Professional Documents
Culture Documents
Simon Aicher
H.-W. Reinhardt
Harald Garrecht Editors
ABC
Editors
Simon Aicher Harald Garrecht
University of Stuttgart University of Stuttgart
Stuttgart Stuttgart
Germany Germany
H.-W. Reinhardt
University of Stuttgart
Stuttgart
Germany
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.
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
Simon Aicher
H.-W. Reinhardt
Harald Garrecht
Sponsors
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
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.
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.
Dead load
Lateral displacement
Wind load
Dead load
Lateral displacement
Wind load
Fig. 2 Post and beam wall segment, see figure 3 for detail B
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.
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
Horizontal Displacements in Medium-Rise Timber Buildings 7
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.
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.
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
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.
Horizontal Displacements in Medium-Rise Timber Buildings 9
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
10 I. Nslund and H. Johnsson
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
Horizontal Displacements in Medium-Rise Timber Buildings 11
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)
12 I. Nslund and H. Johnsson
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 13
DOI: 10.1007/978-94-007-7811-5_2, RILEM 2014
14 J. Humbert et aal.
2 Experimental Tests
T
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.
Improving the Moment Resistance of a Concealed Timber Post Base Joint 117
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
18 J. Humbert et aal.
Fig. 5 Failure of short-lengtth welds on the lower face of the upper horizontal plate of thhe
connector base
Fig. 7 Bilinear model (EEEP curve) for the original configuration, displaying a low ductility
20 J. Humbert et aal.
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.
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
Fig. 14 EEEP curve (bilinear model) for the full-length weld configuration. Mu = 155.9
kN.m.
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.
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.
Improving the Moment Resistance of a Concealed Timber Post Base Joint 29
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
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 31
DOI: 10.1007/978-94-007-7811-5_3, RILEM 2014
32 S. Loebus, S. Ott, and S. Winter
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
Structure
Hygrothermal
Economy
behaviour
Existing
building
Ecology Fire safety
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
34 S. Loebus, S. Ott, and S. Winter
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.
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.
The Multifunctional TES-Faade Joint 35
Fig. 4 Development of platform frame, (a) balloon-, and semi balloon-frame (b-c) to TES
(d) in analogy to (c)
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.
The Multifunctional TES-Faade Joint 37
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.
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
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
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.
42 S. Loebus, S. Ott, and S. Winter
14 Outlook
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)
The Multifunctional TES-Faade Joint 43
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 45
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).
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.
Green-Glued Products for Structural Applications 49
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.
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.
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
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.
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
Fig. 3 Measured twist after various climatic conditioning schemes for series III
Green-Glued Products for Structural Applications 53
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.
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
54 E. Serrano et al.
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
Green-Glued Products for Structural Applications 55
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 57
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.
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.
60 F. Wanninger and A. Frangi
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.
200 800
150 600
M [kNm]
P [kN]
100 400
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
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
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
64 F. Wanninger and A. Frangi
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.
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)
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
Experimental Analysis of a Post-tensioned Timber Connection 65
displacement due to the load, which only depends on the geometry and the modulus
of elasticity of timber perpendicular to the grain.
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.
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.
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
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.
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 67
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
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:
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
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
Risk Based Investigations of Partly Failed or Damaged Timber Constructions 71
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
72 G. Fink and J. Kohler
decisions concerning the further procedure, such as repair alternatives or more de-
tailed investigations, can be made.
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
Risk Based Investigations of Partly Failed or Damaged Timber Constructions 73
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
74 G. Fink and J. Kohler
Table 4 Compilation of the estimated material properties and the corresponding relability
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]
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
Risk Based Investigations of Partly Failed or Damaged Timber Constructions 75
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.
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 77
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
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
80 M. Frese and H.J. Bla
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.
Naturally Grown Round Wood Ideas for an Engineering Design 81
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)
82 M. Frese and H.J. Bla
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.
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.
84 M. Frese and H.J. Bla
18.0 400
15.5 300
Rax [kN]
Ft [kN]
13.0 200
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
Naturally Grown Round Wood Ideas for an Engineering Design 85
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.
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)
86 M. Frese and H.J. Bla
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.
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)
88 M. Frese and H.J. Bla
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 89
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.
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].
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
Recycling and End-of-Life Scenarios for Timber Structures 93
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.
94 A. Hafner, S. Ott, and S. Winter
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.
Solid-CS 1415 25 9 6 91
Lattice 836 19 12 5 41
96 A. Hafner, S. Ott, and S. Winter
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
Recycling and End-of-Life Scenarios for Timber Structures 97
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.
98 A. Hafner, S. Ott, and S. Winter
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
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
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 99
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
2.1.2 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.
102 J. Wehsener et al.
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.
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).
Advancements for the Structural Application 103
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
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.
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
108 J. Wehsener et al.
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 109
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
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.
Table 2 Strength and stiffness results of the isolated sole plate components
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
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.
114 J.M. Menendez, K. Leitch, and R. Hairstans
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
Table 5 Isolation and combination test results for open and closed panel sole plate details
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).
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.
Sole Plate Fixing Details for Modern Methods of Timber Construction 117
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.
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.
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.
118 J.M. Menendez, K. Leitch, and R. Hairstans
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
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.
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 119
DOI: 10.1007/978-94-007-7811-5_11, RILEM 2014
120 B.P. Gilbert, S.B. Hancock, and H. Bailleres
Fig. 1 Examples of cold-formed steel profiles, with and without intermediate stiffeners
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
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
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.
3 Material Testing
4 Stub-Column Tests
(a) (b)
Fig. 6 Stub-column test set-up (shown for Type A cross-section), elevations: (a) flange
view, (b) web view
where MORapp is given in Table 3 and A is the nominal cross-sectional area given
in Table 1.
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.
126 B.P. Gilbert, S.B. Hancock, and H. Bailleres
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
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
6 Conclusion
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
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.
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 129
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.
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.
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
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).
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
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.
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 135
DOI: 10.1007/978-94-007-7811-5_13, RILEM 2014
136 S. Aicher and C. Stritzke
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
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].
138 S. Aicher and C. Stritzke
Table 1 Cross-sectional size and geometry ranges of Keel-web elements according to [2]
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)
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
and
, , = , , / 35 (7)
35
, , = , , = , , 35 / 70 . (8)
/
140 S. Aicher and C. Stritzke
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)
=1 / < 30 (11a)
= 0.1124 + 772( / ) 30 / 61 (11b)
=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:
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
[ ] = 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
, [0] ,
, , = = = (15)
= +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
Novel Lightweight Timber Composite Element 143
where
= ( /(2 )) (18)
= = 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.
[ ] + (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
Novel Lightweight Timber Composite Element 145
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
146 S. Aicher and C. Stritzke
, ( ) , ( ) , ( )
(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
15000
-15000
vertical load = 2000 N
15000 -30000
7500
2nd order theory
0
3rd order theory
-7500
vertical load = 1000 N
-15000 15000
moment [Nmm]
-7500
vertical load = 0 N
-15000
0 hw/2 hw
Fig. 9 Moment distribution along web height while support force increases
Novel Lightweight Timber Composite Element 147
KSE380
1.0
0.75
0.4
0.25
1.0
0
0
-1000 0 1000 2000 3000 4000
zero vertical load [N]
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
148 S. Aicher and C. Stritzke
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 149
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.
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.
152 F. Miebach and D. Niewerth
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
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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 157
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
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
considered to yield too conservative results. Equation (3) resulted from a regression
analysis of embedment test results on hardwoods:
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
Timber Joints with Very High Strength Steel Dowels 161
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
162 J.-W. van de Kuilen, C. Sandhaas, and H.J. Bla
Table 3 Joint test results and ratio of Fmax with vhss dowels over Fmax with hss dowels
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
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]
3
200 200 200 5
[kN/mm]
[kN/mm]
[kN/mm]
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.
164 J.-W. van de Kuilen, C. Sandhaas, and H.J. Bla
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 .
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)
Timber Joints with Very High Strength Steel Dowels 165
1 Motivation
Keywords: nailed connections, rivet connections, brittle, ductile and mixed failure
modes, elastic spring model, ultimate capacity, design 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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 167
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.
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).
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
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.
170 P. Zarnani and P. Quenneville
Grain
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
Wood Load-Carrying Capacity of Timber Connections 171
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);
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).
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.
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.
Fig. 8 Specimen and failed block exhibiting the brittle/mixed modes of wood failure
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.
Wood Load-Carrying Capacity of Timber Connections 175
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.
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
RECX1-G 198 634 N/A N/A 198 200 (12%) 0.99 BRT BRT
RECX2-G 196 634 N/A N/A 196 256 (15%) 0.77 BRT BRT
RECX4-G 196 634 N/A N/A 196 255 (9%) 0.77 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.
Wood Load-Carrying Capacity of Timber Connections 1777
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)
Wood Load-Carrying Capacity of Timber Connections 179
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 181
DOI: 10.1007/978-94-007-7811-5_17, c RILEM 2014
182 F. Bruhl, J. Schanzlin, and U. Kuhlmann
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.
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]
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)
Ductility in Timber Structures: Investigations on Over-Strength Factors 185
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)
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)
with:
Ductility in Timber Structures: Investigations on Over-Strength Factors 187
581
180
320
2x4 mean value 441.7 33.8
test
12
3x3 mean value 449.6 33.9
1
5x2 mean value 440.3 33.4
COV 0.04 [21] 0.001 0.1 [22] 0.15 [22] 0.0025 0.0015 0.1 [21]
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:
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.
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)
190 F. Bruhl, J. Schanzlin, and U. Kuhlmann
[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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 191
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.
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.
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.
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.
194 A. Ohtsuka et al.
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]
6
6
6
3 Specification of a Specimen
200 24
100 200 24
224 100
112 112
42 Steel Hardware
24
0
0
08 3
06
1 08
Plywood
=17mm,19mm
26mm
P-Max
Pc3
Rigidity 2
Pc2
Pc1
Displacement
Rigidity 1
backlash
4.1 Load--Relations
The outline of the load displacement relation to the experimental result of this
paper is shown in Fig. 6.
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]
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]
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.
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
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.
200 A. Ohtsuka et al.
7 Conclusion
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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 201
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
2 Experimental Investigations
copper wires
Fax
F [N]
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
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
Investigations Concerning the Force Distribution 205
% Fmax
% Fmax
50% 50% 50%
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
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
206 A. Ringhofer and G. Schickhofer
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 .
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
Investigations Concerning the Force Distribution 207
= 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
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
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
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)
210 A. Ringhofer and G. Schickhofer
[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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 211
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.
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%.
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].
214 P. Kobel, R. Steiger, and A. Frangi
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.
4 Embedment Tests
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.
216 P. Kobel, R. Steiger, and A. Frangi
120 100
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]
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).
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
218 P. Kobel, R. Steiger, and A. Frangi
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.
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)
220 P. Kobel, R. Steiger, and A. Frangi
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 221
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.
1
LVDT 2
3
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.
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]
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.
f h,0
f h, = (2)
k90 sin + cos 2
2
Where:
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
25
20
15
10
5 = 25 = 55
= 90 = 0
0
0 1 2 3 4 5 6 7 8
Displacement u [mm]
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.
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
The Embedment Failure of European Beech 227
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]
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)
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.
The Embedment Failure of European Beech 229
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
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.
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 231
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)
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
dw/dx dw/dx
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.
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.
Cm Rm 1,m Rm , m +1
236 D. Brandon et al.
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):
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.
238 D. Brandon et al.
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)
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)
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
Modelling of Non-metallic Timber Connections at Elevated Temperatures 239
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)
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.
240 D. Brandon et al.
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)
Modelling of Non-metallic Timber Connections at Elevated Temperatures 241
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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 243
DOI: 10.1007/978-94-007-7811-5_23, RILEM 2014
244 B. Franke and P. Quenneville
b
lr loaded perpendicular to grain
ar
h
m
he
ac
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
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
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
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
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
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
Analysis of the Brittle Failure and Design of Connections Loaded Perpendicular 249
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
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).
250 B. Franke and P. Quenneville
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 )
Analysis of the Brittle Failure and Design of Connections Loaded Perpendicular 251
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
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)
Analysis of the Brittle Failure and Design of Connections Loaded Perpendicular 253
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
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
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
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 255
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)
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.
258 D. Brandon and A. Leijten
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
Structural Performance and Advantages of DVW Reinforced Moment 259
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.
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
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
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.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.
Structural Performance and Advantages of DVW Reinforced Moment 261
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)
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.
262 D. Brandon and A. Leijten
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
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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 265
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.
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.
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).
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.
3 Analytical Approach
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
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
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.
270 R. Jockwer, R. Steiger, and A. Frangi
40 40
= 90 = 90
35 = 60 35 = 60
= 45 = 45
30 30
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.
Table 1 Test results compared to the proposed design model and to values from literature
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
272 R. Jockwer, R. Steiger, and A. Frangi
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
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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 273
DOI: 10.1007/978-94-007-7811-5_26, RILEM 2014
274 D. Wrzesniak, M. Fragiacomo, and A. Jorissen
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)
2 Experimental Tests
T
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
a3 90 90 Varies 90 90 a3
a4 Varies
Varies
Varies
Varies
3 Test Results
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
Alternative Approach to Avoid Brittle Failure in Dowelled Connections 279
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.
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).
282 D. Wrzesniak, M. Fragiacomo, and A. Jorissen
Fig. 8.a
Fig. 8.b
Fig. 8.c
Alternative Approach to Avoid Brittle Failure in Dowelled Connections 283
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.
= .
(66)
Fig. 10 Effective number of fasteners neff versus end distance a3 for specimens with tw
wo
rows of 3 dowels
Alternative Approach to Avoid Brittle Failure in Dowelled Connections 285
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.
286 D. Wrzesniak, M. Fragiacomo, and A. Jorissen
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:
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.
Alternative Approach to Avoid Brittle Failure in Dowelled Connections 287
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
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
1 Introduction
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 289
DOI: 10.1007/978-94-007-7811-5_27, RILEM 2014
290 U. Mahlknecht et al.
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.
Fig. 1 Test setup: upper loading plate with hole matrix and the fixed supporting steel plate
with a rectangular opening
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.
failure # W / # S
specimen length
12,mean [kg/m]
lef / d (temb / d)
source, height,
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
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
Resistance and Failure Modes of Axially Loaded Groups of Screws 295
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
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.
296 U. Mahlknecht et al.
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
Resistance and Failure Modes of Axially Loaded Groups of Screws 297
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
298 U. Mahlknecht et al.
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)
Resistance and Failure Modes of Axially Loaded Groups of Screws 299
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
Common rules and rules for buildings
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)
300 U. Mahlknecht et al.
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
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.
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].
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 301
DOI: 10.1007/978-94-007-7811-5_28, RILEM 2014
302 M. Steilner and H.J. Bla
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
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
3 Experimental Results
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.
304 M. Steilner and H.J. Bla
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.
306 M. Steilner and H.J. Bla
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
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.
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 307
DOI: 10.1007/978-94-007-7811-5_29, RILEM 2014
308 M. Jamil, P. Quenneville, and C. Clifton
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
Un-bonded post-
tensioned tendon
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
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
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].
314 M. Jamil, P. Quenneville, and C. Clifton
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
Top brackets,
multi-linear
plastic
Supports,
friction isolator
Fig. 13 Drift at level 3 (top level) for the seven selected ground motions; left: Ultimate
Limit State; and right: Maximum Credible Earthquake
Low-Damage Design Using a Gravity Rocking Moment Connection 317
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.
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.
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
320 M. Jamil, P. Quenneville, and C. Clifton
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
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)
Low-Damage Design Using a Gravity Rocking Moment Connection 321
[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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 325
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.
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.
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)
2.8 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
Table 2 Tensile strength and different fracture path of finger joints, SD in parenthesis
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.
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.
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.
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)
Finger Jointing of Freshly Sawn Norway Spruce Side Boards 335
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.
336 M. Sterley et al.
strength, but instead, strength is only affected by the (high) moisture content at
testing.
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 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.
Finger Jointing of Freshly Sawn Norway Spruce Side Boards 337
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)
338 M. Sterley et al.
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)
Finger Jointing of Freshly Sawn Norway Spruce Side Boards 339
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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 341
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.
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
344 G. Stapf and S. Aicher
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.
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
Pressure Distribution in Block Glue Lines Analyzed by Theory of Beams 345
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:
346 G. Stapf and S. Aicher
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.
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).
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)
348 G. Stapf and S. Aicher
where
wblock
=
2
EI = Eclamps Iclamps /2 (compare with Eq. 1)
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).
Pressure Distribution in Block Glue Lines Analyzed by Theory of Beams 349
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.
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 :
350 G. Stapf and S. Aicher
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
Pressure Distribution in Block Glue Lines Analyzed by Theory of Beams 351
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.
!
ness
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)
Pressure Distribution in Block Glue Lines Analyzed by Theory of Beams 353
[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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 355
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.
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.
O
C
N N
C
O
N
C
O
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
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
362 K. Grstad and R. Bredesen
40
35
Strength(N/mm)
30
25
20
15
10
5
0
0 50 100 150 200
Time (minutes)
EPI MUF MF
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.
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.
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.
364 K. Grstad and R. Bredesen
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
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.
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).
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 365
DOI: 10.1007/978-94-007-7811-5_33, RILEM 2014
366 Y. Jiang et al.
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))
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).
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.
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
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. 4 Influences of surface preparation (left) and aging (right) on the wettability of wood
surfaces
372 Y. Jiang et aal.
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.
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
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
Keywords: timber, wood, steel, glued connections, fatigue test, ductile, two
component adhesive system.
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 375
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. 2 shows the load displacement curve of another test specimen with its
plastic capacity due to the steel failure.
steel failure
Plastic performance
yield
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
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.
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.
378 L. Bathon et al.
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 = 1.200 N
Kser = 7.000 N/mm
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 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
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 381
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.
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.
384 C. Winkler and U. Schwarz
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
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.
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
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).
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
Multifunctional Wood-Adhesives for Structural Health Monitoring Purposes 389
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
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
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)
Multifunctional Wood-Adhesives for Structural Health Monitoring Purposes 393
[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)
394 C. Winkler and U. Schwarz
[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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 395
DOI: 10.1007/978-94-007-7811-5_36, RILEM 2014
396 B. Franke, F. Scharmacher, and A. Mller
d - cut off
Lamella
Glueline
mm
R l = 80
h
mm T
R l = 80 L
T
R L
L l
T b
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.
3 Experimental Results
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
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
Total percentage of delamination
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. 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)
400 B. Franke, F. Scharmacher, and A. Mller
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.
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
Assessment of the Glue-Line Quality in Glued Laminated Timber Structures 401
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
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)
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
402 B. Franke, F. Scharmacher, and A. Mller
18% 18%
Shear test EN 392
16% Shear test EN 392
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).
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
Assessment of the Glue-Line Quality in Glued Laminated Timber Structures 403
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
European Committee for Standardization (CEN), EN 391:2001, Glued laminated timber,
Delamination test of glue lines, Brussels
European Committee for Standardization (CEN), EN 392:1995, Glued laminated timber,
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
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),
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 405
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
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).
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
Review of Recent Research Activities on One-Component PUR-Adhesives 409
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
410 C. Lehringer and J. Gabriel
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.
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.
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
414 C. Lehringer and J. Gabriel
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.
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.
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.
Review of Recent Research Activities on One-Component PUR-Adhesives 417
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
[11] EN 15416-3:2006. Adhesives for load bearing timber structures other than phenolic
and aminoplastic Test methods - Part 3: Creep deformation test at cyclic climate
conditions with specimens loaded in bending shear
[12] EN 15416-4:2006. Adhesives for load bearing timber structures other than phenolic
and aminoplastic Test methods - Part 4: Determination of open assembly time for
one component polyurethane adhesives
[13] EN 15416-5:2006. Adhesives for load bearing timber structures other than phenolic
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)
418 C. Lehringer and J. Gabriel
[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)
Review of Recent Research Activities on One-Component PUR-Adhesives 419
[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)
420 C. Lehringer and J. Gabriel
[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)
Keywords: Glulam, reinforced timber, FRP, LVL, rigid bond, polymer concrete.
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 423
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.
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.
426 M. Jahreis et al.
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.
Development of a High-Performance Hybrid System Made of Composites 427
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 Experimental Investigations
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
Fig. 6 Shear test: Specimen with test setting (left); first crack (centre); final crack (right)
Development of a High-Performance Hybrid System Made of Composites 429
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.
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.
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
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.
1 Introduction
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 433
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).
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
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.
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.
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
Experimental Study of the Composite Timber-Concrete SBB Connection 437
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
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.
438 M. Manthey et al.
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.
Experimental Study of the Composite Timber-Concrete SBB Connection 439
Fig. 5 Load-slip curves and Phenomenology for the SBB 26-170 Push-Out tests and the
SBB 26-250
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
442 M. Manthey et al.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 443
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
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).
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.
The Predictive Model for Stiffness of Inclined Screws as Shear Connection 447
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
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).
(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.
448 F. Moshiri, R. Shrestha, and K. Crews
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.
3 Experimental Tests
Fig. 3 Test configuration full-hole test in compliance with ASTM D 5764-97a [10]
The Predictive Model for Stiffness of Inclined Screws as Shear Connection 449
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.
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)
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 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%
40000
30000
20000
10000
0
0 15 30 45 60 75 90
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
452 F. Moshiri, R. Shrestha, and K. Crews
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.
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)
The Predictive Model for Stiffness of Inclined Screws as Shear Connection 453
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
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.
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 455
DOI: 10.1007/978-94-007-7811-5_41, RILEM 2014
456 S.-J. Lee et al.
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.
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)
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).
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.
Shear Performance of Wood-Concrete Composite 461
Wood-to-wood composite
1.630.19 7,527778
(Lee et. al., 2012)
4 Conclusions
This study was performed for evaluation of wood-concrete composite with
different anchorage length of inserted steel rebar. Results were compared with
462 S.-J. Lee et al.
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
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.
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 463
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.
3 Shear Connector
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.
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
Development of Prefabricated Timber-Concrete Composite Floors 467
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%.
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
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
ra
ange 43,1 56,7 37,0 56,0 51,6 56,2
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.
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.
470 P. Kuklk, P. Nechanick, and A. Kuklkov
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,
Bruxelles, Belgium (2010)
[12] Comit European de Normalisation, EN 1611-1: Sawn Timber Appearance grading
of softwoods Part 1: European spruces, firs, pines, Douglas firs and larches,
Bruxelles, Belgium (2000)
[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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 471
DOI: 10.1007/978-94-007-7811-5_43, RILEM 2014
472 A. Fadai, M. Fuchs, and W. Winter
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.
Fig. 1 Multi-layer floor elements made of timber, light wood-concrete and self-compacting
concrete
Fig. 3 Different variants of multi-layer floor elements made of timber, light wood-concrete
and self-compacting concrete)
Wood-Based Construction for Multi-story Buildings 477
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).
32.8
Legend
100.0
CLT
SCC
wood-concrete (Velox)
profile for load introduction
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.
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.
beam A: = ; =
beam B: = ; =
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).
482 A. Fadai, M. Fuchs, and W. Winter
width b = 0.5m
length l = 7.2m
thickness modulus of elasticity modulus of shear
Table 4 Comparison of normal stresses in the cross-section at the middle of the beam
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:
Wood-Based Construction for Multi-story Buildings 483
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)
484 A. Fadai, M. Fuchs, and W. Winter
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
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
structural elements.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 485
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.
Fig. 1 Burgers Model; Creep and Relaxation Fig. 2 Test setting long term
according to Hying [1] behaviour under compression
load
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
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.
Rehabilitation, Upgrading and Repair of Historic Timber Structures 491
Fig. 6 Test with double side mounting in different shapes (left and centre); end anchorage
with supporting by CFRP perpendicular to the beam (right)
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]
492 M. Jahreis and K. Rautenstrauch
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
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
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
1 Introduction
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 493
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.
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
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
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
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.
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 499
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
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).
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.
3 Experimental Investigations
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
Hybrid Wall-Slabs for Multi-storey Buildings 503
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.
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.
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
Hybrid Wall-Slabs for Multi-storey Buildings 505
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.
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 507
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.
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.
(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.
An Innovative Prefabricated Timber-Concrete Composite System 511
Table 3 Ultimate withdrawal capacity of the screws inserted into the concrete with different
penetration lengths
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.
40
20
0
0 20 40 60 80
Deflection at mid-span f (mm)
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
)
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.
An Innovative Prefabricated Timber-Concrete Composite System 513
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.
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.
514 R. Crocetti et al.
Fig. 9 Relative humidity and temperature observed in the laboratory during the period 01-
02-2013 / 30-11-2013
An Innovative Prefabricated Timber-Concrete Composite System 515
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.
516 R. Crocetti et al.
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 519
DOI: 10.1007/978-94-007-7811-5_48, RILEM 2014
520 G. Rinaldin and M. Fragiacomo
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.
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.
(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
522 G. Rinaldin and M. Fragiacomo
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.
(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
A Component Model for Cyclic Behaviour of Wooden Structures 523
(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)
(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.
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)
(a) (b)
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.
(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 Component Model for Cyclic Behaviour of Wooden Structures 527
(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.
(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]
528 G. Rinaldin and M. Fragiacomo
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
A Component Model for Cyclic Behaviour of Wooden Structures 529
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)
530 G. Rinaldin and M. Fragiacomo
[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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 531
DOI: 10.1007/978-94-007-7811-5_49, RILEM 2014
532 M.O. Amini et al.
1 Background
Characteristic Behavior
Develop Model
Analyze Models
Evaluate Performance
Document Results
Fig. 1 Overview of the process for quantifying and documenting seismic performance
factors (after FEMA P695, 2009)
534 M.O. Amini et al.
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 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.
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
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.
Overview of a Project to Quantify Seismic Performance Factors 537
Fig. 6 Loading Paths and Parameters of Modified Stewart Hysteretic Model (after Pang et
al., 2010)
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 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
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
540 M.O. Amini et al.
12 Closure
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)
Overview of a Project to Quantify Seismic Performance Factors 541
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 543
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.
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
546 M. Popovski et al.
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.
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.
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
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
5 Concluding Remarks
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
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
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 555
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.
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.
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
100
Riberholt
80 Buchanan
Gustafsson
60
EC5 1996
40
Experiment
20 Regression
0
0 50 100 150 200 250 300
Anchorage Length la (mm)
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.
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.
Experimental Testing of a Portal Frame Connection Using Glued-In Steel Rods 559
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
560 J. Walker and R. Xiao
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.
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.
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
Experimental Testing of a Portal Frame Connection Using Glued-In Steel Rods 561
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.
+ + 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.
562 J. Walker and R. Xiao
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)
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
564 J. Walker and R. Xiao
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)
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)
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
566 J. Walker and R. Xiao
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 569
DOI: 10.1007/978-94-007-7811-5_52, RILEM 2014
570 R. Widmann et al.
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.
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.
572 R. Widmann et al.
Table 2 Glulam beams (GL1 and GL2) and automatically produced finger-jointed lamellas
(FJ GL1) tested in bending
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].
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.
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
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
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
3
Queensland Collage of Arts, Griffith University, Australia
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 577
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.
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
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
580 I.D. Underhill et al.
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.
Fig. 2 Split hardwood plantation thin- Fig. 3 Furniture designed from hardwood plan-
ning round wood logs (from [2]) tation thinning timber (from [2])
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)
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
4.1 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.
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).
Fig. 6 Sample of profiles manufactured at Griffith University (a) circular hollow section,
(b) rectangular hollow section, (c) Cee-section, and (d) I-section
(a) (b)
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
Structural Veneer Based Composite Products from Hardwood Thinning 585
(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.
586 I.D. Underhill et al.
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.
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
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)
588 I.D. Underhill et al.
[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
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
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
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 589
DOI: 10.1007/978-94-007-7811-5_54, RILEM 2014
590 Y. Xiao et al.
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
(+/- 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].
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.
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.
Glue Laminated Bamboo (GluBam) for Structural Applications 593
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.
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%.
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.
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.
Glue Laminated Bamboo (GluBam) for Structural Applications 597
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)
And the ultimate strength of specimens can be obtained from equation (3),
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.
Fig. 15 Applications
Glue Laminated Bamboo (GluBam) for Structural Applications 601
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 603
DOI: 10.1007/978-94-007-7811-5_55, RILEM 2014
604 G. Dill-Langer and S. Aicher
2 Material
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.
3 Experimental Investigations
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.
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
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.
Glulam Composed of Glued Laminated Veneer Lumber Made of Beech Wood 607
7
Compression load Fc [MN]
0
0 5 10 15 20 25 30 35 40
Deformation [mm]
a) b)
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
610 G. Dill-Langer and S. Aicher
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.
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
Glulam Composed of Glued Laminated Veneer Lumber Made of Beech Wood 611
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
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.
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.
Glulam Composed of Glued Laminated Veneer Lumber Made of Beech Wood 613
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
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.
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 615
DOI: 10.1007/978-94-007-7811-5_56, RILEM 2014
616 J. Marcroft et al.
Table 1 Studies comparing structural properties under service class 1 and service class 3
conditions
Notches
SC1 specimen subsequently cut of SC3 specimen
shear specimens
50 270 270 50
57
32
90mm long steel
bearing plate
55 50
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.
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
626 J. Marcroft et al.
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 -
Determination of some physical and mechanical properties
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)
Structural Performance of Accoya Wood under Service Class 3 Conditions 627
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
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.
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 629
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.
Structural Veneer Based Composite Products from Hardwood Thinning 631
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.
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 2 Average MOE and MOR for the four poles tested in bending
Tested material properties for the two poles tested in predominant shear can be
found in [13].
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
636 B.P. Gilbert et al.
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.
Structural Veneer Based Composite Products from Hardwood Thinning 637
Fig. 5 Simple failure mechanism in bending, (a) pole cross-section, (b) strain distribution
and (c) stress distribution
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
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.
5 Conclusion
This paper presented bending, shear and material tests on VBC hardwood hollow
utility poles manufactured from plantation thinning. A simple failure mechanism
Structural Veneer Based Composite Products from Hardwood Thinning 639
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)
640 B.P. Gilbert et al.
[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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 641
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
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]
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]
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]
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.
646 S. Aicher and G. Stapf
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:
Glulam from European White Oak: Finger Joint Influence on Bending Size Effect 647
Table 4 Finger joint bending strengths of investigated white oak lamination configurations
Table 5 Finger joint tensile and bending strengths normalized to lamination and ratio of
finger joint tensile strength
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
Glulam from European White Oak: Finger Joint Influence on Bending Size Effect 649
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
Fig. 2 Bending strength of white oak glulam beams with different cross sectional sizes
650 S. Aicher and G. Stapf
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.
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
, , , = 0.8 , , , , , , = , , . (3a,b)
For the case of grade LS 10 the applicability range of Eq. (2) is slightly violated as
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
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
.
, ,( / ), , = 35.3 N/mm = 1.13 35.3 = 39.8 N/mm (7)
, , = 63.7 N/mm < 1.4 , ,, = 1.4 52.6 N/mm = 73.64 N/mm . (8)
652 S. Aicher and G. Stapf
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.
6.1 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)
= + . (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)
= 1 (1 ) 1 ( ) 1 ( ) 1 ( ) 1 ( ) (12)
where = = ( 2), = ( ), = 2 .
Glulam from European White Oak: Finger Joint Influence on Bending Size Effect 653
=1 (13)
= 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)
Then Eq. (15) which fully disregards the contribution of the wood itself delivers
for the small and large beams the 5%-quantile bending strengths
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
Glulam from European White Oak: Finger Joint Influence on Bending Size Effect 655
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
Determination of some physical and mechanical properties
[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)
656 S. Aicher and G. Stapf
[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
Common rules and rules for buildings
[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
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
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 657
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.
Fig. 1 The metaxylem vessel cavitys position distribution and density variation
: 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.
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.
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.
Fig. 6 The samples thermal conductivity variation from external surface to internal surface
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.
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 667
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.
2 Analytical Approach
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
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
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)
Tapered Beams Made of Cross Laminated Timber 671
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)
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 .
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
3 Experimental Investigations
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
Tapered Beams Made of Cross Laminated Timber 673
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
,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
F F
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.
Tapered Beams Made of Cross Laminated Timber 675
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
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 677
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)
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.
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
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
Influence of the Connection Modelling on the Seismic Behaviour 681
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].
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.
682 I. Sustersic, B. Dujic, and M. Fragiacomo
300
250
Base shear [kN]
50
0
0,00 0,02 0,04 0,06 0,08 0,10
Top floor displacement [m]
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
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
684 I. Sustersic, B. Dujic, and M. Fragiacomo
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.
Influence of the Connection Modelling on the Seismic Behaviour 685
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.
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.
Influence of the Connection Modelling on the Seismic Behaviour 687
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 689
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.
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
692 I. Gavric et al.
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).
Fig. 2 CLT wall test configurations tested at CNR IVALSA by Gavric (2013) (dimensions
in mm)
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)
tot = r + sl + sh + b (3.1)
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
Behaviour of Cross-Laminated Timber Panels under Cyclic Loads 695
90
80
Rocking [%]
70
60
Coupled Walls
50
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%).
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
Behaviour of Cross-Laminated Timber Panels under Cyclic Loads 697
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.
698 I. Gavric et al.
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.
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
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.
Behaviour of Cross-Laminated Timber Panels under Cyclic Loads 699
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.
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.
702 I. Gavric et al.
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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 703
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
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.
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.
706 G. Hochreiner et al.
Fig. 3 Load deformation curves of 3-layered (left) and 5-layered (rigth) CLT plates
Failure Mechanisms of CLT Plates 707
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
708 G. Hochreiner et al.
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).
Failure Mechanisms of CLT Plates 709
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
710 G. Hochreiner et al.
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.
Failure Mechanisms of CLT Plates 711
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.
712 G. Hochreiner et al.
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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 713
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.
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
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.
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.
Seismic Strengthening of Existing Concrete and Masonry Buildings 717
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
3 Experimental Testing
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
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
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.
Seismic Strengthening of Existing Concrete and Masonry Buildings 721
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)
Table 3 Story drift comparison (for the scaled Landers accelerogram (0.7 g)) of the un-
strengthened and strengthened RC frame with masonry infill
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)
Seismic Strengthening of Existing Concrete and Masonry Buildings 723
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
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.
1 Introduction
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 725
DOI: 10.1007/978-94-007-7811-5_65, RILEM 2014
726 J.M. Branco, M. Kekeliak, and P.B. Loureno
2 Experimental Campaign
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.
728 J.M. Branco, M. Kekeliak, and P.B. Loureno
1
K II = (2)
1 / K ser ,ax ,1 + 1 / K ser ,sax , 2
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
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
732 J.M. Branco, M. Kekeliak, and P.B. Loureno
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 )
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
In-Plane Stiffness of Traditional Timber Floors Strengthened with CLT 733
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.
734 J.M. Branco, M. Kekeliak, and P.B. Loureno
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.
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]
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), Coefficient of friction = 0,5 456 2036 429 626 2690 593 531 1993 499
K (N/mm), Coefficient of friction = 0,8 460 2040 433 629 2693 597 534 1996 502
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].
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)
In-Plane Stiffness of Traditional Timber Floors Strengthened with CLT 737
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 739
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
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)
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.
742 S. Nakashima et al.
Center Plastic
l/2 of Hinge
Rotation
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
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].
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)
744 S. Nakashima et al.
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)
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)
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
3.2 Experiment
=16mm
CLT l=150mm Drift Pin
700 Load cell 100kN
224
Fig. 6 Sampling of specimens and apparatus of tensile test for CLT - drift pin joint
(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
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)
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)
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.
748 S. Nakashima et al.
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 751
DOI: 10.1007/978-94-007-7811-5_67, RILEM 2014
752 R. Brandner and G. Schickhofer
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
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
754 R. Brandner and G. Schickhofer
CN 1
, with C = , , (4)
C1 ln ( N ) + 1 C
C
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
Length Effects on Tensile Strength in Timber Members With and Without Joints 755
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)
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
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
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)
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.
758 R. Brandner and G. Schickhofer
Fig. 2 Length effects on expectation and 5 %-quantiles of tensile strength and power re-
gression models: timber members with joints
4 Conclusions
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)
760 R. Brandner and G. Schickhofer
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
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
*
Corresponding author.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 761
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
local
grain
deviation
slope of grain slope of grain top rupture
with knot without knot
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.
3.2 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.
New Perspectives in Machine Strength Grading 765
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
4 Results
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
768 J.K. Denzler and A. Weidenhiller
series
top
regular
rupture
80% 14%
20% 86%
Fig. 7 Identifying specimens in the test sample including a top rupture, n = 116 specimens
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 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)
New Perspectives in Machine Strength Grading 771
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
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 773
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.
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
3.2 Methods
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).
776 G.J.P. Ravenshorst, P.A. de Vries, and J.-W. van de Kuilen
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.
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
Aspects of the Difference between the Local and Global Modulus of Elasticity 777
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.
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.
778 G.J.P. Ravenshorst, P.A. de Vries, and J.-W. van de Kuilen
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).
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
Aspects of the Difference between the Local and Global Modulus of Elasticity 779
5 Analysis
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
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)
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
782 G.J.P. Ravenshorst, P.A. de Vries, and J.-W. van de Kuilen
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 )
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
Aspects of the Difference between the Local and Global Modulus of Elasticity 783
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
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.
1 Background
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 787
DOI: 10.1007/978-94-007-7811-5_70, RILEM 2014
788 H.R. Milner and C.Y. Adam
The standard will not unduly limit options available to the manufacturer by
attempting to prescribing any of the following.
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
790 H.R. Milner and C.Y. Adam
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.
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
f t, g ,k = 0.8 f m , g ,k (3)
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
A Study of Australian Glulam 793
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.
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)
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)
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.
796 H.R. Milner and C.Y. Adam
EI x = ( bt 3 12 ) Ei + ( bt ) Ei yi2 (7)
i i
where
A Study of Australian Glulam 797
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
798 H.R. Milner and C.Y. Adam
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
A Study of Australian Glulam 799
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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 801
DOI: 10.1007/978-94-007-7811-5_71, RILEM 2014
802 J. Oscarsson, A. Olsson, and B. Enquist
Splitting
1325 or 1521.5
300
120 50 50
Fig. 1 Wet-glued beams before (left) and after (right) splitting, drying and planing
along the boards. The method applied for calculation of such profiles has recently
been presented by Olsson et al. (2013).
2 Sampling of Material
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.
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.
804 J. Oscarsson, A. Olsson, and B. Enquist
Fig. 2 WoodEye scanner (left), tracheid effect (middle), and scanned fibre direction field
around a traversing face knot (right)
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).
806 J. Oscarsson, A. Olsson, and B. Enquist
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
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
Fig. 4 MOE profile of board no. FJ25 and possible sections, defined by MOE dips D1-D6,
to be removed
1
FJ board number Potential IP improvement per selected section [MPa]
1 2 3 4 5 6
1. Figures 1-6 refer to the number of sections simultaneously selected for removal.
808 J. Oscarsson, A. Olsson, and B. Enquist
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.
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)
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
810 J. Oscarsson, A. Olsson, and B. Enquist
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
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)
Improving Strength of Glulam Laminations of Norway Spruce Side Boards 811
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
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.
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
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 813
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
F522 F123
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]
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
90
90
90 45 90 45 90 90 45 90 45 90
5
15
15
15
15
400 400
F F
200 200
NSM SG4 SG5 SG6
90
200 200 200 200
1000 1200 800 800 1200 1000
2400 1200 2400
6000
Top view
NSM.20
70
SG4 SG5 SG6
400
130
SG7 SG8 SG9
Botton view
6000
Top view
NSM.40
400
SG4 SG5 SG6
SG7 SG8 SG9
Botton view
6000
F F
200 200
EBR
90
200 200 200 200
400
Botton view
6000
Table 1 Main results obtained on the mechanical characterization of the adhesive (average
values)
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.
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.
Double Span Continuous Glulam Slabs Strengthened with GFRP 819
(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
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.
822 J.M. Branco, M.P. Jorge, and J. Sena-Cruz
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
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.
S. Aicher et al. (eds.), Materials and Joints in Timber Structures, RILEM Bookseries 9, 823
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.
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.
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.
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.
826 R. Hofmann and U. Kuhlmann
0.12
0.10
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
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.
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
828 R. Hofmann and U. Kuhlmann
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
Fig. 5 Utilization of shear due to the torsional moment at the end support of a straight single
span girder with parallel chords and three 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
Fig. 6 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
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
Torsional Moment in Lateral Torsional Stability Analysis 829
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
Fig. 7 Utilization of shear due to the torsional moment at the end support of a straight single
span girder with parallel chords and three horizontal supports at the upper chord according to
bending-torsion second order theory with sinusoidal curvature of imperfection between the
end supports
timber girders for torsional moment needs not to distinguish between several h to b
proportions.
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)
830 R. Hofmann and U. Kuhlmann
[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
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