccentricities due to the non-linear
distributions, and in regard to recommendations concerning load introduction
perpendicular to grain.
Keywords: Axial loading, Self-tapping screws, Push-pull, Withdrawal capacity,

Force distribution, Strain gauge measurement.
1 Introduction
Modern self-tapping screws are usually produced with tensile strengths of more
than 1,000 N/mm which enables assembling up to lengths of 1,500 mm without
pre-drilling. Steel hardening, necessary to reach these high strength values, led
to an optimisation of axial bearing resistance and consequently to inclined
DOI: 10.1007/978-94-007-7811-5_19, RILEM 2014
202 A. Ringhofer and G. Schickhofer
positioning. Concerning those connections, three failure modes of the single screw
are worth to be mentioned: (i) withdrawal failure, (ii) steel failure in tension and
(iii) pull-through failure of the screw head. While scenarios (ii) and (iii) are easy
to handle or rarely decisive in the design process, the withdrawal behaviour, influ-
enced by several material, environmental and constructive conditions, may be
regarded as the essential research topic concerning the axial bearing resistance of
self-tapping screws. Partially influenced by several scientific activities in the past,
shown e. g. in [2,9,16], European standardisation modified the design process of
self-tapping screws over the last decade which currently results in the determina-
tion of the characteristic withdrawal capacity Fax,,Rk of a screwed connection, in
[N], according to EN 1995-1-1:2009 [7], see
nef f ax , k d lef kd
Fax , , Rk = lef k ,
0.5 0.1
with f ax , k = 0.52 d
0.8
(1)
1.2 cos ( ) + sin ( )
2 2
where fax,k is the characteristic value of the withdrawal strength, in [N/mm],

which depends on the geometrical parameters d as the thread diameter of the
screw and lef as the effective penetration length, both in [mm] on the one side and
on the material parameter k as the characteristic density of the timber product, in
[kg/m] on the other side. Further parameters in this model are nef as the effective
number of fasteners and kd as reduction factor for screw diameters smaller than 8
mm. With regard to these parameters which base on empirically determined
regression models, it is worth mentioning that none of them provide explicit
information about the mechanical behaviour of the axially stressed composite
timber-screw. This model thus enables a simplified application for practical use,
but is not able to represent the real behaviour of the connection analytically which
requires adequate physical parameters. Motivated by that, several investigations in
this topic were carried out in the last decades:
On the one hand, two tests series [1,4] where force distributions of axially
loaded rods were measured by several strain gauges (SG) positioned along the
inserted lengths in timber specimen with numerous varying geometrical parame-
ters are worth to be mentioned. Both showed qualitatively similar and nonlinear
distributions of elongations and consequently lead to nonlinear distributions of
axial force in the rods. They provide an empirical background for further numeri-
cal and analytical investigations.
On the other hand, there are many research activities to be noted, whose scope
was to describe the withdrawal behaviour analytically, mainly of axially loaded
glued-in rods (in steel, glass fibre or carbon fibre), but also of lag or self-tapping
screws (parallel to grain direction), see e. g. [5,11,12,18,19]. All of these men-
tioned contributions principally base on the theory of Volkersen [20] which has
been adjusted considering different bond behaviour (linear-elastic, nonlinear-
elastic, ideal-plastic, etc.) or in combination with maximum or mean stress failure
criterions basing on linear elastic fracture mechanics (LEFM). In contrast to
Eq (1), those models contain mechanical parameters such as fracture energy Gf,
Investigations Concerning the Force Distribution 203
maximum shear stress f as well as elastic modulus E to describe the withdrawal

behaviour.
Considering the fact that force distribution of self-tapping screws has not been
determined extensively by measuring their elongations if loaded axially, the main
scope of the project presented in this paper was to fill this lack of information.
Therefore, the following sections illustrate those experimental investigations
which contained strain gauge measurements in the frame of withdrawal tests on
diameter 12 mm screws under the variation of the parameters penetration length,
lef and angles of screw axis to grain, . Consequently, test results are used to verify
the applicability of the mentioned analytical models for self-tapping screws under
the consideration of other test series which mainly focused on the determination of
geometrical parameters for screwed connections (see [10,15,17]). Finally, recom-
mendations, not only for practical use but also for further investigations in this
research field, are given.
2 Experimental Investigations
2.1 Test Configuration and Procedure

The experimental program for determining the force distribution of axially loaded
self-tapping screws in the frame of withdrawal tests was carried out in three main
steps. The first test series, containing a variation of lef (60, 120, 180 and 240 mm,
two tests for each step) while = 90 has been kept constant, was executed. Every
single test was done as follows:
1. preparation of solid timber (ST) test specimen of Norway spruce (Picea abies);
2. cutting test specimen in the middle (perp. to grain direction) followed by press-
ing together;
3. pre-drilling and inserting of the screw centrally in the saw kerf;
4. removing of screw and application of strain gauges (SG);
5. non-destructive tension test of the screw for calibration of the strain gauges;
6. embedding the screw in the screwed hole and bonding of timber pieces;
7. withdrawal test of the screw (pulsating, cyclic load with nonlinearly increasing
load steps and constant loading rate)
In the next step, tests were carried out with = 0 and = 45 where, based on
first results, lef was only varied from 180 mm to 240 mm. For determination of the
elastic modulus, Ec,0 and the compression strength, fc,0 (both parallel to grain
direction), useful parameters for verification with analytical models, small sam-
ples (40 x 40 x 80 mm) were cut out of each timber specimen and exposed to
compression tests. Additionally, density 12 (for 12 % moisture content) was de-
termined. It should be noted that the height of these small samples deviates from
the regulations for the determination of Ec,0 and fc,0 given in EN 408:2010 [6].
copper wires
Fax
F [N]
strain gauges (SG)

adhesive films
t [sec]
2 SG
20 SG saw kerf
60
60 240 mm
10 SG 24 46 220 250 mm
60
lef
SG
6 SG
130 160 mm
60
6 SG
60
2 SG
Fig. 1 Withdrawal test configuration using strain gauges to determine force distribution
2.2 Test Results and Discussion

In Table 1 mean values of the basic parameters 12, Ec,0, fc,0 as well as the with-
drawal strength fax,corr (density correction according to [3], required for compari-
son of rather inhomogeneous values of 12,mean in dependence of ) reached for the
tests are shown. With regard to the given values, a significant difference of fax,corr
can be observed from = 45 to 0 while results of = 45 and 90 are more or
less equal. This corresponds to conclusions from several investigations done in the
past, see e. g. [16]. Furthermore, similar relations of Ec,0, fc,0 and also fax (for
= 45 and 90) in dependence of 12 can be noted.
Table 1 Basic parameters and results of withdrawal tests
N lef No. SG fax,mean 12,mean fax,corr,mean E0,mean fc,0,mean

[] [-] [mm] [-] [N/mm] [kg/m] [N/mm] [N/mm] [N/mm]
90 8 60240 2446 14.5 401 13.0 9303 35.2
45 4 180240 4046 16.2 448 13.3 12637 42.1
0 4 180240 4046 9.98 403 8.92 9503 36.1
A compilation of the determined force distributions along the screw axis, as the
main objective of these investigations, is shown in Fig. 2 (axes are scaled in [%]).
Therefore, the measured electric values from the strain gauges during the with-
drawal tests had been calibrated for two different load levels considering (i) 30 %
of Fmax as loading rate situated in the linear elastic area of the force-deformation
relationships and (ii) 90 % of Fmax which already is located in the nonlinear part
close to the total failure load. Inconsistent values measured by the strain gauges at
100 % Fmax level made assessment at this point impossible. The data points deter-
mined for each test at these load levels have a certain degree of variation which
depends on different parameters like eccentric load transfer or measuring inaccu-

racies of the strain gauges. Therefore, polynomial trend lines (TL, 2nd order for 1st
load level and 3rd order for 2nd load level data points) were used to describe the
measured force distributions shown in Fig. 2. With regard to these distributions,
some principle facts have to be discussed more in detail:
Firstly, measurement errors occurred during three tests ( = 90, lef = 60 mm
(both) and 240 mm (one)) which had therefore to be rejected for further investiga-
tions. Secondly, all test results given in Fig. 2 indicate clearly a slight nonlinear
force distribution along the penetration length lef depending on the load levels
described before. All determined trend lines at 30 % Fmax have a similar and con-
vex shape with small deviations while those at 90 % Fmax show higher variation in
their course, especially for = 90. Thirdly, it is remarkable that apart from this
higher dispersion for = 90 the influence of different angles from screw axis to
grain direction and also penetration lengths lef on the qualitative course of the
lines appears to be negligible.
100% 100% 100%

= 90 = 45 = 0
75% 75% 75%

% Fmax
% Fmax
% Fmax
50% 50% 50%
25% 25% 25%
0% 0%
0%
0% 25% 50% 75% 100% 0% 25% 50% 75% 100%
0% 25% 50% 75% 100%
% lef % lef % lef
120_30_01 120_90_01
180_30_01 180_90_01 180_30_01 180_90_01
120_30_02 120_90_02
180_30_02 180_90_02 180_30_02 180_90_02
180_30_01 180_90_01
240_30_01 240_90_01 240_30_01 240_90_01
180_30_02 180_90_02
240_30_02 240_90_02 240_30_02 240_90_02 240_30_02 240_90_02
Fig. 2 Distributions of Fax in dependence of : polynomial regression graphs of measured

data
3 Verification with Models Adapted for Self-tapping Screws
Within this section the former discussed force distributions along the screw axis
are compared with the solution of Volkersens differential equation considering
linear-elastic bond behaviour declared in Eq. 2 (according to [18], some parame-
ters are renamed).
sinh( x ) G 1 1
N ax ( x ) = Fax , with =
2
dc + , (2)
sinh( lef ) t ( EA ) st ( EA )w
where Nax(x) is the force distribution along the screw axis, in [N], G the shear
modulus of the shear layer, in [N/mm], t the thickness of the shear layer and dc
the core diameter of the screw, both in [mm]. Consequently, only distributions
determined for forces situated in the linear-elastic part (at 30 % of Fmax) of the
force-deformation relationship are considered in this section. All parameters nec-
essary to describe the behaviour analytically are given in Table 2. In addition,
Fig. 3 shows the assumed specifications of the geometrical parameters which dif-
ferentiate in dependence of .
Table 2 Parameters for the analytical determination of force distribution
lef E00 E90 | E45 a GLR | GLT aw,1 bw,1 aw,2 bw,2 ash bsh t
[] [mm] [N/mm] [N/mm] [N/mm] [mm] [mm] [mm] [mm] [mm] [mm] [mm]
90 120 10118 440 389 42.0 30.0 123 67.5 36 18 22.0
120 9367 407 360 42.0 30.0 123 67.5 36 18 22.0
180 8847 385 340 42.0 30.0 123 67.5 36 18 22.0
180 9245 420 356 42.0 30.0 123 67.5 36 18 22.0
240 8820 383 339 42.0 30.0 123 67.5 36 18 22.0
45 180 12767 1505 491 42.0 30.0 97.5 77.0 36 18 22.0
180 13190 1554 507 42.0 30.0 97.5 77.0 36 18 22.0
240 12144 1431 467 42.0 30.0 97.5 77.0 36 18 22.0
240 12447 1467 479 42.0 30.0 97.5 77.0 36 18 22.0
0 180 9366 - 446 30.0 30.0 118 80.0 18 18 14.5
180 10495 - 500 30.0 30.0 118 80.0 18 18 14.5
240 8791 - 419 30.0 30.0 110 67.5 18 18 14.5
240 9359 - 446 30.0 30.0 110 67.5 18 18 14.5
Screw (steel) parameters: Est = 230,000 N/mm; d = 12 mm; dc = 7 mm
( )
1
a
determined according to [13,14]: E45 = E00 1/ 23 sin3 (45) + 1/ 23 cos3 (45)
On the one hand, material parameters such as the elastic and shear modulus
dont have a large impact on the results if varied in a realistic bandwidth and thus
are determined empirically using E90 = E00/23, GLR = E00/26 and GLT = E00/21
(according to [13]).
On the other hand, geometrical parameters, especially the size of the effective
area of the timber, Aw influences the results in a significant way. Due to this fact,
two different approaches lead to the values of aw and bw given in Table 2. The first
approach (index 1, Aw determined as ellipse) assumes aw and bw to be minimum
distances necessary to reach withdrawal failure of an axially loaded screwed con-
nection. In this case, aw and bw are equal to a1/2 and a2/2 (a1 and a2 as minimum
distances according to [7]) which were determined in several investigations on
self-tapping screws and glued-in rods, e. g. [10,15,17], see
= 90 | 45 : aw = a1 / 2 = 3.5d ; bw = a2 / 2 = 2.5 3.5d

(3)
= 0 : aw = bw = 2.5d
The second approach (index 2, Aw determined as rectangle) assumes aw and bw

to be maximum values reachable and therefore equal to the timber specimen di-
mensions used for the withdrawal tests, see Table 2.
In addition to that, ash and bsh, necessary for the determination of the shear layer
thickness t (for = 90 and 45 an equivalent circular cross-section has been used)
result from observations made when withdrawal failure occurred. Both values are
assumed (according to [8]) to be three times the distance from the screw axis to
the fracture zone which was observed to be 0.5d x 0.5d for = 0 and 1.0d x 0.5d
for = 90 and 45, see
= 90 | 45 : ash = 3d ; bsh = 1.5d = 0 : ash = bsh = 1.5d (4)
= 45|90 = 0
Aw Aw
aw,2
aw,2
Ash Ash
ash aw,1
aw,1
dc
dc
l
ash
t Ast t Ast
bsh bw,1 bsh bw,1
bw,2 bw,2
b b
Fig. 3 Assumed specifications of the geometrical parameters used in the model
Finally, a comparison of the determined force distributions with those predicted

by the model approach according to Eq 2 is shown in Fig. 4 to Fig. 6 in depend-
ence of lef and . The two different possibilities of Aw are referenced by M1 and
M2. With regard to the given results, three main aspects should be discussed
more in detail.
First of all and already mentioned before, the interpretation of the effective area
of the timber, Aw has a high impact on the quality of the prediction. All predicted
force distributions (except those with = 45) with Aw determined by the second
approach (maximum condition) correspond very well to the experimental results
while those where the first approach (minimum condition) was applied show a
distinguished convexity over all, significantly underestimating the measured be-
haviour. In addition to that, the angle of the screw axis to grain direction, , influ-
ences the deviation between experiment and prediction in a major way. Compared
to the other angles, the qualitative shapes of the prediction lines for = 45 devi-
ate more from those of the experimental distributions which may is caused by the
assumed values of the model parameters for this configuration, especially E45, aw
and bw. Furthermore, it can be noted that the difference between test results and
prediction slightly increases with increasing lef. It is worth mentioning that a good
correlation is given up to lef = 240 mm (slenderness = lef/d = 20), which is al-
ready located close to the upper limit where withdrawal failure is barely decisive
for connection design.
8.00 12.00 16.00
= 90 = 90 = 90
6.00 9.00 12.00
Fax [kN]
Fax [kN]
Fax [kN]
4.00 6.00 8.00
2.00 3.00 4.00
0.00 0.00 0.00

0 30 60 90 120 0 45 90 135 180 0 60 120 180 240
lef [mm] lef [mm] lef [mm]
M1 01 M1 01 M1 01 M1 02 M1 02
M2 01 M2 02 M2 01 M2 02 M2 02
TL 01 TL 02 TL 01 TL 02 TL 02
Fig. 4 Comparison of force distributions: test results vs. predictions, = 90
12.00 16.00
= 45 = 45
M1 01
9.00 12.00 M2 01
Fax [kN]
Fax [kN]
M1 02
6.00 8.00 M2 02
TL 01
TL 02
3.00 4.00
0.00 0.00
0 45 90 135 180 0 60 120 180 240
lef [mm] lef [mm]
8.00 12.00
= 0 = 0
M1 01
6.00 9.00 M2 01
Fax [kN]
Fax [kN]
M1 02
4.00 6.00 M2 02
TL 01
TL 02
2.00 3.00
0.00 0.00
0 45 90 135 180 0 60 120 180 240
lef [mm] lef [mm]

4 Conclusion
A measurement technique which enables the determination of elongation and

force distributions along the inserted threaded part of an axially loaded self-
tapping screw has been developed and presented in this paper. In addition, test
results have been used to verify the suitability of Volkersens theory in depend-
ence of essential parameters such as penetration length, lef and angle of screw axis
to grain direction, . With regard to that a good predictive capability of this ana-
lytical model can be reached for = 0 and 90 under certain conditions which
especially concern the geometrical parameters Aw and t. Consequently it can be
stated in principle that this model approach is also applicable for the design of
axially loaded self-tapping screws.
Investigations concentrating on the prediction of the withdrawal behaviour in
the nonlinear part of the force-deformation relationship as well as the considera-
tion of timber products inherent orthogonal characteristics by this model approach
are seen as the next steps to be carried out in this field.
The fact, that the stress centre is located in between the first third of the
threaded penetrated part of an axially loaded self-tapping screw should not only be
considered for the declaration of minimum distances of screwed connections but
also for practical cases of load introduction perp. to grain influencing the position-
ing of the threaded part in the timber product.
Acknowledgement. The authors want to thank all persons involved in the research project
which led to this contribution, especially Gernot Pirnbacher, Thomas Krpfl, Uwe Flp
and Bernd Heissenberger.
References
[1] Bernasconi, A.: Tragverhalten von eingeleimten Gewindestangen. Presentationat 2.
GraHSE 2008 Verbindungstechnik im Ingenieurholzbau. Graz University of Tech-
nology, Graz (January 30, 2008)
[2] Bla, H.J., et al.: Tragfhigkeit von Verbindungen mit selbstbohrenden
Holzschrauben mit Vollgewinde. In: Band 4 der Reihe Karlsruher Berichte zum
Ingenieurholzbau. Universittsverlag Karlsruhe (2006)
[3] CUAP 06.03/08 Common Understanding of Assessment Procedure for European
Technical Approval according to Article 9.2 of the Construction Products Directive
Self tapping screws for use in timber constructions (version December 2010)
[4] Ehlbeck, J., Siebert, W.: Praktikable Einleimmethoden und Wirkungsweise von
Gewindestangen unter Axialbelastung bei bertragung von groen Krften und bei
Aufnahme von Querzugkrften in Biegetrgern. Teil 1: Einleimmethoden,
Meverfahren, Haftspannungsverlauf. Versuchsanstalt fr Stahl, Holz und Steine,
Universitt Karlsruhe, Karlsruhe (1987)
[5] Ellingsb, P., Malo, K.A.: Withdrawal capacity of long self-tapping screws parallel to
grain direction. In: World Conference on Timber Engineering (WCTE), Auckland,
New Zealand (July 2012)
[6] EN 408: 2010-07, Timber structures Structural timber and glued laminated timber
Determination of some physical and mechanical properties
[7] EN 1995-1-1: 2009-07, Eurocode 5: Design of timber structures Part 1-1: General
Common rules and rules for buildings
[8] Foschi, R., Longworth, J.: Analysis and design of griplam nailed connections. J.
Struct. Div. ASCE 101(12), 25372555 (1975)
[9] Frese, M., Bla, H.J.: Models for the calculation of the withdrawal capacity of self-
tapping screws. In: 42nd CIB-W18 Meeting, Duebendorf Switzerland, paper 42-7-3
(2009)
[10] Gehri, E.: Eingeklebte Anker Anforderungen und Umsetzungen. Paper Presented at
Internationales Holzbau-Forum (IHF 2009), Garmisch-Partenkirchen (December
2009)
[11] Gustafsson, P.J., et al.: A strength design equation for glued-in rods. In: International
Symposium on Joints in Timber Structures, Stuttgart. RILEM proceedings, vol. 22
(2001)
[12] Jensen, J.L., et al.: A simple unified model for withdrawal of lag screws and gluedin
rods. Eur. J. Wood Prod. (2010), doi:10.1007/s00107-010-0478-y
[13] Keylwerth, R.: Die anisotrope Elastizitt des Holzes und der Lagenhlzer. VDI-
Forschungsheft 430. Ausgabe B, B and 17 (1951)
[14] Kollmann, F.: Technologie des Holzes und der Holzwerkstoffe, vol. 2. Band 1.
Springer, Gttingen (1951)
[15] Mahlknecht, U.: Widerstand und Bruchverhalten axial beanspruchter
Schraubengruppen in Vollholz und Brettschichtholz. Report. Graz University of
Technology (2013)
[16] Pirnbacher, G., Brandner, R., Schickhofer, G.: Base parameters of self-tapping
screws. In: 42nd CIB-W18 Meeting, Duebendorf Switzerland, paper 42-7-1 (2009)
[17] Plieschounig, S.: Ausziehtragverhalten axial beanspruchter Schraubengruppen. Mas-
ter Thesis, Graz University of Technology (2010)
[18] Prtner, C.: Untersuchungen zum Verbund zwischen eingeklebten stiftfrmigen
faserverstrkten Kunststoffen und Holz. Dissertation, University of Kassel (2005)
[19] Serrano, E., Gustaffson, P.J.: Fracture mechanics in timber engineering Strength
analyses of components and joints. Materials and Structures 40, 8796 (2006)
[20] Volkersen, O.: Die Schubkraftverteilung in Leim-, Niet- und Bolzenverbindugnen.
Energie und Technik, Teil 1-3 (1953)
Experimental Analysis on the Structural
Behaviour of Connections with LVL Made
of Beech Wood
Peter Kobel1, Ren Steiger2, and Andrea Frangi1

1
ETH Zurich, Institute of Structural Engineering IBK, Zurich, Switzerland
kobel@ibk.baug.ethz.ch
2
EMPA, Swiss Federal Laboratories for Materials Science and Technology,
Structural Engineering Research Laboratory, Dubendorf, Switzerland
Abstract. Despite its higher strength and stiffness properties as compared to most
softwood species, beech wood is today almost entirely used for energetic purposes
or non-structural applications. Benefitting from elevated mechanical properties
and a reliable high degree of homogeneity, Laminated Veneer Lumber (LVL)
made of beech has a great potential for applications in high performance structural
elements, for instance in large span truss structures. As the performance of timber
truss structures predominantly depends on the efficiency of the connections, an
experimental analysis of dowel-type connections in beech LVL was performed. A
series of embedment tests was carried out according to EN 383:2007, the parame-
ters being the dowel diameter and the end distance of the connectors. The tests
showed a very ductile behaviour of the material, high values of embedment
strength and a low scatter in the results (CoV < 5%). These findings were
confirmed in subsequent tensile tests on full dowel connections. The tested con-
nections consisted of four dowels and two slotted-in steel plates, the examined
parameters were the dowel diameter and the spacing. Provided that an adequate
spacing is guaranteed, the full connections also showed a very ductile behaviour.
The common problem of premature splitting failure did not occur due to the fa-
vourable effect of the cross-layers in the LVL. It was further found, that the
adequate spacing has to be determined with regard to shear plug failure. The
experimental analysis has confirmed the potential for efficient dowel-type connec-
tions with LVL made of beech.
Keywords: laminated veneer lumber, beech LVL, dowel type connections,

embedment strength, ductility.
1 Introduction
Timber is highly complex due to its anisotropic nature and natural inhomogeneity.
The mechanical and physical properties are influenced by duration of load,
moisture and temperature. As a consequence, the reliability of structural timber
DOI: 10.1007/978-94-007-7811-5_20, RILEM 2014
212 P. Kobel, R. Steiger, and A. Frangi
elements is often inadequate and the full potential of timber in the building and
construction sector has not been exploited yet.
The higher strength and stiffness properties of beech wood as compared to most
softwood species are well known. In Switzerland and other European countries
beech is available in large quantities. However, beech wood is today almost en-
tirely used for energetic purposes or non-structural applications (e.g. in the wood
furniture industry). In Switzerland, for example, almost 60% of the harvested
hardwood is used directly for energetic purposes without adding value to it by
considering other applications. A current research project at ETH Zurich and
Empa aims at developing sustainable innovative and reliable timber structures
using Laminated Veneer Lumber (LVL) made of beech wood. Due to its industri-
alised production, reliable and high strength and stiffness properties, improved
dimensional and form stability, structural elements made of beech LVL have a
great potential to be strong and reliable as steel and sustainable as wood.
With its large ratio between weight and strength timber in general is very suit-
able for large span structures. For large spans, usually truss structures are applied,
as they allow an efficient utilisation of the material due to predominantly axial
forces in the members. However, due to the brittle behaviour of timber, the con-
nections in timber truss structures generally become expansive and complex.
Therefore, the connections are often governing the design of the whole structure
and thus can lead to overdesigned timber members. This constitutes technical as
well as economical limits for conventional truss structures.
In order to improve the performance of timber truss structures, the presented
project focuses on developing more efficient connections by applying LVL made
of beech. In a first step, dowel-type connections were investigated, as this is a very
common type of connection for truss structures.
To provide a basis for the design of dowel-type connections using LVL made
of beech a series of embedment tests was carried out. Based on the results from
these tests the load-carrying capacity of full connections can be calculated accord-
ing to the Johansen theory, taking into account additional design criteria given in
current design codes (e.g. Eurocode 5). Finally, the validity of the current design
criteria for dowel-type connections in LVL made of beech was verified in a pre-
liminary tensile test series with full connections.
2 LVL Made of Beech
As a result of its layered structure, LVL is a very homogeneous material with a

very low scatter in its properties. Compared to softwood, the use of beech veneers
significantly improves the strength values. Furthermore, specific properties can be
enhanced and adjusted by applying cross-layers. Most significantly, the tensile
strength perpendicular to the grain can be increased by a multiple (Table 1). This
is particularly favourable in dowel-type connections, where splitting along the
grain is a major problem.
Experimental Analysis on the Structural Behaviour of Connections 213
The LVL material used in this project was manufactured from 2.5 mm thick ro-
tary-peeled beech veneers. In the manufacturing process the veneers were
compacted to a thickness of about 2.3 mm. The cross-sectional layout (Figure 1)
corresponds in principle to that of Kerto Q (established LVL product made of
softwood, usually Norway spruce), with a proportion of cross-layers of around
23%. The average density of all 38 specimens was 762 kg/m3, with a coefficient of
variation of CoV = 1.1%.
Table 1 Influence of cross-layers on the tensile

strength perpendicular to the grain [1]
Beech LVL
Tensile strength
No cross- Cross-layers
perp. to grain
layers (20-25%)
Fig. 1 Cross-sectional layout of the
ft,90,k [N/mm2] 1.6 24.5
used beech LVL (II-III--III-II)
3 Dowel-Type Connections
Connections play an important role with regard to stiffness and load-carrying ca-
pacity of timber structures and thus have been subject of several research studies.
However, so far the research mainly focused on the development of optimised
connection systems and design rules for softwood, rather than hardwood.
Connections with dowel-type fasteners are very common in timber structures.
Their structural behaviour is very complex as it depends on several parameters
like the geometry (fasteners spacing, edge and end distances) and the mechanical
properties of wood and steel. Usually, dowel-type connections are designed
according to Johansens yield theory [2]. This theory is based on a simplified
approach, approximating the behaviour of both timber and steel as ideally rigid-
plastic. By applying equilibrium conditions, the load-carrying capacities of differ-
ent ductile failure modes can be estimated, depending on the geometry and the
material properties. For steel-to-timber connections with slotted-in steel plates the
following three basic failure modes are relevant [3]:
(1) Fv , Rk = f h , k t d
4 M y , Rk
(2) Fv , Rk = f h , k t d 2 + 1
fh ,k t d
2

(3) Fv , Rk = (1.15 ) 2 M y , Rk f h d
Fig. 2 Basic failure modes acc. to Johansen for dowel-type connections with slotted-in steel
plates. (1) Embedment failure, (2) combined failure, (3) bending failure of the dowel [3].
The smallest value from equations (1) - (3) determines the failure mode and the
capacity per dowel and shear plane. For connections with several shear planes the
capacity per dowel can be derived by determining the failure mode for each shear
plane and then adding up the corresponding capacities.
Johansens yield theory provides the basis for the design of dowel-type connec-
tions. However, in current design codes additional design criteria apply, in order
to cover further failure modes, group effects as well as beneficial effects due to
friction (friction between members, rope effect) which are not considered in
Johansens theory.
Eurocode 5 (EC 5) [4] prescribes a minimum spacing according to Table 2 to
prevent premature brittle failures due to splitting along the grain and plug shear
failure of the timber. To take into account the negative effect of several fasteners
in a row, a reduction factor nef (effective number of fasteners) has to be taken into
account:
n

Fv ,ef , Rk = nef Fv , Rk with nef = min a1 (4)
n 4
0.9
13d
Depending on the failure mode EC 5 also allows for beneficial effects related to
friction [3]. In failure mode (3) the deformation of the dowel implicates a lateral
contact force between the timber and the steel plate, which in turn leads to friction
between these elements. For this case EC 5 suggests a friction factor of 15%, as
included in equation (3).
Generally, EC 5 also allows for a rope effect. However, as for dowels the with-
drawal resistance is solely dependent on friction between dowel and timber, this
effect is neglected in the design of connections with dowels.
As the mentioned design rules have been defined mainly for softwood solid tim-
ber and glulam, the relevant parameters still have to be determined for beech LVL.
Table 2 Spacing according to EC 5 for dowel-type

connections
Spacing and Minimum spacing or

Angle
edge/end distance edge/end distance
a1 0 360 (3 + 2|cos|)d
a2 0 360 3d
a3,t -90 90 max(7d; 80mm)
a3,t 90 150 max(a3,t |sin|)d; 3d)
150 210 3d
210 270 max(a3,t |sin|)d; 3d)
max((2 + 2sin)d;
a4,t 0 180
3d)
Fig. 3 Spacing according to EC 5
a4,c 180 360 3d
for dowel-type connections
4 Embedment Tests
4.1 Specimens and Test Setup

To determine the embedment strength for LVL made of beech a series of tensile
embedment tests was carried out. A total of 30 specimens (Table 3) were tested
according to the test standard EN 383:2007 [5], each specimen being equipped
with two dowels (Figure 4). The tests were carried out with four different dowel
diameters. The geometrical properties of the specimens B, C, E and F correspond
to the requirements specified in EN 383. In all specimens of groups D and G, the
prescribed end distance l3 was reduced from 7d to half the value (3.5d) to investi-
gate the effect of the end distance on the embedment strength.
High quality steel dowels with a tensile strength of 960 - 1,100 N/mm2 were
used, in order to prevent a failure of the dowel. The holes for the dowels were pre-
drilled with an equal nominal diameter as the respective dowel.
The tests were run on a displacement controlled servo-hydraulic universal test-
ing machine. The loading procedure included one reloading cycle between 10%
and 40% of the estimated maximum load. During testing the applied load was
recorded, as well as the relative displacement between the dowel and the timber.
Table 3 Specimens for embedment tests
Dowel End No. of

Dimensions
diameter distance specimens
Configuration
L B t
d [mm] l3 [-] n [-]
[mm] [mm] [mm]
B 20 7d 5 880 120 49
C 16 7d 4 704 96 49
D 16 3.5d 5 592 96 49
E 12 7d 5 528 72 21 Fig. 4 Tensile embed-
F 8 7d 5 352 48 21 ment test acc. to EN 383:
G 8 3.5d 5 296 48 21 a1 = 3d; l3 = 7d; l4 = 40d
4.2 Results
According to EN 383 the embedment strength is derived from either the maximum
load or the load at a displacement of 5 mm, whichever occurs first:
{F=5 mm ; Fmax }
fh = (5)
d t
For specimens with a full end distance according to EN 383 the load at a dis-
placement of 5 mm always became decisive. Figure 5 shows the load-
displacement behaviour of all conducted embedment tests.
120 100
Embedment strength fh [N/mm2]

B Test results
100 80
80
Load F [kN]
C 60
Beech
60
D 40
40 Spruce
E 20
20
F
G
0 0
0 5 10 20 30 40 50 0 6 8 12 16 20 30
Relative displacement [mm] Dowel diameter d [mm]
Fig. 5 Load-displacement behaviour of all Fig. 6 Comparison of test results. Acc. to

conducted embedment tests (for notation of EC 5 design rules: fh,m=0.082(1-0.01)m with
groups see Table 3). Limit displacement m(spruce)=470kg/m3 , m(beech)=760kg/m3.
acc. to EN383: =5mm. : l3=7d; x: l3=3.5d.
All configurations with the full recommended end distance of 7d (i.e. B, C, E,

F) showed a very pronounced plateau after an initial linear elastic phase, and the
maximum embedment stresses fh,max occurred at very large displacements and
were significantly higher than the values fh,=5mm derived according to EN 383.
The tests were continued until failure occurred on one side of the specimen.
Hence, fh,max was only reached for one out of two dowels. In Table 4 both these
values are listed for each dowel diameter together with the coefficients of varia-
tion (CoV). For the values derived according to EN 383 the average CoV was
below 5%.
Table 4 Embedment test results
fh,=5mm (EN 383) fh,max

Configuration n fh,=5mm CoV n fh,max CoV (Fmax)
[-] [N/mm2] [%] [-] [N/mm2] [%] [mm]
B (d=20mm) 10 75 5.1 4 104 2.2 39
C (d=16mm) 8 77 3.6 4 100 5.0 21
E (d=12mm) 10 84 3.4 4 101 4.7 29
F (d=8mm) 10 87 6.7 5 101 4.8 16
Table 5 shows the influence of a reduced end distance. In Figure 5 it can be

seen that the initial slopes of the curves are very similar for specimens with half
and full end distances. However, for specimens of groups D and G with only half
the suggested end distance, failure in the form of a shear plug occurred at a much
lower displacement.
Table 5 Influence of the end distance l3 on the embedment strength
d = 16 mm d = 8 mm
fh,=5mm (EN 383) fh,=5mm (EN 383)
Configu- Configu-
l3 n fh,=5mm CoV l3 n fh,=5mm CoV
ration ration
[-] [-] [N/mm2] [%] [-] [-] [N/mm2] [%]
C 7d 8 77 3.6 F 7d 10 87 6.7
D 3.5d 7 72 3.9 G 3.5d 5 81 2.6
Relative loss: -6.5% Relative loss: -7.3%
In Figure 6 the results are compared with the expected values according to the
design rules in EC 5 for solid timber and glulam made of Norway spruce and
beech. The obtained values for beech LVL are significantly higher, not only
compared to solid spruce timber, but also compared to solid beech timber,
which confirms the beneficial effect of the cross-layers. The influence of a re-
duced end distance resulted in a loss in embedment strength of around 7% (for
d = 8 / 16 mm).
5 Full Connection Tensile Tests
5.1 Specimens and Test Setup

To investigate the behaviour of full dowel-type connections with LVL made of
beech a series of tensile tests was carried out. A total of eight specimens were
tested, each specimen being equipped with two connections, and each connection
consisting of four dowels and two slotted-in steel plates (Figure 7). All steel parts
were of the quality S355. The examined parameters were the dowel diameter and
the spacing (Table 6). The dowel diameter was varied in order to study the behav-
iour for rigid dowels (d = 20 mm) and slender dowels (d = 8 mm). Spacing
according to EC 5 was applied, as well as a reduced spacing with only half the
distances, to investigate the influence of the spacing.
Table 6 Configurations of the tested connec-

tions. Specimen dimensions: L=1,275mm,
B=240mm, t=100mm.
Dowel
Configu- Spacing
diameter
ration
d [mm] a3 a1 a4
20F 20 7d 5d 3d
20H 20 3.5d 2.5d 3d
8F 8 7d 5d 3d
8H 8 3.5d 2.5d 3d Fig. 7 Schematic of the tested connections
The tensile tests were carried out according to EN 1380:2009 [6] and
EN 26891:1991 [7] on a displacement controlled servo-hydraulic universal testing
machine. The loading procedure included one reloading cycle between 10% and
40% of the estimated maximum load. The tests were continued until failure oc-
curred on one side of the specimen. During testing the applied load was recorded,
as well as the relative displacement in the connection between the steel plates and
the timber.
5.2 Results
Figure 8 shows the load-displacement behaviour of all conducted tests. The ob-
tained maximum loads and displacements including the observed failure modes
are listed in Table 7.
500
20F
400
Load F [kN]
300
20H
200
8F
100 8H
0
0 5 10 15 20 25
Relative displacement [mm]
Fig. 8 Load-displacement behaviour of the Fig. 9 Failure modes for d=20mm. Left: full
tested connections (notation as in Table 6) spacing; right: half spacing
The tests have shown a very ductile behaviour of connections with a spacing of
the dowels according to EC 5. In configuration 8F the ductility resulted mainly
from the bending deformations of the dowels. In configuration 20F, however, the
dowels remained rigid during the whole test, which means in this case the ductility
was provided by the ductile embedment failure in the LVL.
Table 7 Test results
Configuration n [-] Fmax [kN] [mm] Failure mode

20F 2 523 13.4 Embedment failure
20H 2 346 3.4 Shear plug failure
8F 2 203 15* Dowel failure
8H 2 137 5.4 Dowel and shear plug failure
*Limit displacement acc. to EN 26 891.
Configurations with only half the spacing failed prematurely due to shear plug
failure. Consequently, lower bearing capacities and a significantly reduced ductil-
ity were observed. Configuration 20H showed a quite brittle behaviour, whereas in
configuration 8H the bending failure of the dowels maintained a certain degree of
ductility. No splitting failures were observed in any of the specimens.
6 Conclusions
The presented experimental investigation was carried out to evaluate the potential
of dowel-type connections with LVL made of beech. The tests have shown that
the favourable material properties of beech LVL are well reflected in the perform-
ance of the connections:
The mean embedment strength values obtained in beech LVL are significantly
higher compared to corresponding values for solid Norway spruce and beech tim-
ber. This confirms the advantage of beech compared to Norway spruce (as the
embedment strength is a function of the timber density) and the beneficial effect
of the cross-layers.
Given an adequate spacing, a very ductile behaviour was observed in the em-
bedment tests, along with a very low scatter (CoV < 5%).
These findings were reflected in tensile tests with full dowel connections. The
results have shown that a notable ductility can be provided by the beech LVL
material, as even connections with rigid dowels (d = 20 mm) showed a ductile
overall behaviour. Furthermore, the problem of premature splitting failure could
be eliminated by the cross-layers. The variation of the dowel spacing yielded that
the adequate spacing has to be determined with regard to shear plug failure.
The observed ductile behaviour leads to the assumption that the negative group
effect should be small or even negligible (nef 1). However, further experimental
and numerical analysis has to be conducted in order to validate this hypothesis.
The experimental investigation has shown the potential for efficient connections
in LVL made of beech. This, along with the high strength and stiffness values, con-
firms the suitability of beech LVL for high performance timber structures, such as
large span trusses.
References
1. Van de Kuilen, J.-W., Knorz, M.: Prfbericht Nr. 10511: Ergebnisse der
Zulassungsversuche fr eine, allgemeine bauaufsichtliche Zulassung (abZ) von
Furnierschichtholz aus Buche. Technische Universitt Mnchen (TUM), Germany
(2012)
2. Johansen, K.W.: Theory of Timber Connections, pp. 5669. IABSE, Publication No. 9,
Bern, Switzerland (1949)
3. Porteous, J., Kermani, A.: Structural Timber Design to Eurocode 5. Wiley-Blackwell
(2007)
4. DIN EN 1995-1-1:2010-12. Eurocode 5: Design of timber structures - Part 1-1: General

- Common rules and rules for buildings. DIN German Institute for Standardization
5. DIN EN 383:2007. Timber Structures - Test methods - Determination of embedment
strength and foundation values for dowel type fasteners. DIN German Institute for
Standardization
6. DIN EN 1380:2009. Timber structures - Test methods - Load bearing nails, screws,
dowels and bolts. DIN German Institute for Standardization
7. EN 26 891:1991. Timber structures; joints made with mechanical fasten-ers; general
principles for the determination of strength and deformation characteristics (ISO
6891:1983). CEN European Committee for Standardization
The Embedment Failure of European Beech
Compared to Spruce Wood and Standards
Steffen Franke and Nolie Magnire
Bern University of Applied Sciences, Architecture, Wood and Civil Engineering,

Switzerland
{steffen.franke,noelie.magniere}@bfh.ch
Abstract. The embedment behaviour of European hardwoods in dowel-type

connections is investigated through the analysis of tests performed on beech wood.
Experimental results are evaluated using the 5% offset method. They are com-
pared with estimations provided by the Eurocode 5 and the SIA 265 and with
similar tests performed on spruce. Influences of the load-to-grain angle and the
fastener diameter d are investigated. Beech has a stronger hardening behaviour
than spruce, and is very influenced by the fastener diameter. The comparison with
the standards shows a general overestimation (at least 20%) of the experimental
results, especially for loading perpendicular to the grain. The model provided by
Eurocode 5 to consider the load-to-grain angle is better than the SIA 265 but is
optimized for softwoods only. Thus a formula optimized for both species is
proposed.
Keywords: hardwood, beech solid wood, Fagus sylvatica, embedment strength,

load-to-grain angle, dowel diameter.
1 Introduction
This paper focuses on the behaviour of dowel-type connections which are widely
used in the building industry. It is thus crucial to estimate accurately the stiffness
and ultimate strength of such connections. The European Yield Model (EYM)
proposed by Johansen in 1949 to design dowel connections is today commonly
accepted. In this model, the overall behaviour of the connection has two compo-
nents: the fastener yield capacity and the embedment strength which directly de-
pends on the wood species. Indeed the embedment strength is defined in the norm
ISO/DIS 10984-2:2009 as the average compressive stress at maximum load in a
piece of timber or wood-based product under the action of a stiff linear fastener.
To achieve smaller connections for higher loadings, European hardwoods are
increasingly used in timber construction. But the embedment strength design for-
mula provided by the EN 1995-1-1:2005 (EC 5) has been obtained experimentally
through many tests performed on softwoods and tropical hardwood species mainly
DOI: 10.1007/978-94-007-7811-5_21, RILEM 2014
222 S. Franke and N. Magnire
(Leijten et al. 2004). Whereas only 75 tests were conducted for European hard-
woods species (Hbner 2008). This results in a large uncertainty regarding the
accuracy of European hardwoods embedment strength values provided by the
standards. Consequently our current standards must be reviewed to ensure
the same reliability both for hardwoods and softwoods designs. Thus this paper
aims at evaluating the embedment behaviour of the most common European
hardwood specie in Switzerland: beech (Fagus Sylvatica).
To achieve this goal, 160 embedment tests were conducted on beech speci-
mens. Both, the EC 5 and the Swiss standard SIA 265:2012 (SIA 265) describe the
embedment strength as a function of:
The wood species (softwood or hardwood)
The wood density [kg/m ]
3

The fastener diameter d [mm]
The load-to-grain angle []
Four dowel diameters d (6, 12, 20 and 30 mm) and four load-to-grain angles
(0, 25, 55 and 90) were tested. Similar tests with the same set of parameters were
carried out on common softwood specie: spruce (Picea Abies). 10 specimens were
tested for each combination of parameters to enables a statistical analysis of the
data obtained. The results of both tests series were compared to each other and to
the estimation obtained using the SIA 265 and the EC 5 formulas.
2.1 Testing Standards

Embedment tests can be carried out either using the half-hole design (described in
the ASTM D 5764-97a:1997 or in the ISO/DIS 10984-2) or the full-hole design
(described in ASTM, ISO/DIS or EN 383:1993). The half-hole design enables to
obtain a result free from any influence of the fasteners bending and is suitable for
hardwoods such as beech. It is thus the design which was chosen for this study
(see Fig. 1). According to the ISO/DIS 10984-2, the tests were carried out with a
1
LVDT 2
3
1 - Steel, 2 - Fastener, 3 - Specimen

Fig. 1 Half hole test set up in ASTM and experiment
The Embedment Failure of European Beech 223
constant speed of 1 mm/min to reach the maximum load in 300 s 120 s. The
tests were stopped when either failure or 8 mm displacement was reached. A pre-
loading cycle between 0.4 Fmax and 0.1 Fmax was performed for each test.
2.2 Beech and Spruce Wood Specimens

Beech and spruce specimens were cut from boards of a constant thickness of
40 mm. Their width and height were obtained from a linear interpolation of the
values provided by the ISO/DIS 10984-2 (see Table 1). The specimens were con-
ditioned at 20 C and 65% of relative humidity until mass consistency was
reached with a mean moisture content of 13.3%. The density distribution of each
series was narrow (see Fig. 2) with, in average, 734 kg/m3 (COV 1.8%) for beech
and 447 kg/m3 (COV 4.4%) for spruce.
Table 1 Sizes of beech and spruce specimens
Width/Height/Thickness [mm]
Load-to-grain angle d = 6 mm d = 12 mm d = 20 mm d = 30 mm
= 0 36/70/40 70/72/40 120/100/40 180/150/40
= 25 45/70/40 85/70/40 143/80/40 180/85/40
= 55 50/70/40 100/70/40 170/85/40 242/75/40
= 90 60/70/40 120/70/40 200/70/40 300/70/40
850
Load F [kN]
800 Beech
750 F5 mm
700
Density [kg/m3]
F5 %
650
600 d=6 mm d=12 mm d=20 mm d=30 mm
d=6 mm d=12 mm d=20 mm d=30 mm
550
500
450 K
400 1
Spruce 0.1 F5 mm
350
0 40 80 120 160 u0 0.05d-offset 5 mm
Sample number [-] Displacement u [mm]
Fig. 2 Density of beech (filled) and spruce Fig. 3 Evaluation methods
2.3 Evaluation of the Results

The evaluation of embedment tests is a point of discussion among experts as various
methods have been used in the past (see for instance in Leijten et al. (2004)). This
led to incompatibility in experimental results. In this study, the curves were evalu-
ated following the 5% offset method to obtain the yield load Fyield. The maximum
load was either the ultimate load or the load at 5 mm displacement, according to
EN 383 and ISO/DIS 10984-2 respectively. Furthermore, the stiffness K was
determined. In the analysis, the yield load was used to calculate the embedment
strength of each test following this formula:
Fyield
fh = (1)
d t
Where t = thickness of the wood specimen [mm]
The load-displacement curves were described and evaluated as well. A
summary of the information extracted from the experimental curves is provided in
Fig. 3.
2.4 Design Standards

The test results were confronted with the values obtained from the EC 5 and the
SIA 265. Both standards predict the embedment strength at an angle based on
the values parallel and perpendicular to the grain, fh,0 and fh,90 respectively. In the
EC 5, the interpolation between fh,0 and fh,90 is performed using a trigonometric
model: the Hankinson formula:
f h,0
f h, = (2)
k90 sin + cos 2
2
Where:
f h,0 = 0.082 (1 0.01d ) k (3)
1.35 + 0.015d for softwoods

k90 = (4)
0.90 + 0.015d for hardwoods
In the SIA 265 however, a linear interpolation between fh,0 and fh,90 is provided:
f h, = ( 9000
5.5
+ 0.15) d 0.3 for softwoods (5)
f h, = ( 9000
2
+ 0.19 ) d 0.3 for hardwoods (6)
For both standards, is the wood density [kg/m3], d is the fastener diameter [mm]
and is the load-to-grain angle [].
3 Results
3.1 Embedment Behaviour

For both series, the load-displacement curves show a strong relation of the em-
bedment behaviour with the angle . Fig. 4 provides the mean curves of the
embedment tests performed on beech with a dowel of 20 mm in diameter. They

are representative for the other series as well. For load parallel to the grain ( = 0),
the samples present a quasi-ideal elastic/plastic behaviour with a high slip
modulus followed by a constant load carrying value until failure in splitting along
the plane of a growth ring. When increases a reduction of the slip modulus and
the embedment strength just after yield appears. This corresponds to a hardening
of the wood due to the crushing of fibres under the dowel. The hardening behav-
iour is already present for 25 and 55 angle to the grain and clearly appears at 90
angle. Such behaviour agrees with results found in the literature both for soft-
woods (for instance Franke and Quenneville (2011) or Sawata and Yasumura
(2002)) and hardwoods (Hbner 2008).
Beech, d = 20 mm - mean curves

45
40
35
30
Load [kN]
25
20
15
10
5 = 25 = 55
= 90 = 0
0
0 1 2 3 4 5 6 7 8
Displacement u [mm]
Fig. 4 Typical load-displacement curves for various angles
3.2 Embedment Strength Results

Beech experimental results are given in Table 2. The stiffness K significantly de-
creases when increases (60% to 85%). The yield strength values fh,,5% obtained
were compared with the maximal strength fh,,max. Large differences between fh,,5%
and fh,,max would mean the evaluation method chosen to calculate the yield load
has a strong influence on the final result. Table 2 shows that, in average, fh,max is
26% higher than fh,,5% for beech whereas it was only 17% higher for spruce.
Therefore beech embedment tests are even more sensitive to evaluation method
than spruce ones. An agreement of the experts to one evaluation method is thus a
necessity for further research about hardwood embedment strength.
3.3 Comparison with the Standards Estimation

The characteristics of each specimen (, d and ) were used to calculate its em-
bedment strength according to the EC 5 and the SIA 265, fh,,EC5 and fh,,SIA respec-
tively. They were then compared with the experimental strength found (see Fig. 5
and Table 2). Generally, the embedment strength is overestimated. For both spruce
and beech, only series with = 0 provided the experimental values higher than the
expected ones. The most accurate estimation is for the beech series using the EC 5
formulas. For this group the mean overestimation is only of 18% whereas it ranges
between 50% and 60% for the three other groups. When looking at the evolution
with the angle , the EC 5 provides slightly better results than the SIA 265, espe-
cially for beech. It is important to note that fh,90 is overestimated by least 20% for
both standards and all diameters. fh,90 is then used as a basis to calculate
embedment strength at intermediate angles. Therefore there is a need to review the
calculation of fh,90 to obtain more accurate values. Otherwise the inaccuracy of
fh,90 estimation impacts the estimation of any embedment strength calculated for
0.
Table 2 Embedment strength for beech

Yield strength Max. strength
f h, ,max f h, , EC 5 f h , , SIA
Group Stiffness fh,,5% [MPa] fh,,max [MPa]
HE-d K N/mm] Mean COV Mean COV f h, ,5% f h, ,5% f h, ,5%
HE0-6 46271 70.4 5% 76.9 4% 1.09 0.81 1.16
HE25-6 22703 42.3 3% 62.6 1% 1.48 1.30 1.81
HE55-6 15698 40.0 5% 68.2 6% 1.71 1.45 1.94
HE90-6 16724 45.1 8% 71.0 12% 1.57 1.22 1.56
HE0-12 65751 59.0 5% 62.5 3% 1.06 0.92 1.15
HE25-12 39812 37.1 4% 46.9 3% 1.27 1.37 1.68
HE55-12 32713 33.7 6% 47.7 4% 1.42 1.49 1.84
HE90-12 22107 36.6 1% 55.3 5% 1.51 1.28 1.54
HE0-20 76953 48.5 6% 49.1 5% 1.01 0.96 1.13
HE25-20 56420 41.3 3% 46.4 2% 1.12 1.13 1.34
HE55-20 35286 32.7 7% 39.6 5% 1.21 1.34 1.68
HE90-20 32285 34.0 7% 45.0 6% 1.32 1.22 1.55
HE0-30 116358 50.0 11% 51.0 9% 1.02 0.86 1.02
HE25-30 72105 36.5 4% 39.1 4% 1.07 1.11 1.36
HE55-30 38863 27.3 4% 31.9 3% 1.17 1.31 1.80
HE90-30 30572 25.8 4% 29.6 4% 1.15 1.18 1.71
Average 5% 5% 1.26 1.18 1.52
Beech Spruce
85
fh,,5% - Standard value [MPa ]
40
fh,,5% - Standard value [MPa ]
75 35
65 30
55 25
EC 5 SIA 265 EC 5 SIA 265
45 20
= 0 = 0
= 25 = 25
35 = 55 15 = 55
= 90 = 90
10
25
25 35 45 55 65 75 85
fh,,5% - Experimental value [MPa ]
fh,,5% - Experimental value [MPa ]
Fig. 5 Standard vs experimental embedment strength for the EC 5 (filled) and the SIA 265
3.4 Influence of the Dowel Diameter

The average value of embedment strength for each group of dowel diameters is
shown in Fig. 6. The whiskers represent the coefficient of variation of the group.
Two different trends appear: beech series present a clear negative correlation be-
tween strength and fastener diameter, spruce series however, do not show an obvi-
ous trend with relatively constant strength.
Recent studies on softwoods such as Franke and Quenneville (2011) or Sawata
and Yasumura (2002) found fh vs d relatively constant. Whereas Hbner, on his
tests carried out on ash (2008), found a negative correlation between fh and d.
These studies agree with the trend of both series. However, for beech, a decrease
of the trend with increasing angles is observed which contradicts the results of
Hbner (2008). The models used to consider the evolution of fh, with d are
common to softwoods and hardwoods in both the SIA 265 and the EC 5. Both
standards result in a decrease of fh, with d which corresponds better to hardwoods
behaviour than softwoods behaviour.
Beech Spruce
75 50
Embedment strenght fh,,5% [MPa ]
= 0 = 25 = 0 = 25
65 = 55 = 90 = 55 = 90
40
55
30
45
20
35
25 10
15 0
5 10 15 20 25 30 35 5 10 15 20 25 30 35
Dowel diameter d [mm] Dowel diameter d [mm]
Fig. 6 Evolution of the embedment strength with the dowel diameter
3.5 Influence of the Load-to-Grain Angle

The embedment strength decreases when increases (see Fig. 7). The reduction
between fh,0 and fh,90 ranges from 30% to 65% with a stronger tendency for spruce
series. The experimental curves in Fig. 7 can be split in two parts. In the left part,
(up to 25 for beech and 55 for spruce) most of the reduction occurs. In the
right part, the strength reduction is less significant with even a slight increase for
beech.
Interpolations of fh,0 and fh,90 experimental values were performed to obtain
values for fh,25 and fh,55. The interpolation models used were linear, as in the
SIA 265 and the Hankinson formula, as in the EC 5. Both models provide curves
less convex than the experimental ones and thus overestimate fh at intermediate
angles.
Beech Exp. Hankinson Spruce Exp. Hankinson

35

d = 6 mm d = 6 mm
70 d = 12 mm d = 12 mm
d = 20 mm 30 d = 20 mm
60 d = 30 mm d = 30 mm
25
50
20
40 15
30 10
20 5
0 20 40 60 80 0 20 40 60 80
Load to grain angle [] Load to grain angle []
Fig. 7 Embedment strength fh,5% vs (full lines) and comparison with the adapted
Hankinson model (dotted lines)
EC 5 model is the most accurate. However with only a correlation R2 of 0.72

for beech compared to 0.92 for spruce, it is optimised for softwoods. An optimiza-
tion of the trigonometric coefficients of the formula was performed. Formula (7)
was obtained which describes accurately the behaviour of beech (R2 =0.95) and
spruce (R2 = 0.94). We can thus hypothesize that it is possible to use a common
model for hardwoods and softwoods to describe the reduction of the embedment
strength with . This model could be based on the approach provided by Hankin-
sons formula. However still further research is necessary to optimize the trigo-
nometric coefficients to more hardwoods and softwoods species.
fh,0
fh, = (7)
k90 sin1.2 + cos1.5
4 Conclusion
Tests of the embedment strength with various load-to-grain angles and dowel
diameters were performed on beech specimens. The results of these tests com-
pared to the ones of identical tests carried out on spruce specimens show that both
species present different embedment behaviours. These differences can be as-
sumed to be representative of the ones between softwoods and European hard-
woods in general. The values obtain from the Swiss SIA 265:2012 and European
EN 1995-1-1:2004 show that a review of these standards is necessary to obtain
values also representative for European hardwoods. But they also show that the
embedment strength values are currently overestimated even for softwoods when
the load-to-grain angle is different from 0. A better estimation of the embedment
strength perpendicular to the grain is needed. A reduction of at least 20% of cur-
rent estimated values would be necessary. This study also shows that a model
adapted from the one of the Eurocode 5 describes accurately the influence of the
load-to-grain angle on the embedment strength for both beech and spruce wood.
References
ASTM International, ASTM D 5764-97a Standard Test Method for Evaluating Dowel-
Bearing Strength of Wood and Wood-Base Products (1997)
European Committee for Standardization, EN 383 Test Methods - Determination of Em-
bedment Strength and Foundation Values for Dowel Type Fasteners (1993)
European Committee for Standardization, EN 1995-1-1 Eurocode 5 Design of Timber
Structures, Part 1-1: General Common Rules for Buildings (2004)
Franke, S., Quenneville, P.: Bolted and Dowelled Connections in Radiata Pine and
Laminted Veneer Lumber Using the European Yield Model. Australian Journal of Struc-
tural Engineering 12(1), 13 (2011)
Hbner, U.: Embedding Strength of European Hardwoods. In: CIB-W18, Paper 4175, St.
Andrews, Canada (2008)
International Organization for Standardization, DRAFT ISO/DIS 10984-2 Timber Struc-
tures - Dowel-type Fasteners Part2: Determination of Embedding Strength and Foun-
dation Values (2009)
Johansen, K.: Theory of Timber Connections. International Association for Bridge and
Structural Engineering Publications 9, 249262 (1949)
Leijten, A., Khler, J., Jorissen, A.: Review of Probability Data for Timber Connections
with Dowletype Fasteners. In: CIB-W18, Paper 37713, Edinburgh, UK (2004)
Sawata, K., Yasumura, M.: Determination of Embedding Strength of Wood for Dowel-type
Fasteners. Journal of Wood Science 48(2) (2002)
Swiss Society of Engineers and Architects, SIA 265 Timber Structures, Zrich, Swiss
(2012)
Modelling of Non-metallic Timber Connections
at Elevated Temperatures
Daniel Brandon1, Martin P. Ansell2, Richard Harris1, Pete Walker1,

and Julie Bregulla1,3
1
BRE CICM, Dept. Civil Eng. & Architecture, University of Bath, UK
{D.Brandon,R.Harris,P.Walker}@bath.ac.uk
2
BRE CICM, Dept. Mechanical Eng., University of Bath, UK
M.P.Ansell@bath.ac.uk
3
BRE, Watford, UK
BregullaJ@bre.co.uk
Abstract. Models estimating the slip modulus and the load capacity, including
temperature dependent effects, of non-metallic timber connections are presented.
Previous studies, including the work of Thomson, have shown that Glass Fibre
Reinforced Polymer (GFRP) rods are suitable connectors for timber and that
Densified Veneer Wood (DVW) functions effectively as a flitch plate material.
Thomsons model predicting the slip modulus of the connection with GFRP rods
and DVW plates is revised in this paper. The revised model is used to predict the
slip modulus and the failure load at room temperature and elevated temperatures.
The latter is achieved by predicting local temperatures in the connection and tak-
ing corresponding reduced material properties into account.
Keywords: non-metallic timber connections, pultruded GFRP dowels, densified

veneer wood, embedment strength, connection slip.
1 Introduction
Previous studies have shown that Glass Fibre Reinforced Polymer (GFRP) rods
are suitable shear connectors for timber (Pederson 2002; Drake, 2003; Thomson,
2010) and that Densified Veneer Wood (DVW) functions effectively as a flitch
plate material (Thomson, 2010). A section of the connection type considered in
the study of Thomson is shown in Fig.1. Thomson et al. (2010) state that the
advantages of this connection type, compared to metallic connections, are higher
corrosion resistance and increased fire safety. The latter is, however, not yet con-
firmed. Therefore, a model able to predict the behaviour of the connection at ele-
vated temperatures, as proposed in this paper, is required. This model is part of an
on-going study of the behaviour of the connection in fire where the model will be
validated experimentally.
DOI: 10.1007/978-94-007-7811-5_22, RILEM 2014
232 D. Brandon et al.
60
50
Distance (mm)
40
30
20
F/2 10
0
axis of symmetry -0.01 0.01 0.03 0.05
Deflection (mm)
Bernoulli beam theory
Structural node Timoshenko beam

theory
Thermal node Higher order beam
theory
Experimental
(Thomson, 2010)
Fig. 1 Section of a failed connection (left) model (middle) and deflection at 1kN load
(right)
Only a small number of analyses of timber connections in fire have been pub-
lished in the past. A part of these consist of 3-dimensional (3D) structural finite
element (FE) analyses and a 3-dimensional heat transfer model (Racher et al.,
2010; Peng et al., 2011; Audebert et al., 2012; Laplanche and Racher, 2004). The
FE analyses are not easy to get accurate because cracking of the timber locally
around the dowel should be taken into account. 3D FE analyses that do not model
the cracking or local failure will predict behaviour which is too stiff and must be
calibrated to experimental results to perform accurately. A good solution for 3D
modelling of the embedment of a dowel in timber at room temperature was devel-
oped by Xu et al. (2012), by implementing a modified Hill yield criterion. How-
ever, this method requires a high level of expertise in finite element modelling
under the condition of at constant temperature.
Konig and Fontana (2001) concluded from experimental research that the num-
ber of dowels has an insignificant influence on the time to failure. This indicates
that instead of modelling the full connection, it is adequate to model just a single
dowel to predict the behaviour of the connection. Moss et al. (2010) successfully
implemented modified Johansen yield equations (Johansen, 1949) to predict fire
resistance. The Johansen equations are, however, based on a yield moment of the
dowel, which is adequate for steel dowels. GFRP dowels in shear connections,
however, fail in shear, making the Johansen equations unsuitable for predicting
the fire resistance of the non-metallic connection (Thomson et al., 2010).
Modelling of Non-metallic Timber Connections at Elevated Temperatures 233
This research presents a model based on the structural FE model of Thomson

(2010). He used beams on an elastic foundation to model the deformation of the
dowel in the connection where the beam represents the dowel and the foundation
represents the timber. To gain more accuracy the element mesh is refined and the
beams on elastic foundation are replaced with shear deformable beams on elastic
springs. A comparable structural model with a beams on springs was recently
introduced in a fire model by Cachim and Franssen (2009). However, their design
is made for timber connections with steel dowels. The beam theory used for the
beam elements is not suitable for predicting the performance of connections with
GFRP rods. To accurately predict the behaviour of these connections shear de-
formable beam theories should be used, since the shear modulus and shear
strength of the dowel have a significant influence on the connection behaviour.
These properties are determined experimentally at elevated temperatures for
GFRP and the results are presented in this paper. Cachim and Franssen used the
stiffness reduction factors provided by Eurocode 5, to take into account the re-
duced embedment stiffness at elevated temperatures. The embedment stiffness is
however highly dependent on local cracking of the timber around the dowel and is
therefore not only dependent on the stiffness of the timber. Timber embedment
tests at elevated temperatures are presented in this paper to qualify this assump-
tion. The FE models are made using Matlab v7.11, but different computational
software can be used to make the models.
2 Modelling at Room Temperature

Thomson (2010) proposed an FE model to determine the slip modulus using
beams on elastic foundation, but it does not correspond well to his test results. In
this paper beams on multiple springs are used to predict deflections and to con-
struct the basis for a temperature-dependent model. The proposed FE model
shown in Fig 1 (middle) contains beam elements, which represent the GFRP
dowel, and springs which represent the stiffness of the timber and the DVW (thick
lines). Only one symmetrical half of the connection is modelled to reduce the cal-
culation time. The other half is replaced by a boundary condition that restrains the
rotation.
Different beam elements are considered. The classic or Bernoulli beam theory
assumes that plane cross sections of a beam remain plane and normal to the mid-
line (Fig.2a). The theory is based on pure bending of a beam and therefor does not
take shear deformation into account, in contrast with shear deformable beams. The
Timoshenko beam is a first order shear deformable beam and allows rotation of
the cross section around the midline (Fig.2b). This rotation corresponds to a shear
stress, but since the cross section in Timoshenko beams still remains plane, the
shear stress is constant over the depth of the beam. However, the shear stress has
to be zero at the top and the bottom of the beam and is highest at the midline of
the beam. Therefore, in reality the cross section curves (Fig.2c). The Timoshenko
theory can be corrected with a shear correction factor. A value of 0.88 for a
circular section is assumed in this study.
dw/dx dw/dx

a) Bernoulli b) Timoshenko c) Higher order

beam theory beam theory beam theory
Fig. 2 Beam theories
Higher order beam theories take the warped section into account. The beam
element published by Eisenberger (2003) is used to simulate the dowel. The mate-
rial properties of the timber are determined with standard embedment tests accord-
ing to ASTM D5764 (2002) and the material properties of GFRP are determined
with torsion tests that are discussed below. The results of the two dimensional FE
analysis parallel to the grain can be seen in Fig.1. The results correspond to a load
F of 1 kN and are compared with experimental results of Thomson et al. (2010).
The point representing the experimental results is an average gained from 95
tested dowels and the deformation of the DVW plate is excluded from it. A
slightly more accurate model can be obtained by adding spring elements near the
shear surface. However, to maintain stability of the heat transfer analysis
discussed in the next section the element mesh is chosen as shown in Fig.1.
From the results, it is concluded that predictions with the Bernoulli theory are
not adequate for describing the connection behaviour. This indicates that the shear
deformation plays an important role in the connection behaviour. The higher order
beam theory resembles the experimental results only slightly better than the Ti-
moshenko beam theory. Because of its simplicity, the Timoshenko beam theory is
chosen for further simulations.
The shear strength of the dowel is determined using an expression for the
maximum shear stress in a solid circular section given by Gere (2004):
4V
max = ,
3A
where V is the shear force on the dowel and A is the cross sectional area.
3 Modelling at Elevated Temperatures
Timoshenkos beam element adopted in the FE model of Fig.1 is used as a basis

for the temperature dependent model. To include the temperature effects, a heat
transfer model is introduced, which estimates the local temperatures in the connec-
tion. An increase in local temperature will result in reduced material properties.
The stiffness and strength of each of the spring and beam elements of the FE
model are adjusted to the local temperature estimated by the heat transfer model.
The test setup to determine the reduced shear stiffness and shear strength of the
individual beam elements is shown in Fig.3a. The setup to determine the reduced
stiffness of the individual spring elements is shown in Fig. 3b.
Fig. 3 Material tests at elevated temperatures
3.1 Heat Transfer Model

The heat transfer model presented in this paper is 1-dimensional in the direction of
the GFRP fibres. In this model a full timber section is assumed i.e. no GFRP heat
transfer is modelled. The timber has a higher conductivity and a lower heat capac-
ity per unit volume than GFRP, indicating that the assumption is conservative for
deep beams. However, mass transfer as well as heat transfer is involved during the
heating of timber. For example, moisture will evaporate which requires energy
and slows the heating process down. The validity of assuming a 1D heat transfer
model consisting of only timber will be validated in the on-going research by
comparisons with 2D and 3D models and by experimental validation. The heat
transfer model consists of nine nodes in the length of the dowel. These nodes are
shown in Fig.1 as circles.
The explicit finite difference solution of Fouriers heat equation to predict the
temperatures in the nodes presented by Gilbert (2005) is used:
t Tmp1 Tmp Tmp+1 Tmp

Tmp +1 = + + T4
p
Cm Rm 1,m Rm , m +1
Where Tmp is the temperature at time p at point m. R is the thermal resistance, Cm

is the heat capacity and t is the time step. The temperature of a node in each
following time step is calculated from the temperature difference with the adjacent
nodes. 1D explicit finite difference solutions require a stability criterion, which is
in this case (Incropera, 2013):
Fo 1
2
Fo is the Fourier number. This means that to maintain stability the time steps
have to be small in the ratio of the node distances. If node distances are too small
there is a requirement for small time steps and long calculation times.
The heat flux due to convection and radiation is calculated as follows
(Buchanan, 2002):
q" = hc (T f Ts ) + eff (T f4 Ts4 ) ,

where hc is the convection coefficient, is the configuration factor, is the Stefan
Boltzmann constant, eff is the effective emissivity, Tf is the fire temperature and
Ts is the surface temperature. The convection coefficient is taken as 25 W/m2K
(Buchanan, 2002). The configuration factor is taken as 1.0 to simulate zero
distance between the heat source and the surface. The effective emissivity eff can
be calculated with the furnace emissivity f and the surface emissivity s (Peng et
al., 2011):
fs
eff =
f + s fs
f and s are both taken as 0.9 (Peng et al., 2011).
Thermal properties of timber required for the heat transfer model are specific
heat, density and conductivity. The mass transfer however should not be ignored.
Different authors in the past have included mass transfer in the heat transfer prop-
erties of timber (Knig, 2006; Frangi, 2001; Fredlund, 1993). The authors treat
complex phenomena like pyrolysis and evaporation and fissures in the charcoal
differently, but all confirmed their set of properties with test results. Therefore, the
practical difference between the works is small.
The properties presented by Frangi (2001) are implemented in the heat transfer
model of the present study. The energy required for evaporation of moisture is
included in the heat capacity curve. Similarly the energy required for pyrolysis is
included in the function. At higher temperatures over 400C the cracking of the
char layer is included by implementing an additional conductivity.
3.2 Material Properties at Elevated Temperatures

Tests to determine the shear properties of GFRP and the embedment properties of
timber at elevated temperatures were performed in an Instron 3111 convection
oven that fits in the testing machine. The test setups are shown in Fig.3. In both
tests thermocouples were used to measure the temperature. One thermocouple was
placed in a drilled hole in the steel fixture, one thermocouple was placed in a
drilled hole inside an unloaded sample to determine the time required to heat the
sample to the right temperature and another thermocouple measured the tempera-
ture of the surface of the specimen. The test was initiated when all thermocouples
showed an error of less than 3C.
In the torsion test the upper clamp was fixed in all degrees of freedom. The
lower clamp was free to move upwards and downwards and was fixed in the other
degrees of freedom. Four samples were tested at temperatures of 20C, 60C,
100C and 140C and two samples were tested at 180C. The GFRP used in these
tests consisted of S-glass fibres with a polyester resin matrix without fire resistant
additives.
The embedment test was performed according to ASTM D5764 because of the
ease of loading. Santos et al. (2010) showed, with a large number of tests, that this
method is equivalent to the method described in EN383 (2004). The specimens
had a small thickness of 20 mm to improve the heat transfer. The test was per-
formed at temperatures of 20C, 60C, 100C, 140C, 160C and 200C. Lami-
nated veneer lumber Kerto-S was used for these embedment tests.
Fig.4 shows test results for the embedment stiffness (left) and the embedment
strength (right) at elevated temperatures. Here the values are normalized to the
average at 20C. At 20C the average embedment stiffnesses are 463 N/mm2 and
190 N/mm2 for the parallel and perpendicular to grain direction respectively. The
average embedment strengths for these directions are 361 N/mm2 and 322 N/mm2
respectively. Fig.4 also shows the stiffness and the strength of timber in compres-
sion at elevated temperatures according to Eurocode 5 (BS EN 1995-1-2:2004,
2006) for advanced fire calculations. It can be seen that the test results do not re-
semble the code. The compressive stiffness and compressive strength of timber are
at room temperature not linearly related to the embedment stiffness and embed-
ment strength. There are no grounds for assuming that these properties would
decrease in similar ratios when temperature increases. A notable stiffness increase
after 100C in parallel and perpendicular to grain direction, can be caused by the
evaporation of moist at that temperature. However, the scatter of results is large
and the number of tests is not sufficient to confirm this stiffness increase.
Timber starts charring at approximately 300C at which the timber has no
significant strength left. For the present model, linear interpolation between the
average (AVG.) test results per temperature is used, as shown in Fig.4. After
200C, a linear decrease is assumed which reaches zero strength or stiffness at
300C.
1.2 1.2
Normalized embedment
Normalized embedment
1 1
0.8 0.8
strength
stiffness
0.6 0.6
0.4 0.4
0.2 0.2
0 0
0 50 100 150 200 0 50 100 150 200
Temperature (C) Temperature (C)
Parallel to grain Perpendicular to grain

AVG. parallel to grain AVG. perpendicular to grain
BS EN 1995-1-2:2004
Fig. 4 Results of embedment tests at elevated temperatures
1.2
Shear strength
1
Normalized value
0.8
Shear modulus
0.6
0.4
Modelled shear
0.2 strength
0
20 60 100 140 180 Modelled shear
stiffness
Temperature (C)
Fig. 5 Results of torsion tests at elevated temperatures
The results of the torsion tests are shown in Fig.5. The average shear strength
and shear stiffness are 40.3MPa and 2.59 GPa respectively. It can be seen that
both shear strength and shear modulus decrease rapidly between 20 and 60C. It
can also be noted that the deviation between results becomes very small at higher
temperatures. Similar relative reduction is assumed for the flexural modulus and
strength. This will be studied in future experiments.
Fig.6 shows the predicted deflections of the symmetrical half of the dowel dur-
ing a standard fire. The load is equal to 40% of the maximum load and the side
member thickness is 48 mm. It can be seen that the deformation accelerates during
a fire. According to the model, shear failure in the dowel will occur after 26
minutes. The shear failure occurs at the shear plane where the temperature reaches
77 C at the moment of failure.
The use of pultruded rods with a polyester resin results in a rapid decrease of
connection properties. The fire resistance can be enhanced by using pultruded rods
that contain a resin with a higher glass transition temperature, like vinyl-ester,
epoxy or phenolic resins.
0.2
0.15
0.1
0.05
Deflection (mm)
0
0 min
0 10 20 30 40 50 60
-0.05
10 min
-0.1
20 min
-0.15
-0.2
-0.25
-0.3
Location (mm)
Fig. 6 Dowel deflection of the symmetrical half during a standard fire
5 Conclusions
A model to describe the connection slip and predict the dowel failure at elevated
temperatures of non-metallic timber connections with pultruded glass fibre rein-
forced polymer dowels and a densified veneer wood flitch plate was presented.
This model is part of an on-going study of the fire performance of these non-
metallic connections. The model consists of a 2-dimensional structural model and
a 1-dimensional heat transfer model. Mechanical material properties at elevated
temperatures required for the model were measured experimentally. In further
studies at the University of Bath the model will be validated with fire tests and the
1-dimensional heat transfer model will be compared with a 2 and 3 dimensional
heat transfer model. The most important findings of this study can be summarized
as follows:
The embedment stiffness and the strength reductions resulted from tests, do not
coincide well with the stiffness and strength reductions of timber in compression
at elevated temperatures according to BS EN 1995-1-2:2004 (2006).
The GFRP dowel of this study has a thermosetting polyester matrix without
fire resistant additives. This material degrades rapidly at elevated temperatures,
resulting in a short time to failure. GFRP rods which perform better in fires will be
considered in future research.
References
ASTM D5764. Standard test method for evaluating dowel-bearing strength of wood and
wood-based products (2002)
Audebert, M., Dhima, D., Taazount, M., Bouchar, A.: Behavior of dowelled and bolted
steel-to-timber connections exposed to fire. Engineering Structures 39, 116125 (2012)
Buchanan, A.H.: Structural design for fire safety. Wiley, Chichester (2002)
BS EN 1995-1-2:2004. UK National Annex to Eurocode 5: Design of timber structures
part 1-2: General Structural fire design, British Standards (2006)
Cachim, P.B., Franssen, J.M.: Numerical modelling of timber connections under fire load-
ing using a component model. Fire Safety Journal 44, 840853 (2009)
Drake, R.D.: The advancement of structural connection techniques for timber buildings.
PhD Thesis, University of Bath (2003)
Eisenberger, M.: An exact high order beam element. Computers & Structures 81, 147152
(2003)
EN383, Timber structures. Test methods: Determination of embedding strength and foun-
dation values for dowel type fasteners. European standard, Brussels (2004)
Frangi, A.: Brandverhalten von Holz-Beton-Verbunddecken. Eidgenssische Technische
Hochschule, Zrich (2001)
Fredlund, B.: Modelling of heat and mass transfer in wood structures during fire. Fire Safe-
ty Journal 20, 3969 (1993)
Gere, J.M.: Mechanics of materials with infotrac. Thomson/Brooks/Cole (2004)
Gilbert, B.: Thermal mass and the effects of dynamic heat flow. Master Thesis, University
of East London, School of Computing and Technology (2005)
Incropera, F.P.: Foundations of heat transfer. John Wiley, Singapore (2013)
Johansen, K.W.: Theory of timber connections. International Association for Bridge and
Structural Engineering 9, 249262 (1949)
Knig, J.: Effective thermal actions and thermal properties of timber members in natural
fires. Fire and Materials 30, 5163 (2006)
Knig, J., Fontana, M.: The performance of timber connections in fire. In: Proceedings of
the RILEM Symposium Joints in Timber Structures, vol. 22, pp. 639648 (2001)
Laplanche, K., Dhima, D., Racher, P.: Predicting the behaviour of dowelled connections in
fire: Fire tests results and heat transfer modelling. In: 8th World Conference on Timber
Engineering, Lahti, Finland (2004)
Moss, P., Buchanan, A., Fragiacomo, M., Austruy, C.: Experimental testing and analytical
prediction of the behaviour of timber bolted connections subjected to fire. Fire Technol-
ogy 46, 129148 (2010)
Pedersen, M.: Dowel type timber connections. PhD Thesis, Danmarks Teknise Universitet
(2002)
Peng, L., Hadjisophocleus, G., Mehaffey, J., Mohammad, M.: Predicting the fire resistance
of woodsteelwood timber connections. Fire Technology 47, 11011119 (2011)
Racher, P., Laplanche, K., Dhima, D., Bouchar, A.: Thermo-mechanical analysis of the fire
performance of dowelled timber connection. Engineering Structures 32, 11481157
(2010)
Santos, C.L., De Jesus, A.M.P., Morais, J.J.L., Lousada, J.L.P.C.: A comparison between
the EN 383 and ASTMD5764 test methods for dowel-bearing strength assessment of
wood: Experimental and numerical investigations. Strain 46, 159174 (2010)
Thomson, A.: The structural performance of non-metallic timber connections. PhD Thesis,
University of Bath (2010)
Thomson, A., Harris, R., Ansell, M., Walker, P.: Experimental performance of non-metallic
mechanically fastened timber connections. The Structural Engineer 88, 2532 (2010)
Xu, B.-H., Bouchar, A., Taazount, M., Racher, P.: Numerical simulation of embedding
strength of glued laminated timber for dowel-type fasteners. Journal of Wood Science,
17 (2012)
Analysis of the Brittle Failure and Design
of Connections Loaded Perpendicular to Grain
Bettina Franke1,2 and Pierre Quenneville2
1
Bern University of Applied Sciences, Switzerland
bettina.franke@bfh.ch
2
The University of Auckland, New Zealand
{p.quenneville,b.franke}@auckland.ac.nz
Abstract. The brittle failure behaviour and the design of double shear connections
with mechanical fasteners loaded perpendicular to grain are analysed using
experimental and numerical test series and the fracture mechanics methods. The
results observed define the important geometry parameters which influence the
load carrying capacity. Based on the failure criteria determined for double shear
connections, a new design approach are proposed and compared to current interna-
tional design standards. The correlation between the new design proposal and the
experimental test results done or published confirms the methods used and the
failure criteria determined.
Keywords: mechanical fasteners, dowel type connection, double shear connec-

tions, design proposal, fracture mechanics, European spruce, Canadian spruce.
1 Introduction
Timber constructions almost include joints with mechanical fasteners. The con-
nections can be loaded under tension, compression and under angle to the grain
orientation of wood. Depending on the size, layout as well as the loading situation,
the connection describes brittle or ductile failure behaviour. The ductile failure
depends on the embedment strength of the wood and the bending capacity of the
dowel. For the prediction of the ductile failure the established European Yield
Model is used in international standards. The brittle failure of connections occurs
due to tension or shear stress in the wood member. Based on the anisotropy of
wood the tension stress perpendicular to grain leads to very brittle failure behav-
iour and it is the most critical case for the failure of connections.
For the prediction of the splitting failure of double shear connections loaded
perpendicular to grain, different design equations are available in various interna-
tional standards. The design standards are based on one hand on a strength crite-
rion which was introduced by Ehlbeck, Grlacher and Werner (1989) and on the
DOI: 10.1007/978-94-007-7811-5_23, RILEM 2014
244 B. Franke and P. Quenneville
other hand on fracture mechanic, introduced by v. d. Put and Leijten (1990).The

brittle failure occurs due to the splitting of the wood under tension perpendicular
to grain and/or shear stress. Current research results and publications show that
there are disagreements between the experimental results and the international
design standards, e.g. Franke and Quenneville(2010, 2011), Schoenmakers(2010),
Jensen (2005).
l Fig. 1 Double shear dowel type connection
b
lr loaded perpendicular to grain
ar
h
m
he
ac
2 Experimental and Numerical Test Program
Experimental and numerical test series are used for the analyses of the failure
behaviour. The experimental test series with multiple dowel-type connections and
different loaded edge distances were conducted in laminated veneer lumber of
Radiata Pine (LVL). The test specimens were in large scale format to provide
comparable results to practical construction details. The detailed description of the
test setup is summarized in Franke et al. (2012).
In addition various numerical test series with single or multiple dowel-type
connections with different loaded edge distances, positions, different number of
dowels per row or column as well as different spacing within the connection are
simulated. The numerical model used is defined as a 2-dimensional model based
on the purpose to especially investigate the splitting of the wood and is presented
in Franke and Quenneville (2010). The numerical model simulates the splitting
failure of wood under tension perpendicular to grain and shear as well as the duc-
tile failure of wood due to compression. However, since the focus is to investigate
the brittle failure mechanism, the bending of the dowel is not included in the
numerical model and the influence is neglected. This is acceptable if the dowel
slenderness ratio is small and ductile behaviour due to dowel bending is mini-
mized. The capabilities of the numerical model were proofed on comprehensive
experimental test series done in Canadian spruce glulam and also with the experi-
mental test series done in LVL. The complete experimental and numerical test
programme is summarized in Table 1. In each test series the dowel diameter is
20 mm.
Analysis of the Brittle Failure and Design of Connections Loaded Perpendicular 245
Table 1 Experimental and numerical test program
Connection Layout
Test Dowel Loaded edge Connection Connection
Material setup Sizes b / h / l [mm] mxn distance he/h width ar height ac
LVL exp. 63/400/1600 3x1 0.2/ 0.4/ 0.6 8d -
LVL exp. 63/600/2400 3x1 0.13/ 0.4/ 0.6 8d -
LVL exp. 63/400/1600 3x2 0.4/ 0.6 8d 4d
LVL exp. 63/400/2400 3x2 0.4/ 0.6 8d 4d
LVL exp. 63/400/2400 3x2 0.4/ 0.6 8d 4d
LVL exp. 63/400/2400 3x2 0.4 8d 4d
LVL exp. 63/600/2400 5x2 0.4 16d 4d
LVL num. 63/(200, 400, 600)/4h 2x2 0.2/ 0.4/ 0.6/ 0.8 3d/ 6d/ -
10d/15d/20d/25d
GL num. 80/(190, 300, 400, 600, 1x1 0.2/ 0.3/ 0.4/ 0.5/ -
-
800)/610 0.6/ 0.7/ 0.8
GL num. 80/190/4h+ar 2x1 0.2/ 0.3/ 0.4/ 0.5/ 3d/ 4d/ 6d/ 8d/ 10d/
-
0.6/ 0.7/ 0.8 15d/ 20d/ 25d
GL num. 80/(300, 400, 600, 2x1 0.3/ 0.5/ 0.7 3d/ 6d/ 8d/ 10d/
-
800)/(4h+ar) 20d
GL num. 80/190/610 1-6x1 0.4/ 0.6 (n - 1) 3d -
GL num. 80/304/1320 1-4x1-3 0.44/ 0.7 16.8d (m-1) 3d
GL num. 80/304/1320 2-3x2 0.44/ 0.7 16.8d 3d, 4.5d,
6d
3 Failure Behaviour of Connections Loaded Perp. to Grain
3.1 Failure Behaviour Observed during the Experimental Tests

The load capacity of double shear dowel-type connections transversely loaded
depends mainly on the loaded edge distance but also on the connection layout, as
shown in Fig. 2 and Fig. 3. Increasing the loaded edge distance increases not only
the splitting load capacity but also allows a load reserve between crack initiation
and separation up to the complete failure. The crack initiation describes the load
when first cracks occur and the crack separation load the load level at complete
separationof the inner part of the connection. The analysis of the distribution of
the load capacity over the number of dowels shows that each row of the connec-
tion carries the same load portion and that the dowels at the edge of the connection
with the largest loaded edge distance carries the greatest load portion and also
fully trigger the failure, Franke and Quenneville (2010), Franke et al. (2012).
Fig. 2 Crack initiation, crack separation and splitting failure load for b/h = 63/400 mm and
63/600 mm with m x n = 3 x 1
Fig. 3 Load distribution for the dowels of the top row for a connection with two rows
3.2 Failure Behaviour Analysed by Fracture Mechanics

For the characterization of the brittle failure behaviour of double shear connec-
tions investigated, the energy balance method together with crack resistance
curves as one of fracture mechanic methods were used. The crack process is clas-
sified in three different modes. The fracture mode I describes the symmetric sepa-
ration under normal tension stress, mode II contains the in plane shear stress and
mode III the out of plane shear stress. Double shear dowel-type connections
loaded perpendicular to grain fail, considering the stress situation, in interaction of
fracture modes I and II.
The crack resistance curves evaluated for double shear connections show a
nonlinear material behaviour. The crack grows stable as long as the crack resis-
tance increases more than the crack extension force under a constant load during
crack propagation. If the crack resistance exceeds the critical value, the crack will
grow in an unstable manner and the system fails. The crack resistance curves
respectively the critical fracture energies determined were split into the fracture
mode I and mode II using the method of Ishikawa et al. (1979). The critical frac-
ture energies determined are not comparable with the fracture energies known for
different materials for the pure fracture modes I or II. Because they are caused by
a stress situation related to one connection layout and do not relate to test setups
for the investigation of a single fracture mode. Therefore, in this paper, the frac-
ture energies determined for dowel-type connections are called specific critical
I , II
fracture energies, spec ,c .
Fig. 4 shows as example the numerical solution for a 3 by 2 dowel type connec-
tion and Fig. 5 the corresponding crack resistance curve for the dowel at the edge
with the largest loaded edge distance. In the case shown, the failure load is 60 kN.
Analysing the critical fracture energies for the complete connection shows that the
dowels at the edge with the largest loaded edge distance trigger the unstable fail-
ure, Franke and Quenneville (2011). The cracks between the outer dowels (inner
part of the connection) become also unstable and the wood between these dowels
is then completely separated. For the splitting failure design proposal the outline
of the connection layout, described by the loaded edge distance he and the connec-
tion width ar, will be used as observed during the experimental test series.
Fig. 4 Numerical solution - transverse Fig. 5 Crack resistance curve for mode I with
stress for 3 x 2 dowel type connection energy lines for the outside side of the dowel
marked in Fig. 4
3.3 Failure Criteria

For each dowel of the connection investigated, the crack resistance curves as well
as the critical specific fracture energies were determined, as described before. The
distribution of the specific critical fracture energies of the dowels at the edge with
the largest loaded edge distance shows a dependency on the loaded edge distance
he, the connection width ar and the depth of the member h, as shown in Fig. 6 and
Fig. 7. In general, the normalized fracture energies for solid wood respectively
glulam shows a different behaviour than LVL. Due to the multi-layered cross
section of the engineered wood product LVL, it is more homogeneous and the
brittle failure behaviour occurs different to solid wood.
The distribution of the specific critical fracture energies in relation to the mem-
ber depth shows an influence on the fracture mode I but not on mode II, as shown
as example for single dowel-type connections in Fig. 7. The test programme con-
siders member depths from 150 mm up to 600 mm. The increase of the member
depth results in decreasing of the normalized specific critical fracture energy for
fracture mode I. Whereas the values respectively the corresponding curves are all
close together for the fracture mode II.
3.0
LVL-test series
2.5 GL-test series
-3
Gspec b/F90 10
2.0
1.5
1.0
I
0.5
0.0 0.0
100 0.2
200 0.4
300 0.6
400 0.8
ar [mm] 500 he/h
600 1.0
Fig. 6 Comparison of specific critical frac- Fig. 7 Dependency of the specific critical
ture energies for mode I forGlulam and LVL fracture energies on the member depth for
test series modes I and II for Glulam test series
The splitting failure of dowel-type connections can be summarized using com-

mon failure criteria. The failure criteria describes the interaction between the frac-
ture modes I and II. The distribution of the numerical test series investigated are
compared with the linear-, quadratic- and Wu-failure criterion (1967), as shown in
Fig. 8. The assessment of the distribution of the specific critical fracture energies
for the test programme shows that the quadratic failure criterion mostly encloses
1.2 Glulam test series

LVL test series
1.0 Linear criterion 3.0
Quadratic criterion 2.5
0.8
-3
Criterion from Wu (1967)

Gspec b/F90 10
2.0
I
Kspecific /Kc
0.6 1.5
1.0
I
0.4
0.5
0.0 0.0
0.2
0.2
100
200 0.4
0.0 300 0 .6
400 0.8
0.0 0.2 0.4 0.6 0.8 1.0 1.2 ar [mm] 500 he / h ]
1
II II
Kspecific /Kc [-] 600 .0
Fig. 8 Failure criteria for dowel-type connec- Fig. 9 3-dim. curve of the normalized
tions loaded perpendicular to grain for fracture energies for mode I compared to
Glulam- and LVL test series the individual test results for solid wood/
glulam
all failure cases for Glulam and LVL test series. The quadratic failure criterion
will be used for the prediction of the splitting failure behaviour of dowel-type
connections. The criteria found takes into account the important parameters of the
connection layout of single and multiple dowel type connections as well as the
ratio between the fracture modes I and II.
4 Design Approach
The load capacities of double shear connections loaded perpendicular to grain

depend on the connection layout, the number of fasteners and rows as well as the
loaded edge distance and the cross section of the beam. An influence on the posi-
tion of the connection along the span of the beam could not be observed and is
therefore not considered in the design approach, Franke and Quenneville (2011).
Depending on the connection layout and its position over the member depth,
the connection fails either ductile such as bending of the dowel or the embedment
failure of the wood or in splitting of the wood. Therefore the design for double
shear connections loaded perpendicular to grain has to be used in combination
with the European yield model for the prediction of the ductile failure behaviour
as given in Eq. (1).
Fductile ( EYM )
Fconnection = min (1)
Fsplitting = F90
For the design proposal for the splitting failure, the quadratic failure criterion de-
termined will be used to consider the interaction between the transverse tension
and shear failure. Substituting the individual critical specific fracture energies with
I , II
the distribution of the normalized fracture energies norm , the splitting load F90
for dowel-type connections in wood becomes:
b 103
F90 = kr (2)
norm
I II
norm
I
+
c cII
WhereF90 in [N] is the load capacity depending on the splitting failure of the
wood. cI and cII [Nmm/mm] are the critical material fracture energies for the
fracture mode I or II and b [mm] the width of the member. The normalized frac-
I , II
ture energies norm enclose all individual critical specific fracture energies of the
connection considered. Therefore the critical specific fracture energy of each con-
nection layout investigated was normalized with the specimen width b and the
splitting load F90, see Eq. (3).
I , II
spec b
I , II
norm = (3)
F90
The distribution of all values for solid wood respectively glulam test series were
expressed with a 3-dimensional group of curves, which depends on the loaded
edge distance he/h, the connection width ar [mm] and the member depth h, as
shown in Eq. (4) and Eq. (5) and for LVL as in Eq. (5) and Eq. (6). Fig. 9 shows
as example the 3-dimensional curve for the member depth h = 190 mm compared
to the individual values of the test series. The empirically determined Eq. (4), Eq.
(5) and Eq. (6) are based on more than 200 different connection layouts investi-
gated in solid wood, glulam or LVL.
I
norm =e
( h ( 20010h h
1
e
0.25
ar )) (4)
he
II
norm = 0.05 + 0.12 + 1 103 ar (5)
h
I
norm =e
( 0.81.6 h h
e
1
3 ar
110 ) (6)
The analysis of the results of the test programme shows, that the load capacity
increases with increasing the number of rows. For an increasing number of rows,
the load capacity increases and later on become constant as well as the stress situa-
tion, as shown in Fig. 10. This behaviour could be summarized using the quadratic
interaction of the areas of the tension stress perpendicular to grain and the shear
stress besides the dowels at the corner of the top row, see Franke and Quenneville
(2011). These results lead to the following factor kr to respect the effect of the
number of rows:
1 for n =1
kr = (7)
0.1 + ( arctan ( n ) )
0.6
for n >1
The results of test series, which enclose different numbers of rows with different
spacing between the rows, show that the load capacity does not increase when
spacing increases, Franke and Quenneville (2011). Therefore a dependency on the
spacing between the rows is not included in the factor kr.
For wider connections with more than two columns, e.g. nail plate connections
as shown in Fig. 11, the splitting load has to be determined as for the whole con-
nection with the complete connection width ar but also as for single connections
with the individual connection widths ar,i. The minimum of the load capacities of
Eq. (8) is the governing splitting load capacity of the connection.
F1 ( ar ,1 ) + F2 ( ar ,2 )
F90 = min (8)
F3 ( ar )
2.0
Numerical test results
1.8 he/h = 0.44
he/h = 0.7
Factor kr [-]
1.6
1.4
1.2
1.0 Interaction of stress
kr Design proposal
0.8
0 1 2 3 4 5 6
Number of rows n [-]
Fig. 10 Factor kr, depending on the numerical Fig. 11 Design for nail plates or double
load capacities and stress situations deter- dowel-type connections
mined
5 Discussion and Comparison to Current Design Standards
5.1 Comparison to Experimental Test Results

The design proposal is compared with experimental test series in solid wood, glu-
lam and LVL. Fig. 12 includes the correlation with the experimental test series in
Canadian spruce glulam done by Reshke (1999), Kasim (2002) and Lehoux
(2004). The test series covers single and multiple dowel-type connections.
Furthermore the design proposal are also compared with the experimental test
series from: Ballerini (2004, 2003, 1999) who investigated mainly different depths
of the member and different loaded edge distances; Mhler and Lautenschlger
(1989) who observed different numbers of rows, connection widths and loaded
edge distances; Ehlbeck and Grlacher (1989) who did tests with nailed steel-to-
wood connections; and Schoenmaker (2010) where test series encloses various
double-shear connections in European spruce. All are summarized in Fig. 12. The
comparisons are always related to Eq. (1), because for all cases of the experimen-
tal test results published the differentiation between the ultimate or splitting load
capacities is not given. The material values used for Canadian spruce glulam and
European spruce are cI = 0.225 Nmm/mm2 and cII = 0.650 Nmm/mm2 , as referenced in
Vasic (2000) respectively Larsen and Gustafsson (1990). Fig. 12 always shows a
close correlation in the comparison of the experimental test series and the design
proposal.
Fig. 13 shows the comparison of the splitting load capacity and the design load
F90 for the experimental and numerical test series in LVL. In this case the splitting
load is known and the direct correlation to the new design approach can be shown.
The material values used for LVL are cI =1.0 Nmm/mm2 and cII = 6.0 Nmm/mm2 , as
referenced in Franke and Quenneville (2012), Ardalany et al. (2012).
Fig. 12 Comparison of design proposal and Fig. 13 Comparison of design proposal and
experimental test series published for Euro- experimental as well as numerical test se-
pean spruce or Canadian spruce ries for LVL
6 Conclusion
A new design proposal is presented for double shear connections in solid wood,
glulam and also LVL which allows the design of the splitting failure of the wood
due to connections loaded perpendicular to grain. The design approach is based on
fracture mechanic methods including the important parameters which influence
the load capacity of the connection. The comparison of the design results with
comprehensive experimental test series done in Canadian and European spruce
confirms the procedure of the design proposal. The good agreement is based on
over 200 different experimental test configurations and 600 numerical test results.
Acknowledgments. The research work was generously supported by the Structural Timber
Innovation Company (STIC) from New Zealand.
References
Ballerini, M.: A new prediction formula for the splitting strength of beams loaded by dow-
el-type connections. In: CIB-W18, paper 37-7-5, United Kingdom (2004)
Ballerini, M.: A new set of experimental tests on beams loaded perpendicular to grain by
dowel-type joints. In: CIB-W18, paper 32-7-2, Austria (1999)
Ballerini, M.: Beams transversally loaded by dowel-type joints: Influence on splitting
strength of beam thickness and dowel size. In: CIB-W18, paper 36-7-8, USA (2003)
Canadian Standards Association, O86-09 Engineering design in wood, Canada (2009)
Deutsches Institut fr Normung e. V., DIN 1052:2008 Entwurf, Berechnung und
Bemessung von Holzbauwerken Allgemeine Bemessungsregeln und
Bemessungsregeln fr den Hochbau, Berlin, Germany (2008)
Ehlbeck, J., Grlacher, R., Werner, H.: Determination of perpendicular to grain tensile
stresses in joints with dowel-type fasteners. In: CIB-W18, paper 22-7-2, Germany
(1989)
European Committee for Standardization (CEN), EN 1995-1-1:2004 Design of Timber
structures, Brussels, Belgium (2004)
Franke, S., Franke, B., Quenneville, P.: Analysis of the failure behaviour of multiple dowel-
type connections loaded perpendicular to grain in LVL. In: Proceedings of 12th World
Conference on Timber Engineering, New Zealand (2012)
Franke, B., Quenneville, P.: Investigation of the splitting behaviour of Radiata pine and
LVL under tension and shear. In: Proceedings of 12th World Conference on Timber En-
gineering, New Zealand (2012)
Franke, B., Quenneville, P.: Numerical modeling of the failure behavior of dowel connec-
tions in wood 137, 186, 155226 (2011)
Franke, B., Quenneville, P.: Analyses of the failure behaviour of transversely loaded dowel
type connections in wood. In: Proceedings of 11th World Conference on Timber Engi-
neering, Italy (2010)
Franke, B., Quenneville, P.: Failure behaviour and resistance of dowel-type connections
loaded perpendicular to grain. In: CIB-W18, paper 43-7-6, New Zealand (2010)
Ishikawa, H., Kitagawa, H., Okamura, H.: J Integral of a mixed mode crack and its applica-
tion. In: ICM 3, Cambridge, England, vol. 3 (1979)
Jensen, L.J.: Splitting strength of beams loaded perpendicular to grain by dowel joints.
Journal of Wood Science 51, 480485 (2005)
Kasim, M.H.: Bolted timber connections loaded perpendicular-to-grain - Effect of row
spacing on resistance. Royal Military College of Canada, Kingston, Canada (2002)
Larsen, H.J., Gustafsson, P.J.: The fracture energy of wood in tension perpendicular to
grain results from a joint testing project. In: CIB-W18, paper 23-19-2, Portugal (1990)
Lehoux, M.C.G.: Bolted timber connections loaded perpendicular to grain. A proposed
design approach. Royal Military College of Canada, Kingston (2004)
Ardalany, M., Deam, B., Fragiacomo, M.: Experimental results of fracture energy and
fracture toughness of Radiata Pine laminated veneer lumber (LVL) in mode I (opening).
Materials and Structures 45, 11891205 (2012)
Reshke, R.: Bolted timber connections loaded perpendicular-to-grain - Influence of joint
configuration parameters on strength. Royal Military College of Canada, Kingston
(1999)
Schoenmakers, J.C.M.: Fracture and failure mechanisms in timber loaded perpendicular to
the grain by mechanical connections. Doctoral thesis, University of Technology Eind-
hoven, Netherlands (2010)
Van der Put, T.A.C.M., Leijten, A.J.M.: Evaluation of perpendicular to grain failure
of beams caused by concentrated loads of joints. In: CIB-W18, paper 33-7-7, The
Netherlands (2000)
Vasic, S.: Applications of fracture mechanics to wood. Doctoral dissertation, University of
New-Brunswick, Canada (2000)
Wu, E.M.: Application of fracture mechanics to anisotropic plates. Journal of Applied
Mechanics, Series E 34(4), 967974 (1967)
Structural Performance and Advantages
of DVW Reinforced Moment Transmitting
Timber Joints with Steel Plate Connectors
and Tube Fasteners
Daniel Brandon1,* and Adriaan Leijten2
1
Dept. of Civil Eng. and Architecture, University of Bath, UK
D.Brandon@bath.ac.uk
2
Structural Design Unit, Eindhoven University of Technology, The Netherlands
A.J.M.Leijten@tue.nl
Abstract. This paper presents a study to the moment-rotation aspects of two 3-

member DVW reinforced timber connections with an inter-connecting steel plate
used as middle member. Previous studies showed that reinforcing dowel-type
timber connections with densified veneer wood (DVW) and using expanded tube
fasteners results in connections with superior structural properties compared to all
conventional connections. In this connection type, the DVW prevents premature
timber splitting. The tube fasteners aid a high initial stiffness, a high ductility and
a high reliability. A drawback of the connection, already in a 3-member connec-
tion, is the total thickness. By using only two side members and a much thinner,
steel middle member, the thickness is strongly reduced. The steel middle member
is used as a connecting interface in a flitch plate connection. This generally results
in a 50% reduction of the rotational stiffness. However, it is shown by an analyti-
cal and numerical study, that the rotational stiffness of two closely spaced, flitch
plate DVW connections acting in series remains unchanged if certain conditions
are fulfilled. Two full connection tests are performed to confirm the analytical and
numerical results. Additionally, the paper presents a comparison to a conventional
connection, which confirms the structural quality of the reinforced connection.
Keywords: densified veneer wood (DVD), DVD reinforced connections, ex-

panded tube fasteners, flitch plate connection, moment capacity, moment-rotation
relationship.
1 Introduction
Moment connections in timber structures in general lead to limitations of struc-
tural strength. One conventional connection of this type is a moment connection
*
DOI: 10.1007/978-94-007-7811-5_24, RILEM 2014
256 D. Brandon and A. Leijten
a
b
Fig. 1 The single connection (a), the flitch plate connection before assembly (b) and during
assembly (c)
with shear dowel-type fasteners. These dowel-type connections usually do not

exceed 20% of the strength of the connected beams. Also the rotational stiffness is
significantly lower than the bending stiffness of the beams.
A connection type developed in the 1990s showed a high rotational stiffness
and moment capacity (Leijten, 1998). This connection is shown in Fig.1a. Ex-
panded steel tubes are used as fasteners and a densified veneer wood (DVW)
plate is glued on each timber member to reinforce the timber. DVW is cross wise
layered plywood compressed at high temperatures and prevents timber splitting in
the connection. This material is significantly stronger and stiffer than timber and
shows almost isotropic material properties.
The low rotational stiffness in traditional moment connections can be blamed
on the low embedment stiffness of timber and the hole clearance necessary to
assemble the connection easily. The DVW reinforcement proposed by Leijten
(1998) has an embedment stiffness that is significantly higher than the embedment
stiffness of timber. Additionally, Leijten used hollow steel tubes instead of con-
ventional solid dowels. These easy fitting tubes are expanded with a hydraulic
jack after positioning them in the hole, Fig.1c. Not only does the hole clearance
disappear as a result of this process, but the connection becomes pre-stressed and
the initial stiffness increases significantly.
The total thickness of the DVW reinforced connection is a downside of the
connection type. This becomes even more significant in 5-member or multi-
member connections of this type. To reduce the thickness of the connection, two
connections are placed in series and have a much thinner steel plate as a middle
member, as can be seen in Fig.1b and Fig.1c. The two connections in series are in
this paper further denoted as one flitch plate connection.
Usually the rotation of a single connection s (Fig.2a) is half the rotation of a
flitch plate connection fp (Fig.2b) for any bending moment. The study presented
in this paper, however, shows in an analytical, numerical and experimental way
that the flitch plate connection can have a similar stiffness and a higher strength
than the single connection. In the flitch plate connection the connected timber and
DVW members that are in the same plane can suppress each others rotation. In
Structural Performance and Advantages of DVW Reinforced Moment 257
Fig.2c it can be seen that the rotation will cause a contact between these members
if the timber members are closely spaced. The contact force that comes to exist
becomes larger when the rotation increases. The moment of first contact between
these members is called gap closure in this paper. The significance of the rotation
suppressing effect is dependent on the size of the initial gap between the timber
members. The study presented in this paper predicts the relationship between rota-
tion fp and bending moment M analytically and numerically. It also confirms the
predictions with experiments.
a b c
s fp = 2s fp /= 2s
M M
M
Fig. 2 The rotation difference of the single (a), the flitch plate connection without (b) and
with (c) rotation suppressing effect
2 Analytical Approach
The analytical approach aims to describe the bending moment-rotation curve for
the flitch plate connection. The rotation suppressing effect is studied for different
initial gap sizes between the timber members. The nonlinear load-slip behaviour
of a DVW reinforced connection with a single tube is previously studied by Lei-
jten (1998) and his results are used for this analysis.
As a basis for the analytical approach, assumptions are made. Firstly, it is as-
sumed that half of the flitch plate connection can be modelled as a single connec-
tion replacing the other half by a symmetry plane, as shown in Fig.3. Therefore,
the analysis is restricted to splice connections. Secondly, in the analytical model it
is assumed that the distance between the tube fasteners stays unchanged. There-
fore, the tubes rotate around one single rotation centre when the connection is
loaded. A translation can hereby be seen as a rotation around an infinitely far
point. Thirdly, it is assumed that applying the steel plate as a middle member does
not influence the load-slip behaviour of the tube connection. Lastly, it is assumed
that tube failure is governing the strength of the connection. Other failure mecha-
nisms are studied in the numerical and experimental work.
Fig.3 schematically shows a symmetrical half of the connection. If a bending
moment rests on the connection, the rotation centre R is initially located in the
gravity centre of the tubes. This will remain unchanged until the timber and DVW
members of both symmetrical halves come into contact. At that moment additional
contact force Fc will come to exist. To maintain equilibrium the rotation centre has
to move over the path of rotation centre given in the figure. An iterative calcula-
tion procedure is used to determine the location of the rotation centre on this path.
The height of the compressed contact surface hc is calculated by using the equilib-
rium of forces and geometrical requirements. If the results of both methods are
similar, the location of the rotation centre is determined.
Fig. 3 Forces in the symmetrical half of the connection before (left) and after (right) gap
closure
3 Numerical Approach
The numerical approach aims to predict the moment rotation behaviour of the
DVW reinforced flitch plate connection including different failure mechanisms.
The numerical study consists of three-dimensional finite element analyses per-
formed using ABAQUS v6.8. This part of the research is only presented briefly in
this paper and will be more thoroughly discussed in a following paper by the au-
thors. The aims of the numerical approach were to study the failure mechanisms
that can occur and to predict the moment-rotation relationship.
Special attention is put on failure of the bond line between DVW and timber in
this study. The fractural mechanical properties of DVW-timber bond lines with a
polyurethane (PU) adhesive and with an epoxy (EP) adhesive that are used, are
presented by Brandon (2010). The bond line stress distribution around one tube is
studied using a local model (Fig.4a). Bond line stress distribution in other areas is
studied in a global model (Fig.4b). Cohesive elements, that are able to describe a
reduction in strength with an increasing slip, are used to describe the bond line
behaviour. The steel and DVW are modelled as isotropic plastic materials and the
timber is modelled as an orthotropic elastic material. Failures or other significant
phenomena are estimated per material as follows: the strain of DVW exceeds
0.07; the stress in the timber perpendicular to grain is smaller than -3.6 MPa or
larger than 1.6 MPa; the stress parallel to grain is smaller than -42 MPa or larger
than 42 MPa; or the tube displacement exceeds 9mm. The modelled thickness of
the timber and the DVW are 60 mm and 15 mm respectively. The modelled tube
diameter (d) is 18 mm and the edge and end distances are 3.5d. In the global
model, the beam depth is 300 mm and the DVW plate is 300 mm square.
Applied bending moment

Applied loads DVW glued on timber
Steel plate
DVW glued
on timber
Steel tube
Symmetry
planes
Symmetry
plane
Non linear
load slip relationship
Steel plate
a) Local model b) Global model
Fig. 4 Geometry of finite element models
4 Experimental Approach
The experimental study aimed to verify the analytical and numerical predictions.
A number of two four-point bending tests (Fig.5) were performed. The depth of
the timber members was 300 mm. The thickness of each member was 60 mm and
was chosen so that the connection was the weakest link in the structure. The initial
length of the tubes before expansion was 16% longer than the connection thick-
ness and the expanded tube diameter was 18 mm. The distances between the
Fig. 5 Full scale connection tests of splice connection

supports and the load points and between the two load points were 1600 mm. Two
hydraulic jacks applied the load simultaneously and both loads were measured
with a force gauge. These loads were kept similar, resulting in practically zero
shear forces and pure bending moments at the location of the connection. The
rotation and translations were measured between the steel plate and the timber
members.
Fig.6 is obtained from the analytical study. It shows the bending moment-rotation
relationship of a single connection (grey curve) and flitch plate connections with
different initial gap sizes, tg (Fig.3) (black curves). The kink in the non-linear parts
of the black curves is caused by gap closure when DVWs compressive stresses
pick up. Current production practice in the Netherlands indicate that gaps are usu-
ally smaller than 4mm and, as such, moment-rotation curves are expected to run
between the two black curves on the furthest left. So although the influence of the
initial gap on the bending moment versus rotation is substantial, accurate assembly
resulting in a small initial gap, will make it a non-issue. Additionally it can be
seen that the flitch plate connection has the same rotational stiffness as a single
connection and an increased moment capacity, if the initial gap is small.
45 single connection
Bending moment (kNm)
40
35 flitch plate connection;
30 0 mm gap
25 flitch plate connection;
20 4.3 mm gap
15
flitch plate connection;
10 8.6 mm gap
5
flitch plate connection;
0
13 mm gap
0 0.05 0.1 0.15
Rotation (rad) flitch plate connection;
30.2 mm gap
Fig. 6 Influence of the initial gap size tg
Fig.7 shows the experimental, numerical and analytical results. Different num-
bered points on the numerically gained curve indicate significant points during
loading. At point 3 the tensile stress of the timber perpendicular to grain exceeds
1.6 MPa, resulting in cracking of the timber near the contact surface. However,
because of the glued-on DVW reinforcement, the size of the crack will be limited
and the structural effect will be insignificant. At point 4 the compression in the
timber will exceed 42MPa, resulting in yielding of the timber parallel to grain.
This will have a significant influence on the structural performance, since the
timber bears approximately 60% of the contact force. The dotted curve represent-
ing the numerical results with yield correction in Fig.7 is obtained by reducing the
bending moment after point 3 with the bending moment that is caused by the addi-
tional contact force in the timber after point 3. This contact force and its moment
arm are determined from the numerical results. The curve stops when the tube
deformation is 9 mm and tube failure occurs. It is concluded that bond lines with
PU and EP adhesives do not fail prior to the tubes in the flitch plate connection
with splice-type arrangement.
In Fig.7 it can be seen that the numerical study predicts the stiffness of the con-
nection well, but overestimates the strength of the connection. This is very likely
due to the yielding of the timber which is not taken into account in the model. The
analytical study gives a good prediction of the entire moment-rotation behaviour.
60
Bending moment (kNm)
50 3 Numerical results
2
40 Numerical results with
1 yield correction
30
Experimental results
20
10 Analytical results
0 Traditional connection
0 0.05 0.1 (Leijten, 1987)
Rotation (rad)
Fig. 7 Analytical, numerical and experimental results
6 Comparison with Conventional Moment Connections
This section aims to show the advantages of the DVW reinforced connection in
comparison with conventional connections. Results of Leijten (1987) who per-
formed tests of a conventional three member moment connection are used for the
comparison. The dimensions of the tested specimens are shown in Fig.8 and the
connection contained 10 steel dowels of 8 mm. The beam depth of the tested con-
nections was similar to the beam depth of the present experiments. The ends of the
beams were pulled inwards as shown by the arrows. Results of these tests are
shown in Fig.7. From this figure it can be concluded that the DVW reinforced
connection has a significantly higher moment capacity and the rotational stiffness
than the traditional connection. An even stiffer and stronger connection can be
obtained by using more tubes in one connection.
24
55
19
19
55
R12
0
30
30
0
30
0 0
15 30
150
0
Fig. 8 Traditional connection test dimensions (Leijten, 1987)
7 Conclusion
An analytical and numerical study to the relationship between bending moment

and rotation of a densified veneer wood (DVW) reinforced flitch plate connec-
tion with expanded tube fasteners is performed. Also, experiments are performed
to verify the analytical and numerical models.
The studied flitch plate connection consists of two single connections with a
steel plate as the middle member acting in series (Fig.1a and Fig.1c). The flitch
plate connection, however, can have a higher rotational stiffness and moment
capacity than two single connections in series. This is caused by the rotation sup-
pressing effect that comes to exist when the DVW members and the timber mem-
bers come into contact due to rotations.
The most important findings can be summarised as follows:
the moment capacity of the flitch plate connection (Fig.1c) exceeds the moment
capacity of the single connection (Fig.1a), under the condition that the timber
and DVW members are closely spaced;
under the same condition, the stiffness of the flitch plate connection exceeds
the stiffness of two single connections;
the analytical approach accurately predicts the bending moment-rotation rela-
tionship of the connection;
the numerical approach accurately predicts the stiffness, but overestimates the
moment capacity, which is likely due to the ignorance of timber yielding in the
model;
a comparison with a more conventional dowel connection showed the structural
quality of the DVW reinforced connection.
References
Brandon, D.: Mechanical properties of bond line between timber and densified veneer
wood, STSM report of COST E55. Short term scientific mission code: STSM-E55-
06127, Brussels, Belgium (2010)
Leijten, A.: Standard short duration tests on locally reinforced moment joints, 8 mm dowels
and bolts. Technische Universiteit Delft (1987)
Leijten, A.J.M.: Densified Veneer Wood Reinforced Timber Joints with Expanded Tube
Fasteners. PhD Thesis, Technische Universiteit Delft (1998)
Fully Threaded Self-tapping Screws Subjected
to Combined Axial and Lateral Loading with
Different Load to Grain Angles
Robert Jockwer1, Rene Steiger1 , and Andrea Frangi2

1 Empa, Swiss Federal Laboratories for Materials Science and Technology,
Structural Engineering Research Laboratory, Ueberlandstrasse 129,
CH-8600 Dubendorf, Switzerland
{robert.jockwer,rene.steiger}@empa.ch
2 ETH Zurich, Institute of Structural Engineering IBK, Wolfgang-Pauli-Strasse 15,
CH-8093 Zurich, Switzerland
frangi@ibk.baug.ethz.ch
Abstract. In order to benefit from the advantages of fully threaded self-tapping

screws as reinforcing elements, it is essential to have detailed knowledge about the
strength and stiffness of screws with different shank to grain angles subjected to
combined axial and lateral loading. In this paper a design model is proposed for
the calculation of the load-carrying capacity and stiffness of screws with different
shank to grain angles subjected to loads perpendicular to the grain by accounting for
the effective embedment of the screw in the timber. The proposed model is based
on commonly used material properties and other established design models and fits
well the influence of the angle between shank and grain direction on the joints
capacity and stiffness. However, detailed knowledge about the specific input param-
eters in the design approaches is necessary in order to achieve a reliable prediction
of the load-carrying capacity and stiffness of the individual screw.
Keywords: self tapping screws, fully threaded screws, inclined screws, embedment
length, design approach, load carrying capacity.
1 Introduction
The use of self-tapping screws increased in recent years due to the development in
materialization and shaping of the screws. In contrast to other types of fasteners
screws provide the advantage of load transfer into the wood not only perpendicular
to the shank of the screw but also in axial direction and thus of making optimal
use of the high pull-out capacity of the screw. For reinforcement purposes screws
can be used to reinforce the timber in shear and perpendicular to the grain. They
are used to strengthen timber structural elements loaded in compression and tension
perpendicular to the grain and in shear.
DOI: 10.1007/978-94-007-7811-5_25, c RILEM 2014
266 R. Jockwer, R. Steiger, and A. Frangi
At notches combined shear stresses and tensile stresses perpendicular to the grain
occur. Screws are wanted to reinforce notches against both these acting stress com-
ponents. Due to the brittle failure behavior of timber, the reinforcement should be
as stiff as possible. However, screws exhibit their highest stiffness in axial direction,
whereas stiffness in lateral direction is lower. Hence, the load carrying behavior (in
terms of strength and esp. stiffness) of screws set in different angles to the grain and
loaded in shear and tension is of focal interest.
In order to benefit from the advantages of fully threaded self-tapping screws as
reinforcing elements at notches it is essential to develop detailed approaches for the
estimation of strength and stiffness of screws with different shank to grain angles
subjected to combined axial and lateral loading.
2 Current Design Approaches for Inclined Screws

Beside geometric parameters (outer thread diameter d or d1 , inner thread diameter
dcore , effective penetration depth le f , edge distances and spacing) the load carrying
capacity of screws is influenced by the yield moment My , the embedment strength
of the timber fh and by the withdrawal strength fax of the thread.
The design equations for screws loaded in lateral direction in Eurocode 5 (EC5)
[1] are identical to the ones for other dowel type fasteners. They are based on inves-
tigations by Johansen [2]. The design approach for screws loaded in axial direction
is based on studies by Bejtka [3]. In the design approaches only pure lateral and
pure tensile loading of the screws are considered.
For combined lateral and tensile loads (Fv and Fax , respectively) in EC5 it is
referred to the interaction formula for the calculation of the load-carrying capacity
of threaded nails (Equation 1). In this interaction formula the load to grain angle is
not taken into account. Information about how to account for the stiffness of screws
subjected to combined lateral and axial loading is not given in EC5.
2 2
Fax,Ed Fv,Ed
+ 1.0 (1)
Fax,Rd Fv,Rd
Bejtka and Bla [4] propose a design approach for shear connections with in-
clined screws for different shank to grain angles . The resistance of a connection
with inclined screws results from the portions of the axial and the lateral resistance
of the screw (Rax and R , respectively) in direction of the interface of the members
and from the friction force between the members. Corresponding to failure mode 3
according to the Johansen model with plastic hinges in the screw, the resistance R
of a shear connection with one single inclined screw loaded in axial tension can be
calculated by Equation 2.
R = Rax ( sin + cos ) + R (sin cos ) (2)
Tomasi et al. [5] performed tests on shear connections with screws with dif-
ferent shank to grain angles . Connections with screws acting in tension, shear,
Fully Threaded Self-tapping Screws Subjected to Combined Axial 267
compression and combined tension and compression were studied. The approach
according to Equation 3 for calculating the stiffness Kser of the connection is pro-
posed, taking into account angle and both the stiffness in axial and lateral direction
(K and K , respectively).
Kser = K sin (sin cos ) + K cos (cos + sin ) (3)
In all these approaches the force acting on the connection runs in the parallel
to the grain direction of the member. This corresponds to a force to grain angle
of = 0 . However, when self-tapping screws are beeing used as reinforcement
for notched beams, combined parallel and perpendicular to the grain loading of
the cracked region occurs. This issue is not covered by the above equations but
will be further analysed in the following and an according design approach will be
proposed.
In equivalence to the model proposed by Bejtka and Bla [4] in Figure 1b a design
model for the calculation of the effective embedment of the screw was developed
as illustrated in Figure 1a. When an inclined screw is loaded perpendicular to the
grain in tension, the embedment stresses are acting in direction of the timber surface.
At the very surface of the timber the layer resisting the embedment loads has zero
thickness. As a result zero embedment stresses can be carried by this layer. With
increasing loads compression and splitting failure will occur. The full embedment
strength fh can be achieved at a distance x1 from the surface, at which the timber
layer perpendicular to the screw shank is thick enough to transmit the embedment
stresses along two shear planes at both sides of the screw. Depending on the screw
shank to grain angle rolling shear fv,roll is the dominating strength parameter. The
resulting length x1 along which only reduced embedment stresses can be transmitted
is given in Equation 4 and illustrated in Figure 1a.
f h de f
x1 = (4)
2 tan fv,roll
Based on Johansens model [2] the load-carrying capacity of an inclined screw

loaded perpendicular to the grain can be calculated using the reduced embedment
length. As a simplification, zero embedment strength is assumed within length x1
along the screw. The resulting lateral load-carrying capacity of the screw is given
in Equation 5. The withdrawal capacity of the screw can be calculated using the ef-
fective length reduced by x1 . The resulting load-carrying capacity on the joint with
an inclined screw loaded perpendicular to the grain can be determined according to
Equation 2. Due to the opening of the joint, no friction between the members will
occur ( = 0).
fh
fax
fax
fh
x
x2

Flat
Fshear
x1 Fax Fax

Flat
Fpull
(a) (b)
Fig. 1 Forces and stresses in joints with inclined screws loaded (a) perpendicular and (b)
parallel to the grain

R = f h x1 d e f + 2My + fh x21 de f fh de f (5)
The stiffness of the joint is markedly influenced by the embedment of the screw. As
a simplification, again zero embedment strength is assumed within length x1 along
the screw. As a further simplification, clamping of the screw beyond length x1 is
assumed for the stiffness in lateral direction K as given in Equation 6. The axial
stiffness is calculated using the effective length reduced by x1 .
3 Esteel dcore
4
K = (6)
64 x31
According to [4] the capacity of the joint consists of the portions of the capacity of
the screw in lateral and in axial direction. The stiffness of a joint loaded parallel to
the grain was proposed by Tomasi et al. [5] to be calculated as a sum of the portions
resulting from the stiffness in lateral and in axial direction. In contrast the stiffnesses
in the joint when loaded perpendicular to the grain are acting in series. The effective
stiffness Kser is equal to the sum of the inverse values of the stiffnesses in lateral and
axial directions (K and K respectively) (Equation 7):
1 1 1
= + (7)
Kser K K
4 Experimental Verification of the Proposed Model

Tests have been carried out in order to determine the capacity and the stiffness of
joints with screws with different shank to grain angles and loaded under different
load to grain angles. Within these test series, the parameters shank to grain angle
, load to grain angle and density have been varied. In two different test se-
tups three test series with = 90 , 60 and 45 were tested using two groups of
110 110

110
350
110
300
[mm]
280
(a) (b)
Fig. 2 Test-setups for (a) pulling and (b) shearing tests of joints with inclined screws
samples with high and low timber density. The test setup for the application of a
tension force perpendicular to the grain ( = 90 ) is illustrated in Figure 2a and
named pulling test. The second test setup for a force applied parallel to the grain
( = 0 ) is similar to a block shear test and given in Figure 2b. The tests were run
displacement controlled with constant speed of load. In order to reach ultimate load
within approximately 5 min a speed between 1.0 and 3.0 mm/min was chosen.
Norway spruce glulam of high and low density was produced out of two sam-
ples (each of around 100 boards) which have been selected from a bulk of over
500 boards with mean = 419 kg/m3 (CoV = 12.4%), k = 344 kg/m3 . The
bulk of boards had a density corresponding to solid timber of strength class C24
according to EN 338 [6] (mean = 420 kg/m3, k = 350 kg/m3). According to
prEN 14080 [7] the laminations of glulam of strength class GL24h correspond
to solid timber of strength class C24. The moisture content of the glulam speci-
mens when tested was approximately MC = 10%. The density of the glulam was
mean,high = 464 kg/m3(CoV = 2.5%) for the specimens with high density and
mean,low = 360 kg/m3(CoV = 1.8%) for the specimen with low density.
Fully threaded self-tapping screws of type SFS WR-T-13xL400 as specified in [8]
have been used in the tests. The screws have an outer thread diameter d1 = 13 mm,
an inner core diameter of dcore = 8.5 mm and a length of l = 400 mm. In the national
technical approval published by the Deutsches Institut fur Bautechnik [8] a yield
moment of the screws of My,k = 80.0 Nm is given. Furthermore the characteristic
withdrawal parameter perpendicular to the grain is specified as fax,k = 80 106 k2 .
Two parallel screws were used per joint. The distance between the screws perpendic-
ular to the grain was 60 mm or 4.5d and the edge distance perpendicular to the grain
was 40 mm or 3d. The effective length of the threaded part of the screw in the side
members of the block shear tests and the pulling tests was le f = 110 mm, 127 mm
and 155 mm or 8.5d, 10d and 12d for screw axis to grain angles of = 90 , 60
and 45 , respectively.
40 40
= 90 = 90
35 = 60 35 = 60
= 45 = 45
30 30
Load per screw [kN]

Load per screw [kN]
25 25
20 20
15 15
10 10
5 5
0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Opening displacement [mm] Shearing displacement [mm]
(a) (b)
Fig. 3 Joint behaviour in (a) pulling and (b) shearing tests. Individual results in grey, nor-
malised values for = 420 kg/m3 in black with solid line for = 90 , dashed-dotted line
for = 60 and dashed line for = 45 .
In the gap between the side members and the central member of the shear tests a
teflon foil of 1.5 mm thickness was placed in order to avoid frictional forces between
the members. With this foil inserted, the members were screwed together tightly.
The members of the pulling tests were screwed together tightly without such a teflon
foil.

The load-displacement diagrams are shown in Figure 3. Shearing and opening dis-
placements had been measured in the timber at the location of the screws. In order to
compare the series with high and low density the results are normalized with regard
to their density to a reference density of 420 kg/m3. The effective stiffness Kser is
determined in the range of 0.1 Fult and 0.4 Fult by linear regression following EN
12512 [9]. In EN 12512 a limit deformation of 30 mm is specified. However, due
to the intended application of the screws as reinforcement, a limit deformation of
15 mm was chosen. Thus, the ultimate load Fult was taken as the minimum of the
highest load and the load at 15 mm deformation.
The ultimate load Fult and the stiffness Kser normalised to a density =420 kg/m3
are given in Table 1. The results of the two test series in shearing and pulling with
= 90 serve as a reference in the common design approaches as specified in EC5
[1] and in Bejtka and Bla [4]. The required material strength parameters used in
these approaches can be found in literature: fh,k , fax,k and K in EC5, fh,mean and
fax,mean in Bejtka [3], My,k and K in [8]. For the calculation of the load-carrying ca-
pacity and the stiffness according to Equation 5 and Equation 7 the material property
values published in [3] and [8] were used.
Table 1 Test results compared to the proposed design model and to values from literature
Load-carrying capacity Stiffness

Tests EC5 [4] Eq. 5 Tests [5] Eq. 7
Fult Rint,k Rmean Rmean Kser Kser Kser
[ ] [kN] (CoV) [kN] [kN] [kN] [kN/mm] (CoV) [kN/mm] [kN/mm]
Pulling 90 21.2 (3%) 16.2 18.7 18.7 28.3 (25%) 17.9
60 21.6 (4%) 12.3 16.3 8.9 (15%) 6.5
45 17.7 (8%) 11.8 15.3 3.0 (17%) 3.4
Shearing 90 10.5 (5%) 6.8 6.2 1.9 (22%) 2.5
60 20.5 (4%) 8.6 15.0 5.9 (8%) 6.3
45 27.4 (10%) 11.8 20.3 13.6 (16%) 10.2
All design approaches give lower load-carrying capacities for = 90 compared

to the ultimate loads reached in the tests. The ratio of the mean values from tests
and the characteristic values of the capacities determined according to EC5 is in the
range of 1.3 to 1.5. Due to the low variation of the test results, the characteristic
value of the test results is expected to be higher than the characteristic value accord-
ing to EC5. The stiffness measured in the pulling tests for = 90 is larger than the
one calculated according to the national technical approval of the screws [8]. The
stiffness in the shearing tests for = 90 is lower than the one calculated according
to EC5.
The influence of the angle is modelled in a appropriate way by the proposed
approach of reducing the embedment length (Equation 5 and Equation 6). The de-
viation of the predicted load-carrying capacities and stiffnesses from the test re-
sults can be eliminated by using more precise input parameters valid for the specific
screws used in the tests instead of the general input parameters proposed by stan-
dards and literature: A better fit of the stiffness Kser in pulling is achieved when a
larger stiffness in withdrawal direction K = 40 d le f is used instead of K = 25 d le f
as proposed in [8]. The mean withdrawal strength as suggested by Bejtka ([3]) for
a variety of self-tapping screws is fax,mean = 13.1 N/mm2 wherereas for the screws
used in the tests a larger value of fax,mean = 14.8 N/mm2 was found.
The length x1 according to Equation 4 is calculated using the material property
values given in prEN 14080 and given by Bejtka [3]. A maximum length of approx-
imately x1 = 30 mm occurs in the pulling tests for = 45 . This length corresponds
well to the observations made in the tests.
6 Conclusions
Joints with fully threaded self-tapping screws show highest load-carrying capacity
and stiffness when the screws are loaded in axial direction. The more the screw is
loaded in lateral direction the lower are the capacities and the stiffnesses. By choos-
ing an appropriate angle between screw shank and grain direction, the behaviour of
the joints can be optimised. Existing design approaches do not cover inclined screws
loaded perpendicular to the grain. The model proposed in this paper gives a good
estimate of the structural behaviour of inclined screws loaded perpendicular to the
grain. By accounting for a reduced embedment length the strong decrease in stiff-
ness can be adequatly modelled. The reduction of the embedment length strongly
depends on the properties of the timber and the screw. The values for these prop-
erties given in literature differ strongly which leads to big variations in estimated
load-carrying capacity and stiffness. For a reliable estimate of capacity and stiffness
a precise knowledge of the actual properties is necessary. Precise property values
for the specific screw used in the joint should be used instead of general property
values for dowel-type fasteners as can be found in EC5.
References
1. European Committee for Standardization CEN, EN 1995-1-1: Eurocode 5: Design of
timber structures - Part 1-1: General - Common rules and rules for buildings + AC (2006)
+ A1 (2008). CEN, Bruxelles, Belgium (2004)
2. Johansen, K.W.: Theory of Timber Connections. IABSE Publications 1949(9), 249262
(1949)
3. Bejtka, I.: Verstarkung von Bauteilen aus Holz mit Vollgewindeschrauben. Ph.D.Thesis,
Fakultat fur Bauingenieur-, Geo- und Umweltwissenschaften, Karlsruhe University,
Karlsruhe, Germany (2005)
4. Bejtka, I., Bla, H.J.: Self-tapping screws as reinforcement in beam supports, Paper 39-
7-2. In: Proc. of CIB-W18 Meeting, Kyoto, Japan, vol. 39 (2002)
5. Tomasi, R., Crosatti, A., Piazza, M.: Theoretical and experimental analysis of timber-to-
timber joints connected with inclined screws. Construction and Building Materials 24,
15601571 (2010)
6. European Committee for Standardization CEN, EN 338: Structural timber - Strength
classes. CEN, Bruxelles, Belgium (2009)
7. European Committee for Standardization CEN, prEN 14080: Timber structures - Glued
laminated timber and glued solid timber - Requirements. CEN, Bruxelles, Belgium
(2011)
8. Deutsches Institut fur Bautechnik DIBt, Z-9.1-472, Allgemeine bauaufsichtliche Zu-
lassung, SFS Befestiger WT-S-6,5; WT-T-6,5; WT-T-8,2; WR-T-9.0 und WR-T-13 als
Holzverbindungsmittel. DIBt, Berlin, Germany (2011)
9. European Committee for Standardization CEN, EN 12512: Timber structures - Test
methods - Cyclic testing of joints made with mechanical fasteners + A1 (2005). CEN,
Bruxelles, Belgium (2001)
10. European Committee for Standardization CEN, EN 26891: Timber structures - Joints
made with mechanical fasteners - General principles for the determination of strength
and deformation characteristics (ISO 6891:1983). CEN, Bruxelles, Belgium (1991)
Alternative Approach to Avoid Brittle Failure
in Dowelled Connections
Daniela Wrzesniak1, Massimo Fragiacomo2, and Andr Jorissen3
1
DIA Department of Engineering and Architecture, University of Trieste,
Piazzale Europa 1, 34127 Trieste
dani_wrz@yahoo.de
2
DADU Department of Architecture, Design and Urban Planning,
University of Sassari, Piazza Duomo 6, 07401 Alghero
3
Eindhoven University of Technology, P.O. Box 513, 5600 MB, Eindhoven, and
SHR Timber Research, Wageningen, The Netherlands
Abstract. Ductile behavior of timber connections with metal fasteners is essential

to achieve a robust structure. Moreover, a ductile behavior of the fastener and the
timber prior to failure is necessary to fulfill the boundary conditions for applying
the Johansen theory (1949). If these boundary conditions are not fulfilled, the
capacity of a connection is overestimated and brittle failure may occur. However,
using sufficient spacing, end and edge distances reduces tensile stresses
perpendicular to the grain, the main stresses initiating brittle failure. If in addition
to that the influence of the relation between fastener diameter and timber volume
is considered, brittle failure mechanisms can be avoided.
This paper discusses quasi static tests carried out on dowelled connections with
different spacing and loaded end distances chosen in accordance to the above
criterion. The results were compared with the capacity calculation for dowelled
connections of Eurocode 5, chapter 8.2, the design approach against block shear
failure of Eurocode 5, Annex A, and the design proposal for the avoidance of
block shear failure proposed by Hanhijrvi and Kevarinmki (2008). The
experimental failure load achieved was in all cases higher than predicted values by
all three design approaches.
Keywords: dowelled connections, brittle failure, block-shear failure, splitting.
1 Introduction
Johansens theory, which was presented in 1949 and later extended by Meyer in
1957, is a widely used and acknowledged design method for timber connections
with metal fasteners. This theory is fully incorporated in Eurocode 5 (EC5, EN
1995-1-1:2004) and some other design codes.
DOI: 10.1007/978-94-007-7811-5_26, RILEM 2014
274 D. Wrzesniak, M. Fragiacomo, and A. Jorissen
It is well known that in Johansens theory a full-plastic behavior of the timber

and the fasteners is assumed. Both the dowel in bending and the timber in
embedding are assumed to attain the plastic state prior to failure.
For both materials this looks a logical assumption since also timber in
compression responses Eplastically. To satisfy these assumptions, Johansen noted
that basic design rules such as sufficient spacing between fasteners and sufficient
edge and end distances have to be taken into account.
For dowelled connections he suggested a minimum spacing between fasteners
of a1=10d, a loaded end distance of a3=7d and an edge distance of a4=3d (d =
dowel diameter) to ensure timber plasticization and the connection failure
according to his theory (Figure 1).
Fig. 1 Required spacing and distances between fasteners for dowelled connections
(Johansen, 1949)
Later studies refined and extended these design rules concerning the spacing,
edge and end distances, the slenderness ratio and the effective number of dowels
in a row (Jorissen, 1998). The slenderness ratio is defined by
t with t = timber
d
thickness and d = dowel diameter. An optimized slenderness ratio was suggested
by Mischler (1998):
,
1.4 (1)
,
with Mu,95 being the upper 5th percentile of the bending strength of the dowel and
fh,05 being the lower 5th percentile of the embedding strength of timber.
As indicated in Eq.(1) and pointed out by Mischler the strength of the steel,
spacing and distances should be related to the properties of the timber.
However, brittle failure in timber structures such as splitting along the row of
fasteners and tearing out of timber parts has been observed ever since. Great effort
was made by numerous scientists to describe especially the plug-shear and block-
shear failure mechanisms, which have been found inadequate in Annex A of
Eurocode 5 (Zarnani and Quenneville (2012), Johnsson (2003), Hanhijrvi et al.
(2006)). An alternative design approach was presented by Hanhijrvi and
Kevarinmki (2008), who suggested that the connection should be divided in inner
and outer parts, and their capacity should be calculated independently as indicated
in Figure 2.
Alternative Approach to Avoid Brittle Failure in Dowelled Connections 275
Inner Part
Outer Parts
Fig. 2 Example of a timber specimen with dowels indicating the division of inner and outer
parts after Hanhijrvi and Kevarinmki (2008)
Due to its complexity, the approach results in difficult application in common

practice.
An extensive literature research has been conducted thereafter by the writers,
realizing that researchers specifically dealing with block and plug shear failure
mechanisms came to similar conclusions for the avoidance of this failure
mechanism.
Kangas and Kevarinmki (1997 and 1998) established a model which optimizes
the distances and spacing between nails in timber joints. They also came to
the conclusion that in order to avoid block shear failure, the spacing between
nails and the joint area have to be increased to achieve the full capacity of the
nails.
Johnsson (2003) suggested that in order to avoid plug shear failure the joint
width should be increased and the number of fasteners in a row should be
decreased.
Kairi (2004) came to the conclusion that slenderness ratio and number of
dowels should be optimized such that fastener yielding is attained before timber
failure occurs.
Although the brittle failure mechanisms dealt with in the above literature were
mainly related to nailed connections, the authors agreed upon the application of
basic design principles to all fastener types such as fastener spacing, diameter and
timber volume to avoid brittle failure mechanisms like plug shear or block shear
failure.
Based on the literature research it was decided to conduct tests on dowelled
connections aiming to show that considering basic design rules, brittle failure
mechanisms like those described by Hanhijrvi and Kevarinmki (2008) and in
Annex A of Eurocode 5 can be prevented.
276 D. Wrzesniak, M. Fragiacomo, and A. Jorisseen
T
35 steel-timber-steel dow wel connection specimens were tested to failure at thhe

laboratory of the Neue Holzbau
H AG in Lungern, Switzerland. The test set up annd
load apparatus is presenteed in Figure 3.
Load
direction
Steel plates
fixed to both sides
of the specimen
S235, t=5mm
Timber Board
ft,mean = 23N/mm^2
Connecting
Steel Block
Front View
Specimen Load
direction
Fig. 3 Test Set-up and Load application
2.1 Geometrical Prroperties

The specimens were consstructed with either one or two rows of 12 mm diameter
dowels. The number of dowels
d in a row as well as edge and end distances varieed
(Figure 4). A maximum of 4 dowels per row were used. The fastener spacing a1
was 7.5d which is less th
han suggested by Johansen but higher than the minimum m
values required accordingg to Eurocode 5 (5d for forces parallel to the grain) annd
Swiss standard SIA 265 (7d). The distance between dowels perpendicular to thhe
grain was chosen to be 5dd (minimum values according to Eurocode 5 and SIA arre
both 3d).
An additional test wass carried out with 4 dowels, a spacing a1 of 15d and a
loaded end distance a3 off 12d. The aim was to investigate the influence of endd-
distance and spacing on the load carrying capacity and the embedment strengtth.
The timber board thickneess is 40 mm. Detailed specimen properties are presenteed
in Table 1.
a3 90 90 Varies 90 90 a3
a4 Varies
a4= 30, 40, 60, 75mm

a3= 30, 40, 50, 60, 70, 90, 120mm
Varies
a4= 30, 40, 60, 75mm

a3= 30, 40, 50, 60, 70, 90, 120mm
Varies
a4= 30, 40, 60, 75mm

a3= 30, 40, 50, 60, 70, 90, 120mm
Varies
a4= 30, 40, 60, 75mm

a3= 30, 40, 50, 60, 70, 90, 120mm
12 Dowels
fu,k=1000N/mm^2
Timber Boards
of ft,mean=23N/mm^2
Fig. 4 Detailed layout of test specimens
2.2 Material and Loading Protocol

Ten boards of European spruce with similar grain pattern, density and moisture
content were visually selected out of 100 boards. The Youngs Modulus was
determined by measuring the sound wave velocity, which ranged from 5850 to
6150 m/s. The timber boards had a characteristic tensile strength ft,0,k = 20 N/mm2
according to their strength grade.
Knots and defects such as compression wood were cut off to achieve
representative results. The density of the boards was measured at a moisture
content of w 10% varied between 420 and 500 kg/m3 and 12 mm diameter
dowels of high strength steel (ETG 100, fu = 1000 N/mm2, fy = 900 N/mm2),
smooth surface, and steel plates (S235) with a thickness of 5 mm were used for the
tests. The holes in the steel plates were oversized by 0.5 mm, whereas the holes in
the boards were drilled with diameter of 12 mm. The tensile force was applied by
a hand pump (ENERPAC). The pump speed was applied to reach failure after
approximately 5 minutes.
3 Test Results
3.1 Observations and Analysis of Failure Loads

The slenderness ratio of the dowel was 3.33 in all cases. With a steel side plate
thickness of 5 mm and a dowel diameter of d = 12 mm the steel plates can be
regarded as relatively thin:
8 4 4 900
< = 6,2
, 3 , 3 30
with fh,k being the embedment strength according to Eq. (2). The fasteners can be
regarded stocky (Jorissen, 1998). Stocky dowels of high strength class were
chosen to avoid any ductile behavior of the fastener in order to verify the approach
presented by Hanhijrvi and Kevarinmki (2008/9).
The observed failure mechanisms were all brittle, such as row splitting, cracks
along the row of the fasteners, tensile failure of the cross section or side parts of
the timber. However, for large spacing (up to a1= 12d) brittle failure was observed
after large deformations, indicating that the brittle failure mode is not governing.
Fig. 5 Photos showing different failure mechanisms, from top: 1st and 2nd =Splitting and
Cracks along the row, 3rd =Row Shear combined with tensile failure, 4th =Tensile failure
Table 1 Summary of Specimen Properties and Test Results

Test Results
One Row a4=30, 4Dowels
No. of
Name of Dowels Number Failure Failure
Specimen Distances in row of rows Density Load Mode
a4 a3
mm mm kg/m^3 kN
30/30-1 30 30 468.0 35.2 SC
30/50-1 30 50 464.4 65.7 SC
4 1
30/70-1 30 70 458.6 64.5 T
30/90-1 30 90 447.5 63.9 Td
One Row a4=40mm, 4Dowels
40/30-1 40 30 430.9 20.3 I/C
40/50-1 40 50 452.3 55.6 SCd
40/60-1 40 60 478.5 67.3 SCd
4 1
40/70-1 40 70 494.8 63.7 SC
40/90-1 40 90 447.7 66.1 SCd
40/120-1 40 120 503.9 69.1 T
60/40-1 60 40 524.3 38.3 SC
60/50-1 60 50 472.2 44.3 SCd
4 1
60/70-1 60 70 466.1 61.7 SCd
60/90-1 60 90 451.7 72.8 Scd
75/60-2 75 60 479.8 - I/C
4 1
75/120-2 75 120 470.9 157.3 T
Two Rows a4=40mm, 4Dowels
40/60-2 40 60 4 2 500.8 122.5 SC
60/60-2 60 60 481.0 62.5 RST
4 2
60/120-2 60 120 472.0 114.6 RST
One Row a4=30, 3Dowels
30/40-1.1 30 40 454.1 33.5 SCd
30/50-1.1 30 50 497.2 48.5 SC
30/60-1.1 30 60 469.0 53.2 T
3 1
30/70-1.1 30 70 484.0 49.1 SCd
30/90-1.1 30 90 447.3 39.1 SCd
30/90-1.1b 30 90 456.6 53.4 T
40/60-1.1 40 60 440.0 55.3 Sd
3 1
40/90-1.1 40 90 459.0 56.1 Sd
60/60-1.1 60 60 451.0 44.9 S
3 1
60/90-1.1 60 90 436.0 53.1 S
40/60-2.1 40 60 469.0 87.0 RST
40/90-2.1 40 90 3 2 464.0 106.3 RST
40/120-2 40 120 489.7 104.2 Td
45/60-2.1 45 60 456.0 76.9 RST
3 2
45/120-2.1 45 120 453.0 103.3 RST
Table 1 contains the geometrical and mechanical properties of the specimens,

and the related failure load and failure mechanisms. The following abbreviations
are used: SC=splitting and d/or cracks along the row, T=tensile failure of the timbeer
cross-section, RST=row w shear combined with tensile failure, I/C=teest
interrupted/test cancelled d. Furthermore, an index d (from primarily ductile) is
added to the failure modees which first showed relatively large timber deformatioon
before brittle failure (e.g. SCd, Td and RSTd).
The different failure modes
m are displayed above (Figure 5).
The experimental failu ure load of each specimen per dowel was then compareed
with the embedding streength which was calculated using the semi empirical
formula of Eq. (2), Euroco ode 5.
, = 0.082 (1 0.01 ) (22)
The calculated embeddin ng strength was then turned into a load per fastener bby
multiplication with timbeer board thickness t and dowel diameter d and compareed
with the load per fasteneer achieved during testing. The results are presented iin
Figure 6 and Figure 7.
Note that the embeddiing strength of each specimen was calculated using thhe
density of each specimen obtained during testing.
It was realized that thee embedding strength of the timber was reached when thhe
loaded end distance a3 waas greater than 60 mm (5d) (Figure 6) or the a3/a4 rattio
was greater 1.5 (Figure 7)).
This indicates that althhough stocky dowels were used, if distances and spacinng
are kept sufficiently larrge, the maximum timber capacity in regards to thhe
embedding strength can sttill be reached.
Fig. 6 Comparison between n experimental failure load per dowel and embedding strenggth
(based on EC5) vs. end distaance a3
Alternative Approach to Avo
oid Brittle Failure in Dowelled Connections 2881
Fig. 7 Comparison between n experimental failure load per dowel and embedding strenggth
(based on EC5) vs. a3 / a4 rattio
Furthermore, the failure load was compared with the design methods for thhe
avoidance of brittle failurre mechanisms due to block shear described in Eurocodde
5, Annex A and by the forrmulas presented in the research report VTTS-07046-009
(Hanhijrvi and Kevarinm mki 2009).
The formulas of Euroccode 5, Annex A for the avoidance of block shear failurre
are:
1.5 , , ,
, = (33)
0.7 , ,
with , and , be
eing the net areas of the cross-section perpendicular tto
the grain and parallel to o the grain and , , and , being the characteristtic
tensile and shear strengthh of the timber respectively. The results are presented iin
Figure 8a to 8d. Altho ough the failure mechanism due to the unfavorabble
relationship between dow wel diameter and board thickness were brittle, the desiggn
approaches discussed ab bove underestimated the load-carrying capacity of thhe
connection by far. The vaalues obtained during testing were on average three timees
higher than predicted usin ng the Finnish approach and twice as high as the valuees
calculated based on the Annex
A A of EC5.
The test results in Figgure 8.d show higher results for the connection with aan
edge distance a4 of 40mm m. A reason for this is assumed to be a more evenly loaad
induction into the group of
o fasteners due to a better relationship between a3 and a4,
that is a3 and a4 being moore equal to one another. In addition, the test results werre
compared with Johansens equations of Eurocode 5 concerning double shear steeel-
to-timber connections witth thin steel plates as presented in Eq. (4).
Fig. 8.a
Fig. 8.b
Fig. 8.c

Fig. 8.d
Fig. 8.a to 8.d Comparison between observed and predicted failure loads using EC5, EC5
Annex A, and Hanhijrvi and Kevarinmki (2008/9) approach versus end distance a3. Fi
indicating the results based on Hanhijrvi and Kevarinmki, a4 is the edge distance a
0.5 , ,
, = (4)
1.15 , , , +
The prediction of the load carrying capacity is the minimum of the values
obtained by the formulas presented in Eq. (4) and being the characteristic
withdrawal capacity of the fastener.
The approximate analytical equation of EC5 was used to calculate the yield
moment:
, = , 0.3 . (5)
with d signifying the dowel diameter of 12mm and , being its characteristic
tensile strength of 1000 / .
The failure load based on the embedding strength was always governing. This
is due to a low slenderness ratio and the great strength of the dowels. The results
are presented in Figure 8.a to 8.d and it was found that the failure load is predicted
well by the Eurocode 5 equations, which are always conservative.
3.2 Analysis of the Effective Number of Fasteners

The effective number of dowels was calculated for each specimen by dividing the
failure load per dowel by the load per dowel derived from the embedding strength
as described in Chapter 3.1.
This number was then compared with the effective number of dowels
calculated using the Eurocode 5 formula, Eq. (6):
= .
(66)
where n is the actual number of fasteners. To obtain a representative value thhe

edge distance for the Eurrocode value was set to 80mm as recommended in Tabble
8.5, Eurocode5. In the serries with one row of 4 dowels and one row of 3 dowells,
the effective number off fasteners was greater than the effective number oof
fasteners calculated accorrding to the Eurocode 5 and equal to the actual number oof
fasteners when the conditiion a35d is satisfied (Figure 9).
Fig. 9 Effective number of fasteners

f neff versus end distance a3 for specimens with one roow
of 4 dowels
In the series with two rows

r of 3 and two rows of 4 dowels the effective numbeer
of fasteners was always greater
g than the effective number of fasteners calculateed
according to Eurocode 5 when
w a3 was greater than 5d (Figure 10).
Fig. 10 Effective number of fasteners neff versus end distance a3 for specimens with tw
wo
rows of 3 dowels
Therefore, the effective number of dowels in a connection can be increased by

simply increasing the end and edge distances as well as the spacing between
dowels.
3.3 Analysis of Embedding Capacity

In the following, the influence of fastener spacing, end and edge distances on the
embedding strength was investigated. During testing it was observed that
specimen 75/120-2 which was constructed with two rows of four dowels and an
end and edge distance of 120 mm (10 d) and 75 mm (6.5d) respectively, failed at a
load which was about 2.4 times higher than predicted based on Eurocode 5, for
which the embedment strength was calculated according to equation (2) using the
actual density.
An additional test with four bolts (two rows of two dowels), an end distance of
12d (144 mm) and a spacing a1 of 15d (180 mm) was carried out. At a spacing of
a1 of 15d, no interaction between dowels is expected.
Fig. 11 Plastic deformation in timber
As a result, the failure load reached in the test was Fmean=22 kN per dowel and
the deformation in the timber was around 5 mm (Figure 11) prior to failure. Based
on EC5, Eq. (2), using a characteristic timber density of = 0.7 = 479
0.7 = 368.5 / , the estimated embedding strength is 26.59 N/mm2.
Converting the experimentally obtained failure load per dowel into the embedding
strength by dividing the load by the timber board thickness and the dowel
diameter, the value obtained is 32.08 N/mm2.
The difference between the two values is considerable.
The derivation of the embedding strength is defined in EN 383 (Chapter 6,
Table 1) and provision is made for the use of an end distance of 7d. This would
indicate that current values for the embedding strength are only valid for an end
distance of 7d. Increasing end distances and bolt spacing leads to a higher load
carrying capacity and, consequently, to higher embedment stresses.
Therefore, the embedment strength seems to be not only dependent on the
density and the diameter of the fastener but also on the design of the connection.
However, further analyses of tests already carried out and probably additional tests
are necessary to confirm this assumption.
4 Conclusions
The aim of this study was to show that focusing on simple design principles
concerning the end and edge distances and spacing is sufficient to achieve a
ductile failure mechanism, hence fulfilling the requirements of the Johansen
theory and therefore predicting the capacity of timber connections with dowel
fasteners accurately. Since the slenderness ratio was kept constant ( = 3,33) the
expected positive effect of a large slenderness ratio on the connection ductility is
not studied.
The test results presented in this paper provided evidence that by applying the
above principles, splitting is delayed and ductile failure mechanisms can be
attained. It was demonstrated that:
1. The embedding strength of the timber based on EC5 semi-empirical formulas

was reached.
2. The effective number of fasteners was equal to the actual number of fasteners.
3. The load reached in the tests was always higher than predicted using the
Finnish proposal for block shear failure and according to Eurocode 5, Annex
A.
4. The load carrying capacity increased in respect to the value calculated using
the EC5 formulas, indicating that the embedding strength may not only be
dependent on the timber density, the fastener yield capacity and the diameter
of the fastener but also on the layout of the joint, i.e. fastener spacing end and
edge distances.
The design method of Eurocode 5 - Chapter 8.2-, based on Johansens equations

predicted the load carrying capacity accurately. This proves that applying the
conditions on spacing and end and edge distances already set by Johansen, this
theory is a reliable design tool for the capacity calculation of connections with
dowel fasteners. However, it should be noticed that the values for spacing, end-
and edge distances must be considerably larger than the minimum values
according to Eurocode 5.
Furthermore, it is suggested that the strength class of timber and the timber
thickness should be correlated with the steel grade and the fastener diameter. To
transfer large tensile forces, alternative options like glued-in rods or screws
inserted parallel to the grain like the SFS system should be considered.
A pilot test program was conducted which will have to be extended by
additional tests to draw conclusions which are not only qualitatively but also
quantitatively representative.
Acknowledgements. The authors would like to express their gratitude to Neue Holzbau
AG Lungern, Switzerland who provided the laboratory facilities and the material which
made the tests possible. Special thanks goes to Prof. E. Gehri for the supervision of the
tests, the background knowledge provided, and the assistance in writing this paper. His kind
support and technical advice are very much appreciated.
References
Eurocode 5 - Design of timber structures - Part 1-1, General-Common rules and rules for
buildings. EN 1995-1-:2004 (E) (2004)
EN 383 Timber structures - Test methods - Determination of embedment strength and
foundation values for dowel type fasteners (2007)
Hanhijrvi, A., Kevarinmki, A.: Timber failure mechanisms in high-capacity dowelled
connections of timber to steel - Experimental results and design, VTT Publications 677,
Espoo, Technical Research Centre of Finland (2008) ISBN 978-951-38-7090-4
Hanhijrvi, A., Kevarinmki, A., Koski, R.Y.: Block Shear failure at dowelled steel-to
timber connections. In: CIB 39-7-4, Working Commission W18 Timber Structures
Meeting, Florence, Italy (2006)
Johansen, K.W.: Theory of timber connections. International Association of Bridge and
Structural Engineering 9, 249262 (1949)
Johnsson, H.: Plug Shear Failure in nailed timber connections. In: CIB 36-7-2, Working
Commission W18 Timber Structures Meeting, Colorado, United States (2003)
Johnsson, H., Parida, G.: Prediction model for the load-carrying capacity of nailed timber
joints subjected to plug shear. Materials and Structures 46(13) (2013) ISSN: 1359-5997
Jorissen, A.J.M.: Double shear timber connections with dowel type fasteners. Doctoral
Thesis, University of Delft, Delft (1998)
Jorissen, A.J.M., Fragiacomo, M.: General notes on ductility in timber structures.
Engineering Structures 33(11), 29872997 (2011)
Kairi M.: Block Shear failure tests with dowel-type connection in diagonal LVL structure.
In: CIB 37-7-7, Working Commission W18 Timber Structures Meeting, Edinburgh,
Scotland (2004)
Kangas, J.: Design on timber capacity in nailed steel-to-timber joints. In: CIB 31-7-4,
Working Commission W18 Timber Structures Meeting, Savonlinna, Finland (1998)
Kangas, J., Kevarinmki, A.: Modelling of Block tearing failure in nailed steel to timber
joints. In: CIB 30-7-2, Working Commission W18 Timber Structures Meeting,
Vancouver (1997)
Kevarinmki, A.: Design method for timber failure capacity of dowelled and bolted
glulamconnections, Research report No VTTS0704609, Finnish Glulam Association,
Helsinki, Finland (2008)
Meyer, A.: Die Tragfhigkeit von Nagelverbindungen bei statischer Belastung. Holz als
Rohund Werkstoff, 15Jg. Heft 2, pp. 96109 (1957)
Mischler, A.: Bedeutung der Duktilitt fr das Tragverhalten von Stahl-Holz-
Bolzenverbindungen. Dissertation Nr. 12561 ETH Zuerich (1998)
NTC - Norme Tecniche per le Costruzioni. Decreto del Ministero delle Infrastrutture e dei
Trasporti D.M. (2008)
Quenneville, P., Bickerdike, M.: Effective in-row capacity of multiple-fastener connections.
In: CIB 39-7-1, Working Commission W18 Timber Structures Meeting, Florence
(2006)
Schoenmakers, J.C.M., Jorissen, A.J.M.: Failure mechanisms of dowel-type fastener
connections perpendicular to grain. Engineering Structures 33(11), 30543063 (2011)
Resistance and Failure Modes of Axially Loaded
Groups of Screws
U. Mahlknecht1,*, R. Brandner1,2, A. Ringhofer2, and G. Schickhofer2
1
Competence Centre holz.bau forschungs gmbh | Inffeldgasse 24, 8010 Graz, Austria
u.mahlknecht@tugraz.at
2
Graz University of Technology, Institute of Timber Engineering and Wood Technology
Inffeldgasse 24, 8010 Graz, Austria
Abstract. Screwed connections provide high resistance in strength and stiffness.

Arranged to a group the screws interact and influence each other in dependency of
their in-between spacings. A test setup was found to investigate (i) the influence of
the spacings in-between the screws, and (ii) the anchoring depths on the failure
modes and resistances of groups of screws. We conducted tests on axially loaded
and under a stress-fiber angle of 90 placed groups of screws in solid timber (ST)
and glued laminated timber (GLT) of Norway spruce. Steel fracture, withdrawal
failure and also block shear failure mode, till now for self-tapping screws not
considered by design codes, were observed. Additionally and based on a simple
mechanic load shearing consideration model for the block shear failure mode was
developed for the investigated axially loaded groups of screws. Verification with
test results confirms congruent but conservative results.
Keywords: groups of screws, failure modes, block shear failure, minimum

spacings, load shearing.
1 Introduction
Latest screw technologies offer multifaceted possibilities to connect timber mem-

bers in load bearing systems. Screw joints allow to submit high loads together
with high stiffness. Joint design depends on the magnitude and direction of the
internal forces, and on the dimension and material of the jointed members. Timber
to timber and recently steel-timber connections with outer steel plates are common
for load transfer by a few dozen or even hundreds of screws. Joints with more than
200 screws have been already used to transfer loads of more than 2 MN. To secure
sufficient reliability of these large joints the need for investigations on groups of
screws focusing on their interaction is obvious. In doing so some failure modes
can be clearly distinguished.
*
DOI: 10.1007/978-94-007-7811-5_27, RILEM 2014
290 U. Mahlknecht et al.
1.1 Failure Modes of Groups of Screws, Screwed Joints

Self-tapping screws are optimized for axially loading. The resistance of a screwed
joint may either be limited by the jointed timber members, e.g. by net cross sec-
tion failure, by tension stress perpendicular to the grain or interactions with shear
stresses caused by the local load insertion, by withdrawal, or by the screw itself.
Thus steel fracture, withdrawal and head pull-through failure modes can be distin-
guished. These are well known and explicitly considered by limit state verifica-
tions provided e.g. in EN 1995-1-1 (EC 5). This standard gives also regulations for
groups of screws involving minimum spacings and an effective number of screws
given as nef = n0.9. Of course by testing the occurrence of quasi brittle failure
modes (i) splitting of each row of fasteners along the grain, and (ii) block shear
failure of the whole group was observed, although the regulations concerning the
minimum spacings of EC 5 were met (Plieschounig (2010) and Mahlknecht
(2011)). The interaction of the fasteners in the group to an amount not considered
so far is found to be responsible for these failure modes.
Usually the minimum spacings in-between the fasteners and to the edges of
jointed members are given as multiple of the screw diameter d. According to EC 5
the minimum spacing in-between two screws in grain and perpendicular to grain
are given as a1 = 7 d and a2 = 5 d, respectively, and that in grain to the end grain
and transverse to the edge as a1,CG = 10 d and a2,CG = 4 d, respectively.
1.2 State-of-the-Art Concerning Groups of Fasteners

Kevarinmki (2002), Bejtka and Bla (2002), Bla et al. (2006) and Krenn and
Schickhofer (2009) studied tension joints with inclined screws of combined axial
and lateral loading. In particular Bejtka and Bla (2002) and Bla et al. (2006)
studied tension joints where two timber members got connected by inclined
screws. Firstly by one row of crossed screw pairs (one in compression and the
other in tension), and secondly a row of parallel arranged tension screws
d = 8 mm). Withdrawal or pull-through failures at spacings in-between of
200 mm, without reduction in resistance its compared with one screw pair or
single screw, respectively, was observed (nef = n). Kevarinmki (2002) proposed
for parallel arranged tension screws a1 = 8 d based on similar findings. For high
loaded tensile joints with lateral outer steel plates Krenn and Schickhofer (2009)
proposed to overlap the tips of the screws within the jointed members to avoid
failures in tension perpendicular to grain.
Gehri (2001) investigated the behavior of groups of 4 and 9 threaded rods at a
stress fiber angle of 0 and observed the same failure mode and resistance as in
single rods so far the amount of surrounding timber is comparable.
Splitting failure of groups of fasteners has been reported only on joints with
other fasteners than screws. Ehlbeck et al. (1989) and Schoenmakers (2010) exam-
ined groups of dowel type fasteners loaded perpendicular to the grain, which pro-
voked failure of the timber members in tension perpendicular to the grain.
Resistance and Failure Modes of Axially Loaded Groups of Screws 291
Stahl et al. (2004), Hanhijrvi et al. (2006) and Zarnani and Quenneville (2012)
analysed the block shear behavior of rivet fasteners transmitting a load parallel to
the grain. Hanhijrvi et al. (2006) reviewed the block shear verification methods
given in EC 5 (Annex A) testing groups of dowel fasteners and observed that just
block shear failure and no plug shear of each dowel row occurred. Thus an inter-
action between the dowel rows must be present. Stahl et al. (2004) proposed an
evaluation model based on the capacity of wood failure planes around the group,
in which the plane with the highest resistance governs the joint resistance. Also
the model of Zarnani and Quenneville (2012) is based on the capacity of fracture
planes by considering load shearing between the planes according to their stiff-
ness. This model is further discussed in chapter 3.
In contrast to EC 5 some European Technical Approvals (ETAs) for screws al-
low lower minimum spacings of a1 = 5 d or a2 = 2.5 d given that a1 a2 25 d.
These minimum spacings base on a test procedure of CUAP 06.03/08, which base
on screwing-in tests developed by Bla and Uibel (2009). Of course with this
procedure the splitting along the grain can be examined by screwing-in and not by
loading, which in fact also provokes splitting.
In fact quasi brittle failure modes can be avoided so far adequate minimum
spacings and anchoring depths are provided. Of course Plieschounig (2010) and
Mahlknecht (2011) showed that the spacings given in EC 5 and ETAs are too
small to exclude them. In addition the parameter anchoring depth is until now not
included in any regulation. As it is generally required to design joints tightly ar-
ranged, simply extending the spacings may not be meaningful for all circum-
stance. Thus a verification method is requested to determine the resistance of
groups of screws including block shear.
2 Tests of Groups of Axially Loaded Screws
We investigated the behavior of groups of axially in tension loaded screws at a

stress-fiber angle of 90. To avoid possible homogenisation effects on the with-
drawal behavior of screws in cases where more than one timber lamination is pen-
etrated (e.g. GLT), the first test series were executed in solid timber (ST). As the
dimensions of ST are limited smaller anchoring depths were investigated. In GLT
higher cross sections can be tested. The anchoring depths were defined equivalent
to that of single screws showing (i) clear steel fracture, as well as (ii) clear with-
drawal failure. For classification of failure modes the spacings between the screws
along and perpendicular to the grain were varied.
To account for the tip-effect of screws the proposal of Pirnbacher et al. (2009)
to subtract 1.17 d from the threaded length in the timber to get the effective length
lef was applied. Also the effect of embedding of the thread on the withdrawal ca-
pacity fax reported in Pirnbacher et al. (2009) was considered. Pilot tests in ST
showed an optimal resistance at an embedding depth temb = 4 d.
Limited by the testing device investigations were done with screws of
d = 6 mm, although in timber engineering diameters of 8 to 12 mm are more
common. Groups of m n are defined by m screws along and n screws perpendicu-

lar to the grain.
2.1 Test Procedure and Setup

The screws were inserted through an almost rigid steel plate providing a matrix of
holes for variation of in-between spacings, see Fig. 1. Every screw was pre-
tensioned with a torque wrench and by a twisting moment of 6 Nm. After some
cycles of pre-tensioning a nearly homogeneous load distribution within all screws
and an equal embedding of all threads could be assumed. During loading the spec-
imen in the push-pull loading setups was pressed on a second rigid but fixed steel
plate with a rectangular opening, large enough to allow local deformations of the
timber surface surrounding the edge screws. The length and width of the opening
was adapted for each variation in spacings and group size. The distances between
the supporting edge and the boundaries of the group were chosen equal to the
spacings in timber a1 and a2, see Fig. 1.
Reference tests were done with single screws of the same type, with the same
anchoring depth and in the same timber soureces. Tests on groups in ST were
way-controlled according to EN 1382 using a constant velocity of 4.5 mm/min.
The maximum load Fmax was reached after 180 60 sec. Tests stopped after a
decrease of 30 % from Fmax or a deformation 10 mm. Also the tests with GLT
were way-controlled, but accomplished according to EN 26891, with 1 mm/min
and stopped at F 0.8 Fmax | t.
Fig. 1 Test setup: upper loading plate with hole matrix and the fixed supporting steel plate
with a rectangular opening
2.2 Test Material

ST and GLT of Norway spruce (Picea abies) was used. The cross section of ST
(C24 according to EN 338) was approximately 163 x 252 mm and that of GLT
(GL24h according to EN 1194) 240 x 180 mm. To provide sufficient supporting
area against compression failures perpendicular to the grain during the test, the
specimens were laterally strengthened by edge bonded boards or smaller GLTs. In
particular the ST specimens were strengthened in depth to provide sufficient stiff-
ness against bending deflection. The serial spacings to the end grain and to the
edge of the glued specimen were > 18 d = 108 mm. For the tests in ST self-
tapping partially threaded screws 6x300/75 from Schmid Schrauben Hainfeld
GmbH were used and for tests in GLT with lef = 17.8 d and 28.3 d a special screw
with a longer shank and/or a longer threaded length were used.
Table 1 Test setup: group of screws with a stress-fiber angle of 90, d = 6 mm
mean of Fmax (*or F1st if available) [kN]
failure # W / # S
specimen length
12,mean [kg/m]
lef / d (temb / d)
source, height,
and (# failure mode)

mean density
single screw
width [mm]
a1 = 10.0 d
a1 = 12.5 d
a1 = 5.0 d
a1 = 7.5 d
[mm]
a2 / d
mn
ST 400- 406 5x5 50 /0 3.5 122* (10B) 141* (9B) 157* (10B) 151*
163 600 11.2 (9B)
256 (4)
GLT 350 453 3x4 0/41 2.5 170/182 (8B/2S) 164/183 (4B/5S) 182/184 (1B/4S) -
240 28.3 3.5 184 (5S) 184 (2S) 184 (2S) -
180 5.0 184 (5S) 185 (2S) 184 (5S) -
GLT 350 453 3x5 17/3 3.5 120* (8B) 141* (8B) 138*/165(7B/1W) -
240 17.8 5.0 124/151 134/162/182 135/156 -
180 (1B/4W) (1B/1W-2M/1S) (3B/2W-1M)
7.0 165(3W-3M) 166(4W-2M) 167(5W-1M) -
Wwithdrawal, Bblock shear, Mmixed failure mode, Ssteel fracture
2.3 Analysis of Failure Modes

After testing each specimen was cut along and perpendicular to the grain deter-
mining cracks and their position. With the developed test-setup only block shear,
withdrawal, screw steel fracture and interactions of these failure modes were
observed.
Steel Failure Mode (S): Testing GLT of lef = 28.3 d and a2 3.5 d all screws
failed by steel fracture, see Fig. 2 a).
Withdrawal Failure Mode (W): In specimen this was only observed at
a2 = 7 d, see Fig. 2 b) and in some of the specimen with a2 = 5 d.
Block Shear Failure Mode (B): In both, ST and GLT specimen, cracks in
depth at the position of the screw tips limited by the boundary screw rows along
the grain were located. Crack planes appeared also in-between both outer screw
rows along the grain at the same depth, see Fig. 2 c). In some GLT specimens a
crack caused by tension perpendicular to grain at the plane of screw tips and an-
other one between the first and second lamella was recognised. The order of crack
formation is difficult to clarify. Of course a shear failure in timber was not
observed visually.
Mixed Failure Modes (M): Three of the 20 tested single screws in GLT at
lef = 18.3 d failed by steel fracture. Even in the group tests steel failure occurred,
but of course, in none of the tests of the whole group. These tests are further allo-
cated to the mixed failure mode. Local stiffness differences caused by knots and
other growth characteristics are seen as possible reasons.
Fig. 2 Specimens with a) steel b) withdrawal and c) block shear failure mode
2.4 Discussion
The load-deformation curves of the tests done in ST and some of them in GLT
with lef = 18.7 d showed a partial failure followed by load redistribution and partly
by a further load increase, see Fig. 3 a). The partial failure caused a reduction of
the initial stiffness from K1 to K2. A partial failure was classified as a first failure
F1st in cases of F2 / F1 0.96 and K2 / K1 0.9. In some tests an increase in load
was observed although the cross section was severely damaged and of the residual
dimension lef (n 1) a2.
A test series was accomplished by Plieschounig (2010) with the same screw,
lef = 11.3 d, groups of 4, 9, 16 and 25 screws and a1 = a2 = 5 d. Block shear failure
was observed in groups of 9 screws. A second test series done by Plieschounig
(2010) compared the resistance of two partially threaded screws with that of a
single screw. At a1 7 d and a2 3 d no influence was observed. Hereupon we
Fig. 3 a) Typical load-deformation curve b) boxplot F1st, Fmax vs. a1 of groups with
m n = 25
did tests of groups in ST with 25 partially threaded screws and an anchoring depth
and embedment length of lef = 11.3 d and temb = 4 d, respectively, whereby
a2 = 3.5 d was kept constant and a1 varied with 5 d, 7.5 d, 10 d and 12.5 d. Test
values of F1st in Fig. 3 b) are shown as grey filled boxes, that for Fmax without
filling. An increase in resistance from a1 = 5 d to 10 d is found, whereby all
specimens failed in block shear.
Fig. 4 shows the results for groups of m n = 12 fully threaded screws, tested
with lef = 28.3 d and lef = 17.8 d, in depending of the area per screw A12. The grey
symbols right of the test results with block shear failure represent Fmax, the black
symbols F1st. Also the reference values of single screws tested at the same anchor-
ing depth are included. Under laboratory conditions it was possible to reach the
capacity of a single screw multiplied by the number of tested screws, so far no
block shear failure occurred.
Fig. 4 F1st and Fmax vs. A12 / d: test results for groups of 12 fully threaded screws in GLT
with lef = 28.3 d and 17.8 d
3 Block Shear Model for Screwed Joints
Zarnani and Quenneville (2012) proposed a model for block shear in timber joints
with outer steel plates anchored by rivets and loaded in grain. The bearing model
considers load sharing and load redistribution between potential fracture planes
surrounding the group (two lateral and a bottom shear plane as well as one plane
stressed in tension parallel to grain) with as the local deformation of the affected
timber volume, see Fig. 5 a).
We use the principle idea of Zarnani and Quennevilles model for axially
loaded joints with groups of screws for a 90 stress-fiber angle. We assume an
equal loading of all screws (deterministic stress-strain relationship) and the possi-
bility of the surface to deform freely. Furthermore for simplicity a linear elastic
material behaviour and brittle failure are assumed for transverse and rolling shear
as well as tension perpendicular to the grain. Of course, non-linear stress-strain
relationships and stochastic material properties are going to be considered in a
later work.
In contrast to Zarnani and Quenneville (2012) in our configuration we have a

bottom plane stressed in tension perpendicular to the grain, two lateral planes each
stressed in rolling as well as transversal shear, with total resistance given as
P = Pt + 2 Pr + 2 Ps, see Fig. 5 b).
Fig. 5 a) Proposed spring model of Zarnani and Quenneville (2012); b) principle definitions
for our model
The stiffness of the plane stressed in tension perpendicular to the grain is de-
termined by the elongation in load direction as function of lef.The stress distribu-
tion along lef is considered non-linear, see Fig. 5 b). Therefore the factor Ct,90 was
established, which gives the integral of stress distribution over lef divided by lef. In
a first approach Ct,90 was assumed with 0.8. The plane is given as At =
(m 1) a1 (n 1) a2 with t,90 = Ct,90 Pt,90 / At = / lef Et,90 and the stiffness
Kt,90 is a function of the ratio Pt,90 / , see
Et ,90 At
K t ,90 = . (1)
C t ,90 l ef
Zarnani and Quenneville (2012) assumed that the volumes adjacent to the
planes stressed in shear are stressed by s = G0 = / X G0 =Ps / As, where is
the shear angle and X the width of the influenced volume. Additionally the bottom
planes of this lateral volumes At,s = (n 1) a2 Xs and At,r = (m 1) a1 Xr are
stressed by t,90. We used Zarnani and Quennevilles considerations for the defini-
tion of the stiffness of the planes stressed in transversal and rolling shear, whereby
it is restricted by the distances between the border screw rows and the rectangular
supporting in the test setup with Xr = a2 and Xs = a1. The stiffness Ks of the plane
stressed in shear with As,s = (n 1) a2 lef, and the stiffness Kr of the plane
stressed in rolling shear with As,r = (m 1) a1 lef are calculated based on Ps / ,
see
G0 As,s E t ,90 At , s G90 As, r E t ,90 At,r
Ks = + ; Kr = + . (2)
Xs 10 l ef Xr 10 l ef
As the load distributions are not constant over lef the factors Cs and Cr, similar
defined to Ct,90 are introduced and in a first approach set to 0.8. The factor Ct for
the plane affected with tension perpendicular to the grain is 1.0 for the equal load-
ing of the whole plane At.
We introduced the variable a = 0, 1, 2 indicating the number of already failed
plane types. At the beginning all five planes interact by load shearing. All springs
are active, see Kj = (Kt + 2 Ks + 2 Kr) | a = 0. The fracture elongations
f,i = fi Ai Ci / Ki of each possible failure mode (transversal shear, rolling shear,
tension perpendicular to the grain) i = (1, 2, 3 a) | a = 0 are ranked in ascending
order, with f ,(1) = min ( f ,i ) and f ,(3) = max ( f ,i ) and multiplied with Kj, see
i i
3 f A C

P1st = K j
j =1
i i i
Ki
, with a = 0 and i = 1, 2, 3 .

(1)
(3)

In contrast to Zarnani and Quenneville (2012), who limited the bearing capacity
with that of the 1st failure we allow for load redistribution and further load in-
crease if possible, considering step-by-step loss of load transferring planes by
fracturing until all planes have exceeded their limits and the joints fails. Thus the
maximum resistance of the joints is given as
3 f A C a = 0, 1, 2
Pmax = max
a
Kj i i i
Ki

, with
( a +1) i = 1, ..., 3 - a
. (4)
j = a +1
The model parameters are defined as follows: The modulus of elasticity per-
pendicular to the grain Et,90,mean set to 370 N/mm for ST (EN 338) and to
300 N/mm for GLT (FprEN 14080), the shear modulus G90,mean to 690 N/mm for
ST (EN 338) and to 650 N/mm for GLT (FprEN 14080), the rolling shear modu-
lus G0,mean for both to 65 N/mm (FprEN 14080), the tensile strength perp. to grain
ft,90,mean to 2.00 N/mm (Stuefer 2011, Bla and Schmid 2001), the rolling shear
strength fr,mean to 1.47 N/mm (Mestek 2011) and the shear strength fv,mean to
7.5 N/mm (Wallner 2004, Jbstl et al. 2008, Hirschmann 2011, Gatternig 2012).
Fig. 6 Comparison between statistics of block shear failures and mean model results
Median values of F1st,50 and Fmax,50 of block shear failures at spacing a2 = 3.5 d
are shown in Fig. 6. The whiskers represent the minimum and maximum of ob-
served F1st. The upper limits based on single screw capacity in steel (median) and
withdrawal (5%- and 95%-quanitles) are also included in Fig. 6.
A comparison with the mean model results shows that they are conservative but
able to follow the trends observed in test results.
4 Conclusions
We investigated groups of screws placed in ST and GLT with a stress-fiber angle
of 90 by variation of the in-between spacings and the anchoring depths. Beside
steel and withdrawal failure also block shear failure mode, even at spacings
allowed by EC 5 or technical approvals, was observed. In cases of steel and with-
drawal failures, and under laboratory conditions nef n was reached, thus it was
possible to utilize the full potential of each screw in the tested groups. Of course,
in cases of block shear failure a significant reduction in the bearing capacities
occurred. The parameter anchoring depth was identified as important parameter
although not considered in current regulations.
Based on the test experiences was formulated in a first approach a block shear
model on deterministic basis, taking into account load shearing and load redistri-
bution between potential fracture planes on the borders of the groups of screws.
Although the prediction are until now conservative the model already allows to
explain some important parameters and phenomenons in respect to in-between
spacings and anchoring depths. The consideration of the non-linear material be-
haviour, the interaction between the screws and the influence of bending moments
within the affected timber are next important steps improving the model.
Acknowledgement. The research project was partly conducted in the frame of the COMET
K-project focus-sts at the Competence Center holz.bau forschungs gmbh (see
www.holzbauforschung.at) and partly in the frame of the bridge project SCREWS at the
Institute of Timber Engineering and Wood Technology of the University of Technology in
Graz. All foundering and support by the involved partners and also the Lignum Test Center
has to be thankfully acknowledged.
References
Bejtka, I., Bla, H.J.: Joints with inclined screws. Paper presented at the 35th CIB-Meeting,
Kyoto, Japan (September 2002)
Bla, H.J., Bejtka, I., Uibel, T.: Tragfhigkeit von Verbindungen mit selbstbohrenden
Holzschrauben mit Vollgewinde. In: Band 4 der Reihe Karlsruher Berichte zum
Ingenieurholzbau. Universittsverlag Karlsruhe (2006)
Bla, H.J., Schmid, M.: Querzugfestigkeit von Vollholz und Brettschichtholz. Report,
Universitt Karlsruhe (2001)
Bla, H.J., Uibel, T.: Spaltversagen von Holz in Verbindungen Ein Rechenmodell fr die
Rissbildung beim Eindrehen von Holzschrauben. In: Band 12 der Reihe Karlsruher
Berichte zum Ingenieurholzbau, Universittsverlag Karlsruhe (2009)
Ehlbeck, J., Grlacher, R., Werner, H.: Determination of perpendicular-to-grain tensile
stresses in joints with dowel-type fasteners. Paper Presented at the 22nd CIB-Meeting,
Berlin, German Democratic Republic (September 1989)
EN 1382: 1999-08-01, Timber structures Test methods Withdrawal capacity of timber
fasteners
EN 1995-1-1: 2009-07-01, Eurocode 5: Design of timber structures Part 1-1: General
CUAP 06.03/08: 2010-12, Self-tapping screw for use in timber constructions
EN 26891: 1991-02-01, Timber structures Joints made with mechanical fasteners Gen-
eral principles for the determination of strength and deformation characteristics (ISO
6891: 1983) (1991)
EN 338: 2009-10-01, Structural timber Strength classes
FprEN 14080: 2012-02-01, Timber structures Glued laminated timber and glued solid
timber Requirements
Gatternig, W.: Prftechnische Ermittlung der Schubfestigkeit von Vollholz und Analyse
des Greneinflusses. Master Thesis, Graz University of Technology (2012)
Gehri, E.: Ductile behaviour and group effect of glued-in steel rods. Report, ETH, Swiss
Federal Institute of Technology, Zrich (2001)
Hanhijrvi, A., Kevarinmki, A., Yli-Koski, R.: Block shear failure at dowelled
steeltotimber connections. Paper Presented at the 39th CIB-Meeting, Florence, Italy
(August 2006)
Hirschmann, B.: Ein Beitrag zur Bestimmung der Scheibenschubfestigkeit von
Brettsperrholz. Master Thesis, Graz University of Technology (2011)
Jbstl, R., Bogensberger, T., Schickhofer, G.: In-plane shear strength of cross laminated
timber. Paper Presented at the 41st CIB-Meeting, St. Andrews, Canada (August 2008)
Kevarinmki, A.: Joints with inclined screws. Paper Presented at the 35th CIB-Meeting,
Kyoto, Japan (September 2002)
Krenn, H., Schickhofer, G.: Joints with inclined screws and steel plates as outer members.
Paper Presented at the 42nd CIB-Meeting, Duebendorf, Switzerland (August 2009)
Mahlknecht, U.: Untersuchung von rechtwinklig zur Faser eingebrachten axial
beanspruchten Schraubengruppen im Vollholz und Brettsperrholz. Master Thesis, Graz
University of Technology (2011)
Mestek P.: Punktgesttzte Flchentragwerke aus Brettsperrholz (BSP) Schubbemessung
unter Bercksichtigung von Schubverstrkungen. Doctoral Thesis, Technische
Universitt Mnchen (2011)
Pirnbacher, G., Brandner, R., Schickhofer, G.: Base Parameters of self-tapping Screws.
Paper Presented at the 42nd CIB-Meeting, Duebendorf, Switzerland (August 2009)
Plieschounig, S.: Ausziehverhalten axial beanspruchter Schraubengruppen. Master Thesis,
Graz University of Technology (2010)
prEN 1194: 2006-06-13, Glued laminated timber Strength classes and determinations of
characteristic values (2006)
Schoenmakers, D.: Fracture and failure mechanisms in timber loaded perpendicular to the
grain by mechanical connections. Doctoral Thesis, Eindhoven University of Technology
(2010)
Stahl, D.C., Wolfe, R.W., Begel, M.: Improved Analysis of timber rivet connections. Jour-
nal of Structural Engineering ASCE 130(8), 12721279 (2004)
Stuefer, A.: Einflussparameter auf die Querzugfestigkeit von BSH-Lamellen. Master The-
sis, Graz University of Technology (2011)
Wallner, G.: Versuchstechnische Ermittlung der Verschiebungskenngren von orthogonal
verklebten Brettlamellen. Master Thesis, Graz University of Technology (2004)
Zarnani, P., Quenneville, P.: Predictive analytical model for wood capacity of rivet connec-
tions in glulam and LVL. Paper Presented at the 45th CIB-Meeting, Vxj, Sweden
(August 2012)
A Method to Determine the Plastic Bending
Angle of Dowel-Type Fasteners
Michael Steilner and Hans Joachim Bla
Karlsruhe Institute of Technology, Timber Structures and Building Construction,

{steilner,blass}@kit.edu
Abstract. In timber joints with dowel-type fasteners, the yield moment of the fas-
tener is an important parameter. For fasteners of hardened steel, e.g. self-tapping
screws, the yield moment cannot be calculated and has to be measured in a bend-
ing test as established in EN 409. The yield moment is defined at a plastic bending
angle of = 45 /d 0.7 . Therefore, the determination of the plastic bending angle is
important to derive the characteristic yield moment of fasteners. For a proper deriva-
tion of the fasteners yield moment, the onset of plastic deformations must hence be
identified in order to be able to determine the correct plastic bending angle. A new
method for the determination of the plastic bending angle was developed which is
independent of a ductile or brittle behaviour of the fastener. The method is generally
applicable to different fastener types with various diameters.
Keywords: dowel type fasteners, yield moment, self-tapping screws, moment-

angle-curve, plastic deformation.
1 Introduction
The design equations of joints with dowel-type fasteners in EN 1995 (EC5) [4] are
based on the theory of Johansen. The strength of the joint depends on the geometry
of the joint, the embedding strength of the timber and the yield moment of the
fasteners for failure modes with one or two plastic hinges per shear plane.
In design equations it is assumed that the fastener is an ideal-plastic material.
But in real joints the plastic bending capacity is often only partially used and the
fasteners bending moment lies between the elastic and plastic bending capacity of
the fastener [1].
Therefore, it is important to determine the actual moment of the fastener in real
joints. For most fasteners, e.g. self-tapping screws of hardened steel, the yield mo-
ment cannot be calculated and has to be measured in a bending test as established
in EN 409 [5].
DOI: 10.1007/978-94-007-7811-5_28, RILEM 2014
302 M. Steilner and H.J. Bla
Fig. 1 Mesured moment in

dependency of the bending
angle
In the CUAP for self-tapping screws the yield moment is defined at a plastic
bending angle of = 45 /d 0.7 (d is the diameter) [2]. Therefore, the determination
of plastic bending angle is important to derive the characteristic yield moment of
fasteners. Figure 1 shows a typical result of a bending test with a self-tapping screw.
Up to a bending angle of about 9 only elastic deformations can be observed (dotted
line in Figure 1). At higher angles, plastic deformations occur in the fastener. The
plastic bending angle is defined as the difference between the measured, plastic
bending angle at the end of test and the elastic bending angle.
For a proper derivation of the fasteners yield moment, the onset of plastic defor-
mations must hence be identified in order to be able to determine the correct plastic
bending angle. If ductile behaviour of the fasteners is observed without final rup-
ture of the fastener, the plastic bending angle can be measured after the bending
test. However, if brittle behaviour occurs with final rupture of the tested fastener,
no measurements are possible and the necessary information must be determined
analysing the moment-angle-curve of the bending test.
Meanwhile the tensile strength of many self-tapping screws is very high (>
1000 N/mm2 ). Especially those screws rupture at the end of the test and the plastic
bending angle cannot be measured after the end of the test.
2 Method
EN 12512 describes a test method for the ductility of mechanical joints under cy-
cling loads. EN 12512 includes a method to determinate the yield point of the
joint. The yield point is defined as the intersection of the gradient of the load-
displacement-curve between 0.1 Fmax and 0.4 Fmax and the tangent on the load-
displacement-curve with a gradient of 1/6 from the first gradient [3].
Plastic Bending Angle of Dowel-Type Fasteners 303
Figure 2 shows a typical moment-angle-curve of a bending test with self-tapping

screws. Similar to EN 12512, the onset of plastic deformations of the fastener in the
bending test is defined as the intersection of the gradient of the moment-angle-curve
between 0.1 Mmax and 0.4 Mmax and the tangent on the moment-angle-curve with a
gradient of 1/6 from the first gradient.
If the behaviour of the fastener is not completly ductile and the fastener ruptures
at the end of the test, it is necessary to determine the angle of failure r , shown in
figure 2. Similar to EN 12512 r is defined as the angle where the moment equals
0.8 Mmax after reaching the maximal moment Mmax . So the plastic bending angle
pl can be calculated with r and y .
pl = r y (1)
The advantage of this method is that the calculation only depends on the stiff-
ness at the beginning of the test and the maximum moment. Therefore, the plastic
bending angle can be calculated and it is not important if the behaviour is brittle or
ductile.
Fig. 2 Determination of the plastic bending angle of a dowel-type fastener from a moment-
angle-curve depending of the stiffness at the beginning of the test and Mmax
For the experimental verification of the method, 145 Screws and dowels with dif-
ferent diameters from 4 mm up to 12 mm and different steel grades were tested.
The tested fasteners are listed in table 1. For the bending tests, a bending machine
according to EN 409 was used.
Table 1 Tested fasteners
Diameter Type Material Qantity

in mm of fastener
4 screw steel 10
4.5 screw steel 10
5 screw steel 10
5 screw stainless steel 10
6 screw steel 19
6.5 screw steel 5
6.5 screw vhs steel 12
7 dowel steel 5
7.5 dowel steel 5
8 screw steel 9
8 screw vhs steel 8
8 dowel steel 4
8.2 screw steel 10
9 screw vhs steel 6
10 screw steel 10
10 screw vhs steel 4
12 screw steel 8
total: 145
For all fasteners, the bending angle at the end of the test was 61 . This includes
the plastic and elastic bending angle. After reaching the maximum angle, the bend-
ing test machine leas the force, the elastic deformation disappears and the plastic
bending angle can be measured. Therefore, it is necessary that the bending machine
allows the release of the elastic deformation, because the fastener must be removed
from the bending machine without any extra deformation. Apart from that it is not
possible to measure the plastic bending angle.
For all fasteners the plastic bending angles were measured and calculated with
the new method. Figure 3 shows the measured and calculated plastic bending angles.
The agreement between the calulated and measured angeles is very good with an
coefficient of determination of R2 = 0.9898 and an gradient of m = 0.9938.
The results show that the onset of plastic deformation can be identified with the
method. Now it is possible to determinate the plastic bending angle also from fas-
teners that ruptured at the end of the bending test. Figure 4 shows at the left side
a moment-angle-curve of a bending test of a self-tapping screw ruptured at an an-
gle of about 39 . The right diagram shows the same moment-angle-curve after the
calculation wiht the new method. A plastic bending angle of 27.4 was calculated.
Plastic Bending Angle of Dowel-Type Fasteners 305
Fig. 3 The calculated plastic bending angle over the measured plastic bending angle for
different dowel-type fasteners and various diameters
Fig. 4 Example for a bending test with a self-tapping screw that ruptured during the test at
the left side, on the right side the determination of the plastic bending angle with the new
method after the test
4 Conclusions
The CUAP for self-tapping screws demands the determination of the plastic bend-
ing angle. But often the screws rupture during the bending test and the necessary
information must be determined analysing the moment-angle-curve of the bending
test.
A workable method to determinate the plastic bending angle is presented. The
advantage of the method is that the calculation only depends on the stiffness at the
beginning of the test and the maximum moment.
A further advantege is that for the determination oft the plastic bending angle
only the moment-angle-curve of the test is necessary. The moment-angle-curve can
be used directly from the bending test machine.
For 145 bending tests with different fastener types with various diameters, the
method was confirmed. The onset of the plastic deformation can hence be identified
and is independent if the screws were ruptured during the test or not. So the plastic
bending angle can be calculateted from the moment-angle-curve of a bending test.
References
[1] Blass, H.J., Bienhaus, A., Kramer, V.: Effective bending capacity of dowel-
type fasteners. In: CIB-W18 Meeting 33, Paper 33-7-5, Delft, The Netherlands
(2000)
[2] CUAP, No 06.03/08: Self-tapping screws for use in timber constructions (2010)
[3] EN12512, Timber structures - Test methods - Cyclic testing of joints made with
mechanical fasteners (2001)
[4] EN1995-1-1, Eurocode 5. Design of timber structures - Part 1-1: General - Com-
mon rules and rules for buildings (2004)
[5] EN409, Timber structures - Test methods - Determination of the yield moment
of dowel type fasteners (2009)
Low-Damage Design Using a Gravity Rocking
Moment Connection
Mamoon Jamil, Pierre Quenneville, and Charles Clifton
Faculty of Civil and Environmental Engineering, The University of Auckland,

New Zealand
mjam057@aucklanduni.ac.nz,
{p.quenneville,c.clifton}@auckland.ac.nz
Abstract. A novel low-damage connection for steel column, timber beam multi-
storey moment frame buildings is being developed. The connection uses gravity
load acting at an eccentricity to the column centre line for moment resistance and
self-centring. There is built-in friction damping for energy dissipation. Beams and
columns are continuous past the joint, resulting in minimal damage to the floor.
1:20 scale shake table tests have shown expected low-damage, self-centring
behaviour. A SAP2000 model with friction isolators for support point gap opening
and friction sliding behaviour is presented. The SAP2000 model simulates peak
drift well but is stiffer than the test model at low drift levels.
Keywords: Low-damage, Seismic, Gravity rocking, Friction damping, Shake

table, SAP2000.
1 Introduction
Timber no longer has a major share in the heavy construction market as it did
several decades ago, especially in the case of multi-storey heavy moment frame
buildings. There is a need to develop, implement and showcase high performance
and economical moment frame timber systems. Hybrid solutions with timber in
combination with steel and/or concrete may be most optimised.
Damage in recent major earthquakes has resulted in the development of
performance based design philosophies which not only provide for life-safety, but
also aim to minimise damage. Structural ductility is used to reduce the magnitude
of forces induced during an earthquake. Capacity design principles are used to
direct inelastic action to dedicated zones in a structural system. A major
earthquake can cause permanent damage to these ductile zones and cause the
overall structure to be rendered in-operational or require demolition. Typically,
damage to a timber moment frame is in the form of bolt yielding, crushing of
timber and, if poorly designed or loaded very severely, cracking of timber
members at the joint.
DOI: 10.1007/978-94-007-7811-5_29, RILEM 2014
308 M. Jamil, P. Quenneville, and C. Clifton
Low-damage systems aim to provide damage-free ductile behaviour. Damage

to the main structural frame, secondary structural systems, floor, walls, cladding
and other non-structural components needs to be minimized. Peak drift during an
earthquake must be controlled for limiting damage and residual drift must be
controlled for post-event serviceability. Some existing systems are briefly
described in the next section, followed by a presentation of the new low-damage.
2 Existing Low-Damage Systems

Ancient Greek and Roman masonry temples exhibit gravity rocking and friction
sliding in a strong earthquake. Figure 1 shows these effects in a two-storey temple
with multi-drum colonnades. The rocking and sliding provides ductility capacity
but is not controlled and thus prone to large drift, soft-storey formation and
collapse.
Fig. 1 Gravity rocking and friction sliding in ancient Greek/Roman temples copied from
[1]
Traditional timber Chinese and Japanese pagoda have a far more intricate
connection detail as shown in Figure 2. The underlying mechanism is still
somewhat similar to those in ancient Greek temples, though not as pronounced,
and there is elongation of structural vibration period with increased drift [2].
The Sliding Hinge Joint shown in Figure 3 was developed for steel moment
frame buildings up to about 10 stories in height and in high seismic zones [3]. The
top flange of the beam is effectively pinned to the column, web bolts carry vertical
shear and a friction slider at the beam bottom flange provides moment resistance
and energy dissipation. The connection can sustain several cycles of large rotation
without damage. Due to the unique hysteresis shape of the asymmetric friction
slider, a building with this connection tends to self-centre after a major earthquake
but does not reliably fully self-centre. Addition of a ring spring for full self-
centring has been investigated [4, 5], however, the ring spring can be costly.
Low-Damage Design Using a Gravity Rocking Moment Connection 309
Fig. 2 Comparison of connection used in Greek masonry temples with that in Japanese
wood temples copied from [2]
Effectively
pinned at top Top web bolts
carry vertical load
Column
Beam
Friction slider at bottom
Fig. 3 The Sliding Hinge Joint copied from [3]
Post-tensioned connections, originally developed under the PRE-cast Seismic

Structural System (PRESSS) programme in the 1990s, are now being applied to
steel [6, 7] and timber [8]. An un-bonded post-tensioned tendon runs through the
beam and clamps the connection across the column as shown in Figure 4. Mild
steel rods or other dampers may be added for energy dissipation as the post-
tensioned system itself has low inherent damping. The systems has shown
excellent self-centring and low-damage behaviour in tests [9, 10]. When used with
timber, it is prone to creep shrinkage of the timber over time and thus stress
relaxation in post-tensioned tendons. This would result in a reduction in the
moment strength of the connection. Steel components can be used in the joint
region, where stress travels perpendicular-to-grain through the column, to reduce
the shrinkage effect. However, this would add to cost.
Mild steel Rocking

rods motion
Un-bonded post-
tensioned tendon
Fig. 4 Post-tensioned timber moment connection copied from [11]
3 New Damage Avoidance System
3.1 Target Minimum Performance
Table 1 Minimum performance target for new system
Limit Mode of Condition of Drift

State operation structure
(return residual/
period) peak
SLS Elastic As new 0.0% /
(25 ~ 0.33%
years)
ULS Inelastic Close to as new. ~ 0.2% /
(500 connection No wood crushing 2.5%
years) rotation. or strength/stiffness
Timber degradation.
members Minimal non-
to remain structural damage
elastic including partitions
and floor slab
MCE Large Structural damage No strict
(2500 inelastic limited to readily limit. ~ 4
years) connection replaceable to 5%
rotation. components. expected
Non-structural
elements need to be
detailed for large
drift.
Structural collapse
reliably avoided
3.2 Connection Form

The connection under development combines gravity rocking and friction
damping. The concept was developed through several iterations and final
refinements are still being made. The development process and some discarded
concepts are presented in [12]. Components of the connection are shown in Fig. 5.
There are three main features: (1) Gravity rocking of the beam on supports
cantilevered from the column; (2) Friction sliding between the reinforced
underside of the beam spacer and cantilever support; (3) Brackets on top of the
beam which restrain it relative to the column horizontally but allow it to uplift
upon joint opening; and (4) continuous beam and column past the joint. The
column is continuous to avoid soft-storey formation and the beam is continuous to
achieve rocking behaviour and also to reduce damage to the floor slab. A steel
column is used instead of a timber column because the column is heavily loaded
while the beam is relatively lightly loaded in this particular system. Furthermore, a
steel column simplifies the connection and has a smaller footprint.
Fig. 5 New low-damage connection. Top left: construction step 1; top right: construction
step 2; and bottom: completed joint.
312 M. Jamil, P. Quenneville, and C. Cliftoon
To assemble the jointt one may: (1) start with steel column with cantileveer
support; (2) place armoured spacers on the cantilever supports; (3) screw a haalf
beam on each side of the spacer and; (4) install top brackets. Figure 6 shows addd-
on rubber pads for a posiitive stiffness after the joint opens up. Post joint openinng
stiffness is important for drift
d control.
Rubber on wedge
of wood attached
to beam
Fig. 6 Connection with addeed rubber pad for post joint opening stiffness
3.3 Connection Meechanics

The connection developss moment resistance through a number of sources aas
shown in Figure 7. The main source of moment resistance is the joint tributarry
gravity load acting at an n eccentricity to the column centreline. The horizonttal
restraint to the top of the beam relative to the column forces the cantilever suppoort
to slide relative to the arrmoured spacer. This generates a friction force that accts
through a lever arm appro oximately equal to the beam depth. The top brackets havve
a small vertical resistancee; the resistance of the second top bracket is not shown iin
the figure for simplicity of
o illustration.
It was found from eartthquake testing of a 1:20 scale model that peak buildinng
drift can be very high esspecially during near source earthquakes. To rectify thhis
issue, a rubber pad may be attached to the inside of the beam so that it beaars
against the underside of the
t top flange of the cantilever support as the joint openns
up. This spring like behaaviour of the rubber pad is illustrated in Figure 7. Thhe
result is an increase in stiiffness of the connection beyond joint opening as show wn
in Figure 8. A rubber pad add-on was implemented in an FE model to verify that it
does indeed resolve the laarge peak drift issue. 1:20 scale testing and FE model arre
detailed in the next two seections.
Fig. 7 Connection mechanics
2000
No rubber
Base shear (N)
1500
With rubber
1000
500
0
0 0.02 0.04 0.06 0.08 0.1
Drift (%)
Fig. 8 Increase in building pushover stiffness post joint opening as a result of added rubber
pads
4 1:20 Scale Shake Table Tests
A 1:20 scale model with timber beams and columns was built and tested following
scaling similitude laws. The 3 storey, 1 by 2 bay model is shown in Figure 9. The
walls on either side were used to limit movement of the test model to the in-plane
direction via guide rollers. Note that timber columns were used because this test
was done before opting for steel columns instead.
The model was based on a prototype design for a full scale 3 storey building for
Auckland, New Zealand which is a low seismic zone. In the absense of a design
procedure for the new system, equivalent static earthquake design was done based
on NZS1170.5 {SNZ, 2004 #1249}. Identical connections, as shown in Figure 10,
were used throughout the model making it a uniform strength moment frame.
The length scale factor of 1:20 was used whereas the Youngs modulus and
acceleration were not factored. This resulted in a time scale factor of 1:4.47. The
model was about 1000 mm long, 540 mm wide and 600 mm high and weighed
about 490 kg overall. More detail on the model can be found in [13].
Fig. 9 1:20 scale model ready for testing on shake table
30 mm
Yielding C
top bracket o
l
u Spaced
m beam
n
Mild steel
bearing Side plate Hardwearing
support on screwed steel under
side plate to column beam
Fig. 10 Close-up of connection in 1:20 scale model
4.1 Test Regime

Push-over, snap-back and earthquake tests were done on the model. Only
earthquake test results are presented here for brevity.
A suite of earthquake ground motions chosen specifically for the North Island
of New Zealand according to the local design standard [14] and recent research
[15] was used. The original earthquake records were scaled for soil type D at
Wellington, New Zealand, which is a high seismic zone. These records were
further scaled for the serviceability limit state (SLS), ultimate limit state (ULS)
and maximum credible earthquake

e (MCE) intensities. Raw acceleration recordds
were acquired from the au uthor of [15]; original sources are given in the referenceed
publication. The earthquaake records were time scaled to be 4.47 times shorter aas
per scaling similitude req
quirements. Acceleration records were double integrateed
and linear baseline correccted to get displacement records as the shake table waas
displacement controlled.
4.2 Test Results

The test model showed no o-damage, fully self-centring behaviour as expected. Thhe
joints remained largely cllosed at the SLS earthquake level. At the ULS and MC CE
earthquake intensity the jo
oint showed sliding and uplift.
Fig. 11 SAP2000 'Componen

nt' model
Top brackets,
multi-linear
plastic
Supports,
friction isolator
Fig. 12 SAP2000 'Componen

nt' model connection close-up
Movement of joint coomponents is shown in Figure 15. One cantilever suppoort

butary gravity load. This support also slides, creating a
takes all of the joint trib
friction force. The amoun nt of joint opening, floor uplift and joint sliding dependds
on connection rotation, geeometry and stiffness.
Fig. 13 Drift at level 3 (top level) for the seven selected ground motions; left: Ultimate
Limit State; and right: Maximum Credible Earthquake
Drift at the top level for the each earthquake at ULS and MCE is plotted in
Figure 13. Drift at the SLS was below 0.3% and is not presented here. While drift
is acceptable for most earthquakes at the ULS level, a very high drift was seen in
near source earthquakes, especially the El Centro 1979 earthquake. El Centro
1940, which is not near fault, resulted in a much smaller peak drift. Some tests at
MCE level with severe earthquake motion were not done to avoid damage to the
model. Residual drift was no more than 0.2% and is hardly visible in Figure 14.

Wd
Zd

W^W

W^WZ

,<
dh
,<
dh
,<
dh

^>^ h>^ D
>^
Fig. 14 Summary of peak and residual drift at level 3 (top level) of the 1:20 scale model
The large drift was expected because the model was designed for a low seismic
zone and tested upto near failure with near fault and MCE earthquakes scaled for
a high siesmic zone. This was done to assess system limits. It was found that the
frame and connection were capable of very large drift without damage. However,
drift needs to be limited for protection of non-structural elements. Further tests at
1:5 scale are planned. The 1:5 scale model will be designed for a high seismic
zone and near fault earthquakes. Furthermore, the option of added rubber pads for
drift control was investigated in the FE model as described below.
5 1:20 SCALE SAP2000 Model
5.1 FEM Setup Rotational Spring Model

A simplified SAP2000 model with rotational springs at the beam-column joints
was first done. Four rotational springs were placed at each joint, each representing
non-linear properties due to moment from gravity load, friction and each of the top
brackets. More detail can be found in [16]. Global behaviour of this simplified
rotational spring model was similar to the component model below so results are
not presented here.
5.2 FEM Setup Component Model

A component FE model with elements to model each physical phenomenon as it
occurs was setup in SAP2000. The overall 2D frame is shown in Figure 11 and
close-up of a connection is shown in Figure 12. Some simplifications were made,
for example, the top bracket is modelled with a short link element which is stiff in
the horizontal direction and has steel yielding hysteresis in the vertical direction.
The support points were modelled with friction isolators with a coefficient of
friction of 0.35. Note that large displacement analysis was not done because the
SAP2000 friction isolator link element cannot model concurrent axial support
behaviour and sliding when large displacement geometry changes are recognised.
Excessive drift seen in testing was reflected in the SAP2000 model. To control
drift, a rubber pad may be used to give the connection stiffness beyond joint
opening as shown in Figure 6, Figure 7 and Figure 8. The rubber pad behaviour is
modelled in SAP2000 by assigning a tension stiffness to the tension/ compression
friction isolator link element. It is assumed that the force-deformation behaviour
of the rubber pad is linear in the range of interest. A rubber pad stiffness was
selected so that the overall building push-over stiffness beyond joint opening was
approximately 15% of the initial stiffness.
5.3 Test and FEM Results

The SAP2000 component model matched test results well during large drift cycles
and predicted peak drift well. This can be seen in Figure 13 and Figure 14
respectively. At lower drift levels, the FEM was stiffer than the test building. This
is seen most clearly in the ULS El Centro 1979 earthquake response.
A FEM prediction is shown for some of the tests which were not done on the
physical model to avoid damage. The FEM failed to converge part way through
analysis for the MCE El Centro 1979 and TCU051 earthquakes.
Addition of rubber pads to the FEM for stiffness beyond joint opening
generally controlled drift successfully. A drift of 12.9% is still predicted for the
very severe El Centro 1979 earthquake at MCE level. If drift is to be controlled for
this earthquake level, the moment at which the joint opens should be increased.
This may be done by increasing the length of the support cantilever or by adding
supplemental friction dampers to the joint. Another option would be to increase
the post joint opening stiffness by changing rubber pad properties. Also, as
mentioned earlier, design of the 1:20 scale model was done for a low seismic zone
but it was tested up to its limit with stronger earthquakes.
The gravity rocking joint is capable of large inelastic rotations without damage
and is targeted at medium and high seismic zones. It is most suitable for buildings
up to about 5 stories high. System limits will become more clear with furhter
testing, FEM and development of design procedure. Findings from furhter work
will be publised in the coming year.
Fig. 15 Movement at Joint level for the El Centro 1979 earthquake at ULS intensity
6 Planned Work
1:20 scale tests were done to approximately evaluate system behaviour. Simplified
parts and connection details were used because of the small scale. 1:5 scale tests
with steel columns and actual connection components are planned. Following
these tests, a preliminary design procedure and numerical time history study are
planned.
7 Conclusions
A novel connection for low-damage multi-storey moment frame buildings is under

development. Gravity rocking and built-in friction damping provide moment
resistance, self-centring and energy dissipation. Continuous beams minimise
damage to the floor system.
Tests done at 1:20 scale have shown:
Excellent self-centring with negligible residual drift
Low-damage to connection components, floor slab and frame members
Undesirable peak drift under near source earthquake ground motion. This may
be mitigated by adding rubber pads for post joint opening stiffness or by
increasing the connection cantilever support length
A FE model with contact and friction sliding behaviour was implemented in
SAP2000. The model agreed well with 1:20 scale test results in terms of peak drift
and residual drift.
Acknowledgements. The authors would like to thank the Ministry of Primary Industries
for funding this research. The guidance and help from various staff and colleagues is
greatly appreciated. Thanks also go to laboratory staff for help with testing.
References
[1] Papaloizou, L., Komodromos, P.: Investigating the seismic response of ancient multi-
drum colonnades with two rows of columns using an object-oriented designed
software. Advances in Engineering Software 44, 136149 (2011)
[2] Hanazato, T., et al.: Seismic and wind performance of five-storied Pagoda of timber
heritage structure. Advanced Materials Research 133-134, 7995 (2010)
[3] Clifton, G.C.: Semi-rigid joints for moment-resisting steel framed seismic-resisting
systems. Dept of Civil and Environmental Engineering, The University of Auckland:
Auckland (2005)
[4] Khoo, H.H., et al.: Influence of steel shim hardness on the Sliding Hinge Joint
performance. Journal of Constructional Steel Research 72, 119129 (2012)
[5] Khoo, H.H., Clifton, C., Butterworth, J.: Experimental studies of the Self-Centering
Sliding Hinge Joint. In: NZSEE, Christchurch, New Zealand, pp. 119129 (2012)
[6] Herning, G., et al.: An overview of self-centering steel moment frames. In: Structures
Congress - Dont Mess with Structural Engineers: Expanding Our Role, Austin, TX,
pp. 14121420 (2009)
[7] Chou, C.C., Chen, J.H.: Seismic design and shake table tests of a steel post-tensioned
self-centering moment frame with a slab accommodating frame expansion.
Earthquake Engineering and Structural Dynamics 40(11), 12411261 (2011)
[8] Newcombe, M.: Seismic design of post-tensioned timber frame and wall buildings.
Civil and Resource Engineering, University of Canterbury, p. 475 (2011)
[9] Pino, D.P.M., et al.: Shake table response of multi-storey post-tensioned timber
buildings. In: NZSEE Conference, Wellington, New Zealand (2010)
[10] Newcombe, M., Pampanin, S., Buchanan, A.: Global response of a two storey Pres-
Lam timber building. In: NZSEE Annual Technical Conference. New Zealand
Society of Earthquake Engineering, Wellington (2010)
[11] Buchanan, A., et al.: Multi-storey prestressed timber buildings in New Zealand.
Structural Engineering International 18(2), 166173 (2008)
[12] Jamil, M., Quenneville, P., Clifton, G.C.: Low-damage moment resisting timber
multi-storey buildings. In: World Conference on Timber Engineering, Auckland,
New Zealand, pp. 367376 (2012)
[13] Jamil, M., Quenneville, P., Clifton, G.C.: New low damage timber frame solution for
multi-storey office type buildings. In: SESOC Conference, Auckland, New Zealand
(2012)
[14] SNZ: Structural Design Actions - Part 5: Earthquake actions - New Zealand.
Standards New Zealand (2004)
[15] Oyarzo-Vera, C.A., McVerry, G.H., Ingham, J.M.: Seismic Zonation and Default
Suite of Ground-Motion Records for Time-History Analysis in the North Island of
New Zealand. Earthquake Spectra 28(2), 667688 (2012)
[16] Jamil, M., Quenneville, P., Clifton, G.C.: New damage avoidance connnection for
multi-storey timber-steel hybrid moment frame buildings. In: 10th International
Conference on Urban Earthquake Engineering. Tokyo Institute of Technology,
Tokyo (2013)
Part III
Glued Joints and Adhesives
Finger Jointing of Freshly Sawn Norway
Spruce Side Boards A Comparative Study
of Fracture Properties of Joints Glued
with Phenol-Resorcinol and One-Component
Polyurethane Adhesive
Magdalena Sterley1, Erik Serrano1,2, Bertil Enquist2, and Joanna Hornatowska3
1
SP Technical Research Institute of Sweden, Box 5609, SE-114 86 Stockholm, Sweden
magdalena.sterley@sp.se
2
Linnus University, 351 95 Vxj, Sweden
3
Innventia AB, Box 5604, 11486 Stockholm, Sweden
Abstract. Finger jointing of unseasoned Norway Spruce was studied with respect
to tensile strength, adhesive penetration and durability. Finger joints were
manufactured with 1) unseasoned wood and one component polyurethane (PUR)
adhesive, 2) dried wood and PUR adhesive and 3) dried wood and phenol
resorcinol formaldehyde (PRF) adhesive. Two levels of wood density were used.
The tensile strength of the finger joints was determined and the deformations
within the joint were studied with an optical measurement system (ARAMIS). The
penetration of the adhesive was studied with x-ray microtomography. The
durability of the joints was determined according to the standard ASTM D 4688.
The results show that the tensile strength and the durability of green glued finger
joints are on the same level as that of dry glued PUR joints. The penetration of the
PUR adhesive is high in the unseasoned wood and cavities within the bonds seem
to be smaller than in dry glued PUR joints. The tensile strength of the finger joints
is dependent on density, independent on the adhesive system used. The strength of
the green glued PUR adhesive bonds in finger joints measured with small scale
specimens did not differ from the strength of the dry glued PUR bonds.
Keywords: finger joint, bond line, fracture, green gluing, digital image
correlation, x-ray tomography.
1 Introduction
Gluing of freshly sawn or high moisture content timber was used more than 20
years ago for finger jointing and was successfully tested by many researchers
(Pizzi et al. 1980, 1984, Throughton and Chow 1979, 1980, Kreibich et al. 1996,
1998, Parker 1994, Verreault 1999 and 2000, Lange et al. 2000, Sterley et al.
DOI: 10.1007/978-94-007-7811-5_30, RILEM 2014
326 M. Sterley et al.
2005, Pommier et al. 2006, Karastergiou and Mantanis 2008, Simon and Bredesen
2009). Different adhesives were used for green finger jointing: tannin and soy-
based, modified phenol-resorcinol formaldehyde (PRF) and one-component
polyurethane adhesives (PUR). Green finger jointing has been industrially applied
in South Africa, Chile, North America, New Zealand, Australia and United
Kingdom and lately it has also been tested in France and in Sweden. The green
finger jointing process using fast and cold curing adhesives is recognized as an
eco-efficient process resulting in less waste during production and resulting also in
savings during the drying and pressing operations. It is commonly used for
upgrading of low-value timber as Maritime Pine (Pommier et al. 2006 and Simon
and Bredesen 2009). Results from these extended studies on the green glued finger
joint showed an increased density in the finger tip, the wood-adhesive interphase
with less of a discontinuity and a better penetration of the adhesive into the wood
compared to conventionally glued finger joints. These properties and also the
presence of covalent bonds between specially designed PUR adhesives and the
wood were mentioned as parameters influencing the high bending strength of
green glued finger joints, which fulfilled the load bearing requirements of
structural timber.
In the present project, green finger jointing of side boards from Norway Spruce
was investigated. Side boards, i.e. boards of narrow dimensions sawn from the
outer parts of a log, are seldom used for load bearing purposes. The costs for their
production are often considerable but their commercial value is low. Side boards,
however, possess excellent structural properties (stiffness and strength (Steffen et
al 1997)) and using such boards as laminations in engineered wood products, a
considerable amount of added value for the sawmilling industry would be
obtained. The flat wise green gluing of centre boards and side boards to produce
green glued beams for load bearing applications was studied previously with
promising results (Sterley et al 2004, 2009, Serrano et al 2010). Since side board
lengths vary and are limited to a maximum of 5-6 m, finger joints have to be used
in order to obtain products of desired lengths.
Testing of finger jointed lamellae can be realized by bending tests or by tensile
tests. Tensile testing is generally the better choice in terms of resulting in a
representative state of stress, since the outermost finger jointed lamination in
glulam beams are mainly subjected to tensile stresses during bending.
Optical measurements of strain distribution on tested finger joints were
reported by Konnerth et al. (2006) and Muszynski et al. (2008). Konnerth et al.
(2006) tested finger joints with PUR and MUF adhesives in the elastic range. The
results indicated that the strain concentrations were similar for the MUF and
the PUR adhesives. In the elastic range, the stress levels should thus be higher for
the MUF as compared to the PUR because of the difference in stiffness of the
adhesives, MUF being much stiffer as compared to the PUR.
Muszynski et al. (2008) concluded that smaller strains were present in the
fingertips of the green glued finger joints as compared to the dry glued finger
joints. This was attributed to the lack of gaps at the tips, thanks to the very tight
(close contact) green glued finger joints.
Finger Jointing of Freshly Sawn Norway Spruce Side Boards 327
The aim of the present study was to compare tensile strength, adhesive
penetration and durability of green glued joints in spruce side boards with the
behavior of dry glued joints using the same one component PUR adhesive and a
more brittle PRF (Phenol Resorcinol Formaldehyde) adhesive. Additionally a
mixed-mode test for adhesive bonds obtained from finger joints is presented. This
study was a part of a larger project carried out at SP Wood Technology and
Linnaeus University, Sweden.
2.1 Boards and Finger Joints

Side boards with dimensions 2570 mm were selected and grouped into two
separate density groups: a high-density group with an average density of 445
kg/m3 (with standard deviation (SD) of 20 kg/m3) and a moisture content of 90%
(with SD of 30%) and a low-density group with an average density of 360 kg/m3
(with SD of 17 kg/m3) and a moisture content of 115% (with SD of 30%). These
densities refer to the density as determined with absolutely dry mass and raw
volume.
Each board was divided into three parts: 1) for green gluing with the PUR
adhesive, 2) for dry gluing with the same PUR adhesive and 3) for dry gluing with
the PRF adhesive. The boards for green gluing were finger jointed immediately
after cutting while the boards for dry gluing were kiln dried at a saw mill together
with green glued finger joints to an average moisture content of 14%. Dry glued
finger joints were manufactured thereafter. Finger joints were cut in the laboratory
with a profile for finger joints for structural use: 153.80.42 mm according to
EN 385 and with the fingers visible on the flat side. Adhesive was applied
manually. A laboratory press was used for pressing the finger joints with a
nominal end pressure of 5 and 10 MPa being applied for the green glued and dry
glued joints, respectively.
2.2 Specimens for Tensile Testing

The finger joints for tensile tests were planed on all 4 surfaces. Exactly 14 fingers
were included in each specimen leaving a fingertip at the edges of the specimens.
In order to avoid failure of the specimens close to the clamping device of the
testing machine, the specimen ends were reinforced. Firstly, the area to be gripped
by the testing machine was made wider by gluing of pieces of wood to the edges
of the specimens at both ends. Furthermore a random-orientation glass fiber mat
was glued on to the ends four steel plates were glued to each specimen using an
epoxy adhesive. The specimen for the tensile testing is seen in Figure 1. 10
specimens of each density and adhesive system type were tested.
Fig. 1 Test set up of tensile testing of finger joints. Steel plates and reinforcement with
glass fiber mats are seen. The spray pattern applied for the optical deformation
measurement is visible on the surface of the specimen. In the foreground, the two cameras
of the ARAMIS system are also shown.
2.3 Specimens for X-ray Tomography

For each of the three adhesive systems four fingers originating from different
joints were tested. In total 12 specimens were studied. The specimens had the
dimensions 4535mm3.
2.4 Specimens for Durability Testing

For each adhesive type 14 side boards were used for manufacturing of finger
joints for durability testing. The mean density was 405 kg/m3 (calculated with dry
mass and raw volume) and varied between 365 and 470 kg/m3. The mean moisture
content was 94% ranging from 44 to 156%. From each joint, three specimens of
dimensions 6.435305 mm were cut. In total 126 durability specimens were
manufactured (see Table 1).
2.5 Specimens for Adhesive Bond Mixed-Mode Testing

Small scale specimens for mixed-mode testing of adhesive bonds and the principle
of loading these are presented in Figure 2. The specimens were prepared from
finger joints that had been manufactured as described above. Notches with
thickness of 0.2 mm were cut. Similar types of specimens cut from green and
flatwise glued bonds tested in Mode I and II were presented and evaluated earlier
(Sterley 2012). Specimens were specially designed for stable performance in order
to make it possible to record the complete load-deformation curve. From such
curves strength, fracture energy and brittleness of the bond can be calculated.
a) b) c)
Fig. 2 Specimens cut from finger joint for small scale mixed-mode testing of adhesive
bonds: a) specimen location in the finger joint, b) dimensions, c) loading principle (the fibre
direction is indicated)
2.6 Tensile Testing

The test set up is seen in Figure 1. A servo-hydraulic MTS 810 testing machine
with a capacity of 100 kN was used. For deformation measurements and control of
loading rate two external Linear Voltage Differential Transducers (LVDTs)
mounted on both edges of the finger joint at the nominal distance of 45 mm were
used. The loading rate was 1 mm/min. Fracture path was determined with respect
to failure within fingers and adhesive, failure in wood at the joint and outside of
the joint. A few of the specimens were also studied with help of the optical
deformation measurement system ARAMIS. One side of the specimen was
covered with a pattern applied by spraying of black paint. The present
measurements of strain distribution with the ARAMIS set-up were accomplished
with a relatively large field of view of 100100 mm2, which implied a so-called
facet size of 0.50.5 mm2. The strains reported in ARAMIS are based on
averaging over an area of 3 3 facets. This means that the strain reported at a
certain point is in reality a representation of some average strain over an area of
1.51.5 mm2 in size.
2.7 X-ray Tomography

The morphology of the adhesive bonds was studied using micro-computed
tomography with an XRadia MicroXCT-200 instrument. The sample examination
was carried out using 4 magnification giving voxel resolution of 4.87 m. The
following analysis parameters were applied: the X-ray source voltage was 30 kV
and the power was 6W, the distances from the detector and the X-ray source were
12 mm and 35 mm respectively, the number of projections was 1800 and the
exposure time was 7s/projection for all the samples.
2.8 Durability
Durability testing was carried out according to ASTM D4688, which is an
American standard for structural finger joints. At present, there is no European
standard for testing of the durability of finger joints. The specimens were tested in
tensile loading parallel to the fiber direction with load rate of 5 mm/min. Tensile
strength and failure type, distinguishing between wood and adhesive failure, was
determined. Three specimens were cut from each joint and testing was carried out
after three different treatments, described in Table 1. A total of 9 groups with 14
specimens in each were tested.
Table 1 Groups of specimens tested according to ASTM D 4688, the symbols are used in
figures
Green Dry Dry

glued glued glued Description of treatment
PUR PUR PRF
RG RD RPRF Dry reference tested in dry state
Wet reference tested in wet state after vacuum-
VG VD VPRF
pressure impregnation with water
BG BD BPRF wet testing after 6 boiling and 5 drying cycles
Fig. 3 Set-up for mixed-mode testing

2.9 Mixed-Mode Testing of Adhesive Bonds

A specially designed set-up for mixed-mode testing is shown on Figure 3. The
specimen is glued into the specimen holder. For deformation measurements two
external Linear Voltage Differential Transducers (LVDTs) were used, monitoring
horizontal and vertical deformation, respectively. The testing was performed with
displacement control. The shear (horizontal) load rate was 0.001687 mm/s and the
normal (vertical) load rate was 0.0002098 mm/s the resulting ratio between them
thus being 0.1244. Horizontal and vertical load and deformation were recorded.
3.1 Tensile Test

As shown in Table 2, the strength of the high density joints was significantly
higher compared to the low density joints, the average strengths being 74 MPa and
54 MPa, respectively. This was valid for all three types of joints. There were no
significant differences between the strength of the green glued and dry glued
joints. However, there were some differences in terms of fracture path for the
different joints, (see Table 2). High density PRF joints failed mainly within the
finger joint itself (i.e. adhesive failure and shallow wood failure within fingers),
high density dry glued PUR joints failed partially in the surrounding wood and
partially in the finger joint area and high density green glued joints failed mainly
outside of the joint area, often in the wood under the steel plates at the specimen
ends, and also partially in the vicinity of the finger joint area. For low density
Table 2 Tensile strength and different fracture path of finger joints, SD in parenthesis
High density wood Low density wood

Dry Dry Green Dry Dry Green
glued glued glued glued glued glued
PRF PUR PUR PRF PUR PUR
Tensile strength
(MPa) 72 (9) 75 (10) 75 (8) 58 (10) 54 (12) 51 (11)
Fracture path
(no of specimens):
Finger and/or
adhesive 6 4 2 5 2 0
Wood at joint 2 1 2 4 5 6
Wood outside joint 2 5 6 1 3 4
joints wood failure percentage was generally higher than for high density joints.
Moreover no adhesive failure was observed in low density joints as for high
density joints. Low density PRF joints failed partially in the finger joint area and
partially in the wood, low density dry glued PUR joints failed to a lesser extent in
the finger joint area but also in the wood outside the finger joint area while the low
density green glued joints failed in the wood mainly outside the joint area, no
finger joint failure was observed (see Figure 4).
a) b) c)
Fig. 4 Failure of specimens a) dry glued PRF, failure in the adhesive and the joint, b) dry
glued PUR, failure in a knot outside of the joint, c) green glued PUR, failure in wood
outside of the joint
Thus, comparing PRF and PUR joints it can be seen that the PRF joints tend to
fail within the adhesive bond line and the joint area while the PUR joints have a
tendency to fail outside of the joint area (see Figure 4). This behavior existed both
in high and low density joints. However the average tensile strength of the joints
did not differ between the two adhesives. This result suggests that the PUR joints
(both dry and wet glued, but especially the wet glued ones) are stronger than the
PRF joints and even stronger than the wood in the joint areas tested, this being
true also for the high density wood specimens.
3.2 Strain Field Measurements

With the ARAMIS system, strains in a coordinate system aligned with the main
board directions were obtained: x (normal strain perpendicular to the loading
direction), y (normal strain in the loading direction) and xy (shear strain). In order
to study the strains along the bond lines of the fingers a rotation of the coordinate
system was also applied (not presented here, see Sterley 2012 for more details).
As described above, the strain calculated in each point by the ARAMIS system is,
in our case, an average over an area of 1.51.5 mm2 in size. The bond line
thickness is estimated to be about 0.1 mm and together with the adhesive
penetration depth the complete bond thickness is in the range of 100-500 m.
Therefore, the measured strains do not represent only the strains in the adhesive
bond but rather an average across the bond line which includes the strains in the
wood. A general conclusion drawn is that with the present set-up no clear
differences between the adhesive types could be observed (Sterley 2012). More
accurate measurements (for example with only one or a few fingers and a smaller
field of view) are needed to study the differences between adhesives in detail.
As an example, Figure 5 shows the distributions of y (normal strain in the
loading direction) on the specimens surfaces obtained at 50 MPa nominal tensile
stress for all three types of specimens. The maximum level of strain is approx. 1%
for all three types. However, one interesting difference can be seen: the area with
strain concentrations at the fingertips is larger in the green glued joint than in the
dry glued joints. This could be explained by the morphology of the green glued
fingers i.e. the deep penetration of the adhesive and the compressed wood in the
finger root, causing larger zones of material with higher stiffness.
x
a) b) c)
Fig. 5 Distribution of y (normal strain in loading direction) at 50 MPa stress: a) dry glued
PRF, b) dry glued PUR and c) green glued PUR. The same specimens as shown in Figure 4.
3.3 Penetration of Adhesive

The penetration of the adhesive in three different finger joint types was
determined with microtomography, an x-ray based, non-destructive, method.
Figure 6 Figure shows the penetration of the adhesive within a finger joint for the
three bond/adhesive types investigated. It is clear from Figures 6 and 7 that
the green glued PUR finger joints show deeper penetration of the adhesive. The
adhesive is seen inside the cell lumen at a depth equal to several cell rows from
the bond. Furthermore, the average size of the bubbles in the adhesive layer
(caused by the formation of CO2 during curing) is smaller compared to the dry
glued joints. The dry glued PRF adhesive has a penetration at the same level as the
dry glued PUR adhesive. The PRF finger joints failed more often in the adhesive
layer or within shallow fiber layer and the explanation for this can be the higher
brittleness of the PRF adhesive. On the contrary, the bond with the more ductile
PUR adhesive, despite the cavities present in the bond, can fail more frequently in
the wood, with the strength of the joint in both cases being at the same level.
Dry glued PRF Dry glued PUR Green glued PUR
Fig. 6 Pictures across fingers illustrating adhesive penetration (upper row) and cavities in
the finger tips (lower row)
a) b)
Fig. 7 Comparison of dry glued PUR bond (a) and green glued PUR bond (b)
Figure 7 illustrates different morphologies of the dry and green glued finger
joints with respect to adhesive penetration and cavities. Furthermore, Figure 7
shows that in green glued joints the finger top can be pressed more tightly to the
finger root and create a more homogeneous close contact joint. The transfer of
stresses between the two different wood pieces in the finger joint can be smoother
with such bond morphology. It is possible that such adhesive bonds better can
counteract the stress concentrations present at the finger tips. This can be a
possible explanation of the fact that the green glued finger joints did not fail
within the finger joints or in the adhesive but mainly in the surrounding wood or at
a long distance from the joint.
3.4 Durability
Durability test results are shown in Figure 8. The results show that there are no
significant differences between green glued joints and dry glued joints with PUR
adhesive. Joints glued with the PRF adhesive show higher strengths after the
boiling cycles than both green and dry glued PUR joints do. The PRF adhesive
was thus more resistant to boiling and water treatment than the PUR adhesive.
This difference in behavior resulted in different fracture paths: the PRF joints
failed mostly in the wood and the PUR adhesives failed in the joint and in the
adhesive. The green glued PUR joints failed to a lesser degree in the adhesive than
the dry glued PUR joints. However, the results show for all three adhesive systems
that the respective wet reference strength values (specimens tested in wet
conditions) is at the same level as the corresponding strength after boiling (i.e. no
significant difference exists for a certain combination of adhesive type and
moisture content during gluing) which in turn means that boiling did not affect the
a) b)
Fig. 8 Tensile strength of finger joints a) according to ASTM D4688 and fracture path in
durability testing b). Number of specimens failed in the adhesive is marked with a black
bar. R= dry reference, V= wet reference, B= wet result after boiling, G= green glued, D=
dry glued PUR, PRF= dry glued PRF.
strength, but instead, strength is only affected by the (high) moisture content at
testing.
3.5 Mixed-Mode Testing of Adhesive Bonds

Preliminary results from mixed-mode testing are presented in Table 3. The
evaluation method and calculation of effective fracture energy Gf,eff and brittleness
ratio are presented in Sterley, 2012. The Mode II results are presented here,
because this mode was dominating in the present geometry of fingers. The results
presented here will be evaluated and discussed separately but the preliminary
statistical analysis shows that with 95% confidence intervals no significant
differences were established between different types of specimens for the
measured properties. The trend is that the dry glued PRF bonds possess the highest
shear strength and are as brittle as the wood. There is no difference in shear
strength between the dry and the green glued PUR bonds. The dry glued PUR
bonds have the lowest brittleness and the highest effective fracture energy. The
green glued PUR bonds are less brittle than the wood and the PRF bonds but more
brittle than the dry glued PUR bonds. Similar results have been published earlier
(Sterley 2012).
Table 3 Results from mixed-mode testing of adhesive bonds Mode II results. SD in

parenthesis
Effective
Shear fracture Brittleness
Type of No of strength energy Gfeff ratio fv2/Gfeff
specimen specimens (MPa) (Nm/m2) (GPa/m)
Wood 12 15,2 (1,77) 584 (145) 402 (47)
Green glued 9 14,8 (1,77) 615 (155) 363 (31)
Dry glued 7 14,9 (1,27) 893 (472) 296 (117)
Dry glued 9 16,7 (1,55) 760 (319) 408 (119)
4 Conclusions
The following conclusions can be drawn from this study:
The tensile strength of finger joints glued with dried wood with PRF and
PUR adhesives and finger joints glued with unseasoned wood with PUR
adhesive is at the same level.
The higher the wood density the higher the strength of finger joints,
independent of adhesive type tested in the present study.
The green glued finger joints are characterized by high adhesive penetration
and smaller size of the cavities within the adhesive bond compared to dry
glued PUR bonds.
The tensile strength of green glued finger joints in the wet state after 6
boiling and 5 drying cycles is the same as the wet strength of non-cycled but
water impregnated finger joints, which implies that the durability of the
green glued joints is sufficient.
The tensile strength of green glued PUR finger joints after boiling/drying
cycles was at the same level as that of dry glued PUR finger joints.
The shear strength of green glued PUR adhesive bonds tested with small
scale specimens was at the same level as that of dry glued PUR adhesive
bonds.
Acknowledgments. The financial support from Sdra Timber AB, CBBT and the
Knowledge foundation (KK-stiftelsen), which made this research possible, is gratefully
acknowledged.
References
Anon. ASTM D 4688-95 Standard Test Methods for Evaluating Structural Adhesives for
Finger Jointing Lumber
Anon. EN 385: 2001 Finger jointed structural timber. Performance requirements and
minimum production requirements
Follrich, J., Teischinger, A., et al.: Tensile strength of softwood butt end joints. Part 1:
Effect of grain angle on adhesive bond strength. Wood Material Science &
Engineering 2(2), 8389 (2007)
Gindl, W., Schberl, T., et al.: The interphase in pheno-formaldehyde and polymeric
methylene di-phenyl-di-isocyanate glue lines in wood. International Journal of Adhesion
and Adhesives 24, 279286 (2004)
Karastergiou, S., Mantanis, G.I., et al.: Green gluing of oak wood (Quercus conferta L.)
with a one-component polyurethane adhesive. Wood Material Science &
Engineering 3(3-4), 7982 (2008)
Konnerth, J., Andreas, V., Gindl, W., Mller, U.: Measurement of strain distribution in
timber finger joints. Wood Science Technology 40, 631636 (2006)
Kreibich, R.E., Hemingway, R.W.: Formulation of tannin-based adhesives of the
honeymoon type for end jointing wood at extreme ranges of moisture content. In:
Christiansen, A.W., Conner, A.H. (eds.) Proceedings of Wood Adhesive Symposium.
Forest products Society, Madison (1996)
Kreibich, R.E., Steynberg, P.J., Hemingway, R.W.: End Jointing Green Lumber with
SoyBond. In: Swanson, J.S. (ed.) Wood Residues into Revenue: Proceedings 2nd
Biennial Residual Wood Conference, November 4-5, pp. 2836. MCTI
Communications, Inc, Richmond (1998)
Lange, D.A., Fields, J.T., Stirn, S.A.: Fingerjoint Application Potentials for One-part
Polyurethanes. In: Proc. No.7260 (EA). Wood Adhesives 2000 Extended Abstracts,
pp. 1718. Forest Products Society, Madison (2000)
Muszyski, L., Clouet, B., Pommier, R.: Optical Measurement of Local Strains
Development in Finger-jointed Wood Subjected to Static and Sustained Loads. In:
Proceedings of the Photomechanics Conference Loughborough, UK, July 7-9 (2008)
Parker, J.R.: Greenweld process for engineered wood products. In: Proceedings the
International Panel and Engineered Wood Technology Exposition, Atlanta, GA, USA,
October 5, 10 p. (1994)
Pizzi, A., Roux, D.G.: The chemistry and development of tannin-based weather- and
boil-proof cold-setting and fast-setting adhesives for wood. Journal of Applied Polymer
Science 22, 19451954 (1978)
Pizzi, A., du T. Rossouw, D., Knuffel, W.E., Singmin, M.: Honeymoon Phenolic and
Tannin-based Adhesive Systems for Exterior Grade Finger-Joints. Holzforschung und
Holzverwertung 32(6), 140150 (1980)
Pommier, R., Elbez, G.: Finger-jointing green softwood: Evaluation of the interaction
between polyurethane adhesive and wood. Wood Mater. Sci. Eng. 1, 127137 (2006)
Serrano, E., Oscarsson, J., Enquist, B., Sterley, M., Petersson, H., Kllsner, B.: Green-glued
laminated beams High performance and added value. In: Proceedings (Poster session)
of 11th World Conference on Timber Engineering, Riva del Garda, Italy, June 21-24,
pp. 829830 (2010)
Simon, F., Bredesen, R.: Innovative Programs Led in France to Promote the Use of
Domestic Woods in the Field of Engineered Wood Products, Wood Adhesives, Lake
Tahoe (2009)
Steffen, A., Johansson, C., Wormuth, E.: Study of the relationship between flatwise and
edgewise modull of elasticity of sawn timber as a means to improve mechanical strength
grading technology. European Journal of Wood and Wood Products 55(2), 245253
(1997)
Sterley, M., Blmmer, H., et al.: Edge and face gluing of green timber using a one-
component polyurthane adhesive. Holz als Roh- und Werkstoff 62, 479482 (2004)
Sterley, M., Blmer, H., Rydholm, D.: Finger jointing of green Scots pine. In: Kllander, B.
(ed.) Proceedings of International Conference / Workshop Green Gluing of Wood-
Process Products Market, COST Action E34 Bonding of Wood in cooperation with
SP Swedish National Testing and Research Institute, Bors, April 7-8 (2005) ISBN
91-85533-31-9
Sterley, M., Serrano, E., et al.: Flat wise green gluing of norway spruce for structural
application. In: Proceedings of Conference Wood Adhesives 2009, Lake Tahoe, USA
(2009)
Sterley, M.: 2012 Characterisation of green-glued wood adhesive bonds. Linnaeus
University Dissertations No 85/2012. Edited by Linnaeus University Press (2012)
ISBN:978-91-86983-57-4
Troughton, G.E., Chow, S.: Finger Jointing Unseasoned Rough Western Red Cedar
Lumber Using the WFPL Method. Forintek Canada Corp., Vancouver, WFPL
Tech.Rep.No.11 (1979)
Troughton, G.E., Chow, S.: Finger Jointing Kiln-Dried and Unseasoned White Spruce
Lumber Using the WFPL Method. Forest Products Journal 30(12), 4849 (1980)
Verreault, C.: Performance evaluation of green gluing for finger jointing. Forintek Canada
Corp. Raport No.2295 (April 1999)
Verreault, C.: Assesment of two green gluing processes for finger jointing. Forintek Canada
Corp. Raport No. 2407 (April 2000)
Pressure Distribution in Block Glue Lines
Analyzed by Theory of Beams on Elastic
Foundation
Gordian Stapf and Simon Aicher
Materials Testing Institute University of Stuttgart (MPA Stuttgart, Otto-Graf-Institut

(FMPA)), Pfaffenwaldring 4 B, 70569 Stuttgart, Germany
{gordian.stapf,simon.aicher}@mpa.uni-stuttgart.de
Abstract. Block gluing of glulam beams is an effective production method to obtain

massive wooden elements of cross-sections with very large dimensions. Due to the
extensive gluing areas, it is difficult to impose clamping pressures similar to those
as in standard glulam productions. The paper analyzes the pressure distribution per-
pendicular and parallel to the clamping bars by means of plain stress continuum
analysis and more notably by theory of beams on elastic foundation. Hereby the
stiffness perpendicular to the grain of the beams is realized implicitly by the bed-
ding stiffness. Closed form solutions are given to determine pressures and desirable
cambers of the clamping bars. The effect of nonlinear pressure distribution was in-
vestigated experimentally with a full scale block glued specimen tested in block
shear, delamination and bending whereby the test results are compared against glue
line thicknesses. The test results showed a good agreement with the analysis for
higher glue line thicknesses at the calculated low pressure areas. In the mechanical
tests, no correlation between the rather thin glue line thicknesses and strength values
was detected.
Keywords: clamping pressure, block glulam, beam on elastic foundation, glue line
thickness, block shear, delamination.
1 Introduction
While glulam can be produced in almost limitless heights (in practice limited by the
planing equipment), the production width is greatly restricted. The reasons for the
limitations in width are twofold. Primarily, it stems from the restricted availability
of wide boards, which by edge gluing of the narrow faces of the boards can be
overcome only to a certain extent. Further, the width is limited by manufacturing
equipment. Finger joint cutters, planers and presses are designed to match standard
lamella dimensions, which are normally up to 24 cm and only in select factories up
to 28 cm in Central Europe.
DOI: 10.1007/978-94-007-7811-5_31, c RILEM 2014
342 G. Stapf and S. Aicher
In building elements that are subjected to high loads, these limitations in one
direction of the cross-section can be rather inhibiting. Compression elements which
require high section moduli in both directions to prevent buckling are one example.
Another example where massive cross-sections exceed the ranges of regular glulam
are bridges where by block gluing, the deck and the load bearing structure can be
manufactured in one element.
The limiting width can be overcome by laminating two or more glulam beams,
referred to as block gluing. In national standards, block glued glulam is a compar-
atively recent phenomenon. In Germany for example, it was first introduced in the
timber design code DIN 1052:2004 and subsequently adopted as a new building
product with its inclusion in the former German List of Building Products (Bau-
regelliste 2006/1) [1].
Block glued beams of rectangular cross-sectionalthough effectively existent
for a long timewill eventually be covered in Europe with the introduction of the
new glulam product standard EN 14080:2013 (expected spring 2014). In contrast,
arbitrary shaped block glued cross-sections like t-beams or box-section slabs cov-
ered by DIN 1052 are not included in EN 14080.
Unlike the German national standards, EN 14080 does not specify a distinct
clamping pressure during block gluing. Instead, the standard references the technical
data sheets of the adhesive producer, where detailed values for clamping pressure,
usually at least 0.3 MPa, are given. The thickness of the glue line must be smaller
than 1.5 mm, which is in line with the requirements for gap filling adhesives accord-
ing to prEN 301:2013 [2]. Another requirement is that the glulam components to
be bonded shall not be subjected to bending stresses during cramping.
2 Clamping During Block Gluing
2.1 Load Distribution along Element Axis

The obvious way to clamp together two or more beams that have a width wblock =
wglulam and modulus of elasticity in bending E0 is to use discrete clamps with a
spacing aspace , see Fig. 1.
Disregarding the distribution of the load under the clamps perpendicular to the
axis of the element for the time being, it is essentially important to know how the
spacing of the clamps influences the distribution of the pressure on the glue line
under and especially in between adjacent clamps. Without considering the fluidity
of the glue line itself in a preliminary approach, the system can be treated by means
of classical mechanics in two dimensions, schematically shown in Fig. 2a. This sys-
tem can be considered approximately as an edgewise loaded plate (in the following
referred to as disc) which can be analyzed by Airys differential equation for peri-
odic concentrated and distributed (Fourier) loads [3]. However, the solution is a very
clumsy expression being further exclusively valid only for the isotropic case. As the
large difference between bending MOE E0 and modulus of elasticity in compression
Pressure Distribution in Block Glue Lines Analyzed by Theory of Beams 343
Fig. 1 Typical way to clamp a block glued element
perpendicular to the grain Ec,90 has a strong influence on the stress distribution, the
isotropic solution is merely useful for comparative considerations.
Another approach to analyze the stress distribution consists in splitting the bend-
ing stiffness virtually from the compression stiffness by arranging a spring system
between the two coupled/glued beams, see Fig. 2b. To enable the calculation of the
system as a single beam on a continuous elastic support, the bending properties of
the beams on either side of the glue line have to be unified into one single beam, so
that the mechanical model of the block gluing process can be reduced to the scheme
depicted in Fig. 2c.
The bedding modulus c is composed of the effective thickness de f f = (1/dA +
E
1/dB)1 and the modulus of elasticity perpendicular to the fiber Ec,90 , i. e. c = dc,90
ef f
.
The beam that represents the bending properties of the system should exhibit the
same deflection wtotal = wA + wB as the sum of the deflection of the two beams
that are clamped together, wA and wB . As the bending of a beam is in general a
function with EI in the denominator, w(x) = EI 1
f (x), the total moment of inertia can
be expressed in a reciprocal relationship as (assuming Em,0,A = Em,0,B = E0 )
1
Itotal = 1
IA + I1B . (1)
The pressure under an elastic beam is the deflection multiplied by the bedding
modulus p(x) = w(x) c. The deflection function for concentrated forces can be
taken from Mechanics textbooks, e. g. [4]. For a single clamp and an element with
infinite length, the pressure in the glue line pglueline (x) depending on the distance x
from the clamp can be expressed as
Fclamp c
pglueline,F (x) = e|x| (cos |x| + sin |x| ) (2)
wglulam 8 3 E0 Itotal
with

wglulam c
= 4
4 E0 Itotal
Fclamp = 2 Ftight
The absolute value |x| in Eq. 2 stems from the solution of the deflection function,
which is derived only for positive x.

Fig. 2 Conversion of the real gluing/pressing system to a beam on an elastic foundation

a) idealized real system
b) system divided into two beams only subjected to bending and an elastic coupling
c) final system: single beam on an elastic foundation
In case the width of the clamps wclamp is large compared to the thickness of the
glulam dglulam (or, better, versus the bending stiffness E0 I of the adherends) and/or
when wclamp is large in comparison to the distance of adjacent clamps aspacing , Eq.
2 does not hold. In this case, the single load Fclamp can be smeared as a line load
p0 = Fclamp /wclamp over the width of the clamp wclamp , which gives by integration

x+wclamp /2
1
pglueline,p0 (x) = pglueline,F (x)dx (3)
xwclamp /2 wclamp

x+wclamp /2
p0 c
= e|x| (cos |x| + sin |x| ) dx
xwclamp /2 8 3 E0 Itotal
However, the discontinuity at x = 0 that results from the solution of the differen-
tial equation for x 0 allows solely forward integration at the locations of interest
x+w /2 x+w /2
0 x < wclamp /2, i. e. only x clamp and not xw clamp/2 . This obstacle can be
clamp
overcome by making use of the symmetry of the deflection function around zero.
The direction of integration at x = 0 can be inversed as long as x < wclamp /2. In
mathematical terms this can be expressed as

|x|+wclamp /2
pglueline,p0 = pglueline,F (x, F = p0 )dx +
0

0
k(x)
wclamp pglueline,F (x, F = p0 )dx
|x| 2

p0 w wclamp
|x|+ clamp
= 4 1e 2
cos |x| + +
E0 Itot 2

w

|x| clamp wclamp
k(x) 1 e 2
cos |x| (4)
2
where wclamp
1 for |x| < 0
k(x) = w 2
1 for clamp
2 |x| 0
A periodic deflection and pressure function that will occur under clamps with
a fixed distance aspacing can be achieved from superposition of Eqs. 2 or 4 via the
summation
n
pglueline,periodic (x) = pi,glueline,single (x + i aspacing) (5)
i=n
where
n S 6/aspacing
1
S= characteristic length

In order to prove the validity of the newly developed method, the beam solution
is plotted in Fig. 3 together with the disc solution at five different sections within
the cross-sectional depth (middle of the cross-sectional depth is denoted by 0.5d,
the loaded edge by 1.0d). The disc solution is of course dependent on the number
of Fourier summands; the presented graphs are based on 500 summands. It can be
seen that there is virtually no difference between the disc and the beam solution at
any position within the cross-section.
In Fig. 4, the isotropic disc and beam solutions for concentrated and distributed
loads are compared with pseudo-orthotropic solutions where the modulus of elas-
ticity (MOE) perpendicular to the grain Ec,90 = 370 MPa is considered by the spring
constant c = Ec,90 /dblock . The following general statements about the different
calculation approaches can be made:
Fig. 3 Comparison of pressure perpendicular to glue line calculated as a disc [3] and as a
beam on elastic foundation (Eqs. 4 and 5) are given
The beam solutions are in good agreement with the disc solutions for both load
cases, when isotropic material properties are assumed.
The pressure distribution for the distributed loads is significantly different from
the pressure distributions of the approximations where concentrated forces are
considered; this is especially the case of isotropic material properties.
When orthotropic material properties are used, the differences between the dis-
tributed load solution and the approximation with a concentrated force are minor.
The orthotropic material properties lead to a better distribution of the glue line
pressure under and between the clamps in comparison with isotropic solutions.
Comparison with finite element simulations with a complete orthotropic consti-
tutive law showed a good agreement with the pseudo-orthotropic beam solution,
partially depending on the shear modulus that cannot be captured easily within the
model of a beam on an elastic foundation.
2.2 Pressure Distribution across the Element Width

The clamps usually consist of two beams with stiffness Em,clamp and moment of
inertia Iclamp that run parallel to the glue line and perpendicular to the axis of the
element, see Fig. 1. The two beams are tightened by two tension rods on each side of
the element with forces Ftight . To avoid excessive bending of the clamps, the tension
rods are ideally arranged with no clearance with respect to the element, adist , but
Fig. 4 Comparison of pressure perpendicular to glue line calculated as disc [3] and as beam
on elastic foundation (Eqs. 4 and 5). Solutions for concentrated and distrubuted loads assum-
ing isotropic and orthotropic materials (beams).
production circumstances and/or lack of knowledge often lead to a significant value

of adist .
If the beams used for clamping are straight with parallel edgeswhich is typical
in glulam companiesthe pressure distribution under the clamping beams can be
derived from the solution of a beam on an elastic foundation with boundary shear
forces Q = Ftight and boundary moments M = Q adist at the locations x = wblock /2
and x = wblock /2 (wblock is the width of the block glulam). This definition of bound-
ary conditions implies that x = 0 lies in the symmetry axis of the system. Hence,
two other valid boundary conditions are maximum deflection and zero shear force
at x = 0: w (0) = 0 and w (0) = 0, see Fig. 5.
The above boundary conditions are sufficient for finding the solutions for the four
integration constants A, B, C and D of the classical deflection trial function
w(x) = e x [A cos( x) + B sin( x)] + e x [C cos( x) + D sin( x)] . (6)
This leads to the final solution that can be written as
e (|x|)
2 |x|
w(x) = e 1 sin( |x|)
2EI 3 (sin( wblock ) + sinh( wblock ))

[ M sin cosh + sinh ( M cos Q sin )] e2 |x| + 1

cos( |x|) [cosh ( M sin + Q cos ) M cos sinh ] . (7)
Fig. 5 Mechanical system to calculate the pressure distribution under a clamp
where
wblock
=
2
EI = Eclamps Iclamps /2 (compare with Eq. 1)
The pressure under the clamp, pclamp , then results in
pclamp = c wclamp (8)
where
Ec,90
c= for clamps with a very high compression stiffness Ec,90 (e.g. steel) and
dblock /2
Ec,90
c= for wooden clamps. (9)
dblock /2 + dclamps/2
In order to evaluate Eq. 7 in quantitative manner, the following example has been
studied: Given is
a glued block with thickness dblock = 200mm, width wblock = 1000mm and
Ec,90 = 370MPa,
a wooden clamp with wclamp = 100 mm, E0 = 11 GPa, dclamp = 150 mm to
500mm,
a tightening with Ftight = 100 kN at a distance of adist = 150mm.
The results are plotted in Fig. 6. It is then evident, that with thicknesses up to
dclamp 300 mm, parts of the cross-section are not subjected to any pressure at
all when linear elastic behaviour of clamps and glulam is assumed. (Note: The the-
ory of beams on elastic foundations is only valid as long as the beam does not lift
from the foundation, i. e. the clamp shall be in contact with the beam over the en-
tire width. Since having regions with zero or negative pressure is highly undesirable
during bonding, this restriction is acceptable).
Fig. 6 Pressure distribution 2

under a clamp for specific
geometry and stiffness con- 0
pressure p under the clamps in MPa

figuration (see text)
2
6
d= 150mm d= 350mm
8 d= 200mm d= 400mm
d= 250mm d= 450mm
d= 300mm d= 500mm
10 400 200 0 200 400
width of block glued glulam w
block in mm
The nonuniform pressure under straight clamps can easily be overcome by cam-
bering the contact area between clamp and glulam. The desired convex form of the
camber can be found by simply regarding the clamp as a single-span beam on two
supports with a partial symmetric line load where the symmetric line load is the
sought after pressure under the clamp, pclamp , and the bolts that tighten the clamp
are the supports as shown in Fig. 7.
Fig. 7 Mechanical system to calculate the required cambering of a clamp
Then the deflection w equals the desired camber. Since the deflection of a partial
line load wtot is not given in beam tables, it can be superimposed as a continuous
line load on a beam w p with the length wblock and the deflection wP of a beam with
two single loads P = pclamp wclamp /2 = Ftight at adist and a total beam length of
lclamp = wblock + 2 adist :
wtot (x) = w p (x) + wP (x) (10)

pclamp x
w p (x) = x3 2 wclamp x2 + w3clamp (11)
24 Eclamp Iclamp
pclamp wclamp adist
wP (x) = 3 lclamp 3 2 a2dist (12)
2 6 Eclamp Iclamp
with
= x + adist
3 Pressure Distribution and Glue Line Thickness

The experimental investigations included the block gluing of two glulam beams,
each of them with the dimensions 280 mm x 1000mm x 5000mm. The adhesive
was a One Component Polyurethane (1K-PUR) adhesive, certified for structural
gluing, however not for block gluing. The pressure was applied as sketched in Fig.
1 with steel clamps, each of them with a bending stiffness of E I = 8 MNm2 . The
calculated pressure distribution under and between the clamps according to Eqs. 4
and 8 is plotted in Fig. 8. (Note: The middle part of the length of the manufactured
element where no glue line thicknesses were measured are omitted from the plot.)
Fig. 8 Calculated pressure distribution perpendicular to glue line plotted together with mea-
sured glue line thicknesses
After production of the block glued beam, cross-sectional slabs were taken to
produce block shear test specimens and delamination specimens according to EN
14080:2013. Before testing, the block shear specimens with dimensions 50 mm x
50 mm were used for determination of the glue line thickness. The glue line thick-
ness was measured for each specimen three times at both specimen sides where the
fibers are parallel to the surface with a measuring microscope at a magnification of

50x. The results of the glue line thickness measurements are additionally plotted in
Fig. 8. It can be seen that due to the comparatively stiff clamps and the high ra-
tio dglulam : aspace = 0.31 the calculated pressure distribution showed a fairly good
correlation to the calculated average value of 0.5 MPa. The minimum and maxi-
mum values ranged between 0.36 MPa and 0.72 MPa, respectively. Nevertheless,
glue lines at regions with lower pressure tended to be thicker than at regions with
a higher pressure. Especially in regions with lower pressure, glue line thicknesses
of over 0.3 mm occurred which are at present usually not permissible for 1K-PUR
adhesives according to EN 15425 [5].)
4 Test Results
All specimens revealed a sufficiently high block shear bond strength of fv > 6 MPa
and fulfilled the combined strength and wood failure percentage criterion of EN
14080:2013 [6] with one exception. Thus, the shear test results were judged to be
acceptable. The results of the block shear tests are plotted in Fig. 9 against the glue
line thickness. No distinct correlation between the block shear results and the glue
line thickness can be observed.

!

Fig. 9 Block shear strength

fv against glue line thick-
ness

The delamination specimens manufactured from slices of the cross-sections that

were neighbouring the locations from where the block shear specimens were taken
were tested according to test method B of EN 14080 (which is identical with test
method B of EN 391 [7]). Six of 16 delamination specimens did not meet the re-
quirement of 8 % total delamination after the second cycle which led to somewhat
different conclusions regarding durability of the bond line. Similar to the block shear
test results, no correlation with the glue line thickness could be detected.
From the middle of the manufactured block glued beam, three bending specimens
were taken to test the bending shear strength according to ASTM D 3737 [8]. All
specimens sustained a shear stress of more than the characteristic shear strength
value fv = 3.5 MPa specified for glulam in EN 14080 while failing unintentionally
in bending.
5 Conclusions
The paper gives analytical solutions for the calculation of the pressure distribution
under and between clamps based on the theory of a beam on an elastic founda-
tion. The theoretical approaches also include continuum solutions based on Airys
stress function and reveal differences between isotropic and orthotropic solutions.
Although the analytical solutions do not account for plasticity perpendicular to the
fiber at high loads and plate effects, they represent a helpful tool to obtain quantita-
tive estimates of the required distances between and the dimensions of the clamps,
respectively. With help of the derived equations, it could be shown that under a
completely straight clamp, there is always a nonuniform pressure distribution and
in some cases no pressure at mid-span at all. This problem can be overcome by cam-
bering of the clamp; a simple equation for the derivation of shape and scale of the
camber is given as well.
The derived equations were applied to a full scale block glued test beam from
which slices of the cross-section were taken after production to measure the glue
line thickness and to investigate the performance of the glue line by means of block
shear and delamination tests. The bond shear strength of the glue line was also tested
in large scale bending tests as well.
The small scale block shear and the bending shear strength values conformed to
requirements and characteristic values given in EN 14080, whereas the requirements
for the delamination tests were not met for all specimens. The tests revealed no clear
correlation between glue line thickness with shear strength and delamination results.
Further investigations will include flow properties of the adhesive in terms of
fluid-solid-interaction to obtain a better understanding of the relationship between
glue line thickness and pressure distribution during block gluing. Further on, it will
be evaluated if a more even pressure distribution in combination with a higher reg-
ularity of the adhesive spread can improve the delamination results.
References
[1] Deutsches Institut fur Bautechnik: Bauregelliste A. DIBt Mitteilungen 37(33), 4 (2006)
[2] EN 301:2013. Adhesive, phenolic and aminoplastic, for load-bearing timber structures
Classification and performance requirements (2013)
[3] Goldner, H.: Arbeitsbuch Hohere Festigkeitslehre. Elastizitatstheorie, Plastizitatstheorie,
Viskoelastizitatstheorie. VEB Fachbuchverlag Leipzig (1978)
[4] Szabo, I.: Einfuhrung in die Technische Mechanik: Nach Vorlesungen, vol. 8. Springer,
Heidelberg (1975)
[5] EN 15425:2008. Adhesives One component polyurethane for load bearing timber struc-
tures Classification and performance requirements (January 2008)
[6] EN 14080:2013 (E). Timber structures - Glued laminated timber and glued solid timber
Requirements (June 2013)
[7] Glued laminated timberDelamination test of glue lines; German version EN 391:2001
(April 2002)
[8] ASTM 3737-07. A7. Test Setup and Data Analysis Procedure for Determining Horizontal
Shear Stress by Full Scale Beam Tests (2007)
EPI for Glued Laminated Timber
Kristin Grstad and Ronny Bredesen
DYNEA AS, Svellevn. 33, 2001 Lillestrm, Norway

kristin.grostad@dynea.com
Abstract. Emulsion Polymer isocyanate adhesive (EPI) is a two component adhe-

sive system which combines an emulsion component and an isocyanate functional
cross-linking component. This combination gives glue line performance with the
benefits from both thermoplastic and thermosetting adhesive systems. The glue
line is cold curing, has high flexibility, low creep, contains no formaldehyde and
gives excellent water resistance in both cold and boiling water.
As the setting and curing behaviour are different from todays primarily used
adhesives for laminated beams (PRF, MF, MUF, 1K-PUR), the optimal produc-
tion parameters have to be verified.
EPI adhesives give very good adhesion and are because of this, well suited for
gluing difficult wood species like hardwood, which is a resource with a big poten-
tial. EPI adhesives can also be used for gluing of wood to metal. In addition, EPI
adhesives open for future production benefits which today are not covered by the
European production standards.
This article also describes the properties of EPI adhesives, status for the use of
EPI in glued laminated timber in Europe today and future possibilities.
Keywords: emulsion polymer isocyanate (EPI) adhesive, emulsion component,

cross-linking component, rheology, pot life, glue line thickness, press time.
1 Introduction
EPI adhesive systems have been used since the early 1970s in Japan [1]. EPI has
been described in the Japanese Industrial Standard JIS K6806 in 1985 [2], and has
in addition to PRF (Phenol Resorcinol Formaldehyde Adhesive) been used for
construction purposes since then. European producers of posts for Japan started to
use EPI for export and the demand for EPI approved for Europe was created. Re-
quirements for Japan do not include gap filling properties and systems used for
Japan do not fulfill the European requirements in this respect.
EPI adhesives are fast setting and cold curing and they give light colored glue
lines. The curing process is a combination of film formation which is a physical
process and chemical reactions of the isocyanate. The adhesive systems can be
optimized for different applications by changing the formulations, but generally
the systems have very good adhesion, glues difficult wood species very well and
glues metal to wood.
DOI: 10.1007/978-94-007-7811-5_32, RILEM 2014
356 K. Grstad and R. Bredesen
The first EPI system approved for glued laminated timber for Europe was
Prefere 6151/6651 in 2005 [3]. The first European standard for EPI adhesives,
FprEN 16254, has recently passed formal vote positively and will be implemented
in 2013. Due to the different behavior of EPI adhesives compared to condensa-
tion resins several investigations have been initiated to understand the behavior of
EPI and adhesives in general, and this has been described in this article.
2 Approvals for the Gluing of Structural Laminated Timber
The testing of the first EPI adhesive, Prefere 6151/6651, started in 2002 at Norsk
Treteknisk Institutt (NTI) with approval for use in the Nordic glulam industry. The
system has been used since 2003 for the production of the K-beam by Kjeldstad
with European Technical Approval (ETA) [4]. One request from the former Nor-
dic Glulam Committee (NLN) was that an approved EPI-system should have gap
filling properties to be able to overcome surface irregularities from the planing
process. Hence, Prefere 6151/6651 was developed and tested according to EN 301
with a gap of 0,3 mm. Maximum glue line thickness was set to be 0,2 mm and in
addition a dimension restriction was set. All tests for approval of service class 3
were fulfilled, but NLN chose temporarily to give limitation for use with respect
to glue line thickness until more experience with the adhesive system was gained.
The restrictions were later withdrawn.
In 2005 approval as adhesive type I was given by Deutsches Institut fr
Bautechnik (DIBt) [3] for the production of laminated beams according to DIN
1052 after testing according to Annex C in EN 14080:2005 and DIN 68141[5].
The approval included productions of structural laminated beams used in service
class 1, 2 and 3.
Since the first approval for the use of EPI for the gluing of laminated beams
was given in 2003, the following tests and approvals have been performed:
- Tested for reaction to fire according to EN 13501-1 by Technology Institute
for Forest Cellulose Wood-constructions Furniture (FCBA). The system confirms
to the class D-s2,d0.
- Tested and approved according to ASTM D 2559-04 (Standard specification
for adhesives for bonded structural wood products for use under exterior exposure
conditions) and ASTM D 3535 (standard test method for resistance to creep under
static loading for structural wood laminating used under exterior exposure condi-
tions by Timber Engineering Company (TECO), which enables our customers to
export glued beams to USA.
- Tested and approved according to JIS K 6806 (Water Based Polymer-
Isocyanate Adhesives for wood) [2] by NTI, Norway, which means that it, can be
used for export of posts to Japan.
- Tested and approved for the gluing of birch, larch, Maritime pine and impreg-
nated pine (Wolmanith CX-8 and Scanimp KF) by NTI.
EPI for Glued Laminated Timber 357
- Tested by NTI and approved in the Nordic for gluing of cold materials. Tests
were performed at 4C and 5C and the approval was given for minimum material
temperature of 10C.
In the close future the revised harmonized standard EN 14080:2013 and the
new EPI-standard EN 16254 will be active. According to the new regulations an
EPI glue line tested with a gap of 0,3 mm will be given a dimension restriction
whereas an EPI tested with a gap of 0,5 mm can be used without dimension re-
strictions. The EPI-family in general will be characterized as adhesive type I, but
only for the gluing and use of beams for service class 1 and 2.
Since glue line thickness has been one of the main issues during development
of the EPI standard this parameter has been studied in several contexts and some
of the results are summarized in chapter 5.
3 EPI Adhesives Composition and Reactions
The EPI adhesive is a two component system, an emulsion component and an

isocyanate functional cross-linking component. The two components and the
chemical reactions occurring in the adhesive system during gluing are described in
this chapter [6].
3.1 Emulsion Component

The emulsion component in the EPI adhesive consists of a mixture of polyvinyl
alcohol (PVA) water based emulsions, filler(s) and additives to obtain good dis-
persion of the filler and to prevent foam and fungi. The PVA is a viscosi-
ty/rheology modifier, prevents the sedimentation of filler and plays an important
part in the cross-linking reactions with the isocyanates to give water resistant glue
line. The emulsions are added to give the required setting time and water
resistance. The emulsions also improve the bond quality and heat resistance.
Emulsions (poly (vinyl acetate)-, ethylene vinyl acetate-, vinyl acetate-acrylate-,
acrylic-styrene- or styrene-butadien emulsion) with post cross linking groups give
the optimal bond line quality. Calcium carbonate is the most frequently filler used
in EPI adhesives but other organic or inorganic materials may also be used. The
functions of the filler in the formulations are to increase the solid content, improve
the heat resistance and to give gap filling properties in the final glue line. For
glued laminated timber it is important to ensure that the filler is soft to prevent
tool wear.
3.2 Cross-Linking Component

The cross-linking component, the hardener, is an isocyanate with preferably two
or more NCO groups. To prevent exposure of isocyanate in the production area it
is important to choose a low volatile and preferably a polymeric compound.
Diphenylmethan diisocyanate (MDI) and/or polymeric versions of MDI (pMDI)

(figure 1) are the most common isocyanates used for this purpose today due to the
low vapor pressure and high reactivity. The reactivity of the hardener is adjusted
by the composition of a mixture of 2,4 and 4,4 isomers of MDI.
O
C
N N
C
O
N
C
O
Fig. 1 Structure of polymeric MDI (pMDI), n= 0-4
3.3 Physical and Chemical Reactions Glue Line Formation

The curing process of the EPI adhesive is quite complex. Since the adhesive
system is a mixture of water based and non-water based components the mixing
efficiency is of great importance. Some of the isocyanate components for this
purpose have been modified with surfactants to simplify and improve the mixing
behavior.
The formation of a water resistant glue line involves both a physical and a
chemical process occurring simultaneously. The physical process is a coalescence
of the emulsion particles forming a film and the water penetrating into the wood.
The chemical process is the reaction of isocyanate (NCO) with functional groups
having active hydrogen; like primary amines, alcohols, water, urethanes and car-
boxylic acids. The reaction with water forms carbon dioxide (CO2) which causes
foam in the glue line (figure 2). It is important to have a glue system with limited
foaming to keep a good strength in the glue line.
R-NCO + R`-OH R-NHCOO-R`

R-NCO + H2O R-NHCOOH R-NH2 + CO2 (g)
Fig. 2 Reaction of isocyanate (NCO) with water and alcohol
The setting speed or how fast the initial strength is built up depends mainly on
how fast the water penetrates into the wood so the emulsion particles form a glue
film. This physical process is less influenced by temperature than chemical pro-
cesses and more dependent on the moisture content in the wood, and this is why
curing of EPI adhesives at elevated temperature does not respond to higher tem-
perature as much as thermosetting adhesives. The time to obtain a water resistant
glue line is a chemical process which is more dependent on the temperature than
the process to gain initial strength (Chapt. 5).
4 Gluing Performance
EPI adhesives have different performance properties compared to condensation

resins. The focus in this chapter has been to illuminate the areas of concern during
introduction of EPI as adhesive type for construction purposes.
4.1 Rheology and Pot Life

The EPI adhesive systems are preferably mixed in a resin to hardener ratio of
100pbw:15pbw. The viscosity will increase and the rheology will change in this
mixing process but the level it ends up on and the degree of shear thinning behav-
ior of the mixture depends on the formulation of the glue and on the type of pMDI
used. The amount of foam formed in the mixing process controls the rheological
behavior to a great extends. The viscosity may increase to a certain level and then
level out or continue to increase but this does not influence the pot life.
To obtain a water resistant glue line a sufficient amount of isocyanate from the
pMDI must be available for cross linking. The pot life is defined as the time peri-
od in which the glue mix can be used and still give water resistant glue line. The
pot life can be estimated by running delamination tests (EN 302-2) at different age
of the glue mixture.
4.2 Glue Line Thickness

Glue line thickness has been extensively discussed related to EPI adhesives. The
first EPI adhesive approved for laminated timber had a restriction of 0,2 mm glue
line thickness. Related to this restriction several investigations were initiated to
find the real glue line thickness in a laminated timber produced at different
production sites. Table 1 shows the variation in glue line thickness from several
producers measured in three different laboratories.
As shown in table 1 the glue line thicknesses are in general 0,1 0,05 mm, and
the mean values are always < 0,1mm, independent of the adhesive type. This
shows that for standard glulam beams production with good planing and applica-
tion control the glue line thickness is not a limiting factor.
Table 1 Glue line thickness in laminated timber
Test Production Number of samples/ Mean Min Max Refer-

institute site/ adhesive dimensions (mm) (mm) (mm) ence
type/wood (width(mm)*
height(mm))/number
of measurements on
each sample
Norwegian Several / 8/ 105*105-105*150 [7]
Institute of EPI/ 4 edge 0,05 0,02 0,08
Wood Tech- spruce, pine 4 centre 0,05 0,02 0,12
nology
Norwegian Several/ 12/105*105- [7]
Institute of PRF/ 105*360 0,09 0,02 0,24
Wood Tech- spruce, pine 3-11 edge 0,08 0,02 0,22
nology 3-11- centre
Norwegian Several/ 5/ 90*180-163*498 [7]
Institute of MUF/ 3-10 edge 0,06 0,02 0,12
Wood Tech- spruce 3-10 centre 0,07 0,02 0,12
nology
University of Rettenmeier, 6/4 / 240*240 (trio) [8]
Karlsruhe Werk Side A 0,09 0,05 0,17
Hirschberg Side B 0,09 0,05 0,11
/EPI/spruce
Technical Several / 4 different man. [9]
University of MUF/ 100*120-200*480
Munich spruce, larch A 0,1 0,04 0,25
B 0,09 0,03 0,21
C 0,07 0,03 0,11
D 0,08 0,02 0,22
4.3 Gluing Temperature and Moisture Content in the Wood

As mentioned, EPI adhesives do not have the same temperature dependence dur-
ing curing as thermosetting adhesives. The main process to develop initial
strength in the glue line is removal of water included in the film forming process.
EPI adhesives can be applied and cured at low temperatures, which means that the
wood materials, production site and storage might have lower temperature than
20C. This opens for potential cost savings and reduced energy consumption.
Gluing of construction materials has to follow the production standards with re-
spect to temperature but there is a possibility for approval of specific processes
which may utilize this possibility.
Glue spread is also an important production parameter. Typical glue spread for
EPI adhesives is 150-250g/m2 compared to 210-400g/m2 for MUF adhesives.
The glue spread required depends on the process, especially the assembly time and
the planing quality.
Figure 3 shows the pressing time (PT) and time to water resistance (WR) for
two typical adhesive systems approved for construction purposes. Pressing time
has been measured according to ASTM D 905 on spruce for thin glue lines, glue
spread of 250g/m2, with the requirement of 3N/mm2. This method differs from
EN 302-5 both with respect to wood species and performance. Time to water re-
sistance has been measured by delamination test according to EN 302-2 where the
time is the period from the beam has been put into press until the vacuum pressure
cycle has been started.
As shown in figure 3 the pressing time, time to built initial strength, does not
depend very much on temperature. The time to water resistant glue line which is
happening through a chemical reaction, depends on temperature to a larger extend.
However, the systems are water resistant within less than 10 hours even at lower
temperature than defined by production standards, and will not be the limiting
factor in a production line.
Strength development has been investigated for finger joints and figure 4 com-
pares EPI and MUF adhesives. This shows that the EPI adhesive has a two step
curve and MUF has slower strength development. This is why it is important to
know the required handling strength for the specific production process to choose
the correct production parameters. From experience strength of 20 N/mm2 has
shown to be equivalent to tough handling which gives a pressing time of 15
minutes for EPI and 60 minutes for MUF.
4
Time (hours)
0
0 10 20 30 40
Temperature (C)
PT System 1 PT System 2 WR System 1 WR system 2
Fig. 3 Pressing time (PT) and time to water resistance (WR) for two different EPI systems
at different temperatures
40
35
Strength(N/mm)
30
25
20
15
10
5
0
0 50 100 150 200
Time (minutes)
EPI MUF MF
Fig. 4 Strength development in a finger joint at 20C, 20 mm finger profile
The moisture content in the wood is of great importance since the main effect
to built initial strength is the removal of moisture from the glue mixture and
the formation of a glue film, being a physical process. Low moisture content in
the wood may reduce the pressing time significantly, and a combination of drying
and heating of the wood surface could give a pressing time of only a few minutes
[10].
Spraying water on the applied glue mix extends the assembly time of the
EPI adhesive systems [11]. This gives more flexibility to the production process.
4.4 Gluing Different Wood Species

EPI adhesives have different penetration properties compared to condensations
adhesives and very good adhesion to high density surfaces. This is the reason why
this adhesive family has shown to be very interesting for several types of wood
species which are known as difficult to glue.
Prefere 6151/6651 has for instance shown good performance on high density
species like Black Locust (Robinia pseudoacasia) and Jarrah (Eucalyptus
marginata). Due to the neutral pH of the adhesive family and the good adhesion it
is also shown high performance on low acidic species like Oak (Quercus) and
Chestnut (Castanea sativa). These species can also be glued with condensation
resins, but the low pH of the wood will influence the curing behavior.
The gluing of the species mentioned above is not approved for production of
engineered wood with EPI today. The approval status for gluing of difficult wood
species is shown below:
- MPA University Stuttgart, has tested and approved both Prefere
6182/6682 and Prefere 6151/6651 according to DIN 1052 for the gluing
of larch (Larix decidua), both with face gluing of laminated beams and
finger jointing.
- NTI has tested Prefere 6151/6651 and issued a Nordic approval for the
gluing of load bearing constructions with Birch (Betula pendula).
- After extensive testing by FCBA, Prefere 6151/6651 has been approved
for the gluing of load bearing constructions of Martime pine (Pinus
pinaster) in France.
- NTI has tested and approved Prefere 6151/6651 for the gluing of pine
(Pinus sylvestris) impregnated with Wolmanit CX-8 and Scanimp.
5 Summary and Future
EPI adhesives for glued laminated timber are fast setting and flexible adhesive
systems with good cold curing properties. The glue lines are light colored
with good moisture and heat resistance which fulfills the requirements for load
bearing constructions. The systems may be cured in a Radio Frequency field to
reduce the pressing time even further. Systems with low glue mixture viscosity
may be used for Cross Laminated Timber. Another advantage is the shear thin-
ning rheological behavior which fits very well in contactless application for finger
joints.
Several studies in Europe are now in process to use new and more difficult glu-
ing wood species for laminated beams. EPI adhesives with high flexibility and
good adhesion work very well without primers etc. required for other adhesive
types. Even gluing of metal to wood is possible for EPI adhesives.
EPI adhesive systems contain no formaldehyde and this is often used to pro-
mote this adhesive type. Both the adhesive and the hardener contain chemical
components with the risk of causing allergic reactions, hence the adhesive system
should be handled according to general industrial practice for handling and use of
chemicals.
Due to the fast setting performance and high flexibility with respect to methods
to improve the application process (humidity and temperature) the systems have
shown to be very promising for future production lines. In order to assess the
complex setting and curing process of new generation EPI adhesives an extensive
research and test program has been initiated by the companies Dynea, Rubner and
MPA University Stuttgart.
The low glue spread compared to condensation resins makes EPI adhesive sys-
tems price competitive or even a price leader.
References
1. Sakurada, S., Miyazaki, H., Hattori, T., Shiraishi, M., Inoue, T.: United States Patent
3,931,088 (1976)
2. Japanese Industrial Standards JIS K6806 (1985), revised (1995)
3. Deutsches Institut fr Bautechnik: General German Building Approval Z-9.1-634
(2005)
4. European Technical Approva; K-beams, ETA-04/0040 (2004)
5. Skretteberg, J.R.: Wiener Leimholz Symposium, Wien (March 23-24, 2006)
6. Grstad, K., Pedersen, A.: Journal of Adhesion Science and Technology 24,
13571381 (2010)
7. Norsk Treteknisk Institutt; 507037-LM01. Measurement of glueline thickness (2010)
8. Amtliche Materialprfungsanstalt Universitt Karlsruhe; Prfbericht nr. 096115 (2009)
9. Technische Universitt Mnchen, Report 10/09/KLE/04-rev, (2009)
10. Mrland, E., Bredesen, R., Pedersen, A.: PCT WO 2008/103055 (2008)
11. Mrland, E., Kaiser, M., Michael, H.: PCT WO 2008/056992 (2008)
Bonding of Various Wood Species Studies
about Their Applicability in Glued Laminated
Timber
Y. Jiang, J. Schaffrath, M. Knorz, and Stefan Winter
Lehrstuhl fr Holzbau und Baukonstruktion, Technische Universitt Mnchen, Germany

yuan.jiang@tum.de
Abstract. In this research project, the gluability of five soft- and hardwoods (ash,
beech, Douglas fir, larch and spruce) in combination with four adhesives (EPI,
MUF, PRF and PUR) are investigated. Factors such as extractive content, pH
value, wettability, etc., that may have significant influence on the bonding strength
and durability of adhesive joints, are analyzed. The results of the research project
achieved so far will be reported in this article.
Keywords: Gluability of softwood and hardwood, glued laminated timber, wood

extractive, wettability, tensile shear strength.
1 Introduction
Ongoing climate change has a severe impact on our forests. Hence, the forest
conversion in Germany is going to reduce the population of weak monocultures of
pine and spruce (Bayerische Staatsregierung 2007). They are replaced by
adaptable, climate tolerant and species-rich mixed forests, so that the percentage
of tree species, e.g. Douglas fir and beech, is distinctly growing
(Bundeswaldinventur 2009). Therefore, the supply of raw material from the forest
may change in the near future. To adapt to these changes, the woodworking
industry needs to adjust the production processes. Alternatively, advantages can be
taken by developing new and improved wooden products. In this context, the
bonding technology plays a key role. Bonding of various wood species with
higher strength and/or durability, will lead to benefits for wood engineered
products. However, experience shows that gluing of timber without sufficient
technical knowledge and elaborateness can lead to serious damages. Investigations
after the collapse of the Bad Reichenhall ice rink in 2006 showed that damages in
structural elements made of glued laminated timber (GLT) are often caused by
delaminated glue lines (Bla and Frese 2007, Dietsch and Winter 2009, Wolfrum
and Winter 2007).
DOI: 10.1007/978-94-007-7811-5_33, RILEM 2014
366 Y. Jiang et al.
Today, in Central Europe GLT is almost exclusively made of spruce or pine

(Mack 2006, Ohnesorge et al. 2009). However, in the recent past, research and
industry have shown increasing interest in using hardwoods such as beech. At
present, for the first time a building with GLT made of beech as structural
elements is being erected in Germany. In Switzerland, where building regulations
are less restrictive, some experience with GLT made of hardwoods such as ash
and beech has been gained in the last decades.
The research project Bonding of various wood species and studies about their
applicability in glued laminated timber focuses on the gluability of five soft- and
hardwoods (ash, beech, Douglas fir, larch and spruce as reference). It is assumed
that the selected wood species show increasing importance due to their growth
conditions, resistance or strength. The main objective of the ongoing research
work is to examine the performance and characteristics of the used wood species
in combination with the adhesive systems EPI, MUF, PRF and PUR. These
investigations shall give information about the applicability of these wood species
in glued laminated timber.
Chemical properties of wood, such as the extractive content and pH-value, play
an important role in adhesive bonding. Bikerman (1961) reports that the water-
soluble and volatile substances in the wood, such as the hydrophobic resin and
fatty acids, move towards the wood surfaces during drying and storage and form a
so-called chemical weak boundary layer. Highly acidic extractives can
accelerate or decelerate the pH-dependent curing reaction of some adhesive
systems (Scheikl and Dunky 1996). Zeppenfeld and Grunwald (2005) point out
that bonding failure caused by wood extractives are always to be expected when
their content (locally) rises above 5%. Therefore, the chemical properties of the
wood species used in this work were determined to provide a basis to increase the
understanding of the chemical interaction between wood and adhesives.
The attractive forces between adhesive and wood, such as Van der Waals
forces, dipole-dipole forces and hydrogen bonding, which are necessary for the
integrity of a glue line, act in the boundary layers with a thickness that varies from
nanometers to micrometers (between 0.1 and 1 nm). In order to provide the
optimal intermolecular attraction, the molecules of adhesive and wood should be
brought close enough to each other (Habenicht 2009). This requires a sufficient
wetting of the wood surfaces. The wettability of the selected wood species has
been quantified by means of contact angle measurements.
In addition to the determination of the chemical properties and wettability,
strength tests have been carried out to show the performance of different wood-
adhesive-combinations. Processing conditions are carefully chosen after
evaluating the results of the foregoing tests. Not only the breaking load and the
percentage of wood failure, but also the quality of the glue line, which was
examined under the microscope, are indicators for the reliability of each bonding.
As the research work is in progress, further results, regarding long-time tests for
example, are expected.
Bonding of Various Wood Species 367
2.1 Materials
Five soft- and hardwoods were selected for this study: ash, beech, Douglas fir,
larch and spruce as reference. Sawn timber of these wood species was chosen with
densities that more or less match the average value of the respective wood species
as given by DIN 68364: spruce (460 50) kg/m3, larch (600 50) kg/m3, Douglas
fir (580 50) kg/m3, beech (700 50) kg/m3 and ash (700 50) kg/m3 at
(12 1) % moisture content. All specimens were conditioned at 20 C and 65 %
RH (standard climate), so that the equilibrium moisture content of (12 1) % was
reached.
Four commercial adhesive systems for load-bearing timber structures were
selected:
- Aerodux 185 with Hardener HRP 155 (phenol-resorcinol-formaldehyde
adhesive (PRF))
- Prefere 6151 with hardener Prefere 6651 (emulsion-polymer-isocyanate
adhesive (EPI))
- Kauramin Leim 690 with hardener 1690 (melamine-urea-formaldehyde
adhesive (MUF))
- Jowapur 686.60 (polyurethane adhesive (PUR))
The development of modern adhesives is currently based on MUF-, EPI- and

PUR-systems. They are regularly optimized regarding curing speed and flexibility in
processing. The PRF adhesives have nowadays only a small market share, especially
due to their dark color. With regard to the existing long-time experience, the
combination of PRF adhesive with spruce is useful as a reference system.
2.2 Chemical Analysis
2.2.1 Sample Preparation

The chemical properties of the wood can highly vary within the same species.
Therefore, instead of using reference values, the chemical properties of the
timbers, which are used and evaluated in this study, were precisely measured. To
obtain reliable results, samples were taken from five different boards of each
wood species and cut into small wood cubes ( 1.5 cm3).
2.2.2 Wood Extractives Analysis

The wood extractive content was determined by means of cold water and organic
solvent extraction. For the extraction process wood flour was prepared from small
wood cubes. The cold water extraction was performed at room temperature.
368 Y. Jiang et aal.
Air-dried wood flour wass mixed with deionized water. The extraction time periood
was 48 hours. Afterwardss, the pecentages of wood extractives that were soluble iin
cold water were determined. The organic solvent extraction was carried out at
approximately the boilingg temperature of the solvents using a Soxhlet extractioon
apparatus. Three differentt organic solvents were used successively as extractantts:
petroleum ether, acetone and methanol. The percentage of wood extractives thhat
were soluble in organic so
olvents was determined.
The mixtures of woo od extractives, which have been obtained though thhe
extraction process, were analyzed. The components of the extractive solutionns
were separated and charracterized using chromatographic separation techniquees
such as thin layer chromaatography (TLC), gas chromatography/mass spectrometrry
(GC/MS) and ion exchang ge chromatography (IC).
2.2.3 Chemical Chara

acterization of Wood Surfaces by Means of
pH-Measuremeents
The influence of the surfaace aging on the chemical properties of wood surface waas
quantified by means of surface pH measurement. Therefore, from each woood
species ten wood cubess were taken and subsequently three veneers with a
thickness of 1 mm were prepared using a microtome (see Figure 1, left). Thhe
surface pH was measured d directly after the veneers were cut off, after 24 houurs
and after 7days storage att standard climate, respectively. The measurements werre
taken at room temperature. The thin veneer was placed in the center of a
specifically manufacturedd Teflon container and fixed with the lid of the containeer
(see Figure 1, right). Thee Teflon lid has a round opening in the center, througgh
which the thin veneer is visible
v from above. Pure water with a pH of 7 was applieed
on the wood veneer. A pH p flat membrane electrode (InLab surface electrodde,
Mettler Toledo GmbH) wasw used to measure the surface pH.
Fig. 1 Preparation of the veneers

v (left) and measuring device with pH flat membranne
electrode (right)
2.3 Contact Angle Measurement

Five specimens of each wood species with a thickness of 5 mm and with a radial
surface area of 40 mm 100 mm were prepared. Contact angles were measured
directly after planing as well as after 24 hours and 7 days conditioning at standard
climate, respectively. In addition, some specimens were slightly sanded prior to
measurement using sandpaper with grit size 100. During the entire process the
surfaces of the specimens were carefully kept free from contamination. The
contact angles were determined with the measuring system EasyDrop (KRSS
GmbH, Germany) with the so-called dynamic method.
2.4 Longitudinal Tensile Shear Tests

The performance of each adhesive system in combination with the selected wood
species was evaluated by means of the tensile shear tests according to
DIN EN 302-1. Specimens were prepared for each combination of wood species
and adhesive system from two bonded boards with a thickness of 2 5 mm.
Processing parameters were carefully chosen after consulting the adhesive
manufacturers. After curing, ten specimens with a length of 150 mm, a width of
20 mm and a defined shear plane of 200 mm2 were cut from the bonded boards.
The specimens were conditioned at 20C/65% RH for 7 days. Prior to the tensile
shear tests, the specimens were subjected to one of the following treatments
according to EN 302-1: A1 (without further conditioning), A2 (4 days in cold
water), and A4 (6 hours in boiling and 2 hours in cold water). After the shear test
the glue line quality of different wood-adhesive-combinations was evaluated by
means of microscopy. Sections of 20 m thickness were prepared from the end
face of the tested shear specimens.
3.1 Wood Extractives Analysis

The extractive contents of the five different wood species are summarized in
Figure 2.
A comparatively high extractive content of approximately 10% was found for
larch when using cold water as extractant. That is because larch contains a lot of
sugars, which are highly soluble in water. The extractive content of the ash is
slightly above 5%. Spruce and beech, on the other hand, contain only small
amounts of wood extractives (< 2%).
An overview of the main extractives of the five analyzed wood species can be
found in Table 1. The extractives are subdivided into the following substance
classes: fatty acids, terpenes/steroids, sugar/sugar derivatives and phenols/tannins.
Extractives such as linoleic acid, sitosterol, arabinose and taxifolin are found in
most of the examined wood species.
Fig. 2 Extractive contents of

o spruce, larch, Douglas fir, beech and ash, determined wiith
organic solvents and cold waater as solvent
Table 1 Overview of the exttractives of the analyzed wood species
sugar/
wood terpenes/ phenols/
fatty acids sugar
species steroids tannins
derivatives
dehydroabietic acid
spruce oleic acid arabinose hydroxymatairesinoll
sitosterol
linoleic acid isopimaric acid arabinose dihydrokaempferol
larch
oleic acid dehydroabietic acid galactose taxifolin
isopimaric acid
dihydrokaempferol
douglas fir linoleic acid dehydroabietic acid arabinose
taxifolin
sitosterol
fructose
beech - Sitosterol -
glucose
linoleic acid glucose
stigmasterol
ash palmitic D-mannitol -
sitosterol
acid saccharose
3.2 Wood Surface pH

p
The lowest pH value is sh
hown by Douglas fir, which varies between 3.25 and 4.225
(Figure 3). The surface pH of the two hardwood species are the highest and varry
between 4.75 and 5.50. The difference between the highest and lowest pH value is
assumed to affect the pH H-dependent curing reaction of adhesive systems (e.g.
MUF, PRF).
An influence of surface aging on the pH value can be stated for all five wood
species. As shown in Figure 3, the surface pH of the majority of the wood species
increase slightly after 24 hours storage at 20C/65% RH. After 7 days, the surface
pH of spruce, larch, Douglas fir and ash rise again. Only the pH value of beech
shows a slight reduction.
Fig. 3 Surface pH of the fresh and aged wood surfaces
3.3 Contact Angle

In general, the low contact angles (< 90) indicate a good wettability (Dunky and
Niemz 2002) of all five wood species (see Figure 4). Compared to the planed
wood surfaces, the surfaces, which are slightly sanded, can be more easily wetted
by water (Figure 4, left). Regardless of which surface preparation is used the
lowest wettability is shown by Douglas fir.
Fig. 4 Influences of surface preparation (left) and aging (right) on the wettability of wood
surfaces
It is clear from the daata presented in Figure 4 (right) that the surface aginng
affects the wettability of the
t wood surface. Only after 24 hours an increase of thhe
contact angle can be seen n for almost all wood species, except spruce. After 7 dayys
the wettability of spruce has
h been reduced most though.
3.4 Tensile Shear Strength

S
The bond strengths off different wood-adhesive-combinations, which werre
determined by longitudin nal tensile shear tests, are displayed in Figure 5. Thhe
minimum requirements on the average longitudinal tensile shear strength oof
adhesive type I and glue line thickness of 0.1 mm are given in DIN EN 301 witth
10 N/mm2 after treatment A1 and 6.0 N/mm2 after treatments A2 and A44,
respectively. However, thhese requirements can only be applied to specimens madde
of beech. As a further reeference value, the tensile shear strength of solid woood
samples with dimensions as specified in DIN EN 302-1 was determined.
Low bond strength was found for the specimens glued with PUR adhesivve.
Nevertheless, the tensile shear strength, which is achieved by the combination oof
PUR adhesive and larch after the treatments A2 with cold water and A4 witth
boiling and cold water, ex
xceed the expectation of the adhesive manufacturer.
Fig. 5 Longitudinal tensile sh

hear strength of different wood-adhesive-combinations
Figure 6 (left) shows the electron micrograph of beech specimen glued witth
MUF adhesive. It is visib ble that the adhesive has penetrated only into the upper
wood surface. The lower surface is only partially wetted. Besides that, considerinng
the amount of adhesive applied
a on the wood surface, the glue line seems to bbe
very thin. After consultatiion with the adhesive manufacturer, the test series beecch
specimen glued with MUF M adhesive was repeated with changed process
parameters (increased ap pplication amount and shortened open assembly timee).
Doing so, a significant in ncrease of the tensile shear strengths was achieved (seee
Figure 6, right).
Bonding of Various Wood Species
S 3773
Fig. 6 Micrograph of beech specimens

s produced using MUF adhesive (left) and comparisoon
between the shear strengths achieved
a by the specimens of first and repeated test (right)
4 Conclusions an
nd Outlook
In this research project, differences regarding the gluability of various woood
species are studied. Dettailed information about the chemical extractives annd
surface pH is presented. The results of contact angle measurements show thhe
wettability of surfaces witth regard to age, preparation and wood species. The bonnd
strengths of different woood-adhesive-combinations are evaluated by means of thhe
tensile shear tests.
The influence of the above
a mentioned characteristics on the curing behavioor
and bonding strength will be further investigated. In addition, various processinng
conditions on selected combinations of wood species and adhesive system will bbe
analyzed. Hereby, delamiination tests, block shear tests and microscopic analyssis
are carried out on small test beams. With the knowledge of the best performinng
processing conditions larrger test beams will be subjected to long-term testinng
under load and various cllimate conditions. The test setup will lead to informatioon
about durability and crreep behavior. Finally, the residual strength will bbe
compared through a four-point bending test.
References
Bayerische Staatsregierung, Klimaprogramm Bayern 2020, Minderung von Treibhausgaseen,
Anpassung an den Klimawandel,
K Forschung und Entwicklung. 35 S. (20077),
http://www.stmugv v.bayern.de/umwelt/klimaschutz/klimaprogra am
m/doc/klimaprogra amm2020.pdf (February 22, 2010)
Bikerman, J.J.: The Science of Adhesive Joints. Academic Press, New York (1961)
Bla, H.J., Frese, M.: Schadensanalyse, Schadensursachen und Bewertung dder
Standsicherheit bestehen nder Holzkonstruktionen, Forschungsbericht der Universittt
Karlsruhe, Lehrstuhl fr Ingenieurholzbau
I und Baukonstruktionen (2007)
Bundeswaldinventur 2 (2009 9), http://www.bundeswaldinventur.de (aufgerufeen
am June 10, 2009)
374 Y. Jiang et al.
Dietsch, P., Winter, S.: Assessment of the Structural Reliability of all wide span Timber
Structures under the Responsibility of the City of Munich. In: 33rd IABSE Symposium
Proceedings, Bangkok, Thailand, September 9-11 (2009)
DIN 68364: 2003-05, Kennwerte von Holzarten - Rohdichte, Elastizittsmodul und
Festigkeiten
DIN EN 301:2006-09, Klebstoffe fr tragende Holzbauteile, Phenoplaste und Aminoplaste
Klassifizierung und Leistungsanforderungen; Deutsche Fassung EN 301:2006
DIN EN 302-1:2004-10, Klebstoffe fr tragende Holzbauteile Prfverfahren Teil 1:
Bestimmung der Lngszugscherfestigkeit; Deutsche Fassung EN 302-1:2004
Habenicht, G.: Kleben Grundlagen, Technologien, Anwendungen. Springer, Heidelberg
(2009)
Mack, H.: Der europische Markt fr Brettschichtholz (BSH). In: Wiener Leimholz
Symposium (2006)
Ohnesorge, D., Henning, M., Becker, G.: Bedeutung von Laubholz bei der
Brettschichtholzherstellung. Holztechnologie 50, 4749 (2009)
Scheikl, M., Dunky, M.: Holzforschung Holzverwertung, 48, 7881 (1996)
Dunky, M., Niemz, P.: Holzwerkstoffe und Leime. Springer, Heidelberg (2002)
Wolfrum, A., Winter, S.: Evaluierung geschdigter Hallentragwerke aus Holz.
Ergebnisbericht fr Holzabsatzfonds - Absatzfrderungsfonds der deutschen Forst- und
Holzwirtschaft, Stand Mai 2006/Juli 2007, Unverffentlicht (2007)
Zeppenfeld, G., Grunwald, D.: Klebstoffe in der Holz- und Mbelindustrie. DRW-Verlag,
Leinfelden-Echterdingen (2005)
Fatigue Performance of Adhesive Connections
for Wooden Wind Towers
Leander Bathon, Oliver Bletz-Mhldorfer, Jens Schmidt, and Friedemann Diehl
Hochschule RheinMain, University of Applied Sciences Wiesbaden Rsselsheim

Geisenheim, Kurt-Schumacher-Ring 18, 65197 Wiesbaden, Germany
bathon@t-online.de,
{oliver.bletz,jens.schmidt,friedemann.diehl}@hs-rm.de
Abstract. Recent work at the University of Applied Sciences in Wiesbaden /

Germany has indicated that wood-steel-connections using adhesive action show a
stiff performance, a high ultimate load capacity, a ductile behaviour and at the
same time a predicable fatigue performance. This paper shows test results for
static as well as dynamic loading conditions. By using adhesive connections in
combination with engineered wooden products it is now possible to build a
wooden wind tower that provides clean energy.
Keywords: timber, wood, steel, glued connections, fatigue test, ductile, two
component adhesive system.
1 Introduction
Recent work at the University of Applied Sciences in Wiesbaden / Germany has

indicated that wood-steel-connections using adhesive action show a stiff
performance, a high ultimate load capacity, a ductile behaviour and at the same
time a predicable fatigue performance. The connection is designed to produce
steel failure in the range of ultimate load under static as well as dynamic loading
conditions. In that respect the ultimate performance of the connection depends on
the plastic performance of the steel component in this innovative coupling system.
The basic idea behind this approach is to design in steel and at the same time build
with timber.
The innovative connection consists of a two component adhesive system along
with a standardized steel connection geometry. The adhesive was designed to be
used either on site or in the shop. The fatigue performance of the adhesive was
tested up to 10.000.000 load cycles. All testing was done at the Material Testing
Laboratories in Wiesbaden, Germany. The paper will show static and fatigue
testing under tension and compression loading conditions.
DOI: 10.1007/978-94-007-7811-5_34, RILEM 2014
376 L. Bathon et al.
2 Tension Test
Fig. 1 shows tension test after it has reached the ultimate load capacity. It shows
that the perforated steel plate is interlocked within the timber cross sections due to
the adhesive dowels (AD). Given an adequate AD design the steel is the weakest
part within the failure mechanism and therefore determines the ultimate load
capacity.
Fig. 1 Tension test
Fig. 2 shows the load displacement curve of another test specimen with its
plastic capacity due to the steel failure.
steel failure
Plastic performance
yield
Fig. 2 Load displacement curve [1]
The research shows that adhesive connections can be designed based on the
load capacity and the stiffness of one AD (adhesive dowel). The authors suggest a
char. load capacity of 1.200 N and a slip modulus kser of 7.000 N/mm per dowel
independent of the grain of the timber and number of ADs.
3 Fatigue Test
The following Fig. 3 shows a ramp-loading tension fatigue test. Fig. 4 shows the
whler-performance of the adhesive connection up to 10.000.000 load cycles.
Fatigue Performance of Adhesive Connections for Wooden Wind Towers 377
Fig. 3 Ramp-loading tension fatigue test [2]
The fatigue testing was performed on small (3 AD per specimen) and big (300
AD per specimen) tension test specimens.
Figure 4 shows the results of the fatigue tests in the hsk-connector. The hsk-
connectors were tested under a sinus type tension ramp-load and a frequency of 3
to 10 Hz. The load was applied at constant strain ratios R = min / max.
Fig. 4 Fatigue performance (whler)
The ultimate tension capacity in the short-time tests is based on the steel
capacity used. The ultimate capacity on the fatigue performance of the connection
system was approx. 100 N/mm2, based on the steel net cross section.
The blue line in Fig. 4 shows a fatigue capacity of a grade S235 steel of this
adhesive connection system based on the fatigue factors of a=5 and b=2 given a
stress ratio R = -0,5. The green line shows the same performance under R = 0,1.
Using EC 5 one can now conduct a fatigue design on this innovative connection
system.
4 Analysis
Based on the EC adhesive connections have to be design for steel failure, wood
failure and adhesive failure. The connection system introduced is designed in the
same manner.
However the adhesive capacity is based on the adhesive capacity of the
adhesive dowel (AD). Therefore the code approval proposal shows capacities
based on one AD (Fig. 5) for strength and stiffness.
RKD,K and kser
RKD,K = 1.200 N
Kser = 7.000 N/mm
per AD (adhesive dowel)
Fig. 5 Design model
The design approach based on the code approval proposal therefore shows the
following design criteria:
- design of the timber capacity
- design of the steel capacity
- design of the AD capacity
In order to provide a robust timber structure it is desirable to design based on
the steel capacity. In that case the basic idea behind this connection approach is to
design in steel (weakest part within the design approach) and at the same time
build with timber.
The fatigue design is shown in the EC 5. In order to perform the fatigue design
you need the factor a and b (Fig. 6) for the connector in use. The design proposal
of the hsk-connector shows a=6 and b=2.
(1)
(2)
(3)
(4)
Fig. 6 Fatigue design (Eq. 1,2,3,4) of EC 5
Fatigue Performance of Adhesive Connections for Wooden Wind Towers 379
5 Conclusions
The research shows that adhesive connections allow new applications in timber
design. They deliver the tool that allows timber structures (Fig. 7) to be used in a
field of applications only known to steel and concrete structures so far.
Fig. 7 FlyingStairs of ESB at UBC
Fig. 8 Timber tower in Hannover

Fig. 7 shows the flying stairs of Vancouver which provide the only support of
the stairs at the individual floors. Therefore the stairs cantilever out between the
individual floors.
Fig. 8 shows a wooden tower of a total height of 100 m. By using adhesive
connections in combination with engineered wooden products it is now possible to
build a wooden wind tower that provides clean energy.
References
[1] Bathon, L., Bletz-Mhldorfer, O., Schmidt, J.: Weil: Windenergietrme aus Holz. Die
neue Quadriga, 3236 (March 2010)
[2] Bathon, L., Bletz-Mhldorfer, Schmidt, J., Weil, M.: Der Strom kommt aus dem
Holzturm. Bauen mit Holz, 3640 (November 2010)
Multifunctional Wood-Adhesives for Structural
Health Monitoring Purposes
Christoph Winkler and Ulrich Schwarz
Eberswalde University of Sustainable Development, University of Applied Sciences,

Faculty of Science and Technology, Eberswalde, Germany
cwinkler@hnee.de
Keywords: structural health monitoring (SHM), conductive filled adhesive, elec-

trical modified wood adhesives, influence of moisture, local heating.
1 Introduction
1.1 Structural Health Monitoring in Timber Engineering

Structural Health Monitoring (SHM) is a concept, which uses integrated sensors to
detect the condition of engineered structures. The sensors can be considered as the
nervous system of the structure, which is designed to measure the various loads
and damages to the structure during its lifetime. By analyzing the information
about loads (wind, snow etc.) and damages (delamination, micro cracks etc.), the
general health of the system can be monitored and a prediction of the lifetime
becomes possible. While the development of SHM roots in applications of high
safety demand (air space and power plant engineering), the decreasing costs of
sensors and microelectronic devices supports its utilization in civil engineering.
Different sensor technologies were developed or extended to fit the needs of
SHM, for example fiber optical sensors, piezo-electric vibration sensors, strain
sensors and corrosion sensors [1, 2, and 3]. While most sensors are attached to the
surface or show the viewable deformations by camera-systems, some materials
allows a positioning of the sensors inside (i.e. optical fibers in concrete) or getting
sensual functionality by modification (i.e. strain sensing concrete by mixing it up
with conductive fibers). In wood engineering, the use of SHM techniques is at the
beginning. The basics can be found in various non-destructive measurement tech-
niques for timber structures, an overview over these is given by Kasal and Tannert
[4].
Initiated by some failures of timbers structures in the year 2006, the need for
SHM of timber buildings is rising [5]. A monitoring of moisture content (MC) was
done by regular MC equipment, inside block-glued laminated elements [6], while
Annamdas uses extern piezo-ceramic sensors to detect the biological degradation
[7] and the stiffness of wooden structures [8]. Another possibility was researched
DOI: 10.1007/978-94-007-7811-5_35, RILEM 2014
382 C. Winkler and U. Schwarz
by Heiduschke, who tried to use the electrical properties of carbon fiber sheets to
measure the deformation of glue-laminated timber [9].
Evaluating the cause of failure in glue-laminated timber [10, 11], the monitor-
ing of deformation, damage growth in the adhesive joint, wood moisture and joint
temperature is necessary to identify the health state of the construction element.
1.2 Multifunctional Polymers

The approach is to modify an existing certified adhesive for glue-laminated timber
with conductive filler to achieve multifunctional properties.
Conductive polymers using fillers with a low resistivity are common in anti-
static applications. Other research activities found that filled conductive polymers
can exhibit sensorial qualities. Equipped with an electrical conductive percolation
network, rubber can be used as tactile sensorial material, which is i.e. used for
contact sensor arrays in robotics [12]. Carbon fibers, embedded in glass-fiber
composites (CFK), are usable for detecting the damage accumulation from micro
cracks by measuring the electrical resistance [13]. The capability of conductive
modified polymers and fiber reinforced polymers (FRP) for SHM techniques by
measuring strain and damage growth by resistance has been researched by several
other authors [14, 15, 16]. With increasing content of electrical conductive fillers,
the resistivity of the composite decreases, while the relationship between filler and
adhesive is strongly non-linear. The filler particles are dispersed in the adhesive
matrix, approaching each other and forming a filler network at a composite-
depending threshold. This so called percolation threshold [17] causes a change to
both the electrical and physical properties. The increase in conductivity above or
below the percolation threshold is significantly less than in the area, where the
network is created. The conductivity between the filler particles is formed by the
approach of a few nanometers, while the small distance can be bridged by quan-
tum tunneling. Together with both phenomena, the theory behind the conductivity
of filled polymers is called the tunnel-percolation model (TPM) [18, 19].
Some years ago, these functional fillers where used to enhance the electrical
properties at strength expense [20]. Then it was found that homogeneity of the
dispersion, size, aspect ratio and agglomerates of filler particles as well as the
adhesion of the fillers to the polymer matrix are the main influences on the
strength of filled polymers. Currently, the filler contents can be decreased to 0.3
wt% while reaching the percolation threshold and functionalization of the fillers
increases their bonds with the plastic/adhesive. Utilization of functionalized car-
bon nanotubes as fillers in epoxy resin can increase the shear strength and fracture
toughness of the composite [21, 22].
1.3 Wood as Adherent

To use a conductive adhesive for monitoring purposes, the bonded adherent needs
to be an insulating material or, at least, it should have a much higher resistivity as
Multifunctional Wood-Adhesives for Structural Health Monitoring Purposes 383
the adhesive. Wood shows these properties partly as it is an insulating material in

dry condition, but in a wet state it is a good conductor. In the hygroscopic range
the conductivity increases linearly. After reaching the fiber saturation point
(around 28%) the impact is considerably less and ends in the range of tap water
[23]. Secondary influences on the resistance of wood next to the moisture are den-
sity, grain direction, temperature and the wood species. In fiber direction the
conductivity is twice as high as measured transversely to the fiber direction. The
influence of the wood species is due to different densities and ingredients, so each
type of wood has its own conductivity or resistivity curve. The resistance meas-
urement is used primarily for wood moisture measurement at constant tempera-
ture, since this is the dominating influence.
As the decay of wood by fungi usually needs high moisture content of around
18%, all healthy wood constructions have to be in a dry state or at least should be
drying fast. Accordingly, the resistivity of a healthy timber frame should be much
higher than the resistivity of the adhesive. These circumstances suggest the use of
wood adhesives for sensual purposes. The majority of structural timber is glued
together from lamella or blocks, thus the adhesive is attached in the production
process, anyway. Accordingly, only an additional contacting and connection to a
data logger is necessary.
Following this idea, electrical modified wood-adhesives were produced and
some orientation tests were made to prove their usability for SHM purposes.
2 Experimental Samples and Methods
2.1 Deformation
The influence of deformation was investigated be using a modified testing ma-
chine inside a climate chamber, where all fixtures isolates the samples electrically
from the testing machine. In this configuration, all tests had been done at a
constant climate of 293.15 K and 65% RH. To evaluate the effect of characteristic
deformations on the adhesive joint, 3-point bending, pressure and tension shear
tests have been performed. Deformation was measured with a camera system for
3D digital image correlation (GOM, Software ARAMIS).
The used modified adhesive was a mixture of 95wt% one-component moisture-
curing polyurethane adhesives (1C PUR, JOWAT 686.60) and 5wt% of carbon
black (Ketjenblack EC-300J). Dispersion and homogenization was done by calen-
daring the premixed adhesive and filler three times with a clearance of 30m.
Electrical contact to the samples was made by copper-sheet-electrodes (thickness
= 0.1 mm), which were bonded inside the adhesive joint. 2-wire measurements
were taken by a digital multimeter (NI PXI-4071); the electrodes were contacted
with alligator clips.
2.2 Humidity
A first test setup was used to measure the influence of humidity on the resistance
of wood, bonded by the modified adhesive. A test stand was used to investigate 4
different settings parallel, the influence of following variables was as small as
possible:
production variation of the adhesive (mass ratio, dispergation and homogeniza-
tion)
time dependent behavior of the adhesive (creep, aging)
Each setting is a combination of a sample design and a climate condition. The
objective of the different settings was an elimination of influencing variables like
wood moisture. The general design of the samples is shown in Figure 1; the varia-
tions of the settings are listed in Table 1.
All samples were produced at the same time under identical climate conditions.
The adherent surface of the wooden samples were planed around 1 hour before
bonding, while the PMMA samples were treated by atmospheric-pressure plasma
Fig. 1 General sample dimensions and materials
Table 1 Settings, sample variations and climate conditions
Setting Sample variation Climate condition

A adherent: spruce wood (thickness x = 10mm) T = 293.15 K
adhesive U and V = modified 1C PUR RH = 25 - 95%
B adherent: spruce wood (thickness x = 10mm) T = 293.15 K
adhesive U = modified 1C PUR RH = 25 - 95%
adhesive V = unmodified 1C PUR
C adherent: PMMA (thickness x = 3mm) T = 293.15 K
adhesive U and V = modified 1C PUR RH = 25 - 95%
D adherent: PMMA (thickness x = 3mm) T = 293.15 K
adhesive U and V = modified 1C PUR RH = 17%
to get a sufficient adhesion of the PMMA surface. Copper film (thickness = 0.1
mm) with a one-sided adhesive film was bonded to one adherent ahead of the
bonding. The modified adhesive was a mixture of a one-component moisture-
curing polyurethane adhesive (1C PUR, JOWAT 686.60) with 3wt% of carbon
black (Ketjenblack EC-600JD). The dispersion was made by calendaring the
mixed materials two times (EXAKT 30). After manual application of the adhe-
sive, the samples were pressed according to the data sheet for structural applica-
tions (2 hours at 303.15 K and 1 MPa). All samples were arranged in a climate
chamber, electrically isolated to metal parts, and contacted with alligator clips.
The samples of setting D were placed in a desiccator to get a dry state (the humidi-
ty was around 17% after a short adjustment phase), but the same temperature as all
other samples. The 2-wire resistance measurements were taken by a multiplexed
digital multimeter (NI PXI-4071 with NI PXI-2527). In this configuration, 8 sam-
ples of each setting could be tested parallel. Parallel to the 32 adhesive bonded
samples, the wood moisture of two wooden samples was measured by common
needle electrodes (Ahlborn). The moisture measurements were taken along the
grain like the resistance measurements through the modified adhesive. Figure 2
shows the test set-up.
Fig. 2 Test set-up influence of humidity (A D = variation A D; E = wood moisture

meter)
3.1 Deformation
The result of one 3-point bending test is shown in Figure 3. The three graphs rep-
resent the relative change of the essential values: induced bending stress and the 2-
point resistance measurement of the two adhesive joints of the bending sample. It
is shown that deformation and both resistance measurements show the same
386 C. Winkler and U. Schwaarz
behavior in dependency of time. While unloading, the resistance shows a smaall

hysteresis. Another clear visible result is the different trend of the resistance iin
dependency of the joint position.
p While measurement of the adhesive joint abovve
the neutral phase shows a falling resistance trend, the resistance in the joint undeer
the neutral phase is rising together with the deformation.
Fig. 3 3-Point bending: relattive change of detection and resistance in the adhesive joints
Regarding the concept of stress in bending samples the resistance change trennd
seems reasonable. While in the upper joint the compression stress causes a shorrt-
ening and perhaps a thick kening of the adhesive layer, the measured resistance is
decreasing with increasin ng deformation. On the other side, the joint below thhe
neutral phase is stressed in tension and so far an elongation and thinning of thhe
adhesive layer could be taaken as reason for the increasing resistance with increaas-
ing deformation (see Figu ure 4).
These assumptions arre supported by the shear test results, which show aan
increasing resistance togeether with the stress (see Figure 5). While being sheared,
the adhesive layer will bee stretched and thinned by the deformation and therefoor
the resistance is increasing
g.
The distinction of com mpression and tension by the resistance trend was reproo-
ducible for all samples. Although
A all tests with carbon black filled 1C PUR show
wa
correlation between deforrmation and ohmic resistance, two phenomena prevent a
reproduction of the measu ured resistance.
Multifunctional Wood-Adheesives for Structural Health Monitoring Purposes 3887
Fig. 4 Thickening (a) and shhortening of the upper joint in bending. Thinning (b) and elon-
gation of the joint below thhe neutral phase. With conductive adhesive, the resistance A
decreases, while resistance B increases.
The first is a hysteresiis of the resistance (see Figure 3) and a time dependinng
behavior of the resistance change in respect of the deformation (see Figure 5).
Fig. 5 Shear displacement of

o both adherents and relative resistance change in respect of
time
The second is the chan

nge of the absolute resistance range after the applicatioon
of loads. Figure 6 shows the resistance graphs of 17 shear tests, all with the sam
me
sample. The number of thhe graph is also the order of the tests. Although all tessts
show the same trend and d a relation of the resistance with the displacement, thhe
measurement range of thee resistance is increasing with the load cycles.
Fig. 6 Shear test: change of resistance

r range in dependency of load cycles
This lack of reproducib ble measurements has long been known and has been de-
scribed in the summaries of tactile sensors in the 80's [24, 25], by measuremennts
of Steinfeld and Kalkner [26] as well as of Bokobza [27]. The observed effeccts
can be divided into plasstic (permanent conductivity changes) and viscoelasttic
effects (creep, hysteresis)), while he worst reproducibility was observed in elastoo-
mers. Considering that th he used PUR adhesive is partly elastic, then some of thhe
explanatory models can probably
p also be used for this behavior. Bierbaum [122]
explains the change in thee resistivity and the non-linearity with a plastic change iin
structure, which is explainned by electrical forces (attraction forces by polarizationn)
and thermal stress on thee electrodes (thermal strain at the contact points). Thhis
explanation is supported by b the research of Thostenson and Chou, who show thhat
measurements are reprod ducible as long as the load remains below the limit iin
which micro cracks occurr [16]. Another explanation for this change is provided bby
the Mullins effect. This describes a stress softening of elastomers by repeateed
voltage stress [28]. The stress
s softening occurs in both unmodified and in carboon
black filled elastomers. Mullins'
M studies ascribes the stress softening of filled elaas-
tomers at very small defo ormations (0.1%) to the breaking of filler agglomeratees
into individual particles and the destruction of polymer-filler compounds. H He
observes also, that at low
w loads, a steady resistivity state occurs after a few loaad
cycles. Mullins therefore recommends conditioning cycles before a correspondinng
measurement. These obseervations correspond to the behavior of the repeated shear
tests. The creep behavior of the electrical resistance and the elongation of carbon
black filled silicone were investigated by Wang and Ding [29]. They show that the
creep of resistance and strain has the same tendency, only the time to regain the
initial value is twice as long for resistivity as for the strain. They also show that in
the case of filled elastomers the increase of modulus of elasticity (MOE) can miti-
gate the creep.
3.2 Humidity
Based on a simple model of two adherents bonded by a conductive adhesive joint,
which is contacted by electrodes the measured resistance is a parallel circuit of 3
resistors (see Figure 7). In case of dry wood as adherent, the resistors R' and R''
have a small influence on the total resistance. The higher the humidity, the higher
is the moisture content of the wood and therefore the influence on the total
resistance.
Fig. 7 Parallel circuit of the resistors R, R' und R'' by contacting the electrical conductive
adhesive joint with the resistance R
Based on these assumptions, a change of humidity should effect the resistance

of the different sample settings (hereafter simply called variant A, B, C or D) in
different ways. Since PMMA can be considered as insulating material and the
position of the samples in the desiccator keeps away the humidity changes, the
resistance of variant D should show only small changes. Possible changes can
occur due to time depending effects on the adhesive joint like creeping from dead
load or local heating at the contact from the measurement current.
Variant C should only be affected by the water absorptive capacity of the modi-
fied adhesive and the polymer adhesive, which is around 2wt% for PMMA [30].
The configuration of variant B forces the conductive path through the wooden
adherents, while variant A is subjected to all influences from moisture changes in
the adhesive and the wooden adherents.
Figure 8 and 9 show the results of the forced humidity change in the climate
chamber. The graphs display the average of 8 samples for each group, while the
wood moisture values are from one moisture meter (the variation between both
was within the measurement accuracy).
At a first glance the meeasurements seems to be in compliance with the assumpp-

tions above. The relative change in resistance of variant A, C and D shows a teen
times larger change of ressistance in the glued wooden samples (variant A) than iin
the PMMA samples, wh hich can be ascribed to the higher moisture sorption oof
wood in contrast to PMM MA. While the resistivity of variant D is decreasing to a
ying in the desiccator, the resistivity of variant C shows a
stable level due to the dry
slightly dependence on thhe humidity change.
Fig. 8 Humidity and relativee change of resistance of variant A, C and D in respect of time
Looking at Figure 9 onne sees that the resistance of variant B can't be measured,
until the moisture contennt reaches a value of around 8%. At this moment the ree-
sistance is decreasing masssively due to the increasing wood moisture. Accordinglly
it can be assumed that suiitable electrodes with less surface corrosion then coppeer,
which are bonded in the adhesive
a layer and separated from each other by an insuu-
lating adhesive in the saame joint, can be used to measure the wood moisturre.
Given that the adhesive application
a can be performed in pattern of insulating annd
conductive adhesive a nettwork of measuring electrodes can easily be used to monn-
itor the wood moisture neear the adhesive joint.
However, the positive trend of resistance with increasing humidity seems to bbe
confusing (see Figure 8) as it disagrees with the mentioned model. Usually thhe
resistance of variant A should also decrease like variant B with increasing humiddi-
ty as the wooden adheren nts are more conductive due to the higher moisture conn-
tent. A look at the absoluute resistances of variant A and B in Figure 10 revealls,
Multifunctional Wood-Adheesives for Structural Health Monitoring Purposes 3991
that the wood resistance (variant B) at its minimum is around 2 decimal poweers
away from the resistance of the adhesive joint in variant A. Accordingly its influu-
ence on the total resistancce is minimal.
Fig. 9 Humidity change, reacction of wood moisture and resistance of the variant B samples
Fig. 10 Humidity and absolu

ute resistance of variant A and B in respect of time
In this case the questio

on remains, which effect of the wood causes the relativve
resistance change, which is ten times larger than without wood as adherent.
A reasonable cause of this effect is the swelling of wood with increasing moiis-
ture content. From the defformation part it is known, that a deformation of the joinnt
exhibits a resistance change. Now, if the wood is swelling due to moisture content
change, the adhesive joint will be stressed in both plane directions and therefore
the resistance will decrease.
3.3 Summary
Starting with an introduction to structural health monitoring (SHM) it has been
shown, how conductive filled adhesives probably could be used as In-Situ non-
destructive monitoring sensors. An overview is given, how conductive fillers
generate multifunctional properties in polymers, so they can be used to detect
deformation and damage accumulation in the component. Since the structural
health of wood depends on its moisture content and the resistivity of wood chang-
es with moisture content, conductive adhesives seem to be promising for structural
health monitoring of wood.
It was shown that:
a contacted adhesive joint between two wooden adherents exhibits a defor-
mation dependent resistance, which shows a positive trend in tension and a
negative in compression,
the correlation between deformation and resistance is linear with a time de-
pendent proportion, which can be divided into a plastic and elastic amount,
this plastic part is load cycle dependent and can therefore probably be reduced
with conditioning cycles before a corresponding measurement,
a contacted adhesive joint between two wooden adherents can be used to meas-
ure the wood moisture content, if the bonded electrodes are separated by an in-
sulating adhesive in the same plane.
Although the general practicability of conductive wood-adhesives for monitoring
purposes of deformation and critical moisture content seems to be answered, many
questions and further investigation fields are opened. Especially the following
fields need more attention to evaluate the different influences in more detail:
developing of a stable filler-adhesive dispersion, research in a more reliable con-
tacting and a repeatable adhesive application technology, which can apply two
different adhesives in various patterns.
References
[1] Majumder, M., Gangopadhyay, T.K., Chakraborty, A.K., Dasgupta, K., Bhattacharya,
D.K.: Fibre bragg gratings in structural health monitoring - present status and applica-
tions. Sensors and Actuators A: Physical 147, 150164 (2008)
[2] Ciang, C.C., Lee, J.-R., Bang, H.-J.: Structural health monitoring for a wind turbine
system: a review of damage detection methods. Measurement Science and Technolo-
gy 19, 120 (2008)
[3] Ou, J., Li, H.: Structural health monitoring in mainland china: Review and future
trends. Structural Health Monitoring 9, 219231 (2010)
[4] Kasal, B., Tannert, T.: In Situ Assessment of Structural Timber. Springer (2010)
[5] Mller, A., Vogel, M.: berwachung von Holztragwerken. Baublatt (September
2009)
[6] Tannert, T., Mller, A.: Structural health monitoring of timber bridges. In: Interna-
tional Conference on Timber Bridges, ITCB 2010 (2010)
[7] Annamdas, K.K.K., Annamdas, V.G.M.: Piezo impedance sensors to monitor
degradation of biological structure. In: Advanced Environmental, Chemical, and
Biological Sensing Technologies VII. SPIE (2010)
[8] Annamdas, V.G.M., Annamdas, K.K.K.: Impedance based sensor technology to
monitor stiffness of biological structures. In: Advanced Environmental, Chemical,
and Biological Sensing Technologies VII. SPIE (2010)
[9] Heiduschke, A., Trmper, W., Haller, P., Cherif, C.: Monitoring von
Holzkonstruktionen mittels Carbonfaser-Sensoren. Bauingenieur 83, 468472 (2008)
[10] Bla, H.J., Frese, M.: Schadensanalyse von Hallentragwerken aus Holz. Technical
report, Karlsruher Institut fr Technologie, KIT (2010)
[11] Aicher, S.: Langzeitbestndigkeit und Sicherheit harnstoffharzverklebter tragender
Holzbauteile. Technical report, MPA Stuttgart (2012)
[12] Bierbaum, A.: Haptische Exploration von unbekannten Objekten mit einer
humanoiden Roboterhand. PhD thesis, Karlsruher Institut fr Technologie, KIT
(2012)
[13] Prasse, T.: Elektrisch leitfhige Funktions- und Strukturverbundstoffe auf der Basis
von Kohlenstoff-Nanopartikeln und -fasern. PhD thesis, Technische Universitt
Hamburg-Harburg (2001)
[14] Schler, R., Joshi, S.P., Schulte, K.: Conductivity of CFRP as a tool for health and
usage monitoring. In: SPIE 4th Annual Symposium on SMART Structures and Mate-
rials, San Diego (1997)
[15] Chung, D.D.L.: Structural health monitoring by electrical resistance measurement.
Smart Materials and Structures 10, 624636 (2001)
[16] Thostenson, E.T., Chou, T.-W.: Real-time in situ sensing of damage evolution in
advanced fiber composites using carbon nanotube networks. Nanotechnology 19,
215713 (2008)
[17] Gilg, R.G.: Ru fr leitfhige Kunststoffe. In: Elektrisch leitende Kunststoffe. Hanser
Verlag (1989)
[18] Balberg, I.: A comprehensive picture of the electrical phenomena in carbon black
polymer composites. Carbon 40, 139143 (2002)
[19] Davidson, T.: Conductive and magnetic fillers. In: Xanthos, M.: Functional Fillers
for Plastics. Wiley-VCH (2010)
[20] Schulte, K., Gojny, F.H., Wichmann, M.H.G., Sumfleth, J., Fiedler, B.: Polymere
Nanoverbundwerkstoffe: Chancen, Risiken und potenzial zur Verbesserung der
mechanischen und physikalischen Eigenschaften. Materialwissenschaft und
Werkstofftechnik 37, 698703 (2006)
[21] Wehnert, F., Heinrich, J., Jansen, I.: Multifunctional adhesives by integration of
carbon nanotubes. In: EURADH 2012 - 9th European Adhesion Conference (2012)
[22] Tjong, S.C.: Deformation mechanisms of functionalized carbon nanotube reinforced
polymer nanocomposites. In: Karger-Kocsis, J., Fakirov, S. (eds.) Nano- and
|Micromechanics of Polymer Blends and Composites. Hanser Verlag (2009)
[23] Kollmann, F.: Technologie des Holzes und der Holzwerkstoffe. Springer, Berlin
(1982)
[24] Mair, H., Roth, S.: Elektrisch leitende Kunststoffe. Hanser Verlag, Mnchen (1986)
[25] Weimantel, H., Gairola, A.: Die Verwendbarkeit von leitenden Kunststoffen fr
taktile Sensoren. Feinwerktechnik & Messtechnik 89, 7984 (1981)
[26] Steinfeld, K., Kalkner, W.: Einstellung und Stabilitt der elektrischen Leitfhigkeit
gefllter Polymerwerkstoffe im Bereich der perkolationsschwelle. Technical report,
TU Berlin, Berlin (1999)
[27] Bokobza, L.: Rubber nanocomposites: New developments, new opportunities.
In:Karger-Kocsis, J., Fakirov, S. (eds.) Nano- and Micromechanics of Polymer
Blends and Composites. Hanser Verlag (2009)
[28] Mullins, L.: Softening of rubber by deformation. Rubber Chemical Technology 42,
339362 (1969)
[29] Wang, P., Ding, T.: Creep of electrical resistance under uniaxial pressures for carbon
black-silicone rubber composites. Journal of Materials Science 45, 35953601 (2010)
[30] Lehmann, J.: Wasseraufnahme von PMMA und PC. KU Kunststoffe 7, 8083 (2001)
Assessment of the Glue-Line Quality in Glued
Laminated Timber Structures
Bettina Franke, Florian Scharmacher, and Andreas Mller
Bern University of Applied Sciences, Architecture, Wood and Civil Engineering,

Switzerland
{bettina.franke,florian.scharmacher,andreas.mueller}@bfh.ch
Abstract. Timber constructions with glulam members have regularly to be

proofed for their performance to avoid structural collapse. For the assessment of
glued laminated timber, it is important to know reliable methods and criteria. The
requirements given in standard EN 386:2001 are valid for the quality control of
the glulam production. The use and application of these two different methods at
existing timber structures were investigated. Problems and issues noted during the
test series and analyses of the results are discussed.
Keywords: glue line quality, structural collapse, assessment of existing timber

structures, core samples, delamination tests, shear strength, modified bending test.
1 Introduction
Nowadays glued laminated timber is the most common structural material for
timber constructions. There were several evolution steps between the introduction
of the glulam by Otto Hetzer in Europe in the year 1906 and today, where the
adhesive, the gluing process and also the produced dimensions have changed. In
general, timber constructions with glulam members are regularly proofed for their
performance and to detect abnormalities to avoid structural collapse. For the as-
sessment and monitoring of glued laminated timber constructions it is important to
know reliable methods and criteria, but the practical engineer has got less guid-
ance. Usually after the first visual inspection where openings along the glue-line
will be detected, the failure reason has to be clarified.
The requirements given in standard EN 386:2001 are valid for the quality con-
trol of the glulam production. These requirements are also used and discussed by
practical engineers for the assessment of existing timber structures, Gaspar et al.
(2010), Tannert et al. (2012). On the other hand nondestructive test methods are
developed and investigated for the detection of defects along the glue line, Dill-
Langer (2012), Sanabria and Sergio (2012). At present these methods are proofed
in first steps under laboratory conditions, a practical application or availability is
not known.
DOI: 10.1007/978-94-007-7811-5_36, RILEM 2014
396 B. Franke, F. Scharmacher, and A. Mller
Thus the paper focuses on the proof of a reliable application of the

EN 386:2001 on existing timber structures. This standard refers to a delamination
test according to EN 391:2001 and a shear test of the glue-line according to
EN 392:1995. The methods according to the two test standards result in different
stress situations within the test specimen, on the one hand transverse tension stress
and on the other hand shear stress. The application of these methods and require-
ments at existing timber structures has been investigated and discussed in compre-
hensive experimental test series. The test methods were extended with the shear
test setup according to EN 408:2010 and a modified bending test series.
2 Material and Method
2.1 Material and Test Methods

Glulam is a well-established engineered wood product of single lamellas of solid
wood glued together. The lamellas are finger jointed in length and glued together
in the depth of the cross sections. Glulam is commonly produced of European
spruce, fir or larch and nowadays also of hard wood like ash, beech or oak, Blass
et al. (2005), Frhwald et al. (2003), Z-9.1-704 (2012). The adhesive used for the
lamination of the lamellas ranges from casein (historical) to a wide range of mod-
ern glues like polyurethane (PUR), resorcinol-formaldehyde (RF), urea-
formaldehyde (UF) and melamine-urea-formaldehyde (MUF).
For the quality control of the glulam production, the requirements for a delami-
nation test according to EN 391:2001 and shear strength test according to
EN 392:1995 are given in EN 386:2001. The delamination test setup is used to
proof the resistance against the climate exposure during the life time of the glulam
member. The test begins with a fully adsorption of water of the specimen by ap-
plying controlled vacuum and compression cycles while the specimen is under
water. Finally the specimen has to be dried climate controlled using an oven with
air circulation. The wet and dry cycle results in tension stress in radial direction
respectively transvers to the glue line in the specimen. For the assessment, the sum
of the openings along the glue line at the end grain developed during the tests has
to be taken in relation to the total glue line length at the end grain of the specimen.
To respect different service classes according to EN 1995-1-1:2004, the delamina-
tion test standard differs between three methods for the wet/dry cycles. The block
specimen, shown in Fig. 1, needed for the quality control of the production can
easily be extracted. But for the assessment of existing structures, the block speci-
men defined in EN 391:2001 cannot be extracted from the structure. Therefore in
addition to the standard test specimen, the test setup was also validated on core
specimens as shown in Fig. 2 and Fig. 3, because this specimen can easily be ex-
tracted from existing structure.
The second required test for the quality control of glulam is a shear test of the
glue line loaded in longitudinal direction. The test standard provides two specimen
shapes, on the one hand the block or bar with a cross section of 50 mm by 50 mm
Assessment of the Glue-Line Quality in Glued Laminated Timber Structures 397
d - cut off
Lamella
Glueline
mm
R l = 80
h
mm T
R l = 80 L
T
R L
L l
T b
Fig. 3 Core test specimen

Fig. 1 Block test specimen Fig. 2 Core test specimen with cut off for
for delamination for delamination
delamination
Loading plate
Loading plate
Supporting plate
Glue line
with teflon layer
Lamella
Glue line
L
L T
T
R
R
Fig. 4 Test setup for core, EN 392:1995 Fig. 5 Test setup for single shear test
including all glue lines of the member depth and on the other hand a drill core
including only one glue line as shown in Fig. 4. The core specimen is here com-
monly used for the assessment of existing glulam structures. Parallel to the shear
strength, the percentage of wood failure (PWF) at the failure plane after testing is
visually examined. In addition to this test standard, the shear strength of the glue
line was tested with a single shear test on a cube (Fig. 5) and a modified three
point bending test setup. The test specimens for both additional test setups were
reduced at midsection to a cross section of 80% to induce the shear stress close to
the glue line, which was located at the center of the test specimen.
2.2 Experimental Test Program

The experimental test program comprises delamination test series according EN
391:2001 method A and shear strength test series. For each test series different
specimens from one glulam member and adhesive were considered. Table 1
summarizes the different specifications classified after the test methods used. All
specimens with the same adhesive are taken from one production series of glulam
members in European spruce. The specimens are conditioned in normal climate,
20 C and 65% relative humidity until testing.
3.1 Delamination Tests According to EN 391:2001

For the assessment of existing structures the application of the delaminations test
was investigated in addition to the standard. Fig. 6 shows the standard test speci-
men after the test with the delaminations observed. On the other hand, a core
specimen without cut of with delaminations is shown in Fig. 7. The comparison of
the test results are shown in Fig. 8 and Fig. 9 respectively. The total percentage of
delamination should be less than 5% for the standard specimen for test method A
arccording EN 386:2001. For the assessment of the delamination of the core spec-
imen, the same criterion was used as for the standard test specimens. The compar-
ison between the standard test setup and the core specimen used shows in some
cases significantly larger total percentages of delaminations for both adhesives
PUR and MUF.
Table 1 Test program for delamination and shear tests

SeriesTest method Standard Shape Sizes [mm] Number Material Adhesive
01 Delamination EN 391-A Block b/h/l = 160/320/75 51 GL28k-spruce PUR
02 Delamination EN 391-A Block b/h/l = 160/160/75 42 GL24k-spruce MUF
35, l = 80 cut off
03 Delamination EN 391-A Core 37 GL28k-spruce PUR
0/9/15
35, l = 80 cut off
04 Delamination EN 391-A Core 34 GL24k-spruce MUF
0/9/15
05 Shear test EN 392 Core 35, l = 80 51 GL28k-spruce PUR
06 Shear test EN 392 Core 35, l = 80 42 GL24k-spruce MUF
09 Shear test - Block b/h/l = 160/200/200 17 GL28k-spruce PUR
10 Shear test - Block b/h/l = 160/200/200 14 GL24k-spruce MUF
11 Shear test - Beam b/h/l = 160/320/1350 17 GL28k-spruce PUR
12 Shear test - Beam b/h/l = 160/160/800 14 GL24k-spruce MUF
The reason for the disagreements is due to the different transverse stress situa-
tion as a result of the moisture gradient reached in the wet/dry cycle. The moisture
gradient generated in the dry cycle is more homogenous and with less amplitudes
for the core specimen without cut offs than for the standard test specimen.
Furthermore the round surface leads to other stress situations than the flat sides of
the standard test specimen or the core specimen with cut offs, as shown in
the numerical solutions in Fig. 10 and Fig. 11. The numerical simulation of the
standard and core test specimens with different cut offs is based on an orthotropic
material law with properties for European spruce and shows as result the stresses
perpendicular to the glue line for a uniform moisture gradient during the drying
process.
Assessment of the Glue-Linee Quality in Glued Laminated Timber Structures 3999
Fig. 6 Standard test specim

men after the test Fig. 7 Core specimen after the test wiith
with marked delamination marked delamination
Adhesive: PUR s
standard specimen Adhesive: MUF standard specimen
40% 40%
c
core specimen no cut core specimen no cut
Total percentage of delamination
c
core specimen cut off 9 mm core specimen cut off 9 m
mm
c
core specimen cut off 15 mm core specimen cut off 15 m
mm
30% 30%
20% 20%
10% 10%
5% 5%
0% 0%
0 1 2 3 4 5 6 7 8 9 10 0 11 12 13 14 15 16 17 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Test memberr Test member
Fig. 8 Delamination test resu

ults for adhesive Fig. 9 Delamination test results for adhesivve
PUR MUF
Fig. 10 Stress perp. to thee glue line due Fig. 11 Stress perp. to the glue line due to
to moisture gradient forr the standard moisture gradient for the core specimen, nno
specimen cut offs a), cut off 9 mm b), cut off 15 m
mm
c)
The results observed for core specimen with cut offs show comparable results
as the standard test setup. This first results reached will be extended in further test
series considering different adhesives and wood species to provide an alternative
test method for the assessment of existing timber structures.
3.2 Shear Tests

In addition to the delamination test series also shear tests were carried out with
specimen taken from the same glulam members. The core samples are used in
all test series to determine the shear strength and the percentage of wood failure
(PWF). Fig. 12 shows the shear failure surface for a core. Fig. 13 shows the
results observed for the test series no. 5 and 6 (PUR and MUF). The failure
criterion for the shear tests depends on the shear strength reached and the corre-
sponding PWF, as given in EN 386:2001 and shown in Fig. 13 for soft wood.
With an increase of the shear strength, the allowing PWF is smaller but must be
at least 20%.
For the evaluation of the shear strength of the glue line, further shear tests
with a different test setup were carried out and compared as shown in Fig. 14
and Fig. 15. Parallel to the shear tests according to EN 392:1995, short three
point bending tests with a reduced depth to span ratio and a reduced cross
section at the glue line as well as single shear tests were carried out. The coeffi-
cient of variation is small in each test configuration. The comparison between
the test setups shows differences in the mean shear strength levels because of
the size effect of the specimen and the resulting interaction of different stress
situations (compression, tension and shear). It is to note, that the shear strengths
determined are not directly comparable with the shear strength classes at the
design standards.
100%
failed
80%
Percentage of wood failure
60%
40%
passed
20% Requirement EN 386:2001
Adhesive PUR
Adhesive MUF
0%
0 2 4 6 8 10 12 14
Shear strenght in N/mm2
Fig. 12 Fracture surface with wood failure Fig. 13 Shear test results for the test series
of a tested drill core with PUR and MUF
Adhesive: PUR Shear test EN 392-core Adhesive: MUF Shear test EN 392-core
14 14
Bending test Bending test
12 Shear test-block Shear test-block
12
Shear strength in N/mm
Shear strength in N/mm

10 9.2 N/mm 10
10.2 % COV 8.6 N/mm
12.9 % COV
8 8
6,8 N/mm 8.1 N/mm
10.5 % COV 7.8 % COV
6 6
5.6 N/mm
5.3 N/mm
4 23.7 % COV 4 11.5 % COV
2 2
0 0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Test member Test member
Fig. 14 Shear strength for glulam (PUR) Fig. 15 Shear strength for glulam (MUF)
4 Discussion and Comparison
The EN 386:2001 requires only one of the two quality tests, the delamination or
shear test, for service class 1 or 2 according to the correlation between the delami-
nation and shear test found by Zeppenfeld and Grunwald (2005). This means theo-
retically that only one of the tests has to be carried out for the quality control of
the production line but may also be in the assessment of existing timber structures
in service class 1 or 2, EN 1995-1-1:2004. However in existing timber structures,
the real moisture content can vary in a wide range and easily reach service class 3
where this agreement is not valid. Furthermore defects in the construction or tem-
porary changes can increase the local moisture content of timber members easily.
Also the correlation reachable can depend on the adhesive and wood species used,
Aicher and Reinhardt (2006), Steiger and Risi (2011), Schmidt et al. (2010).
Fig. 16 and Fig. 17 show the comparison of the delamination standard test and
shear tests for the test series carried out with PUR and MUF. For the direct com-
parison of the delamination and shear test series, the minimum required percent-
age of wood failure (PWFmin) is calculated reversely using the shear strength
determined. If the ratio between the effectively (PFW) and the calculated required
percentage (PWFmin) is less or equal to one, the specimen passed the require-
ments. For the delamination test series the same criteria of 5% is used as presented
before.
A correlation of 76% for the test series with the adhesive MUF and only a cor-
relation of 59% for the test series with adhesive PUR are reached. The results
reached show no satisfied correlation between the shear and delamination tests.
The disagreements observed confirm the dependency on the adhesive used for the
production of glulam. Furthermore the delamination test is obviously the more
demanding test setup, because in the cases where specimens passed the shear test,
they failed in the delamination test which discloses the defects.
The delamination test replicates the different climate situation during the life
time of the timber member. On the other hand shear stresses are always existent in
the structure. Therefore it is important to proof the quality of new glulam members
Adhesive: MUF Adhesive: PUR

20% 4.0 20% 4.0
Delamination test EN 391

Delamination test EN 391
18% 18%
Shear test EN 392
16% Shear test EN 392
Shear test (PWFmin/PWF)

16%
Shear test (PWFmin/PWF)

3.0 3.0
14% 14%
12% 12%
10% 2.0 10% 2.0
8% 8%
6% 6%
1.0 1.0
4% 4%
2% 2%
0% 0.0 0% 0.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
Test member Test member
Fig. 16 Comparison of delamination test Fig. 17 Comparison of delamination test

and shear test for MUF and shear test for PUR
or to assess existing timber structures using both test setups, delamination and
shear. However, both methods only allow a local assessment and cannot be as-
sumed to be valid for the complete timber structure, Tannert et al. (2012).
5 Conclusion and View
For the objective to find reliable methods for the assessment of existing timber
structures, both test setups, delamination and shear, according to EN 386:2001
were used and compared between each other and also further delaminations and
shear test setups. For the delamination test, generally the test specimen required
cannot be extracted from existing timber structures. Therefore core samples with
different cut offs are prepared and tested with the current specified wet/dry cycles
in EN 391:2001 in addition to the standard test specimen. The test results reached
and the numerical simulation show that it is important to reach a comparable stress
situation between these different test specimens. The first results observed show a
positive correlation between the standard test specimen and the core samples with
cut offs. The first delaminations test series with core samples with cut offs shows
a good correlation and shows a promising method. But has to be extended before
the method can reliably be applied for the assessment of existing timber structures.
Secondly the shear strength in combination with the percentage of wood failure
are determined on core samples and compared with further shear tests. The com-
parison of the shear strength points out a size effect on the strength. Furthermore
the results observed on core samples represent only the local condition of the tim-
ber structures and is not valid for the complete structure.
Finally the correlation between the delamination and shear test according to
EN 386:2001 show that both test setups are necessary for a comprehensive as-
sessment of timber structures. The two test setups are totally different in the stress
situation resulting in the specimen; on the one hand transverse tension stress on
the glue lines due to the wet/dry cycle and on the other hand the shear stress paral-
lel to the glue line. The delamination test represents the loads/stresses according to
the climate condition during the life cycle of the timber structure, whereas the
shear test reflects more the structural loading situation. A comprehensive assess-
ment of existing timber structures is not possible with only one test setup. To in-
clude the delamination test setup in the assessment of timber structures, a new test
procedure has to be developed for small samples like the core sample with cut offs
used in first steps.
References
Aicher, S., Reinhardt, H.W.: Delaminierungseigenschaften und Scherfestigkeiten von
verklebten rotkernigen Buchenholzlamellen. European Journal of Wood and Wood
Products 65, 125136 (2006)
Blass, H.J., Denzler, J., Frese, M., Glos, P., Linsemann, P.: Biegefestigkeit von
Brettschichtholz in Buche. Publisher University of Karlsruhe, Germany (2005)
Dill-Langer, G.: Detektion von Fehlverklebungen Forschungsergebnisse und
Praxisrelevanz. In: 2. Stuttgarter Holzbau-Symposium, Germany (2012)
European Committee for Standardization (CEN), EN 386:2001, Glued laminated timber,
Performance requirements and minimum production requirements, Brussels
Delamination test of glue lines, Brussels
Shear test of glue lines, Brussels
European Committee for Standardization (CEN), EN 408:2010, Timber structures,
Structural timber and glued laminated timber, Determination of some physical and
mechanical properties, Brussels
European Committee for Standardization (CEN), EN1995-1-1:2004, Eurocode 5, Design of
timber structures - General - Common rules and rules for buildings, Brussels
Frhwald, A., Ressel, J.B., Becker, P., Pohlmann, C.M., Wonnemann, R.: Verwendung von
Laubhlzern zur Herstellung von Leimholzelementen. Research report, University
Hamburg, Germany (2003)
Gaspar, F., Gomes, A., Cruz, H.: Assessment of natural and artificial ageing of glued lami-
nated timber - Core drilling, shear and delamination tests. In: World Conference on
Timber Engineering, Italy (2010)
Sanabria, M., Sergio, J.: Air-coupled ultrasound propagation and novel non-destructive
bonding quality assessment of timber composites, doctoral and habilitation theses.
ETH-Zuerich, Switzerland (2012)
Schmidt, M., Glos, P., Wegener, G.: Verklebung von Buchenholz fr tragende
Holzbauteile. European Journal of Wood and Wood Products 68, 4357 (2010)
Steiger, R., Risi, W.: Qualittskontrolle von Brettschichtholz: Vergleich der Prfverfahren
Blockschertest und Delaminierungstest. Research report, Empa, Duebendorf, Switzer-
land (2011)
Tannert, T., Valle, T., Mller, A.: Critical review on the assessment of glulam structures
using shear core samples. Journal Civil Structural Health Monitoring 2, 6572 (2012)
Zeppenfeld, G., Grundwald, D.: Klebstoffe in der Holz- und Mbelindustrie. DRW-Verlag
Weinbrenner, Leinfelden Echterdingen (2005)
Z-9.1-704, Bauaufsichtliche Zulassung VIGAM Brettschichtholz aus Eiche und BS-Holz
Hybridtrger von Elaborados y Fabricados Gamiz S.A. Spain (2012)
Review of Recent Research Activities
on One-Component PUR-Adhesives
for Engineered Wood Products
Christian Lehringer* and Joseph Gabriel
Purbond AG, Sempach Station/Switzerland

christian.lehringer@purbond.com
Abstract. Manufacture of engineered wood products requires safe adhesive

systems. Since their introduction in the mid-1980s, one component polyurethane
(1C-PUR) adhesives have proven to provide a reliable alternative to commonly
water based aminoplast (UF, MF, MUF) and phenoplast (PR, PRF) adhesive
systems. No formaldehyde, no mixing and fast curing at room temperature are
some of the reasons, why 1C-PUR adhesives have continuously experienced
increasing acceptance in the field of structural timber gluing. Driven by the
intention to establish a highest possible level of adhesive performance and
security, this successful development was accompanied by extensive scientific
and industrial research activities. The present article aims at providing an
understandable overview of the most relevant results from those research activities
during the past years. Major aspects such as chemical and physical bondline
integrity, adhesive behaviour at elevated temperature and fire, and gluing
of alternative wood species will be addressed in order to present an insight into
1C-PUR adhesive-technology and an outlook to its future development.
Keywords: One-component PUR adhesives (1C-PUR), bond line, elasticity,

ductility, engineered wood products, heat resistance, fire behaviour.
1 Introduction
For more than ten decades, engineered wood products have been used for creating
innovative wood buildings. The first known glued laminated timber (glulam)
construction was the meeting hall of the King Edward College in Southampton in
1860. At the turn of last century, the first patents for glulam were issued in
Switzerland and Germany (e.g. [1]) and signaled the true beginning of glued
laminated timber construction.
Always following the pattern of raw material disintegration (stem to board) and
subsequent re-assembly (board to lamella to engineered wood product),
*
DOI: 10.1007/978-94-007-7811-5_37, RILEM 2014
406 C. Lehringer and J. Gabriel
appropriate adhesives were used in order to provide a safe and durable wood
bonding. Traditionally, adhesives from organic sources such as casein were used
for gluing. However, main disadvantage of these natural adhesives was their low
stability against moisture which significantly limited their field of application.
During the 1930s, urea-formaldehyde-based (`aminoplastic, UF) adhesives
offered reasonable alternatives on synthetic basis but their tendency to hydrolyze
under the influence of increased moisture presented a certain risk.
A big development in the engineered wood industry was the introduction of
fully waterproof phenol-resorcinol-(formaldehyde) (PR/PRF) resins in the 1940s.
This allowed the production of engineered wood products to be used in exposed
exterior environments without thread of bondline degradation.
Further progress was made with the introduction of melamine-(urea)-
formaldehyde (MF/MUF) resins in the 1970s, which show good water resistance,
but are known to form rather brittle bondlines.
During the past 30 years, one-component polyurethane (1C-PUR) adhesives for
engineered wood products have experienced a rapid and successful development.
Originally invented by Otto Bayer/Germany in 1937, the first 1C-PUR adhesives
were initially used for non-structural gluing of wood, for leather and fabric etc.
The first applications of 1C-PUR adhesives for engineered wood products were
realized in 1985 in Switzerland. Only nine years later in 1994, the first
commercially successful 1C-PUR adhesive (PURBOND HB 110, Purbond
AG/Switzerland) was approved by German building authorities for the
manufacture of engineered wood products. Since then, many different types of
1C-PUR adhesives were introduced into the market, offering an attractive
alternative to conventional adhesive systems [2, 3].
The introduction of 1C-PUR adhesives revolutionized the production of
engineered wood products for several reasons:
1. One component system no mixing, easy handling
2. Reduction of press time
3. Fast bonding at room temperature
4. Reduced coating weight
5. 100 % solid content, no solvents
6. Inert bondline, no formaldehyde or volatile organic compounds (VOC)
7. Ductile bondline characteristics
8. Wood colour invisible bondlines
9. Gap filling properties
10.Long shelf life
By the end of 2012, 6 producers in Europe distributed overall 42 certified 1C-
PUR adhesives that were in conformity with DIN EN 1052:2008 [4] and/or
EN 14080:2008 [5] for the manufacture of finger-jointed and/or face glued
engineered wood products [6-8]. The high acceptance of 1C-PUR adhesives by
manufacturers of engineered wood products led to the continuous replacement of
conventional adhesive systems in products such as glulam and finger jointed solid
timber. Cross laminated timber (CLT) is a product almost exclusively produced
Review of Recent Research Activities on One-Component PUR-Adhesives 407
with 1C-PUR adhesives. Nowadays, modern engineered wood manufacture is not

imaginable without 1C-PUR adhesives anymore.
Chemistry and reaction kinetics of 1C-PUR adhesives are fundamentally
different from aqueous, two-component formaldehyde based phenolic and
aminoplastic systems [9]. However, until the introduction of specific requirements
and test standards [10-14], the 1C-PUR adhesives were evaluated and certified on
basis of the requirements for phenolic and aminoplastic adhesive systems [15]. It
was thus the intention of extensive scientific research activities to provide a solid
platform on which the different adhesive systems can be compared and where the
1C-PUR adhesives can display their specific performance characteristics.
As innovative and upcoming adhesive systems, the 1C-PUR adhesives are
continuously matter of critical scientific investigations which cover manifold
aspects of adhesive evaluation. Proposing a simple structuring, these research
activities shall be divided into three main subsections.
Firstly, many resources are directed to the evaluation of the chemical and
physical characteristics of 1C-PUR adhesive bondlines. Topics such as mechanical
characteristics of wood-adhesive interface (e.g. fracture mechanics, ductility), wet
adhesion and also green gluing are subject of recent and ongoing research
activities.
Secondly, due to their special chemistry, 1C-PUR adhesives are believed to
display adversely performance under high temperature and fire exposure.
Thorough investigations have been issued on the question of 1C-PUR adhesives
short-term and long-term thermal stability.
Thirdly, market requirements and raw material availability are increasingly
diversifying the sector of engineered wood products, resulting in an increasing
demand for processing other wood species than Norway spruce, fir and pine the
main wood species used in Europe and also modified wood. Correspondingly,
much research is directed towards gluing softwood species with coloured
heartwood, hardwood species and e.g. acetylated wood.
The present article aims at providing an overview on research activities that
have been conducted in the field of 1C-PUR adhesives during the past years.
Focus is laid on these three main topics as they are considered being relevant for
the performance and safety of 1C-PUR adhesives when used for the manufacture
of engineered wood products. Results will be presented to provide an insight into
1C-PUR adhesive-technology and an outlook to its future development.
2 Integrity of 1C-PUR Adhesive Bondlines
2.1 Rheology and Wood-Adhesive-Interphase Characteristics

1C-PUR adhesives are commonly known to be rather ductile adhesive systems
[9, 16]. Several works of Konnerth et al. [17-19] gave evidence of higher elastic
modulus and more creep of 1C-PUR polymers than MUF and PRF. 1C-PUR
adhesives showed 15 times lower elastic modulus than MUF adhesives, resulting
in significant higher ductility (Fig. 1).
Fig. 1 Left: Correlation of elastic modulus measured using mechanical extensometer

(Emacro) and elastic modulus measured using ESPI (EESPI) [17]. Right: Comparison of the
reduced elastic modulus from nanoindentation of pure polymer films and polymers in
adhesive bond lines [18].
Based on these results, Konnerth et al. [20] assumed in a later work on finger
joints that the corresponding stress concentrations are highest in the stiff
bondlines, compared to a specimen with a less stiff bondline under the same
external load. Hence, MUF bondlines possibly suffer from more intense stress
concentrations than PUR bondlines.
Mller and Veigel [21] showed in a study about fracture energy of 1C-PUR
adhesive bondlines that the higher ductility can result in significant higher failure
load for certain load situation. Furthermore it was concluded that stable and
reliable bondlines can be achieved with significant lower adhesive spread rates
than with MUF or PRF adhesives.
When Klusler et al. [22] investigated the influence of moisture on stress-strain
behavior of films of 1C-PUR, PUR prepolymers, PRF and MUF adhesives, they
could confirm the ductility of the tested 1C-PUR polymers on several climate
stages in contrast to the brittleness of MUF and PRF polymers. The fracture strain
of PRF and MUF specimens remained below 5 %, whereas that of the 1C-PUR
adhesives and the PUR prepolymers reached at least 20 %. After redrying, all
adhesives re-gained or even improved their mechanical characteristics. And, at
any stage of relative humidity exposure, the tensile strength and the modulus of
elasticity of MUF and PRF were as high as those of the best performing
polyurethanes [22].
Nonetheless, testing of pure adhesive films might not display the true and real
situation as it occurs in a bondline between two wood surfaces as found by Ren
et al. [23]. They revealed significant wood/PUR interactions that probably had an
influence on the size and size distributions of soft and hard dispersed phases in the
cured 1C-PUR adhesives. It was thus recommended that PUR studies should be
conducted under conditions that simulate real wood/PUR bondlines.
Also Mller et al. [16] picked up the matter that different adhesives show
considerable differences in their stiffness and mechanical properties. According to
their investigations, testing of adhesive bond strength by means of the lap joint test
according to EN 302-1 [24] delivered no clear differentiation of the adhesives,
because of the high percentage of wood failure. Even though polyurethane
adhesives showed lower percentage of wood failure, they reached higher fracture
strength than brittle adhesives. For different loads, more ductile polyurethane
resins reached higher fracture strength. This can be explained by the ductility of
the bondline which lowers peak stresses at the wood-adhesive interface (Fig. 2).
For brittle adhesives the deformation mainly takes place in the wood structure,
causing a high wood fracture percentage in the substrate beyond the wood-
adhesive interface. Due to the ductile behavior of polyurethane polymers, much
higher fracture energy was found for 1C-PUR adhesives.
(a) (b)
Fig. 2 Shear stresses in lap joints of (a) PUR and (b) MUF as shown by [16]
A deeper view into the failure mechanisms of adhesive bondlines was realized
by Hass et al. [25] at the scanning-electron-microscopic level via in situ tensile
tests. The interaction of cracks with adhesive bondlines under various angles to
the crack propagation was the focus of this study, as well as the respective loading
levels for UF, PUR, and PVAc. It could be shown that brittle UF bondlines
provide new crack starters beginning with a curing damage. The crack pattern can
be accelerated and shows the same unstable behavior as through sound wood.
Bondlines of PUR and PVAc can slow down and stabilize the crack pattern by
forming adhesive bridges between two adherents. The adhesive layer itself is able
to deform plastically, leading to blunting of the crack tip. This is in contrast to the
traditional belief that the failure of a bonding should occur in the wood part, away
from the bondline (wood failure).
Obersriebnig et al. [26] found indications also by means of nanoindentation that
the interface between the cell wall and the 1C-PUR adhesive was weaker than
with UF-resins. However, quantitative analysis of indentation curves revealed a
very different pattern of interaction between the different cell wall layers and the
UF and PUR adhesives, respectively. While UF showed a better adhesion to the
freshly cut and thus more hydrophilic S2-layer and less adhesion to the uncut and
thus more hydrophobic inner cell wall surface (S3), 1C-PUR adhesives showed
the opposite trend. Atomic force microscopy revealed a less polar character for the
inner cell wall surface (S3) compared to the freshly cut cell walls (S2). The
authors proposed that differences in polarity of the adhesive and of the cell wall
surfaces account for the observed trends. This could be an indication that 1C-PUR
adhesives show better adhesion on aged wood surfaces, thus opening the
discussion, if the common requirement of gluing max. 24 hours after planning
could maybe extended.
2.2 Adhesive Penetration and Reaction Kinetics

Regarding the penetration of different adhesives into wood cell walls, it was
shown by means of scanning thermal microscopy (SThM) that 1C-PUR adhesives
did not change thermal properties in the cell walls, whereas cell walls adjacent to a
PRF showed a distinct change in thermal properties [27]. This was attributed to
the diffusion of liquid PRF adhesive components into the cell wall prior to curing,
which does not occur for 1C-PUR adhesives due to their higher molecular weight
as also shown by Gindl et al. [28].
Despite serious effort, the chemical and physical interaction of 1C-PUR
adhesives with the adherent wood cell wall is not fully understood until now.
Studies on wood/isocyanate reactions by Bao et al. [29] showed that dry wood can
react with PMDI to form urethane structures, and that dry hemicelluloses and
lignin can react with PMDI to form urethane, but dry cellulose cannot, under
normal processing conditions.
Waver et al. [30] proposed that the formation of covalent bonds between the
adhesive polymers and the hydroxyl groups of the wood cell walls under realistic
gluing conditions is rather improbable. Infrared spectroscopy was applied to
investigate the reaction between phenyl isocyanate and glucose, cellulose, lignin,
and wood. The results showed that the isocyanate reacts with all these entities, but
that its reaction with water is quicker than with any of the other hydroxyl-
containing compounds, and when water is present the water-isocyanate reaction
dominates all others. For wood at higher moisture levels than oven dried, the
infrared results suggested that the isocyanate reacts with the water in preference to
the wood polymers. Hence, it must be assumed that the adhesion performance of
1C-PUR adhesives of the wood surface is owed to physical anchorage effects and
possibly the formation of van-der-Waals-interactions and hydrogen-bonds.
1C-PUR adhesives need a certain amount of water for a complete curing
reaction. Dry climate conditions in the production facilities can induce a rapid
drying of the wood surfaces with equilibrium moisture contents below 8 %. Below
this limit, curing kinetics are significantly reduced due to lack of water availability
and pressing time must be adapted correspondingly in order to guarantee a safe
bonding. Kgi et al. [31] investigated this phenomenon and found that controlled
water spraying on the wood surface prior to adhesive application improves the
bonding quality.
In contrast, higher moisture contents are not a problem for the curing process of
1C-PUR adhesives. By investigating the gluability of green wood i.e. raw
sawn without successive drying Sterley et al. [32] demonstrated the feasibility of
this approach, which however has not been implemented broadly into the
industrial processes to date.
3 1C-PUR Adhesives at High Temperatures and Under Fire

Exposure
3.1 Test Methods

The determination of adhesive performance in structural wood members at high
temperatures and under fire exposure is covered by various research activities in
Europe and North America. However, worldwide no uniform test procedures have
been defined yet.
Building authorities in Canada, the Unites States of America and Japan require
full-scale fire-resistance tests in order to evaluate the materials used for structural
wood members according to specified requirements of ASTM E119-05a [33] and
CAN/ULC S101-07 [34]. Since full-scale tests are very expensive and time
consuming, small scale tests were developed to test the shear strength of adhesives
at elevated temperatures. The standard ASTM D7247-07ae1 [35] defines a method
performing block-shear tests after preheating the specimens to elevated
temperatures near wood ignition of 232 C. But no link between these tests and
the performance in fire could be established [36].
In Canada, adhesives are commonly tested at elevated temperature of 180 C by
means of a creep test according to CSA O112.9-04 [37]. In this regard, a new test
method was recently developed at Forintek, Canada to test finger-jointed
specimens at elevated temperature by Craft et al. [38]. Forinteks method
combines the ASTM D4688-99 [39] and the CSA O112.9-04 [37] standards where
a small finger-jointed specimen is tested in a tensile creep test at 220 C.
Craft et al. [38] found that the testing temperature should be increased to
220 C since adhesives fulfilling the tests at 180 C failed in full-scale fire testing.
Since PRF and MF adhesives fulfilled the requirements of the tests, the test
temperature was considered not too high, irrespectively of the fact that 1C-PUR
and PVA samples failed in these investigations.
In fact, testing at these high temperatures is questionable. Firstly, such
temperatures do not occur to timber constructions during service life. Secondly,
testing at temperatures close to wood ignition over a longer preheating period does
not reflect a fire situation, as shown by Klippel et al. [40]. Some 1C-PUR
adhesives have passed the requirements of full-scale fire-resistance tests in the
past. It thus remains questionable, if the test method displays an appropriate
method to determine a short intensive temperature influence as it occurs in a fire
situation.
In Europe, to date no testing procedure has been established on basis of a

European EN-standard for the evaluation of adhesive performance under elevated
temperature or fire exposure. Even though, with the revision of the adhesives
standards EN 301 [15], EN 15425 [14] and the future implementation of
FprEN 16254 [41] the creep test according to EN 15416-2 [10] becomes
mandatory for all structural adhesives up to 90 C, no reliable statement can be
made neither for elevated temperatures nor for fire exposure.
Thus, a broad research project on the fire resistance of bonded structural timber
members is currently in place, aiming at the development of simplified design
models for the fire resistance of bonded structural timber elements [40, 42]. The
results show that the structural behavior of finger joints at elevated temperature is
influenced by the behavior of the adhesive, confirming the results of Knig et al.
[36]. But also, dominated failure modes were due to the random occurrence of
weak zones (e.g. finger joints, knots and other defects) in commercial graded
bonded structural timber elements and not defects in the finger joint bondlines
so that more experimental and numerical analysis will be performed [40, 42].
Another study investigating the shear behavior of different adhesives at
elevated temperatures was carried out by Frangi et al. [43]. The results obtained
from furnace tests revealed that performance of 1C-PUR adhesives under fire
exposure depends strongly on its chemical composition and varies from one 1C-
PUR adhesive to another. This finding is supported by Clau et al. [44] on the
chemical structure of 1C-PUR adhesives and on the influence of filler materials in
their formulations [45].
The creep behavior of adhesive bonds was investigated by Richter et al. [46]
and Na et al. [47]. A higher initial strength (due to higher concentrations of
isocyanate) caused a reduction of the creep in the low temperature range up to
50 C. It was found that the variation of a few chemical parameters had a big
impact on the thermal stability of 1C-PUR adhesives. These parameters were the
relative proportion of remaining-NCO groups in the polyurethane, the degree of
polymerization of the prepolymer and the rate of reaction. This confirms that the
mechanical and thermal behavior strongly depended on the individual chemical
composition of the product.
Moreover as shown by the activities of George et al. [48] and Richter and
Steiger [49] even the methodology of measurement is decisive for a realistic
determination of creep values for 1C-PUR adhesives in wood bondings in order to
avoid contradictory results not representing full scale situations in engineered
wood products.
3.2 Engineered Wood Products

Studies by Clau et al. [50] showed that all adhesive groups (except PVAc, which
fail much earlier) reveal strong decreases in tensile shear strength at temperatures
between 180 and 200 C (Fig. 3). Even though there are slight differences between
the different adhesive brands, the reduction occurs more or less according to the
same pattern and most of the adhesives provide sufficient shear strength over a
wide temperature range of up to 200 C.
Fig. 3 Shear strength of different adhesives according to EN 302-1 of specimens against

temperature [50]
Wood has a very low thermal conductivity; therefore, it acts as efficient thermal
isolation for the bondlines. Hence, the area of temperatures above 100 C is very
small compared to the entire cross-section for a properly designed glued laminated
timber member at time of failure. Under fire exposure, temperatures between 180
and 200 C are reached only in a very small zone at the outer area of the wood
member. This means that adhesives used in glulam beams need sufficient thermal
strength at temperature around 100 C to show sufficient fire performance (Frangi
and Klippel 2012, personal note).
Furthermore fire behavior of multi-layered engineered wood products depends
highly on the charring rate of the single wood layers and the heat resistance of the
adhesive has only a minor influence. The charring leads to reduction of the timber
cross-section and was still governing the load-bearing capacity for finger-jointed
specimens tested by Klippel et al. [40].
The fire behavior of cross-laminated solid timber panels (CLT) was
experimentally and numerically studied by Frangi et al. [51] whereby particular
attention was given to the comparison of the fire behaviour of homogeneous
timber panels. The results showed that the fire behavior of CLT panels is strongly
influenced by the behaviour of the single layers. Charring rate is considered to be
constant over time [52], however, if the charred layers fall off, an increased
charring rate needs to be taken into account. The same effect is observed for
initially protected timber members after the fire protection has fallen off. Thus the
fire behaviour of cross-laminated solid timber panels can strongly depend on the
thickness and the number of layers.
Frangi et al. [53] found that the outer wood-layers of multi-layered CLT panels
provide a safety margin and can be sacrificed during fire exposure without a
significant effect on the load-bearing capacities of a CLT element. Taking into
account that the second layer on the fire exposed side is usually statically
ineffective, for common multilayered CLT panels with the thickness of the single
boards in the range of 20 to 35 mm the effect of falling off of charred layers on the
load-bearing capacity in fire is therefore relatively small [53].
When large scale fire tests with CLT panels glued with PRF and with 1C-PUR
adhesives were conducted at FP-Innovations/Canada [54], there was a significant
difference detected in the ability of the CLT panel to prevent integrity failure
during the fire exposure, which in these specific tests, may be attributed to the
difference in the adhesive. It appeared that the 1C-PUR adhesive sealed up any
gaps within the panel during manufacturing which prevented any air flow through
the panel during the test. This was not observed with the PRF adhesive and
resulted in integrity failure occurring significantly earlier [54]. The work
demonstrated that with specific chemical adaptations a 1C-PUR adhesive can
show a better performance than a PRF resin exposed to fire.
Klippel et al. [40, 42, 55] revealed in experimental and numerical investigations
that failure of glued laminated timber due to fire exposure is very improbable
since the failure mode is rather dominated by the bending resistance of the beam
than by the adhesive type itself for most applications. Based on the numerical
analysis, it could be concluded that even if the adhesive strength in the bondline is
strongly temperature-influenced, failure in most common design applications will
occur because the bending resistance of the wood member is exceeded. Failure in
the bondline is hardly to be expected.
4 1C-PUR Adhesives for Alternative Wood Species
In central Europe, engineered wood products such as glulam or cross laminated

timber (CLT) are most commonly manufactured with Norway spruce (Picea
abies) and fir (Abies alba). However, the use of alternative wood species for
engineered wood products has been attempted ever since. Softwood species with
colored heartwood such as Larch (Larix spp.), Douglas fir (Pseudotsuga
menziesii) and have successfully been used in the past, even though gluing gets
more challenging due to higher contents of resin and other wood extractives
The gluing especially of larch wood displays certain obstacles not only for 1C-
PUR adhesives when used for service class 3 and thus tested in delamination tests
according to EN 302-2 [56], most probably due to the high content of water
soluble hemicelluloses such as arabinogalactan. However, even though some
MUF and PRF brands have shown rather good performance with larch wood [57]
it remains questionable, if this can be transferred to all other products of these
adhesive groups.
Climate change and forest restructuration in central Europe results in a
continuously reducing availability of Norway spruce and an increasing offer of
hardwood species. Thus, growing efforts have been made in the recent past to
develop innovative products and markets to increase hardwood use for engineered
wood products. Strongest focus was laid on beech wood (Fagus sylvatica L.), but
also ash (Fraxinus excelsior), oak (Quercus robur/petraea) birch (Betula spp.) and
chestnut (Castanea sativa) were integral part of research activities, resulting in
national approvals for some engineered wood products [58, 59].
According to a poll among glulam manufacturers [60], the gluability of beech
wood was described as rather problematic due to strong deformations during kiln
drying, higher losses during planning and generally a more demanding
mechanization for handling and pressing than for softwoods.
Studies with shear tests according to EN 392 or EN 302-1 [24, 61] on the short-
term strength of defect-free, standardized specimens have revealed a variety of
suitable adhesive types [62-64]. The studies showed sufficient performance of
different adhesive systems even on red heartwood or at different steaming levels
However, the challenge with gluing beech wood is to ensure a bond durability
that provides longterm security for structural products. The poor performance of
beech glulams bonded with different adhesives in the obligatory delamination tests
according to EN 302-2 [56] is one of the reasons why beech wood has not yet
been broadly used for engineered wood products [62, 65]. A significant
improvement of longterm gluing quality can be achieved by reduction of lamella
thickness resulting in reduced internal stresses and increase of assembly-,
pressing- and postcuring-times during pressing [66]. In this context, an interesting
approach is the production of laminated veneer lumber (LVL) by a German beech
wood processor, who aims at obtaining a national approval for its product in the
close future.
Brandmair et al. [67] give an overview of beech, ash and oak wood gluing for
structural timber elements. While the application of PRF and some MUF resins is
shown to be feasible, more research activity is recommended for the application of
1C-PUR adhesives.
5 Summary and Outlook
One-component moisture curing Polyurethane adhesives proved over the last

years to be a very versatile class of adhesives for the production of engineered
wood elements. Much scientific investigation and practical work had been
necessary to establish a high level of security for the use of 1C-PUR adhesives.
One of the main differences between the aqueous, formaldehyde-curing- and
the isocyanate-curing adhesives is the elasticity/ductility of the bondline.
Formaldehyde-based adhesives like PRF or MUF have usually a high crosslink
density after curing, which results in rather brittle, high-modulus bondlines.
Polyurethane adhesives can be formulated from very low-modulus adhesives to
hard and brittle systems. For wood bonding, it is advisable to align the modulus of
the adhesive to the modulus of the wood itself. In this way, higher fracture
strength can be reached, because peak stresses are not absorbed mainly in the
wood (as it is the case with brittle adhesives) but equally in the wood and in the
bondline. This may however lead to a lower percentage of wood failure.
The heat resistance also depends to a large extend on the crosslink density of
the cured adhesive, and thus on the formulation. The 1C-PUR adhesives currently
in use for engineered wood application have no problem to withstand temperature
loads of a normative use case. Prolonged extreme temperatures in the range
of 100 C or above for special applications have still to be examined for any kind
of adhesive. The development of 1C-PUR adhesives with a higher heat resistance
than the currently approved systems is possible, but with the risk to sacrifice some
of the advantages of a more ductile adhesive.
The case of fire has to be looked at differently. The work of Klippel et al. [40]
showed that in a fire case, the correct dimensioning of the wood construction is
much more important than any small differences between the adhesive systems in
high temperature laboratory-scale heat tests.
The main focus for all the work presented in this paper is targeted on the
attainment of the highest possible level of security for load-bearing applications.
Security comes however not only from adhesives with the best performance under
particular situations. At least as important is the use of adhesives, which have no
quality deviation from production batch to production batch, which are easy and
safe to apply and which are capable to perform well even when the production
occurs under not ideal conditions regarding temperature, humidity and wood
preparation.
The gluing of some softwood species with colored heartwoods presents certain
challenges for a gluing according to existing requirements not only 1C-PUR
adhesives struggle with the bonding of larch when delamination tests according to
EN 302-2 are required. A very recent approach on laboratory scale has revealed a
significant improvement of gluing quality when the larch wood surface was
treated with an aqueous pretreatment prior to adhesive application. Industrial trials
are up to follow and could bring a big step ahead for larch gluing with 1C-PUR
adhesives.
Gluing of hardwood species is nowadays being done on basis of requirements
that were originally defined for softwood gluing. Thus, test procedures and most
probably also threshold values should be reconsidered in the future, when
engineered wood products from hardwood species shall enroll their full capacities
and advantages. CEN TC 124 WG 3 has just recently taken the initiative to
develop a European standard for glulam made of hardwood species.
Against the background of the increasing demand for gluing alternative wood
species, it is worth to mention that also in this case a pretreatment of the wood
surface with an aqueous primer briefly before adhesive application can
significantly enhance bondline quality and provide reproducible delamination
results within the normative requirements of EN 15425. Preliminary results from
laboratory testing and industrial trials show a promising trace towards a reliable
and safe gluing of these alternative wood species in the future.
Acknowledgements. The authors express their profound gratitude to Prof. Dr. Klaus
Richter (TU Mnchen) and Michael Klippel (ETH Zrich) who contributed a valuable
scientific input to this article.
References
[1] Hetzer, O.: DRP No. 197773 - Patent for curved, glued laminated beams composed
of two or more lamellas (1906)
[2] Allgemeine bauaufsichtliche Zulassung Z-9.1-606. PUR-Klebstoffe "Purbond HB
110", "Purbond HB 440" und "Purbond HB 530" fr Keilzinkenverbindungen in
Verbindung mit dem Auftragsystem KEBA-Kompakt. Purbond AG, Schweiz
[3] Allgemeine bauaufsichtliche Zulassung Z-9.1-616. PUR-Klebstoffe "PURBOND HB
110", "PURBOND HB 120", "PURBOND HB 440", "PURBOND HB 480" und
"PURBOND HB 530" fr die Herstellung verklebter tragender Holzbauteile.
Purbond AG, Schweiz
[4] DIN EN 1052:2008. Design of timber structures - General rules and rules for
buildings
[5] EN 14080:2008. Timber structures - Glued laminated timber and glued solid timber -
Requirements
[6] Klebstoffliste I. MPA Universitt Stuttgart betreffend geprfter Klebstoffe im
Geltungsbereich der DIN 1052 und allgemeiner bauaufsichtlicher Zulassungen
(Stand: February 28, 2013)
[7] Klebstoffliste II. MPA Universitt Stuttgart betreffend Brettschichtholz nach EN
14080 (Stand: Stand: February 28, 2013)
[8] Limliste. Lim testet og godkjent for produksjon av konstruksjonslimtre og
fingerskjtt konstruksjonslast i Norge, Sverige, Finland og Danmark (State: October
18, 2012)
[9] Pizzi, A., Mittal, K.L.: The Handbook of Adhesive Technology. Dekker, New York
(2003)
[10] EN 15416-2:2006. Adhesives for load bearing timber structures other than phenolic
and aminoplastic - Test methods - Part 2: Static load test of multiple bondline
specimens in compression shear
and aminoplastic Test methods - Part 3: Creep deformation test at cyclic climate
conditions with specimens loaded in bending shear
and aminoplastic Test methods - Part 4: Determination of open assembly time for
one component polyurethane adhesives
and aminoplastic Test methods - Part 5: Determination of conventional pressing
time
[14] EN 15425:2008. Adhesives - One component polyurethane for load bearing timber
structures - Classification and performance requirements
[15] EN 301:2013. Adhesives, phenolic and aminoplastic, for load-bearing timber
structures - Classification and performance requirements
[16] Mller, U., Veigel, S., Follrich, J., Gabriel, J., Gindl, W.: Performance of 1C
polyurethane in comparison to other wood adhesives. In: ICWA 2009 International
Conference on Wood Adhesives, Lake Tahoe, Nevada, USA, Presentation,
September 28-30 (2009)
[17] Konnerth, J., Gindl, W., Mller, U.: Elastic properties of adhesive polymers. I.
Polymer films by means of electronic speckle pattern interferometry. J. Appl. Polym.
Sci. 103, 39363939 (2006)
[18] Konnerth, J., Jger, A., Eberhardsteiner, J., Mller, U., Gindl, W.: Elastic properties
of adhesive polymers. II. Polymer films and bond lines by means of nanoindentation.
J. Appl. Polym. Sci. 102, 12341239 (2006)
[19] Konnerth, J., Stckel, F., Mller, U., Gindl, W.: Elastic properties of adhesive
polymers. III. Adhesive polymer films under dry and wet conditions characterized by
means of nanoindentation. J. Appl. Polym. Sci. 118, 13311334 (2010)
[20] Konnerth, J., Valla, A., Gindl, W., Mller, U.: Measurement of strain distribution in
timber finger joints. Wood Sci. Technol. 40, 631636 (2006)
[21] Mller, U., Veigel, S.: Determination of fracture energy for polyurethane bondlines
at different adhesive spread rates. Wood Kplus Scientific report (2009)
[22] Klusler, O., Clau, S., Lbke, L., Trachsel, J., Niemz, P.: Influence of moisture on
stress-strain behaviour of adhesives used for structural bonding of wood. Int. J.
Adhes. Adhes. 44, 5765 (2013)
[23] Ren, D., Frazier, C.E.: Wood/adhesive interactions and the phase morphology of
moisture-cure polyurethane wood adhesives. Int. J. Adhes. Adhes. 34, 5561 (2012)
[24] EN 302-1:2013. Adhesives for load-bearing timber structures - Test methods - Part 1:
Determination of bond strength in longitudinal tensile shear strength
[25] Hass, P., Wittel, F.K., Niemz, P.: Generic failure mechanisms in adhesive bonds.
Holzforschung 67, 207215 (2013)
[26] Obersriebnig, M., Konnerth, J., Gindl-Altmutter, W.: Evaluating fundamental
position-dependent differences in wood cell wall adhesion using nanoindentation. Int.
J. Adhes. Adhes. 40, 129134 (2013)
[27] Johannes, K., David, H., Lee, S.-H., Rials, T.G., Wolfgang, G.: Adhesive penetration
of wood cell walls investigated by scanning thermal microscopy (SThM).
Holzforschung 62, 91 (2007)
[28] Gindl, W., Schberl, T., Jeronimidis, G.: The interphase in phenolformaldehyde and
polymeric methylene di-phenyl-di-isocyanate glue lines in wood. Int. J. Adhes.
Adhes. 24, 279286 (2004)
[29] Bao, S., Daunch, W.A., Sun, Y., Rinaldi, P.L., Marcinko, J.J., Phanopoulos, C.: Solid
State NMR Studies of Polymeric Diphenylmethane Diisocyanate (PMDI) Derived
Species in Wood. The Journal of Adhesion 71, 377394 (1999)
[30] Weaver, F.W., Owen, N.L.: Isocyanate-Wood Adhesive Bond. Appl. Spectrosc. 49,
171176 (1995)
[31] Kgi, A., Niemz, P., Mandallaz, D.: Influence of moisture content and selected
technological parameters on the adhesion of one-part polyurethane adhesives under
extreme climatical conditions. Holz als Roh - und Werkstoff 64, 261268 (2006)
[32] Sterley, M., Gustafsson, P.J.: Shear fracture characterization of green-glued
polyurethane wood adhesive bonds at various moisture and gluing conditions. Wood
Material Science and Engineering 7, 93100 (2012)
[33] ASTM E119-05a, Standard Test Methods for Fire Tests of Building Construction and
Materials. ASTM, Philadelphia (2006)
[34] CAN/ULC S101-07, Standard Methods of Fire Endurance Tests of Building
Construction and Materials. Underwriters Laboratories of Canada, Scarborough,
Canada (2007)
[35] ASTM D7247-07ae1, Standard Test Method for Evaluating the Shear Strength of
Adhesive Bonds in Laminated Wood Products at Elevated Temperatures. ASTM,
West Conshohocken, PA (2007)
[36] Knig, J., Norn, J., Sterley, M.: Effect of adhesives on finger joint performance in
fire. In: CIB W18: Proceedings, Meeting 41.Saint Andrews (2008)
[37] CSA O112.9-04. Evaluation of Adhesives For Structural Wood Products (exterior
exposure). Canadian Standards Association, Mississauga, ON (2004)
[38] Craft, S.T., Desjardins, R., Richardson, L.R.: Development of small-scale evaluation
methods for wood adhesives at elevated temperatures. In: 10th World Conference on
Timber Engineering 2008, vol. 2, pp. 583590 (2008)
[39] ASTM D4688-99, Standard Test Method for Evaluating Structural Adhesives for
Finger Jointing Lumber. In: Annual Book of ASTM Standards, Philadelphia, vol.
15.06 (2005)
[40] Klippel, M., Frangi, A., Hugi, E.: Experimental Analysis on the Fire Behavior of
Finger-Jointed Timber Member. Journal of Structural Engineering (in print, 2013)
[41] FprEN 16254:2013, Adhesives - Emulsion polymerized isocyanate (EPI) for load-
bearing timber structures - Classification and performance requirements (2013)
[42] Frangi, A., Bertocchi, M., Clau, S., Niemz, P.: Mechanical behaviour of finger
joints at elevated temperatures. Wood Sci. Technol. 46, 793812 (2012)
[43] Frangi, A., Fontana, M., Mischler, A.: Shear behaviour of bond lines in glued
laminated timber beams at high temperatures. Wood Sci. Technol. 38, 119126
(2004)
[44] Clau, S., Dijkstra, D.J., Gabriel, J., Klusler, O., Matner, M., Meckel, W., Niemz,
P.: Influence of the chemical structure of PUR prepolymers on thermal stability. Int.
J. Adhes. Adhes. 31, 513523 (2011)
[45] Clau, S., Dijkstra, D.J., Gabriel, J., Karbach, A., Matner, M., Meckel, W., Niemz,
P.: Influence of the filler material on the thermal stability of one-component
moisture-curing polyurethane adhesives. J. Appl. Polym. Sci. 124, 36413649 (2012)
[46] Richter, K., Pizzi, A., Despres, A.: Thermal stability of structural one-component
polyurethane adhesives for wood-structure-property relationship. J. Appl. Polym.
Sci. 102, 56985707 (2006)
[47] Na, B., Pizzi, A., Delmotte, L., Lu, X.: One-component polyurethane adhesives for
green wood gluing: Structure and temperature-dependent creep. J. Appl. Polym.
Sci. 96, 12311243 (2005)
[48] George, B., Simon, C., Properzi, M., Pizzi, A., George, B., Pizzi, A., Elbez, G.:
Comparative creep characteristics of structural glulam wood adhesives. Holz als Roh
- und Werkstoff 61, 7980 (2003)
[49] Richter, K., Steiger, R.: Thermal stability of wood-wood and Wood-FRP bonding
with polyurethane and epoxy adhesives. Advanced Engineering Materials 7, 419426
(2005)
[50] Clau, S., Gabriel, J., Karbach, A., Matner, M., Niemz, P.: Influence of the adhesive
formulation on the mechanical properties and bonding performance of polyurethane
prepolymers. Holzforschung 65, 835844 (2011)
[51] Frangi, A., Fontana, M., Knobloch, M., Bochicchio, G.: Fire behaviour of
cross-laminated solid timber panels. Fire Safety Science, 12791290 (2008)
[52] Frangi, A., Fontana, M.: Charring rates and temperature profiles of wood sections.
Fire Mater. 27, 91102 (2003)
[53] Frangi, A., Fontana, M., Hugi, E., Jbstl, R.: Experimental analysis of cross-
laminated timber panels in fire. Fire Saf. J. 44, 10781087 (2009)
[54] Craft, S.T., Desjardins, R., Mehaffey, J.R.: Investigation of the behaviour of CLT
panels exposed to fire. Preliminary CLT Fire Resistance Testing Report, Project
No.301006155, FP-Innovations (2012)
[55] Klippel, M., Frangi, A., Fontana, M.: Influence of the Adhesive on the
Load-Carrying Capacity of Glued Laminated Timber Members in Fire. In: 10th
International Symposium on Fire Safety Science (IAFSS 2011) Fire Safety Science:
Proceedings of the Tenth International Symposium, pp. 12191232 (2011)
[56] EN 302-2:2013. Adhesives for load-bearing timber structures - Test methods - Part 2:
Determination of resistance to delamination
[57] Knniger, T., Fischer, A., Bordeanu, N.C., Richter, K.: Water Soluble Larch
Extractive - Impact on 1 P-PUR Wood Bonds. In: 5th International Symposium on
Wood Structure and Properties Arbora Publishers Zvolen, Slovakia, pp. 7176
(2006)
[58] Allgemeine bauaufsichtliche Zulassung Z-9.1-679. BS-Holz aus Buche und BS-Holz
Buche-Hybridtrger (Glulam from beech and glulam from beech-hybrid).
Studiengemeinschaft Holzleimbau e.V., Germany
[59] Allgemeine bauaufsichtliche Zulassung Z-9.1-704. VIGAM Brettschichtholz aus
Eiche (VIGAM glulam from oak wood); Gamiz S.A., Spain
[60] Ohnesorge, D., Henning, M., Becker, G.: Bedeutung von Laubholz bei der
Brettschichtholzherstellung. Holztechnologie 6, 4749 (2009)
[61] EN 392:1995. Glued laminated timber - shear test of glue lines
[62] Frhwald, A., Ressel, J.B., Bernasconi, A.: Abschlussbericht Forschungsprojekt
"Hochwertiges Brettschichtholz aus Buchenholz". Institut fr Holzphysik und
mechanische Technologie des Holzes, Bundesforschungsanstalt fr Forst- und
Holzwirtschaft (heute: Thnen-Institut) (2003)
[63] Phler, E., Klingner, R., Knniger, T.: Beech (Fagus sylvatica L.) Technological
properties, adhesion behaviour and colour stability with and without coatings of the
red heartwood. Ann. For. Sci. 63, 129137 (2006)
[64] Ohnesorge, D., Richter, K., Seeling, U.: Glueability of Beech wood containing red
heartwood. In: Proc. 5th Int. Symp.Wood Structure and Properties, Sliac Sielnica,
Slovakia, pp. 471474 (2006)
[65] Aicher, S., Reinhardt, H.: Delaminierungseigenschaften und Scherfestigkeiten von
verklebten rotkernigen Buchenholzlamellen. Holz als Roh- und Werkstoff 65,
125136 (2007)
[66] Schmidt, M., Glos, P., Wegener, G.: Gluing of European beech wood for load
bearing timber structures. European Journal of Wood and Wood Products 68, 4357
(2010)
[67] Brandmair, A., Clau, S., Hass, P., Niemz, P.: Bonding of hardwoods with 1C PUR
adhesives for timber construction. Bauphysik 34, 210216 (2012)
Part IV
Timber and Concrete/
Cement/Polymer Composites
Development of a High-Performance Hybrid
System Made of Composites and Timber
(High-Tech Timber Beam)
Markus Jahreis, Martin Kstner, Wolfram Hdicke, and Karl Rautenstrauch
Bauhaus University Weimar, Department of Timber and Masonry Engineering

Marienstrae 13A, 99423 Weimar, Germany
markus.jahreis@uni-weimar.de
Abstract. For increased requirements in terms of load-carrying capacity and long

spans, a hybrid composite beam made of glulam and high-performance materials
was developed at the Bauhaus University Weimar in cooperation with a
local SME. The compression zone at the top of the beam is reinforced by an ep-
oxy-based Polymer Concrete (PC). The tension zone at the bottom is strengthened
either with Fiber Reinforced Plastics (FRP) - or steel-reinforced lamellas made of
Laminated Veneer Lumber (LVL). Further reinforcing elements increase the shear
resistance or improve the transverse load-carrying capacity at the bearings. By the
use of PC all parts are surface mounted with a stiff connection.
Keywords: Glulam, reinforced timber, FRP, LVL, rigid bond, polymer concrete.
1 Introduction
Wood in Europe, especially in Germany, is a good available row material. It has a

high load capacity and stiffness in comparison to the low death weight. Timber is
traditionally used for bending beams and columns with low processing. Since it is
a natural material, there are variations in properties and quality. There is a demand
for homogenisation for using it in modern engineering and for the exploitation of
the good properties of timber. Therefore, the timber is sliced, sorted and re-
mounted as glued laminated timber or even as veneer lumber, to eliminate defects,
knots or cracks.
Reinforcement with high-tech materials enables more effective upgrade of load
bearing capacity or larger span the girder, made of glulam. Furthermore, the
bending strength decreases at beams with larger height (> 600 mm) and long span
(effect at higher volume). Especially at combined elements, the load-bearing ca-
pacity of the tension-zone is of particular importance. For optimal exploitation of
the properties it is necessary, that the materials can be glued together and have at
least the same strength and stiffness compared to timber. Therefore, an epoxy
DOI: 10.1007/978-94-007-7811-5_38, RILEM 2014
424 M. Jahreis et al.
based Polymer Concrete (PC) at the compression zone and steel or Fibre
Reinforced Plastics (FRP) at the tension zone have been used to reinforce the
timber beam. The advantage of the combined use of PC, glulam and FRP is the
extraordinary strength at low death weight and the possibility to produce almost
every length. The handling is uncomplicated by good possibilities of prefabrica-
tion and mounting on-site.
The new created High-tech Timber Beam (HTB) has a wooden surface, so the
optical acceptance is high, fire resistance is good and the ongoing mounting
process can be done with usual connectors. By its higher stiffness and strength
constructions with long span width or higher load can be realized with smaller
height and a higher security level. Especially as girder in Timber-Concrete-
Composite bridges the HTB enables an improved competiveness compared with
other established bridge-construction methods.
2 Design and Construction
2.1 Compression and Tension Zone

The main material of the developed High-Tech Timber Beam is still glulam with
more than 90 % of the girders volume. Replacing one or two lamellas at the
bottom side of the glulam by laminated veneer lumber (LVL) allows a significant
homogenization of the timber beams properties. The influence of inconstancies of
the wood, as knots, finger joints and defects, will be reduced and the relation of
strength between compression and tension over the cross section can be optimized
and the load capacity increases in comparison to conventional glulam beams. For
the actual investigations a LVL made of spruce with parallel longitudinal veers
was used (STEICO ultralam RTM). However, the main increase of stiffness
and load bearing capacity was accomplished with the help of high performance
materials.
For upgrading the compression zone of the bending beam, the upper lamella
of the glulam was replaced by a decking of PC. This was made of a mineral
mixture with a special grain-size distribution and a binder on base of two-
component epoxy resin (Bennert COMPONO). The PC has high compression
strength and also high stiffness because of its high rate filling with mineral
grits. The binding resin helps connecting the grid and mounts the PC to the
surface of wood. The properties of the used PC are summarized in Table 1.
New epoxy resins are in development on base of phytogenetic oils with high
strength and durability. By replacement of the petrochemical resins, used in
actual tests, with bio-based resins the HTB can be produced with more than
95 % of renewable products.
Development of a High-Performance Hybrid System Made of Composites 425
Fig. 1 Prototypes of High-Tech Timber Beam
To reinforce the tension zone, four versions have been developed and
optimized with the help of numerical simulation and verified with prototypes in
experimental examinations. The reinforcing materials are Carbon-FRP with a high
stiffness (E = 210 000 MPa) and very high tension strength (ft,k = 2 500 MPa) as
lamellas with a thickness of 1.4 mm and a width of 25 mm or 50 mm (S&P rein-
forcement), reinforcement bars of Glass-FRP (ft,k = 580 MPa, Emena = 60 000 MPa)
with a diameter of 16 mm (Schck Combar) and reinforcement steel
(fy,k = 500 MPa) with a diameter of 16 mm. The Lamellas are mounted to the
surface or into pre cut saw grooves with glue made of the epoxy resin and a filling
of sand and mineral flour. The GFRP and steel bars are casted into trenches in an
additional board of LVL with a mortar comparable to the used PC but with differ-
ent grading curve for more flexibility in handling with smaller molding size.
The types of HTB are:
HTB-1 with four 50 mm CFRP lamellas, flat mounted beneath LVL
HTB-2 with eight 25 mm CFRP lamellas, upright mounted into saw grooves in
a second layer of LVL
HTB-3 with eight reinforcing steel bars, casted in pre-cut trenches in a second
layer of LVL
HTB-4 with five prestressed GFRP bars, casted in pre-cut trenches in a second
layer of LVL
Due to the significant higher strength and stiffness of the used materials
compared with timber, the load capacity increases, the deflection is reduced. Fur-
thermore, the behaviour of collapse is more convenient because the reinforcement
carries the beam after rupture of timber in the tension zone.
2.2 Bearings, Anchorage Area and Shear-Reinforcement

The load capacity of glulam beams can be upgraded by reinforcement of the com-
pression and tension zones. Therefore, a higher grade of load will be enforced in
several details. Bearings and areas for load impact receive higher compression
stress, mainly perpendicular to the grain. There are peaks of shear stress in the
end of the beam and at the anchorage zone of the reinforcement. New details and
special junction elements are developed and tested in connection with the HTB
experiments.
The most important details are the bearings. These consist of a block of PC as
thick as a LVL layer and a length of 18 cm, which is a small size for an eight me-
ter long beam with high load capacity. For better stress distribution the bearings
are completed by additional PC-filled drill holes with optimized angles to the
gradient of strain. This is an effective and comparatively easy way to enforce the
load capacity and reduces the deformation of the timber perpendicular to the grain.
The short length of bearing is expedient for girders with large span wide on elastic
bearings to reduce the angel of rotation.
Fig. 2 Bearing with PC and stress distributers (left); Effect of reinforcement (right)
Additionally, the bearings serve as end anchorage zones for the reinforcements.
By embedding the CFRP-lamellas in the PC-blocks of the bearings at the end of
the beam (HTB-1), the adhesive area of the end-anchorage zone of each lamella is
duplicated. Under service conditions the glued-on CFRP-lamellas work as a
continuously bonded reinforcement. However, due to the extreme difference in
stiffness of the bonded materials stress peaks in the ending area are unavoidable,
so an effective end-anchorage of the lamellas is important.
There is a demand on anchorage bodies for prestressing the GFRP-bars to
afford the link to threaded bars as connection to external frames and hydraulic
jacks. After prestressing, the GFRP-bars are embedded in a special PC and the
anchorage bodies casted into the bearings. The reaction force helps to counteract
the high stress peaks at the end of reinforcement by inserting the reinforcement
into the bearings.
To counter the shear stress in the beam's end-sections, additional to the special
bearings with finger-shaped extensions towards the load, glued-in steel rods are
added. The mounting angle is 45 to the grain. During the bending tests the strains
on the surface of the reinforced areas was measured with CRP (Fig. 3, left side).
The strengthening effect was derived by a comparison with the results of an
FE-simulation without shear- and bearing-reinforcements (Fig. 3, right side).
Fig. 3. Strain distribution as test result of reinforced HTB (left) and as simulation without
shear and bearing reinforcement (right)
3.1 Investigation of PC Compression and Tension

An epoxy based resin with several fillings of minerals is the main material for the
reinforcement works. The two component resin reveals a high adhesive potential
for many materials and especially for wood, minerals, glass and steel. Hence, it is
highly convenient to provide rigid bonds between those materials. Furthermore,
the shrinkage of these resigns during the chemical reaction is low. Thus, it is
applicable in use for grouting or to glue rods into holes. In these investigations a
pure epoxy-system was used with different mineral fillings from rock flour to
gravel with a special grain-size curve to 7 mm. For diverse applications at the
HTB special recipes were created. To design the reinforcement and evaluate the
dimensions, placements and dimensions as well as to create a realistic numeric
model, it is necessary to be acquainted with the material properties of the compo-
nents. Therefore, advanced investigations where done on the different kinds
of polymer mortar or concrete. Tests for determine the properties of PC were done
as well as several tests for bounding behaviour towards wood, steel, CFRP or
GFRP.
Additional to traditional measurement equipment, contact free measurement
by industrial closer-range photogrammetry (CRP) was used. The CRP-technique
enables the measuring of the progression of deformations, cracks and deteriorations
during loading and unloading of specimen by computer controlled high resolution
cameras. Thus, it is possible to setup and calibrate a numeric simulation model.
Subsequently the test, the strain distribution, on base of the images, is implemented
in a finite elements model and complimented with a constitutive approach.
In different test series the properties of PC were investigated. The deformation,
for instance under increasing compression and tension load, was recorded by CRP
to determine the Modulus of Elasticity and the behaviour of strain in longitudinal
and rectangular direction. Additional tests under displacement controlled cyclic
compression force are done to determine the deterioration. After the point
of maximum load capacity was exceeded, the degree of deterioration were
determined due to the hysteretic loss under cyclic loading, subsequent.
Fig. 4 Test serial of three specimen, two Fig. 5 Curve of deterioration as logarithmic
specimen under cyclic loading function
Experiments to investigate the shear properties of a material or a connection are

difficult to realize. All known tests include secondary effects like transversal
strain. Those have an impact to the result, but they are difficult to define exactly.
A new test setting was designed to reduce and monitor the side action. It enables
to impact the force as shear stress directly as possible. Therefore, a disc shaped
specimen was chosen and fitted into gripping jaws with finger joints and low
friction support for parallel shift. The specimen has a reduced thickness in the area
of interest. The path of crack could be controlled with the width and depth of an
upper and lower groove. Simulations with FEM helped to design the specimen and
the experimental set-up.
Fig. 6 Shear test: Specimen with test setting (left); first crack (centre); final crack (right)
All specimens revealed almost the same fracture appearance with two small
cracks starting from the arc at the end of groove developing diagonal towards the
outer edge. This was accompanied by sudden but little decline of reaction force.
Under further vertical shift the load increased with little less stiffness. The final
failure divided the specimen, while the crack, starting from the opposite side of
groove to the first crack, runs diagonal opposite to the other groove.
The next table displays the material properties for polymer compound mortar or
concrete, used for the HTB for the different applications. Composition PC_7-7
was used to reinforce the compression zone at HTB in the described tests. In
ongoing investigations a new mixture of PC (PC_3-8) with significant higher
strength and stiffness due to stronger filling materials and a higher filling degree
is tested and optimized for further additional applications in timber and hybrid
constructions, respectively.
Table 1 Material Properties of Polymer Concrete
Polymer Composite PC_1-2 PC_3-4 PC_3-5 PC_7-7 PC_3-8

Application Glue Grouting Bearings Compres- In Test
sion Zone
Max grain size Sand / Flour 3 mm 3 mm 7 mm 3 mm
Resin : Aggregate 1 : 1.75 1:4 1:5 1:7 1 : 8.4
Compression Strength [MPa] 101.5 97.3 96.9 90.3 142
Tension Strength [MPa] 28.3 18.8 19.2 16.1 35.9
Shear Strength [MPa] 3.54 3.37 2.65 2.45
MOE (compression) [MPa] 10 200 16 800 20 700 45 000
Radical strain coefficient (under 0.30 0.26 0.24 0.24
compression)
Bulk density [g/cm] 1.68 1.87 1.99 2.03 2.54
3.2 Bending Beams

Experimental investigations with full-size compounded elements confirmed the
principal of heavy duty structural elements of timber with comparable slight rein-
forcement by high performance materials. Due to the direct bond, the synergetic
effects between the materials can be used to create a High-Tech Beam based on
block bonded glulam. Short time bending tests at the four prototypes were ar-
ranged to determine the behaviour of load bearing, deformation performance and
the failure of the hybrid beams. The dimensions of the specimen were
W x H x L = 40 x 60 x 800 cm. Thus, the span width was 7.80 m. The tests were
carried out as 6-point-bending tests partly according to DIN EN 408. The 6-point
load set-up approximates the distribution of shear forces and bending moments
like under linear load in usual service. During the tests the load, the deflection, the
deformation and the strain at several points of the beam and at every component
were recorded. Areas of advanced interest were monitored by 3D-CRP additional

to the traditional measurement with differential transformers (LVDT) and strain
gauge (Fig. 7).
Fig. 7 Six-point bending test of an eight meter long specimen
The glulam beams with strength class GL24h were examined in deflection tests
to determine the Youngs Modulus before reinforcing. The result was a mean
value of 12 000 MPa for deflection stiffness. The properties of the other materials
are known from product guidelines (steel, GFRP, CFRP) or have been determined
in material-tests, like tension tests for LVL and different test series for the PCs.
The test results of the HTBs were compared with theoretical values for stiff-
ness and load capacity of a normal glulam beam with the same dimension (cross
section of 40 x 60 cm) to characterise the reinforcement effects. The next figure
(Fig. 8) shows the load-deflection curve of the four variations of HTB up to the
maximum load. FE-Simulation revealed curves in good accordance to the experi-
mental results. The graphs are almost in a linear slope until the break down. There
is one curve added of a unreinforced glulam beam made of GL24
(MOE = 12 000 MPa, MOT = 650 MPa) with the characteristic value of bending
strength (fm,k = 24 MPa) and extended to a realistic value from several tests and
literature of fm,mean = 36 MPa.
Fig. 8 Load-deflection curve for HTB-1 to HTB-4 and calculated behaviour of

GL24h-glulam beam and HTB-3 with PC_3-8
In reference to the unreinforced glulam beam (100%) an enhancement of stiff-

ness, compared at the middle deflection, was denoted to about 130 % (HTB 1, -2
and -4) and 144 % (HTB 3), respectively. That means the beam can carry 44 %
more load at the same deflection. After substitution of PC_7-7 with PC_3-8 it is
possible to increase the stiffness for further 15 % to 25 % (FE-Simulation)
depending on the HTB version. The bending strength of timber is assumed with
fm = 36 MPa to compare the load capacity. Thus, the increase can be declared in
a realistic dimension of 122 % to 155 %. Further information are displayed in
Table 2.
Table 2 Increase of Load bearing Capacity of HTB compared to unreinforced Glulam
Specimen HTB Prototype HTB-1 HTB-2 HTB-3 HTB-4

Maximal Loading max Ffail [kN] 1 077,6 1 297,4 1 373,4 1 199,9
Increase-factor of Load Capacity 1.82 2.20 2.32 2.03
referenced to fm,k = 24 MPa
Increase-factor of Load Capacity 1.22 1.46 1.55 1.35
referenced to fm,k = 36 MPa
All beams broke down without noteworthy pre-announcement. The tension re-
sistance of the timber failed at HTB-1, HTB-2 and HTB-4. However, there was no
complete collapse. Due to the bearings were still preserved, allowing the rein-
forcement working like a belt beneath the tension zone. The remaining load bear-
ing capacity after bending failure was up to 73 % of maximum load at HTB-4. The
highest stiffness was measured at HTB-3, due to the high rate of steel reinforce-
ment with high stiffness in tension. Therefore, higher shear forces occurred and
the beam finally failed in shear of the timber. The shear crack was located near the
joint between glulam and LVL with the embedded reinforcement bars. After this
deterioration the effect of tension zone reinforcement was dissolved and the glu-
lam failed by flexural tension. The consequence was a remaining load at only 8 %
of the measured breaking force. However, this HTB-3 revealed the highest load
capacity of 1 374 kN.
A further simulation was done with the material properties of later tested
PC_3-8 with high MOE. In order to the version of HTB, the stiffness of the beams
can be increased with 15 % to 25 %, if this PC is used at the compression zone.
4 Conclusion
The investigations on hybrid timber beams show good results regarding the
enhanced load capacity and stiffness at reduced construction height by reinforce-
ments with modern high-performance materials. Further studies on important
details with high loads are necessary.
Acknowledgments. Thanks are going to the Bennert GmbH as cooperation partner at the inves-
tigation project and the European Union (EFRE) as well as DFG for co-funding.
Experimental Study of the Composite Timber-
Concrete SBB Connection under Monotonic
and Reversed-Cyclic Loadings
Manuel Manthey1,2,*, Quang Huy Nguyen1, Hugues Somja1,

Jrme Duchne2, and Mohammed Hjiaj1
1
Laboratory of Structural Mechanics INSA, Rennes, France
Manuel.Manthey@insa-rennes.fr, m.manthey@a-i-a.fr
2
AIA Ingnierie, Angers, France
Abstract. The present paper investigates the mechanical behavior of a novel dow-
el-type Timber-Concrete Composite system, namely SBB, under monotonic and
reversed-cyclic loadings. This system consists of timber beams connected to a
concrete slab using a dowel type connection. Because the structural behavior of
timber-concrete composite slabs is mainly governed by the shear connection
between concrete and timber, an extensive experimental program was carried out
in order to assess its behavior under both monotonic (serviceability design) and
cyclic loadings (seismic design) and to identify failure modes for each possible
configuration. In order to fully characterize the load-slip behavior of the SBB
connection under both monotonic and reversed cyclic loading, 24 Push-Out tests
(12 under monotonic loading, 12 under reversed cyclic loading) were performed.
The experimental program and the results (parameters and phenomenology) are
discussed in this paper.
Keywords: Composite structures, Timber, Concrete, Push-Out tests, SBB shear

connector, cyclic loading.
1 Introduction
1.1 Timber Concrete Composite Structures

Timber-Concrete Composite systems are competitive technical solution in
refurbishment as well as in new building construction. With this system, the best
properties of timber and concrete materials can be exploited since tensile stresses
induced by gravity loads are resisted primarily by the timber beam and compres-
sion by the concrete slab. Timber-Concrete Composite (=TCC) Structures offers
*
DOI: 10.1007/978-94-007-7811-5_39, RILEM 2014
434 M. Manthey et al.
many advantages over traditional floors. Yeoh [14] presents a non-exhaustive list
of these advantages. Currently, there are no standards for the design of Timber-
Concrete Composite structure. Nevertheless, design methods have been proposed
in the technical literature [3, 6, 7, 8, 14]. The design of TCC must account for two
main phenomena: the partial composite action resulting from the flexibility of the
shear connection and the time dependant properties of the component material
(creep, concrete shrinkage, mechano-sorption, thermal strains and hygroscopic
strains).
1.2 SBB Timber-Concrete Connection System

Various connectors are available in the market with a wide range of stiffnesses
and load capacities which are crucial design parameters for TCC and empirically
determined by Push-Out tests. Ceccotti [2] summarizes the most commonly used
methods for joining concrete to timber. Dias [4] describes the typical load-slip
behavior for different types of joints. The dowel type fasteners used in his work
(see Figure 1) are dowels with a 10mm diameter, so it cannot be directly com-
pared to SBB shear-connectors which have diameters from 21 to 26 mm. The
joints made with dowel type fasteners have smaller strengths and stiffness but
much higher plastic deformation capacity.
Fig. 1 Position of the SBB Load-slip curves compare to the typical behavior of different
type of joints (adapt from Dias, 2005, Figure 2-9)
SBB Timber-Concrete System (see Figure 2) has been developed by the French
company AIA Ingnierie for the last decade and now is widely used in the coun-
try. The SBB system consists of timber beams connected to a concrete slab using a
large diameter dowel type connection patented at the French patent bureau. Two
SBB configurations will be investigated in this paper SBB 26-170 (170 mm
length) and SBB 26-250 (250 mm length).
Experimental Study of the Composite Timber-Concrete SBB Connection 435
Fig. 2 SBB Timber-Concrete System
For seismic design purposes, AIA Ingnierie decided to fully characterize the
load-slip behavior of the SBB connection under reversed-cyclic loading. Under
seismic loads, the relative stiffness of the diaphragm and its connectors has impli-
cations on the seismic response of a building and affects how the floor system can
be modeled. Studies at the University of Canterbury [10] showed that for TCC
floor unit under diaphragm action, the diaphragm displacement was less than 5%
of the total displacement, so TCC can be modeled as a rigid unit. Eurocode 8 [3]
requires that the concrete slab depth is at least equal to 7 cm to consider that the
slab act as a rigid unit. TCC floors with SBB system are built with a minimum
concrete slab depth of 7 cm in seismic area. Others considerations from Eurocode
8 [3] concerning floors dimensions should be respected. Under seismic action, all
the TCC floor (concrete slab + timber beam) moves as a rigid unit, so only the
connectors between TCC floor and LLRS are subjected to reversed cyclic action.
2 Experimental Set-up and Loadings
2.1 Experimental Set-up

As the aim of this study is to analyze the mechanical behavior of timber-concrete
connections, a particular attention has to be paid to the parameters that may affect
it, specifically in the experimental shear test set-up. In their paper, Monteiro et al.
[9] identified different shear test set-ups with a database based on a literature
survey. The database showed that three types of experimental test set-ups exist for
Timber-Concrete Connections: Double Shear tests, Asymmetric-Shear tests and
Pure Shear tests. Others SBB configurations were already tested in double shear
Push-Out test in the past. In 2011, complementary Push-Out tests were performed
with Glue Laminated Timber Beam (GL24h) and two SBB dowel dimensions
(26-170 and 26-250). As SBB 26-250 do have a 250 mm dowel length without the
head (compare to the 170 mm for SBB 26-170), and considering the 70 mm
thickness of the concrete flange, a 60 mm concrete render was necessary for the
SBB 26-250 tests. This concrete render was deliberately not reinforced with steel.
Concerning the concrete, minimal class on site is C25/30. According to Eurocode
4 (Annex B Clause 2.3.), Push Out tests need to be performed with a reduced
concrete class, which in the present case is C16/20. Concreting of slab 2 was done
two days after concreting of slab 1. A plastic foil was applied between timber and
concrete in order to prevent the bleeding of concrete in the timber member during
the concreting phase. These layers also allowed minimizing the friction forces
between timber and concrete during the shear test. Concerning measures the three
points of interests were: a) the relative slip between the concrete slab and the
timber member measured near the connectors, in load direction; b) the separation
between the timber beam and the concrete slab; c) the load applied on the speci-
men. Those measurements allowed calculating the relevant mechanical properties
of the connection such as slip moduli and shear strengths. Concerning boundaries
conditions (see Figure 3), timber beam was clamped thanks to a thick steel plate
placed on the top of it and restrained with four pre-stressed steel links. In order to
apply cyclic loading, a steel device was designed for the campaign. This device
allowed achieving a reversed-cyclic vertical loading directly on the concrete
flanges.
Fig. 3 a) Test set up and b) Views of the designed Push-Out Device
2.2 Loading
As unfortunately, the design of TCC is not addressed by standards, a combination
of timber standards and composite steel-concrete structure was used to determine
the loading program. Monotonic Push-Out tests were carried out according to
European Standard EN 26891 [12], and Eurocode 4 [3] (Annex B). Considering
Reversed-cyclic loadings, two references were used: the European Standards EN
12512 [11] and the ECCS n45 [5]. Both references offer the possibility to define
the reversed-cyclic displacement path according to y, with y being the conven-
tional limit of elastic range measured in preliminary classical monotonic tests.
Many procedures exist to deduce the limit of elastic range from a load-slip curve;
the one chosen is discussed later in this paper. The specimens considered here do
have the same behavior in compression and tension, if this is not the case, two
limits of elastic range should be investigated, one in compression and one in
tension [5]. The reversed-cyclic load program has been developed according to
EN 12512 [11]. Nevertheless some modifications were made: (1) three cycles
were made at the +/-0,25.y step and at the +/-0,5.y step instead of one, (2) addi-
tional intermediate steps were added, basically one intermediate step between each
steps of the standard, for example a +/-0.375. y step was added between the +/-
0,25.y step and the +/-0,5.y step, a +/-3.y step was added between the +/-2.y
step and the +/-4.y step (see Figure 4). The cyclic loadings were carried out till
failure of the system (in tensile and compression). The tests were performed under
load control during the elastic part and then under displacement control for the
plastic range.
y : Limit Slip
Fig. 4 Reversed-cyclic loading defined for the SBB Push-Out tests
3 Results for the Monotonic Push-Out Tests
From the values measured during the monotonic Push-Out tests, particular empha-
sis is put on the ultimate load (Pmax) and on the slip moduli (K0,4 and K0,6) because
they are the parameters usually considered to design a Timber Concrete Compo-
site structure. Limit slip, ultimate slip and static ductility ratio are also quantified.
Strength is quantified as the maximum load applied when failure occurs in the
Push-Out specimen. However designers need to know the design shear strength,
PRd, which is obtained as a 5% fractile value of the maximum load according to
Eurocode 0 [3], (Annex D-clause D.7.2) and then divided by 1,25 as recommended
in the Eurocode 5-2 [3], (clause 2.4.1.-[3]). Stiffness is quantified by the slip
modulus at two different load levels: 40% and 60% of the mean maximum
load corresponding to the service and ultimate load levels, as recommended by
Ceccotti [2]. Limit slip y and yield load Py are values of the slip and the load
corresponding to the transition from elastic behavior to plastic range. To deter-
mine these values, two methods were used, the one from the EN 12512 [11] and
the one proposed by Lachal and Aribert [1] for steel concrete shear connector.
As both methods showed similar results, for normative consideration, the EN

12512 method was adopted. According methods previously described, shear con-
nector properties were determined for monotonic tests series, the one with SBB
26-170 Shear Connectors in Glue Laminated Timber and the one with SBB 26-
250 Shear Connectors in Glue Laminated Timber. Key parameters are listed in
Table 1. It is worth to mention that Push-Out tests with the SBB 26-250 were built
deliberately with unreinforced concrete render. As most of the degradation during
the test was observed on the concrete render, it can be assumed that with a rein-
forced concrete render, higher strength and stiffness should be achieved. The
Push-Out tests with the 26-250 SBB Shear connectors are showing a lower stiff-
ness than the Push-Out tests with the 26-170 SBB Shear connectors. This can be
explained by the lower flexibility of shorter connector.
Table 1 Monotonic Push-Out tests, key parameters. (*: EN 12512 limit)
SBB 26-170 SBB 26-250

Maximum load applied per connector Pmax [daN] 3075 4676
Design shear strength PRd [daN] 2460 3741
Stiffness K0,4 (service load level) [daN/mm] 3516 1942
Stiffness K0,6 (ultimate load level) [daN/mm] 2321 1282
Ultimate slip u [mm] 30* 30*
Limit slip y [mm] 1,32 2,17
Ductility ratio u/y 22,73 13,83
Having analyzed the test results, concerning monotonic loading, SBB connec-
tion exhibits an excellent ductile behavior (see Figure 5). Even at a 40 mm slip the
system is showing high remaining shear strength. Moreover stiffness and
shearstrength were experimentally characterized for the SBB, providing all pa-
rameters needed for TCC floors design in serviceability uses. Timber-Concrete
separation measured during the tests was not significant, less than 1 mm before
failure and about 8 mm at the failure. They were monitored during the Push-Out
tests to control the relative out-of-plane movement of the timber-concrete compo-
site structures. Figure 5 summarizes the phenomenological observations from the
Push-Out tests under monotonic loading with SBB 26-170 and SBB 26-250. Until
Py, the shear connector is considered to behave elastically. After Py, the shear con-
nector behavior becomes plastic with a positive hardening. Plastic hinge appears
on Shear Connector in timber part only for SBB 26-170 and in timber and con-
crete parts for SBB 26-250.Once maximal shear resistance is reached, concrete
degradation is clearly observed.
Fig. 5 Load-slip curves and Phenomenology for the SBB 26-170 Push-Out tests and the
SBB 26-250
4 Results Reversed-Cyclic Push-Out Tests
In reversed-cyclic Push-Out tests, main features highlighted are yield point of

the connection, maximal strength and ultimate displacement as well as ductili-
ty. A particular attention is paid to the strength degradation after 3 load-cycles
at the same displacement level, the so-called Action Reduction Factors. From
the Push-Out tests, load-slip curves showing hysteresis curves are obtained (see
Figure 6). Properties are determined following EN 12512 (see [3]). Analysis
should be done on the backbones curves of the reversed-cyclic tests. Parame-
ters should be quantified in tension and compression to check the symmetrical
behavior of the system. Maximum strength is quantified as the maximum load
applied during the Push-Out test. According to EN 12512, ultimate slip u is
defined as the lowest displacement value between: a) Displacement at failure;
b) Displacement at 80 % of the maximal strength, post peak; c) 30mm. Ulti-
mate strength, Pu, is the corresponding strength to the ultimate displacement.
Limit slip y and yield load Py define the transition zone between elastic
and plastic behavior. To determine those, EN 12512 method [11] was used.
Action Reduction Factor was evaluated for both tests series, in compression
and tension for two displacement levels: 4.y and 6.y. Indeed, according
Eurocode 8-1 [3], Clause 8.3, Action Reduction Factor shouldnt exceed 20%
within 3 cycles at the same displacement level. When this condition is satisfied
for a ductility ratio from 4, the connection system can be used as a dissipative
element in a DCM structure and when this condition is satisfied for a ductility
ratio from 6, the connection system can be used as a dissipative element in a
DCH structure. Under reversed-cyclic loading, the SBB connection shows
a ductile behavior without a significant reduction in the shear strength with

large displacements values. According to the Action Reduction Factor and to
the Eurocode 8 [3], connection with SBB 26-170 could be used as dissipative
element in DCH structures. Due to the unreinforced concrete render and to the
large length of SBB 26-250, an important degradation of the connection with
SBB 26-250 was observed so that SBB 26-250 could be only used as dissipa-
tive element in DCM structures. It is important to notice that Timber-Concrete
Diaphragm are primary element in seismic design and are considered as non
dissipative element in a structure under seismic solicitations. The high ductility
of the connection under cyclic loading is still a comforting fact even though it
is not a mandatory point for TCC floors diaphragm design in seismic area. SBB
Shear Connectors do have a dissipative role when connecting the floor dia-
phragm to the lateral resisting system. If SBB shear connectors have a dissipa-
tive role, as for example connecting the floor diaphragm to the lateral resisting
system, accidental cyclic design shear strength (=PRd,ACC) should be used.
Cyclic design strength can be defined as a 5% fractile value of the ultimate load
Pu (=80% Pmax) applied on the specimen under reversed-cyclic loading, accord-
ing to Eurocode 0 [3] (Annex D-Clause D.7.2), and then divided by 1,0
considering accidental solicitations. As well as monotonic Push-Out tests, Tim-
ber-Concrete separation measured during the tests was not significant. They
were monitored during the Push-Out tests to control the relative out-of-plane
movement of the timber-concrete composite structures. The Equivalent Viscous
Damping Ratio (EVDR) as specified in CEN-EN 12512 [11] (2000) is used
to compare the energy dissipation capacity between different types of
connections. This non-dimensional parameter expresses the hysteresis damping
properties of the connection. It is determined as the ratio between the dissipated
energy in one half cycle and the work performed by the applied force. In the
elastic stage, SBB 26-170 show EVDR values twice as high as the EVDR
values obtained with SBB 26-250. In the plastic stage, both connections show
similar EVDR values. Those EVDR values indicate a quite high energy dissipa-
tion capacity of the SBB connectors with EVDR values from 5% to 12% in the
elastic stage and, for both test series, an increased value for plastic defor-
mations with EVDR values of 17% and 22% for SBB 26-170 and SBB 26-250
connections, respectively. Initial behavior of the shear connector is elastic.
Nevertheless, one cycle after another, reversed-cyclic loading gradually dam-
age timber and concrete material near the SBB shear connector. Irreversible
deteriorations occur, in timber, the hole is ovalised and in concrete, cracks
appear. Those deformations depict the particular shape of the load-slip curves.
On Figure 7, three consecutive hysteresis loops at an advanced loading stage
can be seen, on which critical points from the hysteretic response are represent-
ed. Concrete and wood have already been damaged earlier, thats why no loop
starts from the origin.
SBB SBB
26-170 26-250
Py [daN] 2742 2774
y [mm] 0,70 1,01
Pu [daN] 3095 2964
u [mm] 7,80 9,00
Ductility u/ y 11,81 9,29
Ductility class DCH DCM
Fig. 6 Load-slip curve for one of the six reversed cyclic loading with the SBB 26-170 connector
and comparison to the monotonic loading tests. Table 4.1: Principle parameters for the
reversed-cyclic loading (mean values).
Fig. 7 Load-slip curve for a 3 load-cycles at the same displacement level; Phenomenology for
SBB26-170 and SBB26-250 Shear connector under reversed-cyclic loading
5 Conclusion
Concerning Push-Out under monotonic loading, having analyzed the test results,
SBB connection shows an excellent ductile behavior with all SBB References
tested. Stiffness and shear strength were experimentally characterized for the
SBB configurations, providing all parameters needed for TCC floors design in
normal uses.
For Push-Out tests under reversed-cyclic loading, the SBB connection with
both SBB 26-170 and SBB 26-250 shows a ductile behavior without a significant
reduction in the shear strength with high displacements values. Concerning EVDR
(Equivalent Viscous Damping Ratio), EVDR values from 5% to 12% are observed
in the elastic stage. For both test series, the EVDR values increase for plastic de-
formations with EVDR values of 17% and 22% for SBB 26-170 and SBB 26-250
connections, respectively. According to the ARF (Action Reduction Factor) and to

the Eurocode 8, connection with SBB 26-170 can be used as dissipative element
in DCH structures and SBB 26-250 can be used as dissipative element in
DCM structures. The high ductility of the connection under cyclic loading is a
comforting fact, allowing SBB system use in seismic area without any beam-slab
connection brittle failure risk.
Acknowledgments. AIA Ingnierie is acknowledged for providing funds for the realization
of the test program. A special word of thanks to D. Cvetkovic, C. Garand and F. Marie of
the INSA Rennes for assisting in all stages of the tests.
References
[1] Aribert, J.-M., Lachal, A.: Formulation de la rupture par fatigue de connecteurs
acier-bton pour des sollicitations de type sismique, revue. Construction Mtallique 4
(2002)
[2] Ceccotti, A.: Composite concrete-timber structures. Structural Engineering Materi-
al 4, 264275 (2002)
[3] CEN Eurocode 0; 4;5; 8, European Committee for Standardization, Brussels
[4] Dias, A.: Thesis: Mechanical Behaviour of timber-concrete joints, Technische
Universiteit Delft, Universidade de Coimbra (2005)
[5] ECCS: European Convention for Constructional Steelwork, n45 Recommended
Testing Procedure for Assessing the Behavior of Structural Steel Elements under
Cyclic Loads (1986)
[6] Fragiacomo, M., Yeoh, D.: The Design of a Semi-Prefabricated LVL-Concrete Com-
posite Floor. Hindawi Publishing Coroporation, Advances in Civil Engineering
(2012)
[7] Girhammar, U.A.: A Simplified analysis method for composite beams with interlayer
slip. International Journal of Mechanical Sciences 51, 515530 (2009)
[8] Lukaszewska, E.: Thesis: Development of Prefabricated Timber-Concrete Composite
Floors. Lule University of Technology (2009)
[9] Monteiro, Dias, Negrao: Assessment of Timber-Concrete Connections Made with
Glued Notches: Test Set-Up and Numerical Modelling, Society for Experimental
Mechanics (2011)
[10] Newcombe, M.P., Carradine, D., Pampanin, S., Buchanan, A.H., Deam, B.L.,
Van Beerschoten, W.A., Fragiacomo, M.: In-Plane Experimental Testing of Timber-
Concrete Composite Floor Diaphragms. In: NZSEE Conference 2009 (2009)
[11] NF EN 12512:2002: Timer structures Tests methods Cyclic testing of joints made
with mechanical fasteners, CEN Brussels (2002)
[12] NF EN 26891:2000: Timber structures Joints made with mechanical fasteners
General principles for the determination of strength and deformation characteristics,
CEN Brussels (2000)
[13] SBB SAS (2000) SBB website, http://bois-beton.fr (accessed March 13,
2013)
[14] Yeoh, D.: Thesis: Behaviour and Design of Timber-Concrete Composite Floor
System. University of Canterbury (2010)
The Predictive Model for Stiffness of
Inclined Screws as Shear Connection in
Timber-Concrete Composite Floor
F. Moshiri, R. Shrestha, and K. Crews
Centre for Built Infrastructure, University of Technology Sydney, Australia

farzad.moshiri@uts.edu.au
Abstract. Interest in timber-concrete composite (TCC) floors has increased over

the last 30 years. TCC technology relies on timber and concrete members acting
compositely together. Both timber and concrete exhibit a brittle behaviour in
bending/tension and compression respectively whilst the shear connection is iden-
tified as the only contributor of ductile behaviour. Therefore, the strength, stiffness
and arrangement of the shear connection play a crucial role in the design parame-
ters of TCC system including deflection and stiffness of floor. Hence, calculation
of stiffness is of interest to study the structural performance of TCC floor. Materi-
al properties of timber, fastener and concrete influence the overall load-
displacement response of shear connection.
There are only few investigations on analytical closed-form equation to predict
the stiffness and strength of TCC joints as input values to design a partially com-
posite floor. For example, Eurocode 5 recommends the empirical equations for the
slip modulus of dowels and screws which are limited to vertically inserted fasten-
ers only. Eurocode 5 only recommends that the strength and stiffness of uncon-
ventional joints should be determined by push-out tests. Previous investigations
reported that the inclination of a fastener significantly increase the initial stiffness
and ultimate strength of the TCC joints and consequently composite floor.
This paper presents a model for prediction of the stiffness of TCC joint using
crossed inclined proprietary screws (SFS Intec). The model assumes the behaviour
of inclined screw as a beam on a two-dimensional elastic foundation, and consid-
ers the timber as the elastic foundation consisting of orthogonal springs with
differing stiffness in the parallel and transverse to the grain directions. The exper-
imental aspect of the research consists of embedding and push-out tests aiming to
verify the stiffness model of TCC joints with inclined screws. The model is rea-
sonably accurate in predicting the characteristic stiffness. This research suggests
the model to facilitate the design of inclined screw shear connections for TCC
construction.
Keywords: timber concrete composite floor (TCC), analytical model, stiffness,

slip modulus, shears connection, push out test, (SFS Intec).
DOI: 10.1007/978-94-007-7811-5_40, RILEM 2014
444 F. Moshiri, R. Shrestha, and K. Crews
1 Introduction
Timber concrete composite (TCC) consists of timber, concrete and shear connect-
or such that timber acts in tension due to bending whilst concrete resists the com-
pression caused by bending [1]. The degree of stiffness for the composite system
is dependent on the stiffness within linear range of response of shear connections
which is aligned with the materials natural properties and behaviours [2, 3]. TCCs
benefit from the composite action, resulting in higher strength and stiffness com-
pared to the timber and concrete acting independently [3]. The literature review
indicates that the inclined fasteners increase stiffness and strength rather than the
vertically inserted fastener. The slip between timber and concrete at TCC joint and
consequently the stiffness of laterally loaded fastener has a significant influence
on the overall behaviour of composite beams including displacement, stiffness,
load carrying capacity and distribution of internal forces. Non-linear load-slip
response of a single connection and its stiffness within the linear range of response
(slip modulus) characterises the overall behaviour of composite beams. The de-
sign of TCC structure is governed by displacement within serviceability limit state
(SLS). Hence, the slip at interface of timber and concrete plays a crucial role on
design of TCCs. Moreover, the linear-elastic approach for design of timber com-
posite proposed by Eurocode5 Gamma method employs the slip modulus of
shear connections to design the TCC beams where inaccurate slip modulus results
in unreliable design for the timber composite beams [4, 5].
This paper present a model based on the theory of beam on elastic foundation
to predict the non-linear slip modulus of TCC joint connected by crossed proprie-
tary screws. The experimental aspect of the research consists of embedding and
push-out tests of shear connections and aims to verify the stiffness model of TCC
joints with crossed (30,45, 60) proprietary screws (SFS Intec).
2 Analytical Models
Herein, for TCC joint, the timber and concrete components are assumed to be a
Wrinkler foundation where elements of continuous beam displaces vertically per-
pendicular to the axis of fastener parallel to compressive force. The stiffness and
deformation of fastener in surrounding concrete and timber can be idealized by a
continuous beam which is attached to the springs over the entire length of the
fastener. Assuming the stiffness of attached springs to be linear elastic, Kuenzi [6]
proposed Equation (1) to predict slip modulus of timber joints whilst Foschi
[7] extended the stiffness of spring to non-linear range assuming an ideal elastic
plastic steel for fastener and smooth elasticplastic for nail. A large number of
shear connections including bolts and screws indicate non-linear responses even
during early stages of loading hence, the non-linear assumption seems to be more
applicable [8, 9].
The Predictive Model for Stiffness of Inclined Screws as Shear Connection 445
d4y (1)
EI +ky=0
dx 4
Where, E is Youngs modulus of vertically inserted screw material and I is se-
cond moment of area of screw cross-section. The reaction forces act perpendicular
to fastener axis and opposing the deflection of the fastener. Thus, there is com-
pression in the surrounding timber and concrete in front of fastener. These com-
pressive forces that are distributed along the fastener are proportional at every
point to the deflection of the fastener at this point multiplied by the constant k
(foundation modulus expressed in kN/mm2) which is calculated in compliance
with different codes such as [10, 11].
Symons. D, Persaud. R [12] extends the model for vertically inserted fastener to
the inclined fastener assuming a two-dimensional elastic foundation by additional
springs resisting vertical translation of screws which is perpendicular to the grain
direction as shown in Fig.1. The timber is considered as the elastic foundation
consisting of orthogonal springs with different stiffness in the parallel and trans-
verse to the grain directions. The screw is inclined at an angle of to the vertical
direction while the grain direction is horizontal. The timber is assumed to be elas-
tic material whilst rigid concrete clamping the fastener at the interface of timber
and concrete. Hence the portion of screw embedded in timber is only considered
for the model. The shear force at interface results in a horizontal slip at interface
of composite materials.
Fig. 1 Vertical and Inclined screws in the timber with rigid behaviour of concrete
Fig.2 shows part of a screw of length of dx from whole length of t embedded in

timber whilst y and represent the transverse and axial displacements of the part
in the timber. The vertical and horizontal springs simulate the foundation moduli
of timber transverse and parallel to the grain direction impeding the vertical and
horizontal displacements of screw. The foundation moduli of timber parallel and
Fig. 2 Element of screw embedded in timber
transverse to the grain kp and kt are accounted for stiffness of the spring per unit
length in vertical and horizontal directions. Axial force, shear force and bending
moment are indicated by means of the variables N, V and M, respectively. It is of
interest to note that the effect of friction between composite components is ne-
glected in calculation of equilibrium of screw.
Foundation modulus of timber in Equation (1) can be given in the form of
Equation (2). Hence, Equation (1) is written as shown in Equation (3).
dV 1
k= (2)
dy dx
d 4 y dV (3)
EI + =0
dx 4 dx
The horizontal and vertical forces which act on the part of screw are determined as
foundation modulus of timber multiplied by the associated resisting displacement
as indicated in Equations (4-5).
FP=k p p=k p dx cos (y cos + sin ) (4)
Ft=k t t=k t dx sin ( cos -y sin ) (5)
Satisfying the equations of equilibrium of the fastener, the shear and axial forces
acting on the part of screw are calculated in the form of combination of horizontal
and vertical forces in the springs as given in Equations (6-7). The friction between
composite components is neglected in the calculations.
d
N = EA = Ft cos + FP sin =k p dx cos ( y cos + sin ) sin +
dx (6)
k t dx sin ( cos -y sin ) cos
V=Ft sin FP cos =k p dx cos ( y cos + sin ) cos +

(7)
k t dx sin ( cos -y sin ) sin
Thus, the derivation of shear and axial forces are determined by Equations (8-9).
dV d4y
= EI 4 =(y cos + sin ) k p cos 2
dx dx (8)
( cos -y sin ) k t sin 2
dN d 2
=EA 2 = -[( y cos+sin ) k p+( cos-y sin ) kt ] sin cos (9)
dx dx
The shear force at the interface of composite material is given as combination of
axial and shear force acting on the part of screw at the interface as illustrated in
Equation (10).
F = N sin V cos (10)

Symons. D, Persaud. R [12] proposed that in the case of long screw with higher t/d
ratio e.g. 30, the axial deformation of the screw is important where the shear force
for most of the length of screw is zero. Hence, for a very long screw, Equation (7)
is equal to zero and the relationship between axial and lateral displacement of
screw is obtained as shown in Equation (11).
(1 - tan )tan
y = (11)
(1 + tan 3 )
where, is ratio of (kt/kp). Symons. D, Persaud. R [12] stated that substitution of
Equation (11) in Equation (9) leads in the differential equation as shown in Equa-
tion (12).
d 2 d
y= (12)
dx 2
(1 + tan ) EA
3
cos
k p tan
Compatibility condition enforces that the axial deformation of fastener at very end
distance from interface is zero. Hence, Equation (12) leads to Equation (13) as the
axial force acting on the screw at the interface of timber and concrete.
0 t
N ( x = 0) = EA tanh
(1 + tan ) EA
3
(1 + tan ) EA
3
cos cos
k p tan k p tan
(13)
Substitution of Equation (13) into Equation (10) gives the force at the interface.
Consequently the slip modulus of a very long inclined screw in TCC joints as
indicated in Equation (14).
F EA t
K= = tanh sin 2
(1 + tan ) EA
3
(1 + tan ) EA
3
(14)
cos cos
k p tan k p tan
It is noteworthy to state that a factor of two takes into account the contribution of
second screw in the crossed screws.
The experimental program is composed of embedding and push out tests.
3.1 Embedding Tests

The embedment strength and foundation modulus of timber for dowel-type fasten-
ers can be determined using different standards such as the ASTM D 5764-97a
[10] and CEN 383:2007[11]. To minimise the bending of fastener in the embed-
ding tests, specific specimen geometries and loading conditions are proposed. The
ASTM standard can be carried out in two different configuration of a full-hole or a
half-hole testing as indicated in Fig.3. The dimension of specimens is defined to
be 38 mm or 2d as minimum thickness whilst 50 mm or 4d are considered to be
the maximum width where d is the diameter of the fastener. Applying a constant
Fig. 3 Test configuration full-hole test in compliance with ASTM D 5764-97a [10]
loading rate of 1.0 mm/minute, the maximum forces were obtained within 1 to 10
minutes. The yield loads were measured using the 5%-offset method. The em-
bedment strength of specimens were determined in compliance with ASTM D
5764-97a [10] as the yield load divided by thickness of timber specimens and
diameter of the fastener. Plotting applied load against the relative displacement of
specimens measured by linear variable differential transducers (LVDT), the initial
slope of the curve indicates the foundation modulus of fastener in timber kf. The
values of foundation modulus of fastener in timber parallel and transverse to the
grain kp and kt are tabulated in Table 1.
3.2 Push Out Test

The experimental aspect of the research consisted of push-out tests of 15 TCC
joint specimens using crossed screws (SFS Intec) inclined at an angle of
30,45,60 which connect laminated veneer lumber (LVL) joists and convention-
al concrete slab. The SFS screws introduced by Meierhofer [13] in the early
1990s have been recognised as one of the first specific shear connections for TCC
structures. The fastener consists of two parts with a diameter of 6 mm as an an-
chor in the concrete and another threaded 165 mm long with outer diameter of 7.5
mm as the anchor in the timber as shown in Fig.4 [14]. Typical geometry and
details of TCCs specimen are shown in Fig.5.
The LVL from the hySPAN range produced by Carter Holt Harvey Wood
Products Australia was employed as timber joist in TCC specimens. In the push-
out tests the load was applied onto the end of the timber component whilst the
Fig. 4 SFS Intec VB-48-7.5x 165 screw
Fig. 5 Cross-section and plan view of specimen (mm)

Fig. 6 Set-up of a Test Specimen in the Test Rig
concrete slab, fixed on the table of the testing rig, resisted the load. The test aimed
to derive the strength and stiffness of the connection. A load cell and LVDTs were
used to measure the load and slip between the timber and concrete (Fig.6). The
experimental load-slip graphs of different test series are depicted in Fig.7.
SFS 30 1
40 SFS 30 2
SFS 30 3
35 SFS 30 4
SFS 30 5
30 SFS 60 1
SFS 60 3
Load (kN)
25 SFS 60 4
SFS 60 5
20
SFS 45 1
SFS 45 2
15
SFS 45 3
SFS 45 4
10
SFS 45 5
5
0
0 1 2 3 4 5
Displacement (mm)
Fig. 7 Load-slip diagrams of different test series
4 Comparison of Model Predictions and Experimental Data

The stiffness model of long inclined screw for TCC joints are presented in
section 2. Table 1 summarises the geometry and material properties of SFS screw
embedded in LVL. A summary of the push-out test results and predictions of ser-
viceability slip modulus of crossed (30, 45, 60) SFS Intec VB-48-7.5x 165
screws is listed in Table 2. A comparison between model and experimental data of
serviceability slip modulus is illustrated in Fig. 8.
Table 1 Summary of mechanical properties of TCC joint with SFS screw
Material properties of TCC joint with SFS screw Value

embedment depth in timber t (mm) for =30,45, 60series 155, 142, 123
Shank diameter of screw, d (mm) 4.6
Youngs modulus of screw (GPa) 210
2
Foundation modulus of timber parallel to the grain kN/mm 391.6
Foundation modulus of timber transverse to the grain kN/mm2 129
Table 2 Summary of experimental and analytical data for stiffness of crossed SFS screws
Test series Mean experimental Ks,0.4 (N/mm) CoV(%) Analytical Ks,0.4 (N/mm) Error
SFS =30 34467 10% 34250.4 0.01%
SFS =45 28800 51% 41686 -44.7%
SFS =60 36784 44% 35637.8 0.03%
60000 Analytical model

crossed SFS screws
50000 Exp. Data crossed SFS
screws
Stiffness (N/mm)
40000
30000
20000
10000
0
0 15 30 45 60 75 90

Fig. 8 Comparison of model and experimental results of Serviceability slip modulus
The analytical model takes into account only axial deformation of screw which
is in agreement with the tensile failure of screw in tension as observed in split
specimens. The experimental test quantified the serviceability slip modulus of the
of different test series specimens representing a negligible difference to the pre-
dicted value for =30 and 60 series whilst a maximum difference of 44% was
observed for the stiffness of =45 series (Table 2). Among different series,
=30 series experienced the maximum slip modulus whereas, =45 series rep-
resented the minimum slip modulus. As seen in push out test result (Fig.7), the
behaviour of the TCC joints was reasonably consistent whereas, load increased
steadily, slowing slightly before reaching the peak load (Fmax). Beyond the peak
load, load gradually reduced to around 80-90% Fmax, before a sudden and com-
plete failure [2].
5 Conclusion
This paper presents a model for the stiffness of TCC joint using crossed inclined
proprietary screws (SFS Intec). The theory of beam on elastic foundation is ex-
tended to derive the serviceability slip modulus of TCC joints with inclined screws
which are loaded in tension and compression. The experimental aspect of the re-
search consists of push-out tests and verifies the stiffness model. Comparing
analytical and experimental results, the stiffness model seems to be reasonably
accurate in predicting the slip modulus of TCC joints at SLS. This research sug-
gests the model to facilitate the design of inclined screw shear connections for
TCC construction.
Acknowledgments. The authors gratefully acknowledge the financial support of Structural

Timber Innovation Company (STIC).
References
1. Branco, J.M., Cruz, P.J.S., Piazza, M.: Experimental analysis of laterally loaded nailed
timber-to-concrete connections. Construction and Building Materials 23(1), 400410
(2007)
2. Moshiri, F., et al.: An investigation on TCC systems using light-weight concrete. In:
World Conference on Timber Engineering (WCTE), New Zealand (2012b)
3. Clouston, P., Bathon, L.A., Schreyer, A.: Shear and Bending Performance of a Novel
WoodConcrete Composite System. Journal of Structural Engineering 131(9), 1404
1412 (2005)
4. Mascia, N.T., de Olirera Santana, C.L., Cramer, S.M.: Evaluation of the equivalent slip
modulus of nailed connections for application in linear analysis of plywood timber
beams. Materials Research 11(11), 151157 (2008)
5. CEN, design of timber structures - part 1-1: General rules and rules for buildings, in
Eurocode 5 Eurocode 52003, European Committee for Standardization: Brussels, Bel-
gium
6. Kuenzi, E.W.: Theoretical design of a nailed or bolted joint under lateral load, USDA
forest products lab: Madison (1955)
7. Foschi, R.: Load-slip characteristics of nails. Wood Science 7(1), 6976 (1974)
8. Gabr, M., Valero, S.: Geotechnical properties of municipal solid waste. ASTM Ge-
otechnical Testing Journal 18(2), 241251 (1995)
9. Patton-Mallory, P., Pellicane, P., Smith, F.: Modeling Bolted Connections in Wood:
Review. Journal of Structural Engineering 123(8), 10541062 (1997)
10. (ASTM), A.S.f.T.a.M., Standard Test Method for Evaluating Dowel-Bearing Strength
of Wood and WoodBased Products, in D 5764-97a1997, Annual Book of ASTM
Standards: West Conshocken, PA
11. CEN, Timber Structures Test methods Determination of embedment strength and
foundation values for dowel type fasteners, in EN3832007, European Committee for
Standardization: Brussels, Belgium
12. Symons, D., Persaud, R., Stanislaus, H.: Slip modulus of inclined screws in timber
concrete floor. Structures and Buildings 16(4), 245255 (2010)
13. Meierhofer, U.: A new efficient system for timber/concrete composite structural ele-
ments. Test, research and development. In: The IUFRO S5.02 Timber Engineering
Conference (1992)
14. Lukaszewska, E.: Development of Prefabricated Timber-Concrete Composite Floors.
Universitetstryckeriet, Lule, Lulea (2009)
Shear Performance of Wood-Concrete
Composite with Different Anchorage Length
of Steel Rebar
Sang-Joon Lee*, Jrme Humbert, Kwang-Mo Kim, Joo-Saeng Park,

and Moon-Jae Park
Korea Forest Research Institute (KFRI), 57 Hoegiro, Dongdaemun-gu, Seoul, Korea

lsjoon2@snu.ac.kr
1 Introduction
High composite action between members is one of the most important concern to
design the wood-concrete hybrid system including the timber bridge (Yeah et al.,
2011; Lee et al., 2012). Researches on the timber bridge have been widely
conducted in European countries and Northern America. After 1930's, timber
bridges started to be studied and to be constructed. Several experimental
approaches which deal with the composite action between wood and concrete
member have been performed. Ritter(1990) and Jutila and Salokangas(2010)
pointed out that wood-concrete hybrid system can be effectively applied for the
timber bridge even the fully-composite action is difficult to be achieved.
Additional researches for concerning high composite action have been tried
after 1960's. They can be devided into three approaches - making notches at
wooden part, inserting steel rebar or plate in wood and concrete and using the glue
(An et al., 2008; Gutkowski et al., 2008; Balogh et al., 2010; Bathon and Bletz-
Muhldorfer, 2010). The durability as well as the structural safety also have been
studied (Nunes and Carlito, 2008). It was found out from these researches that the
wood-concrete hybrid system can be effectively applied for the timber
superstructure however the concept of the fully-composite action still remains
main concern for using wood-concrete composites.
There is little approach for using wood-concrete composite system in Korea,
moreover there is no wooden bridge for vehicle and written specifications and/or
standard for constructing the wooden bridge. This makes it difficult to concern
recent rising concern about structural usage of wood and wooden culture in Korea.
Therefore, this study was planned for concerning the structural performance of
wood-concrete composite. The shear performance with different anchorage length
in wood (Lee et al., 2012) was firstly studied and the shear performance with
different anchorage length in concrete was tried to be studied in this paper. After
the prediction with EYM (European Yield Model), comparison study was
conducted between wood and concrete members.
*
DOI: 10.1007/978-94-007-7811-5_41, RILEM 2014
456 S.-J. Lee et al.
Keywords: timber-concrete composite, timber bridge, steel rebar, chemical

anchor, anchorage length, shear failure, pinus koraiensis Sieb.
2.1 Materials
The species of wood for manufacturing the wood-concrete composite was pinus
koraiensis Sieb. and moisture content and oven-dried density were 13.5 2.04 %
and 0.43 0.04 g/cm3 respectively. The shape and size of the wood-concrete
composite (fig. 1) were referred to previous research (Jurita and Salokangas, 2010;
Lee et al., 2012). While anchorage length of steel rebar in wood fixed to 180 mm,
anchorage length in concrete differed with eight steps (20 mm, 40 mm, ..., 160
mm). Eight wood-concrete composites were used for this study.
Table 1 Mix properties of concrete
Design Mixture (kg/m3)

Slump W/C
strength Air
(mm) (%) C W S G Admix.
(MPa) (%)
21 120 53.8 261 165 902 902 4.51.5 2.92
Fig. 1 Wood-concrete composite
2 m3 of ready-mixed concrete which was made by Korean standard (KSF 4009)

was used for concrete material. Table 1 shows the mix properties of concrete
which's design strength targeted to 21 MPa. The concrete was cured in shady place
about 28 days and the compressive strength after the curing shows 20.1 MPa.
Shear Performance of Wood-Concrete Composite 457
2.2 Manufacture of the Wood-Concrete Composites

After anchoring of steel rebar in wood with eight different length, form work for
making concrete part was performed (fig. 2). FIS V 360S chemical anchor
(Fischer co., ltd., Germany) was used for anchoring steel rebar in wood and form
work was done after two days of curing the chemical anchor. Fig. 2 also shows the
specimens after pouring ready-mixed concrete and concrete part was cured in
shady place about three days. After removing the form, each specimens were
additionally cured about twenty five days.
Fig. 2 Manufacturing process of making wood-concrete composite
2.3 Evaluation of the Shear Performance

EYM was applied for prediction of shear performance of wood-concrete
composite. Reference design value(Z)s were derived from each determined yield
mode. As the material properties, the embedding strength of wood and concrete
materials (Few, Fec) was determined to 31.6 MPa and 42 MPa (KBC Wood Design
Manual, 2008) and the yield strength of steel rebar (Fyb) was determined to 294
MPa as the property of SD30A steel rebar. Wooden part was assumed as the main
member and concrete part was assumed as the side member. While the anchorage
length of steel rebar in wood (tw) was fixed to 180 mm, EYM results were derived
with seven steps of anchorage length (tc) in concrete.
Universal Testing Machine (UTM) (Instron, co., ltd, USA) was used for shear
test (fig. 4). The testing procedure was referred to previous research (Lee et. al.,
2012).
Fig. 3 EYM of wood-concrete composite
Fig. 4 Wood-concrete specimen on shear test
3 Results and Discussions
3.1 Prediction of Yield Mode and Reference Design Value(Z)

Table 2 shows results of EYM due to anchorage length of steel rebar in concrete
member. Below 10 mm of anchorage, minimum yield loads were calculated at Is
mode which makes bearing failure at concrete. And the reference design values at
anchorage length of 5 mm and 10 mm derived 2,100 N and 4,200 N respectively.
IV mode which makes plastic hinge at shear plane was derived over 15 mm of
anchorage. And the reference design value was calculated to 5,945 N.
Table 2 Results of EYM due to anchorage length of steel rebar in concrete (Unit : N)
Anchorage Yield Mode

Reference
Length in Determined
Design
Concrete Is II IIIs IV Yield Mode
Value (Z)
(mm)
5 2,100 38,477 11,708 5,945 Is 2,100
10 4,200 39,544 16,246 5,945 Is 4,200
15 6,300 40,685 19,630 5,945 IV 5,945
20 8,400 41,897 22,429 5,945 IV 5,945
40 16,800 47,402 30,841 5,945 IV 5,945
60 25,200 53,838 37,145 5,945 IV 5,945
80 33,600 61,053 42,425 5,945 IV 5,945
Results show similar tendency with previous research (Lee et. al., 2012) which
applies the wood-to-wood composite. Yield mode was changed from Is mode to
IV mode at 15 mm
20 mm and 10 mm
15 mm of anchorage length for
wood and concrete side member respectively. At IV yield mode, reference design
values show 5,565 N and 5,945 N respectively for wood-to-wood and wood-to-
concrete composites. Reference design value at IV mode of wood-to-wood
composite shows about 94 % of wood-to-concrete composite even the embedding
strength of wood (31.6 MPa) is about 75 % of concrete (42 MPa).
3.2 Shear Performance

Fig. 5 shows the load-displacement curves for tested wood-concrete composite
and table 3 shows derived shear performance with initial stiffness, yield load and
maximum load. Initial stiffness and yield load show 2.54 0.42 kN/mm and 7,932
1,266 N respectively for the eight tested wood-concrete composites. Results at
20 mm and 40 mm of anchorage show a little bit smaller values, however there
was not increasing tendency with increment of the anchorage length. It is
considered that all specimens yield samely in IV mode as predicted with EYM. In
the case of maximum load, there was no increasing tendency with increment of the
anchorage length except composite with anchorage of 20 mm which shows pullout
failure of inserted steel rebar.
Fig. 5 Load-displacement curves
Table 3 Shear performance of wood-concrete composites due to anchorage length in

concrete
Anchorage Length Initial Stiffness Yield Load Maximum Load

in Concrete (mm) (kN/mm) (N) (N)
20 2.5 7,527 10,602
40 2.7 7,956 26,985
60 3.2 9,652 27,403
80 2.9 11,728 32,055
100 2.6 10,558 27,869
120 2.0 9,808 29,148
140 2.0 10,144 29,802
160 2.4 11,956 25,807
Table 4 shows comparison of shear performance with initial stiffness and yield
load between wood-to-wood and wood-to-concrete composites. Results of wood-
to-wood composite refer to previous work (Lee et. al., 2012). Same with the result
of wood-concrete composite, there was no increasing tendency of initial stiffness
and yield load with increment of anchorage length of steel rebar in wood-to-wood
composite. Average initial stiffness and yield load of wood-to-wood composite
shows about 65 % and 93 % than those of wood-concrete composite. Reference
design values of wood-to-wood and wood-to-concrete composites calculated to
5,565 N and 5,945 N respectively, and actual yield loads showed about 130 % of
their values.
Table 4 Comparison of shear performance between wood-to-wood and wood-to-concrete

composites
Initial Stiffness Yield Load

(kN/mm) (N)
Wood-to-wood composite
1.630.19 7,527778
(Lee et. al., 2012)
Wood-to-concrete composite 2.540.42 7,9561,266
3.3 Shear Failure

Fig. 6 shows photographs after failure of tested wood-concrete composites at 20
mm and 100 mm of anchorage. After the yield load, local part around steel rebar
at concrete part pulled out with steel rebar in the case of 20 mm of anchorage (fig.
7, left). However, there was not pull out failure over 40 mm of anchorage and
brittle failure of concrete occurred different from the wood-to-wood composite
(Lee et. al., 2012). Concrete cracked along the loading direction as confirmed
from fig 7(right). And bending of steel rebar also can be confirmed which
indicates the plastic hinge at the shear plane.
Fig. 6 Photographs after shear failure of wood-concrete composites (left: 20 mm anchorage,

right: 100mm anchorage)
4 Conclusions
This study was performed for evaluation of wood-concrete composite with
different anchorage length of inserted steel rebar. Results were compared with
previous work (Lee et. al., 2012) which dealt with the wood-to-wood composite.
EYM was applied as the preliminary prediction and mode conversion from Is
mode to IV mode occurred before and after anchorage length of 10 mm 15
mm which is about 5 mm shorter anchorage length then the wood-to-wood
composite.
There was not increasing tendency of shear performance with increment of
anchorage length from 20 mm to 160 mm in concrete. Initial stiffness and yield
load show 2.54 0.42 kN/mm and 7,932 1,266 N respectively and those of
wood-to-composite were about 65 % and 93 %. Actual yield loads derived from
shear test showed around 130 % of reference design value(Z)s from EYM for
wood-to-wood and wood-to-concrete composites. Wood-concrete composite
showed brittle failure after the yield point while wood-to-wood composite showed
ductile failure.
References
1. An, H.J., Kim, S.C., Moon, Y.J., Yang, I.S.: Experimental study of composite beams
consisting structural laminated timber beam with concrete slab. In: 2008 Proceedings
of the Korea Concrete Institute Annual Meeting, vol. 20(1), pp. 233236 (2008)
2. Balogh, J., Fragiacomo, M., Miller, N., Gutkowski, R., Atadero, R.: Testing of Wood-
Concrete Composite Beams with Shear Key Detail. In: Proceedings of International
Conference on Timber Bridges, Lillehammer, Norway, pp. 393398 (2010)
3. Bathon, L., Bletz-Muhldorfer, O.: Performance of single span wood-concrete-
composite bridges under dynamic loading. In: Proceedings of International Conference
on Timber Bridges, Lillehammer, Norway, pp. 399402 (2010)
4. Construction and Maintenance Branch. Ministry of Transportation and Infrastructure,
Standard Specification for Highway Construction (2009)
5. Customary U. S. Units 4th Edition, AASHTO LRFD Bridge Design Specifications
(2007)
6. Gutkowski, R., Brown, K., Shigidi, A., Natterer, J.: Laboratory tests of composite
wood-concrete beams. Construction and Building Materials 22, 10591066 (2008)
7. Julita, A., Salokangas, L.: Wood-Concrete Composite Beams - Finnish Speciality in
the Nordic Countries. In: Proceedings of International Conference on Timber Bridges,
Lillehammer, Norway, pp. 383392 (2010)
8. Lee, S.J., Eom, C.D., Kim, K.M.: Shear Performance of Wood-Concrete Composite I -
Shear Performance with Different Anchorage Length of Steel Rebar in Wood. J.
Korean Wood Sci. & Tech. 40(3), 186193 (2012)
9. Nunes, J.L., Carlito, C.J.: Field Load Test Behavior of Composite Log-Concrete
Bridge - P01. In: Proceedings of the 51st International Convention of Society of Wood
Science and Technology, Chile WS-36 (2008)
10. Ritter, M.A.: Timber Bridge: Design, Construction, Inspection, and Maintenance. U.S.
Department of Agriculture, U.S. Forest Service, EM 7700-8 (1990)
11. Yeah, D., Fragiacomo, M., De Franceschi, M., Boon, K.H.: State of the Art on Timber-
Concrete Composite Structures: Literature Review. Journal of Structural Engineering
137(10), 10851095 (2011)
Development of Prefabricated Timber-Concrete
Composite Floors
Petr Kuklk, Pavel Nechanick, and Anna Kuklkov
Faculty of Civil Engineering, Czech Technical University in Prague,

Czech Republic
kuklik@fsv.cvut.cz
Abstract. The aim of the current research is to develop a new composite timber-
concrete structure element as a prefabricated panel. In the Czech Republic, it is
possible to see an increased interest in the constructions of timber and wood-based
materials in recent years. Increased productivity is evident particularly in
realization of family houses, where the number of realizations was doubled during
last year in comparison with the situation five years ago. A little bit worse
situation is in realization of multi-storey buildings. To some extent, this may be
the result of not only the strict fire safety regulations, but also not so good timber
properties in terms of serviceability limit states and acoustics. There is a will to
spread this kind of construction in our region and one of the possibilities to help
solve these constraints is usage of timber-concrete composites primarily for floor
structures.
Keywords: timber-concrete composite, prefabricated panel, shear anchor,

punched metal plate, load-slip behaviour.
1 Introduction
Timber-concrete composite structures are becoming very important in the housing
sector. They have many advantages compared to traditional timber floors and are
widely used as an effective method for refurbishment of existing timber floors.
Due to the many benefits they are now being used more in new multi-storey
timber framed houses.
The presented research is focused on the development of prefabricated
composite floor element due to advantages of prefabrication. An important part of
the research is to design a new mechanical shear connector on the basis of timber
frame fittings.
2 Materials and Technology

Timber and concrete are commonly used materials in our region and thus
development has been focused on shear connection between timber and concrete
DOI: 10.1007/978-94-007-7811-5_42, RILEM 2014
464 P. Kuklk, P. Nechanick, and A. Kuklkov
and technology used for shear connector assembly. To ensure low cost on
production, neither gluing processes nor adjustments of timber beams like notches,
dimples or holes have been willed.
Development of a new type of prefabricated composite timber-concrete elements
(primarily for floor slabs) has been focused on mechanical fitting of new shear
connectors between a pair of timber beams in whole length at once. Two technology
procedures can be used for achievement of composite structure. Pouring fresh
concrete mixture in a factory over semi-prefabricated timber beams with installed
shear connectors or concreting shear connectors into a slab and after hardening of
concrete assembling of timber beams. In first case, caution with water cement ratio
and removable formwork are needed. In second case, a special assembling machine
has to be used. An engineering solution is shown in Figure 1.
Fig. 1 An exemplary solution of timber-concrete structure (1-concrete slab, 2-timber

beams, 3- shear connector, 4-reinforcement mesh)
3 Shear Connector
For connecting timber and concrete, a modification of standard punched metal

plate was chosen. Extensive research was performed at University of Delft,
Karlsruhe and Stuttgart [6], [7], [8], where different modifications of punched
metal plates were investigated. In collaboration with a Czech manufacturer of
connectors for timber structures BOVA s.r.o. [16], an innovative punched metal
plate (shown in Figure 2) has been developed.
Fig. 2 Prototype of an innovative shear connector

Development of Prefabricated Timber-Concrete Composite Floors 465
A first prototype for first two series of direct shear tests was manufactured. It is
a double-sided punched metal plate made of hot-dip galvanized thin steel sheet
with zinc coat. Dimensions of the connector are 84 x D mm, where D is a length
parameter and can be produced in multiple of 105 mm.
4 Tests and Results
First two series A and B of timber-concrete specimens using double-sided

punched metal plate were experimentally tested for describing mechanical beha-
vior of shear connectors and determining slip moduli for simplified design
methods given in Eurocode 5 [10]. Each series consists of five T shape
specimens with material properties mentioned below.
Two KVH profiles 60x240 mm, 400 mm long were made of timber strength
class C24 according to EN 338[11], visually graded according to EN 1611-1[12].
As their surfaces are planed and edges are trimmed, the quality corresponds with
standard used elements in traditional timber structures based on light timber
frames.
Timber density and moisture content were measured before shear tests.
Average values were about 425 kg/m3 with 11% moisture content.
Two timber elements are jointed together with innovative shear connector on
the upper edge (where the timber-concrete contact is) and one small standard
double-sided punched metal plate 35x84 mm inserted nearby the bottom edge.
This small plate is inserted for better pressing and securing timber beams position.
Concrete desk was made of handmade concrete mixture and reinforced with
steel welded mesh with 6 mm rods and 150 mm distance of mesh. Concrete
mixture was made from a dry concrete mix with the addition of appropriate
amounts of water. Then the mixture was poured into handmade formworks and let
to harden. Thickness of desk is 80 mm and dimensions were 600x400 mm. This
unusual solution was chosen to ensure protection of an intellectual property in the
beginning of research and will not be realized anymore.
Due to significant human factor, corresponding to concrete making, cube
compressive strength was checked at five small cubes of each series according to
EN 12390. Expected cube compressive strength was about 25 MPa and average
reached value was about 24,5 MPa. Characteristic value according to EN 14358
was just about 20 MPa. For further calculation and numerical modelling, strength
class C16/20 is taken into account.
The difference between both series is in length of shear connector where one
and a half times the length of the metal plate is used in series B. Different length
was chosen for experimental investigation of relationship between geometrical
properties and failure mode in variable load levels and gaining reference data for
refining numerical 3D FEM model. Also practical aspects of pressing double-
sided punched metal plate of different length were verified. The longer plate is
more difficult to punch between timber beams due to the need of higher pressing
force and to secure the right beams position.
As we verified during pressing the metal plates to timber beams, it is better to

use one small standard double-sided punched metal plate nearby the bottom edge
of the beams. This plate helps to secure optimal position of timber beams during
pressing.
Because the pressing of punched metal plates on both sides to an optimal
position is very complicated, a direct shear tests using an asymmetrical specimen
were performed. This type of set up was selected for the reason that the specimens
are much easier to construct than the standard push-out tests specimens, where the
concrete slab is on both sides. These specimens are also lighter, cheaper and faster
assembled for experimental testing. Consideration of a disadvantage of an asyme-
trical shear test set-up is important. Due to eccentricity of the shear force, an
overturning moment will lead into a compressive force at the interface between
timber and concrete, which increases the friction and thus improves mechanical
behavior of the shear connector.
Shear test set-up is presented in Figure 3. An asymmetrical specimen is placed
on a special self-made anchoring frame made of steel and inspired by testing
device used for large experimental testing in Sweden [9]. Stability of specimen is
secured by a sliding support at the top of the timber elements. The support is made
of two Teflon plates; one is screwed to the steel frame, the second one to the
timber. To ensure ideal sliding and to minimize friction, the Teflon plates were
oiled.
Fig. 3 Shear tests set-up
The shear tests were carried out according to EN 26891 [13], where also
provisions for the determination of the connector slip moduli are given. The first
sample of series A was loaded with parameters corresponding with estimated
failure Fest=50 kN. The real maximal reached shear force was about 36 kN and
thus the loading scheme was modified to Fest=40 kN. The tests were carried out
within 15 minutes until the failure load was reached. Slip between concrete and
timber was measured in the middle of the timber beams and on both sides using
Linear Voltage Displacement Transducers (LVDT), type 15D. Timber-concrete
layer opening was measured on both sides 100 mm from the upper edge of timber
beams with LVDT 5G. All transducers were glued to timber surface. The tests
were performed under a controlled force increment on INOVA PRAHA hydraulic
press with key points at 0,1xFest, 0,4xFest and 0,7xFest and data retrieved with a
frequency of 2 Hz. Load range of hydraulic press is 0-200 kN with an accuracy of
1-2%.
Fig. 4 Examples of typical failure of shear connection
Probable failure model of the samples could not be reliably deduced. One of the
following failures was predicted: a shear failure of the timber beams; a shear
failure of metal plate between holes of the first timber row; plastic deformation
of teeth in combination of embedment of timber holes; cutting of concrete slab
and combinations of previous. Currently two series of five specimens each were
experimentally tested. As the results, dominant failure model observed was shear
failure of punched metal plate in the plane intersecting the middle part of the first
row of holes in timber beams area. Twisting of first row of teeth punched into the
timber with combination of plastic deformation of the area of first two rows of
teeth and holes can be observed too. Small shear cracks occurred in some of the
timber beams with lower density. As evident, just two rows of teeth and holes in
timber area are deformed. No visible deformation or cracks at concrete slab and no
significant deformations of concrete part of the steel plate were observed. This
observation will help with rearrangement of teeth and their geometrical properties.
As the shear connection is usually characterised by non-linear force-slip beha-
vior, two different slip moduli are considered for simplified design purposes [10]:
Kser for serviceability limit state and Ku for ultimate limit state design. The slip
modulus Kser corresponds to the secant value at 40% of the load-carrying capacity
of the connection and is presented by K04 value. For the slip modulus Ku, the use
of the secant modulus at 60% presented by K06 value is recommended.
468 P. Kuklk, P. Nechanick, and A. Kuklkovv
Whereas steel failure is

i dominant, very similar behavior described by load-sliip
diagram can be observed d in series A. Except the specimens A3 and B44,
where stiffer behavior was
w reached due to some teeth in first row punched iin
timber were twisted of. As
A evident, there is a very low scatter of load-slip curvees
in series A.
Fig. 5 Result of direct shear test specimen A3
For longer plate in serries B is higher scatter of load-slip curves in the rangge
above 0,5Fmax. An Averaage slip is about 5,5 mm in both series and measureed
failure force is about 35 kN
k and 54 kN respectively. Average values of slip moduuli
and maximal force of bo oth series are presented in Table 1. Although the mettal
plate in series B is onee and a half times longer, only maximal shear force waas
reached with the ratio of 1,5.
1
Table 1 Slip modulus valuess
SLIP MODULUS [kN/mm]

SERIEES Fmax [kN]
K04 K06
ra
ange 34,0 38,4 27,3 33,1 34,2 36,1
A aveerage 36,7 29,5 35,1
1,5 2,0 0,8
ra
ange 43,1 56,7 37,0 56,0 51,6 56,2
B aveerage 51,0 48,1 53,6
4,9 7,2 1,7

5 Numerical 3D FEM Model
Preliminary numerical 3D FEM model of direct shear test has been created in
ANSYS WORKBENCH 13 software for comparison of FEM model solution
and experimental behavior of innovative shear connector in timber-concrete
composite structure. Orthotropic elasticity model was used to design timber.
Concrete was designed as an isotropic material adequate to concrete strength class
C16/20 and shear connector was modelled as non-linear steel with bilinear
isotropic hardening. Contact regions were added on all concrete-steel and timber-
steel surfaces. Model was loaded with force F= 20 kN that corresponds to 0,5Fest.
Current results of 3D FEM simulation is presented in Figure 6, where results for
deformation of metal plate and equivalent von Mises stress are shown.
Fig. 6 Numerical model
6 Conclusions
This paper presents an overall research work of Department of Steel and Timber
Structures, CTU in Prague as an introduction. Detailed information about current
research of prefabricated timber-concrete composites is described in the second
part of paper. Descriptions of innovative engineering solution and developed shear
connector are presented. Experimental study on two modifications of the shear
connector was performed. Currently obtained results of direct shear tests are
presented in load-slip diagrams and calculated 5-percentile characteristic values of
slip moduli.
Acknowledgement. This outcome has been achieved with the financial support of Czech
Science Foundation, project No. 104/10/215.
References
[1] Ceccotti, A.: Timber-concrete composite structures. Timber engineering STEP2,
Centrum Hout, NL (1995)
[2] Frangi, A., Fontana, M.: A design model for the fire resistance of timber-concrete
composite slabs. In: Proceedings of the IABSE Conference on Innovative Wooden
Structure and Bridges, Lahti, Finland (2001)
[3] Kuklik, P.: Timber structures. Czech Chamber of Authorized Engineers and Tech,
Prague (2005) ISBN 80-86769-72-0
[4] Kuklkov, A.: Kompozitn devobetonov konstrukce, VUT v Praze, Fakulta
stavebn (2004)
[5] ONeill, J.W.: The fire performance of timber-concrete composite floor. Doctoral
thesis, University of Canterbury, Christchurch, New Zealand (2009)
[6] Aicher, S., Klck, W., Dill-Langer, G., Radovi, B.: Nails and nailplates as shear
connectors for timber-concrete composite constructions. Otto-Graf-Journal 14, 189
210 (2003)
[7] Blass, H.J., Ehlbeck, J., van der Linden, S.M.: Trag- und Verformungsverhalten von
Holz-Beton-Verbundkonstruktionen, Bericht T 2710, Un Universtitt Karlsruhe
(1995)
[8] Van der Linden, M.-L.R.: Timber-concrete composite floor systems. Doctoral thesis,
Technical University of Delft (1999)
[9] Lukaszewska, E., Johnsson, H., Fragiacomo, M.: Performance of connections for
prefabricated timber-concrete composite floors. Materials and Structures, 15331550
(2008)
[10] Comit European de Normalisation, EN 1995-1-1: Eurocode 5: Design of timber
structures - Part 1-1: General - Common rules and rules for buildings, Bruxelles,
Belgium (2006)
[11] Comit European de Normalisation, EN 338: Structural Timber Strength classes,
[12] Comit European de Normalisation, EN 1611-1: Sawn Timber Appearance grading
of softwoods Part 1: European spruces, firs, pines, Douglas firs and larches,
[13] Comit European de Normalisation, EN 26891 - Timber structures. Joints made with
mechanical fasteners. General Principles for the determination of strength and
deformation characteristics, Bruxelles, Belgium (1991)
[14] Engineers Handbook; Reference tables coefficient of friction,
http://www.engineershandbook.com
[15] Astley, R.J., Stol, K.A., Harrington, J.J.: Modelling the elastic properties of
softwood. Holz als Roh und Werkstoff 56, 4350 (1998)
[16] BOVA s.r.o., http://www.bova-nail.cz
[17] Portable Lumber Grader. Information on http://www.fakopp.com/
Wood-Based Construction for Multi-story
Buildings: Application of Cement Bonded Wood
Composites as Structural Element
Alireza Fadai, Michael Fuchs, and Wolfgang Winter
Department of Structural Design and Timber Engineering,

Vienna University of Technology
{fadai,fuchs,winter}@iti.tuwien.ac.at
Abstract. Modern housing and architecture are orientated towards high living
quality and low energy consumption. The city of Vienna has put emphasis to in-
troducing wood as a sustainable and ecological material for the building industry
in the last decade.
With the purpose to optimize structural performance, usability, energy-efficient
and ecological profile of wood-based composite systems in several research pro-
jects of the Department of Structural Design and Timber Engineering (ITI) the
combination of timber products with other conventional building materials and
components was explored [3, 4, 5, 6].
These technologies provide structurally efficient components for low-energy
constructions, support rapid-assembly modular construction with increased effi-
ciency using prefabricated dry elements, thereby open opportunities to reduce
carbon emissions.
As a representative example for these developments the application of cement
bonded wood composites as structural sandwich panels illustrates the extent of the
interrelationships involved in the development of complex system solutions to
increase resource efficiency.
Keywords: Timber-lightweight concrete, Wood chip concrete, Building construc-

tion, Structural engineering, Sustainable building, Sustainable redevelopment,
Wood-based structural elements, Resource efficiency.
1 Introduction
The European countries have to face profound changes in the demand for build-
ings. The basic requirements for new apartments or new office spaces are very
different from the post war decencies, e.g. the quantity, the quality, the size, the
functionality, the flexibility, the expected lifetime for new or for refurbished
buildings have enormously changed. This leads to the need to develop other types
DOI: 10.1007/978-94-007-7811-5_43, RILEM 2014
472 A. Fadai, M. Fuchs, and W. Winter
of buildings: polyvalent, faster erection and transformation, low energy consump-

tion for heating and cooling etc. The urban planning also needs other development
strategies, less traffic, higher density, less one family houses and more densifica-
tion and remodeling of the existing buildings. The adaptation of the built
environment to these new challenges needs a sound evaluation of the currently
dominating building materials and technologies and a neutral comparison with
other options.
In Europe, more than 80% of all buildings consist of reinforced concrete. At the
same time, European forests were overexploited in the past centuries. Forestry
laws set out protected them only since the end of the 19th century. These laws
target a sustainable forestry management, and forests have indeed fully recovered
thanks to reforestation. In fact, they could produce 30 to 50% more wooden raw
material than is harvested today.
The building codes are in constant revision, actually nearly all European coun-
tries allow at least four-story buildings in timber as regular solutions and pilot
projects had actually been built up to ten stories. In the past few years, an increas-
ing number of wood-based buildings had been constructed in Central and Western
Europe, e.g. single-family houses, multi-story condominiums, schools, hospitals
and office buildings. Timber constructions with more than two stories make up
less than 5% of the building market in Austria [1].
Buildings play an important role, especially when the challenge is to control
climatic change through the reduction of CO2 emissions in the production of
building materials. Unless the maintenance and exploitation of buildings as well as
the construction of new buildings changes fundamentally, the European Union
(EU) objectives to reduce CO2 emissions and the consumption of non-renewable
resources may not be reached [2].
2 Motivation
The renewable raw material wood is available in a sustainable way; the basic
technologies are well known and good examples of wood-based building already
exist. However, the potential of using wood in the building sector is still highly
underestimated. Wood as a sustainable and ecological material has an excellent
chance to be an alternative construction material in urban areas.
Modern housing and architecture are orientated towards high living quality and
low energy consumption. Whereas mineral-based buildings dominate European
cities, the city of Vienna has put emphasis to introducing wood as a sustainable
and ecological material for the building industry by political policy in the last
decade. Today, on the free market, timber-based buildings face cost challenges to
conventional building technologies.
The building market had always been the most important mass market for tim-
ber products. The European wood and forest industry plays a key-role in ensuring
a sustainable and environment-friendly use of an important natural resource,
which can positively influence the climate and living conditions. Research issues
Wood-Based Construction for Multi-story Buildings 473
and practical applications regarding the use of renewable resources are important
future-oriented priorities.
Strategies to increase the use of timber in multi-story buildings are:
full potential exploitation of prefabrication and reduced erection time
optimized combination of wood with (cheaper) mineral layers for economically
attractive structural members,
combination of wood with (noncombustible) mineral layers for enhanced fire
resistance,
initiate and realize attractive pilot projects and diffuse information on it interna-
tionally,
increase the specific knowledge about timber of professional planners, builders,
students,
create an international network of builders, planners and universities to ex-
change ideas and experiences.
With the purpose to optimize structural performance, usability energy-efficient
and ecological profile of wood-based composite systems in several research pro-
jects of the Department of Structural Design and Timber Engineering (ITI) at the
Vienna University of Technology (VUT) the combination of timber products with
other conventional building materials and components, e.g. glass, steel and light-
weight wood particle concrete was explored [3, 4, 5, 6].
These technologies provide structurally efficient components for low-energy
constructions, support rapid-assembly modular construction with increased
efficiency using prefabricated dry elements, thereby open opportunities to reduce
carbon emissions.
As a representative example for these developments the application of cement
bonded wood composites as structural sandwich panels illustrates the extent of the
interrelationships involved in the development of complex system solutions to
increase resource efficiency.
3 Application of Cement Bonded Wood Composites as

Structural Element
Mixtures of cement and wood components, e.g. as cement-bonded wood particle

or strand panels, had already been developed around 1900 but they found their
application mainly in non-structural elements such as insulating boards for thermal
and acoustic insulation, as a support for stucco, as fire protection, or for lost
formwork.
Development and applications of wood-concrete-based technologies had also
been achieved in countries where gravel and cement for concrete production are
scarcely available, or where wood is available in abundant volume and the local
climate demands less insulation. Many of the products were developed during the
petrol crisis of the 1970s, in particular. Early and recent research of cement-bound
wood particles or strand boards focused on dimensional stability and durability,

while more recent research also investigates other options such as thermal storage
capacity of the lightweight material or finite element modelling.
Relatively simple systems of timber-concrete composite elements for floor el-
ements as well as shear connections have also been developed and applied in prac-
tice; however, these structural systems generally use heavy regular concrete and
shear connectors with time-consuming installation methods. Additionally, these
systems often need further installations to meet fire safety, thermal and acoustic
insulation requirements. The demolition of concrete structures is time-consuming,
and the incurring concrete waste is laborious to recycle; furthermore, recycled
concrete is only at the beginning of its practical acceptance for its application in
structural elements.
The use of wood-based, cement-bonded products could be considerably in-
creased if applicability in structural elements is given. These gaps can be filled in
by multi-layer elements (light wood-concrete compound instead of heavy regular
concrete, in composite action with structural timber elements) with a maximum
use of wood, appropriate shear connections and corresponding structural design
approaches. The development of these lightweight wood-concrete systems leads to
a new, highly competitive generation of polyvalent multi-material building com-
ponents (Figure 1).
Fig. 1 Multi-layer floor elements made of timber, light wood-concrete and self-compacting
concrete
Various research projects of the ITI at the VUT focused on

analysis and design particularities of elastically composed systems, especially
with regard to girders in bending,
mechanical fasteners or other devices used to generate the composite action,
the long-term behavior of such composite structures due to shrinkage and
creep,
appropriate methods for numerical and spreadsheet calculations [3].
To increase the effective depth, structural timber elements are combined with
the wood-concrete compounds, since timber is an excellent material for tensile
loading (parallel to the fibers), and the wood-concrete compounds should
primarily be loaded in compression and shear. It is principally possible to develop

wood-concrete compound compositions such that they provide a sufficient re-
sistance in compression, and in tension to absorb local stress concentrations.
The innovative technology combines existing concrete tradition with organic
renewable resources as an alternative to conventional concrete or masonry, there-
by open opportunities to reduce carbon emissions through the reduction of cement
use and increased use of wood for construction since the carbon dioxide stocked in
the wood is preserved for several decades instead of burning it. By using renewa-
ble resources, waste and manufactured wood products of the forest industry, this
technology provides structurally efficient components for low-energy construc-
tions, supporting rapid-assembly modular construction with increased efficiency
using prefabricated dry elements.
3.1 Scientific and Technological Objectives, Relevant Results

The use of composite products of timber and lightweight wood-concrete was to be
supported and stimulated by comprehensive and scientifically robust background
data, which was presented in user-friendly and adapted tools for engineers. The
following stages were proposed to achieve relevant results compatible to industrial
applications:
Development and optimization of wood lightweight concrete using recycled
wood particles, new additives; application of recycled wood particles and new
additives,
Strength tests and evaluation of the physical properties of the lightweight
wood-concrete,
Design of wall components made of lightweight concrete connected to timber
sections,
Development of various sets of prototypes for testing,
Shear and bending tests of the prototypes,
Evaluation of the thermal and sound insulation behavior,
Analysis of the experimental results to develop design concepts,
Optimization of the manufacturing methods,
Studies regarding the ecological impact and the fire resistance by explorative
tests.
3.1.1 Development and Optimization of Lightweight Wood Concrete
As a structural model analogy on material level, the wood fibers in different

forms, i.e. saw dust, chips, strands, wool etc., in the wood-concrete compound
allow for a load-bearing truss embedded in the concrete, being able to transmit
compression and also a certain tension (Figure 2).
Different wood-concrete compound compositions can be arranged in the
section of a structural element such, that they provide the necessary structural
performance where it is needed:
higher compression strength in the compressive (bending) zone where little

tensile strength is required;
well-balanced tensile and compression strength in the section core to absorb
shear forces;
higher tensile strength and little compression strength in the tension zone to
support the structural timber elements in tension and for absorbing local stress
concentrations in shear connections to the timber components.
Fig. 2 Development of lightweight wood-concrete
3.1.2 Load Bearing Behavior of Floor Elements
Under typical bending load, the self-compacting concrete layer is stressed by

pressure, while a tensile force stresses the timber layer. The lightweight wood-
concrete layer in the middle connects these layers by shear forces; typically major-
ly close to the supports. Therefore, in this context the most important material
parameters of lightweight wood-concrete are the modulus of shear and the critical
shear stress.
Special attention was provided on the experimental research activity addressed
on the structural characterization of lightweight wood-concrete assemblies to be
used as wall or floor structures particularly on in-plane shear tests on large wall
elements (shear wall tests), as well as in bending tests on large floor elements. The
experimental campaigns were focused on the three most promising prototypes and
were performed on three specimens per prototype (Figure 3).
Fig. 3 Different variants of multi-layer floor elements made of timber, light wood-concrete
and self-compacting concrete)
The goal of the experimental analyses was the evaluation of the global behavior
of a building as well as derivation of further mechanical properties and optimiza-
tion of lightweight wood-concrete compositions. To this purpose, several tests
were scheduled in order to optimize the wood-concrete elements, and to obtain a
complete mechanical characterization (strength, stiffness) of those elements, so to
be able to use the values obtained for the structural modelling of a whole building
(Figure 4).
Fig. 4 Shear tests on wood lightweight concrete
3.1.3 Joints and Connectors

The load bearing behavior is not only dependent on the properties of the three lay-
ers, but also on the joints as well as the connectors between timber and concrete.
The lower joint between the timber and the lightweight wood-concrete layer
was subject of various shear and bending tests. The glued connection was the most
promising one. Due to the material parameters of the used lightweight wood-
concrete, the requirements concerning the mechanical properties of the tested
glues were not high. In order to test whether the inner strength of the glue was
higher than the inner strength of the lightweight wood-concrete.
The upper joint located between lightweight wood-concrete and self-
compacting concrete is fully connected since self-compacting concrete is
poured on the lightweight wood-concrete during the production process.
To increase the load carrying capacity of the multi-layer (sandwich) element
wood screws as shear connectors between timber and concrete were tested addi-
tionally. (Figure 5). Compared to the specimens with glued joints and without
screws there is a significant increase of resistance against shear failure and stiff-
ness of the connection. (Table 1).
Table 1 Comparison of shear connection with and without screws
without screws additionally 6 screws increase

Modulus of displacement k 45,000 N/mm 126,319 N/mm 281%
F 135 kN 354 kN 262%
32.8
Legend
100.0
CLT
SCC
wood-concrete (Velox)
profile for load introduction
Fig. 5 Shear test with mechanical connector
3.1.4 Bending Tests of the Prototypes
Figure 6 shows a specimen where shear connectors were used in the areas close to
the supports. In this specific test, three screws in one row on each end of the beam
were applied. The rows of screws had a distance of 25 cm.
The bending test performed with composite floor elements showed that the
wood screws were able to guarantee a rigid full composite action between timber
and concrete layers.
Fig. 6 Specimen BT 3 with 7.20 m length
3.1.5 Analysis of the Experimental Results

The most important material parameters of lightweight wood-concrete are the
modulus of shear and the critical shear stress. The mechanical properties of light-
weight wood-concrete, which considered in the structural modelling (based on
short-term tests), are as follows:
Modulus of shear - characteristic mean value:
N
= 13.5 (1)
mm
Modulus of shear - characteristic 5-percentile value:

N
= 11.3 (2)
mm
Shear resistance - characteristic mean value:

N
= 0.27 (3)
mm
Shear resistance - characteristic 5-percentile value:

N
= 0.21 (4)
mm
3.1.6 Critical Summary on the Experimental Results

The research activity was addressed on the structural characterization of light-
weight wood-concrete assemblies to be used as wall or floor structures. The
results were compared to the corresponding properties of regular structural con-
crete and preliminary proposals for structural design values for the lightweight
wood-concrete compounds are deduced [7].
Glue
Cement-based glues are tending to dry too fast. The timber and lightweight wood-
concrete soak up the water, therefore there is not enough water left for the glue to
harden properly. The quality of the joint and the processing time depend on the
processability of the glue. The bearing behavior depends on the quality of the
joint.
Screws
To create composite action between the lightweight wood-concrete compounds
and the structural timber, mechanical fasteners as shear connector are optionally
used in prototypes (Figure 7). It was easy and fast to process and handle the
screws. In the tests, the screws had not been pulled out of the self-compacting
concrete, but out of the timber and some of them ruptured. When some of the
screws or the joint between timber and lightweight wood-concrete fails the result-
ing displacement between wood and concrete layer leads to a combined tensile
and bending load of the screws, which are still intact.
Fig. 7 Multi-layer floor elements, screws as shear connectors

Lightweight Wood-Concrete Layer

In most of the tests, the rupture of lightweight wood-concrete on shear loads was
the first failure. However, in most cases the joint was not properly manufactured.
Timber and lightweight wood-concrete were not glued together in the total area,
which leaded to a higher local shear stress.
Timber and Concrete Layers

The failure of one or all of the shear connectors results in a rearrangement of
stresses. Normal forces in the outer layers change to bending moments, which
leads to higher normal stresses in these layers and a higher deflection. After these,
first failures cracks in concrete and failures in the timber layer appeared.
3.1.7 Developing Design Concepts

Practice-friendly approaches for the design of floor and wall elements made of
timber; lightweight wood-based concrete and self-compacting concrete were
developed, supported by illustrative examples for practical engineers, and verified
by comparisons to experimental test results. Analytical and numerical modelling
was envisaged for reliably predicting the structural behavior, to be subsequently
broken down into simpler models formulated for the daily engineering practice.
Embedment of the findings in codes, standards and regulations, is also sought.
Finite Element Method (FEM)

Preparation of analytical formulas were preceded by calibration of three-
dimensional numerical model - finite element method (FEM) - to gain even more
accurate information about stresses and strains in the system elements (Figure 8).
The results of the tests could satisfactorily be verified by FEM.
Fig. 8 Analysis model Finite element simulation of specimen BT 3; shear stress
Shear Analogy Method

The shear analogy method according to the DIN 1052 [8] is the most promising
method for calculating these systems statically. The bending and shear properties
of the single layers are transformed into a system consisting of two beams A and
B, where A takes all the bending stiffness values and B all the shear stiffness val-
ues (Figure 9). B can also take into account the flexibility of the joints.
beam A: = ; =
beam B: = ; =
Fig. 9 Sandwich layers and shear analogy
Where
(5)
This method is very flexible considering arbitrary number of layers, the static
system and the arrangement of the loads. Even local changes to materials can be
taken into account (e.g. screws).
Gamma-Method
The Gamma-method according to the EC 5, Appendix B [9] is an analytic method
for beams; consist of two or three parts, which are connected flexibly.
As it is not possible to calculate shear-flexible materials, the middle layer is
considered as a flexible joint between timber and concrete. One of the conditions
for mathematically correct solution is a constant stiffness of the joint, which is not
given when using screws near the supports. Nevertheless, it is possible to deter-
mine an equivalent stiffness of the joints, as if it was done for birds mouths in
wood-concrete beams [10].
At least this method is good enough for estimating the dimensions of the beam
and calculate deflections.
Comparison of Results
The following comparisons based on the tested specimen BT 3, which is shown
by Figure 6, while Figure 8 shows the corresponding FEM-model. Table 2 con-
tains the basic dimensions and mechanical parameters for the calculations.
Table 3 and Table 4 show comparisons between results of different kind of cal-
culation methods (deflection and normal stress). The level of load is dead load
( = 1.53 kN/m) plus a variable load of = 1.5 kN/m.
The Gamma-Method (Gamma) and the shear analogy method (SA) show
very good relations, as expected.
Two FEM models were calculated. The first one (FEM) is equivalent to the
other systems in Gamma-method and shear analogy method (no additional me-
chanical connectors); the second one (FEM with screws) implements screws at
the support (cf. Figure 8).
Table 2 Basic information of tested specimen BT 3
width b = 0.5m
length l = 7.2m
thickness modulus of elasticity modulus of shear
concrete 6.0 cm (*) 19,650 N/mm --

wood-concrete 15.0 cm ~1,000 N/mm (*) 13.5 N/mm
timber 11.8 cm 11,600 N/mm --
(*)results of tests on used materials
Table 3 Comparison of deflection in the middle of the beam
Gamma SA FEM FEM with screws test with screws

u (l/2) 11.7 mm 11.4 mm 10.6 mm 8.84 mm 8.1 mm
Table 4 Comparison of normal stresses in the cross-section at the middle of the beam
[N/mm] Gamma SA FEM FEM with screws

, -3.72 -3.57 -3.04 -3.04
, -1.17 -1.24 -0.83 -0.84
, -0.17
, 0.13
, -0.24 -0.11 0.58 0.59
, 2.73 2.61 2.32 2.31
4 Conclusions
The research project results benefit at several levels of the European industry.
Firstly, it offers utilization for waste products of the wood industry, thereby im-
proving the exploitation of the utilization chain of forest products. Secondly, it
provides a marketable component system for building construction. Thirdly, it
contributes to reducing carbon emissions by storing forest waste in construction
elements for a long period instead of burning it for warmth or energy. In a larger
context, such products support rapid-assembly construction methods that use pre-
fabricated dry elements to increase construction efficiency.
To reach a comparable or higher market share in the multi-story sector neces-
sary to guarantee a constant high demand of timber products able to absorb the
production capacity of the reforested forests timber-based buildings must be
economically competitive. The strategies to increase use of timber in multi-story
buildings are:
Combination of wood with cheaper, non-combustible mineral materials for cost

and fire resistance optimization,
Full use of prefabrication with the goal of reducing erection time,
Increase specific knowledge about timber by professionals, students and build-
ers,
Realize intelligent and attractive pilot projects.
Incontrovertibly, there are also certain pending questions related to the
research: Manufacturing performance of the multi-layer system and market
acceptance of the proposed building technology due to the economic competitive-
ness. The newly developed system shall be promoted through the initiation of pilot
applications. Initiation of pilot projects will provide the possibility to experience
the practicality and applicability of the developed composite systems, and create
different design and application solutions. Any modification proved to be neces-
sary at architectural and structural design stages may be reached as a result.
The elements provide a certain fire resistance, beneficial effects with regard to
building physics and thermal energy after use, and above all, lead to more exten-
sive and cascaded use of wood through substitution or completion of other build-
ing construction technologies.
Acknowledgments. The authors gratefully acknowledge the financial support of the Aus-
trian Science Fund (FWF, project number I 976-N20) and the Austrian Research Promotion
Agency (FFG, project number 824892), companies VELOX Werk Ltd., Dr Karlheinz
Hollinsky & Partner Ziviltechnikergesellschaft mbH and Architect Dipl.-Ing. Werner
Hackermueller for funding the research work. The partnership with the industrial company
VELOX Werk Ltd. manufacturing cement-bonded boards made of wood chips, allowed the
development and optimization of the production and manufacturing process.
References
1. Stingl, R., Zukal, M.L., Teischinger, A.: Holzbaustudie sterreich Stingl Teischinger,
Holzbauanteil in sterreich. Statistische Erhebung von Hochbauvorhaben. Zuschnitt At-
tachment att., 23, proHolz Austria, Wien (2011)
2. Decision No 406/2009/EC of the European Parliament and of the Council of 23 April
2009 on the effort of Member States to reduce their green-house gas emissions to meet
the Communitys greenhouse gas emission reduction commitments up to 2020
3. Fadai, A., Winter, W., Gruber, M.: Wood Based Construction for Multi-Storey
Buildings. The Potential of Cement Bonded Wood Composites as Structural Sandwich
Panels. In: World Conference on Timber Engineering, Auckland, New Zealand,
pp. 125133 (2012)
4. Winter, W., Fadai, A.: Mischbauansaetze beim mehrgeschossigen urbanen Bauen - Zur
moeglichen Rolle von Naturbaustoffen. In: Holz-bautage Innsbruck 2012 - Nachhaltige
Stadtentwicklung mit Holz, Universitaet Innsbruck; Holzbaulehrstuhl, Innsbruck, pp. 85
89 (2012)
5. Hochhauser, W., Winter, W., Fadai, A.: Entwicklung von verklebten Holz-
Glaskonstruktionen, Bemessung und Anwendung. In: Glasbau 2013, Technische
Universitt Dresden, Institut fr Baukonstruktion, pp. 185200. Ernst & Sohn, Verlag fr
Architektur und technische Wis-senschaften Berlin, Berlin (2013), doi:10.1002/
stab.201390063
6. Fadai, A., Tavoussi Tafreshi, K., Winter, W.: Wood-based construction in urban
context concepts for increased resource efficiency. In: DREVOSTAVBY 2013, a
Stredni PrMyslova Skola (Hrg.): Vyssi Odborna Skola a Stredni PrMyslova Skola,
Volyne, DREVOSTAVBY 2013, Tschechien, March 27-28, pp. 20132028 (2013)
7. Department of Structural Design and Timber Engineering (ITI): For-schungsprojekt
824892; Weitgespannte Flachdeckensysteme in Hol-zspanbeton-Verbundbauweise,
Research Report, Vienna (in progress)
8. DIN Deutsches Institut fuer Normung e.V.: DIN 1052: Design of timber structures
General rules and rules for buildings, Berlin (2008)
9. European Committee for Standardization (CEN): EN 1995-1-1:2009: Eurocode 5:
Design of timber structures - Part 1-1: General - Common rules and rules for buildings
(2009)
10. Winter, S., Kreuzinger, H., Mestek, P.: Holzbau der Zukunft, TEILPROJEKT 15
Flaechen aus Brettstapeln, Brettsperrholz und Verbundkonstruktionen. Fraunhofer
IRB Verlag (2009)
Rehabilitation, Upgrading and Repair of
Historic Timber Structures with Polymer
Concrete and FRP-Reinforcement
Markus Jahreis and Karl Rautenstrauch
Bauhaus University Weimar, Department of Timber and Masonry Engineering,

markus.jahreis@uni-weimar.de
Abstract. Historic load bearing structures are often made of timber. The repair of
elements, deteriorated by biotical influences or an upgrade for new fields of
application with higher loads can require special solutions and methods in
rehabilitation. Several studies with new technologies and materials for restoration
and strengthening have been carried out at the Bauhaus University Weimar.
Reinforcement and repair was realized by high performance materials in direct
bond with timber. An epoxy based resin with diverse mineral fillings was used to
create a material with high strength and good adhesivity. The properties of this
material can be adapted by the fillings in accordance to the application. This can
be used for instance as Polymer Concrete (PC) with high stiffness and
compression strength. Further materials are Fiber Reinforced Plastics (FRP) with
high tension strength for the use as reinforcement or glued-in rods to link
Keywords: rehabilitation, reinforced timber, FRP, rigid bond, Polymer Concrete.
1 Introduction
The restoration and rehabilitation of buildings and constructions has become rising
importance in Europe. For instance in residential buildings the percentage of work
contracts between rehabilitation works and new constructions is almost split to the
half. The energy life cycle balance of a building is mainly caused in its erection.
Therefore, rehabilitation is more and more indicated due to ecological and
economical reasons.
Rehabilitation of a construction means to replace destroyed elements or to
reinforce structure to enable higher duty. The traditional method of replacing parts
of wooden elements is to cut out the decayed areas and replace them with a new
piece of timber. The connection is mostly made of thick boards and dowel type
connectors. Another field of application is reinforcement of wooden structure,
because of leaks like knots, little cracks, cuts and recess or the need of higher load
DOI: 10.1007/978-94-007-7811-5_44, RILEM 2014
486 M. Jahreis and K. Rautenstrauch
capacity in future use. The reinforcement works in a similar way to the repair
systems. Larger boards or even steel sections are placed laterally to the structure
and connected by screws. Therefore, it is necessary to remove all the decking
construction and the inserted ceiling over the length of the area of repair and
further over the area to mount the connecting parts. This is a problem because of
the request for maximum preservation of the inventories especially in historic
buildings. Additional, the connection with dowel type fasteners is ductile and
needs high afford to reduce the deflexion.
For these reasons there is a request for repairing systems with low intervention
and rigid connections in rehabilitation of wooden structures.
2 Materials
Epoxy resin has a high adhesivity to different materials like minerals, wood, steel
and glass. The resin can be filled to control the properties. For high flexibility, the
pure resin-system with the trade name COMPONO was used in current
investigations. Silicates as flour, sand or grading in accordance to Fuller curves
were added. The material properties of selected formulations and further
explanations have also been published in Jahreis, Kaestner, Rautenstrauch:
Development of a high-performance hybrid system made of composites and
timber (High-Tech Timber Beam). They have been determined in experiments
with classic equipment and complementary methods.
The use of light weight materials with high load capacity is expedient for repair
and reinforcement. Fibre Reinforced Plastics (FRP) have high stiffness and
tension strength. There are Carbon Fibre Reinforced Plastics (CFRP) with a MOE
as high as this of steel and a tension strength which is more than five times higher
than usual steel. Stripes, made of Carbon Fibres embedded into an epoxy based
matrix with a width of 50 mm or 100 mm and a thickness of 1,4 mm to reinforce
bending systems, can be used. The advantages of these lamellas are almost non
weight, easy transport and high flexibility in application. For operation procedure
it is enough to open the ceiling from bottom side in the area the reinforcement has
to be placed. Thus, removing the inserted ceiling between the beams and the
decking structures is not necessary any more. In special cases, it is possible to
open the ceiling only for a gap. The CFRP-lamellas can be mounted to the cleaned
wood surface or into saw cut grooves with a filled epoxy resin. The presence of
CFRP reinforcement arrests the crack opening, confines local rupture and bridges
local defects or knots in the timber. The reinforcement carries the load as a more
ductile system in case of the failure of wood.
Another application is the repair of parts of the wooden structure. A connection
between the original wooden element and a renewed part can be realised with
glued in rods, for instance with Glass Fibre Reinforced Plastics (GFRP). The
material has low weight, the MOE has a triple value and the tension strength a 40
times higher value than that of wood. The benefit in comparison to steel rods is a
lower difference of stiffness between the material of the rod and the wood, so a
Rehabilitation, Upgrading and Repair of Historic Timber Structures 487
lower peak of stress at the end of the glue line is induced. There can also exist the
necessary demand for cutting the material without hot sparkles when used in
historic buildings, which is possible with GFRP by an ordinary hand-saw.
3 Methods and Experiments
3.1 Long Term Behaviour

Polymer Concrete (PC) can be used in areas with high compression forces. These
are the bearings, replaced end sections of timber structure or the compression zone
of bending beams. Additionally to the short term properties like diverse
characteristics of strength and elasticity, the long term behaviour is of great
interest. The polymer concrete consists of a polymer resin and a high percentage
of mineral particles. If the mineral material has a good balanced grading, the
influence of creep will be low. Rehm [5] published the results of investigations
about long term behaviour of filled polymer resins in 1980. He found the relation,
as higher the filling the less the creep. He found out that after 1 000 h the main
part of creep is done. If the pressure is not more than 50 % to 60 % of load
capacity, the gradient of creep is approximately linear after this period.
According to the investigation of Rehm, the following tests were done with
cylindric specimen with a diameter of 70 mm and a height of 210 mm. The
deformations are recorded each with three strain gauge in longitudinal as well as
in perpendicular direction. An epoxy based resin was filled with different Fullers
grading of silicate gravel with maximum grain of 3 mm or 7 mm. The mass
mixture ratio of resin and 3 mm-gravel was 1:5, the ration of resin and 7 mm-
gravel was 1:7. From maximum grain and mixture ratio the designation for the
specimen was built as PC_3-5 and PC_7-7. The test-load at PC_7-7 was at 25 %,
33 % and 50 % of the compression load capacity. For PC_3-5 this is at 25 % and
33 % and still in process. Rehm advices a test load at less than 40 % as the
standard for long term tests for aerated concrete EN 1355 does. There are no
standards for tests of PC in constructions. However, there is a guideline by the
German Center of Competence in Civil Engineering (DIBt) [2] with the advice of
a test load at 20 % of the load capacity. Further, there is a German standard for PC
in mechanical engineering DIN 51 290 T3 with the same rate of load in
compression test over a period of 90 days. The long time load is the dead weight
with approximately 60 % of the load in practical use for instance in ceilings for
combined bending beams of timber and PC. The live load with 40 % has mostly
an impact-time shorter than 1 000 h. The degree of efficiency is expected at less
than 33 % in practical application, if the appropriat safety factors and usual
utilisation are considered. The period under load for some specimen was more
than one year, while there was a time of relaxation after half a year because of a
necessary stop for service at the load machines. The curve of relaxation was
recorded too.
The curve of creep can be described as an extended exponential function and

can be used to forecast creep behaviour beyond the experiment. With the results of
measuring, it is possible to specify an equation for creep behaviour of a material.
The Burgers Model gives an expedient approach to describe the creep, as well
as the relaxation of synthetic polymers like the used PC but also for the natural
material timber (see Rautenstrauch [3]), which contains polymers too.
Fig. 1 Burgers Model; Creep and Relaxation Fig. 2 Test setting long term
according to Hying [1] behaviour under compression
load
The Model is a combination of a Maxwell Chain, a serial connection of

spring and dashpot to describe the viscoelastic part of deformation and a Kelvin-
Voigt-Element, a parallel connection of spring and dashpot to describe the
relaxing part. Creep and relaxation can be appreciated with the following
equations:
Creep under load:
t
k = s + c t d + a 1 e (1)

Relaxation without load after a time of stress:
t t2
t
k = s + c t d + a 1 e 1 e
(2)

In this juncture k can be a coefficient or absolute rate of upsetting or
elongation by creep. Due to s an offset can be chosen, which is commonly as
zero for coefficient kdef or the value of deformation after the loading is
completed. The next term contains the time related damping modified by d,
while c is the relation between stress and damping. The model describes the
elastic modulus as a Hooks spring and the coefficient of viscosity of the material
as a Newtonian damper. Third part of the equation includes the material properties
elasticity and viscosity as models of spring and dashpot too. Thereby, the factor
a includes the relation between stress and elasticity. The rate of change with
respect to time regulated by the damping is considered in the quotient . The
Kelvin-Voigt material has a good correlation for polymers. However, it is not

enough in case of relaxation, because it declines to zero, which is unrealistic for
viscoelastic materials with irreversible plastic flow. Therefore, the modified
equation is necessary for relaxation.
Figure 3 shows the experimental results and the evaluation with the Burgers
Model. There is a good correlation between test and modelling in case of
deformation under load as well as relaxation after the load is relieved. Tests with
different load and time were done. A forecast for a twelve years period with the
help of Burgers model is shown in figure 4.
Fig. 3 Test results and simulation with Fig. 4 Long term compression test with
Burgers Model for creep and relaxation forecast up to 12 years based on Burgers
Model
3.2 End Anchorage Zone

CFRP can be used to reinforce bending beams by surface mounting with an epoxy
based glue in a very powerful and rigid connection. There are high peeks of shear
stress at the end of the mounted partners because of the high difference between
the stiffness of timber and CFRP. This is a well-known effect in gluing materials
with different stiffness at all. The tension will be different and increases to the end
of the bounded zone because of the different stress at same elongation of both
materials lateral to the glue at the same force. Furthermore, the singularity in the
system of the intersection and in the stiffness leads to tension peaks.
Different bending tests on high reinforced timber beams with PC in the
compression area and CFRP in tension zone have shown that CFRP debonding
starts in the anchorage zone. The cracks run from both ends to the middle like the
unzipping of a slide fastener. The reinforcing becomes effectlessly and the system
breaks down. There are two approaches for troubleshooting. One possibility is to
reduce the stiffness of the adhesive within the anchorage zone, the other is the
enlargement of the bounding area. Reducing the stiffness of the adhesive can
moderate the intersection between the materials and the tension peaks will abate.
The handling and the allocation of different adhesives in a relative small area is
complicated and almost impracticable for realization at the building site. It is
easier to enlarge the anchorage zone. In this investigation, there was used the
possibility of double side mounting.
The effects at the anchorage zone have been tested with CFRP lamellas, one-
side-mounted to timber in a length of 25 mm to 200 mm and with double side
mounting. The failure always starts with a crack in timber at the end of the
glue-line near the adhesive surface and propagates deeper into the timber section.
So it can be postulated, that peaks of stress appear with reaction forces lateral to
the fibers of timber. Timber is an anisotropic material with a very small load
capacity in its orthogonal direction. The method of Close Range Photogrammetry
(CRP) was used to analyze these effects. The value of the deformation of the
timber in its longitudinal as well as in its perpendicular fiber direction could be
detected. Also the characteristic of the elongation of the CFRP-lamella could be
displayed, even if the elongation of the CFRP-Lamella is very small because of
the high stiffness compared to the small load. It was possible to show, that the
elongation has a higher rate in the beginning of the joint and decreases to its end.
The opposite result was seen at the timber as the deformation rises with the
distance to the abutting face. The observed effects decrease with separating
distance to the joint (Fig. 5). Furthermore, a displacement across the grain was
detected in the same degree as the longitudinal deformation.
Fig. 5 Elongation of the CFRP-lamella (left) and of the timber section nearby joint (right)
Further tests were done with the embedded end of the CFRP-lamellas in
different forms of the anchorage body. Thereby, the load could be allocated to a
larger area at the timber. One possibility is to carve a groove into the wood, launch
the lamella with a slight bending into the PC-filled grove and cover it with a
supporting material. This also has to be mounted beside the grove to distribute the
stress to a larger area. Sheets of CFRP can be used as belts with larger forces
perpendicular to the beam, for instance at jogs or gaps. If the CFRP-lamellas are
mounted into pre-cut grooves over the whole beam, circle shaped anchorage
bodies can be made by large drillings from side. In this case, the force at the
lamella will be supported directly to the grain of wood. In pull-out tests, the failure
for this case was at the lateral wood, which is not the problem at beams. Tests with
double side mounting show that the resistance at short glue line is almost as twice
as high as with one side mounting.
Fig. 6 Test with double side mounting in different shapes (left and centre); end anchorage
with supporting by CFRP perpendicular to the beam (right)
3.3 RepairSystem with Glued in Rods

Glued in rods, made of several materials, are convenient to create a repair system
for wooden structures with low intervention to the inventory. A butt joint can be
realised, while the bending, tension or transverse forces are conducted by the rods.
The connection is stiff and can carry high load if epoxy based glue is used. The
problem of stress peaks at the end of the rods is comparable to the mounted
CFRP-lamellas. The correlation of length of glue line and load capacity could be
confirmed in several pull out tests, as well as it is also known from literature. The
more interesting fact was detected with CRP at an opened specimen with glued
in rod. Therefore, a board was repaired with a gap as wide as the drill-hole and
filled with a PC and embedded steel rod. During the pull-test, the deformation
around the mounting zone could be recorded. The images of measurement are
implemented into a FEM model with special approach for the materials to display
the stress distribution around the mounted rod inside the resin and the timber.
Thus, it was possible to detect the significant change of stress distribution in the
moment when the bond to the end-grain failed. Until this point, the strain is
distributed almost equally over the length of lateral side. Figure 7 shows the FEM-
model with the values of the stress directly before the bond at the bottom of drill
hole fails (only the timber is displayed). The bond to the grain-cut erasures the
stress peak at the end of anchorage zone especially at small slenderness. With the
Fig. 7 Distribution of shear stress [MPa] in adjacent wood to the adhesive (PC). From left:
specimen; shear stress before bond failure at the end grain; initial crack; shear stress after
initial crack; failure. [4]
loss of the bond, at about 80 % of full load, the stress at the end of the lateral bond
line peaks out. The sudden shift of load leads quickly to the damage of the entire
connection.
If used as a repair connection for bending beams, the rods can be glued in the
lower area of cross section in accordance to reinforced concrete. Due to the stiff
connection, the bending moment can be transmitted as compression force by direct
contact and as tension force by the glued in rods. Several tests with different
versions of joints were investigated. There are the possibilities to link two pieces
of timber or timber with a replacement made of PC. To test the influence of
bending moment and cross force, the joint was placed to the middle of the beam
and nearby the bearings. The rods were glued by a flurry PC into drill holes over
both parts of timber lateral to the beam near the bottom side. It is very important
in this connection that there are connectors perpendicular to the grain to prevent
fracture by lateral force. The strain distribution around the joint was observed
during the bending tests and the assumption of stress distribution confirmed. By
this way, it is possible to create a connection with a very high load capacity. The
regulations for glued in rods can be used for dimensioning the reinforcement.
4 Conclusion
Polymer concrete, in convenient formulations, can be used for several applications

in rehabilitation systems. It is possible to use it for reinforcement of compression
zones as well as for surface mounting or gluing in reinforcement bars at the
tension zone of timber structure elements.
References
[1] Hying, K.: Analyse der viskoelastischen Eigenschaften von Poly (tetrafluorethylen) im
Bereich des beta-bergangs. Dissertation. RWTH Aachen. Fakultt fr Mathematik,
Informatik und Naturwissenschaften (2003)
[2] IfBt: Vorlufige Richtlinie fr die Kennwertbestimmung, Zulas-sungsprfung und
Gteberwachung von Reaktionsharzmrteln und Reaktionsharzbetonen fr
zulassungspflichtige Anwendungen. Mitteilung des IfBt (2), S.39S.44 (1977)
[3] Rautenstrauch, K.: Untersuchung zur Beurteilung des Kriechverhal-tens von
Holzbiegetrgern. Dissertation. Universitt Hannover, Hannover. Institut fr
Bautechnik und Holzbau (1989)
[4] Rautenstrauch, K., Hdicke, W., Kstner, M.: Contactless Meas-urement with Close-
Range Photogrammetry (CRP). In: COST FP 1004: Enhanced Mechanical Properties
of Timber, Engineered Wood Products and Timber. COST Action FP1004. Bath (Cost
Action FP1004, Zagreb, Croatia, April 19-20), pp. S.150S.157 (2012)
[5] Rehm, G., Franke, L., Zeus, K.: Kunstharzmrtel und Kunstharz-betone unter
Kurzzeit- und Dauerstandbelastung. Deutscher Ausschuss fr Stahlbeton Heft 309.
Hg. v. Deutscher Ausschuss fr Stahlbeton, Berlin (309) (1980)
Fatigue Performance of Single Span
Wood-Concrete-Composite Bridges
Leander Bathon1 and Oliver Bletz-Mhldorfer2
1
Faculty of Civil Engineering, Timber Structures and Building Technology,
HS-RM Wiesbaden University of Applied Sciences, Kurt-Schumacher-Ring 18,
65197 Wiesbaden, Germany
leander.bathon@hs-rm.de
2
TU Darmstadt and HS-RM Wiesbaden University of Applied Sciences,
Kurt-Schumacher-Ring 18, 65197 Wiesbaden, Germany
oliver.bletz@hs-rm.de
Abstract. Wood-concrete-composite structures in residential and commercial

applications are gaining market shares in recent years. The application of the
wood-concrete-composite bridges however is very limited. This is mainly because
there is a lack of knowledge on the fatigue behaviour of the wood-concrete-
composite bridges. The authors believe there is a great potential to design these
bridges using wood-concrete-composite cross sections with single span of 30
meters and more.
Keywords: wood-concrete-composite, glued in shear connectors, bridges, fatigue,

cyclic loading test.
1 Introduction
The use of wood-concrete-composite systems for highway bridges in Germany

was not possible in the past due to a lack of code approval of the shear connectors.
The research work of this paper however was the base for the first code approval
given in Germany for this kind of application. The authors believe that there is
now a great potential for highway bridges using the wood-concrete-composite
technology.
2 Testing
The performance of a wood-concrete-composite bridge depends solemly on the
performance of the shear connection. The shear connetors used in this application
is a hbv-shear connector which consists of a steel mesh of S 235 (equvalent to
A36). On half of the hbv-shear connector reaches 40 mm into a 3,2 mm wide
DOI: 10.1007/978-94-007-7811-5_45, RILEM 2014
494 L. Bathon and O. Bletz-Mhldorfer
channel within the timber and is secured through adhesive action. The other half
(50 mm) is reaching into the concrete. In order to provide a durable bridge system
it is desirable to use this stiff but ductile connection system.
2.1 Shear Test

The shear tests were conducted as push-out-tests. The specimens were 600 mm
long. The wooden cross section for the beam solutions was 80 by 140 mm and for
the plate solution was 400 by 80 mm. The concrete cross section was 70 by 400
mm. The hbv-shear connector reaches 40 mm into the wooden and 50 mm into the
concrete part of the composite system. All shear tests show steel failure within the
shear connectors under ultimate load conditions (Fig. 1). Figure 1 shows a great
plastic performance with almost now variation due to the steel failure of the shear
connectors.
Fig. 1 Shear testing on shear connector
2.2 Fatigue Testing

The fatigue testing was performed on the shear connector itself (tension shear) and
the wood-cocnrete-composite system (compresion shear). Again all test showed
steel failure within the shear connectors. Figure 2 shows the results of the fatigue
tests in the hbv-shear connector. The shear connectors were tested under a
25
20
[Load KN]
15
10
0
1,0E+00 1,0E+01 1,0E+02 1,0E+03 1,0E+04 1,0E+05 1,0E+06 1,0E+07
Cycles [-]
Distance 2 mm Distance 4 mm
Fig. 2 Fatigue testing on shear connector

Fatigue Performance of Single Span Wood-Concrete-Composite Bridges 495
sinustype tension ramp-load and a frequency of 2 to 20 Hz. The load was applied
at constant strain ratios = u / o. Two test-series differing in a gap between the
two clampings were conducted ( = 2 mm, = 4 mm).
The ultimate load in the short-time tests arise in dependancy of the gap to
Fu,2mm = 20,7 kN respectively Fu,4mm = 19,2 kN. The fatigue strength identified
for a an expected load cycles of 2 x 106 shows a value of 8,47 kN at a gap of 2
mm and 7,98 kN for the gap of 4 mm.
3 Analysis
3.1 Truss Model

The analysis of the wood concrete composite bridge is performed through a truss
model. The model is shown in Figure 3. The truss model consists of a top and
bottom member representing the concrete and wood respectively. The spring
diagonals represent the shear connector based on the shear test of the steel mesh.
The truss model provides both the prediction of the test performance as well as a
design tool for the commercial application of the system.
Fig. 3 Truss model
The analysis showed that the wood concrete composite system produces a
nearly fully composite action. The fatigue design is shown in the EC 5. In order to
perform the fatigue design you need the factor a and b (Fig. 4) for the shear
connector in use. The code approval of the hbv-shear connector shows a=2,5 and
b=4.
(1)
(2)
(3)
Fig. 4 Fatigue design (Eq. 1,2,3) of EC 5
496 L. Bathon and O. Bletz-Mhldorfer
3.2 Fatige Analysis

The fatigue analysis is based on the EC 5. It uses the cumulative linear fatigue
theorie of Palmgren-Miner. Figure 5 shows the comparison of the Palmgren-Miner
Rule for a strain ration = u / o.of =0,09. Figure 5 shows that the application
of the Palmgren-Miner-Rule will produce conservative design values for the
fatigue design of the wood-concrete-composite bridges.
Fig. 5 Fatigue analysis vs. test data
In recent years a number of wood-concrete-composite bridges have been build

using the data presented (Figure 6). The authors are convinced that with the
knowledge gained based on this research there will be more opportunities for the
timber industry to build wood-concrete-composite bridges throughout the world.
Fig. 6 WCC-bridge in Winschoten (NL)
4 Conclusions
This paper introduces a fatigue design approach for a single span wood-concrete-
composite bridge. The design is based on 200 shear test as well as 60 fatigue test
Fatigue Performance of Single Span Wood-Concrete-Composite Bridges 497
that have been performed at the Test Laboratories in Wiesbaden, Germany. The
test data provides the information needed to design the wcc-bridges according to
the code approval of the system used. The code approval paper determines the
stiffness and strength of the shear connector for static loading conditions as well
as the factors a/b for fatigue loadings conditions according to EC 5.
Acknowledgement. The reasearch was supported by the following companies: Derix,

Grber, TiComTec, Duscheck & Duscheck. In addition the federal agency of AIF
supported the research and therefore the Ph.D. work of the co-author Oliver Bletz-
Mhldorfer. The authors are very grateful of all support brought to them.
References
[1] Bathon, L., Bletz, O.: Holz trifft Beton, bauen mit holz 6/2008, Seite 28 33 (2008)
[2] Bletz, O.: Beitrag zur Entwicklung von Holz-Beton-Verbundk on struktionen mit
eingeklebten Streckmetallen. Dissertation in Arbeit, TU Darmstadt, Institut fr
Stahlbau und Werkstoffmechanik
[3] DIN - Deutsches Institut fr Normung. DIN 1074, Holzbrcken (2006)
[4] DIN - Deutsches Institut fr Normung: DIN 50100, Werkstoffprfung,
Dauerschwing-versuch, Begriffe, Zeichen, Durchfhrung, Auswertung (1978)
[5] DIN - Deutsches Institut fr Normung: DIN-Fachbericht 101, Einwirkungen auf
Brcken (2003)
[6] Kuhlmann, U., Aldi, P.: Fatigue of timber-concrete-composite beams:
characterisation of the connection behaviour through push-out tests. In: Proceedings
of the 10th World Conference on Timber Engineering, Miyazaki, Japan (2008)
[7] Bathon, L., Bletz, O.: Long term performance of continous wood-concrete-composite
systems. In: Proc. of the 10th World Conference on Timber Engineering, Portland,
Oregon, USA (2006)
[8] Bathon, L., Bletz, O., Bahmer, R.: Concrete bearings a new design approach in
wood-concrete-composite applications. In: Proc. of the 10th World Conference on
Timber Engineering, Portland, Oregon, USA (2006)
[9] Bathon, L., Clouston, P.: Experimental and Numerical Results on Semi Prestressed
Wood-Concrete Composite Floor Systems for Long Span Applications. In: Proc. of
the 8th World Conference on Timber Engineering, Lathi, Finnland, June 14-17,
vol. 1, pp. 339344 (2004)
[10] Fa. Duscheck & Duscheck GmbH Ingenieurholzbau, A-3032 Eichgraben,
http://www.d2-duscheck.at
[11] Bathon, L.A., Graf, M.: A Continuous Wood-Concrete-Composite System. In:
Proc., World Conference of Timber Engineering, Whistler, B.C. (2000)
[12] Fa. TiComTec GmbH, D-63808 Haibach,
http://www.holz-beton-verbund.de
Hybrid Wall-Slabs for Multi-storey Buildings:
Made of Timber with a Directly Applied
Mineral Cover Layer
Christian Dorn, Alexander Stief, Markus Jahreis, and Karl Rautenstrauch
Bauhaus University Weimar, Department of Timber and Masonry Engineering,

christian.dorn@uni-weimar.de
Abstract. A sustainable handling of the worldwide available resources requires an

innovative and comprehensive ecological concept especially for multi-storied
timber buildings. Therefore, profound investigations were done with the aim to
develop an efficient composite system, which combines the distinguished proper-
ties of wood with common mineral building material anhydrite. This mineral has a
high capacity to accumulate thermal energy and moisture for a balanced indoor
climate.
Anhydrite layered walls, made of stacks of boards or solid crosswise nailed
timber, are developed and investigated for modern residential multi storey build-
ings. The mineral layers are in direct connection with the timber. The serviceabil-
ity under vertical and horizontal load was examined. Numerical structural models
were developed and verified with the results of measurement by Close-Range
Photogrammetry (CRP) experimental tests.
Keywords: hybrid elements, shear walls, timber-anhydrite-slabs, mineral surface,

photogrammetry.
1 Introduction
Wood is the most ecological construction material with low death weight but very
good properties in load-bearing capacity as well as in physical behaviour. This
natural material is suitable for thermal protection as well as for thermal and mois-
ture compensation. There is a high renewable capacity of timber in the forests of
Europe, which has not been used as possible yet. Nowadays there is a great de-
mand on high quality material for producing glulam or scantlings for several uses.
Most of the timber has to be sorted for strength category C24 or better. However,
there accrues a high rate of timber with a worse category at the saw mills, which is
hard to sell. At full timber building components, also low quality wood can be
treated, even for load-bearing elements. While growing the trees assimilates a lot
DOI: 10.1007/978-94-007-7811-5_46, RILEM 2014
500 C. Dorn et al.
of carbon dioxide from the environment and accumulates it in the wood. So there
is a positive aspect in integrating also the timber with worse properties into the
buildings. At the assignment as construction material, it does not need much en-
ergy to process and the carbon dioxide is stored inside the construction. After a
practical lifetime, the material can be re-used as an energy source with a balance
almost at zero.
Two kinds of timber walls were used in the investigation. The first type was a
stack of boards, connected with hardwood dowels (see Figure 1). This kind of wall
has a high load capacity for vertical load but a low stiffness in case of horizontal
forces. Another type was a solid timber wall, with the trade name MHM - Massiv
Holz Mauer by Hundegger Maschinenbau GmbH (see Figure 2, [5]). It is made
of boards, which are connected cross wise with small threaded aluminum nails.
Therefore, the walls have a better rate of diffusion compared to CLT, where glue
is used for the connection of the boards. Additional, the insulation of the walls is
better because of small grooves in the boards, which encloses air.
Fig. 1 Shear wall element stack of Fig. 2 Shear wall MHM element [5]
boards
Anhydrite is a mineral, which can be used as a natural material with very low
need of energy while processing. It has solely to be mined, grinded and maybe
mixed with sand. There is no demand for energy intensive processes like kiln.
Anhydrite has a high capacity for accumulating thermal energy and moisture.
Furthermore, it has an excellent usability for decking because of its compression
strength and better tension strength compared to cement compound materials.
Both materials are combined to create a hybrid system for the use as walls and
ceilings in multi-storey residential buildings with low cost, comfortable life cli-
mate and an excellent carbon footprint. The inner surface of the timber wall and
the upper side of the ceiling are covered with a layer of mineral plaster or anhy-
drite floor. There are no metal connectors in this hybrid system. The fresh plaster
is placed directly on the wooden surface. In this context, it is helpful to use only
saw cut timber without planning. In addition, the little differences between the
single boards support the bonding. In case of walls, made of stack boards, this
connection is strong enough to increase the stability of the walls to be used as
shear walls.
Hybrid Wall-Slabs for Multi-storey Buildings 501
2 Design and Construction
Hybrid elements made of timber and anhydrite are a benefit for the use at residen-
tial buildings because of their excellent properties concerning the life climate. A
high bearing capacity for vertical load is heeded for the possibility to tread as to
use the wall elements for multi-storied buildings. Therefore, the board stacks are
placed in vertical direction over the whole height of the building. There are no
vertical beams planed for construction in the connection joint between the levels.
The horizontal forces are increasing in case of elevated constructions, e.g. of
higher wind load. These reasons are the base for investigations on shear stability
of board stack elements with reinforcing by decks of anhydrite floor or plaster.
The tested walls are made of 45 spruce boards with a cross section of 4 x 10 cm
and a length of 250 cm in customary quality. They are connected with eight hard-
wood dowels with a diameter of 20 mm. The dowels are placed every 30 cm over
the length of the wall.
In addition to wall elements, consisting of board stacks, MHM elements were
examined. These consist of crosswise layered boards, connected by threaded alu-
minium nails. Grooves are cut into the external boards, for a better mechanical
interconnection of the plaster with the wooden surface. This could be done while
prefabrication with the joinery machine. The surface of the wood was treaded with
special vapour permeable primers to counter high water impact while the process
of plastering ([5], see Figure 3 and 4).
Fig. 3 Application of plaster on the Fig. 4 Plaster on profiled wooden surface

MHM-element
In a first step the shear stability of pure board stacks and MHM walls were ana-
lysed. Afterwards a 3-4 cm strong layer of anhydrite floor or plaster was placed on
the surface of the wooden elements. In addition, the convenience anhydrite screed
or plaster was modified for instance with wooden fibres, to equalize the water
content during processing and to reinforce the anhydrite to resist the cracking
caused by the shrinkage while hardening or under load.
502 C. Dorn et al.
For the experiments, a steel frame was created to investigate whole size walls with
variable vertical load. By technological reasons, the displacement of the top of the
wall, expected from the wind, was simulated by displacing the bottom of the wall.
Several modes, simulating the positioning of the wall inside the building, can be
tested by restraining the torsion of the bearing (see Figure 5).
Fig. 5 Test Modes for simulating diverse mounting situations of the wall inside the building
[1]
The elements were placed at the steel frame and fixed over the whole length
with customary screws like they could be used to assemble the walls on building
site. During the test the death load, depending on the height of the building, was
simulated by the vertical load generated with constant ballast. A hydraulic jack
system was used for displacing the wall element at the bottom. The reaction forces
were measured by electronic load cells, the displacement by linear variable dis-
placement transducers (see Figure 6). In addition, the deformations of the hybrid
elements were monitored by 3D-Photogrammetry. Therefore, two high resolution
greyscale CCD-cameras were used. Targets at the specimens helped to detect de-
formations in each direction (see Figure 7). On this way, it was possible to visual-
ise the stress distribution over the whole wall.
Fig. 6 Test frame with embedded shear wall Fig. 7 Close Range Photogrammetry
Every wall element was tested as row timber element and with a decking made
of modified anhydrite floor or plaster. The basic tests for every test-specimen were
done by displacing the bottom of the wall element according to the load history,
shown in Figure 8. Furthermore, tests were concluded with repeated cycles in
different deflexion up to 60 mm in each direction.
Fig. 8 Load history (left: basic tests, right: cycle tests)
The MHM walls do not need reinforcement. The load capacity for shear force
is high enough to carry the horizontal load, for instance induced by wind impact.
The assignment is, to find a formulation of the mineral layer, which endures the
deformation of the wall. This is important for earthquake resistance. The wall for
itself is yieldable enough to absorb much energy because of the many small nails.
However, it is important, that the connected structural elements are also resistant
and do not become a hazard for the inhabitants.
The quantity of the dissipated energy Ed can be determined using the enclosed
area of the hysteresis curves. The equivalent viscous damping coefficient eq as
ratio of dissipated energy and potential energy Ep can be used to compare different
systems (Figure 9).
E
= d (1)
eq 4 E
p
Due to the multiplicity of nails, which connect the layers of timber, the MHM-
element offers an acutely ductile behavior under cyclic loading. Figure 9 shows
the deformation behavior of a MHM-wall element. The entire energy dissipation
of the wall element is characterized by the plastic behavior of each connector (af-
ter reaching the yield moment), the embedment strength and the friction between
the single boards. Compared with framework walls and cross-laminated timber
(CLT), the stiffness of the MHM element is lower and more ductile. Thus, energy
could be dissipated over the whole area of the wall element, not only in local
zones. Especially the hold downs and the connecting brackets, made of slim metal
sheets, are the failing elements at very stiff walls under cycling load. The equiva-
lent viscous damping coefficient of the raw MHM elements was determined with
eq = 0.195 (see equation (1)) in the last cycle with the maximum displacement.
504 C. Dorn et al.
During the first cycles, with a lower displacement, the coefficient was measured
with eq = 0.23. This amount is essentially dependent on the vertical load and the
mounting situation of the wall in the testing-frame. Therefore, for a better compa-
rability, all shear walls were tested with a constant vertical load of 25 kN/m.
Fig. 9 Determination of equivalent viscous damping coefficient of a raw MHM element;

cycle with the maximum displacement, [4]
The comparison between raw board stacks and a MHM element without deck-
ing of plaster or screed shows significant differences in stiffness. Due to the multi-
tude of nails, which connect adjacent layers, the MHM element has a considerable
higher stiffness (see Figure 10, left). Furthermore, the stiffness of the hybrid ele-
ments was increased effectively by the direct interconnection between timber and
the mineral covering layer. The decrease of stiffness after three cycles with a dis-
placement of 30 mm in each direction is very low, as shown in Figure 10 at the
right side. Therefore, a nearly faultless interconnection could be assumed. Further
investigations under dynamic load, e.g. to determine the fatigue behaviour are still
a matter of ongoing research.
Fig. 10 Force-displacement curve, left: tested types of raw elements, right: stiffness-
increasing by application of plaster in direct interconnection to the raw timber element by
MHM
There are obviously effects in raising the stiffness of the board stacks by rein-
forcing with anhydrite plaster. These effects are demonstrating the potential of the
ecological material anhydrite for load bearing building structures. Further investi-
gations should consider the opportunities of the material and the natural bond to
increase the load bearing capacity. This natural bond is induced on one hand by
adhesive reactions and on the other hand by embedded fibres. Because of this
sterical connection with the fibres, the ductility increases in the same way as the
reliability in serviceability.
The load-deformation-curve with different mineral layers at stack of board
elements is shown in the next diagram (Figure 11). The profile of the elements
surface was varied. A strong profiling of the surface influences the stiffness signif-
icantly after the mineral top layer is applied.
Fig. 11 Comparison of stiffness between different versions of plaster
Several kinds of application technologies and different formulations of plaster

were tested at wall elements with a non-profiled and a profiled surface. The influ-
ence of the materials and technologies is shown in the next diagram. First tests are
made at wall elements without profiled surface, whereby the saw cut boards are
mounted parallel in one plane. Customary gypsum lime plaster was applied wet in
wet in two layers and compared with a special formulation of plaster on the base
of anhydrite as ground layer with dry application machine in combination with a
final pass of gypsum lime plaster. For dry application, the material is transported
with high pressure through the pump and hose and is mixed with water directly at
the nozzle. The advantage of this technology is the lower percentage of water in
the material and the high application pressure. Thus, a good interconnection
and less impact of water to the timber is induced. For further tests the wall ele-
ments get a profiling due to the integration of boards with larger or smaller depth
with rough surface. The plaster can run into the notches whereby the load capacity
and the stiffness perpendicular to the boards increase obvious. The same applica-
tion technologies and variations of plaster are tested with these kinds of wall
surface.
506 C. Dorn et al.
4 Conclusion
The investigations with hybrid elements made of timber and mineral plaster con-
firm the suitability of excellent ecological materials for load-bearing systems. It is
possible to use mineral plaster for load bearing systems in combination with tim-
ber. The interconnection is realized by natural bond with a high shear resistance as
well as a high ductility for cycle load history. Further investigations can extend the
possibilities of this kind of mounting and have to find customized methods of
calculating.
References
[1] Rautenstrauch, K., Dorn, C., Jahreis, M., Mller, J., Stief, A.: Innovative nachhaltige
Bauwerke durch effiziente Kombination von nachwachsenden Rohstoffen und einfach
in Kreislufe integrierbaren mineralischen Baustoffen in einer Hybrid-Bauweise,
Abschlussbericht zum Forschungsvorhaben 22024505 der Fachagentur fr
nachwachsende Rohstoffe (2008)
[2] Grosse, M., Rautenstrauch, K.: Numerical modelling of timber and connection
elements used in timber-concrete-composite constructions. In: Proceedings CIB-W18,
International Council for Research and Innovation in Building and Construction
Commission W18 Timber Structures, Meeting 17, Edinburgh UK (2004)
[3] Lehmann, S.: Untersuchungen zur Bewertung von Verbundbauteilen aus
Brettstapelelementen im Flchenverbund mit mineralischen Deckschichten. Disserta-
tion, Bauhaus-Universitt Weimar (2004)
[4] Chopra, A.K.: Dynamics of Structures, 3rd edn. Prentice Hall (2006)
[5] Z-9.1-602, MHM-Wandelemente (Massiv-Holz-Mauer-Wandelemente), MHM-
Entwicklungs GmbH, Pfronten-Weibach, Deutsches Institut fr Bautechnik (DIBt),
valid until 04.04.2016 (2011)
[6] EN 1995-1-1: Eurocode 5: Design of timber structures, General rules and rules for
buildings (2005)
An Innovative Prefabricated Timber-Concrete
Composite System
Roberto Crocetti1 , Tiziano Sartori2, Roberto Tomasi2 , and Jose L.F. Cabo3
1 Department of Structural Engineering, Lund University, Sweden
roberto.crocetti@kstr.lth.se
2 Department of Civil, Environmental and Mechanical Engineering,
University of Trento, Italy
{tiziano.sartori,roberto.tomasi}@unitn.it
3 ETS of Architecture, Polytechnic University of Madrid (UPM), Spain
jose.fcabo@upm.es
Abstract. A novel type of timber-concrete composite floor, consisting of longitudi-

nal glulam beams with a fibre reinforced concrete (FRC) slab on the top is proposed.
In order to check some relevant mechanical properties of such a floor, full-scale lab-
oratory tests along with numerical analyses were carried out. The shear connector
system used in the investigation consisted of self-tapping screws driven at an angle
of 45 to the grain direction of the glulam beams. The manufacture of the structure
occurred according to the following steps: (a) the screws were inserted on the top of
the glulam beams; (b) the beams were rotated 180 about the longitudinal axis and
placed in a concrete formwork; (c) the FRC was cast into the formwork; (d) after
curing of the FRC, the composite floor was again rotated 180 about the longitu-
dinal axis into its right position, i.e. with the FRC slab on the top side. Long term
tests and quasi-static bending tests were performed. It was found that the proposed
connection system showed a very high degree of composite action both during the
long-term testing and at load levels close to the failure load. Furthermore, the as-
sembly of the prefabricated timber-concrete composite system revealed to be very
fast and easy.
Keywords: timber-concrete composite, fiber reinforced concrete slab glulam, self-

tapping screws, floor system, long-term test, load-carrying capacity, stiffness.
1 Introduction
Timber-concrete composite structure consists of timber beams effectively intercon-
nected to a concrete slab cast on top of the timber members. Most of the studies
performed to date have focused on composite systems where wet ordinary concrete
was cast on top of timber beams with mounted shear connectors. Even though such
systems have proven to perform very well from the point of view of statics and dy-
namics, in-situ concrete casting has some clear disadvantages, for example, waste
DOI: 10.1007/978-94-007-7811-5_47, c RILEM 2014
508 R. Crocetti et al.
of time due to concrete curing, low stiffness and high creep, concrete shrinkage ef-
fects on the composite beam, high cost of cast-in-situ concrete slabs, etc. Recently,
composite systems where the concrete slab is prefabricated off-site with shear con-
nectors already embedded and then connected to the timber beams on site have been
investigated ([4] and [7]). The research presented herein focus on the use of com-
posite structure with very high prefabrication level, good performance and short
construction time. Such composite structures are floor modules consisting of two
glulam beams with a a concrete slab on the top.

In order to investigate the behaviour of the proposed composite system, three full-
scale floors were built at the laboratory of Structural Engineering, Lund University.
The main dimensions of the floor system are reported in Table 1.
Table 1 Geometry of the A, B and C composite system (dimensions in [mm]). See also
Fig. 1.
Span (l) Slab width Slab thickness(h1 ) Beam width (b2 ) Beam depth (h2 ) Beam spacing (i)
7200 800 50 115 360 585
The geometry of the three tested floors was nominally identical. The timber used
for the manufacture of the floors was glulam GL30c. The moisture content of the
beams was approximately 12%. For the production of 1 m3 of fiber reinforced con-
crete, 45 kg of steel fibres and 480 kg of cement were used, which gave a mean
value of compression strength fc = 51 MPa for the concrete. In the following text,
the three tested specimens will be referred to: A, B and C. Specimens A and B were
tested on short term bending, whilst specimen C was tested on long term bending.
During the short-term bending tests, the load was applied by an actuator in a dis-
placement controlled manner. The load was distributed on four lines perpendicular
to the longitudinal direction of the floor in order to induce stresses and deformations
in the floor similar to those induced by a uniformly distributed load q, see Fig. 2.
The total load applied to the specimen, the mid-span deflection, and the relative
slip between slab and beam at the supports were continuously measured during test-
ing. For the long-term bending test, a uniformly distributed load of 1 kNm2
was applied
on the slab by means of sacks of cement. The mid-span deflection was measured
over time in order to investigate the creep effects of both timber and concrete.
2.1 Shear Connectors and Manufacture of the Floor Systems

In order to achieve composite action between the timber beams and the concrete
slab, self-tapping screws with dimensions d = 11 mm and l = 250 mm were driven
into the timber beams before the concrete was cast. The screws were driven at an
An Innovative Prefabricated Timber-Concrete Composite System 509
Fig. 1 Geometry of the floor system
Fig. 2 Test setup
angle of 45 to the longitudinal directions with a spacing of 200 mm close to the

supports and 300 mm in the middle part of the floor respectively (Fig. 3). The main
function of the inclined screws is to transfer the shear force from the concrete slab
to the timber beam both by shear in the direction parallel to the slip interface and
mainly by tension in the direction of the screw axis. The ultimate tensile strength of
the screws was approximately fu = 1250MPa.
The screws were inserted on the top of the glulam beams, then the beams were
rotated 180 about the longitudinal axis and placed in a concrete formwork. The
FRC was cast into the formwork, see Fig. 4. After curing of the FRC, the composite
floor was again rotated 180 about the longitudinal axis into its right position, i.e.
with the FRC slab on the top side.
Fig. 3 Screw positions
(a)
(b)
Fig. 4 (a) Timber beam with inserted screws- (b) upside down floor system (including the
formwork) directly after concrete casting
3 Test
3.1 Preliminary Tests
Preliminary tests were performed in order to obtain modulus of elasticity of timber
beam, strength of concrete, withdrawal resistance of screw to concrete connection
and strength of steel screws.
Table 2 Geometry and compression strength of the tested concrete cubes
ID Side Side Depth Compression

strength
a b h m
# [mm] [mm] [mm] [N/mm2 ]
A 150 150 150 51, 11
B 150 150 150 50, 89
C 150 150 150 51, 11
Table 3 Ultimate withdrawal capacity of the screws inserted into the concrete with different
penetration lengths
ID specimen penetration lengths [mm] Ultimate tensile capacity [kN]

A 50 27, 49
B 50 19, 75
C 50 23, 90
Mean value 23,71
A 75 40, 03
B 75 40, 04
C 75 38, 32
Mean value 39,46
A 100 40, 68
B 100 42, 92
C 100 40, 58
Mean value 41,39
Standard compression tests were carried out on three concrete cubes. The geom-
etry and the compression strength of the tested specimens are resumed in Table 2.
Screws with different penetration length were inserted in the concrete cubes. The
value of withdrawal tests on these specimens are reported in Table 3.
Non destructive bending tests were performed on two timber beams in order to
estimate the modulus of elasticity. The results are reported in Table 4.
Table 4 Modulus of elasticity and density of timber beams
ID Lenght [mm] Depth [mm] Width [mm] m [kg/m3 ] Em [MPa]

A 7198 355 112 477 12480
B 6595 356 111 449 12189
3.2 Short-Term Bending Tests

Load-deflection curves and load-slip curves are shown in Fig. 5 and Fig. 6
respectively.
Equivalent uniformly distibuited load q ( kN

m2
)
Floor A
80
Floor B
EImax
EImin
60
40
20
0
0 20 40 60 80
Deflection at mid-span f (mm)
Fig. 5 Equivalent uniformly distributed load vs mid span deflection
2.5
Floor A
Floor B
2
Slip at the support (mm)
1.5
0.5
0
0 20 40 60 80
Equivalent uniformly distibuited load q ( kN
m2
)
Fig. 6 Load slip deflection vs equivalent uniformly distributed load
The curves in Fig. 5 show the relationship between the equivalent uniformly dis-
tributed load q (i.e. the total load applied divided by the slab area) and the deflection
f at mid-span. As it ca be observed, the behaviour is linear up to a load level of ap-
proximately 80 kN m2
, which is well above the design load used in design of common
floor structures. The stiffness of both composite floors shows, after a slightly non-
linear initial part, a constant trend up to the failure of one of the two timber beams.
The curve of Fig. 6 shows the slip at the support related to the equivalent distributed
applied load. Failure of floor A occurred at a load q 84 kN m2
, with the propagation
in one of the two beams of two large cracks in the direction parallel to the grain.
The failure of the floor type B, on the other hand can be attributed to local failure of
a finger joint of the lowest lamination located close to mid span of one of the two
beams. The collapse of floor A occurred at q=80 kN m2
firstly due to bending failure at
a finger joint in one beam and secondly due to a shear failure located along a line
running through the tips of the screws used as shear connectors.
3.3 Long Term Bending Test

For the long-term bending test, a uniformly distributed load of 1 kNm2
was applied
on the slab by means of sacks of cement(see Fig. 7). The purpose of the long-term
test was to investigate the time-dependent behaviour of the prefabricated timber-
concrete composite system. The long-term test results for the specimen B is pre-
sented in Fig. 8 in terms of time vs mid-span deflection. The variables monitored
during the entire test were the mid-span deflection through 2 inductive transducers,
positioned at the mid-span of each glulam beam.
Fig. 7 Long term test setup
Also the temperature and the humidity in the laboratory was continuously moni-
tored. The value of these parameters are presented in Fig. 9.
As it can be seen in Fig. 8 the mid-span deflection increase to about three times
the instantaneous deflection after three months. In March the floor has been down-
loaded for two days. When the floor was reloaded it achieved again the same level
of deflection as before unloading. This operation was performed again in the first
days of May and similar results were obtained.
Fig. 8 Increase in mid-span deflection of the floor C with time
Fig. 9 Relative humidity and temperature observed in the laboratory during the period 01-
02-2013 / 30-11-2013
4 Efficiency of the Composite Beams

Deformations at the shear connectors generate horizontal movement, i.e. slip at the
interface between concrete and timber. Such a behaviour is due to as partial com-
posite action and, as the slip increases it reduces the efficiency of the cross section.
The efficiency of a shear connection for a composite beam can be estimated using
the following equation, see [9] and [10].
EIreal EImin
= (1)
EImax EImin
where is the efficiency, EImax is the bending stiffness of the floor with full com-
posite action, EImin is the bending stiffness of the floor with no composite action
and EIreal is the actual bending stiffness of the floor. At load levels comparable to
those at the serviceability limit state (i.e. 1 kN
m2
) the efficiency is approximately
1.0. The efficiency at a load of 20 m2 or more remains constant,i.e. 0.85.
kN
5 Conclusions
This paper presents the main results of a research project conducted on a novel
prefabricated timber-concrete composite system. The system provides several ad-
vantages compared to cast in-situ concrete slabs, e.g. reduced time of construction
and considerable reduction of the effects of concrete shrinkage. Experimental tests
were carried out on three 7.2 m long strip floor specimens to investigate on stiff-
ness and the strength and stiffness of the prefabricated systems, of which two tests
to failure and one long-time test. The principal observations from the experimental
investigations are:
The tested system showed considerably higher stiffness and strength properties
than a similar system with concrete deck not able to transfer shear stress
The load carrying capacity was very high. The equivalent uniformly distribute
load at failure was approximately 80 kN/m2 , which is considerably larger than
common designs load for floor structures.
The stiffness of the system was also very high. This depends primarily on the
ability of the shear connectors to transmit shear without (or with minor) slipping.
In the tested specimens the efficiency of the system was approximately 1 for
loads well above common design loads. For extremely high loads (q 20kN/m2 )
the efficiency of the system was approximately 0.85, which is also very high
compared to similar composite system with more traditional shear connectors,
i.e. screws or bars inserted perpendicularly to the plane of the slab.
The instantaneous deflection of the floor increased roughly by a factor 3 after a
period of approximately 7-8 months. This relatively large increase in deflection
is believed to be due mainly to the creep of the concrete slab and of the timber
beam and - in some minor extent - to the long-term deformation of the shear
connectors.
Last but not least, the easiness of manufacture of the proposed system should
not be underestimated, since it allows for a quick construction whit a reduced
possibility of human errors.
Acknowledgements. The authors wish to gratefully acknowledge the Mr. Franco Moar who
has performed his Masters thesis on this topic. The timber material was supplied by the
glulam mill Moelven Toreboda AB, Treboda, Sweden. The screws were supplied by Rotho
Blaas srl, Cortaccia, Italy. The fibers for the FR concrete were supplied by Bekaert Svenska
A.B. All the suppliers are kindly acknowledged.
References
1. Bathon, L.A., Bletz, O., Bahmer, R.: Concrete bearings a new design approach in wood-
concrete-composite applications. In: Proceedings of World Conference Timber Engineer-
ing, Portland - Oregon, USA (2006)
2. Blass, H.J., Bejtka, I.: Screws with continuous threads in timber connections. In: Inter-
national RILEM Symposium on Joints in Timber Structures, pp. 193201
3. Moar, F. Prefabricated timber-concrete composite system. Master thesis, Lunds Tekniska
Hogskola, Lunds Universitet (2012)
4. Crocetti, R., Sartori, T., Flansbjer, M.: Timber-Concrete Composite Structures with Pre-
fabricated FRC Slab. In: Proceedings of World Conference Timber Engineering, Riva
del Garda, Italy (2010)
5. Gutkowski, R., Brown, K., Shigidi, A., Natterer, J.: Laboratory tests of composite wood
concrete beams. Construction and Building Materials 22(6), 10591066 (2008)
6. Lukaszewska, E.: Development of Prefabricated Timber-Concrete Composite Floors.
PhD thesis, Department of Civil, Mining and Environmental Engineering Division of
Structural Engineering, Lulea (2009)
7. Lukaszewska, E., Johnsson, H., Sthen, L.: Connections for Prefabricated Timber Con-
crete Composite Systems. In: Proceedings of World Conference Timber Engineering,
Portland-Oregon-USA (2006)
8. Sjostrom, A., N.-Montero, J., Bard, D., Crocetti, R.: Vibratory investigation of a fiber
reinforced concrete floor supported by wooden beams: part I. In: Proceedings of Joint
Baltic-Nordic Acoustics Meeting, Odense, Denmark (2012)
9. Piazza, M., Turrini, G.: Sulle strutture composte legno-legno. In: Proceedings Italian
Workshop on Composite Structures, Department of Mechanical and Structural Engineer-
ing, University of Trento, Villa Madruzzo, Trento, Italy, pp. 349370 (1993)
10. Piazza, M.: Restoration of timber floors via a composite timber-timber solution. In:
RILEM Workshop Timber: a Structural Material from the Past to the Future, Trento,
pp. 167187 (1994)
Part V
Cyclic, Seismic Behaviour
A Component Model for Cyclic Behaviour
of Wooden Structures
Giovanni Rinaldin and Massimo Fragiacomo
DADU Department of Architecture, Design and Urban Planning, University of Sassari,

Piazza Duomo, 6, Alghero, 07041, Italy
{grinaldin,fragiacomo}@uniss.it
Abstract. Wooden structures have become widespread in several regions of the

world, including earthquake-prone areas. An estimation of their dissipative capac-
ity is of fundamental importance for an accurate design for seismic actions. To
reproduce the structural behaviour under earthquake excitation, a numerical model
was developed and validated on experimental results. The cyclic behaviour of
connections for cross-laminated (X-lam) buildings, light-frame construction and
moment-resisting frames were schematised by a piecewise linear hysteretic rela-
tionship taking into account strength/stiffness degradation, pinching and friction
between elements. This relationship was assigned to a non-linear elasto-plastic
spring implemented in Abaqus software package. The spring has to be calibrated
on experimental tests performed on single connections, which are the only com-
ponents dissipating energy in the structure. All the other timber components (X-
lam panels, timber beams, sheathing panels, etc.) are regarded as linear elastic. In
this work, several examples of cyclic analyses of wooden structures are presented,
including X-lam buildings, light-frame construction, and moment-resisting frames.
Numerical predictions are compared with experimental results demonstrating the
effectiveness of the model.
Keywords: wooden structures, cyclic behaviour, timber connections, hysteretic

rule, pinching effect, X-lam buildings, light frame constructions.
1 Introduction
In the last decade, timber multi-storey buildings have rapidly spread all over the
world. Advantages such as sustainability, aesthetic appearance and high strength-
to-weight ratio are among the main reasons for their increasing success. Different
structural systems can be used, including cross-laminated (X-lam) panels, light-
frame construction, and moment-resisting frames. An important issue is the seis-
mic resistance of these systems. Although timber is light-weight and, as such,
subjected to reduced seismic actions compared to heavier construction materials
DOI: 10.1007/978-94-007-7811-5_48, RILEM 2014
520 G. Rinaldin and M. Fragiacomo
such as reinforced concrete, it is inherently brittle. Energy dissipation can only

occur in the connection systems where there are proper details to ensure ductile
behaviour.
Investigation of the seismic behaviour of timber structural systems is not
straightforward. Eurocode 8 [6] contains little provisions for seismic design, and
the X-lam system is not even mentioned explicitly. The cyclic behaviour of timber
systems and connections is rather complex as it is characterized by pinching, post-
peak softening branch, strength and stiffness degradation. Widespread software
packages do not provide specific hysteretic rules specifically developed for timber
systems and connections. On the other hand the use of advanced models where the
cyclic behaviour of timber systems and connections is properly modelled is crucial
to extend the experimental results of small components (e.g. a hold-down connec-
tor for X-lam structures, a nailed connection between wood framing and sheathing
panel, a beam-column joint) to larger subassemblies (e.g. a X-lam wall, a light-
frame wall, a moment-resisting frame) and to entire building systems. These mod-
els can be used, for example, to derive values of the behaviour factor q without the
need for expensive shaking table tests such as those carried out in Japan on X-lam
buildings [1].
The aim of this paper is to discuss the use of a newly developed component
model for cyclic behaviour of timber connections in different timber systems.
After introducing the basic of the proposed hysteretic rule, the model is calibrated
to results of cyclic tests carried out on connectors for X-lam, nailed connections
used in light-frame construction, and a beam-column moment resisting joint. The
model is then used to predict the cyclic behaviour of a light-frame wall, a mo-
ment-resisting frame, entire X-lam and light-frame buildings that were experimen-
tally tested. The comparisons between predicted and measured values demonstrate
the accuracy of the proposed model.
2 Component Approach
The developed approach has been initially developed to model X-lam walls and
subassemblies, with the aim to predict the cyclic (seismic) behaviour with high
accuracy. Each connector (angle bracket, hold-down, self-tapping screws) is
schematized with a non-linear spring characterized by a hysteretic behaviour [15].
The model has been implemented through a non-linear spring element in a wide-
spread software package such as Abaqus using an external user subroutine written
in Fortran.
2.1 Constitutive Model of the Components

Each metal connection has been modelled as a non-linear spring with hysteretic
behaviour. The actual curves have been approximated with piecewise linear rela-
tionships, more specifically tri-linear curves, which have been parameterized to
A Component Model for Cyclic Behaviour of Wooden Structures 521
allow the user to fully control their shape. Two different types of curve have been
developed: for connectors with symmetrical behaviour, and for connectors with
non-symmetrical behaviour. Each curve is made of several branches composing
the backbone curve and the hysteretic cycle.
The non-linear spring has zero length as it connects two coincident points in the
undeformed state. In the most general case, every spring returns to the solver the
three forces that develop in its plane and the corresponding three stiffness values.
Although only planar springs with three degrees of freedom have been considered
for the sake of simplicity as this is the most important case, the theory can be eas-
ily generalized to the case of a spatial spring with six degrees of freedom.
2.2 Hysteretic Relationships

Figure 1a displays the shear force vs. shear displacement (slip) piecewise linear
relationship used to model the behaviour of symmetric connectors such as screws
and angle brackets loaded in shear. Figure 1b displays the axial force vs. axial
displacement piecewise linear relationship used to model the behaviour of non-
symmetric connections such as hold-downs and angle-brackets axially loaded.
Nine independent input parameters and thirty-two state variables are used to fully
describe the cycle. Four additional parameters are used to calibrate the strength
and stiffness degradation, which is a feature of great importance when modelling
timber structures and connections. A strength domain correlating different degrees
of freedoms (e.g. slip and uplift in a metal connector such as a hold-down) was
also implemented, as well as allowance for friction at the wall-foundation and
wall-floor panel interfaces. More details can be found in [15].
(a) (b)
Fig. 1 Piecewise linear relationship of screws and angle bracket springs loaded in shear (a)
and hold-down springs axially loaded (b)
3 Spring Calibration
A calibration has been conducted on each type of connection and for each type of
building system. Starting from X-lam, the springs representing each metal connec-
tor have been calibrated, for each degree of freedom, over experimental cyclic
tests. For the light-frame construction system, the nailed connection is modelled.
Finally, a beam-column joint tested under cycling loading has been used to obtain
the calibration of a rotational spring to be used in timber moment-resisting frames.
3.1 X-lam Connectors

X-lam structures are made from X-lam solid panels connected to each other and to
the foundation using metal connectors such as hold-downs, angle brackets, and
screws. Hold-downs are used at the end of the walls to mainly resist uplift, whilst
angle brackets are used at the base of the walls to mainly resist shear. Screws are
used to connect adjacent wall and floor panels, and for the floor-wall connection.
Each type of spring was calibrated on the experimental results reported in [8,9].
The calibration was done by following the steps listed herein after:
the yielding and peak force were extracted from the experimental results;
the elastic stiffness was estimated once a good fit of the yielding displacement
was obtained;
the hardening stiffness of the plastic branch was chosen on the basis of the
backbone curve;
other parameters, such as the stiffness and the strength degradation factors,
were evaluated in an iterative way until a good fit between the experimental
curve and the model was obtained.
(a) (b)
Fig. 2 Calibration of angle bracket springs loaded in shear (left, a) and in tension (right, b)
To speed up the calibration process, the software So.ph.i. (acronym for SOft-
ware for PHenomenological Implementations) has been developed using the Vis-
ual Basic .NET language [13]. So.ph.i. [16] allows the user to calibrate the spring
parameters over experimental data following the procedure proposed in EN
12512:2001 [4]. This allows the user to obtain automatically elastic stiffness and
yielding force values according to the code. The mean properties of single spring
components have been used in all subsequent analyses. As an example, the
comparison between experimental and approximating model curves is displayed in

Figure 2 for angle bracket in shear and tension.
3.2 Light-Frame Construction

Light-frame systems are widely used in the North America and Europe. The basic
structural unit is a wall made of wooden framing elements (studs and plates),
wood-based sheathing panels, and either nails or screws connecting the sheathing
to the frame. Additional connectors such as hold-downs are used to rigidly con-
nect the framing to the foundation/lower walls. In this paper, an experimental
cyclic test of a light-frame wall loaded horizontally is reproduced in order to vali-
date the proposed numerical approach. The analysed wall was tested by Dolan [2],
and is 2.442.44 m (88 ft) dimensions. The Young moduli used are 3000 MPa
for the panels and 8400 MPa for the frame, while the Poissons ratio is 0.35 in
both cases. Frame elements and sheathing panels, considered as elastic, are con-
nected with 2 degrees-of-freedom (DOF) spring elements used to model couples
of nails. The 2 DOFs are the horizontal and vertical displacement in the plane of
the wall.
The nails characteristics have been obtained from an experimental programme
carried out by Fischer et al. [7]. Since the experimental programme gave similar
results for both directions, a single calibration for both horizontal and vertical
shear in the spring is done (Figure 3a).
(a) (b)
Fig. 3 Nail spring calibration based on experimental tests carried out by [5] (a) and com-
parison of numerical prediction and experimental cyclic response for the light-frame timber
wall tested by Dolan [2] (b)
The numerical cyclic response of the modelled wall is reported in Figure 3b in

terms of total base shear vs. top displacement. Obtained results fits well with the
experimental curves and provide a validation of the proposed model.
3.3 Moment-Resisting Frame

The investigated moment-resisting timber structure is composed by timber beams
and columns, connected to each other using several steel tube connectors.
(a) (b)
Fig. 4 Photo of the test setup of a typical steel tube connector [12] (a) and cyclic results
obtained by Leijten (red curve) [12] and relative calibration (black curve) (b)
The steel tube dowel type connectors (Figure 4a) are first inserted into pre-
drilled holes in the connection region and then expanded, to achieve a non-slip
connection. Furthermore, densified veneer wood plates are used between the con-
nected beams and columns, to prevent the timber from splitting, allow the dowels
to work independently from each other, and enable dowel plasticization according
to the ductile modes of the Johansen equations (Johansen [11], EN 1995-1-1:2009
[5]).
A T-shaped specimen with a beam-column connection made of four 28 mm
diameter steel tubes was tested in the Netherlands [12]. A cyclic imposed horizon-
tal displacement with increasing amplitude was applied at the top of the vertical
member causing a push-pull action. The obtained experimental moment-rotation
hysteresis loops are depicted in Figure 4b. These curves were approximated using
the hysteretic piecewise linear model described in section 2.2 and shown in Figure
1. The comparison between experimental and approximating cycles is displayed in
Figure 4b.
4 Numerical Analyses
Numerical analyses have been performed for every structural system on entire
buildings.
4.1 X-lam Building

A single-storey X-lam building was tested at University of Trento by CNR-
IVALSA [14]; this building is composed by an assemblage of X-lam panels con-
nected to the foundation with angle brackets and hold-downs. The connectors have
the same characteristics (number of nails and connector type) as the ones tested
separately in [8,9]. The building is 77 m in plan and 3.1 m high. The walls are 85
mm thick and the floor is 142 mm thick with a central opening (Figure 5a).
Fig. 5 3D views of the single-storey X-lam building model with springs marked with
crosses (a) and in its deformed shape with 20 times amplification (b)
Pseudo-dynamic tests were performed in three different configurations obtained

by varying the width of the door openings. In this paper, the configuration with
non-symmetric layout with respect to the direction of the actuator (Figure 5b) was
investigated. This configuration has two external openings 4.0 m and 2.25 m wide.
All the wooden parts have been modelled with shell elements (S4R in Abaqus);
the size of the adopted mesh is a convenient value to ensure that the connections
are placed as close as possible to their actual position and that the elements are
regular.
The building was subjected to a pseudo-dynamic test that simulated the earth-
quake of Kobe JMA 0.5g. A non-linear static analysis was performed; a typical
deformed shape can be seen in Figure 5b, while results in terms of total base shear
force vs. top floor displacement are presented in Figure 6a. An overall good accu-
racy of the experimental behaviour can be noticed.
(a) (b)
Fig. 6 Experimental-numerical comparison of the hysteretic cycles (a) and time-history of

total energy (b) of the single-storey X-lam building tested by CNR-IVALSA at the Univ. of
Trento [14]
The numerical model slightly overestimates the backbone response of the build-
ing. In Figure 6b, numerical and experimental responses are compared in terms of
total energy. The final numerical value is 2.85% higher than the experimental one,
due essentially to some approximations of the narrow inner cycles that the model
predicts near the origin of Figure 6a.
4.2 Light-Frame Building

A light-frame 2-storey building tested by Fischer et al. [7] has been modelled us-
ing the proposed approach. The building is 4.886.11 m (1620 ft) in plan and
2.44 m high (Figure 7b).
(a) (b)
Fig. 7 Schematization of light-frame wall in the numerical model (a) and layout of the
resisting walls (b)
The same wall distribution as in the numerical model developed by Du [3] has
been used for comparison purposes. Dus model used the cyclic results of the
2.442.44 m light-frame wall tested by Dolan [2] and modelled in section 3.2 to
characterise the building. For this reason, the geometry of the building had to be
slightly modified to accommodate the different wall length compared to the actual
wall distribution. To reduce the computational burden of the analysis, each wall
has been modelled using two diagonal springs, which have the hysteretic
behaviour displayed in Figure 1a, and rigid truss elements pinned connected to
each other representing the wood framing, as depicted in Figure 7a. The diagonal
springs have been calibrated upon the experimental results presented in Figure 3b.
The model has been restrained at the base using simple supports and the
hypothesis of rigid floor diaphragms has been applied. A modal analysis has been
performed, returning a natural vibration frequancy of 5.25 Hz, which is close to
the value of 5.03 Hz found by Du. A dynamic analysis has then been performed,
using the Northridge recorded accelerogram (Newhall station, 90 compoenent,
1994); the overall response of the building is plotted in Figure 8a as base force
versus top displacement curves.
(a) (b)
Fig. 8 Comparison between the presented model and Dus model in terms of top
displacement vs. base shear of the building (a) and first floor drift vs. time (b)
The analysis has been stopped when the first storey reached a drift equal to 3%
of the building height, which is regarded as an upper limit corresponding to the
attainment of the collapse limit state in [3]. A comparison in terms of time-history
of the first interstorey drift is presented in Figure 8b. Overall good agreement
between the results predicted by Dus model and by the model developed in this
paper can be noticed.
4.3 Moment-Resisting Frame

To demonstrate the potentialities of the presented approach, non-linear analyses of
a 4-storey timber moment resisting frame have been carried out [17]. The studied
planar frame is taken from a building designed in accordance with Eurocodes 5
and 8 in SantAngelo dei Lombardi, Campania, Italy.
Figure 9a shows the geometry of the analysed frame. A static analysis using the
response spectrum, according to Eurocode 8 [6], was carried out adopting a
q-factor of 4. The frame-column joints are realised with twelve 35 mm diameter
dowels. To extrapolate the hysteretic properties of the 35 mm diameter tubes from
the 28 mm diameter tubes presented in section 3.3, the analogy in the monotonic
response between dowels of different diameters proposed by Jaspart and Maquoi
[10] has been used. The relationship is presented in Figure 9b.
(a) (b)
Fig. 9 Analysed frame (b) (dimensions in mm) (units in mm) (a) and non-linear regression
curve of a 35 mm diameter dowel per shear plane [12]
The frame analysed has been modelled with a series of rotational springs placed
at the end of every beam. These springs have been calibrated upon the backbone
curve obtained with the regression curve using the same parameters for the cyclic
branches of the 28 mm tubes. The resulting calibration is presented in Figure 10a.
All the other wooden parts of the frame are modelled with elastic beams, with
columns simply supported at the base.
To demonstrate the capabilities of the developed springs, a dynamic analysis
has been performed on the frame using a generated accelerogram, spectrum-
consistent with the Eurocode 8 design spectrum used in design. The history of
base shear is presented in Figure 10b, shown for both an elastic and a non-linear
analysis. Further investigations have been performed in [17] to estimate the behav-
iour factor of the analysed frame.
(a) (b)
Fig. 10 Comparison between extrapolated experimental and numerical results (in rad-kNm)
(a) and comparison between elastic and inelastic time histories of base shear for a generated
ground motion compatible with the Eurocode 8 elastic spectrum (b)
5 Conclusions
The developed model represents the cyclic response of steel connectors in timber
structures, and can be applied to a variety of systems, including X-lam buildings,
light-frame construction, and moment-resisting frames.
In X-lam buildings, based on experimental evidence, all energy dissipation is
assumed to take place in the steel connectors, whilst the timber panel are regarded
as linear elastic. Two different hysteretic loops characterized by a tri-linear back-
bone curve with significant pinching effect have been implemented in Abaqus
software package using external user-subroutines. Such hysteretic relationships
have then been fitted with the experimental results available for hold-downs, angle
brackets, and screwed connection. An entire single-storey X-lam building sub-
jected to pseudo-dynamic test has been modelled and compared with experimental
results available.
The model proposed for light-frame construction is developed at two different
levels: (i) by modelling the hysteretic behaviour of each nail within the wall, or
(ii) by modelling the global hysteretic behaviour of the wall via a couple of
diagonal springs. Model (i) is used to predict the global wall behaviour from the
behaviour of the wall components (nails/screws, framing, sheathing). Model (ii) is

computationally more efficient and can be used to model entire buildings. A com-
parison between numerical predictions and experimental behaviour has been pre-
sented for a single light-frame wall (Model (i)). Furthermore, the capabilities of
the second approach (Model (ii)) to represent entire buildings have been high-
lighted, by comparing the cyclic response predicted by the proposed model with
another model available in literature.
Moment-resisting frames can be modelled using the same phenomenological
springs used for rotational degrees of freedom. After a calibration made on
experimental tests at the connection level, several dynamic analyses have been
performed in moment-resisting frames to demonstrate the capabilities of the pro-
posed approach to estimate the dissipative capacity at the structural level.
The proposed hysteretic model is very robust and unlike other software pack-
ages does not suffer from convergence problem. As such, it can be effectively
used to model the seismic performance of timber buildings and subassemblies.
The cyclic behaviour of the steel connectors allows a correct estimation of the
dissipated energy and, consequently, a reliable prediction of the seismic capacity
of an entire timber building, which is a fundamental matter in capacity based de-
sign. The model will be used in the future to carry out extensive parametric studies
on X-lam buildings, light-frame construction, and moment-resisting frames and
derive important information for seismic design such as the value of the behaviour
factors for implementation in the new generation of the Eurocode 8.
References
[1] Ceccotti, A.: New technologies for Construction of Medium-Rise buildings in
Seismic Regions: The XLAM case. Journal of the International Association for
Bridge and Structural Engineering 2, 156165 (2008)
[2] Dolan, J.D.: The Dynamic Response of Timber Shear Walls. PhD thesis, University
of British Columbia (1989)
[3] Du, Y.: The Development and Use of a Novel Finite Element For the Evaluation of
Embedded Fluid Dampers Within Light- Frame Timber Structures with Seismic
Loading. PhD thesis, Department of Civil & Environmental Engineering, Washing-
ton State University (2003)
[4] EN12512, Timber structures - Test methods - Cyclic testing of joints made with
mechanical fasteners. CEN, Brussels, Belgium (2001)
[5] Eurocode 5, Design of Timber Structure Part 1-1: General Common rules and
rules for buildings, EN 1995-1:2009 (2009)
[6] Eurocode 8, Design of structures for earthquake resistance - Part 1, General rules,
seismic actions and rules for buildings. EN 1998-1:2005 (2005)
[7] Fischer, D., Filiatrault, A., Folz, B., Uang, C.M., Seible, F.: Shake table tests of a
two-storey woodframe house, CUREE-Caltech Woodframe Project, Department of
Structural Engineering, University of San Diego, California, US (2001)
[8] Gavric, I., Ceccotti, A., Fragiacomo, M.: Experimental cyclic tests on cross-
laminated timber panels and typical connections. In: Proceedings of the 14th ANIDIS
Conference, Bari, Italy, September 18-22. DVD (2011)
[9] Gavric, I., Fragiacomo, M., Ceccotti, A.: Strength and deformation characteristics of
typical X-Lam connections. In: 12th World Conference on Timber Engineering,
Auckland, New Zealand (2012)
[10] Jaspart, J.P., Maquoi, R.: Survey of existing types of joint modeling, Semi Rigid
behavior of Civil Engineering Structural Connections. In: COST C1 Workshop,
Strasbourg, France (1992)
[11] Johansen, K.W.: Theory of timber connections. International Association of Bridge
and Structural Engineering 9, 249262 (1949)
[12] Leijten, A.J.: Locally reinforced timber joints with expanded tube fasteners.
Heron-English Edition 44(3), 131162 (1999)
[13] MSDN - Microsoft Developer Network (2011), Internet site http://msdn.
microsoft.com
[14] Progettosofie, new architecture with wood (2008), Internet site http://www.
progettosofie.it/
[15] Rinaldin, G., Amadio, C., Fragiacomo, M.: A Component approach for the hysteretic
behaviour of connections in cross-laminated wooden structures. In: Earthquake
Engineering and Structural Dynamics (in print, 2013), doi:10.1002/eqe.2310
[16] So.ph.i. Software (2012), Internet site http://giovanni.rinaldin.org/
[17] Wresniak, D., Amadio, C., Rinaldin, G., Fragiacomo, M.: Non-linear cyclic model-
ling of moment-resisting timber frames. Submitted to ANIDIS 2013, Conference,
Padua, Italy (2013)
Overview of a Project to Quantify Seismic
Performance Factors for Cross Laminated
Timber Structures in the United States
M. Omar Amini1, John W. van de Lindt1, Shiling Pei2,

Douglas Rammer3, Phil Line4, and Marjan Popovski5
1
Civil and Environmental Engineering, Colorado State University, Fort Collins, CO, USA
jwv@engr.colostate.edu
2
Department of Civil and Environmental Engineering, South Dakota State University,
Brookings, SD, USA
3
Forest Products Laboratory, Madison, SD, USA
4
Structural Engineering American Wood Council, Leesburg, VA, USA
5
FPInnovations, Advanced Building Systems, Vancouver, BC, Canada V6T 1Z4
Abstract. Cross-laminated Timber (CLT) has been extensively used in Europe

and is now gaining momentum in North America; both Canada and more
recently the U.S. Construction projects have shown that CLT can effectively be
used as an alternative construction material in mid-rise structures and has
significant potential in commercial and industrial buildings. In the United States,
the CLT system is not currently recognized in seismic design codes and
therefore a seismic design can only be performed through alternative methods
specified in the codes. The FEMA P695 report published in 2009 presents a
methodology to determine seismic performance factors namely the response
modification factor, overstrength factor, and deflection amplification factor for a
proposed seismic resisting system. The methodology consists of a number of
steps to characterize system behavior and evaluate its performance under
seismic loading. The additional benefit of the methodology is that it considers
variability in ground motions and uncertainties in tests, design, and modeling.
This paper presents an overview of the P695 methodology and more specifically
the approach adopted to apply the methodology to Cross Laminated Timber
(cross lam) systems in the United States. The type of tests and testing
configurations conducted as part of this study and development of the CLT
archetypes are discussed. Nonlinear models used to simulate CLT behavior at
the connection, wall, and system levels are presented and the procedure to
determine collapse margin ratio is explained.
Keywords: Cross-laminated timber, seismic performance factors, FEMA P695

methodology.
DOI: 10.1007/978-94-007-7811-5_49, RILEM 2014
532 M.O. Amini et al.
1 Background
Cross-laminated Timber (CLT) was initially introduced during the 1990s in

Austria and Germany and since then has been gaining popularity in Europe and
more recently in North America. The increased popularity of this innovative wood
product as a viable alternative in residential and non-residential construction is
due to a number of advantages such as the potential for mass production,
prefabrication, rapid construction, and sustainability as an environmentally
friendly renewable construction product. Very good thermal insulation, acoustic
performance, and fire ratings are some additional benefits of this system (CLT
Handbook, 2013; Ceccotti, 2008).
2 CLT as a Lateral Force Resisting System
A number of studies on CLT systems undertaken in Europe and Canada (i.e.

Ceccotti and Follesa, 2006; Ceccotti, 2008; Dujic and Zarnic, 2006; Dujic et al.,
2008; Popovski et al., 2010) have demonstrated that a CLT system can be utilized
effectively as a lateral force resisting system. The studies included tests on various
types of connections; quasi-static and dynamic tests conducted on isolated CLT
walls and CLT assemblies; and full scale shake table tests on a three- and seven-
story building (SOFIE project). However, in the United States, CLT is not a
documented lateral force resisting system and a design may only be performed
through the alternative methods within the governing design code. In 2009, the
Applied Technology Council (ATC) proposed a methodology published as
Federal Emergency Management Agency (FEMA) report P695 which provides a
rationale to evaluate seismic performance factors (SPFs) including the response
modification factor (R-factor), the system overstrength factor, and the deflection
amplification factor for seismic design in the U.S. The objective of the
methodology is to provide an equivalent level of safety for all the structures
comprised of different seismic force-resisting systems, i.e. approximately a 10%
or lower probability of collapse when subjected to an earthquake having the
intensity of an earthquake with a 2500 year return period (known in the U.S. at the
Maximum Credible Earthquake). The FEMA P695 methodology uses nonlinear
static and dynamic analyses along with statistical analysis and takes into account
the variation in earthquake records and uncertainties inherit in the test data and
modeling methods.
CLT prefabrication and ease of handling can facilitate production of large
monolithic walls; for example, in some cases one-piece story walls. However,
in this project, seismic analysis for CLT will be performed similar to light-
frame wood shear panels (WSP) where analysis is conducted for certain wall
aspect ratios and the result can be used to develop shear design tables similar to
WSP.
Overview of a Project to Quantify Seismic Performance Factors 533
3 Overview of FEMA P695 Methodology

As mentioned earlier, the P695 methodology will be used to evaluate seismic
performance factors known as the response modification factors, R, overstrength
factor, o and deflection amplification factor, Cd. R is defined as the ratio of the
shear developed in the system if the system were to remain entirely linearly elastic
under design ground motions VE to the design base shear value V. o is the ratio
of maximum shear strength Vmax of the yielded system to the design base shear. Cd
is defined as the ratio of the roof drift of the yielded system under design
earthquake ground motions to the roof drift under design base shear considering
the system to behave linearly elastic E, multiplied by the R factor. SPFs are best
described using the following equations:
R = Response Modification Coefficient = VE/V
o = Overstrength Factor = Vmax / V
Cd = Deflection Amplification Factor = (/E) R
The procedure is iterative in nature and includes the following steps: (1)
establish design requirements and develop specifications that is based on
Develop System Concept
Obtain Required Information
Characteristic Behavior
Develop Model
Analyze Models
Evaluate Performance
Collapse Margin Ratio
Document Results
Fig. 1 Overview of the process for quantifying and documenting seismic performance
factors (after FEMA P695, 2009)
applicable codes and standards; (2) identification of a number of archetypes to be

representative of the full design space from low-rise single family buildings to
mid-rise mixed-used buildings including multi-family buildings and office
buildings; The archetypes are categorized based on key design variables such as
geometric variations, load intensities, and other variables that are known to have
an effect on system performance; (3) a series of experimental tests on panels with
varying holddown conditions and aspect ratios; the results are then used to
calibrate the nonlinear numerical models; (4) the development and validation of a
nonlinear computer model; the developed model takes into account degradation in
stiffness and strength in the inelastic range; (5) comprehensive static (pushover)
and incremental dynamic analysis (IDA) are performed to compute the median
collapse that is then used to evaluate margin against collapse for the archetypes;
and (6) determination of whether the seismic performance factors are acceptable
based on FEMA P695 requirements. The entire procedure is overseen by a
technical peer panel and their involvement is critical to each step of the process.
The peer review panel will write a peer panel report for inclusion in the code
adoption process. Figure 1 explains the procedure and its flow.
4 Required Information
Design requirements for CLT are based on ASCE/SEI 7-10, national design
specification for wood construction (ANSI/AF&PA, 2012), and U.S. CLT
handbook (2013). Tests will be conducted to predict strength, stiffness, and
deformation characteristics of the system under consideration when subjected to
simulated seismic loading. All testing will be performed in accordance with the
applicable standards and specifications. Test results can also be used in developing
and validating design methods for the system under consideration. According to
the P695 methodology, testing is to be conducted at various levels to reliably
capture and predict structural response including:
Material test data

Components and connections test data
Assembly and system test data
Material testing will not be conducted as part of this study since the data can be
obtained from the previous studies. Recently the American Paper Association
(APA) published the ANSI/APA PRG 320 (2011) standard that provides
information on performance and requirements for Rated Cross Laminated Timber.
Three different types of tests that consist of connection testing on brackets,
reversed cyclic loading of single isolated walls, and reversed cyclic test of two
walls with a diaphragm and two walls in box type configurations will be
conducted on CLT specimens. CLT wall connection testing will be performed at
Forest Products Laboratory (FPL) in Wisconsin and all the other tests will be
conducted in Structures Lab at Colorado State University (CSU). Simple testing
configurations are shown in Figures 2-5.
Overview of a Project to Quaantify Seismic Performance Factors 5335
5 Archetype Dev
velopment
Archetypes or baseline sttructures are used to investigate seismic performance oof
the proposed lateral forcee resisting system. According to the P695 methodology,
archetypes are prototypical presentation of a seismic force resisting system. Theeir
development is an essen ntial part of the methodology since they determine thhe
applicable range and the design space for the lateral force resisting system. Thhe
design space is divided innto various performance groups which consist of severral
archetype models each. Each
E performance group is categorized based on variablees
such as seismic design cattegory, gravity load, and building height variations.
Three different categorries for the archetypes are considered for the purpose oof
this study: single-familyy dwellings, multi-family dwellings, and commerciial
(including mixed-use) mid-rise buildings. Archetypes will be developeed
considering the variables shown in Table 1.
Fig. 2 Isolated wall test setu

up (out-of-plane Fig. 3 Two wall assemblies with a
bracing not shown) diaphragm (weight will be placed on thhe
diaphragm in lieu of force controlleed
actuators)
Fig. 4 Box type configu

uration with a Fig. 5 Box type configuration with a
diaphragm diaphragm using 0.6 m x 2.4 m (2x 88)
panels
Table 1 Variables considered in developing archetypes
Variable Range
Number of stories Seismic 1 to 10
Design Categories (SDC) Dmax and Dmin
Story height 2.44m to 4m
Interior and exterior nonstructural wall Not considered
finishes
CLT shear wall aspect ratios High/Low
6 Nonlinear Model Development
Based on a preliminary numerical study performed for a mid-rise CLT buildings,

an R-factor of 4.5 was determined reasonable for that CLT structure (Pei et al.,
2012). Therefore, preliminary analysis for CLT buildings as part of this study will
be based on an R value of 4.5, o value of 3, and Cd value of 4.5. This will likely
have to be revised but will serve as a reasonable starting point.
According to the P695 guidelines, the proposed nonlinear numerical models
should simulate all significant deterioration mechanisms that can lead to collapse
i.e. degradation in stiffness and strength, and inelastic deformation. In order to
evaluate collapse of CLT structures, failure criteria should be defined in terms of
story drift for simulated collapse modes. Based on previous studies, CLT systems
have been shown to withstand a lateral drift of 3.5% without collapse; however,
story drift leading to lateral instability is still unknown for CLT, since
comprehensive collapse testing has not been performed.
If the models are unable to simulate all the collapse mechanisms, additional
non-simulated collapse modes will be considered in the analysis since they are
important in determining reasonable median collapse intensities. Analytical
models used to simulate CLT behavior at the component and at the assembly level
is explained in the following.
7 CLT Connections
Based on the previous studies conducted on CLT wall connections, they are
determined to follow the 10-parameter CUREE model. This type of model is
currently used to simulate component behavior of light-frame wood structures
and adopting this model for CLT offers additional advantage of consistency
with the light-frame wood components. A reverse calibration has already been
performed for various connection types and presented in the US CLT
handbook. Similarly typical calibration will be performed for the wall
connections as part of this project. A generic 10-parameter hysteretic model is
shown in Figure 6.
8 CLT Wall Modeling

The numerical model proposed as part of this study for simulating wall behavior
is based on the recently published US CLT handbook (CLT Handbook, 2013).
Previous studies and test observations indicate that CLT walls exhibit rocking
behavior which forms the basis of this model. Assumptions considered in this
numerical model are shown in Figure 7 and explained as follows:
The CLT wall panel exhibits in-plane rigid body behavior
Under lateral load, the CLT wall will rotate around the bottom corner
Lateral slip between the wall and the floor diaphragm is neglected
Gravity acts through center of the wall
The wall panel hysteresis is based on panel connection deformation
during rocking motion
The numerical model used as part of this study will be calibrated using the test
results.
Fig. 6 Loading Paths and Parameters of Modified Stewart Hysteretic Model (after Pang et
al., 2010)
Fig. 7 Kinematic model of a single CLT wall panel

9 Building System Modeling
Analysis will be performed using the SAPWood software (Pei and van de
Lindt, 2007) that was developed as part of the NEESWood project for analysis
of light-frame wood structures. The software is based on the shear-bending
coupled model and the assumptions involved in this model are as follows:
The floor diaphragm behaves as a rigid plate having 6 degrees of

freedom.
Shear resistance can be represented using hysteretic springs.
Overturning restraint can be represented using multi-linear springs
(different in tension than compression, if needed).
The effect of finish materials ignored for now (may be added).
The accuracy and reliability of the software has been validated through a number
of studies (Pei and van de Lindt, 2009; van de Lindt et al., 2010).
The basic kinematic model to be used in SAPWood is shown in Figure 8.
Fig. 8 SAPWood kinematic model for nonlinear history analysis (after US CLT Handbook,
2013)
10 Nonlinear Analysis
Analyses will be performed based on the FEMA P695 methodology to determine

the seismic performance factors that meet all the requirements within the
methodology. This consists of nonlinear static and dynamic analysis of the
archetype models that will be performed using SAPWood software.
Nonlinear static analysis is performed in accordance with Section 3.3.3 of
ASCE/SEI 41-06 (2007) and the purpose is to determine period based ductility
and over-strength factors for the archetypes under consideration. A generic static
pushover curve is shown in Figure 11 and a pushover curve for a 2 ft x 8 ft CLT

wall is illustrated in Figure 12.
Incremental Dynamic Analysis (IDA) (Vamvastikos and Cornell, 2002) will be
performed on all the archetypes for a set of 22 predefined large-magnitude ground
motion records termed Far-Field earthquakes available in P695. All the
archetypes are analyzed for the Maximum Credible Earthquake (MCE) and the
results of the IDA are used to plot the cumulative distribution function (CDF)
which then leads to the determination of collapse spectral acceleration (Ibarra et
al., 2002). Median collapse intensity, SCT, is defined as the intensity corresponding
to 50% probability of collapse.
30
25
20
Force (kN)
15
10
0
0 20 40 60 80 100 120 140
Displacement (mm)
Fig. 9 Pushover curve for 0.61 m x 2.44 m (2ft x 8ft) CLT wall
11 Performance Evaluation
The seismic performance factors will be evaluated and the adjusted collapse
margin ratio (ACMR) determined (FEMA, 2009). The collapse margin ratio is
defined as the ratio of median collapse intensity, SCT, to the MCE intensity, SMT,
and can be calculated as
= (1)
A study conducted by Baker and Cornell (2006) showed that spectral shape of the
ground motion record had a profound influence on the collapse margin ration
(CMR) which is incorporated in the methodology by applying a spectral shape
factor. This factor is determined based on the fundamental period of the archetype,
T, and the period based ductility, T. The adjusted collapse margin ratio (ACMR)
is then computed as
= (2)
Its important to note that regardless of the type of analysis conducted, there are
always uncertainties and variability involved in the analysis. Four major sources
of uncertainty that significantly influence CMR calculations are identified in P695
namely record-to-record variability, design requirements, test data, and modeling.

There values are based on completeness and accuracy of the pertinent information.
Detailed discussion on each of the uncertainties and calculation of the total
uncertainty is provided in FEMA P695 report and are omitted here for brevity.
Acceptable ACMR values are based on the calculated total uncertainty and
accepted collapse probability and are obtained from Table 7.3 of P695. The
collapse probability for MCE ground motions considered in the P695 is 10% and
20% for an average across each performance group (described earlier) and each
index archetype, respectively. If the ACMR meets the collapse criteria, the trial R
value and other calculated seismic performance factors are considered acceptable
per the methodology. However, if not, the archetypes are re-designed and entire
process repeated until an acceptable ACMR is obtained.
12 Closure
CLT is an innovative wood product that is gaining popularity as a viable

alternative for masonry, concrete, and steel in mid-rise construction. Research has
shown that CLT can be used as an effective lateral force resisting system.
However, there are no specific seismic design guidelines in the U.S. pertaining to
this new product and the system must be designed based on alternative methods
allowed by the codes, which will limit its use. This paper provides an overview of
the FEMA P695 methodology which will be applied to CLT in the United States
in order to make it a viable alternative in high seismic regions. At the time of this
paper the project is in the first Phase and is anticipated to be completed in 2015.
References
American National Standards Institute/American Forest and Paper Association
NDS(ANSI/AF&PA), National design specification for wood construction, Washington,
D.C. (2012)
APA - The Engineered Wood Association. Standard for Performance-Rated Cross
Laminated Timber, ANSI/APA PRG 320. Tacoma, Washington, U.S.A. (2011)
ASCE, Minimum Design Loads for Building and Other Structures. ASCE Standard
ASC/SEI 7-05, American Society of Civil Engineers, Reston, Virginia (2005)
ASCE, Minimum Design Loads for Building and Other Structures. ASCE Standard
ASC/SEI 7-10, American Society of Civil Engineers, Reston, Virginia (2010)
ASCE, Seismic Rehabilitation of Existing Buildings. ASCE Standard ASC/SEI 41-06,
American Society of Civil Engineers, Reston, Virginia (2007)
Baker, J.W., Cornell, C.A.: Spectral shape, epsilon, and record selection. Journal of
Earthquake Engineering and Structural Dynamics 34(10), 11931217 (2006)
Ceccotti, A., Follesa, M.: Seismic Behaviour of Multi-Storey X-Lam Buildings. In: COST
E29 International Workshop on Earthquake Engineering on Timber Structures, Coimbra,
Portugal, pp. 8195 (2006)
Ceccotti, A.: New Technologies for Construction of Medium-Rise Buildings in Seismic
regions: The XLAM Case. Structural Engineering International SEI 18(2), 156165
(2008)
CLT, CLT Handbook, U.S. Edition, FPInnovations and Binational Softwood Lumber
Council (2013)
Dujic, B., Zarnic, R.: Study of Lateral resistance of Massive X-Lam Wooden Wall System
subjected to Horizontal Loads. In: COST E29 International Workshop on Earthquake
Engineering on Timber Structures, Coimbra, Portugal, pp. 97104 (2006)
Dujic, B., Klobcar, S., Zarnic, R.: Shear Capacity of Cross-Laminated Wooden Walls. In:
Proceedings of the 10th World Conference on Timber Engineering, Myazaki, Japan
(2008)
FEMA, Quantification of building seismic performance factors: FEMA P695 Federal
Emergency Management Agency (2009)
Ibarra, L., Medina, R., Krawinkler, H.: Collapse assessment of deteriorating SDOF
systems. In: Proceeding of the 12th European Conference on Earthquake Engineering,
London, UK, Paper reference 665, September 9-13. Elsevier, Oxford (2002)
Pang, W., Rosowsky, D.V., Pei, S., van de Lindt, J.W.: Simplified direct displacement
design of a six-story woodframe building and pre-test performance assessment. J. Struct.
Eng. 136(7), 813825 (2010)
Pei, S., van de Lindt, J.W.: Users Manual for SAPWood for Windows: Seismic Analysis
Package for Woodframe Structures, Colorado State University, Fort Collins, CO (2007)
Pei, S., van de Lindt, J.W.: Coupled Shear-Bending Formulation for Seismic Analysis of
Stacked Shear Wall System. Earthquake Engineering and Structural Dynamics 38,
16311647 (2009)
Pei, S., van de Lindt, J.W., Popovski, M.: Approximate R-Factor for Cross Laminated
Walls in Multi-Story Buildings. ASCE Journal of Architectural Engineering (May 2012)
Popovski, M., Schneider, J., Schweinsteiger, M.: Lateral Load Resistance of Cross-
Laminated Wood Panels. In: World Conference on Timber Engineering, pp. 2024
(2010)
Vamvatsikos, D., Cornell, C.A.: Incremental Dynamic Analysis. Journal of Earthquake
Engineering and Structural Dynamics 31(3), 219231 (2002)
van de Lindt, J.W., Pei, S., Liu, H., Filiatrault, A.: Seismic Response of a Full-Scale Light
Frame Wood Building: A Numerical Study. ASCE Journal of Structural
Engineering 136(1), 5665 (2010)
Force Modification Factors for CLT Structures
for NBCC
Marjan Popovski1, Shiling Pei2, John W. van de Lindt3, and E. Karacabeyli4
1
FPInnovations, Advanced Building Systems, Vancouver, BC, Canada
marjan.popovski@fpinnovations.ca
2
Department of Civil and Environmental Engineering, South Dakota State University,
Brookings, SD, USA
shiling.pei@sdstate.edu
3
Civil and Environmental Engineering, Colorado State University, Fort Collins, CO, USA
jwv@engr.colostate.edu
4
Advanced Building Systems, FPInnovations, Vancouver, BC, Canada
erol.karacabeyli@fpinnovations.ca
Abstract. Cross-laminated timber (CLT) as a structural system is not introduced

in European or North American building codes, and there is limited information
available on the seismic factors for design of such structures. The objective of the
study was to derive suitable ductility-based force modification factors (Rd-factors)
for seismic design of CLT buildings in the National Building Code of Canada
(NBCC). For that purpose, the well-known six-storey NEESWood Capstone
wood-frame building was redesigned as a CLT structure. Non-linear analytical
models of the building designed with different Rd-factors were developed
using the SAPWood program. CLT walls were modeled using the output from
mechanics models developed in Matlab that were verified against CLT wall tests
conducted at FPInnovations. Each of the 48 building models was subjected to a
series of 22 bi-axial input earthquake motions suggested in the FEMA P-695 pro
cedure. Results showed that an Rd-factor of 2.0 is appropriate for the building
studied.
Keywords: Cross-laminated timber (CLT), CLT structures, force modification

factor, R-factor, q-factor, ductile nonlinear response, shear walls.
1 Introduction and Previous Research

With three producers already in operation in Canada and one in the US, the use of
cross-laminated timber (CLT) is gaining popularity in North America. Since
CLT is not introduced as a structural system in the building codes and material
standards, one of the most important issues for designers of CLT structures in
DOI: 10.1007/978-94-007-7811-5_50, RILEM 2014
544 M. Popovski et al.
earthquake prone regions are the values for the force modification factors
(R-factors) for this system. The R-factors in building codes in North America
account for the capability of the structure to undergo ductile nonlinear response,
which dissipates energy and increases the building period. In the 2010 edition of
the NBCC (NBCC 2010), the elastic seismic load is reduced by two types of
R-factors, Ro-factor which is related to the over-strength of the system and Rd-
factor that is related to the ductility of the structure. In the major model codes in
the United States, the International Building Code and the ASCE 7 (ASCE 7,
2010), there is only one R-factor, called the response modification coefficient to
reduce the seismic design force. Eurocode 8 (EN 1998-1:2004), the European
seismic model code, also uses only one factor (q-factor) for reduction of the
seismic design force.
Efforts have already been made to quantify the q-factor for CLT structures in
Eurocode 8 using incremental non-linear dynamic analyses on analytical models
of three-storey structures verified by testing of components. As there is no specific
methodology for determining the q-factor in Europe, two approaches are common:
the acceleration-based approach, and the base shear approach. In the acceleration-
based approach the q-factor was calculated as a ratio of the peak acceleration of an
earthquake record that causes near collapse condition in the structure and the
design acceleration in the code for the location for which the building was de-
signed (Ceccotti et al. 2006, Ceccotti 2008, Schdle & Bla 2010). In the base
shear approach the q-factor was calculated as the ratio of the base shear force ob-
tained from linear elastic analysis and the base shear force at near collapse state
of the non-linear analysis for every input ground motion (Pozza et al. 2009). This
method also takes into consideration the influence of the input ground motion on
the elastic response of the structure. Using the acceleration-based approach
(Ceccotti 2008) the average q-factor was found to be q=3.4 while q=3.8 was
reported in Pozza et al. 2009. According to the base shear approach the average
q-factor was found to be q=3.15 (Pozza et al. 2009).
An initial estimate for the R-factors in North America was conducted using the
AC130 equivalency criteria (Popovski & Karacabeyli 2011). According to the
criteria, assigning an R-factor for a new wood shearwall assembly can be made
by showing equivalency of its seismic performance in terms of maximum load,
ductility, and storey drift (all obtained from quasi-static cyclic tests), with respect
to corresponding properties of wood-frame nailed shearwalls that are already in
the code. Based on the experimental tests conducted at FPInnovations (Popovski
et al. 2011, 2012), it was found that although not every single CLT wall configura-
tion satisfied the response parameters as defined in AC130, the average values for
the set of CLT walls did satisfy the AC130 criteria. Consequently, one may
assume that the CLT walls tested can share the same seismic modification factors
with regular wood-frame shearwalls in the US, which means using an R-factor of
6.5. This corresponds to having the product of RdRo equal to 5.1 in Canada
(Rd=3.0; Ro=1.7) which are the factors used in NBCC for nailed wood-frame
shearwalls. However, at this early stage of acceptance in the design practice, the
authors recommended that a more conservative set of factors (Rd=2.0; Ro=1.5) be
Force Modification Factors for CLT Structures for NBCC 545
used for CLT structures with ductile nails or screws and hold-downs. It was also
recommended that further studies such as the analyses presented in this paper and
analyses according to the FEMA P-695 procedure (FEMA 2009) be considered.
2 Objectives and Approach
The research presented in this paper contributes further to determining the

R-factors for seismic design of CLT structures. The objective of the research pre-
sented in this paper is to: (a) develop and verify analytical models for prediction of
the seismic response of CLT walls as main lateral load resisting elements in CLT
structures, (b) apply the verified models in a design of a mid-rise CLT residential
building, and (c) recommend the appropriate Rd-factors for seismic design of CLT
mid-rise buildings in NBCC. The well-known NEESWood wood-frame building,
also called Capstone Building (Fig. 1), was used as a reference building for this
study.
Fig. 1 The Capstone wood-frame building tested during the NEESWood Project
The Capstone wood-frame building was tested on the E-Defense shaking table
facility in Miki, Japan. During the tests, the building satisfied all performance
targets imposed during the design, with only non-structural damage present even
at Maximum Credible Earthquake (MCE) level of shaking with probability of
exceedance of 2% in 50 years (Pei et al. 2010). For the purposes of this study the
wood-frame Capstone building was redesigned as a CLT structure and was used as
a typical mid-rise CLT structure in all analyses. The results from the quasi-static
tests on CLT walls performed at FPInnovations (Popovski et al. 2011) were used
as input information for design and modeling of the CLT walls, the main lateral
load resisting elements of the CLT structure.
3 CLT Wall Modeling
Modeling of the CLT walls as main lateral load resisting elements of the structure
was done using a kinematics model developed in Matlab (Fig. 2a). The model was
calibrated for various input parameters such as size of the walls, gravity load level,
number and type of brackets and location, number of hold-downs, number of
|step-joints (if present), number of fasteners per-bracket, etc., based on the CLT
wall test data (Popovski et al. 2011), and using the basic kinematics formula
shown in Equation (1). The fastener behavior in the model was represented using
the well-known ten-parameter CUREE model, which is widely used for wood
based shear wall and connection modeling. Example of the tested and the calibrat-
ed model hysteretic response of a 2.3 m long CLT wall, which uses brackets with
eighteen 16d spiral nails per bracket and has step-joint with 8 4x70 mm (SFS1)
screws in the middle, is shown in Fig. 3. A total of 19 different models were
developed for CLT walls (assemblies) with 2 different bracket types and
hold-downs, with each of the 10 parameters calibrated for every model.
Fig. 2 Basic kinematics used for developing of the simplified CLT wall models
( )= ( )+ = + (1)
With the numerical model and connector parameters calibrated, the hysteresis
curve for any given CLT wall configuration with different connectors can be esti-
mated numerically using the kinematics (Matlab) model. Consequently, backbone
Force Modification Factors for
f CLT Structures for NBCC 5447
curves and lateral load deesign values for various CLT walls using three differennt
fasteners in the brackets were
w developed. Since at this point no design loads exiist
for CLT walls in Canad da, the design levels were developed by dividing thhe
ultimate load obtained froom the hysteresis loop (by the Matlab model) by a factoor
of 2.5, thus obtaining ap pproximately the same level of safety as with regulaar
wood-frame walls.
Fig. 3 An example of tested d and modeled (calibrated) response of 2.3 m long CLT waall
with 16d spiral nails in the brackets
b and a step-joint in the middle
In North America a typ pical design practice for CLT walls at this point is to aas-
sume that entire shear fo orce along the wall is taken by the bracket connectionns,
while the hold-downs are placed for vertical continuity and to take the wall uplifft.
In other words, the contrribution of the hold-downs is ignored when determininng
the shear capacities for th
he wall panels. It was recognized from experimental testts,
however, that due to rock king response of CLT walls, hold-downs also contribuute
to the shear wall strengtth. Consequently, the lateral resistances for CLT wallls
were derived with the tw wo options considered with and without the hold-dow wn
contribution taken into account. The capacities derived without hold-dow wn
contribution are thereforee conservative compared to the walls actually installed iin
practice that contain hold--downs.
Some of the configurattions for which analytical models and design values werre
derived are shown in Fig.. 4. The notation S stands for Single sided brackets foor
each location, DE stand ds for Double sided brackets at the End of the panel only,
and DA stands for Doub ble All, meaning all brackets are double sided. For a casse
of a CLT wall with 2 bracckets only, configurations DE and DA are identical.
548 M. Popovski et aal.
Fig. 4 CLT panel configuratiions for deriving design values
4 Calibration of the Rd-factor for the CLT Building
To calibrate the appropriaate ductility related force modification factor (Rd-factoor)

for the National Building Code of Canada (NBCC), the value for the over-strengtth
related modification facto or Ro that is mostly related to the over-strength, was choo-
sen to be constant (Ro=1.5) throughout this process. This value is the same witth
the other heavy timber systems in NBCC. As mentioned the 6-storey Capstonne
CLT building was used as a reference structure for developing the different buildd-
ing models used in the an nalyses. The seismic demand for the CLT models (strucc-
tures) was determined acccording to the Equivalent Static Force Procedure (ESFP P)
given in 2010 NBCC for 8 different values of the Rd factor (Rd=1.5, 2.0, 2.5, 3.0,
3.5, 4.0, 5.0, and 6.0). It was assumed that the buildings were located in Vancouu-
ver, BC (maximum spectrral acceleration of 0.96g at T=0.2s) with a design spectrral
acceleration Sa=0.743g at a the building fundamental period of 0.4s. According tto
NBCC requirements, buiildings located on firm soil with Rd 1.5 shall not bbe
designed for a seismic forrce that is greater than the 2/3 of the force at T=0.2s. Thhis
force cut-off value for th he design base shear was found to govern all designns,
meaning that the buildiings were actually designed for force equivalent tto
Sa=0.64g.
With the seismic demaand determined for each storey, CLT walls were selecteed
for each storey to satisfy the demands. Since design values for CLT walls with 3
types of connectors were developed (16d spiral nails, 4x70 mm SFS1 screws, annd
5x90 mm SFS2 screws),, for each of the two hold-down design considerationns
(with and without hold-d downs considered), a total of 6 building designs werre
developed for each cho osen Rd-factor. Since eight different Rd factors werre
considered, a total of 48 CLT Capstone building designs were generated. SAP P-

Wood models for every building
b design configuration were developed. The buildd-
ing models were developed to reflect the realistic as-built system, which includees
the impact of gravity load
d and presence of the hold-downs. On the other hand, thhe
resistances were generated without hold-downs considered in the resistance whicch
will lead to conservativee results. FEMA P-695 (FEMA 2009) suggested earthh-
quake records were used as input ground motions in the non-linear dynamic anaal-
yses. Because all FEMA A P-695 ground motions are biaxial, first the strongeer
ground motion componen nt of each pair was scaled at the building natural periood
(Ta=0.4s) to match the Saa=0.743g from the Vancouver design spectrum. The otheer
component was then scaaled with the same scale factor so that the PGA ratiio
between two componentss was not altered. Fig. 5 shows the response spectra foor
the un-scaled and scaled ground
g motions.
Fig. 5 Response spectra for the

t unscaled FEMA P-695 earthquake ground motions
A series of 44 bi-axiall nonlinear time history analyses was conducted for eacch
of the 48 building design ns (models) developed. For each analysis, the absoluute
maximum inter-storey driift from the building non-linear dynamic response at anny
storey and in any directiion was recorded and rank-ordered. The distribution oof
these maximum drift valu ues represents the performance of each particular desiggn
under the chosen hazard for site as Vancouver, BC. Examples of the cumulativve
distribution functions (CDFs) for the maximum inter-storey drifts for buildinng
configurations with 16d nails
n in the brackets and different Rd-factors are shown iin
Figures 6 and 7.
550 M. Popovski et aal.
Fig. 6 Fragility (CDF) curvees for the building for various Rd-factors for case of 16d spirral
nails in the brackets and the hold-down influence not included in the wall resistance valuess
Based on the performaance of different building designs (different connectors iin

the brackets and Rd facto ors), a comprehensive evaluation of the appropriate Rd
factors for achieving a preescribed performance target can be made. For example, if
the acceptable performan nce level for the buildings is assumed as not to exceeed
2.5% inter-storey drift in 80% of the cases (80% probability of non-exceedancee),
one just need to find the performance
p point corresponding to 2.5% drift on the X X-
axis and 0.8 CDF value on o the Y-axis of the plots. All the curves above that peer-
formance point will be ab ble to satisfy the criteria. The Rd factor that correspondds
to the curve that just satissfies the performance will be the most appropriate valuue
for that performance objeective. Based on this procedure, the Rd factors that satisffy
several different perform mance objectives are shown in Table 1. The minimum m
values for Rd factors for each
e performance target are given in bold font. It shoulld
be noted that not all of thee performance objectives need to be satisfied at the samme
time. It will be up to the design engineer, the jurisdiction of interest or the codde
committees to decide which performance level should dictate the Rd value used foor
the seismic design.
It can be seen from Table 1 that for each performance level the calibrated Rd
values are smaller in cases where the hold-down capacity was included in the CL LT
wall resistance than in thee cases where it was neglected. This is very logical sincce
neglecting hold-down con ntribution automatically builds in extra safety level in thhe
design. At this point wh hen the methods for deriving design values for CL LT
walls are getting developeed and are not agreed upon, it is prudent to take the morre
conservative approach forr determining the calibrated Rd factor for CLT structurees.
If the objective of the dessign is to effectively control the damage and prevent cool-
lapse (2.5% storey drift) during
d a 2% in 50 years event, it is recommended that aan
f all connectors considered for the buildings analysed.
Rd factor of 2.0 is used for
o the single wall components as well on the tests on 3-D
Based on the test results on
CLT structures conducted so far (Ceccotti 2008), the 2.5% inter-storey drift is
achievable in CLT structu ures without inducing excessive building damage.
Fig. 7 Fragility (CDF) curvees for the building for various Rd-factors for case of 16d spirral
nails in the brackets and the hold-down influence included in the wall resistance values
Table 1 Calibrated Rd facto

ors for the CLT buildings for the 2010 NBCC equivalent stattic
procedure
Hoold-downs NOT accounted in Hold-downs accounted in

Performance targets in
resistance values for CLT walls resistance values for CLT walls
terms of storey drifts
and probabilities of non-
16dd nails 4x70m 5x90mm 16d nails 4x70mm 5x90mm m
exceedance (PNE)
3.9x89mm Screws Screws 3.9x89mm Screws Screws
1.5% drift & 50% PNE 4
4.0 4.0 4.0 2.5 3.0 2.5
2.0% & 80% PNE 3
3.5 3.5 3.0 2.0 2.25 2.0
2.5% & 80% PNE 3
3.5 4.5 4.0 2.75 2.75 2.5
4.0% & 80% PNE 6
6.0 5.0 4.0 4.0 3.0 2.5
It should be noted that changes in the building configuration (unsymmetrical

floor plans), building height (fundamental period), the hazard characteristics of the
location, and the influence of the boundary conditions (effects of the perpendicular
walls and floor slabs), will have an influence on the calibrated Rd factors shown
here. However, the authors are of the opinion that such influences will either not
make significant changes to the values of the Rd factors suggested or will add
additional conservatism to the values (in case of the boundary conditions). There-
fore the Rd values presented here are good estimates for symmetrical CLT build-
ings located in high seismic region in Canada. The values of Rd=2.0 and Ro=1.5
will be proposed to the NBCC Standing Committee on Earthquake Design for
acceptance.
5 Concluding Remarks
Based on the analyses conducted in this study, CLT as a structural system is a

viable option for mid-rise buildings in moderate and high seismic regions. When
adequately designed, CLT structures with symmetrical plans can sustain
only limited damage under MCE earthquakes. By selecting appropriate R-factors,
the Equivalent Static Design Procedures can meet the performance objectives
selected. The results showed that although the type of fasteners used in the
brackets connecting the CLT walls has effects on the R-factors, the impact was not
significant. The design lateral resistance values for CLT walls were found to
have more significant impact on the R-factors. For that reason, both cases,
when CLT wall design resistances include and exclude the influence of the
hold-downs, were used in this study. At this time when methods for deriving
design values for CLT shearwalls are in the development stage and not agreed
upon, it is prudent to take a more conservative approach by taking into account
the influence of the hold-downs in deriving CLT wall design values. In such
case the recommended values for R-factors in Canada would be Rd=2.0 and
Ro=1.5.
Variation in the R-factor values as a function of the floor plan and the building
height is likely to exist. However, having in mind the spectra of the records
used and the responses of taller buildings of other structural systems subjected to
the same records (FEMA 2009), it is not expected that the proposed values for
the R-factors will change significantly for taller buildings, provided that the build-
ing floor plans remain symmetrical in both directions. It is recommended
that further studies with a wider scope look into issues related to structures with
different archetypes and non-symmetrical floor plans according to the FEMA
P-695 procedure. Such procedure is planned to be undertaken soon in the
US.
Force Modification Factors for CLT Structures for NBCC 553
References
ASCE7-10, American Society of Civil Engineers, Minimum Design Loads for Buildings
and Other Structures. ASCE, Reston, Virginia (2010)
Ceccotti, A.: New Technologies for Construction of Medium-Rise Buildings in Seismic
regions: The XLAM Case. Structural Engineering International SEI 18(2), 156165
(2008)
Ceccotti, A., Follesa, M., Kawai, N., Lauriola, M.P., Minowa, C., Sandhaas, C., Yasumura,
M.: Which Seismic Behaviour Factor for Multi-Storey Buildings made of Cross-
Laminated Wooden Panels? Proceedings of the 39th CIB W18 Meeting, paper 39-15-4,
Firenze, Italy (2006)
EN 1998-1:2004 (E). Eurocode 8: Design of structures for earthquake resistance Part 1:
General rules seismic actions and rules for buildings. European Committee for Standard-
ization, Brussels, Belgium
FEMA 2009, FEMA P695 Quantification of Building Seismic Performance Factors.
Federal Emergency Management Agency, Washington, D.C. (2009)
NBCC. National Building Code of Canada. Institute for Research in Construction, National
Research Council of Canada, Ottawa, Ontario (2010)
Pang, W., Rosowsky, D.V., Pei, S., van de Lindt, J.W.: Simplified Direct Displacement
Design of Six-storey Woodframe Building and Pretest Seismic Performance
Assessment. ASCE Journal of Structural Engineering 136(7), 813825 (2010)
Pei, S., van de Lindt, J.W., Pryor, S.E., Shimizu, H., Isoda, H.: Seismic testing of a
full-scale six-story light-frame wood building: NEESWood Capstone test. NEESWood
Report NW-04 (2010)
Popovski, M., Karacabeyli, E.: Seismic Performance of Cross-Laminated Wood Panels. In:
Proceedings of the 44th CIB W18 Meeting, Alghero, Italy (2011)
Popovski, M., Karacabeyli, E., Ceccotti, A.: Seismic Performance of Cross-Laminated
Timber Buildings. Chapter 4 of the FPInnovations CLT Design Handbook, Canadian
Edition (2011)
Popovski, M., Karacabeyli, E.: Seismic Behaviour of Cross-Laminated Timber Structures.
In: Proceedings of the World Conference on Timber Engineering, Auckland,
New Zealand (2012)
Pozza, L., Scotta, R., Vitaliani, R.: A non linear numerical model for the assessment of the
seismic behaviour and ductility factor of X-lam timber structures. In: Proceeding of Inter-
national Symposium on Timber Structures, Istanbul, Turkey, June 2527, pp. 151162
(2009)
Schdle, P., Bla, H.J.: Earthquake behaviour of modern timber construction systems. In:
Proceedings of the 11th World Conference on Timber Engineering, Riva del Garda,
Italy (2010)
van de Lindt, J.W., Pei, S., Pryor, S.E., Shimizu, H., Isoda, H.: Experimental seismic
response of a full-scale six-story light-frame wood building. ASCE Journal of Structural
Engineering 136(10), 12621272 (2010)
Experimental Testing of a Portal Frame
Connection Using Glued-In Steel Rods
James Walker* and Robert Xiao
London South Bank University, London, Great Britain

walker.james.n@gmail.com
Abstract. The use of glued in rods as a method for connecting timber has received
much attention over the past decade with an increasing number of examples of
their use in industry. The objective of this study is to understand the failure mech-
anism of a moment resisting portal frame connection in timber using glued in steel
rods. A series of pull out tests are used to determine axial strength of glued in rods
with varying anchorage length and bond line thickness. The results are compared
with the current design equations and the discrepancies are highlighted. Further
pull out tests using rods glued in at angles are used to assess the influence of
lateral loading on the axial pull out strength. Small amounts of lateral loading
combined with axial loading are seen to significantly reduce the pull out strength
of glued in rods and splitting failures are seen at relatively low levels of load. The
findings from the pull out tests are used to design a portal frame connection and
the moment capacity is estimated using linear elastic analysis. Results from the
destructive testing of the portal frame connection are presented and parallels are
drawn between the observed behaviour in the angled pull out tests. The linear
elastic analysis is found to accurately predict the failure load of the tested connec-
tion. The moment rotation (M ) relationship of the portal frame connection is
extracted from the measured test data.
Keywords: Portal frame connections, glued in rod, rod pull out anchorage length,
failure mechanism, moment capacity, moment rotation relationship.
1 Introduction
With the advancement of manufactured timber such as Glulam and LVL
(Laminated Veneer Lumber), the design of efficient connection systems is becom-
ing increasingly important. Today's timber engineering industry is demanding
more than can be fulfilled by traditional wood connectors such as bolts, dowels,
nails or screws. The use of glued in rods as a method of connecting timber began
in Scandinavia in the 1980s for applications involving the transfer of high loads
and moments [Riberholt 1983]. For a number of years glued in rods have been
*
DOI: 10.1007/978-94-007-7811-5_51, RILEM 2014
556 J. Walker and R. Xiaao
used as a method of rep pairing damaged timber in historic buildings due to thhe
minimal impact on the ex xisting fabric of the building [TRADA 1992]. Other studd-
ies have focused on the use
u of glued in rods to reinforce timber in the weak zonees
around notches [Coureau u 2000] or to improve the compression resistance oof
timber perpendicular to thhe grain [Madsen 2000] [Crocetti 2012].
Over the past four decades much research has gone into the manufacture oof
glued in rod connections. Most studies have focused on the strength characteristiccs
of the rod, adhesive and timber interface under axial loading and attempt to definne
guidelines for the rod material,
m rod anchorage length, bond-line thickness annd
the minimum spacing of rods. There are a number of different design equationns
available to the structuraal engineer to predict the pull out strength of an axiallly
loaded rod. The strength of a laterally loaded rod can be predicted by assuming it
to behave in the same waay as an equivalent dowelled connection [IStructE 20077].
However, there is very liittle guidance in the literature to predict the strength oof
multiple rods, or a single rod under combined axial and lateral loading. Any realiis-
tic timber connection willl require multiple rods to resists a combination of loaad
cases that will rarely prod
duce a pure axial force.
In an effort to improvve the efficiency of long span beams, moment resistinng
connections are being inccorporated into timber structures, most commonly usinng
steel plates and arrays of bolts to transfer the loads between members. The laterral
loading of bolts or screw ws inserted between the wood fibres will always creaate
tension perpendicular to the grain and consequently require a large number oof
fasteners to distribute thee load and avoid splitting failures. The use of glued iin
rods offers exciting prospects for the design of improved timber connections.
The aim of this study is to design a series of experiments to demonstrate thhe
failure characteristics of a moment resisting portal frame connection using glueed
in rods. The design impliications of using glued in rod connections to resist moo-
ments will be discussed. The connection under consideration is a 45 mitred joinnt
illustrated in fig. 1.
Fig. 1 Portal frame connectio

on using glued in rods
Experimental Testing of a Portal Frame Connection Using Glued-In Steel Rods 557
2 Rod Pull Out Tests
The first series of tests were designed to look at the influence of anchorage length
on the failure load of the connection. All tests were carried out using a pull pull
configuration as this replicates most closely the loading exerted on a rod in the
portal frame connection. Tests using three different anchorage lengths were car-
ried out, 100mm, 150mm and 200mm. Increasing the anchorage length beyond
200mm shows a more gradual improvement in ultimate load of the connection
[Riberholt 1986]. The first series of tests used M12 threaded steel rods, grade 8.8,
bonded into 20mm diameter holes. To provide the reaction at the other end M16
grade 8.8 steel rods were bonded into 25mm diameter holes. The steel rods were
bonded into LVL cut to 450mm in length, 118mm wide and 90mm deep.
After a cure time of seven days the specimens were tested in a universal testing
machine fitted with a 100kN load cell. The rods were clamped using wedge grips.
A constant cross head displacement of 1mm/min was applied and the load and
cross head travel was recorded every 0.2 seconds. Displacement of the steel rod
was recorded using two Linear Variable Differential Transducers (LVDT) at-
tached to the sample. The datum plane was created by clamping a flat aluminium
plate to the threaded rod. A sample fitted with LVDTs ready for testing is shown
in fig. 2.
Fig. 2 Direct tension pull out sample with LVDTs
A plot showing the ultimate load with anchorage length is given in fig. 3, show-
ing the experimental data with a least squares linear regression along with the
predictions from the design equations [Riberholt 1988], [Buchanan and Deng
1996], [Gustafsson et al. 2001] and [BS EN 1995-2:2001].
558 J. Walker and R. Xiao
140
Ultimate Load Fax,k (KN) 120
100
Riberholt
80 Buchanan
Gustafsson
60
EC5 1996
40
Experiment
20 Regression
0
0 50 100 150 200 250 300
Anchorage Length la (mm)
Fig. 3 Plot showing increase in ultimate load with anchorage length
It can be seen that the design equation that most accurately matches the exper-
imental data is the theoretical model proposed by Gustafsson [2000]. However,
the unknown quantities required by the equation; shear stress, stiffness and frac-
ture energy have been adjusted to fit the experimental data. Riberholts [1986]
equation is conservative as it limits the maximum ratio of rod diameter to hole
diameter, whereas Buchanan and Dengs [1996] equation over predicts the pull
out load. All the empirical design equations predict a more rapid improvement in
the ultimate load with increasing anchorage length. Ideally the design equations
would be slightly conservative in their prediction yet following the trends of the
experimental data closely.
3 Angled Rod Pull Out Tests
Glued in rods have been studied extensively when loaded in tension. When sub-
jected to a pure bending moment the arrangement of the rods within the portal
frame connection creates a couple to resist the moment through direct tension and
compression as long as whole assembly remains rigid. However, due to the flexi-
bility of the connection, the moment will be resisted laterally as well as axially. By
bending the M12 rod the embedded end could be aligned parallel to the grain of
the timber while the protruding end could be axially aligned with the M16 reaction
rod, Figure 6. This arrangement allows a lateral load to be introduced into the
connection while loading the assembly in tension using the universal testing
machine.
Fig. 4 Dimensions of angled pull out test specimen
An experimental test campaign was designed to pull out angled rods with an
anchorage length of 100mm with a range of angles () between 0 and 45. The
M12 threaded rods were bent to a radius of 10mm and glued parallel to the grain
in 20mm diameter holes. The 16mm rod was then glued into 25mm diameter holes
using a jig to ensure the protruding ends of the rods would be axially aligned.
Aluminium brackets were glued to each side of the timber to enable LVDTs to
be mounted onto the sample. A thick aluminium plate was bolted to the 12mm
threaded rod to provide the reference plane for the LVDTs. The load was applied
using a constant cross head displacement of 1mm/min and measured using
a 100kN load cell. Figure 5 shows the ultimate load of the angled rod with the
axial load on the vertical axis and the lateral load on the horizontal axis.
50
40
Axial Load (kN)
30 Experiment
Theory
20
Linear
10 Elliptic
0
0 2 4 6 8 10 12 14
Lateral Load (kN)
Fig. 5 Angled rod pull out load with varying angle of load
As the angle of load increases, introducing lateral loading into the connection,
the load capacity of the connection decreases rapidly. Following this initial drop
off, the load capacity of the connection reduces linearly with increasing lateral
load. Beyond the limit of lateral load produced experimentally, at a rod angle of
45, it is expected that the ultimate load of the connection will be equal to
the equivalent laterally loaded dowel connection, which according to the design
equation in EC5 [BSI 2008] and a modification factor for embedment strength in
end grain [ANSI 2011] will be approximately 12.2kN.
An elliptic expression can be seen in fig. 5 as a dashed line showing the
allowable combined axial and lateral load as defined in EC5 [BSI 2008] for a
screwed connection. A linear combination of axial and lateral load is shown as the
dotted line.
The failure mechanism of the glued in rod changes after a lateral load of 3kN is
reached. Up to 3kN the failure is characterised by a brittle shear failure (mode II
fracture) in the timber originating from location of maximum shear stress. After
3kN the failure is characterised by a crack opening failure (Mode I fracture) at the
top of the hole, followed by an out of plane shear failure (Mode III fracture) at
each side and finally an in plane shear failure in the timber around the adhesive
(Mode II fracture), fig 6.
Fig. 6 Angled rod pull out failure mechanism at 20
As the angle of the load increases the stiffness of the connection drops rapidly.
This is due to the changing failure mechanism which at high angles is dominated
more by the progressive mixed mode fracture and compression perpendicular to
the grain, rather than the brittle shear fracture of the timber.
4 Portal Frame Connection Tests
This section describes the construction and testing of three moment resisting por-
tal frame connections using glued in steel rods. An estimate of the moment capac-
ity of the portal frame connection is given based on linear elastic beam theory and
results from the pull out tests. The manufacturing process is described along with
the procedure used to load the test specimens until failure. The failure mechanisms
of the three specimens are compared and described and the moment rotation
(M - ) curve of the connection is extracted and discussed.
4.1 Moment Capacity Prediction of the Portal Frame

Connection
The objectives of the experimental tests performed on the portal frame connection
are to ascertain the moment capacity and the moment rotation relationship.
Assuming that the connection is rigid, the bending moment distribution (M) with
the angle of rotation () and deflection () can be easily calculated using elastic
beam theory and can be written in terms of the applied point load (P).
A prediction of the ultimate load of the connection was made using a
modified approach outlined by Fragiacomo [2012]. The internal forces in the
connection can be decomposed into the tension (Ts) and compression (Cs) resis-
tance of the glued in rod and the bearing resistance of timber in compression
(Ct). By taking moments about the point of application of load and assuming
that all materials behave elastically, the balance of internal forces can be found
in terms of the distance to the neutral axis (y), the vertical height of the column
(lv), the breadth of the timber section (b), the depth of the timber section (h),
the tensile stress area of the rod (At), the edge distance of the rods (d e) and the
elastic modular ratio of steel to timber (mst), yielding the 3rd order polynomial
expression (4.1).

+ + 2
6 6 2
(4.1)
+ +2 =0
2
Given the geometry of the connection and the material properties of Kerto LVL
[VTT 2009] the location of the neutral axis can be calculated from the real solu-
tion to equation 4.3. Using the second moment of area of the composite section,
the stress in the rods can be found and used to predict the applied force exerted on
the glued in rods for any given loading.
When the applied point load reaches 22kN the tensile force needed to resist the
moment is at the limit of the axial pull out capacity of the glued in rod according
to the pull out tests described in section 2.0. This corresponds to a bending
moment of 19.7 kNm. The compressive stress in the timber is below the character-
istic strength of LVL in compression. It should be noted that the moment capacity
of the timber section using the characteristic bending strength of the LVL [VTT
2009] is 86.9kNm.
4.2 Portal Frame Connection Manufacture and Test Setup

The design moment capacity for the chosen arrangement of rods was significantly
less than the moment capacity of the LVL beams to ensure that failure of the
connection occurred, rather than a bending failure of the timber. M12 steel rods
were welded to a 300mm x 60mm x 6mm steel plate to form the connection. Two
lengths of 360mm x 90mm LVL were cut to form a mitred joint and the welded
connection was positioned to give an edge distance of 45mm between the central
axis of the rods and the edge of the timber. 20mm diameter holes were drilled into
the end grain of the LVL to a depth of 200mm.
The portal frame was hung upside down and fixed to the reaction frame of a
500kN beam bending machine. The load was applied using a hydraulic jack
mounted between the floor and the beam element and fitted with a 60kN load cell.
A ball bearing fixed between plates was used to ensure that the load continued to
be applied vertically while the beam rotated under the applied load. The rotation
of the connection between the beam and the column was measured using six
LVDTs to assess the location of the neutral axis. Three LVDTs were mounted on
each side of the connection, two measuring separation over the inner rods, two
measuring separation over the outer rods and two half way between the inner and
outer rods, the experimental setup can be seen in fig. 7.
Fig. 7 Experimental setup for testing of the portal frame connection

4.3 Experimental Test Results of the Portal Frame Connection

Three portal frame connections were tested to destruction. During the application
of the load, significant cracks were heard indicating brittle failures. The measured
deformations across the mitre joint provide a means of understanding the gradual
failure of the connection. Figure 8 shows the deformation on the compression and
tension sides of the connection averaged between the LVDT measurements on
each side.
24
20
Load P (KN)
16
12 PF1
8 PF2
4 PF3
0
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Connection Displacement (mm)
Fig. 8 Averaged LVDT displacements measured across the connection
As the applied load approaches 4kN, the joint deforms more in compression
than in tension. Between 4kN and 8kN, all three joints display a brittle failure on
the tension side of the connection, corresponding to a large slip. The deformation
in compression shows an almost linear relationship with load. After the first
failure, PF2 displays very gradual plastic yielding up until the ultimate load is
reached, while PF1/3 show linear behaviour followed by sudden brittle failures up
until the final catastrophic failure. All three test specimens fail by rod pull out on
the tension side caused by fracture in the timber close to the adhesive interface.
The load capacity of the connection predicted using linear elastic theory is very
close to the observed values. However, there are significant brittle failures that
occur before the ultimate load at relatively low loads. The characteristic failure
demonstrates wood fibre fracture, similar in appearance to the failure mechanisms
seen in the angled pull out tests, fig 6.
Assuming that all the compression resistance in the connection comes from the
glued in rod, the tension force can be found by equating moments about the
applied load with moments about the neutral axis. The tension in the rod does not
exceed 60kN, and fails at approximately 90% of the ultimate load for an axially
loaded glued in rod with anchorage length of 200mm. The initial stiffness of the
connection is an order of magnitude greater than the stiffness of the axial pull out
tests. This extra stiffness is attributed to bonding between the steel plate and tim-
ber caused by adhesive flowing into gaps as the separate parts of the connection
were united. After the first brittle failure of the connection, between 15kN and
22kN, the stiffness of the connection has a similar magnitude to the equivalent
200mm rod pull out test. The initial brittle failure in the connection occurs due to
the debonding between the steel plate and timber and does not constitute a major
failure of any of the constituent materials. As the full capacity of the glued in rod
is approached, progressive brittle failures occur to shift the position of the neutral
axis and relieve the tension in the connection. The maximum displacement of the
glued in rod under tension at ultimate load is 2mm.
The tensile force in the glued in rod plotted against the averaged displacement
from LVDTs can be seen in fig. 9.
60
50
Tension Ts (KN)
40
30 PF1
20 PF2
10 PF3
0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Displacement (mm)
Fig. 9 Tensile force in the rod with LVDT measured displacement
For the successful design of moment resisting connections the moment rotation
(M-) relationship must be defined. Using the calculated position of the neutral
axis and the displacement readings from the LVDTs, the rotation across the mitre
joint can be found. The rotation of mitre joint must then be added to the rotation of
the portal frame, found from elastic beam theory to give the final rotation of the
full connection. Figure 10 shows the moment rotation relationship for the three
glued in rod portal frame connection tests.
24
20
Moment M (KNm)
16
Theory
12
PF1
8
PF2
4
PF3
0
0 2 4 6 8 10 12 14 16
Rotation (mrad)
Fig. 10 Moment rotation relationship for the portal frame connection

The moment rotation relationship show very good agreement between different
tests up until a moment of 12kNm is reached. After 12kNm the rotation relation-
ships diverge as the failures become more severe. Only PF3 manages to reach the
predicted failure moment of 19.7kN despite the higher than predicted applied
load. As the connection progressively fails the neutral axis shifts towards the
compression face and the tension in the rod is reduced.
Significant splitting of the timber was observed before the ultimate load of the
connection is reached. Following these splitting failures the connection goes onto
display progressive failure up to the design capacity of the connection for the short
term load tests carried out. However, in a real structure subjected to long term
loading and variations in the environmental conditions, any splitting of the timber
would progressively grow, meaning that the long term load capacity of the con-
nection could be significantly less than observed in the experimental testing. By
modifying the connection there could be a way of sacrificing some of the stiffness
of the connection in order to reduce the splitting failures at low loads and improve
ductility.
5 Conclusion
This study has focused on the design and testing of a portal frame connection
using glued in rods, with the objective of establishing suitable design guidelines to
predict the failure load and moment rotation relationship of the connection.
The pull out capacity of glued in rods with different anchorage lengths have
been established by experimental testing and compared to the currently available
design equations highlighting significant differences. Between different design
equations, the variation in predicted pull out strength was as much as 50%.
Riberholts design equation provides the only conservative estimate of strength
and limits the strength gain after a certain anchorage length has been reached, a
trend observed in many experimental investigations and verified by theoretical
studies of single lap joints. Gustafssons design equation gives the most accurate
prediction of pull out strength, but fracture properties of the glued connection
must be established before it can be used, which makes this approach cumbersome
for the practicing structural engineer.
Further experimental tests demonstrate how the strength of a glued in rod is
reduced when loaded laterally as well as axially. As the proportion of lateral load
on the rod increases the failure mode changes, which accounts for the reduced
load capacity at failure. In all real structures any connection will be subjected to
axial and lateral loading, so further investigations are required to understand
the geometrical parameters that influence the strength of a glued in rod under
combined loading.
Destructive testing was carried out on half portal frames connected using a 45
mitred joint. The beam and column elements were connected using steel rods
welded to a steel plate bonded into the timber. The results from experimental test-
ing show that glued in rods can be used to create an efficient moment connection
system. However, there are significant brittle failures that occur before the ulti-
mate load is reached that could have implications for the long term performance of
the structure. A theoretical model has been presented which estimates the ultimate
load and bending moment at failure with a good degree of accuracy.
References
ANSI/AWC, NDS-2012 National Design Specification (NDS) for Wood Construction.
ANSI/AWC (2011)
British Standards Institution, BS EN 1995-2:2001 Design of timber structures. Bridges.
London BSI (2001)
British Standards Institution, BS EN 1995-1-1:2008 Design of timber structures. General.
London BSI (2008)
Broughton, J.G., Hutchinson, A.R.: Experimental verification of design calculations for the
development of design procedures for repairs. Low Intrusion Conservation Systems for
Timber Structures LICONS Task 4.1 CRAF-1999-71216 (2004)
Buchanan, A.H., Deng, X.J.: Strength of epoxied steel rods in glulam timber. In: Proceed-
ings of International Wood Engineering Conference, IWEC 1996, New Orleans, October
28-31, vol. 4, pp. 488495 (1996)
Coureau, J.L., et al.: Strength of PGF reinforced end-notched beams. In: Proceedings of the
International RILEM Symposium on Joints in Timber Structures, Stuttgart, Germany,
pp. 413422. RILEM Publications SARL (2000)
Crocetti, R., et al.: Compression strength perpendicular to grain full scale testing of glu-
lam beams with and without reinforcement. COST Action FP1004 Enhance Mechanical
Properties of Timber, Engineered Wood Products and Timber Structures, Zagreb,
pp. 5156 (2012)
Fragiacomo, M., Batchelar, M.: Timber Frame Moment Joints with Glued-In Steel Rods.
I: Design. Journal of Structural Engineering ASCE, 789801 (2012)
Gustafsson, et al.: A strength design equation for glued-in-rods. In: Proceedings of the
International RILEM Symposium on Joints in Timber Structures, Stuttgart, Germany,
September 12-14, pp. 323332. RILEM Publications SARL (2000)
IStuctE, TRADA, Manual for the design of timber building structures to Eurocode 5.
TRADA (2007)
Madsen, B.: Behaviour of Timber Connections. Timber Engineering Ltd. (2000)
Riberholt, H.: Indlimede bolte til indfstning af vingerne p Nibe-Mlle B. Technical
University of Denmark Series R No. 167 (1983) (in Danish)
Riberholt, H.: Glued bolts in Glulam Technical University of Denmark Series R No. 210
(1986)
TRADA, Assessment and repair of structural timber. Wood Information: Section 1, Sheet
34 (1992)
Part VI
Hardwood, Modified Wood
and Bamboo
Bending Strength and Stiffness of Glulam
Beams Made of Thermally Modified Beech
Timber
Robert Widmann1, Wilfried Beikircher 2, Jos L.F. Cabo3, and Ren Steiger1
1
EMPA, Swiss Federal Laboratories for Materials Science and Technology, Structural
Engineering Research Laboratory, Duebendorf, Switzerland
robert.widmann@empa.ch
2
University of Innsbruck, Institute for Structural Engineering and Material Sciences,
Innsbruck, Austria
3
Universidad Politcnica de Madrid (UPM), ETS Arquitectura (ETSA), Structural
Department (SD), Madrid, Spain
Abstract. The paper describes tests carried out on structural glued laminated tim-
ber (glulam) beams and finger-jointed boards made out of thermally modified
hardwood (beech, fagus sylvatica) in the following named as TMTB. The finger
joints were bonded with a two-component PRF adhesive and the lamellas were
edge-bonded using a two-component MUF adhesive. The finger jointed lamellas
were tested in tension, flatwise- and edgewise bending. While automatically pro-
duced finger joints mostly showed unsatisfactory strengths, it was possible with
manually produced finger joints to achieve higher strength values. Fifty glulam
TMTB beams were produced to evaluate their load carrying behaviour. The beams
were tested in 4-point bending and the integrity of the glue lines was verified by
means of delamination tests and shear tests. Usually it is expected that combining
lamellas of a certain strength class to a glulam beam will enhance certain charac-
teristic mechanical properties of the final product compared to the properties of
single boards. The results of the tests could not confirm this behaviour for the
TMTB glulam beams even if the bond lines proved to be of a satisfactory quality.
Hence, a structural use of TMTB glulam seems to be restricted to a limited range
of applications.
Keywords: Glulam, thermally modified timber (TMT), beech wood, finger joints,
bending strength, adhesives, phenol-resorcinol formaldehyde (PRF), melamine
urea formaldehyde (MUF).
1 Introduction
In the past 5 to 10 years products made out of thermally modified timber (TMT)
have been increasingly used for a wide field of applications. For outdoor use its
DOI: 10.1007/978-94-007-7811-5_52, RILEM 2014
570 R. Widmann et al.
improved durability and dimensional stability upgrades TMT as a potential substi-

tute for tropical hardwoods or impregnated softwoods. For indoor use as furniture
or flooring TMT is becoming a competitor to dark coloured tropical hardwoods
due to the wide range of possible colours resulting from the thermal treatment.
Within the EC-funded FP6 project Holiwood [1] which was running from 2005
to 2009, it was intended to widen the field of uses for TMT made out of European
hardwoods to structural applications. In [2] it was shown that thermally modified
beech (fagus sylvatica) solid timber (TMTB) with an intensive modification in
general has some limitations regarding its structural use. In particular the brittle-
ness of the material and high variations of the strength values were identified as
reasons for its relatively low performance. Therefore the lamination of glued
products like glulam was seen as one possible measure for enhancing the proper-
ties of TMTB due to the homogenization effect. In addition the available range of
cross cut timber from hardwoods especially from beech as well as the require-
ments coming from the heat treatment process limit the possible dimensions and
cross sections of solid thermally modified timber members. Therefore it is re-
quired to use glued products within structures like it is the case in many common
timber constructions made out of softwoods. The heat treatment process at the
company involved in the research project (Mitteramskogler GmbH, Austria) is
limited to thicknesses of about 60 mm. However, square-cut timber members
made of TMTB with these thicknesses showed considerable deformations like
bow and twist which finally lead to reduced usable maximum thicknesses to about
45 mm to 50 mm. As a consequence glued products instead of solid timber have to
be used already for relatively small dimensioned (cross-section) timber members.
For the development of glued structural products made out of TMTB the
production and testing of finger-jointed boards is the first step. However, some
preliminary series of finger-jointed TMTB boards produced according to standard
procedures showed partly very low minimum bending strengths (6 N/mm2), thus
the assessment of suitable finger geometries, adhesive(s) and procedures for the
production of improved finger-joints in TMTB was necessary.
In the following the results of bending and tension tests on finger-jointed
TMTB lamellas and untreated beech lamellas as well as bending, delamination
and shear tests of glulam TMTB-beams are presented.
2 Experimental
2.1 Material
The raw material for the finger-jointed lamellas and the glulam beams consisted
of high visual quality boards made of TMTB "Buche forte" [3] produced by Mit-
teramskogler GmbH in Austria (A). The thermal treatment details are confiden-
tial can be considered as being a strong treatment that enables the TMTB being
assigned to durability class 3 according to EN 350-1 [4]. The finger joints
were produced manually at Mitteramskogler GmbH as well as automatically at
Bending Strength and Stiffness of Glulam Beams 571
Obermayr GmbH (A). Obermayr also produced the glued laminated members
using standard procedures and equipment.
2.2 Finger-Joints and Glulam Beams

For the production of the fingers two different cutter heads were selected, result-
ing in finger geometries of 20 mm x 6.2 mm (Series 1, "FJ1") and
15 mm x 3.8 mm (Series 2, "FJ2") as shown in Figure 1. The lamellas were cut in
two pieces, the fingers were cut and the same pieces were rejoined and glued. All
joints were produced manually using a two-component phenol resorcinol formal-
dehyde (PRF) adhesive (Dynea PRF Prefere 4099 and hardener Prefere 5827). The
bonding pressure was applied manually by clamps.
Fig. 1 Finger jointed boards made of TMTB and untreated beech. The finger geometries
were 20 mm x 6.2 mm "FJ1(left) and 15 mm x 3.8 mm "FJ2 (right). A PRF adhesive was
used and the bonding-pressure was applied manually by means of clamps.
The lamellas for the glulam beams were finger-jointed automatically and
had the geometry "FJ2". For gluing these fingerjoints as well as the lamellas a
melamine urea formaldehyde (MUF) adhesive (Dynea Prefere 4535 adhesive in
combination with a Prefere 5035 hardener) was used.
Tables 1 and 2 give an overview on the finger-jointed specimens (FJ) subjected
to bending and tension tests and on the tested glulam beams (GL) respectively.
2.3 Procedures
The bending tests of the finger-jointed boards as well as of the glulam beams
were executed as 4-point bending tests according to EN 408 [5]. For the tension
tests a special testing device with a clamping length of 300 mm and a free length
between the clamps of 200 mm was used (Fig. 2). Delamination and block shear
tests were carried out according to EN 391 (Method B) [6] and EN 392 [7]
respectively.
Table 1 Finger-jointed TMTB- and untreated beech (B) specimens
Test Flatwise bending Edgewise bending Tension

series FJ1 FJ2 FJ1 FJ2 FJ1 FJ2
[mm] 620 2005 800
eff [mm]* 540 1800 200
b [mm] 120 30 120
h [mm] 30 120 30
TMTB 18 18 16 16 10 12
n[]
B 16 18 15 16 12 14
TMTB 610-750 590-785 600-730 610-790 650-715 675-760
u [kg/m3]
B 640-800 635-810 645-790 650-830 685-815 675-795
TMTB 5.4-7.4 5.3-6.0 5.1-7.4 5.3-6.2 5.5-6.9 5.1-6.3
u [%]
B 10-12 10-12 11-12 10-12 11-12 10-12
* eff corresponds to span for bending tests and free length between the clamps for tension tests
Table 2 Glulam beams (GL1 and GL2) and automatically produced finger-jointed lamellas
(FJ GL1) tested in bending
Glulam bending Flatwise bending

GL 1 GL 2 FJ GL 1
Length [mm] 2050 5140
Span eff [mm] 1845 4860 540
b [mm] 140 120 120
h [mm] 120 270 30
Lamella h [mm] 30 30
n[] 10 40 56
u [kg/m3] 620-680 630-700 620-680
u [%] 5.0-6.5 5.7-6.8 5.0-6.5
3.1 Finger Jointed Boards

Within all test series the reference beech specimens (B) failed in the adhesive
layer or at the wood-adhesive interface of the fingers. As wood fibers could only
be observed on less than 40% of the glued finger surface, adhesive failure was
determined as the major failure mode. In contrary the TMTB samples failed
predominantly in the wood. Failure at the base of the fingers could be observed as
well as wood failure that appeared to be independent of the presence of the joint.
This was in particular the case for the specimens subjected to tensile tests, where
only 20% of the specimens failed predominantly in the adhesive layer.
The test results are shown in Figure 2. It can be seen that for every combination
of wood and finger geometry the strength ranking (from high strength to low
strength) flatwise bending edgewise bending tension is the same.
Set up
120
fm,f fm,e ft
100
80
60
40
20
CoV 7.8 28 15 27 13 25 10 23 15 28 12 31 [%]
0
FJ1-B FJ1-T FJ2-B FJ2-T FJ1-B FJ1-T FJ2-B FJ2-T FJ1-B FJ1-T FJ2-B FJ2-T
FJ
Fig. 2 Boxplots of flatwise and edgewise bending (fm,f and fm,e) as well as tension strength
(ft) in N/mm2 of finger-jointed boards for two finger-joint (FJ) geometries (1 and 2) made
of unmodified (B) and thermally modified (T) beech. The bending strength of the TMTB
corresponds to strength class GL24 according to EN 1194. The boxplots represent
minimum and maximum values as well as quartiles and mean values (x).
About 50% of the tension specimens FJ2-T failed inside the clamps of the
tensile testing machine under low failure loads.
The small number of specimens strongly influenced the determination of 5-
percentile strength values (which will be lower than the shown minimum values)
and therefore the results can only be taken as an estimation of the potential of such
types of finger-joints. Whereas the strength of the untreated beech samples was
comparable to values found in literature [8], the TMTB samples performed con-
siderably weaker. Anyhow the characteristic bending strength of the FJ2-T sample
was sufficient to use these lamellas for the production of GL24 glulam according
to EN 1194 [9].
3.2 Glulam Beams

The glulam beams showed typical bending failures. The initial failure occurred in
the tension zone most often at or close to a finger-joint. Other lamellas in the
tension zone failed due to excessively deviated grain or due to existing (micro)
cracks resulting from the production process. The overall bending strength was
found to be poor (Figure 3) in regards of the superior raw material that had been
used. In addition the strength values showed a high variation. It also has to be kept
in mind that the reference depth for glulam beams is 600 mm [9] and the tested
specimens had depths of only 120 mm and 270 mm respectively. Therefore the
bending strength of specimens, which would meet the reference depth, can be
expected to be even lower than those shown here.
70 20
60 19
50
fm [N/mm2]
18
E0 [kN/mm2]
40
17
30
16
20
10 15
CoV 31.8 32.3 34.9 [ %] CoV 6.0[ %]
0 14
GL 1 GL 2 FJ-GL 1 GL 2
Fig. 3 Left: Bending strength of two series of TMTB glulam beams. Right: Bending
MOE of one series of TMTB glulam beams. The MOE was not determined for beam
series 1.
The unsatisfactory bending performance can be mainly attributed to the brittle

behavior of the TMTB. Already in the production process cracks appeared in the
course of automatically finger-jointing and in the gluing of the lamellas. This was
also confirmed by the low flatwise bending strength of the series 1 finger joints
(Figure 3). If these values are compared to those shown in Figure 2 it can be
clearly seen that the performance of automatically produced finger-joints is
considerably weaker compared to the manually produced joints. However, the
tests with manually fabricated finger joints shown above show, that with an
adapted production process it should be possible to provide finger joints with
sufficient strength.
In general the obtained strength values were disappointing regarding the
superior wood quality (raw material) and these values in addition were inferior to
those found for TMTB solid wood subjected to the same heat treatment [2]. On
the other hand regarding stiffness the TMTB beams showed promising values
(Figure 3, right) with E0,mean = 17.8 kN/mm2 well above all glulams listed in EN
1194. These values also corresponded to such found in [2] for solid TMTB timber.
The delamination tests of sections taken from the glulam beams according to
EN 391 [6], method B showed a high degree of delamination. The unsatisfactory
results can be seen in Figure 4, left. All specimens outside the shown envelope
with a maximum delamination > 4% and/or 40% maximum delamination of single

glue lines did not meet the requirements. Only 30% of the tested specimens
fulfilled the EN 386 [10] requirements. On base of the delamination tests the
adhesive and/or production process has to be regarded as not being suitable for
TMTB to be used in load-carrying structures.
Opposite to that, the block shear tests according to EN 392 showed good results
with a mean shear strength of 13.8 N/mm2 and a mean wood failure percentage
exceeding 80%. Therefore, on base of the shear tests and regarding the require-
ments of EN 386 the used adhesive can be considered as being suitable for gluing
TMTB lamellas. However, about 2.5% of the single values did not meet the
requirements (Figure 4, right). In some of these cases an incomplete or insufficient
gluing could be found. The production process and/or quality control have to be
adapted in order to guarantee that such gluing failures cannot develop.
60 100
Maximum delamination [%]
50 80
Wood failure WFP [%]
40
60
30
40
20
10 20
0 0
0 5 10 15 20 0 5 10 15 20
Total delamination [%] Shear strength fv [N/mm2]
Fig. 4 Results of the delamination tests (left) according to EN 391 (method B) and shear
tests according to EN 392 (single glue lines, right). All specimens within the indicated
envelopes fulfill the requirements (for softwood, according to EN 384). For the delamina-
tion tests a second cycle (dashed line) was not applied.
4 Conclusions
Manually produced finger-joints in TMTB boards showed higher strengths than

automatically produced finger-joints. An adapted processing of the finger joints is
needed for a successful industrial production of high strength finger-joints in
TMTB.
Glulam beams made of TMTB showed high stiffness, but poor bending
strength (at the 5% level) and a high variation of the strength values. In
consequence the tested TMTB glulam could only be used for a limited range of
The used MUF adhesive turned out to be suitable for gluing lamellas and finger
joints made of TMTB and the used PRF showed a good performance for gluing
finger joints in TMTB.
Acknowledgement. The presented work was financially supported by the European

Commission under contract No. NMP2-CT-2005-011799 (HOLIWOOD project).
References
1. Schftner, R.: Holiwood International Research and Development for Innovative
Products made out of Thermal Modified Timber. In: 3rd European Conference on
Wood Modification, Cardiff, UK, pp. 235238 (2007)
2. Widmann, R., Fernandez-Cabo, J.L., Steiger, R.: Mechanical properties of thermally
modified beech timber for structural purposes. European Journal of Wood and Wood
Products 70(6), 775784 (2012)
3. Mitteramskogler, MIRAKO-Thermowood (2008)
4. CEN, EN 350-1: Durability of wood and wood-based products - Natural durability of
solid wood - Part 1: Guide to the principles of testing and classification of the natural
durability of wood (1994)
5. CEN, EN 408: Timber structures - Structural timber and glued laminated timber - De-
termination of some physical and mechanical properties (2010)
6. CEN, EN 391: Glued laminated timber - Delamination test of glue lines (2001)
7. CEN, EN 392: Glued laminated timber - Shear test of glue lines (1995)
8. Blass, H.J., et al.: Bending strength of glulam made of beech (in German:
Biegefestigkeit von Brettschichtholz aus Buche). In: Karlsruher Berichte zum
Ingenierholzbau No. 1. Universittsverlag Karlsruhe, Karlsruhe (2005)
9. CEN, EN 1194: Timber structures - Glued laminated timber - Strength classes and
determination of characteristic values (1999)
10. CEN, EN 386: Glued laminated timber - Performance requirements and minimum
production requirements (2001)
Structural Veneer Based Composite Products
from Hardwood Thinning Part I: Background
and Manufacturing
Ian D. Underhill1, Benoit P. Gilbert1, Henri Bailleres2, Robbie L. McGavin2,

and Dale Patterson3
1
2
3
Queensland Collage of Arts, Griffith University, Australia
Abstract. In Australia, plantation forests have increased in area by around 50% in

the last 10 years. While this expansion has seen a modest 8% increase for soft-
woods, hardwood plantations have dramatically increased by over 150%. Hard-
wood plantations grown for high quality sawn timber are slow to mature, with a
crop rotation time potentially reaching 35 years. With this long lead-time, each
year the risk from fire, pests and adverse weather events dramatically increases,
while not translating into substantially higher financial returns to the grower. To
justify continued expansion of Australias current hardwood plantation estate, it is
becoming necessary to develop higher value end-uses for both pulpwood and
smaller sawlog resources. The use of the low commercial value stems currently
culled during thinning appears to be a necessary option to improve the industry
profitability and win new markets. This paper provides background information
on Australian forests and plantations and gives an overview of potential uses of
Australian hardwood plantation thinning logs, as their mechanical properties.
More specifically, this paper reports on the development of structural Veneer
Based Composite (VBC) products from hardwood plantation thinning logs, taking
advantage of a recent technology developed to optimise the processing of this
resource. The process used to manufacture a range of hollow-form veneer lami-
nated structural products is presented and the mechanical characteristics of these
products are investigated in the companion paper. The market applications and
future opportunities for the proposed products are also discussed, as potential
benefits to the timber industry.
Keywords: veneer based composites structural products, Australian hardwood

plantation forests, plantation thinning.
DOI: 10.1007/978-94-007-7811-5_53, RILEM 2014
578 I.D. Underhill et al.
1 Introduction
When establishing hardwood plantations for sawn timber, significant initial in-
vestment is essential to produce high quality logs at an early age (between 30 to
35 years) that have minimal knots and sufficient log diameter to produce high
value sawn boards. Early pruning (i.e. removal of lower branches) and thinning
(i.e. removal of low quality trees that are crooked, small or have too many branch-
es), which increases the available light, moisture and nutrients to the remaining
stand of trees, is essential during the early years of the plantation [1, 2].
For improved efficiency, a recommended stocking rate of around 1000 trees per
hectare is usually adopted [3], and up to 700 trees can be removed before the plan-
tation is clear felled in usually two main thinning operations. Nearly half of the
trees are typically cut at 1.5 to 3 years in the first thinning, with another 300 being
removed in the second thinning at 10 to 15 years.
Trees removed during the second thinning process usually have a Diameter at
Breast Height Over Bark (DBHOB) of 150 mm to 300 mm and are generally
deemed to have little commercial value. To justify continued expansion of Aus-
tralias current hardwood plantation estates, it is becoming necessary to develop
high value end-use products for these small logs [2], as there are no clearly identi-
fied viable markets for these resources. Pulpwood is currently regarded as the only
large scale viable option for these logs if the plantation is in close proximity (less
than 100 km) to a process facility and/or forms part of an overall harvest scheme
[2].
This paper presents the initial joint research undertaken at Griffith University
and the Salisbury Research Centre, Department of Agriculture, Fisheries and For-
estry (DAFF), Queensland Government, to manufacture a range of high-value
hollow-form laminated Veneer Based Composite (VBC) products from hardwood
plantation thinning. Specifically this paper provides an overview of Australian
hardwood plantations, published various thinning applications and thinning me-
chanical properties. Laboratory based manufacturing process of the VBC products
is introduced. This paper provides a review of potential market applications and
future opportunities for the proposed VBC products. Planned future research is
also discussed.
2 Background on Australian Forests and Plantations
It is estimated that about 30% of Australia was covered by forests prior to the
arrival of European settlers in the 18th century. Today, about 40% of these forests
have been lost, with the remaining native vegetation being highly fragmented [4].
Specifically, 149.4 million hectares of Australia is currently covered by forests,
principally dominated by more than 700 different species of Eucalyptus. About
two million hectares (1.34%) of these forests are managed as plantations [5].
When establishing a plantation, a single species is usually chosen. To allow
maximum yield in the shortest possible rotation time, industry requirements and
Structural Veneer Based Composite Products from Hardwood Thinning 579
environmental conditions are analysed and matched with an appropriate timber

species [6]. In Australia, the yield from hardwood plantations can be up to 14%
higher than from native forests [7] offering many financial and environmental
advantages to the growers.
Overall, from 2000 to 2010 plantation forests in Australia have increased by
about 51%. This equates to a significant increase of 150% for hardwood planta-
tions and an 8% increase for softwood plantations, as shown in Figure 1. This
increase can be attributed predominantly to changes in government policies [8].
Today, a greater understanding exists between the need to balance the use of tim-
ber and the environmental and social importance of maintaining a diverse range of
forests. Agreements that ban logging native forests have been signed around
Australia [9], resulting in the promotion of alternative options, such as plantation
timber.
Fig. 1 Plantation expansion between 1995 and 2009 [8]
Less than 10% of the hardwood plantations in Australia are grown for sawn
timber. The remaining 90% are mainly grown for pulpwood and exported as un-
processed woodchips. Yet, partly due to the price of Australian woodchips being
internationally uncompetitive, Australia has seen the recent collapse of major
companies such as Willmott Forests, Great Southern Plantations and Timbercorp
[10]. With this in mind, many growers have delayed or abandoned harvesting their
pulpwood plantations until the market improves. These plantations are usually
clear-felled at 10 to 12 years of age without prior thinning or pruning. In this type
of plantation, species selection and breeding have focused on achieving high pulp
yield, density and volume, which may adversely affect important mechanical
properties for high value-added solid wood products [11, 12]. These low-quality
logs of similar size to thinning need also to be considered in creating higher end-
value products, such as the proposed VBC products in this study, and shift the
focus of plantation management and harvesting practices. Moreover, as seen in
Figure 1, more than 400,000 hectares of hardwood plantations have been planted
around the year 2000 and a large quantity of thinning logs will be available in the
coming years, reinforcing the need to develop a market for this type of resource,
as well as pulpwood logs.
3 Literature Review on Applications for Plantation Thinning
3.1 Round-Wood Form

The most cost effective ways of utilising plantation thinning is to keep the logs in
their natural round-form. Studies have highlighted potential applications for de-
barked logs, such as vineyard posts [13] or timber piles and posts for the building
and landscaping industries [14-16]. These markets are well established in Austral-
ia, and are dominated by plantation softwoods, which are easy-to-work and readily
available. Yet, hardwood plantation thinning logs have the advantage of being
more durable and having higher mechanical properties, but for these logs to com-
pete favourably, current problems with excessive splitting and checking, as
depicted in Figure 2, need to be addressed [2].
Fig. 2 Split hardwood plantation thin- Fig. 3 Furniture designed from hardwood plan-
ning round wood logs (from [2]) tation thinning timber (from [2])
3.2 Solid Wood Products

The use of hardwood plantation thinning for solid wood applications has also been
successfully investigated in the fabrication of furniture by Training and Further
Education (TAFE) students, as shown in Figure 3. For more industrial applica-
tions, trials were also conducted where sawn timber roof trusses and pallets were
successfully manufactured [2]. High rates of rejection (more than 60%) were
reported in these studies due to defects in the material or subsequent material
processing [2, 17].
3.3 Structural Products

The Salisbury Research Centre investigated the use of thinning for structural ap-
plications from three hardwood species, namely Gympie messmate (Eucalyptus
cloeziana), blackbutt (Eucalyptus pilularis) and red mahogany (Eucalyptus
pettita), and found that sawn thinning logs make attractive structural products if
appropriately graded [2].
Material testing has also shown that, whilst having lower strength, hardness and
density than the mature timber, the juvenile timber of these three species still has
favourable properties when compared to radiata pine (Pinus radiata), a common
softwood plantation timber grown in Australia. Additionally, the juvenile timber
was found to have more favourable shrinkage and stability results than the mature
tree [2]. Table 1 shows average reported material properties for juvenile and ma-
ture Gympie messmate and radiata Pine.
Other structural products investigated in [2] included plywood sheets and, more
specific to the VBC products presented in this paper (see Section 4), Laminated
Veneer Lumber (LVL) products. LVL products are manufactured by gluing ve-
neer sheets together to form composite timber boards. Unlike plywood, the grains
of all veneers are orientated in the same direction. The laminated manufacturing
process randomises the location of the defects in the veneers and produces more
uniform structural products, with higher strength and typically less variability in
mechanical properties, than solid wood [18, 19]. These improved mechanical
properties are especially important for thinning veneer based products, as thinning
logs present a high proportion of natural defects (knots, resin veins etc.), as shown
in Section 4.
Table 1 Average material properties (from [2, 19])
Gympie Messmate
Juvenile (less Mature (between
Material Properties than 8 year-old) 25 to 35 year-old) Radiata Pine
Basic Density 648 kg/m3 810 kg/m3 400 kg/m3
Shrinkage Tangential
4.6% 5.9% 4.5%
(green to 12% MC)
Shrinkage Radial
2.02% 4.2% 3%
(green to 12% MC)
Janka Hardness Test 6.8 12 3.3
Modulus of Elasticity
11.8 GPa 17 GPa 10 GPa
(MOE)
Bending Modulus of
106 MPa 137 MPa 81 Mpa
Rupture (MOR)
The peeling of hardwood thinning logs using traditional peeling technology

with driving dogs proves difficult [20] and generates poor yield, as the lathe is not
well suited for low quality (i.e. large proportion of defects) and small diameter
logs. On the other hand, spindle-less veneer lathe has proved well suited for pro-
cessing small diameter logs, opening opportunities for the proposed thinning VBC
products in Section 4. This type of lathe generally converts 80 to 90% of a log into
green veneers, leaving an unpeeled core of about 45 mm diameter, as opposed to
more than 130 mm diameter unpeeled cores in traditional peeling systems, as
shown in Figure 4.
(a) (b)
Fig. 4 (a) Cores from spindle-less (left) and traditional (right) lathes, (b) thinning logs plus
spindle-less core
4 Current Research, Development of Veneer Based

Composite
4.1 General
Samples of VBC products were manufactured at Griffith University using 15 year
old plantation thinning Gympie messmate (Eucalyptus cloeziana). This species is
currently one of the preferred hardwood plantation species for future development
by the Department of Agriculture, Fisheries and Forestry, Queensland Govern-
ment, in Queenslands sub-tropical areas [2]. Specifically in native forest, Gympie
messmate is largely found in South-east Queensland. It predominantly occurs in
tall open-forests and woodlands and is usually the dominant species in the stand.
(a) (b) (c) (d)

Fig. 5 Examples of 15 year-old juvenile Gympie Messmate veneers used in this study from
(a) low to (d) high proportion of defects
Under ideal growing conditions this species can attain 55 m in height with
DBHOB of 2 m. The heartwood is typically yellow-brown of even texture with a
basic density of about 810 kg/m3 and the highest durability class [19].
The logs used in this study had diameters of about 230 mm and were cut in 1.2
m lengths before being rotary peeled into nominal 1.7 mm thick veneers at the
Salisbury Research Centre. Figure 5 shows some typical veneers ranging from
veneers with low (Figure 5 (a)) to high (Figure 5 (d)) proportion of natural defects.
Samples of hollow-formed VBC products, namely circular hollow sections
(CHS) (see Figure 6 (a)), rectangular hollow sections (RHS) (see Figure 6 (b)),
Cee-sections (see Figure 6 (c)) and I-sections (see Figure 6 (d)) were manufac-
tured using the process described in Section 4.2. In addition to being manufactured
from waste material, advantages of the new products over sawn timber sections lie
with the products (i) having efficient cross-sectional shapes, (i.e. hollow, Cee or I
shapes) and (ii) being able to be manufactured in large sizes currently not availa-
ble in timber. Similar cross-sectional shapes have been successfully used in timber
construction, mainly with glued composite box beams and I-beams. The latter
products typically have a wood based panels for the web (i.e. plywood and parti-
cleboard for instance), and structural timber for the flange (i.e. LVL or sawn
timber for instance).
(a) (b) (c) (d)
Fig. 6 Sample of profiles manufactured at Griffith University (a) circular hollow section,
(b) rectangular hollow section, (c) Cee-section, and (d) I-section
4.2 Manufacturing Process

A self-reacting frame was designed and manufactured (see Figure 7 (a)) to con-
struct the VBC sections. The manufacturing concept uses the natural tendency of
the veneers to roll about one side, making their gluing into half-shapes achievable
around a mandrel, with the veneer grain orientated in the same direction. Specifi-
cally, a pre-glued veneer stack is first sandwiched between Teflon sheets to mini-
mise friction and allow the stack to shape around the mandrel. The stack is then
positioned inside a series of flexible outer straps (see Figure 7) and a hydraulic
jack drives the mandrel down, compressing the veneer stack between the mandrel
and the outer straps. Rubber sheets are also inserted between the straps and the
Teflon sheets to uniform the applied pressure and compensate non-uniform thick-
ness of the veneers. By changing the shape of the inner mandrel, a variety of
(a) (b)
Fig. 7 (a) 30T Self-reacting frame and (b) detailed view
different VBC sections can manufactured, as shown in Figure 8 (a) for circular
hollow sections and Figure 9 (a) for rectangular hollow sections, Cee-sections and
I-sections.
For circular hollow sections in Figure 8 (a), the applied pressure P to the veneer
stack is given in terms of the applied force F as,
F
P= (1)
2 RL
where R is the radius of the cross-section and L the length of the profile. For rec-
tangular cross-sectional shapes in Figure 9 (a), external pressure has to be applied
to the flat sides of the profile. This process is currently being improved, using a
vacuum press for instance.
To form complete cross-sections, two half-shapes are typically butt joined to-
gether glued back-to-back as shown in Figure 8 (b) and Figure 9 (b).
(a) (b)
Fig. 8 (a) Manufacturing of half-round shapes and (b) forming the circular hollow sections
(b)
(c)
(a)
Fig. 9 (a) Manufacturing of Cee-sections, (b) forming the I-sections and (b) forming the
rectangular hollow sections
4.3 Opportunities
Various market opportunities have been identified for the proposed products.
Circular hollow sections are currently being structurally investigated in the
companion paper [21] as an alternative solution for Australias supply shortage
of high quality utility poles (also referred to as power poles). Hollow poles
would be slid in a cylindrical concrete footing, or galvanised steel tube bolted
to a concrete pad. Connecting the hollow poles above ground would considera-
bly limit termite attacks and premature rotting conditions, thus allowing using
less harmful wood treatment chemicals than currently practised in solid poles.
Moreover, the proposed poles would offer lower transportation and installation
costs when compared to solid timber, concrete and steel poles.
Potential applications of rectangular hollow sections include the need high-
lighted by the energy industry for large sections of hardwood timber for cross-
arms of utility poles.
Beyond the energy industry, since these components have high stiffness and
strength per unit weight, it is believed that the hollow cross-sectional shapes
have many potential applications in the building industry, either in large or
small cross-sections. The proposed VBC products can be industrially produced
in standard lengths and marketed in conjunction with a range of click-lock fit-
tings or more permanent connections, designed to rapidly join together lengths
of sections in trusses, beams or columns for instance. Specifically, the click-
lock fittings would allow rapid assembly and disassembly of pre-designed
structures, such as flat-packed house frames, roof trusses, emergency housing
structures or DIY projects. Additionally, the hollow central void can be used to
accommodate services, such as electrical, lighting, gas or water.
I-sections manufactured from plantation thinning can be both used for structur-
al and decorative applications in residential and the commercial buildings,
where architecturally exposed elements are sought.
The use of these forms is currently being investigated with architecture students at
Griffith University for both structural and architectural applications.
5 Future Researches and Challenges
A constraint of the rotary peeling process is the short log lengths that spindle-less
lathes can accommodate, 1.2 m at the Salisbury Research centre for instance. To
manufacture usable lengths (typically 3 m for columns and 8 m for beams) of
VBC products for engineering applications, subsections need to be jointed togeth-
er or manufactured in a continuous process, similar to LVL products.
Due to their cross-section shape, the walls of the proposed VBC products in
bending will mainly undergo compression, tension or shear, and various joint
details were therefore preliminary investigated in tension in a 500 kN MTS uni-
versal testing machine, at ambient temperature and humidity. LVL flat panels
were manufactured from thinning veneers using a two parts phenol-resorcinol
formaldehyde structural adhesive and six coupons (dog bone samples) were cut
from each panel. Three coupons were used as control samples, while the remain-
ing coupons were cut in half and joined back together. Lap, finger, butt and vari-
ous scarf joints were investigated using either resorcinol adhesive or structural
epoxy. Likely due to higher level of extractives present in juvenile Gympie mess-
mate than in mature trees [19, 20], the strength of the joints were significantly less
than the control samples, ranging from an average of 9.4% (butt joint with Phenol-
resorcinol) to 42.6% (300 scarf joint with epoxy) of the strength of the control
samples. Therefore, joining lengths of VBC products together present a challenge.
The Salisbury Research Centre is currently working on improving adhesives for
use with juvenile hardwood thinning and steel-timber connections, consisting of
inserting a steel sleeve in the hollow-form of similar stiffness and strength to the
VBC product, are currently looked at.
Finally, for VBC products to be marketed, research is also required to develop
design rules, enabling engineers and architects to safely use the proposed prod-
ucts, as further detailed in the companion paper [21].
6 Conclusions
This paper presented the development of Veneer Based Composite structural

products manufactured from hardwood plantations thinning. Currently, hardwood
plantations grown for sawn timber only achieve financial benefits on tree maturity
(i.e. at 25 to 35 years) and create a challenging economic business model. Devel-
oping high-value products from thinning can generate revenue at plantation
mid-cycle (i.e. between 10 to 15 year) and therefore significantly benefit the Aus-
tralian plantation and timber industries. Specifically, if markets can be developed
for the proposed products, growers in both the sawn timber and pulpwood markets
may switch and clear fell their crops early, solely for the production of VBC
products.
This paper also introduced the manufacturing process of the proposed
VBC products, potential applications and research opportunities currently being
investigated.
Acknowledgments. The authors would like to thank the Department of Agriculture, Fisher-
ies and Forestry, Australian Government and Forest and Wood Products Australia for their
financial support through the 2012 Science and innovation award for young people in
agriculture, fisheries and forestry, Forestry category.
References
[1] Geoff, R., Smith, B., Brennan, P.: First thinning in sub-tropical eucalypt plantations
grown for high-value solid-wood products: a review. Northern Research Unit, For-
ests NSW Plantations Division (2006)
[2] McGavin, R.L., Davies, M.P., Macgregor-Skinner, J., Bailleres, H., Armstrong, M.,
Atyeo, W.J., Norton, J.: Utilisation Potential and Market Opportunities for Planta-
tion Hardwood Thinnings from Queensland and Northern New South Wales,
PN05.2022, Department of Primary Industries and Fisheries, Queensland Depart-
ment (2006)
[3] Qld DPI, Managing Hardwood Plantations,
http://www2.dpi.qld.gov.au/hardwwodsqld/1815.html(accessed
on March 22, 2010)
[4] Corey, B.J.A.: Little Left to Lose: Deforestation and Forest Degradation in Australia
since European Colonization. Journal of Plant Ecology (2011)
[5] Bureau of Rural Science, Australian Forest at a Glance, Department of Agriculture,
Fisheries and Forestry, Australian Government (2010)
[6] Underhill, B., Watts, H.: An overview of the commercial growing management and
processing of forest products in Queensland (2004)
[7] A.G. Department of Primary Industries and Fisheries, A.G. Department of Primary
Industries and Fisheries, Plantations and Farm Forestry,
http://www.daff.gov.au/forestry/australias-forests/
plantation-farm-forestry (accessed on May 24, 2013)
[8] Bureau of Rural Science, Australian Plantation 2010 Inventory Update, Department
of Agriculture, Fisheries and Forestry, Australian Government (2010)
[9] Australian Government, Department of Agriculture Fisheries and Forestry, Regional
Forest Agreements, http://www.daff.gov.au/forestry/policies/
rfa (accessed on June 01, 2013)
[10] White, A.: Gunns growers challenge PPB in: The Australian, Australia (2012)
[11] Bailleres, H., Gerard, J., Fournier, M., Thibaut, B.: Wood quality of eucalyptus from
plantations. 1. Spatio temporal variations and influence factors of three basic proper-
ties. IAWA Journal 16, 910 (1995)
[12] Bailleres, H., Gerard, J., Fournier, M., Thibaut, B.: Wood quality of eucalyptus from
plantations. 2. End splitting and sawing distorsion. IAWA Journal 16, 10 (1995)
[13] McCarthy, K.J., LCookson, L.J., Mollah, M., Norton, J., Hann, J.: The Suitability of
Plantation Thinnings as Vineyard Posts. Forest and Wood Products Research and
Development Corporation (2005)
[14] Nolan, G., Washusen, R., Jennings, S., Greaves, B., Parson, M.: Eucalypt Planta-
tions for Solid Wood Products in Australia. A Review, PN04.3002, Forest and
Wood Products Research and Development Corporation, Australian Government
(2005)
[15] Ranta-Maunus, A.: Round small-diameter timber for construction, Final report of
project FAIR CT 95-0091, VTT Technical Research Centre of Finland, Finland
(1999)
[16] Dickson, M., Hopewell, G., MacKenzie, C., Bailleres, H., Switala, J., Thomas, C.: A
market assessment and evaluation of structural roundwood products from hardwood
pulp plantations, Project PRA154-0910 Forest & Wood Products Australia, Mel-
bourne, Australia (2011)
[17] Washusen, R.: Processing plantation-grown Eucalyptus globulus and E. nitens for
solid-wood products-is it viable? Tech. Report 209, CRC for Forestry (2011)
[18] Carrick, J., Mathieu, K.: Durability of Laminated Veneer Lumber made from Black-
butt. In: Proceedings of the 10DBMC International Conference on Durability of
Building Materials and Components, France (2005)
[19] Bottle, K.R.: Wood in Australia, Types, properties and uses. McGraw-Hill, Sydney
(1983)
[20] Ozarska, B.: A review of the utilisation of hardwood for LVL (CSIRO). Wood Sci-
ence and Technology 33 (1997)
[21] Gilbert, B.P., Underhill, I.D., Bailleres, H., McGavin, R.L.: Structural Veneer Based
Composite products from hardwood thinning Part II: Testing of hollow utility
poles. In: Proceedings of the RILEM Conference Materials and Joints in Timber
Structures - Recent Advancement of Technology, Stuttgart, Germany (2013)
Glue Laminated Bamboo (GluBam)
for Structural Applications
Y. Xiao1,2,*, B. Shan1, R.Z. Yang1, Z. Li1, and J. Chen3
1
Key Laboratory of Building Safety and Energy Efficiency of China Ministry of Education,
Hunan University, Changsha 410082, China
2
Department of Civil and Environmental Engineering, University of South California,
Los Angeles CA 90089, USA
3
Advanced Bamboo and Timber Technologies, Ltd. Changsha, China
yanxiao@hnu.edu.cn, yanxiao@usc.edu
Abstract. In todays trend of sustainable development, there is a renewed interest to

use bamboo for modern building and bridge structures. However, traditional use of
raw bamboo culms is not the only and nor the most effective application. The
authors developed a laminated bamboo or glubam for general structural
applications. This paper describes the manufacturing process of glubam,
investigates and analyzes its energy consumption and carbon dioxide emission, and
provides main mechanical properties through material testing. Analysis results and
comparison with other comparable construction materials show the eco-friendly
performance of glubam. The mechanical properties of glubam are promising for
general use in construction. Research on connections using steel bolts shows the
good connectivity of glubam components similar to timber structures. The paper
also summarizes the authors extensive experimental tests on various glubam
components, such as full-scale girders under static and fatigue loads, wall panels
under monotonic and cyclic loads, full-scale room models under simulated
earthquake load and fire, etc. Several recent practical design and construction of
residential and industrial buildings are also briefly introduced.
Keywords: glubam, production, energy consumption, carbon dioxide emission,

mechanical properties, components, design and construction.
1 Introduction
Bamboo has been indispensable in human life and its history being used as building
and bridge structural materials can be traced back as long time as wood. Its natural
properties are quite comparable to those of wood. However, the original
geometrical shape of bamboo culms makes it difficult to be used in modern
*
DOI: 10.1007/978-94-007-7811-5_54, RILEM 2014
590 Y. Xiao et al.
construction. Based on abundant material and existing production technology in

China, a new type of laminated bamboo that can be used as structural elements are
invented by the authors with a trademark of GluBam [1-2].
This paper presents a summary of the authors recent work on research and
development of glue laminated bamboo or glubam for general usage in building and
bridge structures.
2 Production of Glubam
The production of Glubam has about five steps consisting of raw bamboo selection;
splitting bamboo strips; netting bamboo curtains or mats; gluing and hot-pressing;
and post-processing, as shown in Fig.1. The process of producing glubam sheets is
similar to that for plywood production [3]. The typical glubam sheets are 30 mm
(a) Glubam sheets
(b) Post-process of structural components

Fig. 1 Production of Glubam
Glue Laminated Bamboo (GluBam) for Structural Applications 591
(+/- 2mm), 1220 mm wide and 2440 mm long. These laminated bamboo sheets after
cutting, gluing, stacking and compressing, finger-jointing can finally form different
structural elements. Figure 1(b) exhibits the production of some glubam columns
with a length of 6 m. Detailed discussions about the glubam production can be
found in [4].
3 Mechanical Properties and Environmental Impact of

Glubam
3.1 Mechanical Properties

The typical glubam structural components use the sheets with bamboo strips
arranged with 80% in the longitudinal direction and 20% in the transverse direction,
referred to as 4:1 sheet. There are usually 15 or less layers of bamboo curtains, and
each is with about 2 mm thickness.
Due to the fact that glubam is a new invention and no testing standard has been
established, the standard for timber structure is referenced in the study of
mechanical properties of glubam. In Chinese timber structure design standard,
design values of mechanical properties of timber are from small clear specimen
tests. Being short of timber material, not much plywood is used in China as
structural material except some thin wood boards for decoration. In view of the
condition of material production and design method, it is reasonable to use small
clear specimen test method to obtain mechanical properties for glubam. A large
number of specimens were tested and the results are summarized in Table 1.
Table 1 Main mechanical properties of Glubam
In-plane Elastic Elastic In-plane

modulus Bending
Materials
tensile
modulus E y
Gxy G yx compressive
strength
Density
strength Ex (MPa) (MPa)
strength
(MPa)
(kg/m3)
(MPa) (MPa) (MPa)
(MPa)
Glubam 82 10400 2600 4.6 7.2 51 99 800~900
3.2 Environmental Impact

The influence on environment of certain material is a quite important aspect in
todays trend towards a sustainable construction industry. It is well known that
timber has less carbon and environmental impacts than other industrialized
materials including concrete, steel, aluminum, etc. The main reason of timber being
greener owes to its less processing energy consumption and more carbon dioxide
storage [4]. An investigation on environmental impact of glubam was conducted at
the authors production base located in Yanling County of Hunan Province, where
moso bamboo abounds.
592 Y. Xiao et al.
3.2.1 General Energy Consumption
According to the mills actual production condition, some data were gathered to
calculate the energy consumption of per cubic meter of glubam sheets. Energy
consumptions of glubam production are found from transportation of raw materials,
electricity usage, phenol formaldehyde resin, hydraulic oil used in hot compressing
machine, heating and cooling water, and fuel in boiler, etc. The energy consumption
nested in the production equipment and the factory facilities is also counted. As the
result, the energy total consumption of glubam sheets is estimated at 2.67GJ/ m3.
A comparison is made in Fig.2 to show the energy consumptions of glubam and
other construction materials including timber, plywood, cement, aluminum and
structural steel. Data for cement are from Hammond and Jones [5], the others
including timber, plywood, aluminum and structural steel are from Buchanan and
Honey [6]. As shown in Fig.2, aluminum and steel are big consumers of energy.
Glubam consumes lower energy amount than cement by about 75%, however
higher than timber.
Fig. 2 Production energy consumption of glubam and other conventional materials
3.2.2 Carbon Emission
Low carbon emission of bamboo products has been widely recognized, making it
very attractive in todays move towards a sustainable society. Research shows that
the carbon dioxide stored in the raw bamboo used in 1m3 glubam is calculated as
2.166t. Different factors including transportation, electricity, lubricating oil, phenol
formaldehyde resin, carbon emission per capita, production equipment (including
post process) and plant inventories that contributing to the carbon emission during
the manufacturing process of glubam are considered in the calculation. Based on
related coefficients, carbon dioxide emission of glubam sheets is calculated as
-261kg/ m. Apparently, glubam is a carbon negative material.
Histogram shown in Fig.3 compares Glubam with other construction materials

including timber, plywood, cement, aluminum and steel in carbon emission. Data
for cement are from reference [5], the others including timber, plywood, Aluminum
and structural steel are from reference [6]. Cement, Aluminum and steel discharge
much more carbon dioxide than the other materials. In comparison, glubam is
carbon emission negative, and outperforms timber and plywood.
Fig. 3 Carbon dioxide emission of Glubam and other conventional materials
4 Experimental Studies on Bolted Connections of Glubam
4.1 Specimens
In this paper, glubam bolted connection with different end distance and side
distance are considered. One end of the specimen is holding end and the other is for
testing, as shown in Fig. 4. The bolt diameter is d=12mm and the hole diameter is
D=14mm. The end distance e and edge distance b/2 of the tested end is
designed.
Fig. 4 Dimension of specimens

594 Y. Xiao et al.
The experiment involved two major groups of specimens, in each of which there
were nine groups of specimens with different end distances and edge distances.
Table 2 shows the groups of specimens, where b is the width of glubam board; e is
end distance; and d is diameter of bolt hole. The thickness of all glubam specimens
is measured at 28mm. There are nine types of specimen configurations, and each
has ten specimens both in longitudinal and transverse directions. There were 90
specimens in total in this testing program. Before testing, moisture content (MC) of
the specimens was measured by handheld wood moisture meter and the average
MC was about 10.0%.
Table 2 Dimension of specimens
V1/H1 groups V2/H2 groups V3/H3 groups

e=2d e=2d e=2d
b=4d e=3d b=6d e=3d b=8d e=3d
e=4d e=4d e=4d
Fig. 5 Fixture of specimen Fig. 6 Test system
4.2 Test Setup

Testing procedures outlined in ASTM D 565295 (R2000) were followed. Figure 5
shows the fixture designed specifically for specimens. In Fig. 6, test system is
shown. Universal testing machine is used to control and apply force, as well as to
record data. Each specimen was subjected to a displacement-control loading with a
rate of 3mm/min until the end of the test, to make sure specimens reach maximum
load in not less than 5min and no more than 20min (ASTM 2000).
4.3 Failure Modes

Most design standards use the Johansens Yield Model. It is a mechanics-based
model to determine the resistance of bolt for various ductile failure modes. In
Johansen Yield Theory, it is hypothesized that wood or composite wood and bolt
will reach total ductility under dowel bearing stress and moment of bolt, as shown in
Fig. 7. This type of failure mode has been considered as the best condition, on
which the bolt, side board and middle board will arrive failure at the same time, so
all the material energy can be used.
Fig. 7 Ideal failure mode in Johansen Yield Theory
In the tests, there are two main types of failure modes, shearing out and net
tension, as shown in Fig. 8 and Fig. 9. Shearing out occurred usually in V group and
net tension usually in H group. For single bolted joint, the predominant failure
modes include bearing, shearing out, cleavage and net tension. The governing mode
and failure load of a joint depend on factors such as orientation of fibers in the
members, the joint geometry, and clamping force [7-8]. For bearing strength, it is
the ideal failure mode in Johansen Yield Theory and is most desirable. But for
specimens in the test, they were clamped by the same prefabricated fixture and no
tension was applied on the bolt, so the specimens tended to fail by glubam fracture.
Fig. 8 Failure mode of V group-shearing out Fig. 9 Failure mode of H group-net tension
596 Y. Xiao et al.
4.4 Analysis
The results of the tests show great consistency that was expected, while some
phenomenon was unexpected. As shown in Fig. 10, the ultimate strength increases
with the increase of side distance. And the same tendency can be found in H group
results. However, within V1, V2 and V3 group, relationship between the ultimate
strength and the increase of edge distance does not show any regularity, neither
does in H group.
Fig. 10 Ultimate strength V group specimens Fig. 11 Ultimate strength H group

specimens
Comparing Fig. 10 and Fig. 11, it can be found that V group has correspondingly
high strength than H group, which means loading parallel to fiber is beneficial to the
ultimate strength of specimen.
In Fig. 11, another phenomenon should be noted that although edge distance is
the first factor to influence the ultimate load, the end distance do obviously affect
the results especially for H2 and H3 groups. This may be explained from stress
distribution around the bolt hole. Many researchers studied the stress distribution
both numerically and experimentally. Echavarra and Salenikovich (1998)
published a paper about model for predicting brittle failures of bolted timber joints.
Their research shows that the stress magnitude around the pin hole decreases with
the increase of end distance, which means the large end distance benefits the
bearing strength of joints[9].
In ASTM D 5652-95 (Re-approved 2000), the connection yield load is
determined by fitting a straight line to the initial linear portion of the
load-deformation curve, offset this line by a deformation equal to 5 % of the bolt
diameter, and select the load at which the offset line intersects the load-deformation
curve. In those cases where the offset line does not intersect the load deformation
curve, the maximum load shall be used as the yield load. The initial stiffness is also
obtained from the method.
Table 3 summaries the test results. All the failures happened in main glubam
board. And according to the two types of failure modes in specimens, equation (1)
and (2) are obtained to count the failure strength of single bolt glubam joint.
F = 2eSt (1)
F = (2b d )Tt (2)
And the ultimate strength of specimens can be obtained from equation (3),
F = min ( 2eSt , (2b d )Tt ) (3)
In which, F is ultimate load, b is side distance, S is the shear strength of Glubam, T

is the tension strength, t is the thickness of Glubam board valued as 28mm, and
, is the coefficient obtained from regression analysis. Based on the equation,
predicted Maximum load is shown in Table 3.
Table 3 Experimental and Analytical Results
Yield Predicted Theoretical Initial

Maximum Failure
Specimen strength Max load ultimate stiffness
load [kN] mode*
[kN] [kN] strength [kN/mm]
V1-1 13.5 15.0 15.6 9.9 5.2 SO
V1 V1-2 21.0 20.2 23.3 14.8 5.0 SO
V1-3 23.8 26.3 31.1 19.8 6.1 SO
V2-1 12.0 13.9 15.6 9.9 5.4 SO
V2 V2-2 22.8 24.1 23.3 14.8 6.4 SO
V2-3 26.5 28.9 31.1 19.8 6.0 SO
V3-1 16.3 17.5 15.6 9.9 6.0 SO
V3 V3-2 25.1 25.9 23.3 14.8 7.1 SO
V3-3 23.2 26.9 31.1 19.8 6.2 SO
H1-1 10.6 12.3 8.62 16.2 3.7 NT
H1 H1-2 11.6 13.1 8.62 16.2 4.1 NT
H1-3 10.8 12.3 8.62 16.2 3.8 NT
H2-1 11.1 12.1 14.7 27.6 4.3 NT
H2 H2-2 14.3 15.1 14.7 27.6 4.4 NT
H2-3 17.8 18.6 14.7 27.6 4.2 NT
H3-1 11.8 12.6 20.8 39.0 4.2 NT
H3 H3-2 14.5 16.3 20.8 39.0 4.4 NT
H3-3 17.5 20.6 20.8 39.0 4.5 NT
Note: SO means shearing out, NT means net tension.
598 Y. Xiao et al.
5 Other Studies and Applications of Glubam
5.1 Static and Fatigue Tests of Full-Scale Girders

Static test of glubam girders (Fig.12) show their excellent bearing capacity [1]. The
use of FRP to reinforce the soffit of the girders can further enhance the load carrying
capacity of glubam girders.
Fig. 12 Fatigue test of full-scale glubam girder
The fatigue experiment reduces the bearing capacity of glubam girders

approximately by 10% due to the development of weaknesses in finger-joint
and gluing face. When the upper value of cyclic load does not surpass its design
value, there was no distinct reduction on the stiffness of specimens compared
with the static tests. It is clear that excellent flexibility of bamboo contributes to the
fine stability of glubam beams in aspect of dynamic response during the fatigue
loading.
5.2 Wall Panel under Monotonic and Cyclic Loads

Significant numbers of experimental tests were conducted to study the lateral
loading behavior of shear wall panels made with glubam sheets and glubam or
wood frame studs. Tests show that the light-weight wood shear walls with glubam
sheathing panel (as shown in Fig. 13) have good seismic performance as well as the
ability to meet the design requirements specified in most timber structure codes.
Besides, the processing of this type of shear wall could satisfy the requirements and
preconditions of industrial production, as well as easy installation.
Fig. 13 Test shear wall details
5.3 Full-Scale Room Models under Simulated Earthquake Load

and Fire
A full-scale lightweight frame glubam room model was tested on shaking table,
with peak input acceleration up to 0.5g. The result shows excellent seismic
resistance. Only some minor damage, such as pulling out or penetrating of nails
from or through sheathing board (see Fig.14), were observed when seismic
acceleration was 0.5g.
The full-scale room model used in the shake-table test was reconstructed with
finishing of the inside surfaces by gypsum boards and voids filled with rock wools.
A wood crib set at the center of the room was then ignited and allowed to burn for an
hour to test the integrity of the lightweight frame glubam building [11]. After one
hour of exposure to fire inside the room, the structures of the walls and the ceilings
were essentially intact. Like timber structures, the existence of carburization layer
under fire can delay further penetrating of fire into the structure.
600 Y. Xiao et al.
(a) penetrating (b) pulling out
Fig. 14 Damage detail of full-scale glubam room under seismic load
6 Applications of Glubam Structures
The authors also made efforts to build modern bamboo buildings and bridges using
the newly developed laminated bamboo structural material, glubam. The structural
design is essentially based on the timber design standards and detailing using the
material property data obtained from this study. So far the authors have designed
and built more than 20 buildings with total building area exceeding 4,000 sq.m. The
practice can be categorized into three categories: modular temporary and permanent
buildings (Fig. 15(a)); single or multi-story residential buildings (Fig. 15(b)); and
heavy space frame buildings (Fig. 15(c)). The temporary glubam buildings were
successfully deployed to the area devastated by the May 12, 2008 Sichuan
earthquake. In the heavy space frame building as shown in Fig. 15(c), the largest
girder had a length of 16.5 m, with a cantilever length of 7 m, and was made of
glubam with a section of 800 mm deep and 120 mm wide.
(a) Mobile building for (c) A park house built with

(b) Residential building
disaster relief glubam space frames
Fig. 15 Applications
7 Conclusions
This paper introduced the current situation of glue laminated bamboo or glubam
research and applications. As newly developed construction material, glubam has
been proven to be applicable in buildings and bridges. The rich resources of
bamboo in many parts of the world, well-established production, good mechanical
properties and low impact on environment make this new material quite promising
in todays trend towards a sustainable construction industry.
Acknowledgements. The research described in this paper was supported by National Natural
Science Foundation of China through a National Key Project (No.50938002) and
Scholarship Award for Excellent Doctoral Student granted by Ministry of Education. The
writers would like to thank Advanced Bamboo and Timber Technologies (ABTT),
Changsha, for their contributions in the testing and data collection.
References
[1] Xiao, Y., Zhou, Q., Shan, B.: Design and construction of modern bamboo bridges.
ASCE Journal of Bridge Engineering 15(Compendex), 533541 (2010)
[2] Xiao, Y., Shan, B., Chen, G., Zhou, Q., She, L.Y.: Development of a new type of
glulamGlubam. In: Xe, Y., et al. (eds.) Modern Bamboo Structures. CRC,
Changsha (2008)
[3] Laboratory, F.P.: Wood handbookWood as an engineering material. In: USDA FS
(ed.). Madison, Wis. (1999)
[4] Xiao, Y., Yang, R.Z., Shan, B.: Production, Environmental Impact and Mechanical
Properties of Glubam. Journal of Construction and Building Materials (accepted,
2013)
[5] Hammond, G., Jones, C.: Inventory of carbon & energy (ICE). Version 16a:
University of Bath, UK (2008)
[6] Buchanan, A.H., Honey, B.G.: Energy and carbon dioxide implications of building
construction. Energy and Buildings 20(3), 205217 (1994)
[7] Rowlands, R.E., Rahman, M.U., Wilkinson, T.L., Chiang, Y.I.: Single- and
multiple-bolted joints in orthotropic materials. Composites 13(3), 273279 (1982)
[8] Patton-Mallory, M., Pellicane, P.J., Smith, F.W.: Qualitative assessment of failure in
bolted connections: Maximum stress criterion. Journal of Testing and Evaluation
26(Compendex), 489496 (1998)
[9] Echavarra, C., Salenikovich, A.: Analytical model for predicting brittle failures of
bolted timber joints. Materials and Structures 42(7), 867875 (2009)
[10] ASTM. D5652 - 95(2013) Standard Test Methods for Bolted Connections in Wood
and Wood-Based Products
[11] Xiao, Y., Ma, J.: Fire simulation test and analysis of laminated bamboo frame
building. Journal of Construction and Building Materials 34, 257266 (2012)
Glulam Composed of Glued Laminated Veneer
Lumber Made of Beech Wood: Superior
Performance in Compression Loading
Gerhard Dill-Langer and Simon Aicher
Material Testing Institute, University of Stuttgart, Department of Timber Constructions,

Pfaffenwaldring 4b, 70569 Stuttgart Germany
Gerhard.Dill-Langer@mpa.uni-stuttgart.de
Abstract. Compared to most softwoods, the hardwood species beech (Fagus syl-
vatica) exhibits higher strength and stiffness properties, but for structural use
also a number of drawbacks. The drawbacks, however, can be overcome partly by
processing beech wood to laminated veneer lumber (LVL). In order to utilise
beech LVL not only for plate-like structures, but also for beams or columns with
deliberate cross-sectional dimensions, the beech LVL can be further processed to
glued laminated beams (glulam) made of LVL laminations. The paper reports on
experimental investigations of innovative high end structural beech glulam and
demonstrates the system effect on the load capacity in compression loading paral-
lel to the grain.
Keywords: hardwood, beech, glued laminated veneer lumber, LVL, glulam, com-
pression strength, load-sharing, system effect.
1 Introduction
Due to the current fundamentally ecological conversion of forestry in Middle

Europe, the ratio of available quantities of hardwoods vs. softwoods will change
considerably. Most notably, the disposable volume of beech-wood will increase in
the next years and decades, whereby the problem of utilisation, especially of the
lower grades, has not yet been solved. The widespread direct thermal utilisation is
not sustainable neither from an economical nor from an ecological point of view.
Thus, considerable attempts have been undertaken by industry and research insti-
tutions to introduce structural beech wood and glued-laminated timber into the
construction sector [17]. In 2011, a German technical building approval [8] was
issued by DIBt for glulam made either entirely of beech wood laminations or as a
hybrid of beech and softwood laminations. However, structural products based on
solid timber have some serious drawbacks, which has limited their utilisation so
far to a narrow niche market: The pronounced radial-tangential anisotropy of
shrinkage-swelling properties and the very low durability cause problems in all
DOI: 10.1007/978-94-007-7811-5_55, RILEM 2014
604 G. Dill-Langer and S. Aicher
conditions except if used in service class 1 conditions. Moreover, the potential

high strength capacity is only valid for high visual or machine strength grades,
which are not currently commercially available to a major extent. A further reason
for the slow gain of market shares results from the significantly different machin-
ing and gluing properties of beech hardwood leading to problems in its integration
into production processes that glulam companies have laid out for handling soft-
woods.
By processing beech round wood into LVL, some of the mentioned drawbacks
can be overcome or minimised: in LVL the year ring structure is homogenized and
thus there are less problems with climate induced stresses and cracking. The lami-
nation effect of veneer layers enables the utilisation of the high strength capacity
of beech wood by additionally using some fraction of lower grade material. A
comprehensive compilation of the major strength and stiffness properties of LVL
made of European beech considered here is given in [9].
Product inherent, the panel product LVL is mainly apt for plate like structures.
When used for beams or columns, the cross-sectional dimensions are limited and
can only be enhanced by further processing of the panel product to glued lami-
nated beam build-ups. In the following it will be shown, based on an extensive
experimental campaign, that the assembly of LVL laminations to glulam cross-
sections not only overcomes the dimensional limits, but also enhances the load
bearing capacity, exemplarily shown for the property compression strength paral-
lel to fibre direction.
2 Material
The structural material investigated consists of several lamellas bonded together

into a glulam cross-section. The primary material of the lamellas is LVL made
of European beech wood (Fagus sylvatica L.). The veneers are 3.7 mm thick and
are glued parallel to fiber direction by a phenolic adhesive. The thickness of the
panels, where the lamellas are cut from is in the range of 40-45 mm and the
lamella thickness (after planing) is in the range of 37 to 42 mm. The raw density
of the veneers is about 680 kg/m3 (mean value). The veneers are jointed in the
longitudinal direction by glued scarf joints. The lamellas are cut from the LVL-
panels parallel to the fibre direction and bonded together flatwise by means of a
phenolic resorcinol formaldehyde adhesive. Prior to bonding the surfaces have to
be planed or sanded.
In the presented study, the lamellas are not finger jointed, i.e. the lamellas have
to be produced in the same length as the final glulam product. As the basic raw
material of the lamellas is a panel product, there are no fundamental limits for
cross-sectional dimensions. In case of the width direction, this is one of the major
differences between glulam from solid wood and glulam from LVL.
In the presented experimental study, which is part of an approval process in
Germany, the height of the glulam beams was limited to 600 mm and the maximal
Glulam Composed of Glued Laminated Veneer Lumber Made of Beech Wood 605
width was 300 mm. Without any jointing in the length direction, however, the
beam length is restricted to the maximum panel length. In the first step of the
industrial production process the beam length will be limited to 18 m. Figure 1
shows the view of the three investigated glulam cross-sections.
Fig. 1 Glulam cross-sections made of beech LVL
3.1 Methods and Test Set-Ups

The investigations presented in this paper are a part of a extensive test program
consisting of tension, compression, bending and shear tests. Additionally the bond
line strength and the durability of bond line strength have been examined.
The tests discussed hereinafter are focused on the compression strength parallel
to the fibre direction. The experiments have mainly been performed according to
EN 408 [10] with specimens of three different cross-sections, composed of 3, 9
and 16 lamellas. The ratio of smallest cross-sectional width vs. length was
throughout chosen as 6, conforming to EN 408, which prevents global buckling
effects. The evaluation of the test data also comprises one data set of compression
tests parallel to fibre direction with specimens from the original LVL panel
product which will be referred to here as single lamella. The tests have been
performed at Technical University of Munich [11] and reported confidentially1.
Table 1 gives an overview of the performed test series.
Table 1 Overview of dimensions and numbers of different compression test series
number of
Configuration width height length specimens
single lamella 100 50 300 36
3 lamellas 50 114 300 15
9-lamellas 300 330 1800 15
16 lamellas 150 600 900 15
Due to the wide span of investigated cross-sectional dimensions with a result-

ing wide range of ultimate loads (between about 250 kN and 9 MN) the tests have
been performed in two different servohydraulic test machines. Figure 2 gives a
view of a 3-lamella specimen mounted in the smaller test machine (ultimate load
1,6 MN); Fig. 3 shows the much larger 9-layer specimen mounted in the 20 MN
heavy duty test machine.
Fig. 2 Test set-up for specimen with 3 Fig. 3 Test set-up for specimen with nine
laminations laminations
1
Use of data in this contribution commissioned by Pollmeier Furnierwerkstoffe GmbH &
Co. KG, Creuzburg.
In order to achieve uniaxial compression loading without any bending moments

the compression load was transferred by a spherical swivel. Whereas the swivel
acc. to EN 408 could be fixed after some preloading for the 3-lamella specimen,
the rotation of the swivel of the test machine used for the larger specimens could
not be limited throughout the whole compression test. However, no global stabil-
ity problems were encountered in these tests. During the compression tests the
load and the deformation parallel to fibre direction were recorded continuously.
3.2 Experimental Results

The failure behaviour of all specimens independent of size and number of lamel-
las could be described as mainly quasi-ductile. The load-deformation behaviour
showed strong non-linearity in advance of the ultimate load and considerable slow
decrease of the load capacity after ultimate load.
Figure 4 exemplarily gives the load-deformation curves for the 9-lamella-
specimens. The non-linearity at the beginning of loading (up to about 1 MN) is not
bound to any material behaviour, but can be explained by adjustments of the lower
part of the testing machine. Between 1 MN and roughly 6 MN, a fairly linear be-
haviour is observed (which of course could also be extrapolated to the lower part
between 0 and 1 MN). Between 6 MN and the ultimate load (around 8 MN),
marked non-linearity occurs. The descending branch of the load-deformation
curves exhibit to differing degrees some deformation reserves in a range of
about 3 to 10 mm before reaching a 50%-level of ultimate load.
7
Compression load Fc [MN]
0
0 5 10 15 20 25 30 35 40
Deformation [mm]
Fig. 4 Load-deformation curves of compression tests with 9-lamella LVL-glulam

specimens
The failure modes of all specimens can be described phenomenologically by

three features, which in many cases occurred in parallel in the same specimen:
- local bucking of lamellas or parts of lamellas

- (local) splitting
- formation of kink-bands with an angle of about 60 to the fibre direction
In Figures 5 to 7 the observed failure modes are illustrated by photographic

views of failed specimens whereby the details of the failure modes of all three
glulam configurations are highlighted.
Figure 5 with the view of a failed 3-lamella specimen shows a typical example
for the formation of kink bands near mid-height of the specimen in combination
with some but not very pronounced local buckling. Figure 6 gives the view of
a failed 9-lamella specimen with distributed local buckling and some minor local
splitting in the vicinity of the lower support. Figure 7 with the view of a
16-lamella specimen shows an example of pronounced local splitting with subse-
quent local buckling of single lamellas.
Most of the visible damage, especially local (macroscopic) buckling and splitting,
occurred beyond ultimate load in the descending branch of the load-deformation
curves. The observation of the damage behaviour suggests that the global quasi-
ductility of the load-deformation behaviour is caused mainly by many small local
(microscopic) damages (being brittle in itself such as buckling, splitting etc.) with
load redistribution to the parallel oriented veneers / lamellas.
a) b)
Fig. 5 View of a failed 3-lamella specimen with a kink-band

a) b)
Fig. 6 View of a failed 9-lamella specimen with the failure modes of local buckling and
splitting next to the lower support
a) b)
Fig. 7 View of a failed 16-lamella specimen: major splitting and subsequent local buckling
The compression strength results and additional values such as raw density and
moisture content are given in Table 2 for all four test series. The characteristic
values were evaluated according to EN 14358 [12]. In order to facilitate the com-
parison of the results of the different test series, the empiric compression strength
results are shown in Fig. 8 as cumulative distributions together with the respective
lognormal distribution functions.
Table 2 Test results fc,0 of compression tests parallel to fibre direction
Test series
Statistical evaluation of
compression strength single
3 lamellas 9 lamellas 16 lamellas
lamella
Number of specimen 36 15 15 15
mean value 59.8 66.2 78.3 77.4
maximum 69 67.7 81.1 82.7
minimum 54.8 64.3 73.7 73.5
standard deviation 3.5 1.1 2.1 2.5
coefficient of variation 0.0586 0.0161 0.027 0.033
5% quantile (lognormal) 54.3 64.4 74.9 73.4
charact. value acc. to
53.7 59.9 71.0 70.0
EN 14358
0,9
0,8
0,7 16 lamellas
cumulative frequency [--]
150 mm 600 mm
0,6 9 lamellas
single lamella 300mm 330 mm
0,5 50 mm 100 mm
0,4
3 lamellas
0,3
50 mm 114 mm
0,2
0,1
0
50 55 60 65 70 75 80 85 90
compression strength parallel to fiber fc,0 [N/mm2]
Fig. 8 Distribution of compression strength parallel to fibre for the basic material (single
lamella) and three glulam configurations with 3, 9 and 16 lamellas; given are the test data
and the respective lognormal distribution functions
3.3 Evaluation of the System Factor

The test results of compression strength parallel to the fibre reveal a clear
tendency of increasing strength and reduced scatter with increasing number of
LVL-lamellas starting from one lamella to several ones glued together in the glu-
lam cross-section. Qualitatively the effect can be explained by load redistribution
in the parallel system of simultaneously loaded lamellas, which is high in case of
glued cross-sections. However, due to limited load distribution range there is a
threshold value of number of lamellas, beyond which the strength is no longer
increasing. This well-known behaviour of parallel systems has been implemented
into Eurocode 5 [13] for design of flatwise bending of glulam or mechanically
connected floor elements. The relative increase of characteristic system strength
related to the original characteristic strength of the single members is defined as
the system factor ksys.
In order to evaluate the system effect in compression parallel to the grain of
glued laminated beech LVL, the test data have been normalised with respect to the
5%-quantile value of the basic LVL material (single lamella). In Figure 9 the
normalised compression test data have been plotted vs. the number of lamellas.
Both, the test data (open boxes), and the 5%-quantile values of the respective log-
normal distributions (filled boxes) are given. The system effect as implemented in
EC5 is depicted as a green line. It can be seen from the figure that the empirically
determined system effect of compression strength parallel to the fibre direction
can qualitatively be described by the ksys-factor acc. to EC5 with a threshold num-
ber of lamellas of 8. However, quantitatively, the effect is distinctly larger with a
maximum value of at least ksys = 1.3 instead of 1.2.
1,5
normalised
test data
1,4
system factor ksys
0.05-quantile
1,3
ksys, compr. acc. to test results
1,2
ksys acc. to EC5
1,1
1,0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
number of lamellas
Fig. 9 System factor dependent on number of parallel lamellas

4 Conclusions
The secondary processing of beech LVL into the innovative structural material
glued-laminated beech LVL yields significant further enhancement of the
already high strength values of beech LVL itself. This has been revealed in the
paper for the example of compression strength parallel to the fibre. The character-
istic compression strength values reach a level of more than 70 N/mm2, which is
in the range of high-strength concrete. In order to highlight the high strength level
of glulam made of beech LVL laminations Tab. 3 gives an overview of the charac-
teristic values of compression strength parallel to grain of different structural
wood based products such as solid timber, glued-laminated timber and LVL made
of soft- and hardwoods.
Table 3 Compilation of characteristic values fc,0,k of compression strength parallel to the

grain
Softwoods Beech
2
C24 21 D30 23
Solid timber
C50 29 D70 34
3,4
GL 24 24 GL 28
Glulam 25
GL 36 36 GL 48
5
Kerto 38 --
6
LVL Ultralam 48 --
7
Pollmeier -- 53
5
Kerto 38 --
Glulam LVL
Pollmeier7 -- 70
The compilation in Tab. 4 underlines that beech LVL glulam is the strongest
wood based structural component made from a European wood species.
2
EN 338 [14].
3
Glulam made of softwoods: EN 14080:2013 [15].
4
Glulam made of hard woods: Technical Approval [8].
5
See [15].
6
See [16].
7
See text.
References
[1] Glos, P., Lederer, B.: Sortierung von Buchen- und Eichenschnittholz nach der
Tragfhigkeit und Bestimmung der zugehrigen Festigkeits- und Steifigkeitswerte.
Research Report No. 98508, Institut fr Holzforschung, Technical University Mu-
nich (2000)
[2] Aicher, S., Hfflin, L., Behrens, W.: A study on tension strength of finger joints in
beech wood laminations. Otto-Graf-J. 12, 169186 (2001)
[3] Glos, P., Denzler, J.K., Linsenmann, P.: Strength and stiffness behaviour of beech
laminations for high strength glulam. In: Proceeding Meeting 37 CIB Working
Commission W18-Timber Structures, paper CIB-W18/37-6-3, Edingburgh (2004)
[4] Freese, M., Bla, H.J.: Beech glulam strength classes. In: Proceedings Meeting 38,
CIB Working Commission W18-Timber Structures, paper CIB-W18/38-6-2, Karls-
ruhe (2005)
[5] Bla, H.J., Denzler, J.K., Freese, M., Glos, P., Linsenmann, P.: Biegefestigkeit von
Brettschichtholz aus Buche. Karlsruher Berichte zum Ingenieurholzbau 1,
Universittsverlag Karlsruhe (2005)
[6] Freese, M.: Die Biegefestigkeit von Brettschichtholz aus Buche Experimentelle
und numerische Untersuchungen zum Laminierungseffekt. PhD thesis, Karlsruher
Berichte zum Ingenieurholzbau 5, Universittsverlag Karlsruhe (2006)
[7] Freese, M.: Buchen-BS-Holz und Buche-Hybridtrger. In: Proceedings of 2.
Stuttgarter Holzbau-Symposium - Neueste Entwicklungen bei Geklebten
Holzbauteilen, pp. 137144. MPA Universitt Stuttgart (2012)
[8] DiBt: Allgemeine bauaufsichtliche Zulassung (General building approval) Z-9.1-679
on BS-Holz aus Buche und BS-Holz Buche-Hybridtrger (glulam made of beech
and beech hybrid glulam) 1st issued (October 2009): Approval holder:
Studiengemeinschaft Holzleimbau e.V., Wuppertal, Germany (2011)
[9] Knorz, M., van Kuilen, J.W.: Development of a high-capacity engineered wood
product - LVL made of European beech (fagus sylvatica L.) In: Quennville, P. (ed.)
Proceedings (Final papers, electronic version) World Conf. Timber Eng., pp. 487
495, Auckland (2012)
[10] EN 408: Timber structures Structural timber and glued laminated timber Deter-
mination of some physical and mechanical properties (2010)
[11] Test Report Nr. 10511 issued by Technical University Munich, Holzforschung
Mnchen, commissioned by: Pollmeier Furnierwerkstoffe GmbH & Co. KG,
Creuzburg European Standart (2012)
[12] EN 14358: Timber structures Calculation of characteristic 5-percentile values and
acceptance criteria for a sample (2006)
[13] Eurocode 5: Design of timber structures Part 1-1: General Common rules and
rules for buildings (2010)
[14] EN 338: Structural timber - Strength classes (2009)
[15] EN 14080: Timber structures glued laminated timber and glued solid timber re-
quirements (2013)
[16] DiBt: Allgemeine bauaufsichtliche Zulassung (General building ap-proval) Z-9.1-
100 on Furnierschichtholz Kerto S und Kerto Q, (laminated veneer lumber
Kerto S and Kerto Q), Approval holder: Finnforest Oyi, Finland (2011)
[17] DiBt: Allgemeine bauaufsichtliche Zulassung (General building ap-proval) Z-9.1-
811 on Furnierschichtholz Ultralam R, Ultralam RS und Ultralam X, (lami-
nated veneer lumber Ultralam R, Ultralam RS and Ultralam X). Approval
holder: MLT Ltd., St. Petersburg, Russia (2010)
Structural Performance of Accoya Wood
under Service Class 3 Conditions
Julian Marcroft1, Ferry Bongers2, Fernando Perez Perez3, John Alexander4,

and Ian Harrison5
1
Marcroft Timber Consultancy Ltd, Suite 5, 4 Lenten Street, Alton, Hants,
GU34 1HG, United Kingdom
julian@marcrofttimberconsultancy.co.uk
2
Accsys Technologies, PO Box 2147, NL-6802 CC Arnhem, The Netherlands
ferry.bongers@accsysplc.com
3
Simpson Strong-Tie, Winchester Road, Cardinal Point, Tamworth, Staffordshire,
B78 3HG, United Kingdom
FPerezPerez@Strongtie.eu
4
Accsys Technologies, Sheet Street, Windsor, SL4 1BE, United Kingdom
john.alexander@accsysplc.com
5
Simpson Strong-Tie, Winchester Road, Cardinal Point, Tamworth, Staffordshire,
B78 3HG, United Kingdom
IHarrison@Strongtie.eu
Abstract. The acetylation of wood gives enhanced resistance against fungal decay
and improved dimensional stability, benefits which make Accoya wood
particularly suitable for use in external applications. With increasing interest in
using acetylated wood in structural applications, Accsys Technologies are
continually undertaking studies which add to the database of structural properties
of Accoya wood. This paper describes one such study investigating the structural
performance of Accoya wood under service class 3 conditions and in particular
the relative structural performance of Accoya wood between service classes 1
and 3 and how Accoya compares with solid timber in this regard.
Keywords: acetylated wood, mechanical properties, Eurocode 5, structural

design, service class 3.
1 Introduction
Since 2007 Accsys Technologies has been commercially producing Accoya

wood that is based on acetylation of Radiata pine (Pinus radiata D. Don). The
acetylation of wood gives enhanced resistance against fungal decay and improved
dimensional stability, benefits which make Accoya wood particularly suitable for
DOI: 10.1007/978-94-007-7811-5_56, RILEM 2014
616 J. Marcroft et al.
use in external applications (Alexander 2007, Bongers et al. 2009, Kattenbroek

2005).. To-date Accoya wood has predominantly been used in external non-
structural applications such as joinery or cladding, but, encouraged by the success
of the two heavy load-bearing traffic bridges constructed using Accoya wood in
Sneek in the Netherlands (Tjeerdsma and Bongers 2009, Jorissen and Lning
2010), there is increasing interest in using acetylated wood in structural
applications.
With the long-term objective of taking the product forward through to an
accepted structural approval, Accsys Technologies are continually undertaking
studies which add to the database of structural properties of Accoya wood.
Recently a Design Guide for Accoya structural wood was published that
demonstrates Accoya structural Radiata pine and Accoya structural SYP to be
assumed as C24 as given in EN 338. The research behind this manual is shown in
Bongers et al. 2013. It should be noted that this structural Accoya product is
based on machine grading and a separate product compared to regular Accoya
wood that is used for non-load-bearing applications. For simplicity in this paper is
referred to Accoya Radiata pine and Accoya SYP, but the samples are based on
the Accoya structural products.
In this paper additionally studies are summarised focused on investigation of
the structural performance of Accoya wood under service class 3 conditions and
in particular the relative structural performance of Accoya wood between service
classes 1 and 3.
As is appropriate for most wood-based materials, the various test EN Standards
stipulate testing specimens conditioned under service class 1 conditions and most
of the structural testing on Accoya wood has been carried out on pieces
conditioned in accordance with these EN Standards. However as the great
majority of Accoyas structural applications are external, for which (in the UK at
least) it is customary to design as being in service class 3, this makes the relative
structural performance between service classes 1 and 3 much more significant for
Accoya wood than it is for solid timber.
The primary purpose of this paper is to determine, for a range of structural
properties, the difference in performance of Accoya wood between service class
1 and service class 3 conditions. However before describing the associated test
investigations, consideration is briefly given to the design rules given for service
class 3 conditions in EN 1995-1-1.
2 Design Rules for Service Class 3 Conditions
2.1 Characterisation of Service Classes

The characterisation of service classes 1-3 in EN 1995-1-1 is as follows:
Service class 1 is characterised by a moisture content in the materials
corresponding to a temperature of 20C and the relative humidity of the
surrounding air only exceeding 65% for a few weeks per year (average
moisture content in most softwoods will not exceed 12%).
Structural Performance of Accoya Wood under Service Class 3 Conditions 617
Service class 2 is characterised by a moisture content in the materials

corresponding to a temperature of 20C and the relative humidity of the
surrounding air only exceeding 85% for a few weeks per year (average
moisture content in most softwoods will not exceed 20%).
Service class 3 is characterised by climatic conditions leading to higher
moisture contents than in service class 2.
Implicit in all the above characterisations, as reference is made to upper bound
moisture conditions, is the expectation that structural properties decrease in
magnitude with increasing moisture content of the timber member. For service
class 3 no specific moisture condition is given but, based on the same expectation,
lower bound structural properties will be found provided that the moisture content
of the timber member exceeds its fibre saturation point.
2.2 Evaluation of Structural Properties of Timber Members in

Service Class 3 Conditions
In EN 1995-1-1 kmod values are given to modify strength values for both load-
duration and service class. For solid timber the ratio of service class 3 kmod value
to service class 1 kmod value ranges from 0.78 to 0.83 depending on which load-
duration class is being considered. This is simply a consequence of rounding
errors with the intended ratio being 0.8, which is the ratio found by interpolation at
the test duration for determination of characteristic values of 5 minutes.
EN 1995-1-1 gives a single set of kmod values for all stress types. This is unlike
some National Codes such as the British Standard BS 5268-2, where the
characterisation of service class is the same as EN 1995-1-1 but where
modifications for service class 3 are markedly different ranging from 0.6 to 1.0
depending on the stress type. It also appears to be at odds with the European
Standard EN 384 (2010) where in clause 5.3.4.2 clearly differing adjustments for
moisture content are made for different stress types.
This may have been a pragmatic decision by the Code-writers, putting ease of
use over accuracy for this design aspect, in view of the fact that the majority of
structural usage of wood-based materials is in service classes 1 or 2. However
whilst such pragmatism might be sensible for solid timber, in view of the fact it is
almost invariably used in external applications, for Accoya wood a more accurate
conversion of service class 1 characteristic values to service class 3 design values
is desirable.
3 Test Investigations to Determine Accoya Wood Structural

Properties in Service Class 3 Conditions
Table 1 summarises six test studies which investigated the relative magnitudes of
structural properties in service class 1 (65% RH / 20 C) and service class 3
(immersed under water for 6 weeks) conditions.
Table 1 Studies comparing structural properties under service class 1 and service class 3
conditions
Structural property Test laboratory at which tests

investigated undertaken
Modulus of elasticity, SHR, The Netherlands
Bending strength
Compression strengths parallel University of Brighton, UK
and perpendicular to grain
Embedment strengths parallel University of Brighton, UK
and perpendicular to grain
Shear strength at a notched end University of Brighton, UK
Strengths of nailed hanger joints Simpson Strong-Tie EU accredited
laboratory
Withdrawal strength of nails Simpson Strong-Tie EU accredited
laboratory
3.1 Modulus of Elasticity and Bending Strength

The effect of wood moisture content on modulus of elasticity and bending strength
was investigated and reported by Tjeerdsma et. al (2007). Bending tests, in
accordance with EN 408, were carried out on 200 mm long and 22 mm x 100 mm
boards (thickness x width) under three different moisture conditions. Both Accoya
Radiata pine and untreated Radiata pine were tested. The three moisture
conditions were conditioned at 65% RH / 20 C, conditioned at 90% RH / 20 C
and water soaked.
The results are presented for modulus of elasticity but similar trends are seen
for bending strength. In table 2 for both the Accoya and the untreated Radiata pine
the moduli of elasticity in each of the two service class 3 conditions (SC3) are
given as a proportion of the modulus of elasticity measured in the service class 1
condition (SC1).
Table 2 Moduli of elasticity under various moisture conditions
Modulus of elasticity under following conditioning as a

proportion of the modulus of elasticity at 65% RH / 20 C
65% RH/20 C 90% RH / 20 C Water soaked
(SC1) (SC3) (SC3)
Accoya 1.00 0.90 0.91
Radiata pine
Untreated 1.00 0.80 0.64
Radiata pine
3.2 Compression Strengths Parallel and Perpendicular to Grain

Compression tests were undertaken on Accoya Radiata pine and Accoya Southern
pine. For both Accoya species the compression tests were carried out on members
with a 38 mm x 89 mm cross-section. Two specimens to be tested in compression
parallel to grain and two specimens to be tested in compression perpendicular to
grain were cut out of each Accoya member. In order to optimise the matching of
each pair of specimens, and bearing in mind that compression test specimens are
small in size, the two extracted specimens had been located immediately adjacent
to one another in the Accoya member. One specimen of each pair was placed until
it reached equilibrium in a 65% RH / 20 C conditioning chamber, whilst the
second specimen was immersed under water to constant mass.
For each pair of both the compression parallel to grain specimens and the
compression perpendicular to grain specimens it was ensured that both the service
class 1 specimen and the service class 3 specimen were tested on the same day. 15
no. test replications were carried out for each permutation of load direction/
species/ service class.
The compression parallel to grain specimens were nominally 240 mm long and
38 mm x 89 mm in cross-section and were tested in accordance with EN 408.
EN 408 stipulates that solid timber compression perpendicular to grain
specimens are 70 mm long, 45 mm wide and 90 mm deep. As this cross-section
was not available in Accoya, the tests were carried out on 70 mm long, 38 mm
wide and 89 mm deep specimens. As these Accoya specimens are more slender
than the EN 408 specification, it is considered that they provided lower bound
compression perpendicular to grain strengths. The compression perpendicular to
grain specimens often continued to increase in load despite obvious distress to the
specimen. In order to enable a consistent approach for comparison purposes, the
load at 10% strain was taken as the ultimate load by which stage the load had
either stopped increasing or was only rising very gradually. This level of strain
was referenced in the 2004 version of EN 1995-1-1 as the approximate strain in
the member at the ultimate limit state.
Table 3 Summary of results of compression tests on Accoya

Type of test Accoya Characteristic Mean ratio of
species compression strength SC3 strength to
(N/mm2) in: SC1 strength
Service Service for matched
class 1 class 3 pairs of
specimens
Compression parallel Radiata pine 40.2 31.2 0.79
Compression parallel Southern 51.3 41.6 0.79
pine
Compression Radiata pine 5.9 4.5 0.73
perpend.
Compression Southern 5.0 4.5 0.81
perpend. pine
A summary of the characteristic compressive strengths and the mean ratios of

service class 3 to service class 1 strengths for the matched pairs of specimens are
given in table 3. The characteristic strengths were evaluated in accordance with
EN 14358.
3.3 Embedment Strengths Parallel and Perpendicular to Grain

The embedment tests were carried out on 20 mm thick, 100 mm wide Accoya
members. The dowel size tested was a 10 mm diameter bolt. The member
thickness-to-dowel diameter ratio of 2 was therefore within the range
recommended by clause 6.2 of EN 383. Two specimens to be tested in embedment
parallel to grain and two specimens to be tested in embedment perpendicular to
grain were cut out of each of the 15 no. Accoya members. In order to optimise the
matching of each pair of specimens, the two extracted specimens had been located
immediately adjacent to one another in the Accoya member. One specimen of
each pair was placed until it reached equilibrium in a 65% RH / 20 C
conditioning chamber, whilst the second specimen was immersed under water for
6 weeks.
For each pair of both the embedment parallel to grain specimens and the
embedment perpendicular to grain specimens it was ensured that both the service
class 1 specimen and the service class 3 specimen were tested on the same day. In
both cases the 10 mm diameter bolt was inserted into a pre-drilled 11 mm
diameter hole. The specimens were tested in accordance with EN 383. Although
most of the specimens continued to develop a gradually increasing reaction above
an embedment of 2 mm, in order to enable a consistent approach for comparison
purposes, the failure load was taken as the load at 2 mm embedment by which
stage significant crushing beneath the dowel had always occurred.
A summary of the characteristic embedment strengths and the mean ratios of
service class 3 to service class 1 strengths for the matched pairs of specimens are
given in table 4. The characteristic strengths were evaluated in accordance with
EN 14358.
Table 4 Summary of results of dowel embedment tests on Accoya

Type of test Characteristic Mean ratio of
embedment strength SC3 strength to
(N/mm2) in: SC1 strength for
Service Service matched pairs
class 1 class 3 of specimens
Embedment strength parallel to grain 49.0 30.6 0.70
Embedment strength perpend. to grain 22.1 17.2 0.80
3.4 Shear Strengths at a Notched End of a Beam

The selection of a method of test for investigating the shear strength of Accoya
took into consideration the following factors:
The method of test given in EN 408 for determining shear strength is

prohibitively expensive involving gluing tapered steel plates to opposite
faces of the timber member. Furthermore there is some evidence
indicating the EN 408 test method produces shear strengths 25-50% less
than the shear strengths found for flexural members from short span
beam tests.
Although short span beam tests can be readily undertaken, they can result
in a proportion of the tests not failing in shear. Trials on Accoya Radiata
pine indicated that even for a span-depth ratio of 6, the beams mostly
failed in bending rather than in shear.
The design method given in EN 1995-1-1 for shear strength at notched
ends introduces two additional factors kv (which takes into the geometry
of the notched end) and a material-dependent factor kn. Clearly a shear
test on a short-span beam with a notched end would give some indication
of the appropriateness of the solid timber values of the kn and kv factors to
Accoya, as well as investigating a condition that will frequently be
limiting in design.
This test study therefore investigated the shear strength at a notched end of an
Accoya Radiata pine member.
The shear tests were carried out using 16 no. members with a 38 mm x 89 mm
cross-section. Pairs of specimens (one for SC1 test and one for SC3 test) were cut
out of the same Accoya member having been located immediately adjacent to one
another within the Accoya member. In order to optimise the matching of each pair
of specimens, the notch was cut out at the end of each specimen where the two
specimens had previously been abutting one another as shown in figure 1. One
specimen of each pair was placed until it reached equilibrium in a 65% RH / 20C
conditioning chamber, whilst the second specimen was immersed under water for
4 weeks.
Notches
SC1 specimen subsequently cut of SC3 specimen
shear specimens
89 Two adjacent 640mm long shear specimens cut out of

Accoya member
Fig. 1 Cutting shear specimens out of each Accoya member

640 mm long, 38 mm x 89 mm beams, end-notched at one end, were tested in

central three-point bending over a span of 540 mm as shown in figure 2. Details of
the support arrangement at the end notch are shown in figure 3. It was ensured that
the matched service class 1 and service class 3 specimens were tested on the same
day.
50 270 270 50
Fig. 2 Set-up for shear tests on notched beams
57
32
90mm long steel
bearing plate
55 50
Fig. 3 General arrangement at notched support
A summary of the characteristic shear strengths and mean ratio of wet

specimen to dry specimen strengths for matched pairs of specimens is given
below. The characteristic strengths were evaluated in accordance with EN 14358.
The characteristic shear strengths at the notched end for the dry
specimens and the wet specimens were 2.3 kN and 2.8 kN respectively.
For Radiata pine Accoya in service class 1 conditions the characteristic
shear strength of a beam at a notched support found from test (2.3 kN)
was in reasonable agreement (+15%) with the equivalent characteristic

shear strength calculated in accordance with section 6.5.2 of EN 1995-1-
1 (2.0 kN) and utilising the solid timber values for the factors kcr (0.67)
and kn (5).
The mean ratio of wet specimen to dry specimen shear strength was
1.20.
For all specimens the mode of failure was a horizontal split emanating
from the corner of the notch.
Although the tests indicated that the shear strength of water saturated Accoya is
greater than the shear strength of dry Accoya, it should be noted that service class
3 conditions extend from moisture contents just greater than those pertaining to
service class 2 up to saturated members. Unusually, in the case of shear strength of
Accoya, the lesser strengths for service class 3 conditions occur for members with
moisture contents at the lower end of the aforementioned range and therefore
consideration should not be given to utilising higher kmod values for service class 3
than those given in EN 1995-1-1 for solid timber in service classes 1 and 2.
3.5 Downward Vertical Load Capacity of an End Grain-to-Side

Grain Connection Formed Using a Three-Dimensional
Nailing Plate (Hanger)
38 mm x 150 mm Accoya Radiata pine and Accoya SYP was used in the study.
Pieces, selected at random, were cut to the lengths prescribed by ETAG 015 for
the joists and headers and then conditioned in a room at 85% RH / 20 C until
constant mass was reached.
The Accoya members were assembled using the stainless steel hanger
SAIX250/38/1.5 fully nailed with stainless steel CNA4,0x35mm. The Accoya
SYP members were predrilled with a drill bit of 4 mm diameter since some trials
showed that nailing without predrilling can result in splitting the timber. Once
they were assembled half of the samples were placed in a room at 65% RH / 20C
and the other half were submerged in water until they reached constant mass.
The tests were carried out at Simpson Strong-Tie EU Laboratory in accordance
with ETAG 015 (Figure 4). Five test replications were done for both Accoya
species.
In view of the relatively small number of test replications comparisons of the
capacities of the connection under service classes 1 and 3 are made in Table 5 at
mean level. The failure mode of the tests was the same for all samples tested. The
header member split before the differential displacement of 15 mm was reached
(Figure 5). As reference a mean value of 12.5 kN is calculated for C24 timber
(Norway spruce) when assuming Fk = 0.75 Fm in service class 1&2 conditions.
Table 5 Summary of results of hanger tests on Accoya
Accoya Mean downward vertical Mean connection capacity in

species capacity of hanger connection SC3
(kN) in: Mean connection capacity in
Service class 1 Service class 3 SC1
Radiata pine 17.7 18.9 1.07
Southern 18.9 21.9 1.16
Yellow pine
3.6 Nail Withdrawal Capacity

Stainless steel annular ring nails CNA4,0x35mm withdrawal capacities were
measured in accordance with EN 1382 for both Accoya Radiata pine and Accoya
SYP under service class 1 and service class 3 conditions. Ten test replications
were done for both Accoya species and comparisons of the nail withdrawal
capacities under service classes 1 and 3 are made in Table 6 again at mean level.
As reference a mean value of 0.81 kN is calculated for C24 timber (Norway
spruce) when assuming Fk = 0.75 Fm in service class 1&2 conditions.
Table 6 Summary of nail withdrawal tests in Accoya
Accoya Mean nail withdrawal capacity Mean withdrawal capacity

species (kN) in: in SC3
Service class 1 Service class 3 Mean withdrawal capacity
in SC1
Radiata pine 2.0 2.1 1.05
Southern 3.2 3.3 1.03
Yellow pine
Fig. 4 ETAG 015 test set up

Fig. 5 Typical failure mode in hanger tests
4 Summary of the Relative Strengths of Accoya Wood

between Service Class 3 and Service Class 1
Table 7 summarises service class 3-to-service class 1 ratios for a range of
structural properties of Accoya wood as found by test and compares these ratios
with the same ratio given in firstly EN 1995-1-1 and secondly the British Standard
BS 5268-2.
Table 7 Service class 3-to-service class 1 ratios for some structural properties of Accoya
wood
Structural property SC3 value for SC3 value for solid timber
Accoya SC1 value for solid timber
SC1 value for given in given in
Accoya EN 1995-1-1 BS5268-2
as found by test
Modulus of elasticity 0.91 No value 0.8
given
Compression parallel to 0.79 0.8 0.6
grain
Compression perpend. to 0.77 0.8 0.6
grain
Bolt embedment parallel to 0.70 0.8 0.7
grain
Bolt embedment parallel to 0.80 0.8 0.7
grain
Shear 1.0 0.8 0.9
Tension perpend. to grain 1.0 0.8 0.9
(Hanger test)
Nail withdrawal strength 1.0 0.8 1.0
The following observations are made in relation to the service class 3 to service
class 1 strength ratios (SC3-to-SC1 ratios) shown in Table 7:
1. A comparison between the SC3-to-SC1 ratios found by test for Accoya
and given for solid timber in BS 5268-2 indicate that:
The properties with the most marked reductions for solid timber
are the same properties with the largest reductions for Accoya.
Across all the structural properties tested the SC3-to-SC1
reduction is significantly less for Accoya than for solid timber.
2. The application to Accoya of the same constant ratio between the service
class 3 kmod factor and the service class 1 kmod factor used by EN 1995-1-
1 for solid timber would be inefficient for several structural properties of
Accoya.
Acknowledgements. The authors are very grateful of the pleasant cooperation with
University of Brighton (Dr. Dave Pope) and SHR (Prof. Dr. Andr Jorissen).
References
Alexander, J.: AccoyaTM. An Opportunity for Improving Perceptions of Timber Joinery. In:
Proceedings of the Third European Conference on Wood Modification, pp. 431438
(2007)
Bongers, F., Roberts, M., Stebbins, H., Rowell, R.: Introduction of Accoya wood on the
market technical aspects. In: Proceedings of the Forth European Conference on Wood
Modification, pp 301310 (2009)
Bongers, F., Alexander, J., Marcroft, J., Crawford, D., Hairstans, R.: Structural design with
Accoya wood. Submitted and accepted by International Wood Products Journal (2013)
BS 5268-2:2002. Structural use of timber Code of practice for permissible stress design,
materials and workmanship
EN 383:2007. Timber structures - Test methods - Determination of embedment strength
and foundation values for dowel type fasteners
EN 408:2010. Timber structures - Structural timber and glued laminated timber -
EN 1382:1999. Timber structures - Test methods - Withdrawal capacity of timber fasteners
EN 1995-1-1:2005. Eurocode 5 - Design of timber structures - Part 1-1: General - Common
rules and rules for buildings
EN 14358:2007. Timber structures - Calculation of characteristic 5-percentile values and
acceptance criteria for a sample
ETAG 015:2002. Three Dimensional Nailing Plates
Jorissen, A., Lning, E.: Wood modification in relation to bridge design in the Netherlands.
In: Proceedings of 11th World Conference on Timber Engineering (2010)
Kattenbroek, B.: How to introduce acetylated wood from the first commercial production
into Europe. In: Proceedings of the Second European Conference on Wood
Modification, pp. 398403 (2005)
Tjeerdsma, B., Bongers, F.: The making of a traffic timber bridge of acetylated Radiata
pine. In: Proceedings of the Forth European Conference on Wood Modification, pp. 15
22 (2009)
Tjeerdsma, B., Kattenbroek, B., Jorissen, A.: Acetylated wood in exterior and heavy load-
bearing constructions. Building of two timber traffic bridges of acetylated radiata pine.
In: Proceedings of the Third European Conference on Wood Modification, pp. 403411
(2007)
Structural Veneer Based Composite Products
from Hardwood Thinning Part II: Testing
of Hollow Utility Poles
Benoit P. Gilbert1, Ian D. Underhill1, Henri Bailleres2, and Robbie L. McGavin2

1
2
Abstract. Australian utility pole network is aging and reaching its end of life, with
70% of the 5 million poles currently in-service nationally installed within the 20
years following the end of World War II. The estimated investment required for
the replacement or remedial maintenance of the aging 3.5 millions poles is as high
as 1.75 billion dollars. Additionally, an estimated 21,700 high-durability new
poles are required each year, representing further investment of 13.5 million dol-
lars per year. Yet, agreements which progressively phase out logging of native
forests around Australia have been signed, giving the industry about 25 years to
make the transition from Crown native forests to plantations and private forests.
As utility poles were traditionally cut from native forest hardwood species, finding
solutions to source new poles currently presents a challenge. This paper presents
tests on Veneer Based Composite hardwood hollow utility poles manufactured
from Gympie messmate (Eucalyptus cloeziana) plantation thinning. Small diame-
ter poles of nominal 115 mm internal diameter and 15 mm wall-thickness were
manufactured in two half-poles butt jointed together, using 9 veneers per half-
pole. The poles were tested in bending and shear, and experimental test results are
presented. The mechanical performance of the hollow poles is discussed and com-
pared to hardwood poles cut from mature trees and of similar size. Future research
and different options for improving the current concept are proposed in order to
provide a more reliable and cost effective technical solution to the current shortage
of utility poles.
Keywords: veneer based composite structural products, Gympie messmate planta-

tion thinning, hollow utility poles, mechanical performance, hardwood poles.
1 Introduction
Utility poles, commonly referred to as power poles, support the energy network
and therefore represent a vital part of the infrastructure in developed countries.
DOI: 10.1007/978-94-007-7811-5_57, RILEM 2014
630 B.P. Gilbert et al.
Yet, Australias utility pole network is aging and reaching its end of life with 70%
of the 5 million poles currently in-service nationally installed within the 20 years
following the end of World War II. Structural failure of one pole may have severe
consequences on local communities and the economy in general.
The estimated investment required for the replacement or remedial maintenance
of the aging 3.5 million Australian poles is as high as 1.75 billion dollars. Addi-
tionally, an estimated 21,700 high-durability new poles are required each year for
extending the network and meeting energy demand, representing further invest-
ment of 13.5 million dollars per year [1].
Hardwood utility poles have been traditionally used and compared to more in-
tensively manufactured steel, concrete or fibreglass poles, they are still thought to
be the most economical and durable solution in terms of life-cycle costs [1]. Yet,
due to growing environmental awareness and concerns over the sustainability of
native forestry practices, agreements which progressively phase out logging of
native forests around Australia have been signed [2]. In South-East Queensland
where this project originated, the South East Queensland Forest Agreement
(SEQFA) [3] was signed in 1999 and phases out the logging of native forests by
2025, giving the industry 25 years to make the transition from Crown native for-
ests to plantations and private forests. Consequently, the shortage of utility poles is
expected to increase dramatically over the next decade while the demand is ex-
pected to rise sharply [1]. Finding solutions to eventually replace the aging poles
and source new poles currently presents a challenge.
Various alternative solutions to solid timber poles or columns have been pro-
posed in the literature. Adam et. al. [4] analysed the feasibility, design strength
and cost of hollow octagonal utility poles manufactured from composite wood
flakes panels. The poles were tapered with the wall thickness decreasing with the
height. Results indicated that the proposed poles were able to be cost competitive
in the market place. Kyoto University has been intensively researching on compo-
site structures ranging from LVL cylinders [5, 6] to spirally cylindrical laminated
veneer lumbers [7-9]. The latter product is manufactured through a continuous
process with veneer tapes winded around a mandrel in clockwise and anticlock-
wise directions to form an interlocking pattern [7]. Low quality logs were used for
the veneer tapes and the distance between butt jointed veneers was found to signif-
icantly influence the strength of the product [8]. Piao [10] manufactured 1.2 me-
tres (small-scale) and 6 metres (full-scale) long nonagon hollow poles from knot
free pine wood strips. While full-scale poles were not tested to failure, cantilever
tests performed on the small-scaled poles showed that shear failure can govern the
strength of the poles.
This paper presents tests on Veneer Based Composite (VBC) hardwood hollow
utility poles manufactured from Gympie messmate (Eucalyptus cloeziana) planta-
tion thinning, a specie traditionally used in Australia for poles due to its favour-
able natural durability and strength properties [11]. Six small diameter poles of
nominal 115 mm internal diameter and 15 mm wall-thickness were manufactured
in short lengths of 970 mm. To determine their mechanical characteristics, four
poles were tested in pure bending and two poles in predominant shear.
Experimental test results are presented in the paper and a simple failure mecha-
nism model of the poles is also introduced and compared to experimental results.
The mechanical performance of the hollow poles is discussed and compared to
hardwood poles of similar size cut from native forests. Future research and differ-
ent options for improving the current concept are proposed in order to provide a
more reliable and cost effective technical solution to the current shortage of utility
poles.
As mentioned in the companion paper [12], research is currently underway on
how best to connect short lengths of VBC products to form useable beams and
columns. The determination of the mechanical characteristics (stiffness and
strength) of these connections is outside the scope of the present paper.
2 Experimental Study
2.1 General
Twelve half-poles were manufactured around a 115 mm diameter steel mandrel in
lengths of 970 mm using the process detailed in the companion paper [12]. Each
half-pole was manufactured at ambient temperature and moisture content using
resorcinol formaldehyde structural adhesive from 9 veneers of 1.7 mm nominal
thickness, i.e. forming a nominal wall-thickness of 15 mm. Veneers were deliv-
ered in sheets of 1.2 m 1.2 m and different sheets were randomly selected for
each layer constituting a half-pole. Two half-poles were butt jointed together using
structural epoxy resin to form a complete section, as shown in Figure 1.
Fig. 1 Example of a manufactured pole
Additionally, to determine the mechanical properties of each half-pole, flat

panels were also manufactured using the same veneer sheets as each half-pole,
glued in the exact same order.
Before testing, hollow poles and flat panels were left at 20C and 65% relative
humidity until they reached equilibrium at an average measured oven dry moisture
content of 14.3%.
2.2 Bending Tests
2.2.1 Test Set-Up

Four timber poles were tested in a 500 kN MTS universal testing machine in four
point bending to determine their bending stiffness and maximum bending capaci-
ty. The tests were run in displacement control and reached failure in about 7 mins
(Tests 1 and 2) and 4.5 mins (Tests 3 and 4).
Each end of the poles was rigidly connected to a steel tube of length L1 = 1,045
mm, as shown in Figure 2, to form beams of length L = 2,930 mm, i.e. about 20
times the external diameter of the tested poles. At the connection with the steel
tubes, polyester resin (pre-filled with 40% sawdust) was poured (see insert in Fig-
ure 2) (i) inside the poles to avoid ovalisation (local crushing) of the timber sec-
tion and (ii) on the outside of the poles to match the inside diameter of the steel
tube and ensure a tight fit between the two elements. Additionally, the steel tubes
were slotted over 300 mm at their top and bottom, and a bolted device was used to
sandwich the timber poles in the steel tubes, creating friction between the two
elements. The butt joints between two half-poles lied in the horizontal plane.
Three Linear Variable Displacement Transducers (LVDT) recorded the dis-
placement of the timber poles at their centrelines (LVDT 1) and at a distance d =
260 mm from their centrelines (LVDTs 2 and 3), as shown in Figure 2.
Fig. 2 Four point bending test set-up
2.2.2 Test Results
Table 1 gives the moment capacity Mb of the poles calculated as,

Fmax
Mb = L1 (1)
2
where Fmax is the total maximum applied load and L1 is the distance from the sup-
ports to the points of application of the loads, as shown in Figure 2.
The bending stiffness EpIp of the poles is also given in Table 1 and calculated
by performing a linear regression on the linear part of the load-displacement
curve,
4E p I p + 3
F= 1 2 (2)
L1d
2 2
where L is the distance between supports, d the distance between transducers and
i is the recorded displacement of LVDT number i. The second moment of area Ip
of the poles, calculated from measured external diameter and wall-thickness, is
also given in Table 1 for information.
In Test 2, failure developed at the compression side of the sample with the pole
opening up, as shown in Figure 3. In Tests 1, 3 and 4, shear failure occurred at the
connection with the steel tubes and therefore the moment capacities Mb given in
Table 1 represent lower bound values of the actual capacity of the tested samples.
Despite successive improvements of the test rig, the latter failure mode was not
eliminated and the rig will need to be redesigned in the future. Yet, the moment
capacities for Tests 1, 3 and 4 are within 2% of the capacities calculated using the
simple failure mechanism model detailed in Section 3, suggesting that shear fail-
ure in the steel/timber connection occurred at a similar applied moment than the
failure mode introduced in Section 3.
Table 1 Bending test results
Test/Pole Mb (kN.mm) EpIp (kN.mm2) Ip (mm4)

1 17,450 2.67108 1.40107
2 17,250 2.30108 1.39107
3 18,380 2.28108 1.47107
4 14,420 1.60108 1.43107
Fig. 3 Failure mode for Test 2

2.3 Shear Tests
2.3.1 Test Set-Up

Two timber poles were tested in a 500 kN MTS universal testing machine in three
point bending to determine their shear capacity. The tests were run in displace-
ment control and reached failure in about 11 mins (Test 1) and 3.5 mins (Test 2).
The poles were simply supported, with a distance between supports of L = 820
mm in Test 1 and 730 mm in Test 2, and the load was applied at the centreline of
the poles, as shown in Figure 4. To avoid local crushing, polyester resin (pre-filled
with 40% sawdust) was poured inside the timber poles at the point of application
of the load and at the supports. The butt joints between two half-poles lied in the
horizontal plane in Test 1 and vertical plane in Test 2.
Three Linear Variable Displacement Transducers (LVDT) recorded the dis-
placement of the timber poles at their centrelines (LVDT 1) and at a distance d
from their centrelines (LVDTs 2 and 3), as shown in Figure 4, with d = 410 mm in
Test 1 (i.e. at the supports) and 260 mm in Test 2.
Fig. 4 Three point bending test set-up
2.3.2 Test Results

The shear capacities Vs of the poles calculated as,
Fmax
Vs = (3)
2
where Fmax is the maximum applied load, was found to be equal to 34.7 kN for
Test 1 and 40.7 kN for Test 2. This corresponds to shear strengths s, calculated
using the following approximate equation for tubes,
Vs
s = 2 (4)
A
where A is the measured cross-sectional area of the poles, of 10.1 MPa for Test 1
and 12.4 MPa for Test 2.
Measured combined bending and shear stiffness can be found in [13].
2.4 Material Testing

For the four poles in bending (see Section 2.2), coupon (dog bone) samples, with
50 mm wide 225 mm long necks, were cut from the flat panels of the half-pole
undergoing tension, while the flat panels of the half-pole undergoing compression
were cut in two, reglued to form 30 mm nominal thick panels and further cut into
nominal 85 mm wide 180 mm long samples. The former samples were tested in
tension and the latter samples in compression following the method proposed in
the Australian and New-Zealand Standard AS/NZS 4357.2 [14] in a 500 kN MTS
universal testing machine, at the same strain rate as their associated poles. Three
samples were tested per half-pole.
The Modulus of Rupture (MOR) in tension f,tens and in compression f,comp of
the tension and compression samples, respectively, were calculated as,
Pmax Pmax
f ,tens = and f , comp = (5)
bd bd
where Pmax is the maximum tensile or compressive applied load, and b and d are
the measured width and depth of the tension or compression samples, respectively.
Table 2 gives the average measured MOR for all poles tested in bending.
Additionally, to accurately measure the Modulus of Elasticity (MOE) E of the
material and avoid measuring out-of-straightness deformations as well as tension
or compression deformations, two extensometers were positioned on opposite
sides of the tension or compression samples. The MOE of each sample was calcu-
lated as the average of the MOE given by the two extensometers, measured by
performing a linear regression on the linear part of the stress-strain curve. Table 2
gives the average measured MOE for the four poles tested in bending.
Table 2 Average MOE and MOR for the four poles tested in bending
Half pole in compression Half pole in tension

f,comp E f,tens E
Test/Pole (MPa) CoV (GPa) CoV (MPa) CoV (GPa) CoV
1 61.3 0.068 17.9 0.071 94.7 0.038 19.2 0.034
2 60.8 0.019 17.9 0.048 90.3 0.049 19.1 0.044
3 66.2 0.020 20.9 0.017 84.9 0.050 18.8 0.025
4 55.9 0.030 15.8 0.030 68.5 0.066 15.7 0.007
Tested material properties for the two poles tested in predominant shear can be
found in [13].
3 Simple Failure Mechanical Model in Bending
This section develops a simple failure mechanism model of the timber poles in
bending, based on perfect elastic-plastic (as encountered in Section 2.4) and brittle
behaviours of the material in compression and tension, respectively. Typically,

timber fails in tension at a higher stress than in compression, as shown in Table 2,
and therefore, plastic deformation would occur in the compressive side of the
poles before reaching the failure tensile stress f,tens in the tension side. In refer-
ence to Figure 5, the stress (x) at a distance x from the centreline of the pole is
given as,
f ,comp if x 0 and (x ) plastic

(x ) = E top (x ) if x 0 and (x ) plastic (6)

E bot (x ) if x 0
where Etop and Ebot are the MOE of the top half-pole in compression and bottom
half-pole in tension, respectively, (x) is the strain at a distance x from the centre-
line of the pole and plastic is the value of the compressive strain where plastifica-
tion occurs, given as,
f , comp
plastic = (7)
E top
The tensile and compressive forces Ftens and Fcomp in the pole, respectively, are
calculated as,
D
x =d x=
Ftens = D ( x )dA( x ) and Fcomp = (x )dA (x )
2 (8)
x = x=d
2
where D is the external diameter of the pole, d is the shift in the neutral axis due to
the plastification in the compressive side of the pole and dA(x) is the differential
area, as shown in Figure 5. The bending moment M in the pole is then given as,
M = Fcomp e = Ftens e (9)
where e is the lever arm between the resultants of the tensile and compressive
forces, as shown in Figure 5.
In sawn timber structures, bending failure is typically triggered by brittle failure
in the tensile side of the structure. A similar typical failure mode is hypothetically
assumed for hollow timber structures, and specifically that failure occurs when the
tensile stress reaches f,tens at x = D/2 + tbot/2 (i.e. at an average stress in the bot-
tom wall of the pole of f,tens), as shown in Figure 5, the moment capacity Mb of
the pole is then calculated by solving Ftens = Fcomp in Eq. (8) and finding the shift
in neutral axis d and lever arm e.
Table 3 gives the moment capacity Mb numerically calculated for each tested
pole in Section 2.2 by using the calculated MOE and MOR given in Table 2, and
measured external diameter D and wall-thickness of the top and bottom half-poles
ttop and tbot, respectively. Comparison with actual moment capacities, given in
Table 1, is also provided in Table 3.
(a) (b) (c)
Fig. 5 Simple failure mechanism in bending, (a) pole cross-section, (b) strain distribution
and (c) stress distribution
Table 3 Simple failure mechanism model results
Test/Pole Mb (model) (kN.mm) Mb (model) / Mb (test)

1 17,630 1.01
2 17,110 0.99
3 17,990 0.98
4 14,760 1.02
Results in Table 3 show that the simple failure mechanism model provides a
good approximation of the actual moment capacity of the poles. Furthermore,
Table 3 suggests that the observed failure modes in Section 2.2.2 result in similar
or slightly lower moment capacity Mb than the typical bending failure mode for
timber structures considered in the present failure mechanism model.
4 Discussion
4.1 Comparison with Existing Poles

The Australian/New Zealand standard AS/NZS 4676 [15] classifies timber utility
poles in seven strength groups ranging from bending characteristic strength fb of
100 MPa (Strength group S1) to 25 MPa (Strength group S7), and shear character-
istic strength fs of 7.2 MPa (Strength group S1) to 2.5 MPa (Strength group S7).
Considering a duration of peak load of less than 5 seconds, a full length preserva-
tive treated, unshaved and not steamed solid pole (see [15] for more details), the
moment capacity Mb of the pole is given as [15],
M b = f 'b Z (10)
where Z is the section modulus, and the shear capacity Vs as [15],
V s = f ' s As (11)
where As is the shear plane area, equal to 3/4 the cross-sectional area for solid
round cross-sections.
Therefore, a 145 mm external diameter solid utility pole cut from mature trees
would have a minimum moment capacity of 29,930 kN.mm in Strength group S1,
i.e. 1.6 times more than the maximum capacity found in Table 1 (pole number 3),
and of 7,483 kN.mm in Strength group S7, i.e. 1.9 less than the minimum capacity
found in Table 1 (pole number 4). Specifically in terms of moment capacity, pole
number 4 in Table 1 would be classified in Strength group S5, while poles num-
bers 1 to 3 in Table 1 would be classified in Strength group S4. While the investi-
gated poles fall in intermediate strength groups, they are 2.7 times lighter than
solid utility poles of the same overall dimension. Furthermore, the manufacturing
process would allow customising the diameter and wall-thickness of the hollow
pole to meet any desired design capacity. For instance, using the simple failure
mechanism model developed in Section 3, a hollow pole of external diameter D =
175 mm, wall thickness t = 25 mm, MOE E = 18 GPa, MORs f,tens = 70 MPa and
f,comp = 55 MPa (similar to tested pole number 4 in Section 2.2) would have a
moment capacity of 30,050 kN.mm, i.e. similar capacity and 1.4 times lighter than
a 145 mm diameter solid utility pole in Strength group S1.
Additionally, the investigated poles in Section 2.3 would be classified in
Strength groups S6 and S7 in terms of shear capacity. Yet, the dimensions of the
pole can also be customised to meet any desired shear capacity Vs. Considering a
shear strength s of 10 MPa (as experimentally found in Section 2.3.2) and Eq. (4),
the 175 mm 25 mm hollow pole in the previous paragraph would have a similar
shear capacity than a 145 mm solid pole in Strength group S4.
4.2 Future Research

As seen in Section 4.1 and previous studies [10], the shear capacity of hollow
poles can govern the design. Immediate future studies will investigate cross-
laminating the veneers at angles of +/- 10o with the pole longitudinal axis (similar
to [8]). This configuration should significantly increase the shear capacity of the
poles while marginally reducing their moment capacity. Additional studies would
also include (i) investigating the influence of the high proportion of natural defects
in waste thinning on the strength and variability in the mechanical properties of
thinning composite products, (ii) understanding the behaviour and particular fail-
ure modes of the hollow poles and (iii) ultimately establishing probability-based
limit state design criteria for the proposed sections, resulting in immediate benefit
to the community.
5 Conclusion
This paper presented bending, shear and material tests on VBC hardwood hollow
utility poles manufactured from plantation thinning. A simple failure mechanism
model in bending was able to reproduce the actual failure capacities. Results show
that the poles are able to compete with hardwood solid poles cut from mature trees
and of similar size. While the shear capacity of the hollow poles needs to be en-
hanced, poles can be custom-built to meet any desired bending and shear capaci-
ties. Future research and different options for improving the current concept are
also proposed, including cross-laminating veneers at angles of +/- 10o.
Acknowledgment. The authors would like to thank the Department of Agriculture, Fisher-
ies and Forestry, Australian Government and Forest and Wood Products Australia for their
financial support through the 2012, Science and innovation award for young people in
agriculture, fisheries and forestry, Forestry category.
References
[1] Francis, L., Norton, J.: Australian timber pole resources for energy networks, A
review, Department of Primary Industry and Fisheries, Queensland Government,
Brisbane, Australia (2006)
[2] Australian Government, Department of Agriculture Fisheries and Forestry, Regional
Forest Agreements, http://www.daff.gov.au/forestry/policies/
rfa (accessed on June 01, 2013)
[3] Queensland Government, Department of Environment and Resource Management,
South East Queensland Forests Agreement (SEQFA),
http://www.timberqueensland.com.au/Growing/SEQFA.aspx
(accessed on June 01, 2013)
[4] Adams, R.D., Krueger, G.P., Lund, A.E., Nicholas, D.D.: Development of utility
poles from composite wood material. In: Proceedings of the 7th IEEE/PES Trans-
mission and Distribution Conference and Exposition (Ed.: IEEE Service center),
Atlanta, U.S.A., pp. 3740 (1979)
[5] Sasaki, H., Kawai, S.: Recent research and development work on wood composites
in Japan. Wood Science and Technology 28, 241248 (1994)
[6] Hara, Y., Kawai, S., Sasaki, H.: Manufacture and mechanical properties of cylindri-
cal laminated veneer lumber. Japan Wood Research Society 81, 2830 (1994)
[7] Hata, T., Umemura, K., Yamauchi, H., Nakayama, A., Kawai, S., Sasaki, H.: Design
and pilot production of a spiral-winder for the manufacture of cylindrical laminated
veneer lumber. Journal of Wood Science 47, 115123 (2001)
[8] Berard, P., Yang, P., Yamauchi, H., Umemura, K., Kawai, S.: Modeling of a cylin-
drical laminated veneer lumber I: mechanical properties of hinoki (Chamaecyparis
obtusa) and the reliability of a nonlinear finite elements model of a four-point bend-
ing test. Journal of Wood Science, 17 (2011)
[9] Berard, P., Yang, P., Yamauchi, H., Umemura, K., Kawai, S.: Modeling of a cylin-
drical laminated veneer lumber II: a nonlinear finite element model to improve the
quality of the butt joint. Journal of Wood Science, 17 (2011)
[10] Piao, C.: Wood laminated composite poles. PhD Thesis, School of renewable natural
ressources, Louisiana State University, Louisiana, U.S.A. (2003)
[11] Bottle, K.R.: Wood in Australia, Types, properties and uses. McGraw-Hill, Sydney
(1983)
[12] Underhill, I.D., Gilbert, B.P., Bailleres, H., McGavin, R.L., Patterson, D.: Structural
Veneer Based Composite products from hardwood thinning Part I: Background
and manufacturing. In: Proceedings of the RILEM Conference Materials and Joints
in Timber Structures - Recent Advancement of Technology, Stuttgart, Germany
(2013)
[13] Gilbert, B.P., Underhill, I.D., El Hanandeh, A., Bailleres, H., McGavin, R.L.: Struc-
tural Veneer Based Composite products from hardwood thinning Part III: Testing
of hollow utility poles. ASCE Journal of Material in Civil Engineering (in prepara-
tion)
[14] AS/NZS 4357.2, Structural laminated veneer lumber, Part 2: Determination of struc-
tural properties - Test methods, Standards Australia, Sydney, Australia (2006)
[15] AS/NZS 4676, Structural design requirements for utility services poles, Standards
Australia, Sydney, Australia (2000)
Glulam from European White Oak: Finger
Joint Influence on Bending Size Effect
Simon Aicher and Gordian Stapf
Materials Testing Institute University of Stuttgart, (MPA Stuttgart,

Otto-Graf-Institut (FMPA)), Pfaffenwaldring 4 B, 70569 Stuttgart, Germany
{simon.aicher,gordian.stapf}@mpa.uni-stuttgart.de
Abstract. Glued laminated timber consists today predominantly of softwoods as

they are currently the main source of structural timber in the northern hemisphere.
Due to several reasons, hardwoods will increasingly gain a more important role as a
sustainable material resource which will of course therefore also impact glulam
production. In Europe, the wood species white oak (Quercus Robur, Quercus
Petraea) is the most important species other than beech, and is being increasingly
used in structural applications. Recently, the first national and European technical
approvals were issued for oak glulam. The design strength values were based on
extensive testing campaigns, with emphasis on the differences inherent between
growth regions of the raw material itself. The paper reports first on basic strength
properties and requirements for oak laminations and beams as specified in the tech-
nical approvals. Further, essential test results which form the basis for the character-
istic strength and stiffness parameters are given, showing a clear effect of size on the
bending strength. In order to assess the beam characteristics based on basic lamina-
tion and finger joint properties, the applicability of the new European strength
equations for softwood glulam is discussed. In the case where the important ratio of
finger joint vs. lamination strength is at least one, as is typical for softwoods, the
model applies for the considered hardwood databases as well. However, when the
lamination strength reaches very high values with means in the range of 80 MPa to
120 MPa, finger joint strength vs. lamination strength drops significantly below one.
It is then shown that the bending capacity and size effect can be well predicted for
specific high strength hardwood glulams by a serial model, effectively including the
bending stress gradient between adjacent laminations.
Keywords: hardwood, white oak, Quercus Petraea, Quercus Robur, glulam, finger
joints, bending strength, tensile strength, failure probability, size effect, serial
system, Weibull distribution.
1 Introduction
Contemporary components for engineered timber constructions are almost exclu-
sively made from softwood species. This is especially true in the Northern
DOI: 10.1007/978-94-007-7811-5_58, RILEM 2014
642 S. Aicher and G. Stapf
hemisphere where in recent years timber has become a more viable alternative to
the still dominating building materials such as reinforced concrete and steel. The
reason for the increased attention towards timber stems from the undeniable eco-
logical advantages and the increased product performances.
Within this mainstream attraction to timber, the use of load bearing components
made of hardwoods has gained considerable momentum as well. This fact is driv-
en by four major reasons: i) hardwoods prevail in Southern and Eastern parts of
the world, ii) the mechanical properties of hardwoods are very often superior to
softwoods, iii) the visual appearance of hardwoods is very attractive, and finally
iv) the price of softwoods is rising due to increased demand. Of further importance
is the fact that reforestations are often done with hardwoods to obtain near to
nature forests andin the case of oakdue to their better aptness for soil and
climate impacts.
Beech and oak are the two hardwood species with the largest growth areas in
Europe. For instance, 11 % and 27 % of the forest area in France is covered with
beech and oak wood, respectively. Contrary to beech, oak wood has been used
since ancient pre-roman times for structural purposes. The reasons were twofold,
one being the considerably higher strengths as compared to most softwoods (when
graded and selected carefully) and the other being significantly increased durabil-
ity even versus softwoods with higher durability level, such as for instance larch.
In the last decade small dimension oak glulam has been used increasingly in
central Europe in so called post and beam window faades, which nowadays rep-
resents a widely used building technology for building envelopes. The manufac-
ture and use of oak glulam were hereby in general enabled by building authority
decisions. In a consequent follow up of the described systems use in faade sys-
tems, the first German technical approvals for oak glulam were granted [1; 2].
Recently, a European technical approval for oak glulam was issued on the basis of
a Common Understanding Approval Procedure (CUAP) [3], the first for hard-
wood.
2 Materials
2.1 Wood LaminationsVisual Grades and Strength

Requirements
The laminations used for the manufacture of the Gmiz/VIGAM [1; 3] and Schiller
[2] oak glulam, come from trees that belong to the genus Quercus, section Quercus
(Lepidobalanus), and are commonly termed as white oaks. Within the section of
white oaks, the prevailing European species are Quercus Petraea and Quercus
Robur. In the case of VIGAM oak glulam, the timber stems exclusively from South-
ern France, whereas the growth region of Schiller oak glulam comprises southern
Germany and the Czech Republic. (Note: in the following, producer Gamiz and
glulam brand VIGAM are abbreviated by G whereas producer Schiller is denoted
by S (EU).
Glulam from European White Oak: Finger Joint Influence on Bending Size Effect 643
Further, tensile strengths, MOE and density properties of American white oak
laminations are discussed. However, due to the unsolved bonding durability is-
sues, no results for glulam beams are presented for these species.
The oak wood used for both considered glulam products (G and S (EU)) must
conform to specific visual grading requirements and continuous verifications of
bending or tensile strength properties of the laminations, as stated in Table 1.
Note: the American white oak laminations, in the manuscript termed S (US) were
graded the same as the European oak boards S (EU) from the producer Schiller,
with regard to knot and pith requirements.
Table 1 Requirements for the laminations of VIGAM and Schiller oak glulam as specified
in the Technical approvals [1-3]
product VIGAM (G) Schiller

(S (EU))
grade DIN 4074-5 LS 10 (inner) LS 13 (outer) LS 13+
lamination dimensions
thickness mm 18-22 19-23
width mm 50-160 50-70
imperfections
single knots 13 16 7 mm
group of knots 12 13 7 mm
pith not permitted not permitted not permitted
density 12 kg/m 600-750
bending strength
unjointed lamination N/mm 38 47 80
jointed lamination N/mm 49 51 60
2.2 Glulam-Build-Ups and Properties According to Technical

Approvals
The ranges of the glulam dimensions and the specific build-ups produced accord-
ing to the technical approvals [1-3] vary considerably. This reflects the fact that
the approval holders aim at different sectors of the building market. VIGAM oak
glulam is approved and manufactured up to beam width and depths of 160 mm
and 400 mm respectively, whereas the dimensions of the Schiller oak glulam are
maximally 70 mm x 250 mm.
Apart from width and depth ranges of the beams and source of raw material, a
major difference between both products exists with regard to the cross-sectional
build-up as well. The VIGAM glulam is built-up as symmetrically inhomogene-
ous, such that the outer 6th of the cross-section, or at least two laminations, consist
of grade LS 13, and the inner part of the cross-section is comprised of grade LS 10
laminations. The Holz Schiller oak glulam is built-up homogeneously with grade
LS 13+ laminations throughout. Note: the technical approval [2] also contains
specifications for a premium brand with no finger joints in the outermost lamina-
tions and hence very high bending strength. However, this case is not considered
here.
Table 2 specifies characteristic strength and stiffness properties of both glulam
products as given in the respective approvals. The compilation reveals the rather
high bending strength, discussed below in detail. Further, it is apparent that the
compressive strength in the direction parallel to the fibre is significantly above the
strength level of todays European softwood glulam strength classes (see EN
14080 [4]). This issue and hereto based options for primarily compression loaded
column type structural elements are highlighted in a separate paper. The strength
and stiffness properties given in Table 2 result from extensive test series and com-
plementary calculations, partly revealed in the following.
Table 2 Requirements on the laminations of VIGAM and Schiller oak glulam as specified
in the Technical approvals [1-3]
oak dimensions strength MOE density

glulam
grade width depth
fm,k fc,0,k fc,90,k ft,0,k ft,90,k E0,mean k
mm mm MPa MPa MPa MPa MPa Gpa kg/m
VIGAM 33[3];
50-160 400 45 8 23 0.6 14.4 690
[1;3] 33.5[1]
Schiller
50-70 280 31.5 48 9 29 0.6 14 650
[2]
3 Test Results for Laminations
3.1 Tensile Strength, MOE and Density of Unjointed and Finger

Jointed Wood
The tension tests were performed with two different lamination widths in each
sample configuration as distinguished by producer, growth area and visual grade.
For the case of VIGAM G glulam, the width of the laminations was 100 mm and
140 mm in grades of both LS 10 and LS 13. The Schiller S (EU) and S (US) lami-
nation widths were 50 mm and 70 mm, respectively. The thickness of the lamina-
tions was 20 mm in all configurations tested, with a sample size of 50 specimens
per configuration.
All laminations and hereof made glulam beams had a moisture content in the
range of 9.3 % to 11.9 %. The average moisture content of the total investigated
material was 10.6 %.
All tension tests with unjointed and finger jointed laminations were performed
according to the relevant provisions in EN 408 [5], i.e. with a free length of 9
times the lamination width for the case of unjointed boards and 200 mm clearance
between the clamps of the test machine for the finger jointed laminations.
Table 3 gives a compilation of the statistical evaluation of the tension test re-
sults (strength, MOE and density) for all tested lamination combinations.
Table 3 Tensile strength results of unjointed and finger jointed white oak lamination
configurations
Pro- Width Grade Un- Tensile Strength ft,0,l and ft,0,j MOE Den
ducer jointed E0,t,mean -sity
or x50 C. xmin x05 12,k
Finger O. 1)
2-P /
Jointed V. 2)
Log 3-P
(FJ) Normal Weibull
2 2 2 2 2 kg/
- mm - - N/mm % N/mm N/mm N/mm N/mm 3
m
2)
- 55.5 32.0 25.3 30.6 30.7 14.3
LS 13 2)
641
100/ FJ 46.9 21.2 29.4 31.3 32.5 -
G
140 - 40.8 29.2 20.8 24.4 24.4
2)
13.1
LS 10 2)
675
FJ 43.9 20.1 26.5 30.3 30.8 -
2)
- 86.4 25.5 37.9 52.6 52.1 13.6
50/ 70 1)
FJ 59.5 18.9 30.8 40.6 39.8 -
S (EU) LS 13+ 1)
606
50 FJ 61.0 18.4 34.3 43.2 41.0 -
1)
70 FJ 58.2 18.2 30.8 40.3 39.2 -
2)
- 115.9 27.0 49.9 69.1 68.3 16.1
S (US) 50/ 70 LS 13+ 1)
697
FJ 63.9 19.2 35.3 43.6 41.9 -
The results for lamination width 100 mm and 140 mm in the case of producer
G and 50 mm and 70 mm in the case of producer S, with a sample size of 50 spec-
imens per each individual width, are regrouped here to groups of now
100/140 mm and 50/70 mm where each group is now comprised of 100 specimens
per investigated lamination grade. Since the results of the separate S (EU) groups
50 mm and 70 mm are needed for further calculations, those values are given here
as well. The 5 % quantiles are given for lognormal as well as for the 3-Parameter
(3P) (or in some cases 2P) -Weibull data approximations.
Figure 2a shows the cumulative frequencies/distributions of the tensile strength
of the unjointed laminations together with the fitted lognomal distributions. Fig 2b
gives the results for the finger jointed laminations together with the 2P (S (EU), S
(US)) and 3P (G) -Weibull fits.
a)
b)
Fig. 1 Experimental and fitted cumulative tensile strength distributions of tested white oak
lamination configurations
a) unjointed laminations, ft,0,l
b) finger jointed laminations, ft,0,j
The values specified in Table 3 as well as the graphs reveal the enormous dif-
ference between the different growth areas/species, which in case of tensile
strength of the laminations results in mean and 5% quantile values twice as large
when comparing batch G LS 10 to S (US). Regarding finger joint tensile strength,
the differences are far less pronounced, which is evaluated quantitatively by the
ratios ft,fj/ft,0,l presented below in Table 5 and discussed later. As employed in the
calculations below, the 2P-Weibull scale and shape parameters obtained for the S
(EU) laminations of grades LS 13+ are:
= 50 mm , = 65.41 N/mm , = 6.33 and

= 70 mm, = 62.34 N/mm , = 6.42 .
The mean values for the MOE range from 13.1 GPa to 16.1 GPa. The characteris-
tic values of density 12 lie between 641 kg/m and 675 kg/m.
3.2 Bending Strength of Finger Jointed Laminations

The bending tests were performed in 4-point bending according to EN 408, with
loads at a distance of a third of the length from each support, and a span length of
18 times the lamination thickness.
Table 4 gives the results of the statistical evaluation of the bending strength of
the finger jointed laminations for the different configurations.
Table 4 Finger joint bending strengths of investigated white oak lamination configurations
Pro- Lamination Grade n Finger Joint Bending Strength fm,j

ducer Width
x50 C.O.V. xmin x05
Log 3-P
Normal Weibull
- Mm - - 2 2 2 2
N/mm % N/mm N/mm N/mm
S13 200 68.5 14.5 35.9 51.1 51.1
G 100/140
S10 200 66.8 13.9 42.7 49.5 50.9
S (EU) 50/70 S13+ 100 84.7 14.7 49.0 63.7 62.8
S (US) 50/70 S13+ 100 90.4 13.7 57.9 69.8 69.55
3.3 Strength Ratios

Table 5 gives some strength ratios which have an impact on the bending capacities
of the glulam beams and on their respective modeling. The table specifies, for the
mean and for the 5 % quantile strength level, respectively, the ratio of finger joint
tensile strength vs. lamination tensile strength ft,0,j/ft,0,l, the ratio of finger
joint bending strength vs. lamination tensile strength fm,j/ft,0,l and the ratio of finger
joint bending strength vs. finger joint tensile strength fm,j/ft,0,j. It can be seen that
the ratio ft,0,j/ft,0,l , which has the dominating influence on the load capacity of the
glulam beams, is predominantly very different from usual ranges obtained or re-
quired in softwood glulams where typically ft,0,j/ft,0,l 1.2 to 1.5, i. e. the finger
joint is stronger than the wood itself. This is here only given for the case of the
lowest lamination grade LS 10 (producer G) where characteristic tensile strength
ft,0,l is 24 N/mm, corresponding to T class T24 acc. to EN 14080:2013 [4] or C40

as given in EN 338 [6] and ft,0,l = 1.24 ft,0,l.
For the next higher grade LS 13 (producer G) the discussed strength ratio is 1
and for grades LS 13+ depending on growth area/species the ratio ft,0,j/ft,0,l is
considerably smaller than one, meaning that the joint is too weak to trans-
fer/activate the high tensile capacity of the wood. This is probably due to
combined influences of an insufficient ratio of shear vs. tensile strength of the
wood, adhesive and eventually finger joint profile, which can have a distinct influ-
ence (see [9], too).
It is further interesting to note that the ratio of finger joint bending vs. finger
joint tensile strength is throughout close to 1.6 on the characteristic strength level,
whereas for the case of finger jointed softwood laminations, the expected ratio
would be about 1.4.
Table 5 Finger joint tensile and bending strengths normalized to lamination and ratio of
finger joint tensile strength
Producer Lamination Grade Ratio ft,0,j/ft,0,l Ratio fm,j/ft,0,l Ratio fm,j/ft,0,j

width at level of at level of at level of
x50 x05 x50 x05 x50 x05

LS 13 0.85 1.02 1.23 1.67 1.46 1.63
G 100/140
LS 10 1.08 1.24 1.64 2.03 1.52 1.63
S (EU) 0.69 0.77 0.98 1.21 1.42 1.57
50/70 LS 13+
S (US) 0.55 0.63 0.78 1.01 1.41 1.60
Average of all configurations - - - - 1.45 1.61
Std. of all configurations - - - - 0.05 0.03
4 Beam Bending Tests
All tests were performed in a four point bending test setup as specified in EN 408
with a span and constant moment length of 15 and 6 times the beam depth, respec-
tively. In each configuration minimally ten beams were tested whereby the sizes
comprised the production ranges specified in the respective approvals. In the case
of the VIGAM G glulam, beams with depths of 100 mm, 200 mm, 300 mm and
400 mm were tested; for the case of Schiller S (EU) glulam, the beam dimensions
were (width x depth) 50 mm x 120 mm and 70 mm x 250 mm. Table 6 presents
the statistical evaluations of the experiments. The beam tests revealed, as antici-
pated and shown in Fig. 2, a considerable effect of size, i.e. of beam depth d on
bending strength. This influence was encountered in a rather similar manner on
the level of the mean and 5% quantile values. In the case of VIGAM G glulam
with four different sizes, the power function strength fit = delivered
closely matching exponents of c = 0.22 and 0.21 for the mean and 5 % quantile
level values, respectively. For Schiller S (EU) beams, tested solely with two con-
siderably different sizes, the size effect exponent is directly derived from the rela-
tionship , , , , = (250120) delivering for the 5 %
quantile level c = 0.22, which matches the results obtained for the VIGAM G
beams. It should be noted that the performed size evaluation does not account for
the presumably existent width effect, which for instance is addressed in [13].
Table 6 Bending strength, MOE and density of the white oak glulam beam tests
Pro- Beam Beam n Bending Strength fm,g Global Density
ducer/ Width Depth MOE 12,k
Build- x50 C.O.V. xmin E0,g,mean
x05
up
Log 2-P
Normal Weibull
- mm mm - N/mm2 % N/mm2 N/mm2 N/mm2 GPa kg/m3
100 120 10 54.4 10.4 48.5 43.7 43.0 10.8 740
100/140 200 20 49.5 11.0 39.3 39.4 39.7 13.5 736

G/C
100/140 300 19 43.0 10.9 35.0 34.1 34.9 14.6 736
140 400 10 42.6 8.9 37.5 35.3 35.1 14.0 747
50 120 30 56.8 20.7 32.4 37.4 35.5 13.1 659

S/H
70 250 15 45.4 17.3 35.5 31.8 31.0 11.9 668
Fig. 2 Bending strength of white oak glulam beams with different cross sectional sizes
The extent of the obtained size effects is larger than that encountered generally
with glulam made of softwoods where extensive investigations, e.g. by [10; 11]
delivered as a good/conservative estimate for spruce and fir species a size expo-
nent of c = 0.1 which is implemented in several design codes/standards (see [12;
13]). The absolute values for bending strength, although decreased considerably
by the size effect are throughout still in the range or well beyond the highest stan-
dardized softwood glulam strength class GL32 (homogeneous or combined) speci-
fied in the European glulam standard EN 14080; see also Table 3.
It will be shown in the following to what extent the given numbers for glulam
bending strength and their respective dependency on size can be predicted by
existing or new strength models.
5 Applicability of the European Softwood Strength Model to

Hardwoods
The new harmonized standard for glulam made of softwoods EN 14080:2013 [4]
specifies the characteristic bending strength equation for homogeneous glulam
based on the investigations reported in [14; 15] and several subsequent changes
within WG3 of CEN TC124/WG3 as
, , = 2.2 + 2.5 , ,, + 1.5 ,, 1.4 , ,, +6 . (1)
The bending strength is based on the two influencial parameters: i) characteristic

value (i.e. 5% quantile) of the tensile strength of the lamination ft,0,l,k and ii) on the
bending strength of the finger joints fm,j,k. Note: the bending strength of the finger
joints was chosen instead of the tensile strength because bending tests are easier to
execute and more widely recognized by glulam producers; in EN 14080 a ratio of
, , / , , , = 1.4 was assumed. The validity of Eq. (1) obtained from fits to
results of Monte Carlo simulations according to input parameters ranges is limited
to the below stated ranges of finger joint bending strength vs. lamination strength.
Hereby, the bending strength of the finger joint has to conform to the range
1.4 , ,, , , 1.4 , ,, + 12 . (2)
Other important glulam strength properties such as tensile and compression

strength parallel to the fibre direction are directly related to fm,g,k by
, , , = 0.8 , , , , , , = , , . (3a,b)
whereas the shear strength of softwoods is considered to be independent of the

respective glulam build-up and given as a fixed value of fv,g,k = 3.5 N/mm. The
strength properties of glulam with inhomogeneous build-ups, i.e. with different
lamination strength classes at different cross-sectional depth ranges, can be com-

puted by means of elastic composite beam theory.
In an attempt to apply the softwood glulam model to the investigated hardwood
glulam beams, it can be noticed that the components of the VIGAM G beams fit or
are at least close to the prescribed validity ranges. For the case of grade LS 13, Eq.
(2) is fulfilled considering the x05 lognormal values given in Tables 3 and 4
, , = 51.1 Nmm > 1.4 , ,, = 1.4 30.6 = 42.84 N/mm, (4a)

, , = 51.1 N/mm < 1.4 , ,, + 12 = 42.84 + 12 = 54.84 N/mm . (4b)
For the case of grade LS 10 the applicability range of Eq. (2) is slightly violated as
, , = 49.5 N/mm > 1.4 24.4 + 12 = 46.2 N/mm.
However, the small excess of the upper application boundary can be regarded
as being marginal. Then, Eq. (1) delivers as input for the inhomogeneous VIGAM
G glulam built-up the homogeneous glulam values for LS 10 and LS 13+ beams
, ,( ), = 33.88 N/mm and , ,( ), = 36.73 N/mm (5)
and hence for the combined inhomogeneous LS 10/LS 13 build-up (1/6 of the
depth at both outer edges made of LS 13) by composite beam theory
, ,( / ), = 35.65 N/mm. (6)
This strength value, which is actually by its derivation inherently bound to a

beam depth of 600 mm, conforms in fact well with the characteristic bending
strength value of 35.3 N/mm derived in the tests for the largest tested cross-
section with a depth of d = 400 mm. By applying the capped size effect rule given
in Eurocode 5 [12] for the characteristic bending strength of glulam, hereby disre-
garding the code-specified upper band of strength increase (10 %) one obtains for
the smallest tested beam size of d = 120 mm
.
, ,( / ), , = 35.3 N/mm = 1.13 35.3 = 39.8 N/mm (7)
This result underestimates the empirically obtained value by 8 %.

In contrast to the VIGAM G glulam, it can be noted that the softwood glulam
strength Eq. (1) is not applicable in the case of the homogeneously built-up
Schiller S (EU), as Eq. (2) is violated by far:
, , = 63.7 N/mm < 1.4 , ,, = 1.4 52.6 N/mm = 73.64 N/mm . (8)
Hence, in order to validate the obtained beam test results of Schiller S (EU)
glulam in a somehow theoretical manner, the derivation of a new model equation
is inevitable. In order to address all influential effects, of which the most im-
portant are: scatter of lamination length, MOE and tensile strength of the lamina-
tions and finger joint strength, a full probabilistic model similar to [14-16], with
Monte-Carlo simulation of the partly random component properties is preferable
and presently established. However, less elaborate models are also suitable, as
shown subsequently.
6 Simple Stochastic Hardwood Glulam Model
6.1 General
In a first approach to model the glulam bending strength, it was proposed in [16]
to employ a simple serial material model based on two components - solid wood
laminations and finger joints - and their respective tensile strength probability
functions. (Note: In the subsequent equations, the failure probability distributions
are denoted by F and the related intact or survivor probabilities by S = 1 F).
According to basic probability rules, the failure probability (function) of a simple
serial model of n components is
=1 =1 =1 (1 ). (9)
In the case of a 2-component system, where = (failure probability of glu-

lam) and i = 1 = wood, i = 2 = fj (fj = finger joint) Eq. (9) delivers
= + . (10)
In the case of glulam, a two-component serial model represents certainly the most
simple approach, however it reveals to a good extent to what degree the strength
distributions of finger joints and of the unjointed wood influence the resulting
glulam bending strength [16]. For the instance where the model consists of wood
and several ( ) serially arranged finger joints, Eq. (9) delivers
= 1 = (1 ) 1 . (11)
In case no load sharing is considered, lower stressed joints in inner laminations

can be easily included by extending Eq. (11) to (tlam = lamination thickness)
= 1 (1 ) 1 ( ) 1 ( ) 1 ( ) 1 ( ) (12)
where = = ( 2), = ( ), = 2 .
For derivation of more easily used expressions, it is convenient to express the

failure probability of the single finger joint by a 2P -Weibull equation
=1 (13)
which, when introduced in Eq. (12) delivers

= 1 (1 ) . (14)
It can be seen that the depicted approach leads to a fictive increase of the num-
ber of finger joints subjected to maximum stress. In the case of the wood being
stronger than the finger joints and of zero overlapping of the strength distributions,
= 0, and then Eq. (14) reduces for instance for the two outer laminations to

= 1 . (15)
6.2 Application to Oak Glulam with Short, High Strength

Laminations
In order to validate the above presented model, the bending strength of S (EU)
glulam is calculated. For evaluation of the equations, the number of finger joints
nfj in the individual laminations is needed. In order to address the uneven moment
distribution, a modified constant moment length based on engineering sense or,
better, by Weibull transformation can be performed. Based on measurements of
the lamination length in the bending tension zone of all beams, the number of
finger joints per lamination was found to be nfj = 1.6 and 2.5 for the beams with d
= 120 mm and d = 250 mm, respectively. In the case of the small beams, the two
outermost laminations (tlam = 20 mm) are considered whereas for d = 250 mm
three laminations are considered in the strength calculations. Employing the
Weibull distribution parameters for the lamination tensile strength specified
above, the fictively increased number of finger joints nfj,eff = nfj [1+ (di/d)m] in
the serial model is obtained as
d = 120 mm: , = [1 + 0.67 . ] = 1.6 1.08 = 1.73 and

d = 250 mm: , = [1 + 0.83 . + 0.68 . ] = 2.5 1.39 = 3.48
Then Eq. (15) which fully disregards the contribution of the wood itself delivers
for the small and large beams the 5%-quantile bending strengths
, , , = 37.5 N/mm and , , , = 32.6 N/mm

The crudely derived characteristic beam bending strengths match the experimen-
tally obtained values (Table 6) very closely. Fig. 3 shows the full cumulative
strength distribution based on Eq. (15) and reveals a rather satisfactory agreement
with the test results. The graph further reveals the strength distributions of the
components (jointed and unjointed laminations) by and, based on Eq. (14), the
very small influence of the component wood which is only noticeable at higher
failure probabilities. The graph also highlights the fact that if joints are considered
only in the outermost laminations, the results would produce overly optimistic
strength estimates, especially when considering deeper cross sections.
Fig. 3 Empiric and theoretical strength distributions of oak laminations and beams
7 Conclusions
The paper has revealed that glulam made of oak hardwood differs in several re-
spects from softwood glulam. This is due to the high inherent hardwood properties
of the wood itself, growth area bound differences, production features and differ-
ent size ranges of the regarded products.
The results from two extensive investigations in the context of technical ap-
provals showed that visual grading of oak laminations based on somewhat modi-
fied German hardwood grading standard DIN 4074-5 [17] gives, for European
white oak (Quercus Robur and Quercus Petraea), very high tensile strength classes
of T24 and T30 (tensile strength 24 MPa and 30 MPa, respectively) which when
compared to standardized softwood strength classes, are at the upper or highest
level. In case knots are limited to very small values (less than 7 mm), characteris-
tic tensile strength arises to more than 50 MPa and in the case of North American
white oaks 70 MPa is achieved. Due to a completely different defect structure of
the wood, the lamination length is considerably smaller as compared to softwoods
whereby the range of average length varies between 0.5 m to a maximum 1.5 m.
This renders finger jointing to be even more important than for the case of soft-
woods.
The investigations have clearly revealed that the high tensile strength potential
of the material can only be activated up to strength classes of maximally T30 and
that beyond the ratio of finger joint tensile strength versus lamination strength
drops significantly below unity. For comparison: in softwoods this ratio is in the
range of 1.2 to 1.5. Regarding glulam bending strength which is influenced by the
wood itself and the finger joints, the investigations revealed an increased or almost
exclusive influence of finger joint strength at high lamination tensile strength clas-
ses with very little influence on the beam bending strength.
It was revealed that the standardized European lamination model for softwood
glulam applies to hardwoods, as well, as long as the model restrictions on finger
joint strength ratios are well fulfilled. For high strength oak laminations, however,
a completely different model should be applied. In this context, it was shown that
a simple modified serial failure probability model accounting primarily for the
finger joint strength and the respective occurrence frequency is well apt to model
the beam bending behavior and to predict the size effect of the beam depth accu-
rately, as the size effect is higher than that of the typical softwood glulam.
As glulams made of hardwoods become increasingly important i. a. due to sig-
nificant changes in the wood supply chain, it is of major importance to improve
the bonding technology of hardwood finger jointing and to further develop and
tailor strength models for hardwoods.
References
[1] DIBt, German Technical Approval Z-9.1-704. Glulam made of oak. Holder of ap-
proval: Elaborados y Fabricados Gmiz. S.A. Spain. issued by DIBT, Germany
(2012)
[2] DIBt, German Technical Approval Z-9.1-821. Holz Schiller oak post and beam glu-
lam. Holder of approval: Holz Schiller GmbH. Germany. issued by DIBT, Germany
(2013)
[3] OIB, ETA-13/0642 European Technical Approval VIGAM-Glued laminated timber
of oak. Holder of approval: Elaborados y Fabricados Gmiz. S.A. Spain. issued by
OIB, Austria (2013)
[4] EN 14080:2013 Timber structuresGlued laminated timber and glued solid tim-
berRequirements
[5] EN 408:2012 Timber structures Structural timber and glued laminated timber
[6] EN 338:2009 Structural timber Strength classes
[7] Larsen, H.J.: Strength of glued laminated beams Part 2 (properties of glulam lamina-
tions). Report No. 8004. Institute for Building Technology and Structural Engineer-
ing. Aalborg University, Aalborg, Denmark (1980)
[8] Aicher, S., von Ruckteschell, N.: Brettschichtholz aus Eiche (Oak glulam). In: 2nd
Stuttgarter Holzbau-Symposium - Neueste Entwicklungen bei geklebten
Holzbauteilen, November 0809, pp. 145154 (2012)
[9] Aicher, S., Radovic, B.: Investigations on the influence of finger joint geometry on
tension strength of finger jointed glulam beams. European Journal of Wood and
Wood Products 57(1), 111 (1999)
[10] Moody, R., Falk, R., Williamson, T.: Strength of Glulam BeamsVolume Effects.
In: Proceedings of the, Int. Timber Eng. Conf., Tokyo, Japan, vol. 1, pp. 176182
(1990)
[11] Lam, F.: Size Effect of Bending Strength in Glulam Beams. In: Proceedings Meet-
ing 44 CIB-W18-Timber Structures, Paper CIB-W18/44-12-5, Alghero, Italy (2011)
[12] EN 1995-1-1:2010 Eurocode 5: Design of timber structures Part 1-1: General
[13] ANSI/AF&PA NDS-2005 National Design Specification (NDS) for Wood Con-
struction with Commentary and Supplement: Design Values For Wood Construction
[14] Bla, H., Frese, M., Glos, P., Denzler, J., Linsenmann, P., Ranta-Maunus, A.: Relia-
bility of spruce glulamModelling the characteristic bending strength, Bd. 11,
Karlsruher Berichte zum Ingenieurholzbau. Universitts-verlag Karlsruhe, Karlsruhe
(2009)
[15] Frese, M., Hunger, F., Bla, H., Glos, P.: Validation of strength models for softwood
glulam. European Journal of Wood and Wood Products 68(1), 99108 (2010)
[16] Colling, F.: Tragfhigkeit von Biegetrgern aus Brettschichtholz in Abhngigkeit
von den festigkeitsrelevanten Einflussgrten. Ph.D. Thesis. University Karlsruhe
(1990)
[17] DIN 4074-5:2008 Sortierung von Holz nach der Tragfhigkeit Teil 5:
Laubschnittholz
Non-homogeneous Thermal Properties
of Bamboo
Puxi Huang1, Wen-Shao Chang1, Andy Shea1, Martin P. Ansell2,

and Mike Lawrence1
1
BRE CICM, Dept. Architecture & Civil Eng., University of Bath, UK
{P.Huang,W.chang3,A.shea,M.lawrence}@bath.ac.uk
2
BRE CICM, Dept. Mechanical Eng., University of Bath, UK
M.P.Ansell@bath.ac.uk
Abstract. A Phyllostachys edulis (Moso Bamboo)samples density, heat capacity

and thermal effusivity were obtained by a series of experiments. The porosity,
thermal conductivity and thermal diffusivity were calculated. Based on these ex-
perimental values, this study discusses the Phyllostachys edulis samples micro-
structure characteristics and the causes of the variation of thermal properties along
the radial direction.
Keywords: Moso bamboo, heat capacity, thermal effusivity, conductivity and

diffusivity, porosity, variation along radial direction, heat transfer models.
1 Introduction
Bamboo has been widely acknowledged and appreciated as an environmental

building material for its growing speed, carbon sequestration and mechanical
properties (Flander 2009). In terms of energy saving, a building materials thermal
performance parameters are often decisive. However, the following reasons mean
that bamboos thermal performance data is difficult to obtain:
Bamboo culm is a hollow cylinder rather than a solid. Therefore, flat board
based thermal performance test equipment cannot be utilised to test the curved
surface.
Although raw bamboo can be manufactured into flat samples, the size is still
restricted by its thickness and radius. Accuracy could not be assured in a num-
ber of main current stationary thermal parameters tests. In addition, the curved
bamboo cutoff materials thermal performance contribution is sacrificed in the
procedure of the rectangle manufacture.
Bamboo is a heterogeneous material and the changes in density and porosity
can cause thermal property variations.
DOI: 10.1007/978-94-007-7811-5_59, RILEM 2014
658 P. Huang et al.
Field study indicates that this variation is ignored in the rough manufacture pro-
cess. Outer green and inner yellow skin are planed along the radial direction with-
out considering adjusting for the thermal performance.
In order to discuss the Phyllostachys edulis made window-frame, prefabricated
wall panel and floor slabs thermal performance, a series of research works are
predicated. This study aims to provide the crucial simulation parameters to the
heat transfer modeling work.
2 Anatomical Observation of Phyllostachys Edulis in Radial

Direction
Anatomically, Phyllostachys edulis culms in radial direction can be classified into

four zones. The first zone is the Peripheral zone, namely the epidermis layer con-
sisting of the hypodermis and primary cortex tissue. Two zones form the middle
part, the Transitional zone and the Central zone. Both of these are composed of
vascular bundles and parenchymatous ground tissue. The difference is that the
vascular bundles remain undifferentiated in the Transitional zone, while they ap-
pear fully differentiated in the central zone. The fourth zone is the Inner zone. The
character of this zone is that simplified vascular bundles start to appear non-
directionally (Grosser and Liese 1971, Liu ed al. 2012).
Microscope observation indicates that the density of the Phyllostachys edulis
culms is inconsistent in radial direction. The Peripheral zones hypodermis and
primary cortex tissue is distinct from the other zones. The vascular bundles mor-
phological variation and the proportion of parenchymatous ground tissue dominate
density change in the other three zones. The metaxylem vessel cavity of the vascu-
lar bundles is the main channel of water transportation.
The vascular bundles metaxylem vessel cavity area is measured by Image J.
The boundary needs to be demarcated manually, and then the area and centre point
coordinates are calculated automatically. The figure 1 could illustrate every
metaxylem vessel cavitys position distribution and area change. This prior study
utilises the metaxylem vessel area so as to calculate porosity. The overall porosity
of this sample is 4.06%.
The figure 1 indicates that more small vascular bundles metaxylem vessels ap-
pear in the transitional zone. From central zone to inner zone, the individual
metaxylem vessel distributes more sparsely whilst the vessels area increases
gradually. The air cavitys position and area distribution could change the thermal
conductivity. For instance, if a large cavity area appears at a high density area, this
areas density will be counteracted; the thermal conductivity will decrease accord-
ingly. The densitys variation characteristics will be further discussed in the next
section.
Non-homogeneous Thermal Properties of Bamboo 659
Fig. 1 The metaxylem vessel cavitys position distribution and density variation
3 Heat Transfer Parameters
Heat transfer may be described by the following Partial Differential Equation

(PDE).
T T
C =k (1)
: Density (kg/m3)
C : Specific heat capacity (J/kg K)
k: Thermal Conductivity (W/m K)
T: Temperature (K)
t: Time (S)
The thermal conductivity, specific heat capacity and density of the
Phyllostachys edulis sample need to be obtained in order to satisfy the PDE equa-
tions definite condition (Incropera 1981).
3.1 Density
In order to quantify the density variation tendency, a CT scanner is utilised to
capture a grey scale image of the Phyllostachys edulis samples. Benchmark grey
scale value is provided by Balsa, Polypropylene, water and Aluminum.
The CT scan direction is from the Phyllostachys edulis samples internal sur-
face to its external surface. The samples thickness is 10.084mm. Therefore, the
samples absolute radial density value can be positioned along this direction. As
demonstrated by figure 1, high density values appear from the external side to the
660 P. Huang et al.
subsequent 2.5mm zone. The reason for this is that the zone constitutes hypoder-
mis and primary cortex tissue and non-differential vascular bundles. Then the
density values fluctuate and decrease along the radial direction. The peak density
values are attributed by the vascular bundles sclerenchyma sheaths and
parenchymatous ground tissues cell wall. The vascular bundles cavity is the
result of bottom values. Due to parenchyma cells aggregation, the density value
rises again at the internal surface side.
3.2 Heat Capacity

The heat capacity of three Phyllostachys edulis samples was tested by the
Differential Scanning Calorimeter (DSC). The samples dimensions and cutting
positions are illustrated by figure 2. The external sample and internal sample are
exactly cut at external surface and internal surface respectively. The middle sam-
ples center point coincides with the whole samples middle point at the radial
section. A specific manufacture device was developed to control the thickness of
the sample while keeping a smooth surface. The manufactured samples could well
fill into the DSC test pan.
Fig. 2 DSC test samples
Fig. 3 Heat capacity of the test samples

All of the samples are pre-heated to eliminate water. At the same temperature,
the middle sample has the highest heat capacity value, followed by the internal
sample. The external samples heat capacity is the lowest. Every samples heat
capacity curve has its peak value. The internal, middle and external samples heat
capacity peak values appear at 70.35, 64.68 and 80.32 respectively.
The DSC heat capacity test result indicates that the heat capacity is variable
with the temperature. Due to relatively low density and high porosity in the middle
zone, the middle samples heat capacity value is higher than the internal sample
and external sample. Every samples heat capacity value increases with tempera-
ture until achieving the peak value.
3.3 Thermal Effusivity

The thermal effusivity of six Phyllostachys edulis samples is tested by the C-
Therm (sensor T301) TCi system. This system is based on the Modified Transient
Plane Source (MTPS) method (Parlouer 2013). The samples are cut from the same
Phyllostachys edulis culm along the radial section. Each samples dimension and
cutting position are shown by figure 4. The samples thickness is 5mm. Both
length and width are 25mm. The external samples upper surface is the outer green
of the Phyllostachys edulis. The internal samples bottom surface is the inner skin.
The other samples are cut stepwise between the external sample and internal sam-
ple. Each step height is 1mm.
Fig. 4 Thermal effusivity test result

662 P. Huang et al.
Sample 3s effusivity value reveals an obvious distinction among the six sam-
ples. The other 5 samples have a similar tendency in that the effusivity value de-
creases along the radial direction from external surface to internal surface. The
reason for this is that a samples density starts to decrease while the vascular bun-
dles cavitys number still remains at a relatively high level at around 3mm.
By these average test values, the thermal effusivity along the radial direction
can be calculated via a data fitting process. The thermal effusivity variation ten-
dency is shown by figure 5.
Fig. 5 The samples thermal effusivity variation from external surface to internal surface
High values appear at 1.5mm from the external surface, then effusivity rapidly
decreases to the bottom value. From the 2.5mm to 5.5mm zone, the effusivity
slightly rebounds to 440 Ws0.5/m2K. The vascular bundles number declines in this
zone. The second descent zone ranges from 5.5mm to 8.5mm. The effusivity value
has a rising trend at the last 1.5mm. The fluctuation interrelates closely with the
density and porosity variation. The effusivity value will be utilised to calculate the
thermal conductivity value with heat capacity and density value in the simulation
process.
4 Thermal Conductivity and Thermal Diffusivity
4.1 Thermal Conductivity Calculation

Since the samples thermal effusivity, density and heat capacity can be set as varia-
bles of radial length, the formula 2 may be utilised to calculate the thermal con-
ductivity.
k = e /C (2)
e: Thermal effusivity (Ws0.5/m2K)
The external surfaces and internal surfaces thermal conductivity are remarka-
bly higher than that of the middle 8mm zone and this is due to high thermal
effusivity and low heat capacity. The average thermal conductivity of this sample
is 0.227W/m K. If using average thermal effusivity, density and heat capacity
value to calculate average thermal conductivity, the result is 0.162 W/m K. In
which case, the thermal conductivity is underestimated.
Fig. 6 The samples thermal conductivity variation from external surface to internal surface
4.2 Thermal Diffusivity Calculation

Thermal diffusivity can be utilised to quantify temperature balance speed. The
formula 3 describes the relationship between thermal diffusivity, thermal
conductivity, density and heat capacity.
= k/C (3)
2
: Thermal diffusivity (m /s)
Fig. 7 The samples thermal diffusivity variation from external surface to internal surface
If the tested variable values are utilised to calculate thermal diffusivity, the re-
sult is also a variable series. The average value of the series is 6.1810-7 m2/s. If
average values of thermal conductivity, density and heat capacity are utilised to
calculate thermal diffusivity, the result is 2.3710-7 m2/s. Therefore, simply using
the average value of a whole Phyllostachys edulis sample is inaccurate to reflect
the heat transfer process along the radial direction.
5 Summary and Future Works
We can see that the Phyllostachys edulis samples microstructure is variable along
the radial direction from anatomical observation. The porosity is dominated by the
664 P. Huang et al.
metaxylem vessels area and geometry distribution. The thermal conductivity and
thermal diffusivity calculation results indicate that it is not appropriate to describe
Phyllostachys edulis samples heat transfer process using average thermal parame-
ters. In terms of the heat transfer through heterogeneous material, like bamboo, the
density, heat capacity, and thermal effusivity vary and these variations cause
thermal conductivity and thermal diffusivity fluctuation along the radial direction
and are crucial factors in the heat transfer procedure.
The subsequent work will be further developed in the following fields:
Numerical heat transfer models from radial, longitudinal and tangential directions
will be built to simulate the heat transfer through these directions. The
Phyllostachys edulis sample from different positions will be tested to build a
thermal parameters database. The proving tests and database will be continuously
utilised so as to modify the numerical models error.
References
Flander, K.D., Rovers, R.: One laminated bamboo-frame house per hectare per year.
Construction and Building Materials 23(1), 210218 (2009),
doi: http://dx.doi.org/10.1016/j.conbuildmat.2008.01.004
Grosser, D., Liese, W.: On the anatomy of Asian bamboos, with special reference to their
vascular bundles. Wood Science and Technology 5(4), 290312 (1971), doi: 10.1007/
bf00365061
Incropera, F.P.: Fundamentals of heat transfer. Wiley, New York (1981)
Liu, K., Takagi, H., Osugi, R., Yang, Z.: Effect of physicochemical structure of natural
fiber on transverse thermal conductivity of unidirectional abaca/bamboo fiber compo-
sites. Composites Part A: Applied Science and Manufacturing 43(8), 12341241 (2012),
http://dx.doi.org/10.1016/j.compositesa.2012.02.020
Parlouer, P.L.: Thermal Conductivity of six bamboo sections SETARAM Instrumentation,
Caluire, H5793 (2013)
Part VII
Cross-Laminated Timber
Tapered Beams Made of Cross Laminated
Timber
Marcus Flaig and Hans Joachim Bla
Karlsruhe Institute of Technology, Timber Structures and Building Construction,

{marcus.flaig,hans.blass}@kit.edu
Abstract. A method to determine strength reduction factors for tapered beams made
of cross laminated timber (CLT) is presented. The method is based on EC5 equa-
tions for the calculation of strength reduction factors for tapered glulam beams. For
CLT-beams, however, the required strength properties, i.e. the shear strength and
the tensile or compressive strength perpendicular to the longitudinal direction, are
determined considering the beam layup and the different failure modes both affect-
ing the characteristics of CLT-beams. The analytical approach was substantiated by
the good agreement that was found between the strength reduction factors obtained
from the analytical approach and the results of tests performed with double tapered
CLT-beams.
Keywords: cross laminated timber, tapered beams, strength reduction factors, fail-
ure modes, beam lay-up.
1 Introduction
Pitched members with variable height are commonly used in timber structures for
both, aesthetic and economic reasons. Typical examples are tapered or double ta-
pered beams and columns in three hinged frames. However, since glulam, the most
established material in engineered timber structures, has poor strength properties
perpendicular to the grain and also a relatively small shear strength compared to
its strength properties in grain direction, transverse and shear stresses arising at the
pitched edges of tapered beams cause a rapid decrease of bending strength of tapered
glulam beams with increasing taper angle. In contrast, beams made of cross lami-
nated timber (CLT), a material consisting of several orthogonally bonded layers of
board lamellae, provide similar strength properties in two directions: the main beam
axis, which is the direction of longitudinal layers, and the direction of transversal
layers. Consequently, the tensile and the compressive strength perpendicular to the
longitudinal direction and also the shear strength of CLT-beams are considerably
higher than the corresponding values for glulam. These comparably high strength
DOI: 10.1007/978-94-007-7811-5_60, RILEM 2014
668 M. Flaig and H.J. Bla
properties make CLT particularly suitable for the production of tapered beams where
both stresses perpendicular to the longitudinal direction and shear stresses occur si-
multaneously at the pitched edges.
In most CLT products neighboured lamellae within individual layers are not bonded
to each other directly, at their facing edge sides, but indirectly via the crossing areas
with lamellae of neighbouring orthogonally arranged layers. Yet the bonding be-
tween the individual longitudinal lamellae is strong enough to ensure that the layers
act as solid units. It can therefore be assumed that the normal stress distribution in
tapered CLT-beams is similar to the distribution in tapered glulam beams and that
the reduced bending strength of tapered CLT-beams can be calculated as a function
of bending strength, shear strength and tensile or compressive strength perpendicu-
lar to the grain, like the reduced bending strength of glulam.
For tapered glulam beams Eurocode 5 [1] provides the strength reduction factors
given in Eq. (1) with values of ks = 0, 75 and ks = 1, 5 for beams with tapered tension
and compression zone, respectively. The factors take into account the interaction of
shear stresses and stresses perpendicular to the grain direction which in glulam is
beneficial for compression and adversely for tension.
1
k = 2 (1)

2
1 + f m
tan +
f m
tan2
ks f v ft(c),90
where:
fm is the bending strength,
fv is the shear strength,
ft(c),90 is the tensile or compressive strength perpendicular to the grain
The application of Eq. (1) to CLT-beams requires the knowledge of the strength
properties used in the equation. For any approved CLT-product the in plane bending
strength, which mainly depends on the quality of the lamellae within longitudinal
layers but is also influenced by the number of longitudinal layers, is provided by the
respective technical approval. Since the transversal layers do not contribute to the
in-plane bending resistance of CLT-beams, the bending strength is usually related
to the net cross section of longitudinal layers. With regard to the strength reduction
factors discussed here it is useful to relate the shear strength and strength properties
in direction of the transversal layers to the same cross section.
In CLT-beams both the shear strength and the strength in direction of transversal
layers do not only depend on the respective strength properties of the lamellae, but
also on the strength of the crossing areas between orthogonally arranged longitu-
dinal and transversal layers. Therefore the number and the arrangement of layers
as well as the width of lamellae have great influence on these strength properties
Tapered Beams Made of Cross Laminated Timber 669
of CLT-beams, too. Since shear stresses and stresses in direction of transversal lay-
ers cause stresses within both the lamellae and the crossing areas different failure
modes have to be considered when determining the strength properties.
Shear Strength
Three different failure modes can be distinguished in CLT-beams subjected to shear
stresses as shown in Fig. 1. The first failure mode takes into account a simultaneous
shear failure in all longitudinal and transversal layers within a beam. In the second
failure mode the possibility of shear failure in sections that coincide with joints be-
tween the unglued edges of lamellae is considered. In these joints only the cross
section of orthogonally arranged layers is available for the transfer of shear stresses.
The third failure mode takes account of shear stresses in the crossing areas of or-
thogonally bonded layers.
Shear stresses in the gross cross-section in failure mode 1 can be calculated ac-
cording to the Euler-Bernoulli theory taking into account the gross cross section of
a beam (see Eq. (2)).
Shear stresses in the net cross-section in failure mode 2 can be calculated accord-
ing to the beam theory taking into account the net cross section of a beam, which is
the smaller value of longitudinal and transversal layers (see Eq. (3)).
Shear stresses in the crossing areas of orthogonally bonded longitudinal and
transversal layers can be derived from the model of a composite beam as shown
in Fig. 2. From the transfer of differential normal forces dNi between longitudi-
nal lamellae two shear stress components result: Shear stresses yx (see Eq. (4)) in
longitudinal direction and torsional shear stresses tor (see Eq. (5)).
Shear stresses within the lamellae and the crossing areas of CLT-beams can be
calculated from the equations given below. Detailed information on the derivation
of Eq. (4) and Eq. (5) can be found in [2].
Vz Sy,gross
xz,gross = (2)
Iy,gross tgross
Vz Sy,net
xz,net = (3)
Iy,net tnet

6 Vz 1 1
yx = 2 (4)
b nCA m2 m3

3 Vz 1 1
tor = 2 3 (5)
b nCA m m
In Eq. (2) through (5) the following symbols apply.

Vz Shear force,
Sy First moment of area about the weak axis of the beam,
Iy Second moment of area about the weak axis of the beam,
t Thickness of the considered cross section,
b Width of lamellae in longitudinal and transversal layers,
nCA Number of crossing areas within the thickness of the beam,
m Number of lamellae within a longitudinal layer
When determining the effective shear strength of CLT-beams the interaction of

simultaneously acting shear stress components within the crossing areas needs to be
considered. The failure criterion given in Eq. (6) has been derived by evaluating test
results performed on different types of CLT-beams [3].
tor yx
+ 1 (6)
fv,tor fR
The effective shear strength of rectangular CLT-beams can generally be defined
as the minimum strength resulting from the three failure modes described. How-
ever, when calculating strength reduction factors for tapered CLT-beams the second
failure mode can be neglected, since shear stresses in transversal layers are small
near the tapered edges. Taking into account the first and the third failure mode only,
the effective shear strength of tapered CLT-beams related to the net cross-section of
longitudinal layers can be calculated according to Eq. (7).

t
fv,lam gross

tnet,long

fv,CLT = min nCA b 1 (7)

2 tnet,long

1 1
1 2 +
2 1
2
1

fv,tor mx fR mx mx
T T T
T T T
T T T
T T T
Fig. 1 Failure modes in CLT-beams subjected to transversal forces acting in plane direction:
shear failure in the gross cross-section (left), shear failure in the net cross-section (middle),
shear failure in crossing areas of orthogonally bonded layers (rigth)
qz t long,2
t long,1 t long,3
V V
a5
h=mb
a2 a4
centre plane
M M
a1
i
t cross,2
x
q
y t cross,1
y
x
z z
Vi Ni Mi N i + dN i
M tor,i
V V + dV
M M + dM dN i
dV i
b M i + dM i V i + dV i
Fig. 2 Side view and cross-section of a six-layered CLT-beam loaded in plane (top) and
internal forces in the beam, in the lamellae and in the crossing areas (bottom, from left to
right)
In Eq. (6) and (7) the following symbols apply.

fv,tor Torsional shear strength of crossing areas,
fR Rolling shear strength of crossing areas,
fv,lam Shear strength of lamellae,
tgross Total thickness of the beam,
tnet,long Summarised thickness of longitudinal layers,
m Number of lamellae within a longitudinal layer at position x
For European softwood the shear strength fv,lam of lamellae can be taken from
EN 338. In CLT, due to the crosswise arrangement of layers, the influence of cracks
on the shear strength of lamellae is low and a crack reduction factor of kcr = 1,0 may
be assumed. The torsional shear strength and the rolling shear strength, determined
by tests during the assessment procedure of CLT-products, can be found in techni-
cal approvals. For spruce (Picea abies) and fir (abies alba) the characteristic rolling
shear strength usually ranges between 0,9 and 1,2 N/mm2 whereas the characteristic
torsional shear strength varies between 2,0 and 2,7 N/mm2 .
Strength in Direction of Transversal Layers

In CLT-beams stresses perpendicular to the longitudinal direction which arise in
longitudinal lamellae at tapered edges can be transferred by transversal layers. The
transfer of these stresses into transversal layers also causes shear stresses within
the crossing areas. The tensile or compressive strength in direction of transversal
layers can therefore be defined as the minimum strength resulting from shear failure
in the crossing areas and failure in transversal layers due to tensile or compressive
stresses. Assuming uniformly distributed shear stresses within the crossing areas the
effective strength direction of transversal layers, again related to the cross section of
longitudinal layers, can be calculated according to Eq. (8).
tnet,cross tnet,cross

fc,0,lam or ft,0,lam

tnet,long tnet,long
f90,CLT = min (8)

n b

fR CA
2 tnet,long
In Eq. (8) the following symbols apply.
fc(t),0,lam Compressive (tensile) strength of transversal lamellae,
tnet,cross Summarised thickness of transversal layers,
nCA Number of crossing areas within the element thickness
b Width of lamellae in longitudinal and transversal layers
Strength Reduction Factors for Tapered CLT-Beams

In CLT, unlike in glulam, stresses acting in transversal direction do not adversely
affect the shear strength. In general a factor ks = 1, 0 Eq. (1) can be used when calcu-
lating the strength reduction factors for CLT-beams according to Eq. (1). However,
in CLT-beams with small gaps between the lamellae (less than 1 mm), compressive
stresses will activate friction forces between the edges of lamellae which increase
the shear strength. Therefore in beams with small or no gaps a factor ks = 1, 5 may
be assumed. As an example the strength reduction factors for CLT-beams with a
total thickness of 150 mm and lamellae of strength class C24 are given in Fig. 3
and Fig. 4. In the chosen example small gaps between the lamellae and an effec-
tive bending strength of fm,CLT = 24 N/mm2 have been assumed. The rolling shear
strength and the torsional shear strength of crossing areas were assumed as fR = 1,0
N/mm2 and fv,tor = 2,5 N/mm2 .
Twenty double tapered CLT-beams with two different taper angles were tested. Half
of the beams of each type were tested with compressive stresses (series C) and
tensile stresses (series T) at the tapered edge, respectively. All beams had the same
layup consisting of four longitudinal and two transversal layers. The dimensions of
the tested beams are given in Tab. 1. All specimens were produced from lamellae of
strength class C24 with a mean density of 438 kg/m3 at an average moisture content
of 10,6%. To determine the bending strength at the tapered edges two single loads
were applied in the third points of the span (Fig. 5).
In all specimens of series C6 failure was caused by compressive stresses at the
tapered edge (Fig. 6, left). In series C10 only in two specimens the compressive
strength was reached whereas in three specimens failure occurred in the crossing
areas between longitudinal and transversal layers (Fig. 6, right). In series T6 and
T10 failure occurred exclusively in the crossing areas. In all specimens a subsequent
splitting-off of the tapered ends of longitudinal lamellae was observed (Fig. 7).
Test results are summarised in Tab. 2. The given bending stresses m, ,net at the
tapered edge are related to the net cross section of longitudinal layers. The values
were calculated according to Eq. 9 taking into account the non-linear distribution of
bending stresses [4]. The given 5%-percentiles of bending strength were calculated
according to EN 14358.
6 Fmax x
m, ,net = (1 3, 7 tan2 ) (9)
tnet,long h(x)2
In Eq. (9) the following symbols apply.
Fmax Maximum load determined in the test,
tnet,long Summarized thickness of longitudinal layers,
x Position of maximum bending stress,
h(x) Height of the beam at position x
For the tested double tapered beams the position x, where the ratio of M(x)/W (x)
becomes maximum, was x = 1933 mm and x = 1697 mm for series C6/T6 and
C10/T10, respectively. The height of the beams at the respective positions was 500
mm for series C6/T6 and 600 mm for series C10/T10.
1,00
0,95 m8
m=7
m=6
0,90
m=5
m=4
0,85
m=3
,c
k
0,80
CLT:
0,75 Lamellae of strength
class C24 GL24h
Fig. 3 Strength reduction
tnet,long = 120 mm
factors k ,c for CLT beams 0,70
nCA = 4
and glulam beams with b = 150 mm
0,65
compressive stresses at the 0 1 2 3 4 5 6 7 8 9 10
tapered edge in
Table 1 Dimensions and layup of the tested double-tapered CLT-beams

Series Length Span Angle Height Total Layer Number
of taper at apex thickness thickness of layers
in mm in mm in in mm in mm in mm
C6 / T6 6000 5800 6 600 150 30 / 15 4/2
C10 / T10 7000 6800 10 900 150 30 / 15 4/2
Fig. 4 Strength reduction 1,00

m8
factors k ,t for CLT beams m=7
0,90
and glulam beams with m=6
tensile stresses at the tapered m=5
0,80
m=4
edge
m=3
0,70
,t
k
0,60
CLT:
0,50 Lamellae of strength
class C24 GL24h
tnet,long = 120 mm
0,40
nCA = 4
b = 150 mm
0,30
0 1 2 3 4 5 6 7 8 9 10
in
Fig. 5 Test setup for double- F F

tapered beams Top: Beams
with compressive stresses at
the tapered edge (series C)
Bottom: Beams with tensile
F F
stresses at the tapered edge L/3 L/6 L/6 L/3
(series T) F F
F F
Fig. 6 Failure in specimens

with tapered compression
zone Left: Failure caused
by compressive stresses in
longitudinal lamellae Right:
Failure caused by shear
stresses in the crossing areas
Both mean and characteristic values of the bending strength determined by tests
were used to calculate strength reduction factors k . Since the bending strength
parallel to the grain of the sample was unknown, the strength reduction factor for
series C6 was calculated from the analytical approach as described above and set
as reference level. The strength reduction factors of the remaining series were then
calculated from the ratios of the experimentally obtained bending strengths. In Tab.
3 the strength reduction factors k evaluated from the test results and the respective
values obtained from the analytical approach, calculated with characteristic values
of fR,k = 1, 0 N/mm2 and fv,tor,k = 2, 5 N/mm2 , are summarised and compared.
Fig. 7 Failure in speci-

mens with tapered tension
zone Left and right: Failure
caused by shear stresses in
the crossing areas and split-
off longitudinal layers
Table 2 Test results

Series No. Fmax m,,net mean,long Series No. Fmax m,,net mean,long
in kN in N/mm2 in kg/m3 in kN in N/mm2 in kg/m3
C6 1 106 39,4 424 T6 1 97,6 36,2 417
2 111 41,1 427 2 99,2 36,7 445
3 86,5 32,0 430 3 75,2 27,8 464
4 100 36,9 440 4 102 37,7 438
5 90,7 33,6 441 5 79,6 29,5 435
Mean value 36,6 432 Mean value 33,6 440
5%-Quantile 28,1 5%-Quantile 23,6
C10 1 155 32,4 429 T10 1 116 24,2 431
2 165 34,5 467 2 128 26,8 455
3 154 32,2 441 3 121 25,3 450
4 130 27,2 430 4 110 23,0 437
5 135 28,2 439 5 101 21,1 424
Mean value 30,9 441 Mean value 24,1 439
5%-Quantile 24,0 5%-Quantile 19,1
Table 3 Comparison of experimentally and analytically obtained strength reduction factors

k for tapered CLT-beams
Series fm,,mean k,exp k,anlyt Ratio fm,,k k,exp k,anlyt Ratio

in N/mm2 in N/mm2
C6 36,6 0,880 1,0 28,1 0,880 1,0
C10 30,9 0,743 0,744 1,002 24,0 0,752 0,744 0,990
T6 33,6 0,808 0,780 0,966 26,3 0,824 0,780 0,947
C10 24,1 0,579 0,608 1,050 19,1 0,598 0,608 1,017
4 Summary and Conclusions

A method for the calculation of strength reduction factors for tapered CLT-beams
is presented. The method is based on EC5 equations for the calculation of strength
reduction factors for tapered glulam beams. For CLT-beams the required strength
properties are determined considering possible failure modes and the beam layup.
Considering the variability of the material involved, good agreement has been found
between the strength reduction factors evaluated from tests and analytically obtained
values. Therefore it can be stated that EC5 equations for the calculation of strength
reduction factors are also suitable for CLT-beams, if the required strength properties
are determined with respect to the materials peculiarities.
The higher strength properties perpendicular to the longitudinal direction of CLT-
beams result in significantly smaller strength reductions than for glulam, making
CLT a viable alternative to the established material.
References
[1] EN 1995-1-1: Eurocode 5. Design of timber structures - Part 1-1: General -
Common rules and rules for buildings (2004)
[2] Bla, H.J., Flaig, M.: Karlsruher Berichte zum Ingenieurholzbau. Bd. 24. In:
Stabformige Bauteile aus Brettsperrholz. KIT Scientific Publishing, Karlsruhe
(2012)
[3] Flaig, M.: Karlsruher Berichte zum Ingenieurholzbau. Bd. 27: Tragfahigkeit
und Steifigkeit von Biegetragern aus Brettsperrholz bei Beanspruchung in
Plattenebene. KIT Scientific Publishing, Karlsruhe (2013)
[4] Riberholt, H.: Tapered Timber Beams. In: CIB-W18 Meeting 11, Paper 11-10-1.
Vienna, Austria (1979)
Influence of the Connection Modelling
on the Seismic Behaviour of Crosslam Timber
Buildings
I. Sustersic1, B. Dujic1, and Massimo Fragiacomo2

1
CBD d.o.o. Contemporary Building Design, Celje, Slovenia
iztok@cbd.si
2
University of Sassari, Dept. of Architecture, Design and Urban Planning, Alghero, Italy
Abstract. The paper investigates the influence of modelling different types

of connections in multi-storey cross-laminated timber buildings. The importance
of modelling the connection flexibility in the prediction of the natural vibration
periods and the base shear force of a crosslam building is demonstrated using
linear-dynamic analysis. The buildings global ductility and peak ground accelera-
tion were compared using non-linear static and dynamic analyses, demonstrating
that the former may lead to non-conservative result. The effect of gravity loads
does not seem to be crucial in the analyses, whereas including friction in the
model leads to a lower seismic response of the structure. This beneficial influence
may however be reduced by the vertical component of the seismic acceleration,
not considered in this study.
Keywords: Cross-laminated timber, Multi-storey buildings, Seismic design.
1 Introduction
Cross-laminated timber (also referred to as crosslam or Xlam) is a wood-based

product that has become widespread over Europe in the last decade. Recently,
it has also found its way to other continents, namely North America, and Austral-
asia. In addition to low-rise residential housing, crosslam has a great potential
in multi-storey timber construction particularly in earthquake-prone areas.
Unfortunately, research on seismic behaviour of crosslam buildings is hardly
keeping up with the increased need of safer and more demanding applications for
the building market. On the one hand, a lot of experiments (both wall racking and
shake table tests) have been conducted over the past years. But on the other hand,
the number of publications systematically dealing with the influence of boundary
conditions such as connection types, friction, vertical loads, floor diaphragm
influence, wall panels sizes, etc. on the global seismic behaviour of multi-storey
crosslam buildings has been fairly scarce. Most of the research work done so far
has been focused on only one type of a crosslam wall assembly, where the walls
DOI: 10.1007/978-94-007-7811-5_61, RILEM 2014
678 I. Sustersic, B. Dujic, and M. Fragiacom
mo
are composed from severaal narrower panels, adjacently connected with a single oor
double step joint [Ceccottti 2008].
This paper focuses onn the influence that different wall connection propertiees
have on the global seismic response of the multi-storey building. The inpuut
parameters for the parameetric study were:
connection types varying in strength;
influence of friction beetween walls and inter-story floor slabs; and
influence of vertical lo
oad on walls.
2 Case Study Bu
uilding
A four-storey crosslam buuilding (Figure 1) was modelled in this study. The buildd-
ing has 140 mm thick, 5-llayer crosslam walls along its perimeter. Inside the buildd-
ing there are only two possts and a beam that support 140 mm thick crosslam slabbs
running from wall A to wall C. Wall A is made from two separaate
panels, which are connectted only with a beam element pinned onto the walls. Thhe
other walls are all made out
o of single panels. The wall panels are connected at thhe
bottom and at the top with
h BMF105 bracket connectors. The spacing of the brackk-
ets was determined accorrding to the base shear force calculated with the laterral
force method (explained more in detail later). The building was modelled in SA AP
2000 [Computers & Strucctures inc.]. For modelling timber panels when perform m-
ing linear elastic and non
nlinear static analysis, shell elements and the reductioon
coefficients proposed by y Blass and Fellmoser [Blass and Fellmoser 20044]
were used. The floor diapphragms were modelled as rigid. The connections amonng
adjacent vertical panels were
w schematized in different ways (explained in detaail
later).
Fig. 1 FE model 3D view (lleft), wall B with number of brackets and their position (righht)
(measures in mm)
The following data was

w used for the modal response spectrum analysis oof
the building according to
o Eurocode 8 [CEN Eurocode 8, 2004]; type 1 elasttic
Influence of the Connection Modelling on the Seismic Behaviour 679
response spectra and type A soil, behaviour factor q = 1.0 and lower bound factor
for the design spectrum = 0.20. Ground acceleration was assumed to be 0.25 g,
with a building importance factor I = 1.0. The permanent load of the floor and
roof was 3.5 kN/m2 and 2.0 kN/m2, respectively. The imposed load on the floor
and roof was 2.5 kN/m2 and 2.0 kN/m2, respectively. The self-weight of the outer
walls was 1.2 kN/m2. The building was assumed to be category of use A
according to EN 1991-1-1 [CEN Eurocode 1, 2004], so the value of 2i for
quasi-permanent load was 0.3 and the factor was 0.5 for all floors except for the
roof where it was 1.0 assuming the roof is accessible. The mass was modelled as
lumped in the centroids of the floors 31.5 tons at the 1st, 2nd and 3rd floor and
19.4 tons at the top floor.
An important design issue is how to model the stiffness of the bracket connec-
tions. In this paper it is suggested that stiffness, strength and ductility of the steel
brackets and screwed connections are determined according to the procedure
proposed by Yasumura and Kawai [Yasumura and Kawai 1997] for wood framed
shear walls. For BMF 105 angular brackets with ten 460 mm nails a tensile stiff-
ness of 6345 kN/m and shear stiffness of 2767 kN/m were calculated [Fragiacomo
et al. 2011]. The ultimate tensile strength was calculated as 21 kN and the ultimate
shear strength was 16.5 kN.
3 Linear Elastic Analysis
The building main vibration period, base shear and story shear forces are calcu-
lated according to Eurocode 8 [CEN Eurocode 8, 2004]. A behaviour factor q
equal to 1 (elastic design) was used so as to compare the design ground accelera-
tion 0.25 g with the maximum peak ground acceleration the building could resist
calculated with nonlinear a static analysis procedure. The ratio of the design and
the maximum ground acceleration provides a simplified value of the behaviour
factor q [Ceccotti 2008]. Such a simplified behaviour factor q was calculated for
different types of connections and wall boundary conditions.
According to the linear elastic EC8 procedure the base shear was calculated to
be 712 kN. Hence the buildings connections were designed to resist the base
shear demand. To withstand 712 kN, 44 brackets are needed in every direction,
hence 24 of them would be installed on each of the parallel walls (Figure 1, right).
The resistance to the overturning moment due to seismic action was conserva-
tively assumed to be provided only by the walls running in the direction of the
seismic load and their corresponding bracket connectors. Under horizontal forces
the walls are assumed to rotate about the perpendicular wall on the perimeter.
When considering also the beneficial contribution of the vertical load assumed to
be applied in the centre of mass of each floor, the tensile strength of the brackets
was found to be sufficient. No reduction for coupled shear and tensile force in the
angle brackets was considered in design.
680 I. Sustersic, B. Dujic, and M. Fragiacomo
Fig. 2 Wall Calibration Procedure
When determining the elastic stiffness of brackets for use in the modal response
spectrum analysis, the values of the shear and tensile stiffness were calculated as
described in section 2. Each wall was then nonlinearly modelled using elastic shell
elements, gap elements (elements very rigid in compression with no stiffness in
tension) at the interface with the wall underneath, and elastic links for brackets
to simulate the exact boundary conditions of (rigid) contact in compression and
elastic behaviour in tension and shear. The wall models were then recalibrated
(Sustersic and Dujic 2012) so that only symmetrically elastic (same stiffness in
tension and compression) links were used, and the target horizontal displacements
(assumed as a sum of displacements caused by shear slip and rotation of the wall)
for the nonlinear and linear cases were the same under the same horizontal
load (Figure 2). With the so-determined fully elastic model, a modal response
spectrum analysis could be carried out.
The floor diaphragms were modelled as rigid. The connections among adjacent
vertical panels were schematized in different ways: (i) with rigid links (full 3D
model with rigid connections); (ii) with linear-elastic springs for the top and bottom
connections of walls, and without any connection between perpendicular walls at the
same level (pseudo 3D model); and (iii) with linear-elastic springs between
top/bottom and same level panels (full 3D model with elastic connections).
Figure 3 compares the natural periods, base shears and top floor lateral dis-
placements of the building for the three different types of models. The stiffness of
the building is very high for model i, where all connections are assumed as rigid.
This results in higher base shears and is conservative. However if the building was
lower and hence even stiffer, it could yield non-conservative results due to natural
periods being in the range of increasing spectrum outside the plateau region.
On the other hand, the top displacements are only 20% of what might actually be
expected. Although this does not compromise the general stability of the buildings
Periods Base shear Top displacement

0,7 350 30
0,6 300 25
0,5 250
20
0,4 i 200 i i
[mm]
[kN]
[s]
15
ii 150 ii ii
0,3
iii iii 10 iii
0,2 100
0,1 50 5
0 0 0
T1 T2 T3 Rx Ry Ux Uy
Fig. 3 Four storey case-study crosslam building Comparison of vibration periods, base
shear and top floor displacements among different models (i, ii and iii)
(unless second order effects are significant), it can lead to underrated damage
estimation. From Figure 3 it can be also observed that the difference between con-
sidering and ignoring vertical connections between perpendicular walls at the
same level is less than 4% when long wall segments are used like in the case under
study. This result may be different when shorter wall panels are used like in the
buildings developed in the SOFIE research project [Ceccotti 2008].
4 Nonlinear Static Analysis and the N2 Method
The nonlinear static analysis (NSA) procedure is more complex than the linear
elastic analysis, however it allows the designer to take into account the actual
inelastic behaviour of the structure. Furthermore, it can be used for Performance-
Based Design (PBD), where the design is achieved for different performance
levels such as no damage, limited structural damage, important structural damage
without collapse, etc.
A number of different methods for the evaluation of results of the nonlinear
static analysis have been proposed, including the N2 method [Fajfar 2000], which
has been adopted by the Eurocode 8. The N2 method was found to provide the
best approximation among various NSA methods for SDOF systems with different
hysteretic models and for MDOF systems [Fragiacomo et al. 2006]. However the
N2 was originally not developed for the design of timber buildings with specific
hysteretic behaviour. Therefore it must be noted that the results derived in this
study could be non-conservative, because hysteresis loops with pinching, slip and
strength degradation (typical of connections in timber structures) dissipate less
energy than bilinear plastic loops with the same ductility. Nevertheless, it should
be also pointed out that in the analysed setups, the SDOF systems equivalent to
the multi-storey building have periods longer than Tc which is usually the value
from where the reduction factor (R) and ductility factor () are considered to be
the same, regardless the type of hysteresis loop. The results of these analyses
should therefore be considered as a preliminary study aimed to investigate the
effect of different wall boundary conditions.
The standard bilinearisation procedure of the pushover curve suggested within

the N2 method assumes the attainment of the structure ultimate displacement
when the first structural component (beam, column, wall) reaches the near
collapse (NC) limit state. In this paper, however, the NC state was defined as a
global condition on the entire Xlam building according to the Yasumura & Kawai
[Yasumura and Kawai 1997] procedure. Therefore the NC state is assumed to be
attained for a top floor displacement such that the base shear force of the structure
drops by 20 % from the peak value. The initial stiffness of the bilinearised
pushover curve is also defined according to the aforementioned procedure instead
of using the standard N2 method formula. It should be noted, however, that the
results in this case could be non-conservative, as the storey shear force in an
individual floor might drop below 80 % of the peak value before the base shear
does the same.
The following types of connectors were analysed in the parametric study; (1)
BMF 105 angular brackets with twenty 460 mm nails 1st cycle backbone curve.
Unlike all other cases where the 3rd cycle backbone curve is used so as to account
for the strength degradation during a seismic event, the 1st cycle backbone curve
with an approximately 25% higher strength was considered in this analysis. (2)
BMF 105 angular brackets with twenty 460 mm nails that exhibited ductile be-
haviour in experimental tests under tensile load and therefore showed a desirable
failure mode. (3) BMF 105 angular brackets with twenty 460 mm nails, without
the influence of the vertical load. In this case the influence of vertical load on the
panels, calculated in accordance with the Eurocode 8 load combination, was
neglected as opposed to all other cases. All the aforementioned cases have friction
between the wall elements and interstory plates neglected. Cases 4, 5 and 6 are the
same as cases 1, 2 and 3 but consider friction. In the FEM program SAP2000
[Computers & Structures inc.] the friction isolator elements were used for mod-
elling the effects of friction. A friction coefficient of 0.4 was used (both for static
or dynamic loading), meaning that 40% of the vertical load on walls was trans-
formed into horizontal resistance. It should be noted, however, that in literature
the friction coefficient values between two pieces of timber range from 0.25 to
0.5. A 2D pushover analysis of wall B was then carried out for the six different
cases listed above.
From Figure 4 and Table 1 it can be clearly seen that the stiffness and peak
force of the analysed wall is markedly lower if the vertical load is neglected (case
3). Conversely, using the first cycles response backbone (case 1) results in an
increase of the walls stiffness and strength. However the buildings global ductil-
ity does not change much if using either the 1st or the 3rd cycles (case 2) backbone
curve. The maximum allowed ground acceleration is higher for the 1st cycles
response, followed by the 3rd and finally the lowest for the case without vertical
load applied. If friction is considered in the analysis, the allowed ground accelera-
tion is generally higher for all cases. The situation is, however, significantly dif-
ferent for the buildings global ductility factor; neglecting the vertical load results
in a higher ductility factor. Nevertheless the allowed acceleration is still smaller
(due to a reduction in shear resistance).
WALL B - EFFECT OF FRICTION

350
300
250
Base shear [kN]
200 1) 1 ST. CYCLE, NO FRICTION

4) 1 ST. CYCLE, FRICTION
2) 3 RD. CYCLE, NO FRICTION
150 5) 3 RD. CYCLE, FRICTION
3) 3 RD. CYCLE, NO FRIC. NO GRAVITY
6) 3 RD. CYCLE, FRICTION, NO GRAVITY
100
50
0
0,00 0,02 0,04 0,06 0,08 0,10
Top floor displacement [m]
Fig. 4 Pushover curves of wall B for different types of boundary conditions
Table 1 Ductility ratios, seismic demand and capacity of the case-study building in terms of
maximum acceleration depending upon the wall connection boundary conditions when the
wall B is loaded in the Y direction
Case No. 1 2 3 4 5 6
building ductility 2.39 2.36 1.93 2.59 2.86 3.13
target displacement [mm] 35 38 45 33 33 36
maximum displacement 83 81 70 80 74 66
max. peak ground acceleration [g] 0.59 0.52 0.38 0.61 0.55 0.45
max. ground acc. / design ground acc. 2.36 2.08 1.52 2.44 2.20 1.80
5 Nonlinear Dynamic (Time-History) Analysis
The non-linear time-history analysis (NDA) was carried out by selecting a set of 7
recorded earthquake ground motions from the European strong motion database
(using the Rexel software [Iervolino et al. 2010]) based on provisions of Eurocode
8-1 [CEN Eurocode 8, 2004]. Individual accelerograms were scaled so that their
average elastic response spectrum value is within the limits of the Eurocode elastic
response spectra (Figure 5) with 5% damping.
Nonlinear springs with hysteretic behaviour characterized by pinching and
strength degradation (pivot type in Sap 2000) were used for modelling the
behaviour of timber connections (angular steel BMF 105 brackets) in shear and
Fig. 5 Elastic EC8 response spectrum (with 5 % damping) and elastic spectra of chosen
scaled accelerograms
tension. The springs were calibrated on the results of experimental cyclic tests in
tension and shear [Dujic and Zarnic 2005]. Nonlinear dedicated friction finite
elements (friction isolator type in Sap 2000) were used (friction coefficient of
0.4) for modelling friction between walls and plates. Reyleigh damping was used
for the calculation of the damping matrix, which is computed as a linear combina-
tion of the mass and stiffness matrixes with coefficients a0 and a1 based on the
assumption of 5 % constant damping in the most important vibration modes.
To reduce the significant computational demand, a truss schematization of wall
B was investigated as an alternative to the use of shell elements. In this simplified
FE model, the Xlam wall panel is schematized with a couple of diagonal truss
elements, pin-connected at each storey. Nonlinear link elements are used for mod-
elling friction between plates, axial and shear forces resisted by angle brackets.
60 60
Displacement [mm]
Displacement [mm]
40 40
20 20
0 0
-20 -20
-40 -40
-60 -60
0 2 4 6 8 10 0 2 4 6 8 10
Time [s] Time [s]
Vertical NO Vertical Friction NO friction
Fig. 6 Comparison of wall B top floor displacements for the case of accelerogram 4675x
(South Island earthquake). On the left, the cases with/without vertical load considered in the
analysis are displayed, with friction neglected in either case. On the right, the cases
with/without friction between the wall and floor panels are compared, with the vertical load
considered in both cases.
In Figure 6 (left), the time-history of the top floor displacements for accelero-
gram 4675x are compared when the effect of vertical load was considered (gener-
ally, a beneficial effect) or neglected. Friction between wall and floor panels was
neglected in both cases. The average maximum displacements are about the same
during the strongest cycles of the earthquake ground motion. However in the
model without the vertical load the displacement amplitudes reach higher values
in an earlier stage and the model continues to sway with larger amplitudes even
after the more intensive part of the ground motion is finished. The gravity load, in
fact, causes a stabilising moment capable to counteract the overturning moment
due to the horizontal seismic forces.
Figure 6 (right) and Table 2 show that friction seems to significantly influence
the maximum top floor displacement. In Figure 6 the difference is most obvious in
the more intensive part of the ground motion, where the model with no friction
exhibits displacement on the top of the wall twice as high as displacement of the
model with friction. The average value of maximum top storey displacements for
the 7 chosen accelerograms (Table 2) shows that neglecting friction in the model
results in doubled displacements.
Based on these comparisons, the vertical load does not seem to noticeably af-
fect the maximum amplitudes of the total displacements. However, since the struc-
ture starts to experience higher displacements at an earlier stage and damps out
later, the structural damage would increase as the connecting elements undergo
higher amplitude cycles, hence accumulating more damage.
Table 2 Top floor displacements for the 7 chosen accelerograms if friction is considered or
neglected. The vertical load was considered in all cases.
Top floor displacement [mm]

Accelerogram
With friction Without friction
287x 14.23 38.75
372y 29.59 52.91
428y 27.91 37.98
1228x 16.83 40.35
1228y 29.17 54.56
4674x 20.56 48.91
4675x 16.74 34.33
Average displacement 22.15 43.97
Friction seems to have a strong effect on the building response. However it

should be noted that the vertical acceleration component was not taken into
account in this study. Vertical acceleration could result in the loss of contact
between walls and plates, hence reducing the beneficial effect of friction. When
comparing displacements from the NDA and NSA assessed with a modified N2
method, the target displacements from NSA cases without friction (1, 2) yield an
average displacement about 18% lower than from the NDA. In addition, dis-
placements from NSA cases with friction (4, 5) yield an average displacement
about 25% lower than from the NDA. The proposed modification (explained more
in detail in chapter 4) of the N2 procedure hence seems to lead to non-
conservative results. Nevertheless, further studies with a higher number of accel-
erograms and more wall types should be carried out before final conclusions on
this issue can be drawn.
6 Conclusions
The paper investigates how friction and different connection properties can influ-
ence a buildings seismic response. The elastic modal analysis has shown that
rigid connections between walls and plates result in very low vibration periods,
hence causing too low displacements. However in the case-study building the
prediction of the base shear force is conservative, as the stiff structures main vi-
bration period corresponds to the plateau of the design elastic spectra, hence re-
sulting in the highest seismic forces.
The results from the nonlinear static analyses show that if the first cycle hys-
teresis backbone is used to model the non-linear response of angular brackets, the
building can withstand about 10% higher peak ground acceleration. When includ-
ing friction in the model, the pushover curves are slightly steeper and the allow-
able ground acceleration a bit higher.
The results from the nonlinear dynamic analyses show that the vertical load
does not noticeably affect the maximum displacement amplitudes. However, the
damage to the structure is greater if the vertical load on walls is neglected in the
model. The connecting elements undergo higher amplitude cycles, hence accumu-
lating more damage during an earthquake. Considering friction, however, results
in a markedly lower response of the structure. Since the vertical acceleration was
not considered the results may be non-conservative.
A comparison of target displacements assessed with the push-over analysis and
a modified N2 procedure, and the average displacement over seven ground
motions compatible with the Eurocode 8 elastic spectrum from the non-linear
time-history analysis, shows that the proposed NLSA procedure may be non-
conservative. Further numerical analyses are needed to obtain final conclusions.
Acknowledgments. The research support provided to the first author by the EU through the
European Social Fund 'Investing in your future' is gratefully acknowledged. The technical
support of the IZIIS institute laboratory team is also highly appreciated.
References
Blass, H.J., Fellmoser, P.: Design of solid wood panels with cross layers. In: 8th World
Conference on Timber Engineering, WCTE 2004, Lahti, Finland, pp. 543548 (2004)
Ceccotti, A.: New technologies for construction of medium-rise buildings in seismic re-
gions: the XLAM case. IABSE Structural Engineering International, Special Edition on
Tall Timber Buildings 18(2), 156165 (2008)
Computers & Structures Inc. SAP2000 - Integrated Finite Element Analysis and Design of
Structures. Computers & Structures Inc., Berkeley (2000)
Dujic, B., Zarnic, R.: Report on evaluation of racking strength of KLH system, University
of Ljubljana, Faculty of civil and geodetical engineering, Slovenia (2005)
European Committee for Standardization (CEN). Eurocode 1: Actions on structures - Part
1-1 (2004)
European Committee for Standardization (CEN). Eurocode 8 - Design of structures for
earthquake resistance - Part 1 (2004)
Fajfar, P.: A nonlinear analysis method for performance-based seismic design. Earthquake
Spectra 16(3), 573592 (2000)
Fragiacomo, M., Amadio, C., Rajgelj, S.: Evaluation of the structural response under seis-
mic actions using non-linear static methods. Earthquake Engineering & Structural Dy-
namics 35(12), 15111531 (2006)
Fragiacomo, M., Dujic, B., Sustersic, I.: Elastic and ductile design of multi-storey crosslam
massive wooden buildings under seismic action. Engineering Structures 33(11), 3043
3053 (2011)
Iervolino, I., Galasso, C., Cosenza, E.: REXEL: computer aided record selection for code-
based seismic structural analysis. Bulletin of Earthquake Engineering 8, 339362
(2010), doi:10.1007/s10518-009-9146-1
New Zealand Standard. Timber structures standard, NZS3603. Published by Standards New
Zealand, Private Bag 2439, Wellington 6020, New Zealand (1993)
Sustersic, I., Dujic, B.: Simplified Cross-laminated Timber Wall Modelling for Linear
Elastic Seismic Analysis. In: CIB-W18, paper 45-15-6, Vxj, Sweden (2012)
Sustersic, I., Fragiacomo, M., Dujic, B.: Influence of connection properties on the ductility
and seismic resistance of multi-storey cross-lam buildings. In: CIB-W18, paper 44-15-9,
Alghero, Italy (2011)
Yasumura, M., Kawai, N.: Evaluation of wood framed shear walls subjected to lateral load.
In: Meeting 39 of the Working Commission W18-Timber Structures, CIB, Vancouver,
paper CIB-W18/30-15-4 (1997)
Behaviour of Cross-Laminated Timber Panels
under Cyclic Loads
Igor Gavric1, Massimo Fragiacomo1, Marjan Popovski2, and Ario Ceccotti3
1
Department of Architecture, Design and Urban Planning, University of Sassari, Italy
gavric.igor@gmail.com, fragiacomo@uniss.it
2
FPInnovations, Building Systems Department, Vancouver, BC, Canada
marjan.popovski@fpinnovations.ca
3
CNR IVALSA Trees and Timber Institute, S. Michele all'Adige, Italy
ceccotti@ivalsa.cnr.it
Abstract. In this paper, the behaviour of cross-lam (CLT) wall systems under
cyclic loads is examined. Experimental investigations of single walls and adjacent
wall panels (coupled walls) in terms of cyclic behaviour under lateral loading
carried out n Italy at IVALSA Trees and Timber Institute and in Canada at
FPInnovations are presented. Different classifications of the global behaviour of
CLT wall systems are introduced. Typical failure mechanisms are discussed and
provisions for a proper CLT wall seismic design are given. The influences of
different types of global behaviour on mechanical properties and energy
dissipation of the CLT wall systems are critically discussed. The outcomes of this
experimental study provides better understanding of the seismic behaviour and
energy dissipation capacities of CLT wall systems.
Keywords: Cross laminated timber, connections, single and coupled wall systems,
cyclic lateral loading, energy dissipation, failure mechanisms, seismic design.
1 Introduction
The key objective of this paper is to understand in detail how lateral systems in
cross-laminated (CLT) structures behave under seismic actions, and how they
influence the overall performance of CLT buildings in a seismic event. This is an
important issue as this technology has become widespread in several regions of
the world including earthquake-prone areas such as Italy, Japan and North
America. Thus, in the past and recently, several experimental projects were
performed on this subject.
A comprehensive study to determine the seismic behaviour of CLT wall panels
was conducted at the University of Ljubljana, Slovenia. Numerous quasi-static
DOI: 10.1007/978-94-007-7811-5_62, RILEM 2014
690 I. Gavric et al.
monotonic and cyclic tests were carried out on walls with lengths of 2.44 m and
3.2 m and a height of 2.44 m or 2.72 m (Dujic et al., 2004, Dujic & Zarnic, 2005).
Wall panels were subjected to different levels of vertical loads and monotonic or
cyclic horizontal load applied according to different loading protocols. Also, the
wall panels were tested at various boundary conditions, which enabled the
development of wall deformations from the cantilever type to the pure shear.
Influences of boundary conditions, magnitudes of vertical load and type of
anchoring systems were evaluated in terms of wall deformation mechanisms and
shear strengths of wall segments (Dujic et al., 2006a). Further, the influence of
opening on the shear properties of CLT wall panels was investigated (Dujic et al.,
2006b, 2007). Numerical models of CLT wall panels were implemented in the
software package SAP2000 and verified against test results (Dujic et al., 2008).
Results of the parametric study were used to derive analytical formulas describing
the relationship between the shear strength and stiffness of CLT wall panels
without and with openings.
This paper presents an extensive study of the behaviour of CLT panels
subjected to cyclic horizontal loads. In previous years, extended experimental
studies on CLT panels were conducted at CNR IVALSA Trees and Timber
Institute (San Michele all' Adige, Trentino, Italy) by Ceccotti et al. (2006), and
Gavric et al. (2011, 2013) and at FPInnovations in Vancouver, Canada (Popovski
et al. 2010). At CNR IVALSA, the first part of cyclic tests on CLT wall panels
was performed in 2006, focusing on single wall panel behaviour and behaviour of
CLT wall panels with openings (Ceccotti et al., 2006). The walls had different
types of connectors (hold-downs, angle brackets) and boundary conditions. A
second part of the experimental programme started in 2010 (Gavric et al., 2011,
Gavric, 2013), with the aim to test single connectors (hold-downs, angle brackets,
screwed joints) and better understand the behaviour of coupled CLT wall panels.
Different types of vertical joints were tested (step joint, spline LVL joint) with
several different configurations of anchoring connectors. Analysis of seismic
performance was done, with detailed investigation of energy dissipation properties
and equivalent damping ratio of CLT timber panels.
Similarly, an extensive experimental programme on CLT wall panels subjected
to cyclic lateral loads was undertaken at FPInnovations in 2010 (Popovski et al.,
2010). Wall configurations included single wall panels with three different aspect
ratios, multi-panel walls with step joints and different types of screws to connect
them, as well as two-storey wall assemblies. Various types of metal connectors
were used, as well as several types of fasteners such as annular ring nails, spiral
nails, and screws with different diameters and lengths.
In this paper, the results of 49 cyclic tests (25 conducted at FPInnovations in
2010 and 24 performed at CNR IVALSA in 2006 and 2010) are discussed and
compared. Test results were analyzed in terms of mechanical properties and
observed failure modes. In addition, evaluation and analysis of energy dissipation
properties and equivalent damping was performed. The influence on seismic
performance of various parameters, including geometry of panels, vertical loads,
connection configuration, number and type of metal connectors, type of fasteners
Behaviour of Cross-Laminated Timber Panels under Cyclic Loads 691
and type of vertical joint between adjacent panels was also studied. Two different
classifications of CLT behaviour types under cyclic loads will be introduced, and
parameters that are influencing overall performance of CLT walls will be studied.
2 CLT Wall Test Experimental Programme
2.1 CLT Wall Tests at CNR IVALSA Research Institute

In 2005, an experimental study to quantify the seismic behaviour of CLT wall
panels subjected to lateral loads was performed at CNR-IVALSA Institute
(Ceccotti et al., 2006). This experimental research was a first step of the SOFIE
project, which included experimental tests of full-scale CLT buildings (Ceccotti,
2008). The CLT wall testing programme involved single wall panels (2.95 m
2.95 m) with different connection layouts, including walls with openings,
subjected to cyclic loading with different levels of vertical loads applied on the
wall (Figure 1). Wall-foundation and wall-floor connections were tested with
different types of metal connectors, which were then used in the SOFIE buildings.
Results from the quasi-static tests on CLT wall panels showed that the connection
layout and design has strong influence on the overall behaviour of the wall.
Hysteresis loops were found to have an equivalent viscous damping of 12% on
average which makes the system suitable for implementation in high intensity
earthquake-prone areas. The cyclic tests showed that the construction system is
very stiff but still ductile (Ceccotti et al., 2006).
Configuration A Configuration B Configuration C Configuration D
Fig. 1 CLT wall test configurations tested at CNR IVALSA by Ceccotti et al. (2006)
(dimensions in mm)
Additional experimental tests were conducted on CLT wall panels in 2010 and
2011. The main goal of this testing programme was to provide a better insight of
the seismic performance of single and coupled adjacent CLT wall panels subjected
to cyclic loads and to understand the differences in their seismic behaviour. Wall
configurations included single panel walls with different connection layouts,
coupled wall panels with half-lap joints and different types of screws to connect
them. In total 16 cyclic wall tests were performed, including single walls and
coupled walls with in-plane screwed connections. Different connection layouts
were used, with the aim to investigate how they affect the entire wall behavior
and, possibly, optimize the structural performance.
Experimental wall tests were performed on 2.95 m 2.95 m single walls and
on 1.48 m 2.95 m coupled walls (Figure 2). Different connector layouts and
applied vertical load were used. The types of metal connectors, screws and nails
used were the same as those used in the 3-story SOFIE building tested on a
shaking table in Japan (Ceccotti 2008).
Configuration S-I Configuration S-II Configuration C-III Configuration C-IV
Fig. 2 CLT wall test configurations tested at CNR IVALSA by Gavric (2013) (dimensions
in mm)
The experimental test results were assessed in terms of strength, stiffness,

energy dissipation, equivalent damping ratio, ductility and strength degradation,
following the standard procedure from EN12512 (2001). The values of
mechanical properties for a single wall specimen were analyzed by taking into
account the results from the experimental hysteretic loops. Detailed assessment of
these properties is presented in Gavric (2013).
2.2 CLT Wall Tests at FPInnovations Research Institute

In 2010 a series of monotonic and cyclic CLT wall tests was conducted at
FPInnovations Research Institute, Vancouver, Canada (Popovski et al., 2010). In
total 12 different configurations including one- and two-storey walls, single and
coupled walls, and different wall-to-floor and wall-to-wall connections details
were performed. Different types of connectors (hold-downs, angle brackets) and
fasteners (ring nails, spiral nails, self-threaded screws, timber rivets) were used as
well as different aspect ratios, levels of vertical loads and loading protocols
(Figure 3).
Results from quasi-static tests on CLT wall panels showed that CLT walls can
have adequate seismic performance when nails or screws are used to connect the
steel brackets to the wall. The use of hold-downs with nails on each end of the
wall enhanced the seismic performance of the wall panel. The use of inclined
screws to connect the CLT walls to the floor panel underneath performed
I II III IV
V VI VII VIII
IX X XI XII
Fig. 3 CLT wall test configurations tested at FPInnovations by Popovski et al. (2010)
relatively poor in terms of seismic performance. The use of half-lap joints in

longer walls proved to be an effective solution to reduce the wall length and
consequently to improve the wall deformation capabilities.
3 Classification of CLT Walls Behaviour Types
Understanding the behaviour of cross-laminated timber (CLT) wall systems

subjected to lateral loads is crucial for a reliable seismic design of CLT buildings.
In order to understand better what affects CLT walls seismic performance in terms
of energy dissipation, hysteretic damping, ductility and displacement capacity
(drifts), different classifications of wall panel behaviour under cyclic loading is

presented in this section.
3.1 Classification Based on Predominant Type of Wall

Deformation during Cyclic Loading
The total top horizontal displacement tot (interstory drift) of a CLT wall panel is a
sum of four components: (i) rocking r; (ii) slip sl; (iii) shear deformation sh; and
(iv) bending deformation b (Figure 4):
tot = r + sl + sh + b (3.1)
Rocking Slip Shear Bending
Fig. 4 Deflection components of a CLT wall panel (Gavric et al., 2011)
In nearly all tests, most of the total top horizontal displacement was a
consequence of panel rocking and sliding. Shear and bending deformation of the
panels were generally negligible as, on average, their contribution on the total
deflection was only 2.77%. The only case where shear and bending deformations
became more important was CLT wall panels with large openings. A correlation
between type of CLT wall panel and predominant type of deformation was found.
Based on the results of numerous CLT wall cyclic tests, the following
classification based on predominant type of wall deformation during cyclic
loading can be proposed:
i) Rocking behaviour
ii) Combined Rocking-Sliding behaviour
iii) Sliding behaviour
While shear and bending contributions were almost negligible (except for walls
with large openings, see Dujic 2007), rocking deformation was governing the wall
behaviour in the cases of coupled wall panels and tall wall panels. It was found
that the geometry (aspect ratio) of the CLT panels influences significantly the
rocking and sliding contributions to the total lateral displacement of the panels. In
case of single wall panels with aspect ratios 1:1, panels with the same number of
metal connectors but using angle brackets at the ends instead of hold-downs
exhibited a significantly higher rocking behaviour. The higher proportion of

sliding in case of walls with hold-downs is due to the lower shear stiffness of
hold-downs in comparison with angle brackets. Furthermore, higher levels of
vertical load decreased the rocking contribution, thus increasing the sliding
contribution of the panel. Long wall panels had combined rocking-sliding
behaviour, again due to the panel aspect ratio. Walls with relatively low number of
metal connectors behaved predominantly with sliding as the resistance to
overturning moments was higher than the panel shear resistance. Figure 5 presents
the percentages of the rocking contribution to the total ultimate top horizontal
displacement of the CLT wall panels tested at IVALSA and FPInnovations.
CLT Walls Deformation: Rocking Contribution

100
90
80
Rocking [%]
70
60
Coupled Walls
50
Single Walls - Brackets Only

40
Single Walls - with Hold-downs
30
0 2 4 6 8 10 12 14 16 18
CLT Walls
Fig. 5 Percentages of rocking contribution to the total CLT wall deformation at ultimate
displacement under cyclic loading
As can be seen in Figure 5, the rocking contribution was the highest in coupled
CLT walls, followed by single walls with brackets only. The lowest rocking
contribution was observed in the case of single CLT walls with hold-downs and
angle brackets. On average, the rocking contribution for coupled walls was 84.2%
(N = 14, COV = 5.0%), for single walls with only angle brackets was 79.3%
(N = 14, COV = 11.1%) and for single walls with hold-downs and brackets was
61.5% (N= 17, COV = 18.3%).
3.2 Classification Based on Panels Interaction during Cyclic

Loading
Theoretically, there are three possible scenarios for the behaviour of adjacent CLT
wall panels subjected to cyclic lateral loads:
i) Coupled wall panels behaving as independent, individual panels

(Coupled wall behaviour).
ii) Coupled wall panels behaving as partly fixed panels with semi-rigid
screw connection (Combined Single-Coupled wall behaviour).
iii) Coupled wall panels behaving as a single wall panel with rigid screw
connection (Single wall behaviour).
In the first case, the vertical joint between wall panels is relatively weak in
comparison to the anchoring connections, thus providing low level of stiffness
between individual wall panels. While being loaded with lateral forces, connected
panels behave as individual panels, rocking around each individual lower corner
(Figure 6a). Conversely, if the vertical connection between coupled wall panels is
very stiff, the behaviour of coupled walls is to the same as the behaviour of a
single wall panel, as shown in Figure 6c. In this case, the vertical connection has
higher resistance than shear forces between wall panels, and is very stiff. The third
possibility is an intermediate, combined behaviour between individual wall
behaviour and fully connected walls behaviour. As vertical connections between
coupled wall panels are semi-rigid, slight deformations (slip) of the vertical
connection can take place (Figure 6b).
(a) (b) (c)
Fig. 6 Types of behaviour of adjacent wall panels: (a) Coupled wall behaviour; (b)
Combined Single-Coupled wall behaviour; (c) Single wall behaviour
Thus, special attention in coupled walls design should be given to the vertical
connection between adjacent panels. By over-sizing the vertical connection may
result in a completely different behaviour of the coupled wall panels. For example,
if the stiffness of the vertical screwed connection is very high because large
diameter screws are used at a small spacing, the behaviour of adjacent wall panels
will become similar to that of a single wall panel. Thus, special care should be
given when designing the vertical joint to achieve the desired wall behaviour. The
''Single wall behaviour'' results in higher strength capacity whereas the ''Coupled
wall behaviour'' has lower elastic stiffness but attains larger ultimate
displacements, which are also very important in earthquake design. Hence, the
type of behaviour of wall subassemblies should be always decided a priori and
then implemented into a proper design of the vertical joints. An example can be
given by comparing wall tests 1.2, 2.1 and 2.3 (Gavric, 2013). Wall test 1.2 was a
single wall (Figure 2, Configuration S-II), while walls 2.1 and 2.3 were coupled
walls with screwed half-lap joint (Figure 2, Configuration S-III). As there were 20
screws connecting adjacent panels in case of Wall 2.1, the coupled panels behaved
virtually as a single panel, exhibiting a very similar behaviour in terms of
mechanical properties and energy dissipation capacity as wall panel 1.2. Wall 2.3
had only 50% screws in the vertical joint compared to Wall 2.1. This resulted in
the so called ''coupled wall behaviour'', as both panels were rocking separately and
not any more as one single panel. Here, the total dissipated energy was 35%-40%
lower than in Walls 1.2 and 2.1. However, Wall panel 2.3 exhibited 33% higher
ultimate displacement in comparison to single wall panel 1.2, showing higher
level of flexibility. The reasons for the higher energy dissipation capacity of
panels with ''single wall panel behaviour'' in comparison to wall panels with
''coupled wall behaviour'' up to a certain level of top horizontal displacement
(interstory drift) is due to the aspect ratio of the panels.
Similarly, coupled wall CA-SN-11 (Figure 3 Configuration III) (Popovski et
al., 2010) exhibited 43% lower energy dissipation in comparison with single wall
CA-SN-03 (Figure 3 Configuration I), both having equal connection layout.
Coupled wall CB-SN-16 with three adjacent panels (Figure 3 Configuration IX)
had 37% lower total dissipated energy in comparison with a single long wall (3.45
m 2.3 m) CB-SN-14 (Figure 3 Configuration VIII). The difference in
unexpected higher level of energy dissipation is attributable to the aspect ratio of
the panels; long single wall had aspect ratio 1.5:1, while each of the three adjacent
panels in the long coupled wall had dimensions 1.15 m 2.30 m (aspect ratio 1:2).
The long single wall had three times higher elastic stiffness and 46% higher lateral
load resistance Fmax, but the long coupled wall exhibited 59% higher ultimate
displacement (4.66% vs. 2.94% interstory drift). On the other hand, the single wall
(CB-SN-14, Figure 3 Configuration VII) had a ductility ratio 26% higher than in
the case of the coupled wall (CB-SN-16, Figure 3 Configuration IX), even if the
ultimate displacement was significantly lower. The reason for that is the almost
twice larger yielding displacement in case of the coupled wall (18.3 mm vs 9.5
mm) compared to the single wall. In terms of global behaviour, the failure
mechanism was similar in both walls, namely: exceeded vertical uplifts in the
corners of walls, while the horizontal displacements were relatively small.
A practical design rule is that all brackets in each individual wall panel should
be placed symmetrically with respect to both panel edges. This provides
symmetrical wall behaviour in both directions, which is needed as horizontal
earthquake and wind loads alternate their direction.
4 Analysis of CLT Walls Behaviour under Cyclic Loads
Each CLT wall behaves under cyclic loading as a combination of the two
classifications introduced in the previous section. How do these different
combinations affect the walls performance under cyclic loads? The answer is
provided in the following sections.
4.1 Hysteresis Loops and Failure Mechanisms

Figure 7 displays the typical force-displacement hysteresis loops of CLT walls
with different predominant types of behaviour.
Wall 4.1: Force - Displacement Wall 1.2: Force - Displacement Wall 1.1: Force - Displacement
100 150 100
80 80
100
60 60
40
50 40
Force [kN]
Force [kN]
20
Force [kN]
20
0 0
-100 -80 -60 -40 -20 0 20 40 60 80 100 -80 -60 -40 -20 0 20 40 60 80 0
-20 -60 -40 -20 0 20 40 60
-50 -20
-40
-40
-60
-100
-60
-80
-100 -150 -80

Displacement [mm] Displacement [mm] Displacement [mm]
(a) (b) (c)
Fig. 7 Typical force-displacement hysteresis loops: (a) Rocking behaviour; (b) Combined
Rocking-Sliding behaviour; (c) Sliding behaviour
In the first case (rocking mechanism), the CLT wall panel under horizontal
cyclic loading failed due to exceeded uplifts as a consequence of rocking of the
panel over the lower corners of the panel and crushing of the wood under the
compressed corner. On average, the compression deformation of the lower corners
of tested CLT panel was 2.6 mm. This value was higher for the walls with high
height-to-width ratio (tall walls) and for the walls with higher vertical loads. Metal
connectors (hold-downs, brackets) progressively yield, starting from the outer
connectors, and progressing to the brackets installed in the central part of the wall.
While in vertical (tension) direction metal connectors yielded and eventually
failed, no significant deformation of connectors were observed in horizontal
direction, thus no significant panel sliding occurred (Figure 7a). In cases of
combined rocking-sliding mechanisms, the metal connectors failed due to
combined shear-uplift displacements, thus yielding of the connectors occurred due
to exceeded shear forces and the overturning moment (Figure 7b). In the last case,
the sliding mechanism, the CLT wall failed due to exceeded shear resistance.
Shear forces are concentrated in the angle bracket connectors, which consequently
failed in shear (Figure 7c). Hold-downs do not have significant shear stiffness and
shear resistance and typically fail due to buckling of the steel part when subjected
to high shear forces.
Walls with rocking behaviour have the so called self-centering ability after
being subjected to horizontal loads, due to the effect of vertical loads and the axial
resistance of metal connectors. This means that after a seismic event, this type of
wall system returns to the initial vertical position without any significant residual
displacements. This kind of behaviour is very desirable in terms of seismic
performance of lateral load resisting systems as even for high level of damages in
the connectors, the CLT building can snap back to the initial vertical position. The
only residual damage will be localized in the nailed connections between the steel
brackets/hold-downs and CLT panel, where some timber crushing at the CLT
panel-fastener connection will occur together with the plasticization of some
fasteners. However, there will be little or no permanent deformation of the
building, as the shaking table tests have clearly demonstrated (Ceccotti 2008). The
damaged components can be easily replaced at the end of an earthquake, in this
way minimizing the disruption caused by the seismic event, and enabling a prompt
reuse of the building.
On the other hand, combined rocking-sliding behaviour has some non-
reversible deformations after cyclic loading, while sliding mechanisms are not
able to return by themself to the original position. Full scale building tests
(Ceccotti 2008) showed that sliding of walls resulted in withdrawal of fasteners in
walls perpendicular to them. That caused a reduction of strength and stiffness of
the wall system in the opposite direction, thus reducing the overall strength and
stiffness capacity of the entire building.
A suggestion is therefore given that at the wall level, plasticization should
preferably occur in the hold-downs and angle brackets loaded in tension, whereas
the angle bracket should ideally remain elastic in shear so that there is no residual
slip in the wall at the end of the seismic event. This condition means that the angle
brackets should be overdesigned with respect to the hold-downs. To achieve this
result, capacity based design should be applied, in order to ensure that ductile
modes of failure should precede the brittle modes of failure with sufficient
reliability.
4.2 Mechanical Properties

In terms of strength and stiffness capacity of CLT walls, the vertical load has
beneficial impact. For example, Wall 3.6 (Figure 2 Configuration C-III) (Gavric,
2013) without any vertical load exhibited 45% lower lateral resistance and 24%
lower initial stiffness in comparison with Wall 3.2 (Figure 2 Configuration C-
III...) with 18.5kN/m vertical load. Similarly, wall CA-SN-00 (Figure 3
Configuration I) (Popovski et al., 2010) without any vertical load had 10% lower
strength and 28% lower initial stiffness in comparison to wall CA-SN-03 (Figure 3
Configuration I) with 20 kN/m vertical load.
Higher elastic stiffness and strength of CLT walls can be achieved
by increasing the number of metal connectors (wall CA-SN-20 vs CA-SN-03,
Figure 3 Configuration IV), where 75% increased number of brackets resulted in

35% higher elastic stiffness and 55% higher strength. Moreover, presence of hold-
downs in CLT walls increases the initial stiffness (80% higher kel in case of wall
CA-SNH-08A compared to CA-SN-03 without hold-downs, Figure 3
Configuration II).
As discussed in the previous section, coupled walls usually have lower elastic
stiffness and strength capacity in comparison to single walls, while their ultimate
displacement capacity is higher. Walls with aspect ratio b/h < 1 (tall walls), result
in lower initial stiffness but higher displacement capacity in comparison with
walls with ratio b/h > 1 (long walls), if the connection layout in both cases has the
same distribution (Popovski et al., 2010).
CLT walls anchored to CLT floor panels have lower lateral stiffness and
resistance capacity in comparison with CLT walls which are anchored to the
foundation. This is mostly due to the larger flexibility of the CLT floor panel
compared to a reinforced concrete foundation, and to the reduced load-carrying
capacity of fasteners used in timber (usually self-threaded screws) compared to
fasteners used for concrete (usually bolts). Thus, CLT wall-CLT floor panel
connections are in general weaker in comparison with CLT-foundation
connections (Ceccotti et al., 2006, Popovski et al., 2010).
Higher ductility ratios did not always correspond to better performance of CLT
walls in terms of being able to dissipate energy in non-linear range with larger
ultimate displacement capacity. For example, walls CS-WT-22 (Figure 3
Configuration VI) and CS-WT-22B (Figure 3 Configuration VI) had relatively
high ductility ratios (7.54 and 4.97, respectively) in comparison with walls which
were able to dissipate larger amount of energy and underwent larger interstory
drifts, like for example wall CA-SN-20 (Figure 3 Configuration IV) (ductility ratio
3.65) with almost three times larger interstory drift capacity and five times more
dissipated energy. A study of parameters which could possibly influence drift
capacity of CLT wall was done; relatively good correlation between absolute
values of rocking deformations and interstory drifts was found (R2 = 0.91).
Rocking deformations are directly related with CLT wall uplifts, which are usually
measured during CLT wall experimental tests. Thus, instead of relative ductility
definition (EN12512, 2001), emphasis should be given to total displacement
capacity of CLT walls (Jorissen & Fragiacomo, 2011), as this quantity better
represents the capacity of CLT walls to undergo large displacements during
seismic events.
4.3 Energy Dissipation Capacity

In the previous sections, differences in total dissipated energy among different
wall types were presented. In addition, equivalent viscous damping ratios were
calculated for all walls (N = 49) at the displacement rate where the maximum
force was attained.
For walls with single wall panel behaviour type with predominant rocking
behaviour, the average hysteretic damping value was 14.50% (COV = 9.32%) for
the 1st cycles and 9.86% for the 3rd cycles (COV = 13.97%), N = 25. For walls
with coupled wall panel behaviour, the damping value was 14.25% (COV =
8.24%) for the 1st cycles and 9.76% for the 3rd cycles (COV = 9.90%), N = 14. No
significant differences in hysteretic damping ratios were observed for walls with
single wall panel behaviour with predominant rocking mechanism and walls with
coupled wall panel behaviour, even though significant differences in total
dissipated energy were calculated (Section 3.2).
Walls with predominant sliding behaviour exhibited higher equivalent viscous
damping values, namely on average 17.50% for the 1st cycles (COV = 7.57%) and
14.71% for the 3rd cycles (COV = 13.29%), N = 6. The higher values of hysteretic
damping ratios in this case are due to the additional friction between CLT wall and
floor panels, as the wall behaviour is predominantly sliding.
Tall walls had the lowest damping values; on average 10.02% for the 1st cycles
(COV = 16.48%) and 5.55% for the 3rd cycles (COV = 14.72%), N = 5.
5 Conclusions
The study in this paper focused on the determination of different types of CLT
walls behaviour, and how these types affect properties such as strength and
stiffness capacity, displacement capacity (drift capacity), ductility ratios, energy
dissipation capacity and hysteretic damping values, which are all important in
seismic design of CLT systems. In addition, observations on the failure modes of
connections provided an insight on how a proper design of typical CLT
connections should be carried out. Different type of global panel behaviour
resulted in different level of energy dissipation capacity. CLT wall panels with
single wall panel behaviour performed better than walls with coupled wall
behaviour in terms of total dissipated energy until a certain level of interstory drift.
On the other hand, panels with coupled wall behaviour can exhibit higher ultimate
displacements. Special attention should be given to the design of vertical joints
between adjacent wall panels, as the global behaviour of wall panels can change
dramatically and, consequently, the wall performance in terms of mechanical
properties, energy dissipation capacity and displacement capacity can be
significantly different.
For classification of CLT wall panels in terms of performance under cyclic
loads, instead of relative ductility ratio definition, different quantities should be
used. CLT wall properties such as interstory drift capacity and energy dissipation
capacity can be evaluated more precisely using: (i) CLT wall uplifts, as a
consequence of wall rocking; and (ii) types of wall global behaviour, namely:
rocking behaviour vs sliding behaviour, and single wall panel behaviour vs
coupled wall panel behaviour.
References
Ceccotti, A., Lauriola, M., Pinna, M., Sandhaas, C.: SOFIE Project - Cyclic Tests on Cross-
Laminated Wooden Panels. In: Proceedings of the 9th World Conference on Timber
Engineering, Portland, Oregon, USA (2006)
Ceccotti, A.: New technologies for Construction of Medium-Rise buildings in Seismic
Regions: The XLAM case. IABSE Structural Engineering International 18(2), 156165
(2008)
Dujic, B., Pucelj, J., Zarnic, R.: Testing of Racking Behavior of Massive Wooden Wall
Panels. In: Proceedings of the 37th CIB-W18 Meeting, paper 37-15-2, Edinburgh,
Scotland (2004)
Dujic, B., Zarnic, R.: Report on evaluation of racking strength of KLH system. University
of Ljubljana, Faculty of civil and geodetical engineering, Slovenia (2005)
Dujic, B., Aicher, S., Zarnic, R.: Testing of Wooden Wall Panels Applying Realistic
Boundary Conditions. In: Proceedings of the 9th World Conference on Timber
Engineering, Portland, Oregon, USA (2006a)
Dujic, B., Klobcar, S., Zarnic, R.: Influence of Openings on Shear Capacity of Massive
Cross-Laminated Wooden Walls. In: COST E29 International Workshop on Earthquake
Engineering on Timber Structures, Coimbra, Portugal, pp. 105118 (2006b)
Dujic, B., Klobcar, S., Zarnic, R.: Influence of Openings on Shear Capacity of Wooden
Walls. In: Proceedings of the 40th CIB-W18 Meeting, paper 40-15-6, Bled, Slovenia
(2007)
Dujic, B., Klobcar, S., Zarnic, R.: Shear Capacity of Cross-Laminated Wooden Walls. In:
Proceedings of the 10th World Conference on Timber Engineering, Miyazaki, Japan
(2008)
EN 12512. Timber structures - Test methods - Cyclic testing of joints made with
mechanical fasteners. CEN, Brussels, Belgium (2001)
Gavric, I., Ceccotti, A., Fragiacomo, M.: Experimental cyclic tests on cross-laminated
timber panels and typical connections. In: Proceedings of the 14th ANIDIS Conference,
Bari, Italy (2011)
Gavric, I.: Seismic Behaviour of Cross-Laminated Timber Buildings. Ph.D. Thesis,
University of Trieste, Italy (2013), http://hdl.handle.net/10077/8638
Jorissen, A., Fragiacomo, M.: General notes on ductility in timber structures. Engineering
Structures, Special Issue on Timber Structures 33(11), 29872997 (2011)
Popovski, M., Schneider, J., Schweinsteiger, M.: Lateral load resistance of cross-laminated
wood panels. In: Proceedings of the 11th World Conference on Timber Engineering,
Riva del Garda, Italy (2010)
CLT Plates under Concentrated Loading
Experimental Identification of Crack Modes
and Corresponding Failure Mechanisms
Georg Hochreiner1, Josef Fussl1 , Josef Eberhardsteiner1, and Simon Aicher2

1 IMWS - Technical University of Vienna, 1040 Vienna, Austria
georg.hochreiner@tuwien.ac.at
2 University of Stuttgart, MPA,
Otto-Graf-Institute and Department of Construction Materials, 70569 Stuttgart, Germany
simon.aicher@mpa.uni-stuttgart.de
Abstract. Since wood products for structural elements, such as cross-laminated tim-
ber (CLT), have gained importance in the building sector, the need for appropriate
and reliable design codes has become essential. For this reason, this work focuses on
global failure mechanisms and the corresponding evolution of different crack modes
in CLT plates, depending on geometric and/or material related influence quantities.
Therefore, plate-bending experiments on 3- and 5-layered CLT-plates were carried
out. In addition to standard evaluation methods, each specimen were cut into cubes
(10/10/10 cm) to get information about the failure modes inside the plates. Areas
and location of dominant shear failure, tensile failure, delamination, and mixed fail-
ure modes could be clearly defined and connected to geometry and loading situation.
Based on this evaluation well known but not yet in detail described effects, such as
the ductile structural behavior of CLT plates, can be explained.
Keywords: cross laminated timber, concentrated loading, evolution of crack modes,

global failure mechanism, mixed failure modes, ductile structural behaviour.
1 Introduction
In recent years, cross-laminated timber (CLT) plates have gained importance in the
building sector, because modern building systems, only consisting of horizontal
plates and vertical columns connected to some bracing core-structure, have been
transferred to wooden structures as well. Despite their application as 2D load-
carrying structural elements, standard testing procedures for the derivation of design
parameters are still based on 1D loading schemes, like 3-point and 4-point bend-
ing tests ([12]), respectively. Thereby, important stress transfer mechanisms typical
for 2D load-bearing systems cannot be captured. Especially global shear deforma-
tions resulting from middle layers with quite low shear stiffness, and global shear
failure mechanisms due to concentrated loads cannot be estimated appropriately
with 1D test arrangements. Compared to concrete slabs, where standard verification
DOI: 10.1007/978-94-007-7811-5_63, c RILEM 2014
704 G. Hochreiner et al.
procedures for punching failure are available for a long time and the load carrying
capabilities beyond the pure elastic regime are well investigated, there is a lack of
knowledge concerning CLT plates.
Basic experimental investigations regarding the influence of orthotropy on the
rolling shear stiffness of CLT plates can be found in [2] and [5], respectively. Re-
sults of traditional testing procedures according to standards are presented in [3]
and [6]. Moreover, the impact of internal stress states induced by a change of mois-
ture content on the load carrying behaviour is analyzed in [9] and [13]. A numerical
model for CLT plates under concentrated loading and a comparison to experimental
results is proposed in [8], and aspects concerning local shear reinforcements are dis-
cussed in [15]. For all investigations, the assessment of failure is limited to surface
information of the specimens and the main focuse is laid on the mechanical behavior
up to the elastic limit.
However, to allow for the development of more accurate and reliable design
procedures, knowledge about the failure mechanisms (type and mode of cracks,
crack propagation and formation) within the plates is necessary. Also with regard
to the development and validation of more sophisticated numerical simulation tools
for structural elements. Investigation and assessment of crack formation has been
proven to be an appropriate method for the derivation of corresponding design pro-
cedures ([1]) for diverse and even more complicated failure modes.
For this reason, this work focuses on a detailed evaluation of plate-bending tests
on CLT elements. In the following, Chapter 2, the experimental setup and the basic
evaluation strategy is described. In Chapter 3 the global load deformation behavior
as well as the recorded failure modes within the plates are illustrated and discussed.
Finally, concluding remarks are given in Chapter 4.
2 Experiments
Tests were carried out on 3- and 5-layered CLT plates with dimensions 1 x 1 and 2
x 2 m2 , and layer thicknesses of 34/28/34 mm and 5 x 19 mm, respectively. The raw
material was Norway spruce (strength class C24) and PUR adhesive was used for
glueing. The mechanical testing of the plates was carried out at MPA Stuttgart fol-
lowing in general the test concept in [8]. The plates were vertically supported along
all four edges (see Fig. 1). The load was applied deformation controlled, with a
loading-rate between 3 to 6 mm/min within the elastic regime and 15 to 20 mm/min
during extensive crack formation, by means of a rigid metal punch without any flex-
ible interlayer to the wooden surface. In total, ten plates with different dimensions
and loading conditions were tested.
Traditional assessment of the cracking behaviour is usually based on surface
views and a few cuts through the specimen at promising locations. But during the
loading process acoustic signals often imply cracking processes before they are vis-
ible from the outside. For this reason, a systematic crack assessment is only pos-
sible by slicing the plates into strips and cubes. A grid size with an interval of
about 100 mm has been proved to be appropriate for a reliable reconstruction of the
Failure Mechanisms of CLT Plates 705
Fig. 1 Test setup at MPA Stuttgart
cracking mechanism and, at the same time, produces a reasonable amount of work
for machining and rearrangement of the cubes. Each single cube were inspected
from every side to evaluate the extent and mode of cracking over the whole volume.

In this section, results from three different sources are presented: (i) from selected
point measurements, (ii) from videos including acoustic signals, and (iii) from the
inspection of the wooden cubes, giving the global crack formation.
With respect to the latter, on the wooden board level the following failure modes
will be distinguished (see Fig. 2):
Fig. 2 Classification of crack modes within CLT plates
L1: Tension failure in L-direction, where the fracture surfaces within clear wood
proceed perpendicular to the grain and are rather compact, with a maximum
length of torn out fibres of 10 mm. Usually L1 failure is initiated at growth ir-
regularities like knots much earlier.
L2: Compression failure in L-direction, which leads to the typical kink-banding
failure mode, well known for fibre composites, due to local instabilities of the
wood cells.
I: Indentation failure perpendicular to the grain, which is characterized by ex-

treme ductile plastic behavior.
R: Tension failure in R-direction, mainly occuring in the early wood cells, which
exhibit a larger scatter of strength values despite of their higher stiffness in com-
parison to the T-direction. This type of failure may be born from either shear
stress parallel to the grain or so-called rolling shear stress perpendicular to the
grain. The corresponding fracture surfaces are usually located in the interior of
the board, following as long as possible the annual rings.
T1: Tension failure in T-direction, occuring in both early- and late wood. This
type of failure either results from so-called rolling shear stress or embarrassed
shrinkage. The fracture surfaces is usually compact.
T2: Plane tension failure in T-direction, occuring in the transition zone to neigh-
boring boards and characterized by a nearly plane crack path, but a staggered
structure of the crack surface. The reason for this special type of failure mode
may be the superposition of global rolling shear stress by either bending stress
perpendicular to the grain or stress perpendicular to the single layer.
In general, failure modes L1 and T2 together lead to global failure and are respon-
sible for the structural failure mode, while modes R and T1 rather remain local and
hardly influence the effective plate stiffness and strength.
Moreover, the following failure modes between boards may appear (see Fig. 2):
EG: Failure of edge-glueing, and
IF: Interface failure between single layers.
For both plate structures (3 and 5 layers) and for all specimens the quasi elastic
limit was at the same level, at about 130 kN (see Fig. 3). With the exception of plates
PK3 and EL2, all specimens exhibit quasi ductile load carrying behaviour, with even
hardening in case of plate EL4.
On the basis of the recorded videos from the upper and lower side of each plate,
the corresponding acoustic signal and the global load deformation behaviour, a tab-
ular failure protocol could be compiled for each test, in which the cracking modes
according to the aforementioned categorization are assigend to global load defor-
mation characteristics. Alternatively, an assignment of fracture modes to the global
load deformation curves (LDC) and the intensity of acoustic signals during the fail-
ure process is illustrated in Fig. 4.
Fig. 3 Load deformation curves of 3-layered (left) and 5-layered (rigth) CLT plates
Fig. 4 Illustrative load deformation curve with acoustic protocol and assigned crack modes
to the different failure stages
In the following, the probably occuring effects and/or cracks in each stage, de-
fined in Fig. 4, are described:
Stage 1, elastic regime: Within the elastic regime there are no visual as well
as acoustic signs which would indicate any kind of damage or crack initiation. As
expected, the average effective system stiffness of the 5-layered plates is higher than
that of the 3-layered plates, despite of the same overall thickness. The reason is the
higher transverse load bearing capacity of the 5-layered plates, where 3 x 19 mm
effective thickness can be activated instead of 1 x 34 mm as for the 3-layered plates.
Stage 2, crack initiation: In this stage cracking starts, which is clearly indicated
by acoustic signals, but the different cracks remain local and therefore the effective
plate stiffness is hardly influenced. Tension failure in R-direction (R) due to global
rolling shear stress is assumed to be the first occuring fracture mode, because of the
smaller tensile strength of early-wood in R-direction compared to the T-direction.
Typically, the cracks run along annual rings and possibly jump to neighboring rings
with advancing crack growth, but clearly being limited to the area where tension
stress in R-direction exists. The resulting global failure pattern consists of in regular
intervals appearing inclined cracks, whereas the inclination angle is between 40
to 60 to the horizontal, depending on the local configuration of the annual rings.
The distance between cracks depends on the width of the individual lamella. These
crack formation strongly reminds of the regular shear cracks in reinforced concrete
structures, and also the development of a truss-like structural system is obvious
(see Fig. 5, left). Figures 6 and 7 show the regular distribution of R-cracks for two
different loading directions, but the same annual ring configurations of the boards
of the middle layer. It can be seen that the crack location is clearly influenced by the
course of the annual rings, but the global crack pattern is basically independent of
the location of the pith. Tension failure in tangential direction (T1) seems to be the
subsequent cracking mode (see Fig. 5, right), also induced by rolling shear stress,
mainly occuring in high loaded areas in the surrounding of the load application.
Furthermore, partial tension failure of the edge glueing (EG) was also observed at
Fig. 5 Truss-like structural

system at Stage 2 (left), and
further cracking modes in
Stage 3 (right)
Fig. 6 Regular pattern of R-cracks in mid- Fig. 7 Regular pattern of R-cracks in mid-
dle layer for concave annual ring course dle layer for convex annual ring course
Stage 2, usually affecting torsional plate stiffness. Nevertheless, since the areas of
high bending stress and dominating twisting stress are hardly overlapping, global
stiffness is not really affected.
Stage 3, controlled softening of the system: At this stage horizontal crack planes
are forming (T2) (see Fig. 5, rigth), and the truss-like structural system, which pre-
served the initial level of global stiffness, is partially going to be destroyed. The
shear stress transfer between bottom and top layer is now strongly reduced and
the structural stiffness decreases. The cracked T2 surface is quite even in the area
of the former compression struts (shear failure) and shows a staggered/rough face
under the former tension struts (tensile failure; see Fig. 5). It can be assumed that the
effective stiffness of the resulting structural system is also dependent on the friction
parameters between these cracked surfaces.
Stage 4, stable ductility of the system: At the beginning of this stage propagating
T2-failure surfaces, as described in Stage 3, freeze, and tension band systems are
activated, which may lead to hardening effects of the global system. Stage 3 and
Stage 4 may occur repeatedly in succession (see EL6 and EL3 in Fig. 3), whereas
the global stiffness decreases for each newly formed system. This cycle is stable as
long as T2-failure does not expand to the boundaries of the plate.
Stage 5, unstable softening of the System: This stage is characterized by an in-
crease of zones with T2-cracks and additional arising interface failure surfaces be-
tween boards (IF). Together with tensile failure in longitudinal direction (L1), often
initiated at growth irregularities, it comes to abrupt stress transfers and resulting
therefrom unstable softening behavior, which is reflected by signal peaks in the
acoustic protocol. While from the LDCs (see Fig. 8, left) T2-failure can not be dis-
tinguished from L1-failure, the abrupt failure characteristic of the latter is clearly
visible in load time relationships (see Fig. 8, right).
Fig. 8 Identification of softening cracking modes (T2 or L1) by means of load time relation-
sships
L1-failure usually indicates that the ductile potential of the system is depleted
and the ultimate load is almost reached. Interestingly, L1-failure was not only be
detected at the tension zone (bottom layer of the plate) but also within the inner
layers and at the top layer, being characterized by dominating compression stress
at the stages 1 and 2. This confirmed the aforementioned activation of tension band
systems, in which tension forces dominate in all layers.
Stage 6, final failure Finally, tension band systems are predominant and as a
result global failure is mainly characterized by L1-cracks distributed over all layers.
If tension band systems cannot develop, e.g., due to a low span to height ratio,
global failure is characterized by cracking of the middle layers (R, T1, T2), which
could be summarized as rolling shear failure. An additional failure mechanism could
be observed at 3-layered plates, where the top layer was locally destroyed by the
loading stamp (I), and subsequently a notched system arises, as illustrated in Fig. 9.
In that case, global failure is superimposed by IF-failure around the load application
area.
Fig. 9 Interface failure between the layers L1 and Q2 due to loading a notched system
In the following, selected failure patterns of 3- and 5-layered plates are shown
and interpreted.
Fig. 10 shows the crack formation of a 3-layered plate with a slenderness (de-
scribing the span to height ratio) of 10. Global failure is characterized by rolling
shear failure modes which spread to one side up to the edge, and interestingly no
L1-cracks appear, suggesting that no tension band systems evolve. The situation is
Fig. 10 Crack formation within a 3-layered CLT plate with a length to heigth ratio of 10
different for 3-layered plates with a slenderness of 20, where final failure was al-
ways indicated by L1-cracks. But, depending on the tensile strength of the lamellas
of the bottom layer, different sequences of cracking can occur. At best, rolling shear
failure spreads from the center to the edge of the plate (as can be seen in Fig. 11,
left) and only then L1-cracks appear and lead to structural failure. In this case, the
load carrying capacity of the plate is perfectly exploited, which is indicated by a
pronounced ductile behaviour or even hardening effects, apparent in the LDC (see
Fig. 3). Fig. 11 (right) shows the opposite case, where L1-cracks in the bottom
Fig. 11 Crack formation within 3-layered CLT plates, (left) dominating rolling-shear failure
modes and (right) dominating L1-cracks
layer (vertical bold bars) occure before rolling shear failure modes can evolve in the
affected boards. The corresponding LDC for EL2 can be found in Fig. 3, and is char-
acterized by a very short plastic plateau. For all 3-layered plates, with a slenderness
of 20, is valid that tension band systems are activated during the failure process and
finally all layers are under tensile load. That is the reason why tensile cracks (L1)
could also be found at the top layers (horizontal bold bars in Fig. 11). The recorded
failure patterns for the other 3-layered plates exhibit a crack formation which can
be classified between the two shown in Fig. 11.
Compared to the 3-layered plates, 5-layered plates exhibit a much more localized
and compact failure pattern, as can be seen in Fig. 12. Global failure due to L1-
cracks, distributed over all layers, occurs before rolling shear cracks could propagate
to the edges of the plate. A higher tensile strength of the raw material (lamellas)
might lead to a better utilization of the plastic potential of the plate.
Fig. 12 Crack formation within a 5-layered CLT plate
4 Conclusions
In this paper, bending tests of 3- and 5-layered CLT plates were illustrated. Thereby,
not only the global load deformation behaviour was analyzed but also the occuring
failure modes within the plates, based on acoustic protocols, videos and a detailed
recording of the final failure mechanism. Based on the obtained results, the follow-
ing conclusions can be drawn:
At the beginning of failure a truss-like structural system develops within CLT
plates, which is influenced by the annual ring configuration of the cross layers
and the width of the boards.
To exploit the ductile potential of a CLT plate the tensile strength of the lamel-
las in longitudinal direction must be sufficiently high. If this is the case, a pro-
nounced ductile behavior and even structural hardening effects could be ob-
served, which are not taken into account in design concepts for CLT plates yet.
During the failure process different load-bearing systems, e.g., truss-like systems
and tension band effects, develop within CLT plates. In general, detailed knowl-
edge about these systems will help to develop as well as validate appropriate
numerical simulation tools and derive new design concepts for punching failure
and reinforcements, respectively.
Acknowledgements. The authors gratefully acknowledge the financial support of the Aus-
trian Research Promotion Agency (FFG, project number 832803 and 839858) and the wood
industry partner CEI-Bois (through the platform Building with Wood) for funding the re-
search work within project MechWood 2.
References
1. Leonhardt, F.: Vorlesungen uber Massivbau. Vierter Tell: Nachweis der Ge-
brauchsfahtgkeit, Rissebeschrankung, Formanderungen, Momentenumlagerung und
Bruchlinientheorie im Stahlbetonbau. Springer, Heidelberg (1976 + 1978)
2. Aicher, S., Dill-Langer, G.: Basic considerations to rolling shear modulus in wooden
boards. Otto-Graf-Journal 11 (2000)
3. Jakobs, A.: Zur Berechnung von Brettlagenholz mit starrem und nachgiebigem Ver-
bund unter plattenartiger Belastung mit besonderer Berucksichtigung des Rollschubes
und der Drillweichheit. Dissertation; Universitat der Bundeswehr Munchen, Fakultat fur
Bauingenieur- und Vermessungswesen, Professur Konstruktive Gestaltung und Holzbau
(2005)
4. Schickhofer, G., et al.: Brettsperrholz - Ein Blick auf Forschung und Entwicklung,
Tagungsband der 5. Grazer Holzbau-Fachtagung
5. Schwar, A.: Das Tragverhalten von beanspruchten Brettsperrholzplatten unter dem Ein-
flu streuender Jahrringorientierungen der Querlagen. Bautechnik 83, Heft 1 (2006),
doi:10.1002/bate.200610008
6. Tobisch, S.: Methoden zur Beeinflussung ausgewahlter Eigenschaften von dreilagigen
Massivholzplatten aus Nadelholz. Dissertation Universitat Hamburg, Fachbereich Biolo-
gie
7. Gsell1, D., Feltrin, G., Schubert, S., Steiger, R., Motavalli, M.: M.ASCE5; Cross-
Laminated Timber Plates: Evaluation and Verification of Homogenized Elastic Prop-
erties. Journal of Structural Engineering (January 2007), doi:10.1061/(ASCE)0733-
9445(2007) 133:1(132)
8. Czaderski, C., Steiger, R., Howald, M., Olia, S., Gulzow, A., Niemz, P.: Versuche
und Berechnung an allseitig gelagerten 3-schichtigen Brettsperrholzplatten. Holz Roh
Werkst 65, 383402 (2007)
9. Gereke, T.: Moisture-Induced Stresses In Cross-Laminated Wood Panels. Dissertation
ETH Zurich (2009)
10. Frangi, A., Knobloch, M., Fontana, M., Bochicchio, G.: Fire behaviour of cross-
laminated solid timber panels. Fire Safety Science 9, 12791290 (2009)
11. Schickhofer, G., Bogensberger, T., Moosbrugger, T.: BSPhandbuch - Nachweise auf Ba-
sis des neuen europaischen Normenkonzeptes (2009) ISBN 978-3-85125-076-3
12. Steiger, R., Gulzow, A.: Validity of bending tests on strip-shaped specimens to derive
bending strength and stiffness properties of cross-laminated solid timber (X-lam). In:
The Future of Quality Control for Wood and Wood Products, Edinburgh, The Final Con-
ference of COST Action E53, May 4-7 (2010)
13. Gulzow, A., Richter, K., Steiger, R.: Influence of wood moisture content on bending and
shear stiffness of cross laminated timber panels. Eur. J. Wood Prod. 69, 193197 (2011),
doi:10.1007/s00107-010-0416-z
14. Sturzenbecher, R., Hofstetter, K., Eberhardsteiner, J.: Hohere Plattentheorien fur die Be-
messung von Brettsperrholz, einem mehrschichtigen schubnachgiebigen Plattenwerk-
stoff. Baustatik-Baupraxis Universitat Innsbruck/TU Graz (2011)
15. Mestek, P.: Punktgestutzte Flachentragwerke aus Brettsperrholz (BSP) Schubbemes-
sung unter Berucksichtigung von Schubverstarkungen. Dissertation Technische Univer-
sitat Munchen (2011)
Seismic Strengthening of Existing Concrete
and Masonry Buildings with Crosslam Timber
Panels
I. Sustersic and B. Dujic
CBD d.o.o. Contemporary Building Design, Celje, Slovenia

iztok@cbd.si
Abstract. The paper deals with the possibility of using crosslam timber panels for
strengthening existing buildings against seismic forces. Three types of buildings
are considered unreinforced masonry (URM), reinforced concrete frames and
reinforced concrete frames with masonry infill. Results from an extensive testing
series are presented, namely quasi-static cyclic test on URM and shaking table
tests of reinforced concrete frames with and without infill.
Keywords: Cross-laminated timber, Concrete, Unreinforced masonry, Seismic

strengthening.
1 Introduction
A majority of the buildings on seismically active areas built before the nineteen
eighties have two major problems; first they are seismically unsafe because of the
lack of (sufficient) seismic design codes at the time they were built. And second,
they have a high energy consumption because of lack of insulation, proper details
etc. The proposed retrofit system could deal with both problems at once; the new
outer cross laminated timber walls stabilise a building against horizontal shear
forces that are caused by earthquakes. On one hand the timber panels have a low
mass and therefore dont contribute much to seismic forces. However they are
very stiff on the other hand (Dujic, 2005) and provide high shear resistance. In
addition the new outer wall if combined with an effective insulation provides a
very good thermal insulation of the building. A 170 mm outer shell (including a
faade) could provide sufficient building thermal insulation and strengthen a rea-
sonably sized building against earthquakes. The new outer shell could have win-
dows, doors and a faade installed already in the manufacturing plant. Another
positive aspect of the outer shell is that there are no harsh interventions to a build-
ing and that people dont have to move out during the construction phase (unlike
when using most of the conventional methods for seismic strengthening). On the
other hand the installation of panels is also possible from the inside with little
DOI: 10.1007/978-94-007-7811-5_64, RILEM 2014
714 I. Sustersic and B. Dujic
influence on the existing structure. All together it makes a unique system that
could solve two major problems in older buildings on seismically active areas and
therefore prolong the lifespan of constructions, contributing to sustainability.
In the following chapters the system is presented more in detail. The back-
ground of seismic analysis and the performance based design N2 method (Fajfar,
2000) used for the evaluation of the seismic resistance of the case structure are
discussed. And the system connection concept development along with the results
from an experimental study of structures strengthened with crosslam panels;
namely URM, RC frame and RC frame with masonry infill tested either with
quasi-static cyclic tests or dynamic shaking table tests are presented.
2 Seismic Retrofit Case Study
2.1 The SPEAR Structure

A case study of seismic strengthening shall be performed on a three-storey plan-
asymmetric structure (Rozman et. al, 2009). The structure was conceived as a
representative of typical non-earthquake-resistant older constructions in Southern
European countries. It was designed for vertical loads only, with the construction
practice and materials commonly used in Southern Europe in the early 70s. This
structure was pseudo-dynamically tested at full-scale and analysed within the
scope of the European project SPEAR.
Fig. 1 Floor plan, cross section and typical reinforcement of the SPEAR structure
The buildings floor plan, cross section and typical reinforcement in columns
and beams are shown in Figure 1. For the analysis in the paper the structure was
initially modelled without any additional retrofitting systems. An elastic modal
spectral analysis and more important a nonlinear static (pushover) analysis were
performed. The latter serves as a basis for the application of the N2 method a
performance based design method with which structural damage can be assessed
for different types of earthquake intensity.
Seismic Strengthening of Existing Concrete and Masonry Buildings 715
2.2 Seismic Analysis

Current codes of practice suggest two different approaches for design of ductile
structures in earthquake-prone regions. The first approach, well-known and widely
used, is referred to as the Force-Based Design (FBD) method since it mainly
focuses on designing the strength of the structure. The second approach, which
explicitly refers to the structural ductility in addition to the strength, is based on a
Non-linear Static Analysis (NSA) procedure. The purpose of this approach is the
evaluation of the actual structural response mainly in terms of ductility demand
and, hence, ultimate displacement induced in the structure by the earthquake
ground motion. The NSA procedures are generally based on the evaluation of the
push-over curve, which represents the response of the structure under a lateral
loading distribution schematising the seismic action. A number of different meth-
ods have been proposed, including the N2 method (Fajfar, 2000), which has been
adopted by the Eurocode 8. The N2 method was found to provide the best ap-
proximation among various NSA methods for SDOF systems with different hys-
teretic models and for MDOF systems (Fragiacomo et.al, 2006) however the N2
was never developed for the design of timber buildings with specific hysteretic
behaviour. Therefore it must be noted that the results derived in this study could
be non-conservative, because hysteresis loops with pinching, slip and strength
degradation (typical for connections in timber structures) dissipate less energy
than bilinear plastic loops with the same ductility. Nevertheless, it should also be
pointed out that for both analysed retrofitted setups, the SDOF systems equivalent
to the multi-storey building have periods longer than Tc which is usually the value
from where the reduction factor (R) and ductility factor () are considered to be
the same, regardless the type of hysteresis loop. The results of these analyses
should therefore be considered as a preliminary study aimed to investigate the
effect that different retrofit panel setups have on the seismic resistance of the case
reinforced concrete frame building.
2.2.1 Modelling of Crosslam Panels

Currently Eurocode 8 does not provide extensive seismic design guidelines for
timber buildings. Crosslam structures are not included in Eurocode 8 and not even
in Eurocode 5. There are, however, scientific papers dealing with modelling
of crosslam (Dujic, 2005, Blass et.al, 2004, Jbstl et.al, 2008, Moosbrugger et.al,
2006). The complex panel layout can be modelled using an orthotropic, homoge-
nised-orthotropic or homogenised-isotropic material, depending on the
possibilities offered by the FEM software. The proposed (Blass et.al, 2004)
homogenised-orthotropic plane-stress reduced cross section method, which is
based on the reduction of a multilayer to a single layer section using reduction
coefficients to modify the stiffness and strength, is precise enough for the needs of
seismic modelling, where building behaviour mostly depends on the behaviour of
connections (Fragiacomo et.al 2011, Sustersic et.al 2011, Sustersic and Dujic
2012). The aforementioned method was used to defy the crosslam panels used in
this study. 120 mm (layers 40-40-40) plates were taken into account.
2.2.2 Reinforced Concrete Plastic Hinge Definition

In the FE model used for the analysis, the plastic hinges that form at the ends of
beams and columns are set in accordance with the standard EN1998-3 (Eurocode 8).
An additional reduction factor of 0.375 is used due to the use of smooth reinforce-
ment bars. The plastic hinges are defined with a bilinearised relation rotation-
bending moment. The cracking in the reinforced concrete (RC) sections and a drop
in load capacity after the near collapse (NC) state is neglected. The limit states are
the following: DL (damage limitation) presents the yielding of reinforcement or the
maximal elastic capacity of a RC cross section. At this point the plastic rotation
(um,pl ) remains 0. SD (significant damage) is defined with 75% of the full plastic
rotation of a cross section. The NC (near collapse presents a full plastic rotation in a
cross section. A value of 3 um,pl defies the TC (total collapse) state, though in our
study it is merely theoretical as we only consider states up to NC.
2.2.3 Connections between the Concrete Frame and Crosslam Panels
The connections are modelled on the slipshear force experimental response of a

BMF 105 angle bracket with ten 60 mm long 4 mm diameter nails. The backbone of
the response of the 3rd cycle is taken as the input for the slip-force relationship in a
non-linear link element used in the FE model for connecting the RC frame with the
retrofitting XL panel. The reason for picking the 3rd cycle is to take into account the
accumulated damage that occurs in the angular bracket. It is visible from Figure 2,
that i. e. the 1st cycle yields about 30% higher peak strength oppose to the 3rd cycle.
The connection is simplified and hence the same response is modelled for both ver-
tical and horizontal direction. The case study presumes that the structure being retro-
fitted (the SPEAR construction) has the possibility of accessing the connections
from the inside since the structure has no outer walls or infills.
Fig. 2 A BMF105 angular bracket (left) and the presumed XL panel r.c. plate connection
(middle). Calibration of the non-linear FEM link on the backbone curve (right) of the 3rd
cycle of the experimental results of BMF105 brackets with ten 60 mm nails subjected to
shear force.
2.2.4 Seismic Resistance of the SPEAR Structure

With the use of the N2 method the seismic resistance of the building in the x
directions is assessed. The basic SPEAR structure, which could itself withstand an
earthquake with peak ground acceleration of less than 0.2 g has been retrofitted in
two different ways (Figure 3).
Fig. 3 The basic SPEAR structure on the left and both cases of the seismically retrofitted
structure. In the middle the option with short panels and on the right the option with long
panels.
The first option was to try strengthening it with shorter crosslam panels. A
BMF 105 bracket was used to attach the outer panel to the main structure at every
30 cm of the panel. That resulted in a not particularly stiff structure, since the
behaviour of the panels was mostly in bending and the short leverages to the con-
nections caused extensive deformations. As a result (Table 1) the allowable
ground acceleration a structure can withstand is raised by roughly 21 percent. On
the other hand the long panels with longer leverages and more connectors resulted
in an almost 90% increase in allowable ground acceleration.
Table 1 Comparison of maximum allowable ground acceleration for the basic and strength-
ened structures
Maximum allowable ground acceleration [g]
Basic structure 0.197

Strengthened with short panels 0.239 (+21 %)
Strengthened with long panels 0.374 (+90 %)
3 Experimental Testing
3.1 New Connection Concept Development

The system was developed using the general seismic strengthening principles
combined with a newly developed connection between the existing building and
the crosslam timber panels. The connection is combined into three steps (as
demonstrated in Figures 4):
connecting the crosslam timber panel with the steel anchoring bracket
connecting the steel anchoring bracket with the steel strap
connecting the steel strap with the existing building
The aim of this multiple-step connection is to allow on the one hand for a quick
assembly and attachment to the existing structure however on the other hand pro-
viding a controlled connection mechanism. The controlled energy dissipation
mechanism is established in the middle joint between the steel anchoring bracket
and the steel strap. This is the point where the connection has its ductile fuse
providing a controlled failure mechanism. The connections of the timber panel to
the steel anchoring bracket and the steel strap to the existing structure are both
completely elastic, providing a strong and stiff connection with as little slip as
possible.
Fig. 4 The connection detail (right) and a retrofitted building example (left). One should
note the steel strap running along the perimeter of the building in order to smear the
stresses.
When dealing with URM structures it is also necessary to smear the force of the
connections along the perimeter of each story, namely avoiding local concentra-
tions of stresses that can result in brittle failure of that connection. This is
achieved by using a steel strap (Fig 4) running along the complete perimeter of the
building (in each floor). The strap is attached to the URM structure with chemi-
cally anchored (epoxy or similar) steel rods on small enough distances to allow a
smother transfer of shear forces from the existing structure to the new outer jacket.
For buildings with RC slabs this is not necessary.
Seismic Strengthening of Ex
xisting Concrete and Masonry Buildings 7119
3.2 Quasi-static Cyyclic Testing of (Un)strengthened URM

During March, April and May 2012, tests of unreinforced masonry walls (URM M)
strengthened with crosslam timber panels (Fig 5) attached in various ways (glued,
anchored etc.) were bein ng performed at the Institute for Civil Engineering iin
Ljubljana, Slovenia. The vertical
v load on the walls was 300 kN in all cases. Hencce
a shear failure mechanism m was established. The results show that URM can bbe
strengthened with crosslam panels as both ductility and resistance can be inn-
creased by using differentt bonding systems. Namely by using only epoxy glue thhe
initial stiffness slightly increases,
i the peak resistance is increased by 34% annd
the displacement capacity y by 25%. However by connecting the Xlam panels witth
the specially developed steel
s connections at the top and bottom of the wall thhe
peak shear resistance is increased
i by 34% and more importantly the ductility bby
100% (Fig 6). Hence the strengthening system seems to have a good potential foor
unreinforced masonry buiildings.
Fig. 5 Experimental testing of URM (left) and URM strengthened with crosslam panells;
glued (middle) and bolted (riight)
150
125
100
75
50
25
base shear [kN]
0
-22 -20 -18 -16 -14
4 -12 -10 -8 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 22
-25
-50
URM
-75
XLAM bolted 2
-100
-125
-150 top displacement [mm]
Fig. 6 The hysteresis respon

nse of an unstrengthened and strengthened (steel connectionns)
URM wall
720 I. Sustersic and B. Dujjic
3.3 Dynamic Shakee Table Testing of (Un)strengthened RC

Frames With and Without Masonry Infill
During November and December 2012, a series of experimental tests waas
performed on the shaking table of the Macedonian IZIIS institute in Skopje. A
reinforced concrete framee and a RC frame with masonry infill were tested botth
un-strengthened and strenngthened (Fig. 7). The strengthening panels were attacheed
to the structure with the specially developed steel anchoring bracket. Tests on a
RC frame were performed m-
d in elastic state in order to preserve the structure undam
aged for later testing witth masonry infill. Two ground motions were used; thhe
original Petrovac earthquuake and the modified Landers (Fig. 9) accelerogram m.
Sweep tests were perform med before and after both earthquakes and the structurees
natural periods were meassured.
R frame with masonry infill; un-strengthened (middle) annd

Fig. 7 RC frame (left), RC
strengthened (right)
On figure 8 it can be observed that a shift in the natural frequencies occuurs

when the crosslam panells are applied onto the RC frame structure (only in thhe
ground floor and in the excitation direction). A structure is stiffened, hence its 1st
frequency changes from 2 Hz to 3.7 Hz.
1 min 2 min 3 min 4 min 5 min
5,0E-03 3,0E-03
4,0E-03 2,5E-03
2,0E-03
3,0E-03
1,5E-03
2,0E-03
1,0E-03
1,0E-03 5,0E-04
0,0E+00 0,0E+00
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Frequency [Hz] RC frame Strengthened RC frame
Fig. 8 Power spectral analyssis of the sweep test as measured by the accelerometer placed in
the top floor and measuringg in the table excitation direction. The retrofitted structure was
strengthened only in the botttom floor in the excitation direction.
2,0
1,8
EC8
1,6
Landers
1,4
1,2
Sa [%/100 g]
1,0
0,8
0,6
0,4
0,2
0,0
0,0 1,0 2,0 3,0 4,0
T [s]
Fig. 9 The modified Landers accelerogram elastic response spectra with 5% damping. The
accelerogram is modified in order to affect structures of different stiffness as it can be seen
from its comparison to the standard EC8 elastic spectra.
The RC frame was than rotated by 180and a masonry infill was constructed it
the excitation direction. First the buildings natural periods were measured for
both the strengthened and un-strengthened cases. Than the unstrengthened struc-
ture was tested until serious damage occurred (masonry infills started to crack and
fall out). The last ground motion applied to the unstrengthened structure was the
modified Landers earthquake, scaled to 0.75 g. Crosslam strengthening was ap-
plied and the loading protocol was repeated. The strengthened structure withstood
the 0.75 g Landers earthquake two times without any additional damage (minor
changes in the vibration periods (Tab. 2)) and also an additional Petrovac earth-
quake with a 0.5 g peak acceleration. The story drifts of the strengthened structure
were reduced by 30% (Tab. 3) at the same ground motion applied. Hence it was
confirmed that the cross laminated timber strengthening system can prevent the
collapse of an already severely damaged building.
Table 2 Vibration period changes of the unstrengthened and strengthened RC frame with
masonry infill before and after testing
Period (Frequency)
Unstrengthened structure before testing 0,144 s (6,925 Hz)

Unstrengthened structure after testing 0,337 s (2,967 Hz)
Strengthened structure before testing 0,161 s (6,227 Hz)
Strengthened structure after testing 0,177 s (5,660 Hz)
Table 3 Story drift comparison (for the scaled Landers accelerogram (0.7 g)) of the un-
strengthened and strengthened RC frame with masonry infill
Story Unstrengthened structure Strengthened structure
2-1 0,88 % 0,68 % (-23%)

1-0 2,16 % 1,52 % (-30 %)
4 Conclusions
In the paper we have presented a new system for seismic strengthening of existing
buildings. As far as the studies show it is most advisable to use longer (if possible)
crosslam panels instead of shorter segments. If shorter segments are used, it is
advisable to join adjacent panels together on the vertical sides, however this op-
tion has not yet been explicitly analysed.
The quasi-static cyclic testing on strengthened URM walls has shown that the
system can reach a 40% increase in strength and more importantly a 100% in-
crease in the ductility of an existing wall even in unfavourable conditions with a
high vertical load on the masonry walls.
The RC frame with masonry infill shake table testing has shown that a building
already severely damaged by an earthquake can be strengthened with the Xlam
panels and its stiffness can be increased to an almost undamaged state level.
Acknowledgments. The research support provided to the first author by the EU through the
European Social Fund 'Investing in your future' is gratefully acknowledged. The technical
support of the IZIIS institute laboratory team is also highly appreciated.
References
Blass, H.J., Fellmoser, P.: Design of solid wood panels with cross layers. In: 8th World
Conference on Timber Engineering, WCTE 2004, Finland, pp. 543548 (2004)
Ceccotti, A.: New technologies for construction of medium-rise buildings in seismic re-
gions: the XLAM case. IABSE Structural Engineering International, Special Edition on
Tall Timber Buildings 18(2), 156165 (2008)
Computers & Structures Inc. SAP2000 - Integrated Finite Element Analysis and Design of
Structures. Computers & Structures Inc., Berkeley, California, USA (2000)
Dujic, B., Klobcar, S., Zarnic, R.: Influence of openings on shear capacity of wooden walls.
Research report, University of Ljubljana and CBD Contemporary Building Design Ltd.,
Slovenia (2005)
European Committee for Standardization. Eurocode 8 - Design of structures for earthquake
resistance. Part 1 - General rules, seismic actions and rules for buildings (2003)
European Committee for Standardization. Eurocode 5 - Design of timber structures - Part 1-
1: General rules and rules for building (2004)
Fajfar, P.: A nonlinear analysis method for performance-based seismic design. Earthquake
Spectra 16(3), 573592 (2000)
Fragiacomo, M., Dujic, B., Sustersic, I.: Elastic and ductile design of multi-storey
crosslammasive wooden buildings under seismic action. Engineering Structures 33(11),
30433053 (2011)
Fragiaomo, M., Amadio, C., Rajgelj, S.: Evaluation of the structural response under seismic
actions using non-linear static methods. Earthquake Engineering & Structural Dynam-
ics 35(12), 15111531 (2006)
Jbstl, R.A., Bogensperger, T.H., Schickhofer, G.: In-plane Shear Strength of Cross lami-
nated Timber. In: Meeting 41 of the Working Commission W18-Timber Structures,
CIB, St. Andrews, Canada (2008)
Moosbruger, T., Gugenberger, W., Bogenperger, T.: Cross-Laminated Timber Wall Seg-
ments under homogeneous Shear with and without Openings. In: WCTE 2006 - 9th
World Conference on Timber Engineering - Portland, OR, USA (2006)
Rozman, M., Fajfar, P.: Seismic response of a RC frame building designed according to old
and modern practices. Bull. Earthquake Eng. 7, 779799 (2009)
Sustersic, I., Fragiacomo, M., Dujic, B.: Influence of connection properties on the ductility
and seismic resistance of multi-storey cross-lam buildings. In: 44th CIB-W18, Alghero,
Italy (2011)
Sustersic, I., Dujic, B.: Simplified cross-laminated timber wall modelling for linear elastic
seismic analysis. In: 45th CIB W18, Vxj, Sweden, paper 45-15-6 (2012)
In-Plane Stiffness of Traditional Timber Floors
Strengthened with CLT
Jorge M. Branco, Milos Kekeliak, and Paulo B. Loureno
ISISE, University of Minho, Department of Civil Engineering, Guimares, Portugal

jbranco@civil.uminho.pt
Abstract. Five full-scale timber floors were tested in order to analyze the in-plane
behaviour of these structural systems. The main objective was to assess the
effectiveness of the in-plane strengthening using cross laminated timber (CLT).
For that, one unstrengthened specimen (original), one specimen strengthened with
a second wood board, two specimens strengthened with 3 CLT panels and one
specimen strengthened with 2 CLT panels were tested. Moreover, because of its
importance in the composite behaviour, the first phase of the experimental
program was composed by push-out tests on specimens representing the shear
connection between the timber beams and the CLT panels. This paper describes
the tests performed and the numerical modelling aimed to evaluate the composite
behaviour of the strengthened timber floors.
Keywords: experimental evaluation, in-plane stiffness, strengthening, connection.
1 Introduction
In many countries, traditional construction comprises timber floor and roof

systems. Current knowledge assumes the need to preserve and to protect existing
timber systems as a cultural value with important advantages to the overall
behaviour of the building. Strengthening and stiffening of old timber floors are
often needed as they were designed to bear moderate loads and may suffer from
excessive in-plane and out-of plane deflections with respect to current
requirements. The structural refurbishment of traditional timber floors can be
achieved, in order to increase the bending stiffness of the main elements, by
including other elements, such as a concrete slab, or timber planks. The structural
behaviour of the resulting timber composite structure is governed by the strength
and stiffness of the mechanical joints adopted to connect the existing timber
beams to the new element.
Another important aspect to be keenly considered is the timber floor diaphragm
in plane stiffness, which may affect the structural performance of a traditional
masonry building subjected to lateral loads: the common configuration of existing
timber floors with a crossly arranged single layer of wooden planks could possibly
DOI: 10.1007/978-94-007-7811-5_65, RILEM 2014
726 J.M. Branco, M. Kekeliak, and P.B. Loureno
need an in plane shear strengthening, in order to assure an efficient redistribution

of lateral seismic load through the bearing walls [1,2,3].
Several techniques exist to strength timber floors, with different effectiveness
in terms of in-plane stiffness [4,5]. One possibility is the use of cross laminated
timber (CLT). This wood based material is easy to apply and has mechanical
properties with high potential to flexural and in-plane strengthening. In both cases,
the connection between the timber beams and the CLT is crucial for the load
capacity and distribution of stresses along the floor cross-section. Therefore, in the
first research step, an experimental campaign of push-out tests was defined to
study the connection between timber beams and CLT elements, using different
screws. The main objective was to assess the load-slip behaviour of the connection
to allow a full and detailed analysis of the results obtained from the tests of full-
scale timber floors done in the following stage. A total of 20 connections were
tested divided in four groups, with 5 specimens each, according to the screw used
and with their inclination in relation to the shear direction: HBS 8x140 and
VGZ 7x140 placed normal to the shear plane and inclined 45 SFS WT-T-8.2x190
and VGZ 7x180. Then, full-scale timber floors were tested in order to analyze the
in-plane behaviour of traditional timber floors and to assess the effectiveness of
the CLT-strengthening. Five specimens were tested: one unstrengthened, one
strengthened with a second wood board; one strengthened with CLT divided in
two panels and two strengthened with CLT divided in three panels. Last two type
of specimen were defined to assess the influence of the number of CLT panels
used in the strengthening.
2 Experimental Campaign
2.1 Test Procedure and Specimens

2.1.1 Timber-to-CLT Push Out Test
A total of 20 timber-to-CLT connections were tested divided in four groups of 5

specimens each, according to the screw used and its inclination with the shear
direction: HBS 8x140 and VGZ 7x140 placed normal (90) to the shear plane and
inclined 45, SFS WT-T-8.2x190 and VGZ 7x180. The central element is made of
C24 [6] solid wood with 100x200 mm2 while the exterior ones are made of CLT
with 66 mm thickness.
The main objective of these tests was to assess the shear stiffness of the
connections used in the full-scale timber slabs analyzed in the next research step.
By using screws placed with different inclination in relation with the shear plane,
namely, 90 and 45, it was possible to quantify the improvement expected by
using screws 45 inclined, enlarged in the case of full threaded screws. Besides,
two types of threaded screws were considered (SFS-WT-T and VGZ) aiming to
compare their efficiency.
In-Plane Stiffness of Traditional Timber Floors Strengthened with CLT 727
The load-carrying capacity and the deformation behaviour of the connections

were determined by push-out tests according to EN 26891:1991 [7]. During the
test, the load and relative displacements (slip deformation) of the joint members
were measured. Load was applied with a hydraulic jack and recorded by means of
a load cell. To measure the slip deformations two transducers with the accuracy of
0.1 mm were used.
Fig. 1 shows the specimens of each group where the geometry of the
connections and the position of the screws is visible.
Fig. 1 Specimens used in the push-out tests
2.1.2 Tests on Full-Scale Timber Floors
Five full-scale timber floors were tested: one unstrengthened floor (S), one floor
strengthened with a second wood board (SS); two floors strengthened with three
CLT panels (CLT3.1, CLT3.2) and one floor strengthened with two CLT panels
(CLT2). Last three type of specimen were defined to assess the influence of the
number of CLT panels used in the strengthening and type of used screws. In
specimen (S) and (SS), with one and two layers of floorboards, respectively, nails
of 2.5x60mm were used in the application of the floorboards. In specimens
strengthened with CLT - CLT2, CLT3.1 and CLT3.2 - screws were used to
connect the CLT panels to the timber beams. CLT2 and CLT3.1 specimens used
VGZ 7x180 screws while CLT3.2 used SFS WT-T-8.2x190 screws, in all cases,
placed inclined 45 in relation to the shear plane. A total of sixty screws, what is
thirty joints with X positioned couples of screws, were used for connection of
CLT panels to timber beams in each specimen. Fig. 2 presents the timber floors
specimens tested.
Fig. 2 Specimens used in the full-scale timber floor tests
The test setup is visualized in Fig. 3. A hydraulic jack, positioned at a height of

2m above base of the floor, apply a transversal force, with a programmed
displacement. In this case, the loading (displacement) was monotonic with a
constant rate of 0,05 mm/s. Seven LVDT sensors have been used for measuring
the displacements on the specimens. In Fig. 3, the measuring positions are marked
by numbers from 1 to 7. The force (F) versus displacement (d) curves were
measured on positions 1-7 for evaluation of the stiffness behaviour and
verification of deformed shape of whole specimens. Central set-up of LVDT
sensors, consisting of measuring positions from 4 to 7, is aligned to the axis of the
central beam and to middle of the beam length.
Fig. 3 Setup of the full-scale timber floor tests
2.2 Test Results and Analysis

2.2.1 Timber-to-CLT Push Out Tests
The experimental load-slip curves of the timber-to-CLT connections analyzed are

presented in Fig 4. The specification of push-out tests and averaged values of
numerical characteristics of connections obtained in push-out tests are presented in
Table 1. The comparison of the same type of screws VGZ, the specimens group
V1-V5 with 90 inclination angle of screws and group V6-V10 with 45
inclination angle of screws, the increase of the load carrying capacity is more than
200% and the increase of the stiffness more than 500%, when the length of screws
is not considered. As it was expected, the performance of the SFS type screws in
specimens group S6-S10 is higher than in case of screws VGZ in specimens group
V6-V10, due to bigger diameter. Considering the load carrying capacity, the
displacements and failure modes of connections obtained during the test, the
ductile behaviour of screw connections in each specimen group was observed.
Mentioned before, the load-carrying capacity and the deformation behaviour of
the connections were determined according to EN 26891:1991 [7], where the
value of Fest was assumed as the average of the maximum load Fmax measured in
particular specimen group. The maximum load Fmax of each test, corresponds to
maximum load measured during the test phase of displacement within 0-15mm. In
general, the inclination angle of the screws had a crucial influence on stiffness and
load carrying capacity of connections, investigated in push-out tests. It is
recognized, that the value of Kser according to EC5 [8] is very low in comparison
with tests values. The effective diameter def defined by [8] as the value equal to
1,1-times of core diameter of the screw, was used for calculation of SLS slip
modulus Kser. The effective diameter def , was used for calculation of slip modulus
of screw connections in analysis of inclined screws in [9]. The assessment of

the SLS slip modulus for connections with 45 inclination angle of screws was
based on the paper [9], wherein the theoretical calculation of connections with
inclined screws was performed and the equation (1), (2), (3), (4) were used and
derived.
K ser = K . cos 2 + K II sin 2 (1)
1
K II = (2)
1 / K ser ,ax ,1 + 1 / K ser ,sax , 2
K II = K ser ,ax ,i (3)
K ser ,ax ,i = 30.s g .d (4)
Where is the inclination angle of the screw, K is SLS slip modulus

according to [8] for one screw and one shear plane, KII is the axial slip modulus
for a screw, Eq. (2) for the double stiffness model; Eq. (3) for the single stiffness
model, Kser is the stiffness for screws crossed in an X position working
simultaneously one under sheartension stress and the other under shear
compression stress, Kser,ax,i is the instantaneous withdrawal/penetration stiffness, sg
is the embedment length, in mm, of the threaded segment of the screw and, d is
the outer diameter of the screws thread.
Fig. 4 Load-slip relation, results of experimental testing - Push-out tests

Table 1 Stiffness of screw connections average results of experimental testing - Push-out

tests
Maximum force Load

reached carrying SLS slip modulus Kser ( N/mm )
Type of (experiment) capacity
screw
Specimen 1 EN 1995-1- Stiffness
Experimental EC5
Specime 4 screws screw 1:2004 according to [9]
n group
one Single Double
Failure shear 1 stiffness stiffness
Reference
Fmax COV Fmax/4 mode FvRk plane COV screw model model
-Length
2 1
(kN) (%) (kN) (kN) (%) 1 screw 1 screw
screws screw
H1 - H5 HBS 8x140 19,01 8 4,75 f 5,48 2738 1369 42 2419 - -
V1 - V5 VGZ 7x140 25,05 11 6,26 f 4,22 1917 959 43 2060 - -
V6 - V10 VGZ 7x180 30,73 5 7,68 f* 5,08 12376 6188 15 - 8165 4598
SFS
S6 - S10 34,09 16 8,52 f* 6,85 18642 9321 23 - 9626 5429
8.2x190
f* - failure mode (f) according to EC1995-1-1:2004 was observed simultaneously with

failure related with the withdrawal capacity of the screw
2.2.2 Full-Scale Timber Floors
As mentioned before, five full-scale specimens were tested to analyze the in-plane
behaviour of timber floors and the effectiveness of the CLT-strengthening. The
specimen (S) represents an unstrengthened timber floor, with timber beams and
floorboards only. The specimen (SS) is strengthened by a second layer of
floorboards placed orthogonal to the first and specimens CLT3.1, CLT3.2, CLT2
represent the strengthening method based in the use of CLT panels connected to
timber beams by screws according to the provisions specified in Table 2. The
specimens evaluated can be divided in two main groups: specimens with
floorboards and specimens with CLT panels. The maximum load Fmax of each test
corresponds to maximum load measured during the test phase of displacement
within 0-100mm. The stiffness K was evaluated according to EN 26891:1991 [7],
where the value 0,4.Fest and 0,1.Fest were assumed to be equal to 0,4.Fmax, 0,1.Fmax
respectively, and the corresponding displacements v0,4, v0,1 were used in
calculation. For calculation of the stiffness K of the timber floors, the force
applied on upper beam and displacement on the opposite side of the beam was
used. Tests results obtained point out that strengthening using a second layer of
floorboards (specimen SS) can double the stiffness of the timber floor system
(specimens S), while the ultimate resistance has been increased almost to a
quadruple value. The specimens strengthened by CLT panels are approximately 5
-10 times stiffer than the unstrengthened specimen (S) and 2-4 times stiffer than
specimen strengthened by a second layer of floorboard (specimen SS).

Experimental results show that specimens with CLT divided in three CLT panels
were 30% stiffer than specimen strengthened with 2 CLT panels. The highest
stiffness was observed on specimen CLT3.2, where the SFS type of screw with
bigger diameter was used. The results of stiffness and ultimate resistance of timber
floor specimens obtained during experimental testing are presented in Fig. 5 and
Table 2.
Fig. 5 Load slip behaviour of the full-scale timber floors
Table 2 Stiffness and ultimate strength - results of experimental testing - Timber floor tests
Maximum
Specimen Description of Stiffness of
Specimen description force
label connection specimen
reached
Number of fasteners / Fmax K
joint ( kN ) ( N/mm )
(S) One layer of floorboards 2 nails 2,5x60 2,13 55
(SS) Two layers of floorboards 4 nails 2,5x60 8,51 132
CLT3.1 Three CLT panels 2 screws VGZ 7x180 23,70 411
CLT3.2 Three CLT panels 2 screws SFS 8.2x190 24,55 563
CLT2 Two CLT panels 2 screws VGZ 7x180 25,20 311
3 Numerical Modelling
The numerical analysis and study of structural behaviour of timber floor
specimens was performed in order to evaluate its composite behaviour. Numerical
models were developed in finite element analysis (FEA) software ANSYS [10].
Considering the capability of the finite elements, the structural models of timber
floor specimens were assumed to be modelled with use of one dimensional (1D)
beam elements and volume solid elements. 1D beam elements BEAM188 were
used for modelling timber beams in floor specimens, and volume elements
SOLID186 were used for modelling of CLT panels. The CONTA174/TARGE170
elements were used for modelling of contact interface and related friction between
elements representing the CLT panels. The eccentric position of timber beams to
CLT panels was neglected and the position of the beam elements was assumed to
be in the middle plane of CLT element. The structural system in numerical model
was fixed by three vertical supports and one horizontal support on the bottom
beam, and horizontal supports in direction out of plane of floor in each connection
between the beams and CLT panels. Structural members in numerical models, the
beams and CLT panels, respected dimensions of timber floor specimens.
Description of numerical models is presented in Fig 6.
Fig. 6 Description of numerical models of timber floors specimens CLT3.1, CLT3.2, CLT2
Connections between beam elements and volume elements (timber beam - CLT
panel) have been modelled with use of MPC184 elements with linear force-
displacement relation in two directions in plane of CLT, considering the slip
modulus Kser achieved by the push-out tests performed. The Kser value of
HBS8x140 screws from push-out tests (group V1-V5) was used for load-slip
relation in direction perpendicular to the beam axis, in joints for numerical model
of CLT3.2. The third displacement, orthogonal to the CLT plane was considered
fixed. All rotational degree of freedom, in the three axes, were release. A
simplified description of the in-plane structural behaviour of timber floor as the
composite of CLT panels and timber beams is showed in Fig. 7, where the rigid
connection of fastener to beam is assumed and slip on CLT is described.
Considering the global displacement shape of timber floor, it can be expected, that
the stiffness of the connection in direction perpendicular to the axis of the beam
affects the stiffness more strongly than the connection stiffness in direction
parallel to the axis of the beam.
Fig. 7 Shear displacement between CLT panels; slip in screw joint
Taking into account the anisotropy of wood, transversely isotropic material

model was defined in the three orthogonal directions, where properties in radial
and tangential direction were assumed to be equal because of reduction of material
characteristics. In terms of material properties, the elastic material characteristics
of C24 defined in standard EN 338 [6] were adopted for the beam elements. The
poissons ratios LT and RT were calculated based on [11] and missing mechanical
parameters were obtained as depended mechanical properties for transversely
isotropic model. Material characteristics of CLT panels were adopted and
modified, based on EN 338 [6] and European Technical Approval ETA-06/0009
[12]. The boards in particular layers of the CLT panels were graded by
manufacturer to strength class C24 [6]. Mentioned before, material characteristics
of CLT panels were modified in order to respect and approximate the stiffness
behaviour of the CLT panel, influenced by different orientation of timber boards
in particular layers of CLT. Elastic properties were calculated as weighted average
of material characteristic of wood for parallel and perpendicular to the grain
direction respecting the thickness of corresponding layers of CLT panel. It is
necessary to note, that shear stiffness in RT-plane for CLT material property is
significant higher than the value provided by technical list, due to dependency of
material characteristics for material model used, but without the influence on
simulation, when considering the centred position of the beam elements and in-
plane direction of the load. As it was observed during the tests, the stiffness
behaviour of whole specimens was influenced particularly by stiffness of the
screw connections, because of this fact, the material model used and the
simplification of the CLT panel as one anisotropic timber member, were assumed
as a proper approach for simulation of timber floor tests. The summary of the
material characteristics of timber floor members are presented in Table 3.
Table 3 The summary of mechanical properties of timber member in numerical models
GLT = GRL
ET = ER
MOC
GRT
RL
RT
LT
EL
m
Mechanical parameter
(kg/m3)
(MPa)
(MPa)
(MPa)
(MPa)
(%)
-
Type of timber member
Timber strength class C24 (1) 12,0 420 11000 370 690 - - - -
(2)
CLT board - strength class C24 12,0 470 11000 370 690 50 - - -
Timber beams (numerical model
- - 11000 370 690 128 0,440(3) 0,015 0,450(4)
characteristics)
CLT panel (numerical model
- - 7492 3878 690 1337 0,440(3) 0,227 0,450(4)
characteristics)
MOC - Moisture content, m - Mean density, E - Elastic modulus, G - Shear elastic
modulus, L - Longitudinal direction, R - Radial direction, T - Tangential direction, -
poissons ratio
1 - material properties used from the European standard EN 338 [6]
2 - material properties used from the European technical a approval Binderholz, ETA-
06/0009 [12]
3 - value obtained as geometric average of poissons ratios LT and LR [11]
4 - value obtained as geometric average of poissons ratios RT and TR [11]
The parametric study by means of finite element modelling was performed in

order to investigate composite behaviour of timber floor specimens. For each
specimen, CLT3.1, CLT3.2, CLT2, three groups of numerical models were
developed, depended on variation of the stiffness characteristics of screw
connections and friction between the CLT panels. The influence of friction
between CLT panels on stiffness behaviour of timber floor specimen was analyzed
by variation of coefficient of friction in numerical models. For each specimen
three numerical models were developed, where the coefficient of friction varied
in range of 0,2 - 0,8.

The results of experimental testing and numerical modelling show the relation
between structural members and its connection. The variation of stiffness
characteristics of screw connections, the values of coefficient of friction used and
results of numerical modelling are presented in Table 4. Comparison of results of
experimental testing and numerical modelling shows good match in case of
CLT3.1 and CLT3.2 specimens. In case of specimen CLT2, the stiffness
behaviour obtained by means of numerical modelling was overestimated. This can
be explained by simplified modelling of connection, where only linear load-slip
relation was used in numerical models. Because of this simplification, the
structural behaviour of elastic phase is considered in comparison. The results of

parametric study shows, that the main influence on in-plane stiffness of whole
timber floor specimens had the stiffness behaviour of the screw connections in the
direction perpendicular to the axis of the beams. The influence of the friction
between the CLT panels was not observed as a significant factor of global
stiffness of timber floor specimens. As it was presented in results of push-out
tests, the inclination angle had major influence on stiffness and load-carrying
capacity of joints. Considering this influence on local joint and composition of
structural members into whole structure, the global behaviour of composite
structure can be significantly changed depending on connection applied, or
modified as required.
Table 4 Specification of numerical models and results of numerical modelling

CLT3.1EXP
CLT3.2EXP
CLT2EXP
CLT3.1Y
CLT3.1X
CLT3.2Y
CLT3.2X
CLT2Y
CLT2X
Model name
Parameter
Kser (N/mm), SLS Slip modulus in X-direction 12376 12376 1917 18642 18642 2738 12376 12376 1917
Kser (N/mm), SLS Slip modulus in Y-direction 1917 12376 1917 2738 18642 2738 1917 12376 1917
K (N/mm), Coefficient of friction = 0,2 452 2032 425 623 2687 589 528 1990 496
K (N/mm), experimental testing 411 CLT3.1 563 CLT3.2 311 CLT2
CLT_EXP - numerical model with connection stiffness obtained by push-out tests, CLT_Y,
CLT_X - numerical model with modified connection stiffness in Y, X-direction, The
orientation of the X,Y-direction is consistent with directions of x, y-displacements
presented in Fig.7, K - stiffness of timber floor specimen obtained as the result of numerical
modelling. The same procedure of evaluation was used as used in full scale timber floor test
[7].
Acknowledgement. The present work is part of a research project supported by program

Quadro de Referncia Estratgico Nacional (QREN), project number 21635, from the
Agncia de Inovao (ADI). The authors would like to thank BinderHolz and Rotho Blaas
for all the support offered, particularly in the preparation and execution of the experimental
program.
References
[1] Branco, J.M., Cruz, P.J.S., Piazza, M.: Experimental analysis of laterally loaded
nailed timber-to-concrete connections. Construction and Building Materials 23(1),
400410 (2009)
[2] Dias, T.I.: Pavimentos de madeira em edifcios antigos. Diagnstico e interveno,
Porto. Dissertao de Mestrado em Engenharia Civil, FEUP (2008) (in Portuguese)
[3] Appleton, J.: Reabilitao de edifcios antigos patologias e tecnologias de

interveno. ORION (2003)
[4] Valluzzi, M., Garbin, E., Benetta, M.D., Modena, C.: Experimental assessment and
modelling of in-plane behaviour of timber floors. In: SAHC 2008 - Structural
Analysis of Historical Constructions, Bath, UK, July 2-4 (2008)
[5] Angeli, A., Piazza, M., Riggio, M., Tomasi, R.: Refurbishment of traditional timber
floors by means of wood-wood composite structures assembled with inclined screw
connectors. In: 11th World Conference on Timber Engineering WCTE 2010, Riva
del Guarda, Trentino, Italy, June 20-24 (2010)
[6] EN 338:2003, Structural timber - strength classes. European Committee for
Standardization, Brussels
[7] EN 26891:1991, Timber structures. Joints made with mechanical fasteners. General
principles for the determination of strength and deformation characteristics.
European Committee for Standardization
[8] EN 1995-1-1:2004 Eurocode 5 Design of timber structures, Part 1-1: General
Common rules and rules for buildings, Brussels
[9] Tomassi, R., Crosatti, A., Piazza, M.: Theoretical and experimental analysis of
timber-to-timber joints connected with inclined screws. Construction and Building
Materials 24, 15601571 (2010)
[10] ANSYS, Academic Research, Release 12.0, ANSYS, Inc. Documentation for
Release 12.0 (2009)
[11] Dahl, K.B.: Mechanical properties of clear wood from Norway spruce. Ph.D
dissertation, Norwegian University of Science and Technology, Trondheim (2009)
[12] European technical a approval ETA-06/0009, Multilayered timber elements for
walls, ceilings, roofs and special construction components. European Organisation
for Technical Approvals (2011)
Propose Alternative Design Criteria for Dowel
Type Joint with CLT
Shoichi Nakashima, Akihisa Kitamori, Takuro Mori, and Kohei Komatsu
Research Institute for Sustainable Humanosphere, Kyoto University, Gokasho,

Uji, Kyoto, 611-0011, Japan
s-nakashima@rish.kyoto-u.ac.jp
Abstract. The load carrying capacity of dowel type joint in Cross Laminated
Timber (CLT) was derived based on the Johansens yield model. The steel plate
inserted drift pin joint with CLT (5 layered; thickness of laminae were same in one
CLT, all of laminae were orthogonally arranged) was chosen as the specimen.
Stiffness and nonlinear load deformation relationships were calculated by nu-
merical analysis using Rigid Body Spring Model (RBSM). Estimation showed the
good agreement with the tensile test results on the joints.
Keywords: cross laminated timber, dowel type joints, Johansens yield model,
load carrying capacity, nonlinear load deformation relationship, rigid body spring
model.
1 Introduction
Cross Laminated Timber (CLT) is especially expected in Japan as a new material

for the structural wall panel and/or floor which might promote the usage of local
low grade Japanese cedar.
We chose the dowel type joint as the connectors to fix wall legs because it has
high stiffness and strength and ductility.
For the drift pin joint in solid timber or glulam the criteria for the yield load de-
pends on the Johansens Yield Model and stiffness depends on the beam on elastic
foundation theory. In this paper, load carrying capacity is derived based on Johan-
sens theory and stiffness were estimated by the numerical analysis with Rigid
Body Spring Model (RBSM) based on the Winklers foundation model[1, 2].
2 Load Carrying Capacity
Assumption
Load carrying capacity by Johansens Yield Model for the six layers CLT was
reported by Uibel et al [3]. Here, newly 5 layers CLT was picked up. Model
DOI: 10.1007/978-94-007-7811-5_66, RILEM 2014
740 S. Nakashima et al.
consists of steel plate, drift pin, and CLT (5 layered; thickness of laminae were
same in one CLT, all of laminae were orthogonally arranged). It was assumed that
the perfectly plastic relationships for the materials could be held good to decide
the load carrying capacity (py). Fig.2 shows the possible yielding models for deriv-
ing py. To derive the tensile performance of drift pin joint of CLT, equilibrium of
the loads is drawn such as shown in Fig.2.
Load carrying capacity was derived using the diameter (d), length (l) and bend-
ing strength (F) of drift pin and density () of timber. In practice, the diameter of
drift pin is the region of 10 - 20mm. Therefore in such region, load carrying ca-
pacity will be decided by Mode2.2 deformation in dFig.2.
Material properties
P Fe : embedment strength
parallel to the grain
(N/mm2)
F : strength of drift pin
CLT (N/mm2)
: =F/Fe
P
l : timber width (mm)
d : diameter (mm)
Steel Plate Drift Pin
Fig. 1 Concept of joint and defenition of material properties
Load carrying capacity (py) of the CLT dowel type joint with slotted in steel
plate should be estimated by following equation.
py = C Fe d l
(1)
where,
C: coefficient depends on joint configuration and failure mode shown in Fig.2
Fe: basic bearing strength of timber (surface layer) (N/mm2)
d: diameter of fastener (mm)
l: effective length of fastener in CLT panel
2.1 Coefficient C
Coefficient C should be determined from the minimum value of following
equation;
Propose Alternative Design Criteria for Dowel Type Joint with CLT 741

4
Mode1
5

58 8 d 6
2
+ Mode2.2
25 3 l 5
d 8
C = Mode3.1
l 3

1 4 d
2
+ +
1
Mode3.2
100 3 l 10

4 8 d 1
2
+ Mode3.3
25 3 l 5
(2)
where,
: ratio of basic material strength of fastener and that of basic bearing strength of
main member (F/Fe)
F : basic material strength of fastener (N/mm2)
Fe : bearing strength of timber (surface layer) (N/mm2)
Equation(2) stands on the following assumptions of equation(3);
l
t1 = t2 = 2t3 = t =
5
f = 2 f = f = F (3)
1 2 3 e
where,
t1 : thickness of first layer (surface layer) (mm)
t2 : thickness of second layer (middle inner layer) (mm)
t3 : thickness of third layer (central layer) (mm)
f1 : bearing strength of first layer (surface layer) (N/mm2)
f2 : bearing strength of second layer (middle inner layer) (N/mm2)
f3 : bearing strength of third layer (central layer) (N/mm2)
The assumption of equation(3) indicates (a) the thickness of each layer is same (b)
bearing strength of perpendicular to the grain layer is the half of that of parallel to
the grain layer. From this assumption, Mode 2.1 of Fig.2 cannot stand.
P/2 P/2 P/2 P/2 P/2 P/2
Center Plastic
l/2 of Hinge
Rotation
P/2 P/2 P/2 P/2 P/2 P/2
f1 f2 f3
My My My My My
My My M
py/2 py/2 py/2 py/2 py/2 y py/2
t1 t2 t3 t1 x t3 x t2 t3 x x t3 x t2 t3
t2-x t1-x
Mode1 Mode2.1 Mode2.2 Mode3.1 Mode3.2 Mode3.3
Fig. 2 Possible yielding models based on equilibrium of the loads
2.2 Figure for Load Carrying Capacity

Fig.3 shows the relationship between load carrying capacity divided by square of
diameter of fastener (py/d2) and ratio of length and diameter of fastener (l/d),
which were shown in Eq.1.
Fig. 3 Relationships (Py/d2-l/d) for the Load Carrying Capacity (py)

3 Determination of Load-Deflection Curve of Dowel Type

Joints
3.1 Analysis
It is important for structural design to determine the load (P) deformation ()
curve of CLT dowel type joint to obtain the stiffness, yield load and the other
characteristic values.
The stiffness of dowel type joints is usually determined by the theory of the
beam on elastic foundation, in which an assumption of homogeneous material is
used to derive the equation. Here, we dared to determine not only the stiffness but
also the shape of curve by simulating the situation of beam on nonlinear multi-
layered foundation as shown in Fig.4. This model called RBSM (Rigid Body
Spring Model) developed by Kawai [6] and applied to bolted joint and drift pin
joint in glulam by Tsujino, et al. [7].
In comparison with past usual 2D FEM model with Winklears foundation
model, each element is replaced by rigid body and normal, shear, and rotation
spring at both nodal points(shown in Fig.4). The foundation consists of normal
spring is also replaced by combination of normal and rotation springs.
The behavior of the dowel itself is assumed as a nonlinear. The stress ()
strain () relationship of the dowel is expressed with three springs as in Fig.4 and
they are expressed in Eq.4 according to Kawai [6] and Tsujino [7].
2EA , 2EA , k = 2EI

kn = ks =
l1 + l2 (l1 + l2 ) (1+ ) m l1 + l2 (4)
I
My = (5)
d
km2 = km , km3 = km
(6)
where;
E = 205000 N/mm2 (modulus of elasticity of steel beam)
A = d2 / 4 mm2 (cross section area of steal beam)
l1 = length of the element of the model 1 (mm)
l2 = length of the element of the model 2 (mm)
= 0.3 (Poissons ratio of steel)
d = diameter of steel beam (mm)
= km2 / km =0.035, = kn3 / km = 0.02
km, km2, km3 = initial, second and third rotational stiffness (Nmm/rad)
kn = initial normal stiffness (N/mm)
ks = initial shear stiffness (N/mm)

= 580 N/mm2 (proportional limit strength of steel beam)
2 = 750 N/mm2 (strength of steel beam)
here, , 2, and are obtained by three point bending test of fasteners as shown
in the sub-figure in Fig.4.
Timbers that are oriented in parallel and perpendicular to the grain direction are
also assumed as a tri-linear Winkler type foundation that generates restoring
stress in response to the deflection of the beam. Them bearing stress () em-
bedment () relationship is obtained by embedment tests for parallel and perpen-
dicular to the grain direction. The spring constants of model expressed in equation
(8), according to Tsujino [7]
For 0 direction layers;
3
K h = k0li , K r = k0 li
12 (7)
For 90 direction layers;

3
K h = k90li , K r = k90li
12 (8)
where;
li = length of element (mm)
k0 = 35.23 N/mm3 (Initial embedment stiffness)
k90 = 11.46 N/mm3 (Initial embedment stiffness)
The other values indicated in Fig.4 are as follows;
0 = 0.48 = k0 / k0
90 = 0.61 = k90 / k90
0 = 0.01 = k0 / k0
90 = 0.05 = k90 / k0
l0 = 20.88 N/mm2 (Proportional limit embedment strength)
l90 = 9.1 N/mm2 (Proportional limit embedment strength)
y0 = 28.04 N/mm2 (Yield embedment strength)
y90 = 12.40 N/mm2 (Yield limit embedment strength)
For the direction loading, we used the Hankinsons equation to obtain k,

, , l, y, k+90, +90, +90, l+90, and, y+90.
Static load increment analysis was performed to get the stiffness and P- curve
with RBSM. Number of rigid beam element is 38. The software SNAP LE Ver.6
was used for the analysis. To perform this analysis on the software, we dared to
use improved model (Fig.5(c)).

y
k 0
limit k0
"
M Bending of Steel Beam
"
k0 Parallel to the grain
M2" km3
limit y
km2
My"
"
"
Perpendicular to the grain
km
"
y 2 y k90
limit k90
Beam
k90
limit y
P Boundary condition
Border condition
uz=0
y=0
ux=0
y=0
L/2
Spring for Dowel type fastener
P 2EA
kN = Normal
l1 + l2
P
d 2GA
l2 kS = Shear
l1 + l2
Material properties of steel beam
Material Properties kM
2EI
Diameter d 16 mm l1 kM = Moment
Timber width L/2 75 mm l1 + l2
Young's Modulus E 205000 N/mm 2 Spring for Timber
Shear Modulus G 79800 N/mm 2
kS kN
Area A 201 mm 2 K h = kh l1
Morment of Inertia I 3217 mm 4 kh l13
Element length l, l 1 , l 2 2 mm Kr =
12
Embemdment stiffness k h 35.23 N/mm 3
Embemdment stiffness k h90 11.56 N/mm 3 Kh Kr
Fig. 4 Numerical Analysis Model (RBSM) for the half of the drift pin joint
AM
G BE
RIN
R SP
BE
TIM
x
Finite Spring
(a) FEM Model with nonlinear
beam and nonlinear spring
Beam on Nonlinear ING AM
Foundation PR BE
RS ID
BE RIG
TI M G
RIN
SP
AM
BE
(b) RBSM Model with Rigid Body (c) Improved Model for
and Nonlinear Springs Conventional Software
Fig. 5 Comparing between FEM model and RBSM model

3.2 Experiment
Test Specimen and Materials

Fig.6 shows the specimens and apparatus for tensile test of joint. Five layered
Japanese cedar CLT (thickness of lamina (t) = 30mm, air-dried specific density
() = 0.43, moisture contents (M.C.) = 13%, all surface of lamina were glued by
AIP. ) were used as specimens. The parameters were the angle of the grain direc-
tion of outer layer with respect to the loading direction 0, 45, and 90. The spec-
imen consists of CLT, steel plate (thickness ts = 9mm), and drift pin (diameter (d)
= 16mm). The thickness of slit in CLT was 11mm, then end and edge distances of
joints were 7d respectively. Loading protocol used was monotonic and cyclic, and,
tensile load and relative displacement between steel plate and CLT were meas-
ured. The detail of test was same as reference [5].
=16mm
CLT l=150mm Drift Pin
700 Load cell 100kN
224
Oil Jack 200kN

90 Steel Plate t=9mm
45 0
Fig. 6 Sampling of specimens and apparatus of tensile test for CLT - drift pin joint
3.3 Results and Discussion
Comparisons between Calculated Characteristics and Experimentally Derived

Ones
Fig.7 shows the example of deformation that obtained by analysis. Analysis were
performed for half of the figure and copied with turn over. Fig.8 shows Load (P)
deformation () curve. Left side figures show calculated results and the right side
are experimental ones.
Deformations (a) and (b) were Mode2.2 in Fig.3. We can say that the theoreti-
cal model estimated deformation curve and P- curves suitably ash shown in Fig.7
and Fig.8.
(mm)
00$ 5$ 0$ 5$ 0$ 5$ 30
30$ 35$ 40$ 45$ 50$ 55$ 60
60$ 65$ 70$ 75
75$ 80$ 85$ 60
90$ 95$ 00$ 05$ 0$ 120
5$ 0$ 5$ 30$ 35$ 40$ 150
45$ 50$
-5$
0$
5$ 20kN
10$ 25kN
15$ 30kN
20$
25$ Steel
35kN
0 Layer 90 0 90 0
30$ Plate
(a) Analysis (b) Experiment

Fig. 7 Example of deformation that obtained by analysis and experiment. (Five layered
CLT, diameter of drift pin = 16mm, thickness of layers = 30mm).
60 60
50 50
Load P (kN)
Load P (kN)
40 40
30 0 30 0
20 45 20 45
10 90 10 90
0 0
0 5 10 15 20 0 5 10 15 20
deformation (mm) deformation (mm)
(a) Analysis (b)Mean curves of experiments
Fig. 8 Example of P- curves that obtained by analysis and experiment
Fig.9 shows the comparisons between calculated load carrying capacities (py) ,
initial stiffness (k) and experimental values. From these figures, it can be seen that
there is satisfactory, agreement between calculated stiffness (k) and experimental
ones. While in the case of yielding prediction, some sorts of discrepancy, were
observed between calculated load carrying capacity (py) and experimental ones.
The differences according to loading angles of in-plane characteristics are not
so much. Especially, the yield load shows almost same values in both calculation
and experimental data.
40 40
Experiment Experiment
Stiffness k (kN/mm)
30 30
Yield Load py (kN)
20 20
Analysis Analysis
10 10
0 0
0 15 30 45 60 75 90 0 15 30 45 60 75 90
Angle (rad) Angle (rad)
(a) Initial Stiffnes (k) (b) Yield Load (Py)
Fig. 9 Comparison in characteristics between analysis and experiment
4 Conclusion
Load carrying capacity of drift-pined joint in CLT is derived using the diameter
(d), length (l) and bending strength (F) of drift pin and density () of timber as the
parameters for prediction.
Load deformation curves were determined by the numerical analysis by RBS-

Model using a commercial computer software. Calculation showed the good
agreement with test results.
The differences of mechanical performances according to loading angles of in-
plane characteristics are not so much. Especially, the yield load shows almost
same values in both calculation and experimental data. This indicates that CLT
drift-pin joint has very weal orthotropic strength property.
References
[1] Kuenzi, E.W.: Theoretical Design of Nailed or Bolted Joint Under Lateral Load,
USDA, No. D1951 (March 1955)
[2] Harada, M.: On Longitudinal Strength of Wooden Ships. Report of the Institute of
Industrial Science University of Tokyo (3) (1951) (in Japanese)
[3] Bla, H.J., Uibel, T.: Forschung an der Universtt Karlsruhe - Stiftfrmige
Verbindungsmittel in Brettsperrholz, Granzer Holzbau-Fachtangung Tagungsband06
Brettsperrholz, pp. I-1I-16 (2006) (in German)
[4] Nakashima, S., Kitamori, A., Mori, T., Komatsu, K.: Evaluation of Tensile Perfor-
mance of Drift Pin Joint of Cross Laminated Timber with Steel Inserted Plate. In:
World Conference on Timber Engineering, pp. 417424 (2012)
[5] Nakashima, S., Kitamori, A., Komatsu, K.: Evaluation of Tensile Performance of
Drift Pin Joint with Steel Plateon Cross Laminated Timber. J. Struct. Constr. Eng.,
AIJ 78(687), 969975 (2013) (in Japanese)
[6] Kawai, T., Chen, C.-N.: A Discrete Element Analysis of Beam Bending Problems In-
cluding the Effects of Shear Deformation. SeisanKenkyu 30(5), 710 (1978)
[7] Tsujino, T., Takeuchi, N., Hirai, T.: Analyses of the Lateral Resistanes of Bolted
Joints and Drift Pin Joints in Timver IV. A Preliminary Study of Applying the Rigid
Bodies-Spring Models, Mokuzaigakkaishi 50(3), 176182 (2004) (in Japanese)
Part VIII
Properties and Testing of Wood
Length Effects on Tensile Strength in Timber
Members With and Without Joints
R. Brandner and G. Schickhofer
Institute of Timber Engineering and Wood Technology, Graz University of Technology,

Inffeldgasse 24/I, 8010 Graz, Austria
{reinhard.brandner,gerhard.schickhofer}@tugraz.at
www.lignum.at
Abstract. Strength properties in timber disperse remarkable. A regressive course

of strength with increasing volume is observed. This phenomenon is known as
size effect, dominated by the stochastic part, the dispersion in strengths locally.
Although boundary conditions of this theory are violated, traditionally, size effects
have been modelled by means of Weibulls brittle failure theory. We address sto-
chastic length effects on the tensile strength parallel to grain of timber members
with and without finger joints by means of a probabilistic approach. For jointed
members regulations of minimum requirements on finger joint tensile strength are
discussed. We assume lognormal distributed strengths and use a second-order
hierarchical model together with equicorrelation to account for within and be-
tween members strength variations. As outcome, the effect of length on the mean
and 5 %-quantile of tensile strength is quantified. Simplified models and parame-
ters for the design of timber structures are provided. Minimum requirements on
finger joint tensile strength, relevant for modelling of timber products as well as
factory production control, are defined.
Keywords: Length effect, Timber, Tensile strength parallel to grain, Lognormal,

Requirements on finger joints, Hierarchical model.
1 Introduction
In general, size effects describe the relationship between strength and volume or
dimension relative to a reference volume or dimension at equal stress conditions
and for a specific material hierarchy. A regressive course in strength with increas-
ing volume or dimension can be observed. In contrast, scaling effects describe the
influences caused by changing the material hierarchies (Sutherland et al. 1999).
Size effects are normally dedicated to brittle or quasi-brittle materials. In con-
trast to perfect brittle material, characterised by randomly distributed microscopic
flaws and without reference dimensions determined by the material structure
(Weibull 1939), quasi-brittle material shows macroscopic flaws which are
DOI: 10.1007/978-94-007-7811-5_67, RILEM 2014
752 R. Brandner and G. Schickhofer
additionally seldom randomly distributed. Modelling of this type of material ne-

cessitates the definition of representative volume elements (RVEs) or characteris-
tic dimensions (Baant 2001). Quasi-brittle materials fail in dependency of their
dimension within a given hierarchical level. In contrast, brittle material fails sud-
denly with the failure of the first, the weakest element. Wood, timber, bone and
fibre composites are typical quasi-brittle materials (Baant 2001).
We focus on stochastic size effects in timber, in particular on length effects on
the tensile strength parallel to grain. Thus effects caused by an ideal serial system
action are concerned. In the 15th century first examinations were made by Leo-
nardo da Vinci, followed by Galileo (1638), Mariotte (1686) and Griffith (1920).
Peirce (1926) formulated the weakest link theory (WLT) based on the extreme
value theory (EVT) of Tippett (1925). According to the EVT, the distribution
function of minima gained from samples of identical and independent distributed
(iid) variables (e.g. the strength of elements) X1 ~ FX1 ( x ) of an arbitrary distri-
bution function FX1 ( x ) is given as
N
FX N ( x ) = 1 1 FX1 ( x ) , with X N = min ( X1i ) , i = 1, , N. (1)
i
Eq. (1) is primary of interest for distributions available in closed form. One of
these was defined by Weibull (1939). Combining physics and stochastics he for-
mulated the brittle failure theory (BFT) and the two- and three-parameter Weibull
distribution (2p, 3pWD), which is one of in total three limiting distributions of the
EVT. He assumed an ideal brittle material according to the WLT, composed of an
infinite number of iid sub-volumes, each characterised by a small but finite num-
ber of randomly distributed flaws. Applying the theory of elasticity and assuming
an ideal serial system the material was treated as continuum. The BFT allows
explaining phenomena in brittle material caused by size effects. Due to its flexibil-
ity in fitting test data and its closed form the 3pWD or solely its kernel power
model have been frequently used for modelling size effects also of quasi-brittle
and other materials which definitely violate boundary conditions of the theory. A
comprehensive discussion can be found e.g. in Williamson (1992), Baant (2001)
and Brandner (2012). The main contradicting points are:
timber is a hierarchically structured, heterogeneous material with macroscopic
flaws in dimension even equal to that of the elements (e.g. knot clusters);
the stress-strain relationship depends on the way elements interact in the system
(serial vs. parallel; e.g. Bohannan 1966); in parallel systems load-sharing and
redistribution of loads can be clearly identified;
physical and geometrical properties in timber are spatially correlated;
the lognormal distribution (2pLND) is often found to represent strengths of
timber better than the 3p and 2pWD; this can be explained somehow by multi-
ple fracture processes before final failure, which coincide with the law of pro-
portionate effect (Gibrat 1930) underlying the 2pLND.
Length Effects on Tensile Strength in Timber Members With and Without Joints 753
Thus and despite its simplicity it is recommended to reconsider the preferred

use of BFT for the description of size effects on timber strengths.
With focus on length effects in tension parallel to grain Madsen and Buchanan
(1985) found a length effect higher in jointed members than in that without joints
(kl,05 = 0.36 vs. 0.29, respectively). An inverse relationship between the length
effect and timber quality was observed (Madsen and Buchanan 1985, Barrett et al.
1992, Williamson 1992, Burger 1998). This is explained by the reduced dispersion
in occurrence and magnitude of global and local growth characteristics with in-
creasing timber quality of the same species. Madsen (1990) and Lam and Varoglu
(1990) found length effects higher on the 5 %-quantile than on the 50 %-quantile.
The reason for that is not at least seen in the tail-fitting executed in both citations.
As already discussed, applying the BFT alone coincides with violation of its
boundary conditions. Thus modifications on BFT were made to account for the
anisotropy of timber (Madsen and Buchanan 1985, Barrett et al. 1992) and by
adding of additional terms, to account for the influence of growth characteristics,
found by multiple regressions (Burger 1998). Later the spatial correlation was
considered by bi-modal distributions (Williamson 1992) or second- and third-
order hierarchical models (Riberholt and Madsen 1979, Kllsner et al. 1997,
Isaksson 1999, Khler 2005, Brandner 2012). These models account for the spatial
correlation of local strengths and the spatial distribution and magnitude of flaws
indicating local strength properties.
2 Length Effects in Timber Without Joints
The stochastic length effects in timber by means of a second-order hierarchical

model are discussed in brief. Readers interested in more details are kindly referred
e.g. to Brandner (2012). The hierarchical model allows to allocate the total disper-
sion of strength in that within and between elements. Adressing ideal serial
systems timber members composed of N 1 elements (weak zones, WZ) fail
suddenly with the failure of the weakest element, see
ft ,0 = R = min Z ij N = n = Y j + min X ij = Y j + X N , (2)

n
with Zij | j = Yj + Xij as random variable representing the strength of one weak zone
in member j, Yj and Xij as random variables representing the average strength of
member j and the ith deviations from the average, respectively, and N as the num-
ber of elements represented by weak zones, with i = 1, , N. The distribution
FR(r) of R is given by the convolution integral

(
FR ( r ) = P Y j + X N r N = n = ) fY ( y ) FX N ( r y n ) dy , (3)
0
with FX N ( x N = n ) according to Eq. (1). For iid X1 ~ 2pLND the distribution

function of XN can be well approximated by XN ~ 2pLND and with
CN 1
, with C = , , (4)
C1 ln ( N ) + 1 C
C
C1 and CN as characteristics of the distribution of X1 and XN, respectively, with

expectation E[X] = and variance Var[X] = 2 (Brandner 2012). Regression mod-
els for C and C, with C = , , are given in Table 1.
Table 1 Regression models for parameters C and C of Eq. (4), for C = , ; degree of
determination r2 and statistics of residua ei, with ei ~ ND ( e , se ) ; CV[.] as coefficient of
variation
Regression models r2 e se
2
= 0.826 CV [X1] + 2.925 CV[X1] > 0.99 0 0.024
= 0.806 CV 2[X1] + 2.858 CV[X1] + 0.681 0.98 0 0.007
= 0.005 CV 2[X1] + 0.160 CV[X1] + 0.309 0.99 0 0.045
= 0.253 CV 2[X1] 0.188 CV[X1] + 0.510 0.78 0 0.016
For modelling of the local tensile strength the regression models of Ehlbeck
et al. (1985) are used, see
( ) ( )
ln ft ,0,ij = 4.22 + ln Et ,0,ij ( 0.876 0.093 KAR ) + f , (5)
with correlation coefficient r = 0.86, [f] = 0.187, and with the local elastic
modulus Et,0,ij given as
( )
ln Et ,0,ij = 8.20 + l3.13 0 1.17 KAR + E , (6)
with r = 0.77, [E] = 0.180, 0 as oven dry density and KAR as knot are ratio.
Furthermore the tensile strength and length effects are discussed for three material
qualities, GI, GII and GIII. The parameter settings are given in Table 2.
Applying Eq. (5) gives CV[ft,0,ij] 31 %. For modelling of the tensile strength
of timber members at given length l and with the distance between the weak zones
dWZ ~ 2pLND also the number of serial elements, N 1, is lognormal distributed.
Eq. (3) can now be adapted to

FZ N ( z l ) = f N ( n ) FZ N ( z n ) dn , (7)
0
with N ~ 2pLND and l as the length of the analysed serial system, the timber
member. Interested in the main statistics the second term in the integral can be
approximated by Eq. (4). Thus Eq. (7) can be solved numerically or simply by
means of random variates; the second approach is applied. The main statistics of
the tensile strength of timber members at reference length lref = 2,000 mm, based
on 104 random variates, assuming iid realisations of dWZ, are given in Table 3. The
statistics base on pure tensile stresses. The interaction of moment and normal
stresses, caused by structural inhomogeneities (e.g. Colling et al. 1991) is not con-
sidered. As expected and in line with Williamson (1992) a decrease in the length
effect with increasing timber quality is observed.
Table 2 Parameter settings (expectation E[.] and coefficient of variation CV[.]) for groups
GI, GII and GIII
GI GII GIII
E[dWZ]; CV[dWZ] [mm]; [%] 450; 60 % 520; 60 % 590; 60 %
E[wWZ] 1); CV[wWZ] 1) [mm]; [%] 70; 40 %
E[KARWZ]; CV[KARWZ] [--]; [%] 0.23; 45 % 0.20; 45 % 0.17; 45 %
equi(ft,0) [--] 0.40 0.45 0.50
E[12] (E[0]); CV[] [kg/m]; [%] 410 (396); 8 % 460 (434); 8 % 500 (472); 8 %
E[N] 2); CV[N] 2) [--]; [%] 6.04; 60 % 5.23; 60 % 4.61; 60 %
1)
longitudinal extension of knot zones as defined in Brandner (2012)
2)
at reference length lref = 2,000 mm
Table 3 Main statistics of the tensile strength parallel to grain of timber members at
lref = 2,000 mm; groups GI to GIII
GI GII GIII
ft,0,mean; CV[ft,0] [N/mm]; [%] 29.4; 30.9 % 35.7; 30.6 % 43.1; 30.3 %
ft,0,05; ft,0,05,corr 1) [N/mm]; [N/mm] 17.1; 16.3 20.9; 19.9 25.3; 24.1
ft,0,mean / ft,0,ij,mean [--] 0.748 0.778 0.804
ft,0,05 / ft,0,ij,05 [--] 0.756 0.780 0.803
1)
bias corrected according to Brandner (2012)
Subsequently, the influence of parameters CV[ft,0,ij] = (20, 30, 40) % and

E[dWZ] = (400, 500, 600) mm on the strength of members, with l = (1.0 12.0) m,
CV[dWZ] = 60 % and equi(ft,0,ij) = 0.45, is investigated in reference to
ft,0 | lref = 2,000 mm, see Fig. 1.
In contrast to serial systems of iid elements, an effect higher on the 5 %-
quantile than on the expectation can be observed in case of equicorrelated ele-
ments. This is because only the variability within the elements is affected by the
serial system action, but the variability between the timber members remains.
Based on tests equal results were found by Madsen (1990), with kl,50 = 0.13 and
kl,05 = 0.22, and Burger (1998), based on a literature survey: kl,50 = 0.12 and
kl,05 = 0.18, his own results: kl,50 = 0.15 or 0.10 and kl,05 = 0.23 or 0.20 for un-
graded or visually graded structural timber without joints. Consequently, even a
minor but with CV[ft,0,ij] and length l increasing CV[ft,0] can be observed.
Fig. 1 Length effects on expectation and 5 %-quantiles of tensile strength and power re-
gression models given dWZ = 500 mm: timber members without joints
The results in Fig. 1 clearly indicate CV[ft,0,ij] as the dominating parameter on

the length effect. For common board material of Norway spruce (Picea abies)
with CV[ft,0] = (30 10) %, knowing that CV[ft,0] | lref = ft,0,B CV[ft,0,ij], and in
regard to current design rules it is proposed to consider a length effect on the 5 %-
quantile by a simple power model with power parameter kl,05 = 0.16, see
kl ,05
lref
ft ,0,05 l = ft ,0,05 lref . (8)
l
Dividing the material in two groups with CV[ft,0] = (25 5) % and

CV[ft,0] = (35 5) % (see Brandner and Schickhofer 2008), the parameters are
kl,05 = 0.13 and 0.21, respectively. Overall, our results are within the range of pub-
lished values. Influences by E[dWZ] are negligible.
3 Length Effects in Finger Jointed Timber Members
If stressed in tension parallel to grain timber members with finger joints can either
fail M-times in timber sub-members or (M 1)-times in finger joints, whereby
each timber sub-member consists of N equicorrelated timber elements. For simpli-
fication the jointed timber sub-members are further treated as iid although in
practise this is not always fulfilled. Furthermore, the number of finger joints per
reference length lref = 2,000 mm is kept deterministic.
For modelling of the resistance of finger joints it is required to define a rela-
tionship to that of the jointed sub-members. As the quality of finger joints depends
to a great extent on the production process this relationship is further defined as a
minimum requirement on the finger joint tensile strength; for the 5 %-quantile see
ft ,0,FJ,05 ( ) = 05 ft ,0,B,05 , (9)
with ft,0,B,05 and ft,0,FJ,05 as 5 %-quantiles of tensile strengths parallel to grain of

structural timber members without joints at lref = 2,000 mm and of single finger
joints, respectively, and with 05 as ratio of both strengths. Following assumptions

are made (Brandner 2012):
iid ft,0,FJ ~ 2pLND; CV[ft,0,FJ] = (15 5) %;
iid ft,0,B ~ 2pLND; CV[ft,0,B] = (30 10) %;
for simplicity it is assumed that the joining sub-members at given member
length and quantity of joints are equal in length.
Three definitions of minimum requirements on ft,0,FJ are discussed: [1] by secur-
ing equal reliability of members with and without joints at reference length, [2] by
securing equal 5 %-quantiles of members with and without joints at reference
length, and [3] by securing an equal failure probability of M jointed timber sub-
members and (M 1) finger joints in a jointed member at reference length, which
is equivalent to equal medians (50 %-quantiles).
Approach [1] is motivated by a probability based design. Due to the in general
lower variability in ft,0,FJ in comparison to ft,0,B the requirements on ft,0,FJ,05 are even
more relaxed than on ft,0,B,05. The influence of CV[ft,0,FJ] on 05 is negligible. The
failure of the jointed member is dominated by joint failure. Approach [2] is formu-
lated in regard to the current semi-probabilistic design concept where strengths
base on 5 %-quantiles. Of course, the influence of serial system action on the vari-
ability and thus on the partial safety factor is not taken into account. The influence
of CV[ft,0,FJ] and CV[ft,0,B] on 05 is small. The third approach [3] bases on a
balanced failure probability of both, timber and joints. Thereby a significant influ-
ence of CV[ft,0,ij] CV[ft,0,B] and CV[ft,0,FJ] on 05 can be found (see Brandner
2012).
Table 4 Parameter 05 in dependency of CV[ft,0,B] and the number of finger joints per refer-
ence length (excerpt from Brandner 2012)
1 FJ / lref 2 FJ / lref 3 FJ / lref

CV[ft,0,B] 25 % 30 % 35 % 25 % 30 % 35 % 25 % 30 % 35 %
05 | CV[ft,0,FJ] = 15 % 1.12 1.21 1.29 1.21 1.30 1.39 1.25 1.34 1.44
In regard to the current semi-probabilistic design concept and the balanced fail-
ure probability of joints and timber, approach [3] is used further in investigations
on the length effect in timber members with joints. Some values of 05 are pre-
sented in Table 4. For calculation of the length effect CV[ft,0,FJ] = 15 % was
applied. The results are presented in Fig. 2. In case of only one finger joint per
reference length the decrease in strength with increasing length is lower than in
case of members without joints. Of course with increasing number of joints this
fact becomes smaller or even reversed.
To account for current design rules, for practical applications and in particular
for the design of finger jointed structural timber members as part of a whole struc-
tural system, in principle and for simplification the same power parameters than
for members without joints can be applied, so far the average number of joints per
reference length does not exceed three.
Fig. 2 Length effects on expectation and 5 %-quantiles of tensile strength and power re-
gression models: timber members with joints
Madsen and Buchanan (1985) found a multiple-member factor (members with

joints) of kl,05 = 0.36 larger than the single member factor (members without
joints) of kl,05 = 0.29. In fact, both values are much higher than in our examina-
tions. This indicates that their material showed higher dispersing tensile strengths
in timber and finger joints. Comparing Fig. 1 and Fig. 2, the results at
CV[ft,0,B] = 40 % are nearly equal. It can be concluded that in cases of
CV[ft,0] > 40 % and / or more than three finger joints per lref a higher length effect
in finger jointed members than in members without joints results.
4 Conclusions
We addressed length effects on the tensile strength parallel to grain of timber

members with and without joints. The examinations focused on parameter set-
tings, common for structural timber and finger joints of Norway spruce (Picea
abies). Thereby a significant length effect on expectation and 5 %-quantile was
observed. A comparison with published test results concerning timber members
without joints showed that our results are in-line.
A slight increase in CV[ft,0] with increasing length was found. This small in-
crease is seen as one important reason for partly contradicting experimental data in
the literature. In particular the preference for the BFT, partly motivated by the
observation of nearly constant CV[ft,0] indicating the suitability of a 2pWD, can be
explained.
Current European design and product standards regulate length effects only for
members shorter than the reference length. As consequence of our and others re-
sults this procedure is definitely inadequate, physically contradicting and even
wrong. There is a need for action. Following current regulations, as shown in
Eq. (8), the power parameters for length effects in timber members without joints
or at least with two joints per lref = 2,000 mm are proposed as kl,05 = 0.16, 0.13 and
0.21 for CV[ft,0,B] = (30 10) %, (25 5) % and (35 5) %, respectively. Of
course these values are only valid for timber members within the hierarchical
material level of structural timber. Thus the proposal is limited, e.g. to

1 m l 20 m.
With focus on length effects of timber members with joints a positive relation-
ship between the power parameter and the number of finger joints per reference
length was found. Aiming on regulations for the minimum requirements on finger
joint tensile strength by ft,0,FJ,05 05 ft,0,B,05, values for 05 = 1.30, 1.20 and 1.40
are proposed for CV[ft,0,B] = (30 10) %, (25 5) % and (35 5) %, respectively.
The presented models allow considering directly the stochastic nature of
strength properties, their longitudinal and even spatial distribution and correlation.
Variability and equicorrelation can be explicitly applied according to the material
quality and the production requirements. The basic model allows also verifying
length effects relevant for different design concepts, e.g. deterministic, semi-
probabilistic and probabilistic. It was applied for investigations on length effects
of timber members with and without joints. This enhances directly the abilities in
modelling and optimisation of existing linear but also two-dimensional structural
members. In addition, the presented concepts and models support the design
process required for the definition of new products, and the judgement of new
structural concepts or structures in uncommon dimensions.
References
Barrett, J.D., Lam, F., Lam, W.: Size effects in visually graded softwood structural lumber.
Paper Presented at the 25th Meeting of CIB-W18, hus, Sweden (August 1992)
Baant, Z.P.: Probabilistic modelling of quasibrittle fracture and size effect (2001),
http://www.civil.northwestern.edu/people/bazant/PDFs/
Backup%20of%20Papers/S44.pdf (accessed April 17, 2013)
Bohannan, B.: Effect of size on bending strength of wood members (1966),
http://www.fpl.fs.fed.us/documnts/fplrp/fplrp56.pdf (accessed
April 17, 2013)
Brandner, R., Schickhofer, G.: Glued laminated timber in bending: new aspects concerning
modelling. Wood Science and Technology 42(5), 401425 (2008)
Brandner, R.: Stochastic System Action and Effects in Engineered Timber Products and
Structures. Dissertation, Graz University of Technology (2012)
Burger, N.: Einfluss der Holzabmessungen auf die Zugfestigkeit von Schnittholz unter
Zugbeanspruchung in Faserrichtung. Dissertation, Technische Universitt Mnchen
(1998)
Colling, F., Ehlbeck, J., Grlacher, R.: Glued laminated timber: contribution to the deter-
mination of the bending strength of glulam beams. Paper Presented at the 24th Meeting
of CIB-W18, Oxford, United Kingdom (September 1985)
Ehlbeck, J., Colling, F., Grlacher, R.: Influence of finger-jointed lamellae on the bending
strength of glulam beams: input data for the computer model. Holz als Roh- und
Werkstoff 43, 369373 (1985)
Galileo, G.L. Discorsi I Demostrazioni Matematiche interno due Nuove Scienze. Elsevier,
Leiden (1638); English edition: Westen T, London (1730)
Gibrat, R.: Une Loi Des Repartitions Economiques: Leffet Proportionelle. Bulletin de
Statistique General, France (1930)
Griffith, A.A.: The phenomenon of rupture and flow in solids. Philos. Trans. Roy. Soc.
London A 221, 163198 (1920)
Isaksson, T.: Modelling the Variability of Bending Strength in Structural Timber. Disserta-
tion, Lund University (1999)
Kllsner, B., Ditlevsen, O., Salmela, K.: Experimental verification of a weak zone model
for timber in bending. In: IUFRO S 5.02 Timber Engineering, Copenhagen, Denmark
(1997)
Khler, J.: Reliability of timber structures. Dissertation, Swiss Federal Institute of Technol-
ogy Zurich (2005)
Lam, F., Varoglu, E.: Effect of length on the tensile strength of lumber. Forest Products
Journal 40(5), 3742 (1990)
Madsen, B., Buchanan, A.H.: Size effects in timber explained by a modified weakest link
theory. Paper Presented at the 18th Meeting of CIB-W18, Beit Oren, Israel (June 1985)
Madsen, B.: Length effects in 38 mm spruce-pine-fir dimension lumber. Canadian Journal
of Civil Engineering 17, 226237 (1990)
Mariotte, E.: Trait due movement des eaux. Mdela Hire, London (1686); English edition:
Desvaguliers JT, London (1718)
Peirce, F.T.: Tensile tests for cotton yarns v. the weakest link Theorems on the strength
of long and of composite specimens. J. Textile Inst. Tran. 17(7), 355368 (1926)
Riberholt, H., Madsen, P.H.: Strength distribution of timber structures measured variation
of the cross sectional strength of structural lumber. Technical University of Denmark,
Report, No. R 114 (1979)
Sutherland, L.S., Shenoi, R.A., Lewis, S.M.: Size and scale effects in composites: I. Litera-
ture review. Composites Science and Technology 59, 209220 (1999)
Tippett, L.H.C.: On the extremes of individuals and the range of samples taken from a
normal distribution. Biometrika 17(3/4), 364387 (1925)
Weibull, W.: A statistical theory of the strength of materials. IVA, Handling Nr. 151, Royal
Swedish Institute for Engineering Research (1939)
Williamson, J.A.: Statistical models for the effect of length on the strength of lumber. Dis-
sertation, University of British Columbia (1992)
New Perspectives in Machine Strength Grading:
Or How to Identify a Top Rupture
Julia K. Denzler* and Andreas Weidenhiller
Holzforschung Austria, Franz-Grill-Strasse 7, 1030 Vienna, Austria

j.denzler@holzforschung.at
Abstract. An important aspect in machine strength grading of timber is the

prediction of strength. This property is mainly based on grain direction and its
deviations from a main direction, which can cause a dramatic loss of strength.
Therefore, an essential demand of the wood industry is to evaluate the direction of
wood fibers in a fast and non-invasive way to identify specimens with low
strength values. For three decades now polarized microwave radiation has been
investigated to identify the main direction of wood fibers in timber in a non-
contact and non-destructive way. This paper deals with the use of microwave
radiation to identify specimens with severe slope of grain to be able to reject them
in the grading process.
Keywords: grain deviation, timber, strength grading, identification.
1 Introduction
Within the last thirty years the development in machine strength grading was
highly pushed forward. From an initial idea to predict strength, stiffness and
density of sawn timber a significant development in sawmilling industry took
place in the last years which has a great impact on economy and product safety.
This development is also present in the techniques used to grade timber: First
machines used bending processes. Later versions work contactless by using
longitudinal frequency e.g. in combination with x-ray technology or camera
systems nowadays. The timber quality of graded material increased due to these
efforts. Unfortunately, the detection of one main characteristic still is not solved
satisfyingly: Grain deviations are exceptionally difficult to identify, but can cause
severe safety problems in the use of the product.
Grain deviations reduce especially the strength of sawn timber. They appear
globally over the whole length or more local over a limited length (Fig. 1). Global
grain deviations mainly result from growth properties of the log like sweep or
*
DOI: 10.1007/978-94-007-7811-5_68, RILEM 2014
762 J.K. Denzler and A. Weidenhiller
spiral grain and influence the whole piece of sawn timber. Local grain deviations
are only present in a limited area of the sawn timber normally resulting from knots
or top rupture. Typical global deviations are small in their absolute value, while
local grain deviations may alter fiber direction very strongly. Bending, tensile, and
compression strength decrease up to 1/10 of the original value while the slope of
grain changes from 0 to 90 (Kollmann 1968). Sawn timber with these low
strength values is unsuitable for construction. Specimens including severe slope of
grain still have to be rejected visually during the grading process. But also for
human beings this characteristic is difficult to identify, especially within the
available time frame of only 1-2 seconds in the grading process for one specimen.
To be able to predict some of these grain deviations in sawn timber knot
detection was developed. The knot size is highly correlated with grain deviation
around the knot. Visual grading as well as some machine strength grading
methods e.g. x-ray use this principle and measure the size of knots to predict grain
deviations and last but not least strength of sawn timber. Severe slope of grain
without the presence of knots or based on top rupture stays undetected.
global
grain
deviation
sweep spiral grain taper
local
grain
deviation
slope of grain slope of grain top rupture
with knot without knot
Fig. 1 Grain deviations and their origin
Industry is looking for alternatives to identify specimens with severe slope of

grain by machine readings. This paper deals with a method to detect local and
global slope of grain automatically, aiming to develop a stable and high developed
machine grading principle. This is realized using microwave transmission.
New Perspectives in Machine Strength Grading 763
2 Literature Review
Microwaves are used for non-contact determination of slope of grain in timber for
some decades now. During the transition of microwaves through wood some basic
characteristics of the waves are changing (Torgovnikov 1993): attenuation and
phase shift take place, mainly caused by water embedded in fibers. The electric
field vector initially linearly polarized becomes elliptically polarized. These
modifications can be measured and the complex transmission coefficients along
(u) and across fibers (v), as well as the grain direction can be determined.
Furthermore, the estimation of moisture content and density is possible (James et
al. 1985; Orhan 2002; Nguyen et al. 2004; Schajer & Orhan 2005; Baradit et al.
2006; Lundgren 2007).
Tiuri et al. (1979) and James et al. (1985) developed initial methods for
determination of slope of grain, density and moisture content based on microwave
transmission. Heikkila et al. (1982), Shen et al. (1994), Leicester et al. (1996) and
Malik et al. (2005) built sensor systems for transmission measurements. Schajer et
al. (2005), Malik et al. (2005) and Bogosanovic et al. (2009) found methods of
microwave data analysis with respect to grain angle.
At the moment, two techniques are most commonly used for non-contact
microwave testing of wood. Based on free space transmission, the focused beam
technique, using transmitting and receiving horn antennas and lenses is reported,
e.g. by Malik et al. (2005). This method is not able to measure grain deviations
locally, since horn antennas in use are immobile and large in size. Secondly the
modulated scattering technique is used, enabling more local information along the
wood surface. Density and moisture content (Johansson 2003; Lundgren 2007) as
well as slope of grain (Leicester and Seath, 1996) have been measured on local
scales using modulated scattering technique. Bogosanovic et al. (2010) give a
comprehensive overview of microwave non-contact measurement techniques.
Beside the microwave technique the tracheid laser technology was developed
within the last 5 years and is used to identify slope of grain via the surface of a
piece of timber (Olsson et al. 2013, Hunger & van de Kuilen 2010). If a laser dot
lights a timber specimen, the original round dot will be deformed to an ellipse
with its main axis in grain direction. These signals can be used to detect the slope
of grain on all four surfaces of a specimen and can be combined to stereoscopic
information. Disadvantages of this method clearly are the prospect of a three
dimensional information based on a two dimensional information and the
difficulty to detect slope of grain diving in the middle of a board with no severe
grain deviations on its smaller sides.

3.1 Material
The material was chosen with respect to the task: An abnormally high amount
of specimens with a top rupture is included in the sample made of spruce
(Picea abies karst.). These specimens were sampled at DOKA, producer of

formwork beams consisting of a web made of different materials and two chords
made of sawn timber with a cross section of 40 x 80 mm. For this product,
specimens with a strength below the 5th percentile due to severe grain deviations
can cause severe damage combined with a loss of product confidence.
The length of the specimens was approximately 4 m. Tab. 1 summarizes the
basic properties of two samples: The first sample was used for training, the second
sample for testing.
Table 1 Basic test data separated for specimen's sample
sample series no. moisture u strength f_m stiffness E density

in % in N/mm in N/mm in kg/m
(COV in %) (COV in %) (COV in %) (COV in %)
1 regular 86 9.6 ( 3) 32.6 (37) 11 904 (22) 448 (10)
(training) top rupture 97 9.3 ( 5) 29.4 (32) 11 109 (17) 449 ( 8)
total 183 9.5 ( 4) 30.9 (35) 11 483 (20) 448 ( 9)
2 regular 60 10.1 ( 7) 33.7 (31) 12 595 (19) 454 (10)
(test) top rupture 56 9.5 ( 6) 24.5 (38) 10 505 (16) 433 (10)
total 116 9.8 ( 7) 29.2 (38) 11 586 (20) 443 (10)
3.2 Methods
3.2.1 General on Microwave Transmission
The basic physical principles during the transition of microwaves through timber
are damping and phase shift. The waves are damped, mostly by the
water embedded. As the water content changes parallel and perpendicular to
grain, damping and phase shift of the waves are changed differently for these
two directions (Torgovnikov, 1993). Fig. 2 shows that the damping parallel
to grain (attu) is bigger than the damping perpendicular to grain (attv).
Respectively, the phase shift parallel to grain (u) differs from the one orthogonal
to grain (v).
Testing the three-dimensional distribution of damping and phase shift in timber
is done using microwaves of two orthogonal planes. The mathematical
superposition of the resulting waves, leads to an electromagnetic field vector
which is rotating along an elliptic path. The semi-minor axis of this ellipse is in
line with the grain direction (Fig. 2). Measuring the angular position of the ellipse
yields the grain angle parallel to the face side. Comparable mathematical models
based on two microwave transmission in different angles to the timber surface
lead to the grain angle rectangular to the face side.
A B
A-A E B-B
A0,u attu Au
A0,v A0 Av
Au
x
A0,u = A0,v = A0
u
A B
E
attv
A0
Av
x z
v ,grain
ull grain
v x
Fig. 2 Principle of transition of microwaves through wood
3.2.2 Specimens Categories

To get information about the accuracy and the possibilities of the microwave
equipment, the specimens have been separated for two series by hand prior to
testing: Specimens not including a grain deviation are called "regular". Specimens
including a top rupture are called "top rupture". Fig. 3 shows some examples of
these two series.
After destructive testing, two additional categories were formed differentiating
between specimens which collapse by the influence of top rupture and specimens
which did not break by the influence of a top rupture. This differentiation is
difficult to perform but a trained person who can evaluate the breakdown reasons
of sawn timber is able to assess the rupture.
a) "regular" specimen b) "top rupture" c) "top rupture"

specimen specimen
(II face side) ( face side)
Fig. 3 Examples for specimens series

3.2.3 Non Destructive Testing and Its Parameters

Each specimen was measured with the microwave prototype built by institute
EMCE of the Vienna University of Technology. Fig. 4 shows an impression of
this prototype. In total, 3 receiving antennas arranged in different angles to the
timber surface and 28 transmitting antennas arranged in two rows perpendicular to
the timber surface can be used to detect grain angles in specimens up to 60 x
300 mm cross section.
Fig. 4 Prototype at Holzforschung Austria
The measurements take place with 12 GHz resulting in a resolution of 12.5 x

12.5 mm on the specimens width. The specimens thickness is evaluated as a
whole so that information about the grain deviation of cuboids with 12.5 x 12.5 x
42.0 mm exists. In total more than 350 different parameters are calculated out of
the microwave transmission. To identify the most suitable ones for the
identification of top rupture specimens linear regression analysis was used.
Additionally, the non destructive measurements of knots, frequency and weight
took place in laboratory.
3.2.4 Destructive Measurement and Its Parameters

All specimens were tested in tension according to EN 408 and EN 384.
Additionally to the categories given to the specimens prior to testing, the reason of
failure was documented especially if an included top rupture influences the
collapse (see section 3.2.2). Specimens where the top rupture influences the
collapse are called "top rupture", specimens where the top rupture was included
but did not influence the collapse are called "no top rupture".
4 Results
4.1 Identifying Specimens Including a Tree Top Rupture

As the main goal of this work is to identify specimens with severe local slope of
grain, specimens with top rupture are over-represented in the population. For the
training sample 97 out of 183 specimens included a top rupture in different
intensity. Fig. 5 shows the two predictors chosen to identify the top rupture for
these 183 specimens.
Predictor 1 includes the grain deviation measured with the prototype over a
length of 75 mm and a specimen`s width of 50 mm. The predictor includes in
principle the total width of the specimen albeit a safety border has to be used
because edge effects and refraction will interfere with the measurement (Denzler
et al. 2013). The microwaves pass through the whole thickness of 42 mm of the
board. Within this volume, the predictor 1 calculates the absolute mean grain
angle on the boards faces, slides per 12.5 mm and does the same again. Predictor 1
includes the maximum of this value per board.
Predictor 2 uses the same information, but performs the grain angle difference
of two following volumes.
Fig. 5 shows that the prototype is able to identify 84% "top rupture" specimens
correctly using predictor 1 >= 4 or predictor 2 >= 3. On the other hand 28%
"regular" specimens are identified to include severe local slope of grain, probably
caused by knots. To find out if these specimens show severe grain deviation
caused by knots, Fig. 6 shows the strength values evaluated in destructive testing
versus predictor 1. 7 out of 12 specimens with a value for predictor 1 >= 4 have a
strength lower than 25 N/mm.
To show that this is not only working for the sample where the parameters
where chosen, Fig. 7 shows similar results obtained using the test sample with 116
specimens. 40 specimens included a top rupture. 86 % "top rupture" specimens are
identified correctly using the same predictors and limits. In this case, only 20 %
"regular" specimens are identified to have a top rupture.
series
top
regular
rupture
72% 16%
28% 84%
Fig. 5 Identifying specimens in the training sample including a top rupture separated for
series, n = 183 specimens
Fig. 6 Strength versus predictor 1 separated for series, n = 183 specimens
series
top
regular
rupture
80% 14%
20% 86%
Fig. 7 Identifying specimens in the test sample including a top rupture, n = 116 specimens
4.2 Identifying Specimens Failing Due to the Impact of the Top

Rupture
This chapter only deals with the specimens including a top rupture. For the
training sample 97 out of 183 specimens are included in the series "top rupture",
for the test sample 56 out of 116 specimens are included in this series. Fig. 8
shows that 89% of 97 specimens including a top rupture fail due to the impact of
the top rupture. The results indicate that 86 of these 97 specimens can be
identified by predictor 1 and predictor 2 correctly.
Fig. 9 shows comparable results for the test sample: 90% of the specimens
including a top rupture failed with an impact of top rupture and are identified
correctly. This results in 50 out of 56 specimens.
collapse
no top rupture
top rupture
collapse
no top top
rupture rupture
25% 11%
75% 89%
Fig. 8 Identifying the impact of top rupture on collapse for series "top rupture" of the
training sample, n = 97 specimens
collapse
no top rupture
top rupture
collapse
no top top
rupture rupture
25% 10%
75% 90%
Fig. 9 Identifying the impact of top rupture on collapse for series "top rupture" of the test
sample, n = 56 specimens
5 Discussion
The results show the potential using microwave transmission to identify

specimens with slope of grain. So far, the results are limited to one relatively small
cross section made of spruce which clearly limits the assignment of the results to
timber industry as a whole. It can be expected that increasing cross sections will
lead to a more differentiated impact of local slope of grain.
The results given in this paper are promising: More than 80% of the specimens
with slope of grain or top rupture, respectively, are identified correctly. More than
85% of the identified specimens break with an impact of the top rupture included.
6 Summary
This paper deals with the possibilities to use microwave transmission to identify
specimens with severe slope of grain. Especially the identification of pieces with
low strength values has a high impact on safety issues. The aim is to reject these
pieces as they contribute to strength values lower than 5th percentile to an
extended amount.
The principle of microwave transmission allows the identification of grain
deviation over the total thickness of the board for different volumes. Two
predictors are identified to be able to detect more than 80% of the specimens
including a top rupture correctly on a training sample. These values are confirmed
with an independent test sample.
Additionally, the impact of identified grain deviation on failure was
investigated. In this case, more than 85% of the specimens including a top rupture
fail due to the presence of top rupture.
Microwave technology shows potential to complement or replace some grading
procedures and to improve the identification of weak pieces including top rupture.
Apart from that, the contactless measurement of moisture content and density
seems feasible. Therefore, the possibility to detect moisture content, density and
grain deviation with one system shows great potential for future use in both small
and large enterprises.
Acknowledgement. This work was supported by the Austrian Research Promotion Agency
- FFG and Wirtschaftsagentur Wien ZIT within the COMET K-project HFA-TiMBER
A.1.3, (project nr. 820501). We also thank our project partners DOKA Group Austria,
MiCROTEC s.r.l, the institute EMCE of the Vienna University of Technology and BOKU.
References
Baradit, E., Aedo, R., Correa, J.: Knots detection in wood using microwaves. Wood Sci.
Technol. 40, 118123 (2006)
Bogosanovic, M., Anbuky, A.A., Emms, G.W.: Microwave measurement of wood in
principal directions. IEEE Sensors, 252256 (2009)
Bogosanovic, M., Anbuky, A.A., Emms, G.W.: Overview and comparison of microwave
noncontact wood measurement techniques. J. Wood Sci. 56(5), 357365 (2010)
Born, M., Wolf, E.: Principles of Optics. Cambridge University Press, UK (1999)
Bucur, V.: Nondestructive characterization and imaging of wood. Springer, Heidelberg
(2003)
Denzler, J.K., Koppensteiner, J., Arthaber, A.: Grain angle detection on local scale using
microwave transmission. International Wood Products Journal 4(2), 6874 (2013)
Heikilla, S., Jakkula, P., Tiuri, M.: Microwave methods for strength grading of timber and
for automatic edging of boards. In: Proceedings of the 12th European Microwave
Conference, pp. 599603 (1982)
Hunger, F., van de Kuilen, J.-W.: Improved Grading by detecting local slope of grain.
Presentation on Gradewood Meeting in Paris (September 9, 2010)
James, W.L, Yen, Y.H., King, R.J.: A microwave method for measuring moisture content,
density and grain angle of wood. Forest Products Laboratory, Research Note FPL-0250
(1985)
Johansson, J., Hagman, O., Fjellner, B.A.: Predicting moisture content and density
distribution of scots pine by microwave scanning of sawn timber. J. Wood Sci. 49(4),
312316 (2003)
Kollmann, F.F.: Principles of wood science and technology. Springer, Berlin (1968)
Leicester, R.H., Seath, C.A.: Application of microwave scanners for stress grading. In:
Conference Proceedings - International Wood Engineering Conference, vol. 2, pp. 435
440 (1996)
Lundgren, N.: Microwave sensors for scanning of sawn timber. Doctoral thesis, Lulea
University of Technology, Skelleftea, Sweden (2007)
Malik, S.A., Ghodogoankar, D.K., Hambaly, A.M., Majid, W.M., Nuruddin, M.F.:
Measurement of wood grain angle using free-space microwave system in 8-12 GHz
frequency range. In: Proceedings of the Asian Conference on Sensors and the
International Conference on New Techniques in Pharmaceutical and Biomedical
Research, pp. 213218 (2005)
Nguyen, M., Leicester, R., Foliente, G., Seath, C.: Identitfication of strength-reducing
characteristics in lumber using microwave scanners. Key Engin. Mat. 270-273, 1513
1520 (2004)
Olsson, A., Oscarsson, J., Serrano, E., Kllsner, B., Johansson, M., Enquist, B.: Prediction
of timber bending strength and in-member cross-sectional stiffness variation on the basis
of local wood fibre orientation. Eur. J. Wood Prod. 71, 319333 (2013)
Orhan, F.B.: Estimating wood physical properties using microwave measurements. MAS
thesis, University of British Columbia, Vancouver (2002)
Schajer, G.S., Orhan, F.B.: Microwave non-destructive testing of wood and similar
orthotropic materials. Subsurf. Sens. Technol. Appl. 6, 293313 (2005)
Shen, J., Schajer, G.S., Parker, R.: Theory and practice in measuring wood grain angle
using microwaves. IEEE Trans. Instrum. Meas. 43, 803809 (1994)
Tiuri, M., Heikkila, S.: Microwave instrument for accurate moisture measurement of
timber. In: Proceedings of the 9th European Microwave Conference, pp. 702705
(1979)
Torgovnikov, G.I.: Dielectric properties of wood and wood-based materials. Springer,
Heidelberg (1993)
Aspects of the Difference between the Local
and Global Modulus of Elasticity of Structural
(hardwood) Timber
G.J.P. Ravenshorst1, P.A. de Vries1, and Jan-Willem van de Kuilen1,2

1
Delft University of Technology, The Netherlands
g.j.p.ravenshorst@tudelft.nl
2
TU Mnchen, Germany
Abstract. The modulus of elasticity of structural timber (MOE) may be deter-

mined by 2 methods according to the European standard EN 408 [1]. The ratio
between the so-called local and global MOE that is found from tests according to
these two methods, cannot be explained by the ratio between the MOE and shear
modulus G that is assumed in the strength class tables of EN 338 [2]. The relation-
ship between MOElocal and MOEglobal from EN 384 [3] is not consistent with
the shear values given in [2]. In this study the shear modulus for samples of the
tropical hardwood species massaranduba and of softwood species spruce was
determined. The shear modulus G was found to be not related to the MOE and was
shown to be constant at around 550 N/mm2 for massaranduba and 190 N/mm2 for
spruce. With these values, the ratios between MOE-local and MOE-global that
were found in the test series could be explained. The found values for the shear
moduli differ from previous research. The study concludes that it is unclear which
parameters determine the magnitude of the shear modulus of a single piece of
timber and that this needs to be investigated.
Keywords: structural (hardwood) timber, ratio of local and global Modulus of

Elasticity, shear modulus, weak zones.
1 Introduction
The Modulus of Elasticity of structural timber can be determined by 2 methods
according to the European standard EN 408. The first method is based on
measurements of the deformation within the zone with a constant bending moment
in a 4-point bending test arrangement. The relation between the applied force and
the measured deformation is used to calculate the so-called MOElocal of the beam.
The second method uses the same four point bending test, but now the mid-span
deflection relative to the supports of the test piece is measured. In this case, the
total deformation is caused by both bending and shear deformations in the beam.
In the 2010+A1:2012 version of EN 408 [1] the expression for this so-called
MOEglobal includes the shear modulus G. However it is stated that the shear
DOI: 10.1007/978-94-007-7811-5_69, RILEM 2014
774 G.J.P. Ravenshorst, P.A. de Vries, and J.-W. van de Kuilen
modulus G should be taken as infinite when the MOEglobal is used to determine the
MOElocal according to the relationship in EN 384, which should be used for
strength class assignments. Therefore in this paper MOEglobal is defined as the
property calculated out of deflection measurements where the shear modulus G
taken as infinite in the calculation.
EN 384 [3] gives the following equation to calculate MOElocal out of the
MOEglobal:
= 1,3 2690
This equation is derived from a dataset with results for softwood specimens,
consequently with a limited range of MOE values. For low values and for high
values of MOEglobal,G= the resulting values for MOElocal are doubtful.
In the strength class tables of EN 338, the shear modulus G is given as
MOE local divided by 16. When this ratio is used to calculate G for individual test
pieces, the ratio between MOEglobal and MOElocal that is found in literature cannot
be explained adequately with the equation for MOEglob in EN 384. Under the as-
sumption that the MOE does not vary along the beam, the expected ratio between
MOElocal and MOEglobal is approximately 1.04 for a ratio MOE local/G =16.
However, Ravenshorst and Van de Kuilen[4] report tests on softwood, temperate
hardwoods and tropical hardwoods in which a ratio MOElocal/MOEglobal of around
1.15 is found for all species. Brandner et al. [5] report from literature that
MOElocal/G values for softwoods can vary between 12 to 36, and that the shear
modulus G for structural softwood timber could be more or less constant around
G=600 N/mm2, slightly increasing for higher MOE.
The aim of this study is to explain the differences between tested MOElocal and
MOEglobal for 2 samples of a tropical hardwood species and 1 softwood sample, to
test the quality of the EN 384-equation and to evaluate the assumed dependency of
the shear modulus with MOElocal.
2 Theoretical Considerations
The difference in the local and global Modulus of Elasticity of structural timber
are influenced by the following two phenomena:
The shear deformation in the side parts of the test set-up.
The influence of zones with low bending stiffness.
2.1 The Shear Deformation in the Side Parts of the Test

Set-Up
As mentioned in the introduction, there is no consensus in literature for the magni-
tude of the shear modulus G in relation to the modulus of elasticity for softwoods.
For tropical hardwoods there is almost no data available for the shear modulus of
structural timber. The effect of the shear modulus is therefore studied on both
softwood and tropical hardwood.
Aspects of the Difference between the Local and Global Modulus of Elasticity 775
2.2 The Influence of Weak Zones with Low Bending Stiffness

In the standard EN 384 an equation for the MOElocal is given when the MOEglobal
determined. According to this equation MOElocal can become lower than
MOEglobal. Denzler et al. [6] confirmed this phenomenon. The relevant studies are
mostly done on softwood, so including pieces with low MOE, with a lower limit
of approximately 5000 N/mm2. That MOElocal becomes lower than the MOEglobal
can only be explained by the occurrence of a zone with a low bending stiffness
within the MOElocal area, which has more influence than the deflection due to
shear in the outer parts. This seems logical as it is prescribed that the weakest part
should be placed in the center of the span. However, for most tropical hardwoods,
the only stiffness reducing characteristic is the grain angle deviation. Since there is
no clear visible weak part, it can be assumed that the MOE is more constant over
the beam length than for softwoods with knots, which cause low local bending
stiffness. This is another argument to test and compare a tropical hardwoods spe-
cies and a softwood species. In this paper the zones with low bending stiffness
will be referred to as weak zones.
3.1 Materials
In table 1 the materials that are used are listed. The tropical hardwood species
massaranduba has no visible weak zones due to defects like knots, so a constant
value of the MOE and G can be assumed over the length of the beam. Two samples
of massaranduba with different dimensions and different moisture contents are stud-
ied. In the softwood sample of spruce knots, causing weak zones are be present.
Table 1 Description of the hardwood and softwood samples used in the study
Dimensions Mean mois-

Sample Mean den-
n Wood species Botanical name source t (mm) x ture content
code sity (kg/m3)
h (mm) (%)
M1 26 Massaranduba Manilkara spp. Brazil 60 x 150 20.2 1088
M2 40 massaranduba Manilkara spp Brazil 50 x 100 11.9 979
S1 31 spruce Picae abies Austria 40 x200 9.0 453
3.2 Methods
3.2.1 Test Set-Up
For all measurements a 4-point bending test arrangement was used, similar to
the setup described in EN408. In the test arrangement special care was taken to
minimize local restrain of deformations. The detailing of both supports and the
devices used to apply the loads ensure this (see figure 1).
MOElocal and MOEglobal were determined in accordance with EN408, where for
the determination of MOEglobal the relation between the applied force and the dis-
placement of the section of the beam where the load is applied (points P in figure
1) was used instead of the displacement of the middle section of the beam. The
displacement of point P was derived from the displacement of the actuator that
applied the force F. Effectively corrections were made for the stiffness of the test
rig elements as well as for local indentations at both supports and loading point
locations.
Fig. 1 Test set-up
To investigate the influence of the shear deformation, the distance A from the
support to the location where the load is applied (points P in figure 1) was varied.
For all specimens the relation between the load F and the deflection wp was meas-
ured for A = i.h, where i=2,3,4,5,6. These 5 measurements were used to derive an
effective shear modulus G and an effective MOE for each specimen.
Using Timoshenko beam theory the deflection wp of point P caused by load F
in the test setup in figure 1 is equal to:
F A2 B F A3 F ks A The measurements for A = i.h result each
wp = + +
4 EI
6 EI G
2 b
h
bending shear
dwp
in a value for
dF i
wp A2 B A3 ks A
= + +
F 4 EI
6 EI
G
2 b
h
bending shear

wp 1 (i h) 2 6h (i h)3 ks i h
= + +
F i E 4 1 bh3 6 1 bh3 2 G b h
12 12
wp 1 18 i 2 + 2 i 3 ks i
= +
F i E b 2G b
dwp

dF i 1 1 ks 1 1 1
= + MRi = + Fi

(
18 i 2 + 2 i 3 E 18 i + 2 i 2 2 G

) E G

b
Fi
MRi
For a rectangular section ks = 1.2. This yields the f values for Fi according to
table 2.
The slope of the linear trend line through the points (Fi, MRi) now yields the re-
ciprocal value of the effective shear modulus G for the specimen. The cutting
point of the trend line with the vertical axis is the reciprocal value of the MOE for
pure bending of the beam. This value will be tagged MOEbeam.
Table 2 Values for Fi
i Fi MR2 trend line
MR
2 0.0136 MR4
MR6
3 0.0083
1/MOEbeam
4 0.0058
5 0.0043 F6 F5 F4 F3 F2
F
6 0.0033
Fig. 2 Principle for deriving of MOEbeam and
G out of MOEglobal measurements
4 Results
In table 3 the values for MOElocal and MOEglobal are given and the ratio between
them. The values found are in line with the ratios found in Ravenshorst and van de
Kuilen [4]. Figure 3 shows a scatter plot for the 3 samples with regression lines
forced through the origin. What can be observed is that the 2 massaranduba sam-
ples almost follow the same regression line but that sample M2 is shifted giving
higher values for both MOElocal and MOEglobal. Thisis is caused by the difference
in moisture content between the 2 samples.
Table 4 gives the values for MOEbeam and G determined with the variable span
method, the ratios between MOElocal and G and the ratio between MOEbend and
MOElocal. Table 4 shows that ratios between MOElocal and G are much higher than
the value of 16 which is used in the strength class tables of EN 338. Remarkable
are the low values for G. Particular for the spruce sample very low values for G
are found. (Remark: to verify these values with an alternative method the dynamic
torsional stiffness Gdyn was determined for samples M2 and S1. Values found with
this method gave similar results as the static G).
Table 3 Test results for MOElocal and MOEglobal
MOElocal MOEglobal MOElocal/MOEglobal

Sample code
Mean (N/mm2) cov Mean(N/mm2) cov mean
M1 22580 0.17 19500 0.15 1.16
M2 26100 0.12 22580 0.11 1.15
S1 13190 0.21 11680 0.21 1.14
Table 4 Calculated values for MOEbeam and G
MOEbeam G MOElocal/G MOElocal/MOEbeam

Sample
code Mean
cov Mean(N/mm2) cov mean mean
(N/mm2)
M1 23200 0.17 500 0.12 45.1 0.97
M2 25220 0.12 590 0.22 44.3 1.04
S1 14790 0.27 190 0.26 69.4 0.89
40000
M2
35000 y = 1.16x
R = 0.67
30000
S1
MOElocal (N/mm2)
25000 y = 1.12x
R = 0.61
20000
M1
15000 y = 1.16x
R = 0.91
10000
S1
5000 M2
M1
0
0 5000 10000 15000 20000 25000 30000
MOEglobal (N/mm2)
Fig. 3 Scatter plots for MOElocal against MOEglobal for the tested samples
5 Analysis
5.1 Prediction of MOEglobal

To study the influence of G on the ratio between MOElocal and MOEglobal, the
MOEglobal is predicted out of the found values of MOEbeam and G, both determined
with the variable span method for every test piece. Out of the share of deflection
of both properties the MOEglobal;pred can then be calculated, assuming an infinite
value for the shear modulus, as if the beam was tested according to EN 408.
The found MOEbend and G-values for a piece are average values over the entire
beam (MOEbeam) and the end parts (G). This is a simplification of reality, because
the weak zones might have a local influence, especially for softwoods. This can
explain that the ratio between MOElocal and MOEbeam deviates more from 1 for
softwoods than for hardwoods. Since the weak zone is placed in the MOElocal zone
it can be expected that the MOElocal for softwood is lower than the MOEbeam, which
is confirmed with the value of 0,89 between MOElocal and MOEbeam for softwood.
In table 5 the mean values for the calculated MOEglobal,pred are given, together
with the ratio between MOElocal and MOEglob,pred. In figure 4 the MOEglobal, is plot-
ted against the MOEglobal,pred. Figure 4 shows that the MOEglobal;pred gives a good
prediction of MOEglobal. This gives evidence that the shear modulus G is the main
cause for the difference between MOElocal and MOEglobal.
Table 5 Calculated values for 30000
MOEglobal;pred M2
25000 y = 1.00x
R = 0.71
M1
S1
MOEglobal (N/mm2)
20000 y = 0.95x
MOElocal/ y = 0.97x
R = 0.89
MOEglobal;pred R = 0.74
Sample MOEglob;pred 15000
code Mean 10000

cov Mean
(N/mm2) S1
5000 M2
M1 20400 0.15 M1
1.11
0
M2 22480 0.11 1.16 0 5000 10000 15000 20000 25000 30000
MOEglobal;pred(N/mm2)
S1 11830 0.20 1.12
Fig. 4 Scatter plots for MOEglobal against the calcu-
lated MOEglobal;pred or the tested samples
5.2 Dependency of G from MOElocal

In figure 5 the G is plotted against MOElocal . Figure 5 shows that within the sam-
ples there seems to be no clear relation between the MOElocal and G. Between the
samples there seems to be a relation between the MOElocal and G, suggesting a
higher MOElocal leads to a higher Gstat. However, in previous studies mean values
for samples of sawn beams of sitka spruce of 500 N/mm2 and Norway spruce of
650 N/mm2 (Kokhar [7]) and spruce glued laminated timber between 600 N/mm2
and 800 N/mm2 (Brandner et al. [5]) were found, with comparable values for
MOElocal as in this study. This suggests that the G is not only dependent of
MOElocal but also on other factors that could not be identified in this study. Op-
tions are the density, the dimensions or the quality. Regarding this last point, in
Nocetti et al. [8] was found that for increasing knot ratio the ratio MOEloc
/MOEglobal decreased, which might be caused by the influence of knots on the G.
The results found in Nocetti et al. [8] also supports also the influence of weak
zones on the ratio MOEloc /MOEglobal.
900
800
700
600
G (N/mm2)
500
400
300
200
S1 M2 M1
100
0
0 5000 10000 15000 20000 25000 30000 35000 40000
MOElocal(N/mm2)
Fig. 5 Scatterplot for MOElocal against G for the tested samples
5.3 The Influence of Weak Zones on the Ratio between MOElocal

and MOEglobal
The influence of weak zones as a function of its position in the beam on the
MOElocal and the MOEglobal can be calculated, and as a result the effect on the rati-
os between them. In figure 6 the influence of the position of a weak zone on the
ratio between MOElocal and the MOEglobal is shown, also depending on the value of
MOElocal and G. There are many possibilities , but the figures assume that the
weak zone has a length of 0,5 the height of the beam and one third of the bending
stiffness of a clear part. For softwood the shear modulus G is kept a constant value
of 190 N/mm2 which was found in this study. The figure gives the effect on the
ratio between MOEloc and MOEglob when the position of the weak zone changes
from the center to the outside of the beam. The weak zone is assumed to be
asymmetric (so a weak zone only on one side, the other side is clear). For tropical
hardwood no weak zone is assumed and a constant shear value of 550 N/mm2 was
used. The ratio between MOEloc and MOEglob is calculated for different values for
the MOElocal.
Figure 6 shows that for a low MOElocal the ratio MOElocal/MOEglobal can become
lower than 1. Based on figure 6 the expected values and ratios for MOElocal and
MOEglobal can be calculated for the shear values found in this research. In the range
of MOEglobal for softwoods in this research (between 10.000 N/mm2 and 18.000
N/mm2) the influence of a weak zone leads to a minimum and maximum value for
every MOEglobal depending on the position of the weak zone. In the range of
MOEglobal (15.000 N/mm2 to 30.000 N/mm2) for hardwoods no weak zone was
assumed, giving one value.
In figure 7 the ratio MOElocal / MOEglobal is plotted against the MOElocal for the 3
samples. For softwood for three values of the MOElocal the ratio MOElocal / MOEglobal
is plotted for the minimum and maximum value found in figure 6 for a combination
of E and G. The influence of the weak zones will lead to a ratio MOElocal / MOEglobal
somewhere between these limits. For hardwoods also for three values of the MOElocal
the ratio MOElocal / MOEglobal is plotted. In this case there is only one value, because
no weak zones are assumed. It must be stated that these are average expected ratios,
where scatter due to normal variation of timber can be expected. Figure 7 shows that
the hardwood test data are around the predicted theoretical line, and that for soft-
woods the test data is between the maximum and minimum theoretical line. That
means that for softwoods the ratio MOElocal / MOEglobal is a combined effect of low G
and weak zones. However, the weak zone effect can only be evaluated properly when
also timber with low MOE is tested.
In figure 8 the predicted theoretical values from figure 7 are plotted for
MOElocal against MOEglobal. In figure 8 also the EN 384 equation is shown. Figure
8 shows that the different regression lines that can be found depending on the
presence of weak zones and the value of the shear modulus. When one single re-
gression line is plotted through all theoretical data points which is then forced
through the origin a slope of 1,15 is found, similar to that for all tested data. The
En 384 equation does not follow exactly this slope, but the error between the 2
lines is relatively small.
1.5
1.4
1.3
MOElocal /MOEglobal
E=10000,G=190
1.2
E=5000, G=190
E=15000,G=190
1.1
E=20000,G=190
1 E=15000, G=550, no weak zone
E=25000,G=550, no weak zone
0.9
0.8
0 1 2 3 4 5 6 7 8 9
position defect *h from center
Fig. 6 Influence of position of weak zone, MOE and G on the ratio MOElocal/MOEglobal
1.6
1.5
M2
1.4
M1
1.3
MOElocal/MOEglobal
S1
1.2
G =190 N/mm2 weak
1.1 zones theory min
G=190 N/mm2, weak
1 zones theory max
G= 550 N/mm2no weak
0.9 zone
0.8
0.7
0 10000 20000 30000 40000
MOElocal (N/mm2 )
Fig. 7 Ratio MOElocal/MOEglobal plotted against MOElocal with theoretical expected values
45000
40000
y = 1.278x - 3032.8
35000 R = 0.9996 theory max softwood G=190
N/mm2
30000 theory min softwood G=190
MOElocal (N/mm2 )
y = 1.7746x - 4349.9 N/mm2

25000 R = 0.9998 theory hardwood G=550
N/mm2
20000 EN 384 equation
15000 y = 1.2957x - 3870.2 Lineair (theory max softwood

R = 0.999 G=190 N/mm2)
10000
Lineair (theory min softwood
5000 G=190 N/mm2)
Lineair (theory hardwood
0 G=550 N/mm2)
0 5000 10000 15000 20000 25000 30000 35000 40000
MOEglobal (N/mm2 )
Fig. 8 Theoretical values from figure 8 for MOElocal plotted against MOEglobal together with
the EN 384 equation
6 Conclusions
The ratio for MOElocal/MOEglobal for the studied dataset can be explained by the
found low values for the shear modulus for the samples determined with the var-
iable span method for the softwood and hardwood samples. For the softwood
sample the presence of weak zones is also expected to have an effect, but this
aspect could not be identified for individual softwood pieces, because of the
lack of low MOElocal values in the tested sample.
The found G-values are not in line with current standardized values. The cur-
rent assumed dependency from the MOElocal is questioned. No final statement
from which parameters the shear modulus G is dependent can be made.
The EN 384 equation for MOElocal out of MOEglobal is not similar with the equa-
tion found in this study, but the differences are relatively small.
References
[1] EN 408. Structural Timber -Determination of some physical and mechanical proper-
ties, CEN 2010+A1:2012
[2] EN 338. Structural Timber- Strength classes, CEN 2010
[3] EN 384. Structural Timber- Determination of characteristics values of mechanical
properties and density, CEN 2010
[4] Ravenshorst, G.J.P., van de Kuilen, J.W.G.: Relationships between Local, Global and
Dynamic Modulus of Elasticity for soft- and hardwoods. CIB W18-42-10-1,
Dubendorf, Switzerland (2010)
[5] Brandner, R., Gehri, E., Bogensperger, T., Schickhofer, G.: Determination of modulus
of shear and elasticity of glued laminated timber and related examinations, CIB
W18-40-12-2, Bled, Slovenia (2007)
[6] Denzler, J.K., Stapel, P., Glos, P.: Relationships between global and local MOE. CIB
W18 paper 41-10-3 (2008)
[7] Khokhar, A., Zhang, H.: The use of torsion test method to evaluate the shear properties
of timber joists. In: World Conference on Timber Engineering, Auckland (June 2012)
[8] Nocetti, M., Brancheariau, L., Bacher, M., Brunetti, M., Crivellaro, A.: Relationship
beyween local and global modulus of elasticity in bedning and its consequence on
structural timber grading. European Journal of Wood Products 71, 297308 (2013)
Part IX
Glulam
A Study of Australian Glulam
H.R. Milner1 and Con Y. Adam2

1
A & M Industrial and Engineering Consulting, Pound Creek, Australia
bwmilner4@bigpond.com
2
Monash University, Department of Civil Engineering, Clayton, Australia
Abstract. Australia and New Zealand are debating how they should develop a
new specification for structural glulam to replace AS/NZS 1328.1:1998. The new
document will include i) performance requirements and ii) methods for computing
characteristic strength and stiffness properties from lamination data. With respect
to item ii) ASTM D3737 and prEN 14080 have been considered as potential mod-
els but, it is argued, both are indirect and involve arbitrary assumptions and a
series of computational steps that are difficult to justify on a rational basis. Two
drafting principles will guide the format of the new standard: it will contain
minimal manufacturing prescriptions and the link between the lamination test data
and the bending strength of the glulam will involve as little simulation and calibra-
tion as possible. Although the testing of both laminations and glulam members has
shown that Australian glulam meets the stated characteristic strength and stiffness
values listed in AS/1720.1, it is difficult to justify this claim on the basis of
lamination test data. The paper addresses those issues and recommends taking an
alternative approach to both ASTM D3737 and prEN14080 methodologies in
developing a new glulam standard. It is a work in progress at the time of writing.
Keywords: Australian structural glulam, performance requirements, lamination

and glulam characteristic strengths, laminating effects.
1 Background
1.1 AS/NZS Standards

Australia and New Zealand write standards for bi-lateral use, which in the case of
structural glulam, is AS/NZS 1328.1:1998. It is a performance based document
in the sense that very little is said about manufacturing processes but it does
prescribe performance levels and the associated test methods.
Subject to committee approval, future version will specify the following.
Deemed-to-comply provisions that enable finger jointed laminations
having defined properties to be manufactured from standard sawn timber
strength classes. There will be accompanying manufacturing prescrip-
tions related to the proximity of knots to finger joints.
DOI: 10.1007/978-94-007-7811-5_70, RILEM 2014
788 H.R. Milner and C.Y. Adam
Qualification provisions that allow a manufacturer to produce either

standard or proprietary lamination grades using in-house manufacturing
methods.
A methodology for converting tension properties of laminations to
glulam bending strengths.
Verification methods for confirming the strength of finger joints
and face/edge bond integrity through prescribed test methods and
performance levels.
Moisture content of the finished product.
Utility requirements.
Preservative and flame retardant treatment where these are used.
Formaldehyde levels where phenolic adhesives are used.
The standard will not unduly limit options available to the manufacturer by
attempting to prescribing any of the following.
Finger joint profiles and end pressures.

Cramping pressures.
Factory environmental conditions.
The methodology in AS/NZS 1328.1:1998 for converting lamination stock and

finger joint bending strength is based on tests by Falk et al (1992) which is unsat-
isfactory given that the data is based on Norwegian spruce. The principal glulam
species in Australia and New Zealand are radiata and slash pine in the case of
softwoods and so-called Victorian ash species which is a mix of lower density
hardwoods having a dry (12% mc) density of 680kg/m3. Attempts were being
made at one point to use some higher density species having densities of
1100kg/m3 which presented considerable problems with achieving acceptable
finger joint strengths. Random testing of glulam showed that the higher density
species used in glulam could not be finger jointed without considerable strength
loss. Recently, some adhesive manufacturers have claimed that new adhesive
technologies are being formulated that can deal with these higher density
hardwoods.
In the event a modified version of the Falk et al (1992) relationships was
adopted for AS/NZS 1328.1:1998 that took the form
f b = 0.75 (1 + 0.05Smin ) f b,ej f b,stock (1)
f b = characteristic strength of a glulam member in bending (MPa)

f b,ej = characteristic strength of a finger joint in bending (MPa)
f b,stock = characteristic strength of the laminating stock in bending (MPa)
S min = minimum spacing of finger joints in outer laminations (m)
A Study of Australian Glulam 789
A corresponding relationship was provided for tension strength. The factor

(1 + 0.05Smin) was included as an encouragement to use longer laminating lengths.
The restriction on the glulam bending strength being no greater than that of the
laminating stock ( fb,stock) is very conservative and it is one of the issues that
needs to be addressed in a revised version. If the Falk et el paper had been fully
implemented this value could have been lifted to at least (1.1fb,stock).
AS/NZS 1328.1:1998 was developed at the same time that a system of glulam
grades was introduced. The strength values were assigned without reference to
international data but, subsequently, they were found to track European GL grades
based on comparing characteristic bending strengths against modulus of elasticity;
see Figure 1. No USA data is included in Figure 1 given that there is not always a
one to one correspondence between modulus of elasticity and bending strength
even with balanced lay-ups. The GL17 and GL18 grades shown in Figure 1 are
associated with the high density species mentioned previously; no manufacturers
are currently attempting to manufacture these grades. The glulam data has been
extracted from Adam and Milner (2012).
Fig. 1 Comparison of prEN 14080 and AS/NZS glulam grades plotted along an elastic
baseline. Also shown are the values of AS/NZS glulam grades obtained using the
prEN14080 formula, equation 2.
The weaknesses in the AS/NZS system for converting lamination stock and
finger joint data into glulam beam and tension strengths are evident. The industry
accepts that a better system needs to be developed and that lamination tension
strengths should form the basis of an improved system. While it is not unknown
for glulam to be used as a tension element the predominant applications involve
bending applications. Two approaches for deriving glulam bending strengths were
considered as replacements for the existing AS/NZS 1328:1998, namely ASTM
D3737-08 and prEN 14080:2012.
1.2 ASTM D3737-08

ASTM documents are very widely used and, in the case of glulam, form the basis
for all of North America and beyond. ASTM D3737 has the considerable merit
that its accuracy has been verified by many industry and university studies accord-
ing to anecdotal evidence provided to the writer by US members of ISO commit-
tee TC 165. Having said that, the amount of data required to implement the D3737
methodology is considerable and the method is onerous. The scope of ASTM
D3737 contains no restrictions with regard to the species that may be used in
making glulam and permits LVL, LSL and OSL laminations.
The characteristic strength in bending of glulam is obtained using what is
known as the Ik/IG method that has its origins in Freas and Selbo (1954). It has
subsequently been modified by Moody (1974, 1977) and through the develop-
ment and updating of ASTM D 3737 itself. The original Freas and Selbo (1954)
paper seems to have influenced much of what followed even though it is diffi-
cult to pin-point a reason why the Ik/IG method works although following the
detail of the procedure in ASTM D3737 is relatively simple.
The method was developed at a time when structural timber strengths were
based on small clears data so that formed a natural starting point for Freas and
Selbo (1954) to develop their method. Intuitive reasoning let them to select IK, the
second moment of area for all knots within 6 inches (150 mm) of the critical
bending moment acting on the beam, as a measure of strength degrade. The value
was then non-dimensionalized as Ik/IG (=R) where IG is the gross second moment
of area of the beam. Tests on full size glulam beams were plotted against R and an
empirical relationship between R and modulus of rupture was developed. More
detail was added to the method by means of placing weighting factors on knots
in laminations that reflect the position of the lamination in the assembly with
respect to the neutral axis and statistical adjustments were made to allow for
variability in knot size within the lamination stock.
In developing a new standard it is not a method that offers a great deal of ap-
peal and it flies in the face with the modern tendency to derive timber properties
by full scale testing.
1.3 prEN 14080:2012

prEN 14080 sets out the European rules for glued-laminated and glued-solid
timber in a single document along with glulam products having large finger joints
and so-called block glued glulam products but restricts itself to coniferous species
and poplar although there is a note that "it may be possible to produce glulam
made from specific hardwood species".
In the case of horizontally laminated glulam, prEN 14080 has the merit that it is
easy to follow but it has the disadvantage that the derivation of characteristic
strengths is outwardly artificial and based on considerable mathematical
simulation studies. In Table 6 for homogeneous glulam (members made from
laminations taken from a single strength class) the characteristic strength in
bending is given by equation 2 as
l ,k + 1.5 ( f m , j ,k 1.4 f t ,0,l ,k + 6 )

0.65
f m ,g ,k = 2.2 + 2.5 f t ,0,
0.75
(2)
The tension strength is given as equation 3
f t, g ,k = 0.8 f m , g ,k (3)
f m , g ,k = characteristic strength in bending of the glulam member (MPa)

f t, g ,k = characteristic strength in tension of the glulam member (MPa)
f t ,0,l ,k = characteristic tension strength of the lamination stock parallel to the
grain (MPa)
f m , j ,k = characteristic strength in bending of the finger joints (MPa)
While equation 2 may well be justified on the basis of computer simulations

that are themselves based on the principles of mechanics and "spot check" com-
parison with test data, its outward appearance gives no inkling that is the case. It is
an oddity that the tension strength of a glulam member should bear any logical
relationship to the bending strength of the finger joints except for the fact that
finger joint bending and tension strengths may be related.
Viewing equations 2 and 3 from an Australian perspective, the predictions from
equations 2 and 3 do not fit especially well with Australian data based on MGP
10, 12 and 15 (Machine Graded Pine) and the visually graded A17 (VSG) hard-
wood favoured by the Australian hardwood industry. The reason for this becomes
obvious when we examine the characteristics of the grading systems used in Aus-
tralia relative to those of the European system. In Australia and New Zealand there
are two strength class (grade) systems in use an F grade system which can be
either machine stress graded (MSG) or VSG. The earlier Australian F- grading
system was a developed preferred mathematical series F8, F11, F14 etcetera that
reflected bending strength where the numeral indicated the approximate bending
strength in MPa based on long term loading for working stress design. What is
now far more popular with softwoods are the MGP or MSG (MGP10, 12, 15)
series based on stiffness criteria where the numeral represents the elastic modulus
in GPa. Industry observed that design was often controlled by serviceability
Fig. 2 European and Australian stress grades tension strengths plotted against an elastic
modulus base (sources EN 338, AS1720.1)
Fig. 3 Cumulative frequency of lamination tension strengths plotted against glulam bending
tension and bending strength using MGP10 radiata pine
(deflection) limits and therefore chose systems that favoured bending applications
of structural timbers where stiffness was predominant. The MGP and A17 systems
have led to strengths taking a back seat as is evident when tension strengths
are compared against an elastic modulus base relative to European grades; see
Figure 2. It is the principal reason why direct adoption of the prEN 14080 cannot
be translated directly into an Australia/New Zealand system. What is also shown
in Figure 3 drawn from Adam and Milner (2012) are cumulative frequencies of
lamination tension strengths plotted against glulam bending and tension strengths.
The stock material was MSG graded MGP10, with E = 10GPa, and a characteris-
tic strength in tension listed in AS1720.1 of 7.7MPa. Further investigation is
required to determine why the high number of low strength laminations at 1200
mm finger joint spacing led to glulam members that showed relatively good
glulam bending and tension strengths.
2 Horizontally Laminated Glulam and Laminating Effects

Horizontally laminated beams almost invariably fail at a tension lamination that is
most highly stressed relative to its strength even if some constrained compression
(plastic) crushing occurs on the compression face first. More than one writer has
drawn attention to the fact that the tension strength of laminations in glulam
assemblies is higher than the strength of an identical lamination tested in isolation.
Colling and Falk (1993) summarise these effects as follows.
Off-centre knots induce bending effects in isolated lamination tension

tests that are substantially resisted in the tension lamination of a glulam
beam because of the restraint of other laminations. The ratio of the ten-
sion strength of a lamination in a glulam beam to the tension strength of
an isolated lamination is designated ktest which is, in effect, a partial
laminating factor.
In glulam members these same knots are also reinforced on one side for
the extreme outer lamination and on both sides for all other laminations,
the partial laminating factor being designated, k reinf .
There is also a statistical effect that arises out of the dispersing the de-
fects; the partial laminating factor is designated k disp .
A composite laminating factor, , given by equation 4 provides the combined
effect of the partial factors.
= ktest k reinf kdisp (4)
There have been numerous attempts to assess the effects of ktest , k reinf , kdisp
separately but inevitably the value of evaluated in this way relies on mathemat-
ical simulation or strength models. A summary of the models used to assess the
strength of glulam beams is provided in the COST E24 document listed as a refer-
ence, the best known, at least in Europe being the Karlsrhue model. This model
has been used by a number of European researchers to investigate glulam beam
strength issues. The general principles upon which the model are based has been
described elsewhere no useful purpose is served by a repetition herein. The model

has been calibrated whereby artifices are introduced in the form of four different
failure criteria are used that are computational devices rather than necessarily
being linked to fundamental material behaviour.
3 Directness of the Method for Linking Lamination and

Glulam Characteristic Strengths
While the ASTM D3737 and prEN 14080 methodologies provide potential path-
ways for AS/NZS standards our inclination is to follow an alternative pathway
based on minimising the number of computational steps involved between the
basic measured data and the glulam bending strengths.
The most direct pathway is to test full sized beams, an approach that is ruled
out on the grounds of cost and the number of parameters involved. This method is
the most accurate and, in research projects, serves as a basis in deciding if a
particular mathematical simulation is effective.
The next most direct method is to move from lamination strengths to member
strengths as directly as is possible. Most glulam members are used in bending
application and that these typically fail in the tension zone it is natural to focus on
tension lamination performance. However, the observation by Colling and
Falk (1993) immediately suggests that at least some of the test method and dis-
persion effects and most of the defect reinforcement effects can be eliminated
by testing laminations in tension that are one-sided reinforced and two-sided rein-
forced. Where glulam is manufactured from timber that has already been assigned
a strength class under national grading rules this could be undertaken on an indus-
try wide basis provided some manufacturing prescriptions were accepted.
However, Australian and New Zealand manufacturers have offered strong re-
sistance to the inclusion of manufacturing prescriptions in AS/NZS standards. In
such cases, the lamination strengths need to be determined on manufacturer basis
that embeds individual practices such as defect docking in the data basic test data.
4 Test Methodology
It is proposed that laminations be tested in tension in both single sided and com-
bined single and double sided reinforced configurations as illustrated in Figure 3.
For the single sided reinforced laminations the tension strength of the combination
is taken as
f t = Ft 2bt (5)
f t = tension strength of a one sided reinforced lamination as produced by a

manufacturer or determined on an industry basis
Ft = failure load in tension of the lamination pair

b, t = lamination cross-section dimensions
The characteristic strength at the 5th percentile, 75 percent confidence level is
extracted by standard statistical techniques such as those detailed in AS/NZS
4063.2. As such a characteristic strength, f t ',ss , is extracted.
f t ',ss = characteristic strength of a single sided reinforced lamination
For double sided reinforced laminations it has to be noted that two of the lami-
nations are single sided reinforced and the middle lamination is double reinforced.
In this case, a tension strength is derived first for each assembly and its character-
'
istic strength determined. Let this value be Ft,dsassembly . The two single sided rein-
forced laminations have a characteristic strength, 2btf t ',ss so that the double sided
reinforced lamination will have strength
'
f t,ds = ( Ft ',dsassembly 2btf t ',ss ) bt (6)
2 x single sided reinforced

+ 1 x double sided reinforced
t
Stress Single sided reinforced
b distribution
Fig. 4 Single and double sided reinforcement of laminations in homogeneous glulam beams
The intention is that the test length for laminations be related to a glulam beam
500 mm deep. It is usual in Australian/New Zealand standards that any beam be
tested in four-point bending over a span of 18 times the depth. For the standard
glulam member that means a span of 9 metres where the middle third is 3 metres.
Thus the tension length will be set at 3 metres.
It is evident that this approach has best prospect of working with homogeneous
lay-ups. Modification may be required in the case of heterogeneous lay-ups. There
is no guarantee that a lamination derived from high grade stock will be reinforced
by a lamination made from low grade lamination stock in the same manner as in
the homogeneous case. However, this is thought not to be a difficulty given that
most beam failures occur in the critical outer tension zones likely to occupy up to
one sixth of beam depth where homogeneity will exist. It is more likely that ten-
sion strengths will be affected but this can be overcome by citing conservative
tension values without the need for additional testing.
5 Lamination Grades and Beam Properties
5.1 Lamination Grades

The basic tension properties of lamination grades would be provided in two forms
one sided and two sided reinforced. These would be linked to Australian/New
Zealand grades for sawn timber including our MGP10, 12, 15 and A17 material.
There is therefore, associated with those grades, a set of characteristic properties
for elastic modulus, compression strength, shear strength, tension and compression
perpendicular to the grain but the tension strengths parallel to the grain will be
enhanced on the basis of the single and double sided tension testing.
Some manufacturers have expressed a wish to operate outside standardized
properties. They are prepared to measure elastic moduli of individual pieces using
the higher stiffness and presumably higher strength pieces in the outer lamina-
tions. Whether this is an advantage is debatable given that laminating factors for
higher grade laminations are typically lower. To operate efficiently they will need
to develop their own set of lamination grades and then run the one and two sided
reinforcement tension tests. Subsequently, the grade data will be provided to a
computer program that will allow the glulam grades to be met. An interesting
aspect of this provision is that manufacturers are not especially enthusiastic about
lay-up details being in the hands of designers external to the company. In other
words, they would undertake the specification of lay-up details in-house and
market their product through company specific literature. The claim is made that
having engineers external to the manufacturer company specifying lay-ups will
present the manufacturer with insuperable difficulty in managing their production
facilities.
5.2 Underlying Theory

Composite beam theory will be invoked which requires both strength and stiffness
data. The stiffness data is usually available through the purchase of lamination
stock that has been assigned a elastic modulus as a consequence of manufacturers
selecting their laminating stock as machine stress graded timber. However, there is
a tendency for manufacturers to identify higher stiffness material for use in the
outer laminations so that any new standard will provide test methods that will
allow such selection. According to composite beam theory, the flexural stiffness
for laminations of constant thickness, t, is given by equation 7
EI x = ( bt 3 12 ) Ei + ( bt ) Ei yi2 (7)
i i
and the stress in the ith lamination is given by equation 8
f t ,i = Ei Myi EI x f t ' (8)
where
Ei = elastic modulus of the ith lamination

M = design bending moment including all load safety factors
yi = distance from the neutral axis to a tension element
= material safety factor
'
f t = tension strength of the lamination based on one and two sided reinforce-
ment as appropriate
A check would be built in to ensure that compression stresses are not in excess
of the laminating stock characteristic strengths. With homogeneous and balanced
heterogeneous beams compression face plasticity will not be an issue given that
compression strengths in a statistical sense exceed tension strengths at least for
Australian and New Zealand laminating species. With unbalanced heterogeneous
lay-ups low grade compression face material can result in compression face
plasticity. The analysis of such beams then involves non-linear analysis but this
presents no difficulty in arriving at a strength failure model.
5.3 Glulam Grades

Australian/New Zealand grades have existed for approximately 15 years and have
become a permanent feature of the design standards. There exists a suite of design
properties that can be used. Like the European glulam grades, composite beam
theory is not required and this suits the vast majority of designers. They have
therefore served a useful purpose and will be retained as prescribed lay-ups based
on standardized lamination properties.
Where a manufacturer chooses to use non-standard lamination grades and at-
tempts to manufacture a standardized glulam grade using the composite beam
theory internally they will need to assure themselves that there are no laminating
effects that have been excluded in the testing and application of composite beam
theory that is proposed herein. The fundamental question is whether or not the
proposed testing adequately takes account of the ktest, kreinf, kdisp identified in
Colling and Falk (1993). On the basis of the St Venant principle it seems unlikely
that any reinforcing effect on a knot or general sloping grain would extend beyond
lamination(s) immediately adjacent to a given lamination and that ktest and kdisp
effects would be mitigated to some extent. However, the possibility remains some
laminating effects might still exist that are not embedded in the lamination test
data. This can only be decided once investigations have been undertaken.
6 General Drafting Principles
Any manufacturing prescriptions that form part of the standard will be included as
deemed to comply provisions linked to glulam grades. Such prescriptions would
not extend beyond restricting maximum knot size in outer zones and restrictions
on the distance between finger joints and the nearest knot. Beam lay-up details
would also be provided for the lamination grades. In following this pathway quali-
fication and verification testing of finger joints only would be required.
There will be no restrictions on manufacturers developing their own lamination
grades without any manufacturing prescriptions and subsequently using computer
software to develop a beam having specific properties. However, individual quali-
fication testing would be required to establish lamination properties under a manu-
facturing regime that is clearly specified. Verification of finger joint strengths
would also be required.
7 Conclusions
1. The paper sets out a potential pathway forward for the development of
an Australian/New Zealand standard for structural glulam.
2. Australia and New Zealand are disinclined to copy either the US or
European method for evaluating lamination characteristic properties
from lamination grade properties.
3. Subject to verification now underway, it will be developing a method-
ology based on testing of reinforced where account is taken wholly
or partially of ktest, kreinf, kdisp in addition to manufacturer specific
production factors.
References
Adam, C.Y., Milner, H.R.: Glulam design based on lamination grades and the use of mill
shorts, Project Number PNB157-0910 (November 2012)
Australia/New Zealand standard AS/NZS 1328.1:1998, Glued laminated structural timber,
Part 1: Performance requirements and minimum production requirements
ASTM D3737 - 08 Standard practice for establishing allowable properties for structural
glued laminated timber (glulam)
Colling, F., Falk, R.H.: Investigation of laminating effects in glued laminated timber. In:
CIB-W18 26th Meeting (August 1993)
COST E24 Probablistic model code for timber Glued laminated timber HJL/
Draft 2004-08-28, http://www.km.fgg.uni-lj.si/coste24/data/
LjubljanaDocuments/Larsen.pdf
EN 338 Structural timber Strength classes
Falk, R.H., Solli, K.H., Aasheim, E.: The performance of glued laminated beams
manufactured from machine stress graded Norwegian spruce. The Norwegian Institute
of Wood Technology, Report no.77, Oslo, Norway (1992) ISBN 82-7120-029-1, ISSN
0546-3637
Freas, A.D., Selbo, K.L.: Fabrication and design of glued laminated wood structural
members, USDA Technical Bulletin No. 1069 (1954)
ISO/IEC 17007:2009 Conformity assessment Guidance for drafting normative documents
suitable for use for conformity assessment
Moody, R.C.: Design criteria for large structural glued-laminated timber beams using
mixed species of visually graded lumber, USDA Forest Services Research Paper, FPL
236 (1974)
Moody, R.C.: Improved utilization of lumber in glued laminated beams, USDA Forest
Ser-vices Research Paper, FPL 292 (1977)
prEN 14080:2012 Timber structures Glued laminated timber and glued solid timber
Requirements
Improving Strength of Glulam Laminations
of Norway Spruce Side Boards by Removal
of Weak Sections Using Optimized Finger
Jointing
Jan Oscarsson1, Anders Olsson2, and Bertil Enquist2

1
SP Technical Research Institute of Sweden, Vxj, Sweden
jan.oscarsson@sp.se
2
Linnus University, Vxj, Sweden
Abstract. Recent research has shown that glulam laminations of Norway spruce
side boards possess excellent structural properties. This investigation concerns the
possibility of improving the performance of such laminations through elimination
of weak board sections by means of finger jointing. Sections to be removed were
identified using profiles of edgewise bending stiffness determined on the basis of
scanned fibre angle fields on board surfaces. The difference in average tension
strength and average tension stiffness, respectively, between a group of finger
jointed boards and a reference group of non-jointed boards was evaluated. Joints
were inserted in the first group with an average distance of 2.4 m. It was found
that the finger jointing gave a considerable increase of strength (36 %), whereas
the stiffness improvement was not as evident. Based upon the results, it can be
assumed that application of finger jointed side board laminations will result in
glulam beams with very high strength.
Keywords: Dot laser, finger jointing, glulam, scanning, side boards.
1 Introduction
Side boards, i.e. boards of narrow dimensions cut from the outer parts of a log, are in
most cases regarded as products of poor profitability owing to low value and costly
handling. Their main outlets are e.g. loading pallets and formwork timber. However,
their structural properties are, in general, excellent (e.g. Steffen et al. 1997) which
indicates that utilization of side boards as laminations in engineered wood products
will result in load bearing components with high structural performance. As a
consequence, considerable increase of board value could be expected.
Research concerning the application of Norway spruce (Picea abies) side
boards as laminations in wet-glued glulam beams has been jointly conducted by
the SP Technical Research Institute of Sweden, Linnus University, and the
DOI: 10.1007/978-94-007-7811-5_71, RILEM 2014
802 J. Oscarsson, A. Olsson, and B. Enquist
sawmilling company Sdra Timber. The principal lay-up of investigated beams is

shown in Fig. 1. After gluing (Fig. 1, left) the beams were split, dried and planed
(Fig. 1, right) whereupon modulus of elasticity (MOE) and bending strength was
determined on the basis of testing procedures laid down in the European Standard
EN 408. In two initial test series, the beams were produced from ungraded side
boards, but in a third series, the three outermost laminations at both beam edges
were selected on the basis of strength grading criteria concerning maximum knot
size and minimum global axial dynamic board MOE. The results were very
promising (Serrano et al. 2011). High performance and low variability in terms of
strength and stiffness was achieved, particularly for beams with graded boards.
Splitting
1325 or 1521.5
300
120 50 50
Fig. 1 Wet-glued beams before (left) and after (right) splitting, drying and planing
This paper concerns the possibility of improving structural properties of graded

side board laminations even further by removal of weak sections using optimized
finger jointing. Previous research (e.g. Hoffmeyer 1995) has shown that strong
indicating properties (IPs) of strength can be defined on the basis of stiffness
measures determined locally, at so called critical sections in timber pieces.
According to EN 384, such a section is defined as the position at which failure is
expected to occur. To what extent elimination of critical sections will contribute to
increased strength and stiffness has been investigated by comparing the difference
in average tension strength, average local static tension MOE, and average global
axial dynamic MOE, respectively, between a group of finger jointed (FJ) boards
and a reference group of non-jointed (NJ) boards. Sections to be removed has been
identified on the basis of the relationship between strength and critical sections,
the latter defined as local dips in edgewise bending MOE profiles determined
Improving Strength of Glulam Laminations of Norway Spruce Side Boards 803
along the boards. The method applied for calculation of such profiles has recently
been presented by Olsson et al. (2013).
2 Sampling of Material
A number of 51 side boards of nominal dimensions 251254800 mm were

sampled from a batch of 360 boards of sawfalling quality delivered from Sdra
Timbers sawmill in Torss, Kalmar County. Before delivery, the boards were
dried to a target moisture content (MC) of 16 %. The sampling was aimed at
achieving a large variation of strength between sampled boards. Boards in which
the amount of compression wood exceeded 10 % of the cross-sectional area were
rejected. This requirement, evaluated on the basis of visual inspection, is in
accordance with the acceptance criteria for strength classes C18 to C30 laid down
in the Nordic visual strength grading rules INSTA 142. After ten months storage
in a standard climate of 20 C and 65 % relative humidity (RH), MC equilibrium
had been reached. Each sampled board was then split longitudinally into two parts,
one randomly assigned to the FJ board group and the remaining one to the NJ
group.
3 Methods and Measurements
The laboratory work included determination of global axial dynamic MOE, laser
scanning, finger jointing and static tensile loading. Board quantities and properties
measured or determined were dimensions, weight, resonance frequency of the first
axial mode of vibration, fibre direction fields determined on board surfaces on the
basis of dot laser illumination and tracheid effect scanning, static tensile strength,
and local static tensile MOE. In addition, the research also comprised application
of beam theory, finite element modeling, and development and application of
optimization algorithms for identification of critical sections to be removed by
finger jointing. The software Matlab was used for these purposes.
3.1 Global Axial Dynamic MOE

The global axial dynamic MOE, Edyn [Pa], was determined for each split board as
Edyn = 4 (f1,a L)2 (1)
where is the density [kg/m3] calculated on the basis of measured weight and
dimensions, fa,1 is the axial resonance frequency [Hz] that corresponds to the first
mode of vibration, and L is the board length [m]. The value of fa,1 was measured
using a Timber Grader MTG, which is a handheld wireless measuring device
certified for commercial strength grading of structural timber.
3.2 Fibre Direction Fields on Board Surfaces Based on Laser

Scanning
By application of the tracheid effect, fibre direction fields on both flat and edge
surfaces of the split boards were measured using an optical scanner of make
WoodEye, see Fig. 2 (left). The mentioned effect, resulting in laser dots entering
an elliptic shape when they illuminate a board surface, is due to the fact that laser
light is more easily spread in the fibre direction than in other directions, see Fig. 2
(middle). As a board is transported length-wise through the scanner, flat and edge
surfaces are scanned simultaneously using four sets of laser sources and
multisensor cameras. Each laser source provides a row of laser dots which is
oriented perpendicular to the longitudinal board direction. The boards were fed
through the scanner with a feed speed of about 60 m.p.s. In combination with the
sampling frequency of the cameras, this resulted in a longitudinal scanning
resolution of 1.2 mm. Corresponding resolution in lateral direction was 3.9 mm.
The total number of fibre direction measurement points was approximately
180000 on each board. A typical fibre direction field scanned around a traversing
face knot is shown in Fig. 2 (right).
Longitudinal board direction
Fig. 2 WoodEye scanner (left), tracheid effect (middle), and scanned fibre direction field
around a traversing face knot (right)
3.3 Identification of Critical Board Sections

As mentioned in the Introduction, edgewise bending MOE profiles were applied
in order to identify critical board sections to be removed by means of finger
jointing. A brief description of the procedures by which such profiles are
determined follows below, whereas a more detailed account is found in Olsson
et al. (2013). An important novelty is the application of fibre direction fields
described above. Such fields provide information about the angle between
local fibre direction and lengthwise board direction, see Fig. 3 (left). Since wood
is an orthotropic material with the highest structural performance in the
longitudinal fibre direction, even small deviations between fibre direction and
lengthwise board direction result in a considerable impairment of structural
board properties.
Transformations taking material properties in different directions into account
serve as a basis for calculation of local material stiffness in the board direction,
which in turn provides data for integration of an MOE profile valid for edgewise
bending. Important assumptions in the method are that
the MOE in the longitudinal fibre direction, El, and density, , are
constant within a board,
other stiffness parameters are linear functions of El,
fibre directions measured on board surfaces are located in the
longitudinal-tangential plane,
the fibre direction coincide with the wood surface, i.e. the so called
diving angle is set to zero, and
the fibre direction measured on a surface is representative for the fibre
direction to a certain depth within the board. Thus, the fibre angle
shown in Fig. 3 (left) is assumed to be valid within the volume defined by
the area dA (see Fig. 3 middle) times the length dx (see Fig. 3 left).
Thus, except from , the only parameter that has to be determined individually for
each board is El. This parameter is determined such that an eigen value analysis on
a one dimensional finite element model of a board result in the same resonance
frequency as the one determined experimentally.
According to Olsson et al. (2013), calculated MOE profiles can also be
utilized for strength grading of structural timber. The IP for a certain board is
defined as the lowest edgewise bending MOE found along its length. In the
research referred to it was found that the relationship between this IP and strength
was dependent on the local scale on which the IP was determined. It could be
determined with a maximum resolution corresponding with the longitudinal
scanning resolution of 1.2 mm, see Fig. 3 (left). However, since critical knots
have a certain extension in this direction, it was found that the strongest
relationship between IP and strength was obtained for IPs determined on the
basis of MOE profiles representing a moving average MOE calculated over an
interval of about half the board depth. Consequently, the research presented here
was based on edgewise bending MOE profiles calculated over intervals of about
30 mm. The reason why bending profiles was applied, although the boards were
loaded in axial direction, was due to the fact that an IP based on bending MOE
correlates better to tensile strength than what an IP based on profiles representing
axial MOE does (Oscarsson 2012).
y y

x
dA
x z y
z
dx
dx = 1.2 mm (longitudinal
scanning resolution)
Fig. 3 Fibre directions measured on a boards flat surface (left), cross-section divided into
sub-areas (middle) implying that the highlighted angle is valid within the volume given
by dAdx, and board segment of length dx (right) the edgewise bending MOE of which is
calculated by stiffness integration over the segments volume
3.4 Critical Sections to Be Removed; Optimization Algorithms

Critical sections to be removed in the FJ boards were identified by means of
algorithms designed to optimize the average tensile strength and stiffness,
respectively, of the FJ group. It was assumed that the tensile strength is directly
proportional to the lowest local bending MOE. As a result, the application of the
algorithms resulted in different number of joints in different boards.
The benefit of finger jointing in terms of improved structural properties had to
be put in relation to the costs incurred by such processing. This issue was
discussed with representatives of Sdra Timber resulting in a decision implying
that joints were to be inserted with an average distance of 2.4 m over the total FJ
board length of 245 m. Thus, a total number of 102 joints, i.e. on average two
joints per FJ board, were to be inserted.
Elimination of a critical section meant that the length of the board in question
was reduced with about 120 mm (see under heading Finger jointing below).
Consequently, for those boards in which several joints were inserted, the length
was considerably reduced. Since the strength of timber is dependent on the length
of tested pieces (Isaksson 1999), it was necessary that all boards, both FJ- and NJ
boards, were strength tested at the same length, determined by the FJ board with
the largest number of inserted joints. For all other boards, cross cutting was
necessary, although this implied that board information was lost. To confine such
loss, the maximum number of joints inserted into a board was limited to six.
By means of the optimization algorithms, the six deepest dips in the MOE
profile of each FJ board were identified as possible sections to be removed. Such
dips coincided with significant knots. As an example, identified dips, denoted D1
to D6, in board no. FJ25 are shown in Fig. 4. For each FJ board, the IP
improvement per section selected for possible removal was calculated. The
equations applied for such calculations varied depending on the number of
sections assumed to be selected simultaneously. When only the deepest dip (D1)
was considered, the potential IP improvement was calculated as the difference
between the second deepest dip (D2) and D1. When the two deepest dips were
considered simultaneously, the potential improvement per section was calculated
as the difference between D3 and D1 divided by 2, i.e. (D3-D1)/2, and so on.
Thus, when six sections were considered at the same time, the potential
improvement per section was calculated as (D7-D1)/6. The described calculation
procedure was carried out for each FJ board and all potential improvement values
were subsequently assembled into a matrix the principle of which is exhibited in
Table 1. Highlighted potential improvement values for board no. FJ25 show that
the largest increase of the IP value of this board was achieved when five sections
were selected at the same time.
It should be noted that MOE dips with a distance shorter than 200 mm was
considered as one single dip, since the minimum length of pieces possible to joint
in the applied finger jointing machine, see under heading Finger jointing below,
was of this length.
Moving average bending MOE [GPa]
D6 = 12.506 D7 = 12.695
D5 = 9.816
D4 = 7.548
D3 = 5.232
D1 = 4.889 D2 = 5.018
Board no. FJ25; Distance from board root end [m]
Fig. 4 MOE profile of board no. FJ25 and possible sections, defined by MOE dips D1-D6,
to be removed
Table 1 IP improvement matrix for selection of sections to be removed by finger jointing
1
FJ board number Potential IP improvement per selected section [MPa]
1 2 3 4 5 6
01 D2-D1 (D3-D1)/2 (D4-D1)/3 (D5-D1)/4 (D6-D1)/5 (D7-D1)/6

: : : : : : :
25 129 172 886 1231 1523 1301
: : : : : : :
51 D2-D1 (D3-D1)/2 (D4-D1)/3 (D5-D1)/4 (D6-D1)/5 (D7-D1)/6
1. Figures 1-6 refer to the number of sections simultaneously selected for removal.
The selection of board sections to be actually removed was carried out using
the matrix exhibited in Table 1 and the following algorithm loop:
1. The matrix element with the largest value was identified.
2. The section (or sections), i.e. the MOE dip (or dips), to which this
element referred was selected for finger jointing.
3. The identified matrix element and possible elements in the same row
to the left of this element were deleted.
4. Possible remaining elements in the same row were renumbered D1,
D2, etc. and new IP improvement values were calculated accordingly.
5. Loop iteration.
The loop was reiterated until 102 sections were selected.
3.5 Finger Jointing

Finger jointing of the FJ boards was carried out using a compact jointing machine
of make Ersson EKS 160, see Fig. 5 (left), with finger cutting tools of dimensions
153.80.42 mm according to EN 385. A melamine-formaldehyde adhesive,
Casco Adhesives 1250, was used together with a hardener, Casco Adhesives 2550.
This adhesive-hardener combination is approved for use in load-bearing structures
in service classes 1, 2 and 3. The fingers were visible on the flat board surfaces
and the applied nominal end pressure was 7.7 MPa.
The length of eliminated defect zones was determined on the basis of the length
over which fibre directions were disturbed due to the presence of the defect, i.e. a
knot or a cluster of knots. It was found that elimination of a board length of 120
mm, i.e. a length of about twice the board depth, was appropriate, although this
meant that EN 385 requirements concerning distance between cross cut and
removed knot were occasionally violated. In some cases, the eliminated length
was increased due to requirements in EN 385 regarding distance between finger
joint and adjacent knots.
Since the application of the optimization algorithms resulted in different
number of joints in different boards, the length of the jointed FJ boards varied
from 3885 mm to 4800 mm. As mentioned before, all boards (both FJ- and NJ
boards) were then cross cut in the boards root end to 3885 mm length. After
subsequent planing to cross-sectional dimensions 2157 mm, tracheid effect
scanning and determination of global axial dynamic MOE, respectively, was
carried out once again.
3.6 Tension Testing

Tension strength and local static MOE, respectively, of both FJ- and NJ boards
were determined in tension parallel to the grain in accordance with EN 408. The
testing machine was of make MFL, see Fig. 5 (right), with a hydraulic force
generation, 2.0 m length of stroke and a load cell with a capacity of 3.0 MN.
The grips were of wedge type and the gripping length at actual testing was
estimated to 735 mm, which means that the board length between the grips was
2415 mm. The load was applied in force control mode with a constant loading rate
of 8 kN/minute and the average time to failure was 283 seconds. Pairs of
corresponding load-deformation values were sampled at a frequency of 2 Hz.
Fig. 5 Finger jointing machine Ersson EKS 160 (left), and tension testing machine MFL
(right)
4 Results and Conclusions
Application of the optimization algorithms resulted in a large variation of the

number of joints allocated to different boards. For example, nine boards were not
allotted any joints at all, see example in Figure 6 (left), whereas two boards were
given the maximum number of six joints each, see example in Figure 6 (right).
Moving average bending MOE [GPa]
Finger
joint
Board FJ21; Distance from board root end [m] Board FJ04; Distance from board root end [m]
Fig. 6 MOE profiles; board with zero (left) and board with six sections to be jointed (right),
respectively
The results from the tensile strength tests and from the determinations of global
axial dynamic MOE, respectively, are presented in Table 2. By means of finger
jointing, the mean tensile strength was increased from 24.2 MPa to 32.9 MPa, an
improvement of no less than 36 %. The strength variation in terms of coefficient
of variation was simultaneously reduced with 10 %, whereas the standard
deviation was somewhat increased. The latter could not be considered as very
surprising, given the considerable increase of strength achieved by the finger
jointing. It was also observed that the number of boards with strength higher than
40 MPa was 12 in the FJ group, but only 4 in the reference group. As regards the
MOE values, the finger jointing resulted in limited improvements of both mean
value and spread.
It should be mentioned that the results from the tensile tests using the MFL
machine were based on evaluation of 46 pairs of boards since 5 pairs had to be
rejected due to tensile board failure within the grips of the machine.
The conclusions that can be drawn from the results are that finger jointing of
narrow side boards based on dips in MOE profiles is an efficient means for
increasing mean strength and reducing strength variation, whereas such treatment
has a limited effect on the board stiffness in terms of MOE. Thus, to investigate to
what extent the strength of laminated beams exhibited in Fig. 1 could be improved
by finger jointing of outer laminations is a very interesting research issue to be
probed into further. The larger number of boards with high strength (> 40 MPa) in
the FJ group indicates that other optimization criteria than average strength of
selected samples should be interesting to investigate thoroughly.
Table 2 Mean value, standard deviation (Std.) and coefficient of variation (CoV) for
tension strength, local static tension MOE and global axial dynamic MOE for investigated
boards
Board group Tension strength Loc. stat. tens. MOE Glob. ax. dyn. MOE
Mean Std. CoV Mean Std. CoV Mean Std. CoV

[MPa] [MPa] [%] [MPa] [MPa] [%] [MPa] [MPa] [%]
Non-jointed (NJ) boards 24.2 12.7 52.6 13674 2010 15.4 14240 2130 14.9
Jointed (FJ) boards 32.9 14.1 42.9 14520 2120 14.6 14710 2062 14.0
References
EN 384. Structural timber Determination of characteristic values of mechanical properties
and density. CEN (2010)
EN 385. Finger jointed structural timber Performance requirements and minimum
production requirements. CEN (2003)
EN 408. Timber structures Structural timber and glued laminated timber Determination
of some physical and mechanical properties. CEN (2010)
INSTA 142 / SS230120. Nordic visual strength grading rules for timber. Inter-Nordic
standardization, Swedish Standards Institute (2010) (in Swedish)
Hoffmeyer, P. (ed.): Strength grading adds value, Part 2 Available technique. Technical
University of Denmark, Technical Report 335-1995 (1995) (in Danish, Norwegian and
Swedish)
Isaksson, T.: Modelling the variability of bending strength in structural timber Length and
load configuration effects. Doctoral thesis, Lund Institute of Technology, Report TVBK-
1015, Lund, Sweden (1999)
Olsson, A., Oscarsson, J., Serrano, E., Kllsner, B., Johansson, M., Enquist, B.: Prediction
of timber bending strength and in-member cross-sectional stiffness variation on basis of
local wood fibre orientation. European Journal of Wood and Wood Products 71(3),
319333 (2013)
Oscarsson, J.: Strength grading of structural timber and EWP laminations of Norway spruce
Development potentials. Licentiate thesis, School of Eng., Linnus University (2012)
Serrano, E., Blixt, J., Enquist, B., Kllsner, B., Oscarsson, J., Petersson, H., Sterley, M.,
Petersson, H., Sterley, M. (ed.): Wet glued laminated beams using side boards of
Norway spruce. Report No 5, School of Engineering, Linnus University,
Vxj, Sweden (2011)
Steffen, A., Johansson, C.-J., Wormuth, E.-W.: Study of the relationship between flatwise
and edgewise moduli of elasticity of sawn timber as a means to improve mechanical
strength grading technology. Holz als Roh- und Werkstoff 55, 245253 (1997)
Double Span Continuous Glulam Slabs
Strengthened with GFRP
Jorge M. Branco, Marco P. Jorge, and Jos Sena-Cruz
ISISE, University of Minho, Department of Civil Engineering, Guimares, Portugal

jbranco@civil.uminho.pt
Abstract. Full-scale slab strips were tested in order to analyze the flexural
response of GFRP-glulam slab systems under monotonic loading. The type of
strengthening technique (externally bonded reinforcement EBR and near-surface
mounted NSM) and the increase target in terms of ultimate load capacity (20%
and 40%) were the main studied parameters. GFRP sheets were utilized in the
EBR technique, while GFRP rods were applied in NSM technique. In this work
the tests are described in detail, and the obtained results are presented and
discussed.
Keywords: Glulam slabs strengthened with GFRP, double span, flexural

strengthening, strengthening techniques, increase of load carrying capacity,
moment redistribution, failure modes.
1 Introduction
Glulam members often become large in cross section where heavy loads should be
carried. In some applications this may cause problems if limitations on height are
posed. A possible solution is to reinforce the member by e.g. bonding fibre
reinforced polymer (FRP) on the beams or between the glulam lamellas.
In fact, strengthening by the use of fibre reinforcement has become a
common method to enhance the bearing capacity and stiffness of timber
members. The strengthening effect is achieved through bonding reinforcement
rods or sheets onto the timber beam by means of glue, commonly epoxy resins.
Application areas range from enhancement of low strength lumber (Gilfillan
(2003), Raftery and Harte (2011)) to upgrading old members assessing the
possibility to on-site applications (Borri (2005), Branco et al. (2005), Corradi
and Borri (2007)). The types of fibres used have been, essentially, glass and
carbon fibres, while aramid fibre is a possibility less suitable because of it
sensitivity to moisture content.
FRP strengthening may increase ductility of timber beams and usable strains of
tension fibers by delaying the onset of tensile cracking of the fibers. In some cases
it was found that the FRP-glulam beams with sufficient reinforcement in the
tension regions not only exhibit significant strength increment, but also they
develop wood ductile compression failure, rather than the typical brittle tension
DOI: 10.1007/978-94-007-7811-5_72, RILEM 2014
814 J.M. Branco, M.P. Jorge, and J. Sena-Cruz
failure of wood (Dagher et al. (1996), Dagher and Lindyberg (1999), Lopez-Anido
and Xu (2002)). However, limited full-scale applications and field demonstration
projects have been conducted (Gentile et al. (2002), Buell and Saadatmanesh
(2005)). The majority of strengthening applications has been related to glulam
beams simple supported under bending moments (Gilfillan et al. 2003, Micelli et
al. 2005). No studies or applications are known on timber-FRP strengthening of
indeterminate structural systems.
In this context, the effectiveness of flexural strengthening configurations for
glulam continuous slab strips of two equal span lengths was explored in terms of
load carrying capacity and moment redistribution. For that purpose, six full-scale
glulam slab strips were tested. The type of strengthening technique (EBR and
NSM) and the increase level of load carrying capacity (20% and 40%) to be
attained are the main parameters considered in this experimental program. GFRP
sheets were used in the EBR technique, while GFRP rods were applied in the
NSM technique. In the following sections the tests are described in detail, and the
obtained results are presented and discussed.
2 Experimental Program
2.1 Test Specimens

The six full-scale two span glulam slab strips are organized in three different
series. The first series is composed by two unstrengthened slabs (reference slabs,
REF.1 and REF.2). The second series is composed by two glulam slab strips
strengthened according to the NSM GFRP using a percentage of GFRP rods in
an attempt to increase the maximum load in 20% (NSM.20 and 40% (NSM.40),
when compared to the average value registered in the average of REF.1 and
REF.2. Finally, in the third series, two glulam slab strips were strengthened
according to the EBR technique applying GFRP sheets, and the purpose was
also to increase the maximum load in 20% (EBR.20) and 40% (EBR.40),
respectively.
Fig. 1 presents the test setup in terms of load configuration, support conditions
and measuring devices. Fig. 1(b) indicates the six linear voltage differential
transducers (LVDTs) used to measure the deflections of the slab strips. The
LVDTs 60541 and 18897, placed at the slab loaded sections, were used to control
the test at a displacement rate of 20 m/s. The load (F522) applied at the left span
(see Fig. 1(a)) was measured using a load cell of 200 kN and accuracy of
0.03%. In the right span, the load (F123) was measured using a load cell of
250 kN and accuracy of 0.05%. To monitor the reaction forces, load cells were
installed at two supports. One load cell (AEP_200) was positioned at the central
support (nonadjustable support), placed between the reaction steel frame (HEB
300 profile) and the slabs support device (see Fig. 1(b)). The other load cell
(MIC_200) was positioned in-between the reaction steel frame and the apparatus
of the adjustable right support of the slab. These cells have a load capacity of
200 kN and accuracy of 0.05%. Fig. 2 presents the transverse geometry and the
Double Span Continuous Glulam Slabs Strengthened with GFRP 815
strengthening arrangements of the tested slabs, whereas Fig. 3 presents the

longitudinal configuration of the strengthened slabs and the corresponding strain
gauges locations.
F522 F123
(a) AEP_200 MIC_200
90
200 1400 1400 1400 1400 200
2800 2800
6000
F F
Suspension Yoke
(b)
LVDT LVDT LVDT LVDT LVDT LVDT
82803 60541 82804 19906 18897 3468
Fig. 1 (a) Test configuration; (b) displacement transducers location. All dimensions in
[mm]
2.2 Characterization of the Materials

Glued laminated timber of strength class GL24h (NP EN 1194:1999) was used for
all the series. The material characterization of the GL24h included compression
and tension tests parallel to the grain according to EN 408:2003. From the
compression tests an average compressive strength of 27.99 MPa with a
coefficient of variation (CoV) of 17.6% and an average modulus of elasticity of
6.62 GPa (CoV=27.8%) were obtained. From the tension tests, an average tensile
strength, modulus of elasticity and strain at the peak stress of 55.93 MPa
(CoV=16.7%), 9.17 GPa (CoV=11.9%) and 6.35 (CoV=12.4%) were obtained,
respectively.
The GFRP rod used in the present work, with a trademark Maperod G, is
provided in bars with 6 meters each and it is produced by MAPEI. These rods
have a diameter of 10 mm and the external surface is sand blasted. Tension tests
were carried out to assess the tensile mechanical properties of GFRP rod
according to ISO TC 71/SC 6 N - Part 1 (2003). Tests were performed at a
displacement rate of 2 mm/min. To measure the modulus of elasticity, a clip gauge
was mounted at middle region of each specimen. From the experimental tests an
average tensile strength of 778.14 MPa (CoV=3.5%), an average modulus of
elasticity of 38.42 GPa (CoV=1.3%) and an average ultimate strain of 20.25
(CoV=2.3%), were obtained.
The unidirectional GFRP sheets used in this study was provided in rolls with
100 meters long and 0.6 meters wide and it was produced by MAPEI. The
GFRP sheet has the trademark MapeWrap G UNI AX 900/60 and its tensile
behavior was assessed by uniaxial tensile tests carried out according to TC 71/SC
6 N - Part 2 (2003) recommendations. An average tensile strength of 2209.60 MPa
(CoV=13.86%), an average elasticity modulus of 101.20 GPa (CoV=4.31%) and
an average ultimate strain of 21.79 (CoV=11.32%), were obtained.
Sagging Region (S1-S1) Hogging Region (S2-S2)
REF.2
REF.1
90
400
GFRP GFRP
S1 S2
NSM.20
90
90
70 130 130 70 70 130 130 70
400 400
GFRP GFRP
S1 S2
NSM.40
90
90
37,5 65 65 65 65 65 37,5 37,5 65 65 65 65 65 37,5
400 400
GFRP (SHEET) GFRP (SHEET)
S1 S2
EBR.20
90
90
35 330 35 35 330 35
400 400
GFRP (SHEET) GFRP (SHEET)

Layer2
Layer1 Layer2 S1 Layer1
S2
EBR.40
90
90
90 45 90 45 90 90 45 90 45 90
5
15
15
15
15
400 400
Fig. 2 Specimens cross-sectional dimensions of sagging (S1-S1') and hogging regions

(S2-S2'). All dimensions in [mm]
Two distinct adhesives were used in the experimental program: MapeWrap 31

and MapeWood Paste 140, being both supplied by MAPEI. The former adhesive
is a medium viscosity epoxy resin for the impregnation of FRP sheets, whereas the
last is a thixotropic epoxy adhesive currently used in the restoration of timber
structural elements. The MapeWrap 31 was used with the EBR strengthening
technique, while the MapeWood Paste 140 was adopted to bond the rods within
the NSM technique. To assess the tensile properties of each hardened adhesives,
six tensile tests for each experimental series were carried out according to ISO
527-2 (1993). After casted, the adhesive specimens were kept in the laboratory
F F
200 200
NSM SG4 SG5 SG6
SG1 SG2 SG3 SG7 SG8 SG9
90
200 200 200 200
1000 1200 800 800 1200 1000
2400 1200 2400
6000
Top view
NSM.20
SG1 SG2 SG3
70
SG4 SG5 SG6
400
130
SG7 SG8 SG9
Botton view
6000
Top view
NSM.40
SG1 SG2 SG3
400
SG4 SG5 SG6
SG7 SG8 SG9
Botton view
6000
F F
200 200
EBR
SG4 SG5 SG6
SG1 SG2 SG3 SG7 SG8 SG9
90
200 200 200 200
1000 1200 800 800 1200 1000

2400 1200 2400
6000
SG1 SG2 SG3 SG4 SG5 SG6 SG7 SG8 SG9

EBR
400
Botton view
6000
Fig. 3 Longitudinal configuration of the strengthened slabs. All dimensions in [mm].
Table 1 Main results obtained on the mechanical characterization of the adhesive (average
values)
Adhesive Fadh,max (kN) adh,max (MPa) Eadh (GPa) adh,max ()

MapeWood
0.65 (5.5%) 15.41 (4.3%) 6.49 (2.6%) 2.82 (9.4%)
Paste 140
MapeWrap 31 18.95
1.75 (3.8%) 44.33 (4.0%) 2.77 (5.2%)
(Specimen EBR.20) (10.5%)
MapeWrap 31 17.33
1.62 (8.8%) 40.47 (8.6%) 2.58 (3.2%)
(Specimen EBR.40) (11.9%)
Notes: Ff,max= maximum force; f,max= uniaxial tensile strength; Eadh=
longitudinal elasticity modulus; adh,max= strain at adh,max; Eadh is the slope of the
curve between 0.0025 and 0.0075 of . The values between parentheses are
the corresponding coefficients of variation.
environment, as the slab specimens, and were tested after the same curing period
of time of the strengthening intervention process of the corresponding slabs when
tested. A universal test machine was used with a displacement rate of 1 mm/min.
A clip gauge mounted on the middle zone of the specimen recorded the strains,
whereas a high accurate cell load has registered the applied force. The obtained
results are presented in Table 1.
3 Preparation of the Specimens

3.1 NSM Strengthening Technique
The GFRP rods were installed into grooves of 15 mm wide and 20 mm deep,
opened in the glulam, on the top face of the hogging region and on the bottom face
of the sagging regions, by a saw cutter machine. To eliminate splinters and dust
resultant from the cutting process, sandpaper was applied inside the grooves and
they were then cleaned using compressed air. The GFRP rods were also cleaned
with acetone to remove any possible dirt, and strain gauges were glued in the
predefined locations. The grooves were filled with the epoxy adhesive MapeWood
Paste 140 using a spatula, followed by the application of a thin layer of adhesive
on the GFRP surface. The rods were then gradually introduced into the grooves
until its final position at about 75 mm from the external surface. Further
information can be found in Jorge (2010).
3.2 EBR Strengthening Technique

To eliminate splinters and dust, a sanding process with sandpaper was applied in
the zones where the FRPs systems were planned to be installed, followed by the
application of compressed air. The GFRP sheets were cut with the desired
dimensions and cleaned with acetone. The sheets were impregnate with the
MapeWrap 31 epoxy adhesive using a broche, and a hard rubber roller was used to
press the sheet to remove possible voids. After curing, strain gauges were installed
on the GFRP sheet surface. Further information can be found in Jorge (2010).
4 Results
The obtained results were analyzed in terms of maximum load, Fmax, deflection at
the peak load, max, failure modes and moment redistribution index,, which is
obtained from:
M red 2.8 FMIC 1.4 F123
=1 =1 =1 (1)
M elast ( 5 6 2.8 1.4) F123
where Mred is the bending moment in the intermediate support after moment
redistribution, and Melast is the bending moment in this section, calculated
according to the theory of elasticity using FMIC and F123 (see Fig. 1). Fig. 4
presents the F- curves, whereas Table 2 summarizes the obtained results.
4.1 Reference Slabs

The load versus deflection experimental curves, F-, obtained in both tests are
presented in Fig. 4a-b. Despite the symmetry of the test setup, the response of the
glulam slabs strips is not symmetric. The maximum values obtained for the
average applied load are similar (46.28 kN and 47.23 kN for REF.1 and REF.2,
respectively). The behaviour is linear elastic until failure (brittle behaviour), which
occurred in the hogging region. After the formation of the first cracks in the
hogging region, quite smaller bending moment redistribution was obtained for the
reference slabs (average value of 2.5%).
4.2 NSM Series

When compared with the unstrengthened slabs, in NSM.20 and NSM.40 the
response obtained for both spans was more similar. However, the most important
effect of this strengthening technique is the non-linear F- response, as visible in
Fig. 4c-d. In fact, the full linear elastic response obtained for the unstrengthened
slabs strips is replaced by an elasto-plastic behavior, with a significant plastic
branch (ductility). The NSM.20 slab strip has achieved a maximum average
applied load (Fmax) of 47.32 kN for a deflection of 53.90 mm, representing only an
increment of 1.2% in the applied load and 15.0% in the corresponding deflection.
Concerning to the NSM.40 slab strip, a maximum average applied force of
53.99 kN was attained with a corresponding deflection of 76.94 mm. At the peak
load, a moment redistribution of about 10% and 25% was obtained for the
NSM.20 and NSM.40, respectively.
4.3 EBR Series

Fig. 4e-f shows the F- experimental curves, where a quite symmetric response of
the slabs can be observed. The EBR.20 presented an ultimate average load of
56.54 kN for a deflection of 53.34 mm, representing an increment of 20.9% in the
applied load and 13.9% in the corresponding deflection, while the moment
redistribution at the peak load was of about 10%. Concerning to the EBR.40 slab
strip, a maximum average load of 67.33 kN was obtained, for a deflection of 65.22
mm, representing an increment of 44.0% in the applied load and 39.2% in the
corresponding deflection. A moment redistribution of about 9% was obtained at
the peak load for both strengthened slab strips. According to the Fig. 4e-f, the
behavior is linear elastic until the maximum load, after which a sudden drop in the
load carrying capacity is observed, revealing a behavior without ductility (fragile
failure). This behavior is confirmed by the linear variation registered during these
tests between the applied load and the strains in the SGs installed in the both
GFRP sheets and glulam, Jorge (2010). In consequence, no significant local
embedment of wood was observed in the compression side.
820 J.M. Branco, M.P. Jorge, and J. Sena-Cruuz
(a) (b)
(c) (d)
(e) (f)
Fig. 4 Experimental curvess F- of the (a) REF.1, (b) REF.2, (c) NSM.20, (d) NSM.440,
(e) EBR.20 and (f) EBR.40 specimens.
s
Table 2 Main results obtaineed in the tests
Slab Fmax (kN) max (mm) (%)

REF.1+REF.2 46.75 46.85 5%
NSM.20 47.32 (+1.2%) 53.90 (+15.0%) 10%
NSM.40 53.99 (+15.4%) 76.94 (+64.2%) 25%
EBR.20 56.54 (+20.9%) 53.34 (+13.9%) 9%
EBR.40 67.33 (+44.0%) 65.22 (+39.2%) 9%
Double Span Continuous Glu
ulam Slabs Strengthened with GFRP 8221
4.4 Failure Modes

Representative failure mo odes for all the tested series (REF, NSM and EBR) arre
presented in Fig. 6. The unstrengthened
u specimens did not show local embedmennt
in the compression reg gions. The failure occurred in the tension side. IIn
consequence of the brittlee nature of the timber behavior in tension, the failure is
sudden and violent.
The slabs strips stren ngthened with the NSM technique have showed aan
extensive cracking locateed at the existing wood between the FRP rods. Tensioon
cracks have formed in the t hogging and saggings regions, despite being morre
evident in the hogging, annd always local embedment in the compression side waas
observed in the hogging region.
For the EBR series, thhe failure always occurred by debonding of GFRP. Loccal
embedment in the com mpression region was only observed in the EBR.440
specimen.
(a) (b) (c)

des: (a) REF, (b) NSM.40 and (c) EBR.20
Fig. 5 Typical failure mod
5 Conclusions
This work explores the potentialities
p of using GFRPs systems for the flexurral
strengthening of continuo ous glulam slab strips, in terms of increasing the loaad
carrying capacity, deformmability and the bending moment redistribution capacity..
The NSM and EBR strengthening techniques were designed to provide aan
increase of 20% and 40% % of load carrying capacity of the reference slab. Thhe
aimed increase of the ultimate
u load was attained in the slabs strengtheneed
according to the EBR tech hnique, and did not occur in the NSM strengthened slabbs.
However, the NSM strrengthening technique has increased significantly thhe
deformational capacity off these statically indeterminate structural systems, leadinng
to a moment redistributio on that attained 25% in the slab strengthened with thhe
highest percentage of GF FRP. In both slabs strengthened according to the EB BR
technique the moment red distribution was 9%.
Due to the intensive crracking process before the rupture of the GFRP system ms
in the NSM flexurally streengthened slabs, a pronounced nonlinear branch betweeen
the applied load and th he deflection occurred before the ultimate deflection.
Therefore this technique has
h provided a significant increase in terms of deflectioon
capacity. In fact, the two span unstrengthened glulam slab strips (reference slabs)
presented an almost linearr relationship between the applied load and the deflectioon
up to failure, with a quitee brittle failure mode, which deformational response waas
similar to the one registtered in the slabs strengthened according to the EB BR
technique.
Acknowledgements. The present work is part of a research project supported by program

Quadro de Referncia Estratgico Nacional (QREN), project number 21635, from the
Agncia de Inovao (ADI). The authors also like to thank all the companies that have been
involved supporting and contributing for the development of this study, mainly: INEGI,
S&P Reinforcement, Portilame, MAPEI and Rothoblaas.
References
Borri, A., Corradi, M., Grazini, A.: A method for flexural reinforcement of old wood beams
with CFRP materials. J. Compos. Part B 36(2), 143153 (2005)
Branco, J., Cruz, P., Dias, S.: Old Timber Beams. Diagnosis and Reinforcement. In:
Tampone, G. (ed.) Proceedings of the Inter. Conf. The Conservation of Historic
Wooden Structures, Florence, February 22-27, pp. 417422 (2005)
Dagher, H.J., Lindyberg, R.: FRP reinforced wood in bridge applications. In: 1st RILEM
Symposium on Timber Engineering, Cachan Cedex, France, pp. 591598 (1999)
Dagher, H.J., Kimball, T., Abdel-Magid, B., Shaler, S.M.: Effect of FRP reinforcement on
low-grade eastern hemlock glulams. In: National Conf. on Wood Transportation
Structures, 9 p. (1996)
EN 408:2003, Timber structures - Structural timber and glued laminated timber.
Determination of some physical and mechanical properties. European Committee for
Standardization (CEN), Brussels
Gilfillan, J.R., Gilbert, S.G., Patrick, G.R.H.: The use of CFRP composites in enhancing
the structural behaviour of timber beams. Jour. of Reinfor. Plast. and Comp. 22(15),
13731388 (2003)
ISO 527-2:1993, Plastics - Determination of tensile properties - Part 2: Test conditions for
moulding and extrusion plastics. International Organization for Standardization
ISO TC 71/SC 6 N Part 1:2003, Non-conventional reinforcement of concrete-test methods:
Fiber reinforced polymer (FRP) bars. International Organization for Standardization
ISO TC 71/SC 6 N Part 2:2003, Non-conventional reinforcement of concrete-test methods:
Fiber reinforced polymer (FRP) sheets. International Organization for Standardization
Jorge, M.A.P.: Experimental behavior of glulam-FRP systems. MSc Dissertation,
Department of Civil Enginerring, University of Minho, 247 p. (2010)
Lopez-Anido, R., Xu, H.: Structural characterization of hybrid fiber-reinforced polymer-
glulam panels for bridge decks. Jounal of Composites for Construction 6(3), 194203
(2002)
NP EN 1194:1999, Estruturas de madeira. Madeira lamelada-colada. Classes de resistncia
e determinao de valores caractersticos. IPQ, Lisbon (in Portuguese)
Micelli, F., Scialpi, V., Tegola, A.L.: Flexural reinforcement of glulam timber beams and
joints with carbon fiber-reinforced polymer rods. J. Compos. Constr., ASCE 9(4),
337347 (2005)
Corradi, M., Borri, A.: Fir and chestnut timber beams reinforced with GFRP pultruded
elements. Composites 38, 172181 (2007)
Raftery, G., Harte, A.: Low-grade glued laminated timber reinforced with FRP plate.
Composites Part B: Eng. 42(4), 724735 (2011)
Buell, T.W., Saadatmanesh, H.: Strengthening timber bridge beams using carbon fiber. J.
Struct. Eng., ASCE 131(1), 173187 (2005)
Gentile, C., Svecova, D., Rizkalla, S.H.: Timber beams strengthened with GFRP bars:
development and applications. Journal of Composites for Construction, ASCE 6(1),
1120 (2002)
Simplified Design of Glued Laminated Timber
Girders for the Torsional Moment Caused by
Stability Effects
R. Hofmann and Ulrike Kuhlmann
Universitat Stuttgart, Institute of Structural Design, Pfaffenwaldring 7,

70569 Stuttgart, Germany
{Reiner.Hofmann,U.Kuhlmann}@ke.uni-stuttgart.de
Abstract. This paper presents a numerical study on the torsional moment in lateral
torsional stability analysis of glued laminated timber girders. A numerical model is
developed solving the bending-torsion problem. This numerical model according to
second order theory also takes single supports at the upper chord into account, which
are typical for wide span girders as roof beams with horizontal bracings in order
to reduce the relative slenderness of the girders. The numerical studies on straight
parallel-chorded glued laminated beams show, that the torsional stresses on buckling
girders are in the range of 4-12% of the allowable shear stresses in cross-section at
the end supports considering two different curvatures of imperfection. In practice the
torsional moment due to stability analysis on the lateral torsional buckling should
be derived in a simple and secure way to be able to verify it against the relatively
low shear strength of timber without using a calculation according to second order
theory.
Keywords: glued laminated timber girders, torsional moment, shear stresses at end
support, slenderness ratio, lateral torsional buckling, second order theory, design
proposal.
1 Introduction
The tendency for increasing slenderness in structural timber buildings leads to an
increase of the importance of additional deformations due to stability phenomena
such as lateral torsional buckling. Due to the wide spans a simple verification of the
bending and shear stresses does not suffice for the ultimate limit state. Imperfections
of the system may generate effects of second order theory. These effects increase
more than proportionally with the span of the girder, so the stability analysis of such
girders becomes more important, because slender girders under bending may fail by
lateral torsional buckling. The usual procedure to verify the lateral torsional stability
refers only to the bending moment (see Equation (1)) and does not consider the
torsional moment, which is accompanying the bending moment caused by second
order theory.
DOI: 10.1007/978-94-007-7811-5_73, c RILEM 2014
824 R. Hofmann and U. Kuhlmann
MyI
Wy
1, 0 (1)
kcrit fm,d
In order to define the influence of the torsional moment on the shear stress of glued
laminated timber girders at the end support (see Figure 1), a computer program has
been developed to perform numerical simulations solving the differential equations
describing lateral torsional buckling of glued laminated timber girders based on
second order theory.
Fig. 1 Forked end support of a glued laminated timber girder
This numerical model also takes single supports at the upper chord into account
in order to reduce the relative slenderness of the girder, which are typical for wide
span girders as roof beams with horizontal bracings.
2 Lateral Torsional Buckling According to Second Order

Theory
Lateral torsional buckling of a girder is mathematically expressed as a system of
three coupled differential equations ([1] [2]).

qz = EJy w
el + (Mz ges )
(2)

qy = EJz v
el + (My ges )
(3)
qy ez,qy qz ey,qz = GJT el EJ el
Myq vges Mz wges
qy el ey,qy
qz el ez,qz (4)
Torsional Moment in Lateral Torsional Stability Analysis 825
The derivation of the differential equations results from a static equilibrium sys-
tem for a girder with imperfections supported at the ends by effective lateral and
torsional restraints. The derivation is based on a simple symmetric cross-section,
linear-elastic behaviour of material and small deformations. An analytic solution of
the system of differential equations is only given for a small number of cases. The
developed computer program is based on Equation (2) to Equation (4) and solves
the bending-torsion problem according to second order theory using the Finite Dif-
ferences method.
3 Basic Principles of the Numerical Model
3.1 Approach of Imperfections

For calculations according to second order theory the effects of imperfections have
to be taken into account. Eurocode 5 [3] assumes an eccentricity e = 0.0025l,
which should be applied to the girder with an initial sinusoidal curvature between
the supports of the structure (scs), acc. to Figure 2a). In general the imperfections
a) b)
Fig. 2 Top view on a girder with discrete horizontal supports at the upper chord with imper-
fections a) sinusoidal curvature between the supports (scs) b) sinusoidal curvature between
the end supports (sces)
should be applied in a way, that leads to the most unfavourable stress distribution in
the girder. That is why another initial sinusoidal curvature between the end supports
(sces) of the girder acc. to Figure 2b) has been investigated. First calculations on a
16 cm wide and 112 cm high cross-section show, that the sinusoidal curvature be-
tween the end supports on a straight girder results in higher torsional stresses at the
end supports of the girder than the sinusoidal curvature between the horizontal sup-
ports of the structure (scs), (comp. Figure 3). Figure 3 shows also, that the torsional
stresses at the end supports of girders depend on the number of supports of the up-
per chord. While the torsional stresses increase on girders with sinusoidal curvature
between the end supports of the girder with an increasing number of supports, the
torsional stresses decrease for girders with sinusoidal curvature between the hori-
zontal supports of the girders with an increasing number of supports. Altogether the
torsional stresses at the end supports of straight girders in Figure 3 are not higher
than 12% of the shear stress resistance.
0.12
0.10
utilization , tor 0.08 Em2s (sces l/400)

Em3s (sces l/400)
0.06 Em4s (sces l/400)
Em2s (scs l/400)
0.04
Em3s (scs l/400)
0.02 Em4s (scs l/400)
0.00
0.45 0.55 0.65 0.75 0.85
slenderness rel,m
Fig. 3 Comparison of the utilization due to torsional stresses at the end supports according
to second order theory on a 16 cm wide and 112 cm high cross-section of a simply supported
girder with two (Em2s), three (Em3s) and four (Em4s) horizontal supports of the upper chord
for different curvatures according to Figure 2
3.2 Shear Strength and Modulus of Shear of Glued Laminated

Timber
The shear strength of glued laminated timber is decisive for the analysis of the tor-
sional moment in lateral torsional stability analysis. [4] performed a considerable
number of experimental series on timber and glued laminated timber girders. For a
feasible calculation of the shear stress due to torsion an isotropic approach for the
anisotropic timber has been chosen. The 5%-fractile of torsional strength of square
cross-section is specified to ftor,k = 3.48 N/mm2 for glued laminated timber due to
the experiments. The shape of the cross-section causes an increase of the torsional
resistance of rectangular cross-sections h/b4, according to [4] depending on the
shape of cross-section a shape factor is determined to 1.2. The shear stress due to
torsion can be verified with

tor,d
1 (5)
1.2 ftor,d
which is quite similar to the current verification in Eurocode 5

tor,d
1 (6)
kshape fv,d

1 + 0.15 hb
kshape = min
1.3
The shear strength due to the shear force has been investigated in several research
projects, see [5], [6], [7], [8]. The characteristic shear strength is determined to
fv,k,0 = 2.5 N/mm2 (7)

while according to [8] the shear strength of a girder depends on its length and depth
0.09
h0 l0
fv,k = fv,k,0 (8)
hl
h0 = 600 mm
l0 = 6000 mm
For the following numerical studies shear stresses are verified by
2 2
tor,d y,d z,d
+ + 1 (9)
kshape ftor,d fv,d fv,d
Aside of the shear strength the modulus of shear and its approach for numerical
calculation of deflections and internal forces is decisive for the results. The modulus
of shear in timber has been subject of several research projects in the past. Based on
[9] and [10], in [7] a constant value for the shear modulus Gg,mean = 650 N/mm2 of
glued laminated timber has been derived and is taken into account for the following
numerical studies.
4 Results of the Numerical Studies

In the first numerical studies straight parallel-chorded glued laminated girders have
been investigated. The dimensions of the girders were adapted to typical dimensions
in practice. The width of the investigated girders has been varied between 12 and
22 cm, the length between 10 and 30 m, the height is dependent on the length and
is assumed to h l/17 according to [11]. In a first series imperfect girders with
a sinusoidal curvature between the supports have been analysed, Figure 4 shows
the utilization due to torsional stresses at the end supports for girders with two and
0.10
0.09
0.08
utilization , tor
0.07
PT2s h/b=7
0.06 PT2s h/b=8
PT2s h/b=9
0.05 PT2s h/b=10
0.04
0.03
0.65 0.75 0.85 0.95 1.05 1.15 1.25
slenderness rel,m
Fig. 4 Utilization of shear due to the torsional moment at the end support of a straight single
span girder with parallel chords and two horizontal supports at the upper chord according to
bending-torsion second order theory with sinusoidal curvature of imperfection between the
supports
0.10
0.09
0.08
utilization , tor
0.07
PT3s h/b=9
0.06 PT3s h/b=10
0.05
0.04
0.03
0.65 0.75 0.85 0.95 1.05 1.15 1.25
slenderness rel,m
span girder with parallel chords and three horizontal supports at the upper chord according to
supports
0.10
0.09
0.08
utilization , tor
0.07 h/b=5 (Em2s)

h/b=6 (Em2s)
0.06 h/b=7 (Em2s)
h/b=8 (Em2s)
0.05 h/b=9 (Em2s)
h/b=10 (Em2s)
0.04
0.03
0.45 0.55 0.65 0.75 0.85 0.95 1.05 1.15 1.25
slenderness rel,m
span girder with parallel chords and two horizontal supports at the upper chord according to
end supports
Figure 5 for girders with three horizontal supports at the upper chord. The utilization
due to torsional stresses stays below 10% of the allowable shear stresses according
to second order theory in all analysed cases. Obviously the torsional stresses at the
end support depend on the proportion of h to b of the cross-section.
In a second series the same girders as in series one have been investigated, but the
imperfection has been changed to a sinusoidal curvature between the end supports.
The results in Figure 6 and Figure 7 show, that the torsional stresses of these girders
are insignificantly higher than the torsional stresses of girders with sinusoidal cur-
vature of imperfection between the supports of the structure. Although the influence
of slenderness rel,m and the h to b proportion of the cross-section is less. Since
the total amount of the torsional stress in cross-section at the end supports for both
curvatures of imperfection is quite small, a simplified design of the glued laminated
0.12
0.11
0.10
utilization , tor
0.09
0.08 Em3s h/b=7
Em3s h/b=8
0.07 Em3s h/b=9
Em3s h/b=10
0.06
0.05
0.45 0.55 0.65 0.75 0.85 0.95 1.05 1.15 1.25
slenderness rel,m
span girder with parallel chords and three horizontal supports at the upper chord according to
end supports
timber girders for torsional moment needs not to distinguish between several h to b
proportions.

The first numerical studies on straight parallel-chorded glued laminated girders
show, that the torsional stresses of girders due to lateral torsional buckling is in
the range of 4-12% of the allowable shear stresses in the cross-section at the end
supports. Eurocode 5 does not explicitly consider the torsional stresses of glued
laminated timber girders at the end support. A design proposal should consider the
torsional stresses in the current verification of lateral torsional buckling according
to Eurocode 5 and should lead to an simple and secure design of girders. Therefore
the present studies will be extended to saddle roof girders and fish-bellied girders.
The aim is to derive a simple and secure design proposal for the torsional moment
avoiding a calculation according to second order theory in practice.
Acknowledgements. This research work (17398N) coordinated by the International Associ-

ation for Technical Issues related to Wood (iVTH e.V.) is supported in the program of Indus-
trielle Gemeinschaftsforschung (IGF) and financed by the German Federation of Industrial
Research Association (AiF).
References
[1] Petersen, C.: Statik und Stabilitat der Baukonstruktionen. Friedr. Vieweg und Sohn
(1980)
[2] Roik, K., Carl, J., Lindner, J.: Biegetorsionsprobleme gerader dunnwandiger Stabe. Ver-
lag von Wilhelm Ernst und Sohn (1972)
[3] EN 1995-1-1:2004: Eurocode 5: Design of timber structures; Part 1-1: General Com-
mon rules and rules for buildings (2004)
[4] Mohler, K., Hemmer, K.: Rechnerischer Nachweis von Spannungen und Verformungen
aus Torsion bei einteiligen Vollholz- und Brettschichtholzbauteilen. Holzbau-Statik-
Aktuell Folge 2 (1977)
[5] Glos, P., Denzler, J.K.: Kalibrierung der charakteristischen Schubfestigkeitskennwerte
fur Vollholz in EN 338 entsprechend den Rahmenbedingungen der nationalen Sortier-
norm/Technische Universitat Munchen, Bericht Nr. 04502 Forschungsbericht (2004)
[6] Schickhofer, G.: Determination of Shear Strength Values for GLT using Visual and
Machine Graded Spruce and Laminations, Paper 34-12-6, CIB-W18-Meeting (August
2001)
[7] Brandner, R., Gehri, E., Bogensperger, T., Schickhofer, G.: Determination of modulus
of shear and elasticity of glued laminated timber and related examinations, Paper 40-
12-2, CIB-W18-Meeting (August 2007)
[8] Klapp, H., Bruninghoff, H.: Shear strength of glued laminated timber, Paper 38-6-3,
CIB-W18-Meeting (August 2005)
[9] Mohler, K.: Strength and long-term behaviour of lumbar and glued-laminated timber
under torsion loads, Paper 7-6-1, CIB-W18-Meeting, Februar/Marz 1977
[10] Gorlacher, R., Kurth, J.: Determination of shear modulus, Paper 27-10-1, CIB-W18-
Meeting (Juli 1994)
[11] Cyron, G., Sengler, D.: Holzleimbau, Bauen mit Brettschichtholz. Informationsdienst
Holz; Dusseldorf; Arbeitsgemeinschaft Holz (1985)
Author Index
Adam, Con Y. 787 Eberhardsteiner, Josef 703

Aicher, Simon 135, 341, 603, 641, 703 Enquist, Bertil 45, 325, 801
Alexander, John 615
Amini, M. Omar 531 Fadai, Alireza 471
Ansell, Martin P. 231, 657 Fink, Gerhard 67
Flaig, Marcus 667
Bailleres, Henri 119, 577, 629 Fragiacomo, Massimo 273, 519, 677, 689
Bathon, Leander 375, 493 Frangi, Andrea 57, 211, 265
Beikircher, Wilfried 569 Franke, Bettina 243, 395
Bla, Hans Joachim 77, 157, 301, 667 Franke, Steffen 221
Bletz-Mhldorfer, Oliver 375, 493 Frese, Matthias 77
Bongers, Ferry 615 Fuchs, Michael 471
Branco, Jorge M. 725, 813 Fssl, Josef 703
Brandner, R. 289, 751
Brandon, Daniel 231, 255
Gabriel, Joseph 405
Bredesen, Ronny 355
Bregulla, Julie 231 Gavric, Igor 689
Brhl, Frank 181 Gilbert, Benoit P. 119, 577, 629
Grstad, Kristin 355
Cabo, Jos L.F. 507, 569
Ceccotti, Ario 689 Hdicke, Wolfram 423
Chang, Wen-Shao 657 Hafner, Annette 89
Chen, J. 589 Hairstans, Robert 109
Clifton, Charles 307 Haller, Peer 99
Crews, K. 443 Hancock, Steven B. 119
Crocetti, Roberto 507 Harris, Richard 231
Harrison, Ian 615
Denzler, Julia K. 761 Hartig, Jens 99
de Vries, P.A. 773 Hjiaj, Mohammed 433
Diehl, Friedemann 375 Hochreiner, Georg 703
Dill-Langer, Gerhard 603 Hofmann, R. 823
Dorn, Christian 499 Hornatowska, Joanna 325
Duchne, Jrme 433 Huang, Puxi 657
Dujic, B. 677, 713 Humbert, Jrme 13, 455
832 Author Index
Ito, Takumi 191 Nguyen, Quang Huy 433

Niewerth, Dominik 149
Jahreis, Markus 423, 485, 499
Jamil, Mamoon 307 Ohtsuka, Akiko 191
Jiang, Y. 365 Olsson, Anders 801
Jockwer, Robert 265 Oscarsson, Jan 45, 801
Johnsson, Helena 3 Ott, Stephan 31, 89
Jorge, Marco P. 813
Jorissen, Andr 273 Park, Joo-Saeng 13, 455
Park, Moon-Jae 13, 455
Kambe, Wataru 191 Patterson, Dale 577
Karacabeyli, E. 543 Pei, Shiling 531, 543
Kstner, Martin 423 Perez, Fernando Perez 615
Kekeliak, Milos 725 Popovski, Marjan 531, 543, 689
Kim, Kwang-Mo 455
Kitamori, Akihisa 739 Quenneville, Pierre 167, 243, 307
Knorz, M. 365
Kobel, Peter 211 Rammer, Douglas 531
Kohler, Jochen 67
Rautenstrauch, Karl 423, 485, 499
Komatsu, Kohei 739
Ravenshorst, G.J.P. 773
Kuhlmann, Ulrike 181, 823
Rinaldin, Giovanni 519
Kuklk, Petr 463
Ringhofer, A. 203, 289
Kuklkov, Anna 463
Sandhaas, Carmen 157
Lawrence, Andrew 129
Sartori, Tiziano 507
Lawrence, Mike 657
Schaffrath, J. 365
Lee, Sang-Joon 13, 455
Lehringer, Christian 405 Schnzlin, Jrg 181
Leijten, Adriaan 255 Scharmacher, Florian 395
Leitch, Kenneth 109 Schickhofer, G. 203, 289, 751
Li, Z. 589 Schmidt, Jens 375
Line, Phil 531 Schwarz, Ulrich 381
Loebus, Stefan 31 Sena-Cruz, Jos 813
Loureno, Paulo B. 725 Serrano, Erik 45, 325
Shan, B. 589
Magnire, Nolie 221 Shea, Andy 657
Mahlknecht, U. 289 Shrestha, R. 443
Manthey, Manuel 433 Somja, Hugues 433
Marcroft, Julian 615 Stapf, Gordian 341, 641
McGavin, Robbie L. 577, 629 Steiger, Ren 211, 265, 569
Menendez, Jesus M. 109 Steilner, Michael 301
Miebach, Frank 149 Sterley, Magdalena 45, 325
Milner, H.R. 787 Stief, Alexander 499
Mori, Takuro 739 Stritzke, C. 135
Moshiri, F. 443 Sustersic, I. 677, 713
Mller, Andreas 395
Takahashi, Sumiya 191
Nakashima, Shoichi 739 Tomasi, Roberto 507
Nslund, Ida 3
Nechanick, Pavel 463 Underhill, Ian D. 577, 629
Author Index 833
van de Kuilen, Jan-Willem 157, 773 Winter, Stefan 31, 89, 365
van de Lindt, John W. 531, 543 Winter, Wolfgang 471
Wrzesniak, Daniela 273
Walker, James 555
Walker, Pete 231
Wanninger, Flavio 57 Xiao, Robert 555
Wehsener, Jrg 99 Xiao, Y. 589
Weidenhiller, Andreas 761
Werner, Tom-Egmont 99 Yang, R.Z. 589
Widmann, Robert 569
Winkler, Christoph 381 Zarnani, Pouyan 167
Keyword Index
A bond line shear strength 45

bridges 493
acetylated wood 615 brittle 167
adhesive bonds 45 brittle damage 129
adhesives 569 brittle failure 273
analytical model 443 Buckling 135
anchorage length 456 building construction 471
Araucaria Cunninghamii 119 building envelope 31
assessment of existing timber
structures 395 C
Australian hardwood plantation
forests 577
Australian structural glulam 787 Canadian spruce 243
axial compression 99, 119 carbon dioxide emission 589
Axial loading 201 chemical anchor 456
azob 157 chestnut 77
clamping pressure 341
closed panel system 109
B CLT structures 543
column tests 119
beam column connection 57 components 589
beam lay-up 667 composite structures 433
beam on elastic foundation 341 compression strength 603
beech 157, 603 concentrated loading 703
beech LVL 211 concrete 433,713
beech solid wood 221 conductive filled adhesive 381
beech wood 569 conductivity and diffusivity 657
bending strength 569, 641 connection slip 231
block glulam 341 connections 689, 725
block shear 341 core samples 395
block shear failure 289 cross-laminated timber (CLT) 531, 543,
Block-Gluing 149 677, 667, 689, 703, 713, 739
block-shear failure 273 cross-linking component 355
bond line 325, 405 cyclic behaviour 519
836 Keyword Index
cyclic later loading 689 Eurocode 5 615

cyclic loading 15, 433 European spruce 243
evolution of crack modes 703
D expanded tube fasteners 255
experimental evaluation 725
damage 67 F
delamination 341
delamination tests 395 faade construction 31
densified veneer wood (DVD) 231, 255 fagus sylvatica 221
densified wood 99 failure mechanisms 555, 689
design and construction 589 failure modes 157, 289, 667, 813
design approach 265 failures of structures 67
design model 167 fatigue 493
design proposal 243, 823 FEMA P695 methodology 531
deterioration 67 fiber reinforced concrete slab glulam 507
digital image correlation 325 finger joints 325, 569, 641
dot laser 801 finite element model 3
double shear connection 243 fire behaviour 405
double span 813 fire safety 31
dowel diameter 221 flexural strengthening 813
dowel type connections 181, 211, 243 flitch plate connection 255
dowel type fasteners 301 floor system 507
dowel type joints 739 force distribution 201
ductile and mixed failure modes 167 force modification factor 543
ductile damage 129 fracture 325
ductile nonlinear response 543 fracture mechanics 243
ductile structural behaviour 703 friction damping 307
ductility 15, 211, 405 FRP 423, 485
durability 45 fully threaded screws 265
duration of load 67
DVD reinforced connections 255 G
global failure mechanism 703

E gluability of softwood and hardwood 365
glubam 589
ECO2 project 90 glue line quality 395
elastic spring model 167 glue line thickness 341, 355
elasticity 405 glued composite structure 135
electrical modified wood adhesives 381 glued connections 375
embedment length 265 glued in rod 555
embedment strength 211, 221, 231 glued in shear connectors 493
emulsion component 355 glued laminated timber 365
emulsion polymer isocyanate (EPI) glued laminated timber girders 823
adhesive 355 glued laminated veneer lumber 603
end of life 90 glulam 45, 149, 423, 569, 603, 641, 801
energy consumption 589 glulam slabs strengthened with
energy dissipation 689 GFRP 813
energy efficiency 31 grain deviation 761
engineered wood products 405 gravity rocking 307
Keyword Index 837
green gluing 45, 325 load paths 129

grillage type structures 129 load shearing 289
groups of screws 289 load-sharing 603
Gympie messmate plantation load-slip behaviour 463
thinning 629 load-to-grain angle 221
local heating 381
H lognormal 751
logs 77
hardwood 221, 603, 641 long-term test 507
hardwood poles 629 low-damage 307
heat capacity 657 LVL 57, 423, 603
heat resistance 405
heat transfer models 657 M
hierarchical model 751
high strength steel dowels 157 mechanical fasteners 243
hollow utility poles 629 mechanical performance 629
hybrid elements 499 mechanical properties 589, 615
hysteretic rule 519 melamine urea formaldehyde (MUF) 569
metallic connections 15
I mineral surface 499
minimum spacing 289
identification 761
mixed failure modes 703
inclined screws 265
MMC 109
increase of load carrying capacity 813
modified bending test 395
indeterminate structures 129
moment capacity 255, 555
influence of moisture 381
moment redistribution 813
in-plane stiffness 725
moment rotation relationship 555
moment-angle-curve 301
J
moment-rotation relationship 255
Monte-Carlo simulation 181
Johansens yield model 739
Joint 192 Moso bamboo 657
moulded wooden tubes 99
K multi-storey buildings 677
Keel web element 135 N
L nailed connections 167

nonlinear bending 135
laminated veneer lumber 211 nonlinear load deformation relationship
laminating effects 787 739
lamination and glulam characteristic non-metallic timber connections 231
strengths 787
lateral displacement 3 O
lateral torsional buckling 823
length effect 751 oak 77
life cycle 90 one-component PUR adhesives
light frame constructions 519 (1C-PUR) 405
lightweight timber element 135 OSB 135
load carrying capacity 265, 739 over-strength factor 181
838 Keyword Index
P resource efficiency 471

reuse 90
parallel system 611, 752 R-factor 543
performance requirements 787 rheology 355
phenol-resorcinol formaldehyde rigid body spring model 739
(PRF) 569 rigid bond 423,485
photogrammetry 499 risk analysis 67
pinching effect 519 rivet connections 167
pinus koraiensis Sieb 456 rocking behaviour 57
plantation thinning 577 rod pull out anchorage length 555
plastic deformation 301 round wood 77
plastic hinges 157
plywood 135, 192 S
poles 99
polymer concrete 423, 485 SAP2000 307
porosity 657 SBB shear connector 433
portal frame connections 555 scanning 801
post and beam 3 second order theory 823
post-buckling behaviour 119 seismic 307
post-tensioning 57 seismic design 677, 689
pot life 355 seismic performance factors 531
prefabricated panel 463 seismic strengthening 713
prefabrication 31 self tapping screws 201, 301, 265, 507
press time 355 service class 3 615
production 589 serviceability limit state 3
pultruded GFRP dowels 231 SFS Intec 443
punched metal plate 463 shake table 307
push-Out tests 443, 433 shear anchor 463
push-pull 201 shear capacity 135
shear connection 433
Q shear failure 456
shear modulus 773
q-factor 543 shear strength 395
quercus petraea 641 shear stresses at end support 823
quercus robur 641 shear wall 109, 499, 543
side boards 801
R single and coupled wall systems 689
size effect 641
racking 109 slenderness ratio 823
ratio of local and global Modulus of slip modulus 443
Elasticity 773 splitting 273
recycling 90 spruce 157
refurbishment 31 S-shaped webs 135
rehabilitation 485 stabilizing element 3
reinforced timber 423, 485 steel 375
reinforcement 181 steel connectors 77
reliability 181 steel rebar 456
requirements on finger joints 751 steel-to-timber joints 157
resistant mechanism 192 stiffness 443
Keyword Index 839
stiffness 507 Timber-lightweight concrete 471

Strain gauge measurement 201 torsional moment 823
strength grading 761 two component adhesive system 375
strength model 77
strength reduction factors 667 U
strengthening 725
strengthening techniques 813 ultimate capacity 167
structural (hardwood) timber 773 unreinforced masonry 713
structural collapse 395
structural design 615 V
structural engineering 471
structural health monitoring (SHM) 381 variation along radial direction 657
support design 135 veneer based composites structural products
sustainable building 471 577, 629
sustainable redevelopment 471 veneers 119
system effect 603
W
T
weak zones 773
tapered beams 667 wettability 365
tensile-shear strength 365 white oak 641
tensile strength 641 wind power plant 99
Tensile strength parallel to grain 751 withdrawal capacity 201
thermal effusivity 657 wood 375
thermally modified timber (TMT) 569 wood chip concrete 471
thin walled timber C-sections 119 wood extractive 365
timber 375, 433, 751,761 wood-based structural elements 471
timber bridge 456 wood-concrete-composite 493
timber concrete composite floor wooden structures 519
(TCC) 443
timber connections 519 X
timber construction 31
timber structures 90 X-lam buildings 519
timber to concrete 109 x-ray tomography 325
timber-anhydrite-slabs 499
timber-concrete composite 149, 456, 463, Y
507
timber-frame 3 yield moment 301

(RILEM Bookseries 9) Ida NÃ Slund, Helena Johnsson (Auth.), Simon Aicher, H.-W. Reinhardt, Harald Garrecht (Eds.) - Materials and PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(RILEM Bookseries 9) Ida NÃ Slund, Helena Johnsson (Auth.), Simon Aicher, H.-W. Reinhardt, Harald Garrecht (Eds.) - Materials and PDF

Uploaded by

Copyright:

Available Formats

RILEM Bookseries

Materials and Joints

RILEM, The International Union of Laboratories and Experts in Construction

For further volumes:

Materials and Joints

ISSN 2211-0844 ISSN 2211-0852 (electronic)

Library of Congress Control Number: 2013949137

Printed on acid-free paper

Regarding mechanical joints, self-tapping screws produced at lengths of up to

The Conference was kindly supported by the following enterprises:

Sole Plate Fixing Details for Modern Methods of Timber

Part II: Mechanical Connections

Modelling of Non-metallic Timber Connections at Elevated

Resistance and Failure Modes of Axially Loaded Groups of Screws . . . . . 289

Part III: Glued Joints and Adhesives

Fatigue Performance of Adhesive Connections for Wooden Wind

Part IV: Timber and Concrete/Cement/Polymer

Shear Performance of Wood-Concrete Composite with Different

Fatigue Performance of Single Span Wood-Concrete-Composite

Part V: Cyclic, Seismic Behaviour

Part VI: Hardwood, Modified Wood and Bamboo

Structural Performance of Accoya Wood under Service Class 3

Part VII: Cross-Laminated Timber

Part VIII: Properties and Testing of Wood

Part IX: Glulam

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 831

Keyword Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 835

Ida Nslund and Helena Johnsson

Division of Structural and Construction Engineering, Lule University of Technology,

Keywords: Stabilizing element, finite element model, lateral displacement, tim-

1.1 Case Study Houses

Fig. 1 Shear wall segment, see figure 3 for detail A

2 Materials and Method

2.1 Light-Frame System

Table 1 Material properties

Timber Glulam Plywood

Table 2 Stiffness of Sylomer

Table 3 Stiffness of connections

Type Plywood Particleboard Anchorage Bolts

2.2 Post and Beam System

2.4 Finite Element Model

Sheating-to-frame joints Frame Sheet

4 Conclusion and Discussion

Kllsner, B.: Panels as wind-bracing elements in timber-framed walls, Stockholm, Sweden

Jrme Humbert*, Sang-Joon Lee, Joo-Saeng Park, and Moon-Jae Park

Korea Forest Research Institute (KFRI), 57 Hoegiro, Dongdaemun-gu, Seoul, Korea

Fig. 1 A traditional Korean timber house Hanok composed of an apparent post-and-beaam

Concerning the lateraal resistance, the post-and-beam structure is known tto

these modifications. In paarallel, a refined 3D finite element (FE) model is derived,

Keywords: post and beam

2.1 Materials and Methods

Fig. 2 Metallic connector

2.2 Experimental Results

Fig. 6 presents the evo

subsequently loaded on thet negative side. In this particular test, an experimenttal

ment M in the joint function of the rotation angle

In order to quantify th he ductility, a possible approach is to use the Equivalennt

Fig. 8 FE mesh of the specim

Concerning the uplift, the measured values show no significant relative

3.1 Numerical Model

Table 1 Material properties of the FE model

Material Parameter Value

3.2 Numerical Sim