You are on page 1of 258

VOL. 111, NO.

2
MARCH-APRIL 2014

ACI
STRUCTURAL

J O U R N A L

A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE

CONTENTS
Board of Direction

ACI Structural Journal

President
Anne M. Ellis

March-April 2014, V. 111, No. 2

Vice Presidents
William E. Rushing Jr.
Sharon L. Wood
Directors
Neal S. Anderson
Khaled Awad
Roger J. Becker
Dean A. Browning
Jeffrey W. Coleman
Robert J. Frosch
James R. Harris
Cecil L. Jones
Cary S. Kopczynski
Steven H. Kosmatka
Kevin A. MacDonald
David M. Suchorski
Past President Board Members
James K. Wight
Kenneth C. Hover
Florian G. Barth
Executive Vice President
Ron Burg

Technical Activities Committee


Ronald Janowiak, Chair
Daniel W. Falconer, Staff Liaison
JoAnn P. Browning
Chiara F. Ferraris
Catherine E. French
Fred R. Goodwin
Trey Hamilton
Kevin A. MacDonald
Antonio Nanni
Jan Olek
Michael M. Sprinkel
Pericles C. Stivaros
Andrew W. Taylor
Eldon G. Tipping

Staff

Executive Vice President


Ron Burg
Engineering
Managing Director
Daniel W. Falconer
Managing Editor
Khaled Nahlawi

a journal of the american concrete institute


an international technical society

235 
Web Crushing Capacity of High-Strength Concrete Structural
Walls: Experimental Study, by Rigoberto Burgueo, Xuejian Liu, and
EricM.Hines
247 Response of Precast Prestressed Concrete Circular Tanks Retaining
Heated Liquids, by Michael J. Minehane and Brian D. ORourke
257 Bond Strength of Spliced Fiber-Reinforced Polymer Reinforcement,
by Ali Cihan Pay, Erdem Canbay, and Robert J. Frosch
267 Flexural Behavior and Strength of Reinforced Concrete Beams with
Multiple Transverse Openings, by Bengi Aykac, Sabahattin Aykac, Ilker
Kalkan, Berk Dundar, and Husnu Can
279 
Experimental Assessment of Inadequately Detailed Reinforced
Concrete Wall Components, by Adane Gebreyohaness, Charles Clifton,
John Butterworth, and Jason Ingham
291 Behavior of Epoxy-Injected Diagonally Cracked Full-Scale Reinforced
Concrete Girders, by Matthew T. Smith, Daniel A. Howell, MaryAnn T.
Triska, and Christopher Higgins
303 High-Performance Fiber-Reinforced Concrete Bridge Columns under
Bidirectional Cyclic Loading, by Ady Aviram, Bozidar Stojadinovic, and
Gustavo J. Parra-Montesinos
313 Analysis of Early-Age Thermal and Shrinkage Stresses in Reinforced
Concrete Walls, by Barbara Klemczak and Agnieszka Knoppik-Wrbel
323 Effects of Casting Position and Bar Shape on Bond of Plain Bars, by
Montserrat Sekulovic MacLean and Lisa R. Feldman
331 
Performance of Glass Fiber-Reinforced Polymer-Doweled Jointed
Plain Concrete Pavement under Static and Cyclic Loadings, by Brahim
Benmokrane, Ehab A. Ahmed, Mathieu Montaigu, and Denis Thebeau
343 Nonlinear Static Analysis of Flat Slab Floors with Grid Model, by
Dario Coronelli and Guglielmo Corti
353 Effect of Steel Stirrups on Shear Resistance Gain Due to Externally
Bonded Fiber-Reinforced Polymer Strips andSheets, by Amir Mofidi
and Omar Chaallal
363 Punching of Reinforced Concrete Flat Slabs with Double-Headed
Shear Reinforcement, by Maurcio P. Ferreira, Guilherme S. Melo,
PaulE. Regan, and Robert L. Vollum
375 Behavior of Concentrically Loaded Fiber-Reinforced Polymer Reinforced Concrete Columns with Varying Reinforcement Types and
Ratios, by Hany Tobbi, Ahmed Sabry Farghaly, and Brahim Benmokrane

Staff Engineers
Matthew R. Senecal
Gregory M. Zeisler
Jerzy Z. Zemajtis
Publishing Services
Manager
Barry M. Bergin
Editors
Carl R. Bischof
Kaitlyn Hinman
Ashley Poirier
Kelli R. Slayden
Editorial Assistant
Tiesha Elam

Contents cont. on next page

Discussion is welcomed for all materials published in this issue and will appear ten months from
this journals date if the discussion is received within four months of the papers print publication.
Discussion of material received after specified dates will be considered individually for publication or
private response. ACI Standards published in ACI Journals for public comment have discussion due
dates printed with the Standard.
Annual index published online at http://concrete.org/Publications/ACIStructuralJournal.
ACI Structural Journal
Copyright 2014 American Concrete Institute. Printed in the United States of America.
The ACI Structural Journal (ISSN 0889-3241) is published bimonthly by the American Concrete Institute. Publication office: 38800 Country Club Drive, Farmington Hills, MI 48331. Periodicals postage paid at Farmington, MI, and
at additional mailing offices. Subscription rates: $161 per year (U.S. and possessions), $170 (elsewhere), payable in
advance. POSTMASTER: Send address changes to: ACI Structural Journal, 38800 Country Club Drive, Farmington
Hills, MI 48331.
Canadian GST: R 1226213149.
Direct correspondence to 38800 Country Club Drive, Farmington Hills, MI 48331. Telephone: 248.848.3700.
Website: http://www.concrete.org.

ACI Structural Journal/March-April 2014

233

387 Repair of Prestressed Concrete Beams with Damaged Steel Tendons


Using Post-Tensioned Carbon Fiber-Reinforced Polymer Rods, by
Clayton A. Burningham, Chris P. Pantelides, and Lawrence D. Reaveley
397 
Study of Composite Behavior of Reinforcement and Concrete in
Tension, by John P. Forth and Andrew W. Beeby
407 Flexural Capacity of Fiber-Reinforced Polymer Strengthened Unbonded
Post-Tensioned Members, by Fatima El Meski and Mohamed Harajli
419 Size Effect on Strand Bond and Concrete Strains at Prestress Transfer,
by Jos R. Mart-Vargas, Libardo A. Caro, and Pedro Serna-Ros
431 
Proposed Minimum Steel Provisions for Prestressed and Nonprestressed Reinforced Sections, by Natassia R. Brenkus and H. R. Hamilton
441 Lateral Strain Model for Concrete under Compression, by Ali Khajeh
Samani and Mario M. Attard
453 Discussion
Cyclic Loading Test for Beam-Column Connection with Prefabricated Reinforcing

Bar Details. Paper by Tae-Sung Eom, Jin-Aha Song, Hong-Gun Park, Hyoung-Seop
Kim, and Chang-Nam Lee
Shear Strength of Reinforced Concrete Walls for Seismic Design of Low-Rise
Housing. Paper by Julian Carrillo and Sergio M. Alcocer
Performance of AASHTO-Type Bridge Model Prestressed with Carbon FiberReinforced Polymer Reinforcement. Paper by Nabil Grace, Kenichi Ushijima,
Vasant Matsagar, and Chenglin Wu

460

In ACI Materials Journal

462

Reviewers in 2013

MEETINGS
2014
MARCH
19-21ICRI 2014 Spring Convention,
Reno, NV, www.icri.org/Events/2014_
Spring/conv_home.asp
22ASA Spring 2014 Committee
Meetings, Reno, NV, www.shotcrete.org

27The Changing Future of Cement


& Concrete: Threat or Opportunity,
Leicestershire, United Kingdom, http://ict.
concrete.org.uk

MARCH/APRIL
30-2ACPA 2014 Convention,
Indianapolis, IN, http://convention.myacpa.
org/indy2014/

24-29ICPI Annual Meeting, New


Orleans, LA, www.icpi.org/node/3996

UPCOMING ACI CONVENTIONS


The following is a list of scheduled ACI conventions:
2014March 23-27, Grand Sierra Resort, Reno, NV
2014October 26-30, Hilton Washington, Washington, DC
2015April 12-15, Marriott & Kansas City Convention Center, Kansas City, MO
For additional information, contact:
Event Services, ACI
38800 Country Club Drive, Farmington Hills, MI 48331
Telephone: 248.848.3795
e-mail: conventions@concrete.org
ON COVER: 111-S23, p. 269, Fig. 3Diagonal reinforcement spiraling around circular openings.

Permission is granted by the American Concrete Institute for libraries and other users registered with the Copyright
Clearance Center (CCC) to photocopy any article contained herein for a fee of $3.00 per copy of the article. Payments
should be sent directly to the Copyright Clearance Center, 21 Congress Street, Salem, MA 01970. ISSN 0889-3241/98
$3.00. Copying done for other than personal or internal reference use without the express written permission of the
American Concrete Institute is prohibited. Requests for special permission or bulk copying should be addressed to the
Managing Editor, ACI Structural Journal, American Concrete Institute.
The Institute is not responsible for statements or opinions expressed in its publications. Institute publications are not able
to, nor intend to, supplant individual training, responsibility, or judgment of the user, or the supplier, of the information
presented.
Papers appearing in the ACI Structural Journal are reviewed according to the Institutes Publication Policy by individuals
expert in the subject area of the papers.

234

Contributions to
ACI Structural Journal
The ACI Structural Journal is an open
forum on concrete technology and papers
related to this field are always welcome.
All material submitted for possible publication must meet the requirements of
the American Concrete Institute Publication Policy and Author Guidelines
and Submission Procedures. Prospective
authors should request a copy of the Policy
and Guidelines from ACI or visit ACIs
website at www.concrete.org prior to
submitting contributions.
Papers reporting research must include
a statement indicating the significance of
the research.
The Institute reserves the right to return,
without review, contributions not meeting
the requirements of the Publication Policy.
All materials conforming to the Policy
requirements will be reviewed for editorial
quality and technical content, and every
effort will be made to put all acceptable
papers into the information channel.
However, potentially good papers may be
returned to authors when it is not possible
to publish them in a reasonable time.
Discussion
All technical material appearing in the
ACI Structural Journal may be discussed.
If the deadline indicated on the contents
page is observed, discussion can appear
in the designated issue. Discussion should
be complete and ready for publication,
including finished, reproducible illustrations. Discussion must be confined to the
scope of the paper and meet the ACI Publication Policy.
Follow the style of the current issue.
Be brief1800 words of double spaced,
typewritten copy, including illustrations
and tables, is maximum. Count illustrations
and tables as 300 words each and submit
them on individual sheets. As an approximation, 1 page of text is about 300 words.
Submit one original typescript on 8-1/2 x
11 plain white paper, use 1 in. margins,
and include two good quality copies of the
entire discussion. References should be
complete. Do not repeat references cited
in original paper; cite them by original
number. Closures responding to a single
discussion should not exceed 1800-word
equivalents in length, and to multiple
discussions, approximately one half of
the combined lengths of all discussions.
Closures are published together with
the discussions.
Discuss the paper, not some new or
outside work on the same subject. Use
references wherever possible instead of
repeating available information.
Discussion offered for publication should
offer some benefit to the general reader.
Discussion which does not meet this
requirement will be returned or referred to
the author for private reply.
Send manuscripts to:
http://mc.manuscriptcentral.com/aci
Send discussions to:
Journals.Manuscripts@concrete.org

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S04


Note: Paper 111-S04 of the January-February 2014 ACI Structural Journal has been reprinted herein with corrected figure art.
Please reference the March-April 2014 version of this paper only.

Web Crushing Capacity of High-Strength Concrete


Structural Walls: Experimental Study
by Rigoberto Burgueo, Xuejian Liu, and Eric M. Hines
This paper discusses the relationship between concrete strength
and web crushing capacity based on results from large-scale tests
of thin-webbed structural walls with confined boundary elements.
Eight walls with concrete strengths ranging from 39 to 138 MPa
(5.6 to 20 ksi) were tested to web crushing failure under cyclic
and monotonic loading. These tests clearly demonstrated differences between elastic and inelastic web crushing behavior and
their dependence on concrete strength. Walls with higher concrete
strengths reached higher levels of displacement ductility due to an
increase in web crushing capacity. Evidence with respect to monotonic tests showed that degradation of the diagonal compression
struts from cyclic loading increases with concrete strength, thus
limiting the inelastic deformation capacity gains. Thus, concrete
compressive strength does not linearly increase web crushing
strength as implied by rational web crushing models; rather, the
relationship is nonlinear, with a decreasing limit as concrete
strength increases. The ACI shear stress limit considerably underestimated the web crushing capacity of the walls. Test results and
observations are reported with the intent of providing physical
insight into the web crushing failure mechanism and the inherent
limits of thin-webbed concrete members in shear.

possibility because they have observed deformed configurations in the inelastic range that could undermine the benefits
of increased concrete strength. The work described herein
represents an attempt to answer this question experimentally
and thereby establish limits for the future analytical prediction
of web crushing failures. This experimental program proceeded
with the intention of testing the following two hypotheses:
1. Web crushing strength increases in proportion to fc as long
as the struts are not damaged. Hence, transformation from an
elastic web crushing failure to an inelastic web crushing failure
can be achieved simply by increasing the concrete strength; and
2. Damage to struts caused by cyclic loading and inelastic
deformations can limit web crushing strength independently
of fc. Hence, increases in fc may not lead to proportional
increases in ductility capacity.
The experimental program designed to test these two
hypotheses consisted of two sets of four structural walls with
a range of concrete strengths. One set of walls was tested
cyclically, and the other set was tested monotonically.

Keywords: ductility; high strength; shear walls; web crushing.

RESEARCH SIGNIFICANCE
Elastic and inelastic web crushing failures were consistently achieved in a series of large-scale structural wall
tests designed to study the relationship between concrete
compressive strength and web crushing strength of thinwebbed members. The results of these tests validate both the
dependence of web crushing capacity on fc and the significant degradation of web crushing capacity experienced for
a range of concrete strengths under cyclic loading in the
inelastic range. Physical insight developed from observations and measurements of these tests provides a firm foundation for establishing the limits of thin-webbed reinforced
concrete member design. Consistency of the test results indicates that it may be possible to design thin-webbed elements
to experience significant inelastic deformations before
failing in shear, opening up new possibilities for acceptable
ductile failure modes of reinforced concrete members.

INTRODUCTION
Over the past 50 years, web crushing capacity of reinforced
concrete members has emerged as a primary design concern in
three distinct contexts: gravity loading of thin-webbed beams in
the 1960s,1-4 seismic loading of structural walls with confined
boundary elements in the 1970s,5-9 and seismic loading of
hollow bridge piers with confined corner elements in the
1990s.10-13 In each context, motivation to design lightweight
members based on physical insight led to large-scale structural testing programs that discovered web crushing capacities significantly in excess of the average shear stress limits
recommended by ACI-318.14 Researchers in charge of these
testing programs have repeatedly emphasized the importance
of understanding shear behavior in terms of diagonal tension
and diagonal compression instead of average shear stresses.
While diagonal compression demands depend on member
geometry and reinforcement, diagonal compression capacities
depend on the size and strength of the most heavily loaded
struts. Previous research programs established consensus
regarding the linear dependence of web crushing capacity on
concrete compressive strength fc and web thickness. Limits
on the value of fc itself, however, were not evaluated. Could
the web crushing capacity of a 30 MPa wall be increased by
a factor of four simply by increasing the concrete strength to
120 MPa? Seismic researchers have hesitated to endorse this
ACI Structural Journal/March-April 2014

WEB CRUSHING SHEAR CAPACITY


Work in the 1960s on thin-webbed girders established
that properly reinforced concrete webs could resist diagonal compression loads in the elastic range almost equal
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. 2011-322.R2, doi:10.14359.51686515, was received May 23, 2013, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

235

Fig. 1Elastic and inelastic shear web crushing: (a) definition of elastic and inelastic web crushing failure modes and critical
regions in wall; (b) free body diagrams used for assessing flexure-shear web crushing capacity of structural wall with confined
boundary elements13; and (c) schematic of strut degradation due to cyclic loading.
to the compressive strength of concrete itself, resulting in
average shear stresses an order of magnitude higher than
the prevailing provisions.15 It was during this work that the
term web crushing was conceived, and a case was made
to consider diagonal compression capacity independently of
average shear stresses. Work in the 1970s, 1980s, and early
1990s on structural walls with confined boundary elements
under seismic loads emphasized the dependence of web
crushing capacity on inelastic deformation and the realignment of cracks in the web crushing region, but recommended
assessment of web crushing capacity in terms of average
shear stresses.9,16 Among these works, the effect of concrete
strength on web crushing capacity was witnessed through
the tests by the Portland Cement Association (PCA)7,9 in the
mid-1970s on walls with boundary elements. Wall B6 with
a concrete compressive strength of 22 MPa (3165 psi) failed
in web crushing at significantly lower deformation capacity
than Wall B7 with a concrete compressive strength of
49MPa (7155 psi). No further high-strength concrete (HSC)
structural walls were tested, however. Work in the late 1990s
and early 2000s on hollow bridge piers with confined corner
elements clearly distinguished between elastic web crushing
and inelastic web crushing, and tied the assessment of
inelastic web crushing to the development of the plastic hinge
region,12 emphasizing the interaction of inelastic flexure and
shear behavior in this zone.13 The physically based method
of assessment by Hines and Seible13 recognized the focus
on principal stresses from the 1960s research, accounted for
inelastic deformations, and allowed accurate assessment of
cross sections with various relations between depth of web
and depth of boundary elements. A review of the model is
provided as follows; however, for a detailed description of
the model, the reader should refer to Reference 13.
The approach to web crushing capacity by Hines and
Seible13 is based on the assessment of capacity and demand

236

on individual struts inside the plastic hinge region as it


spreads up the height of a wall. As shown in Fig. 1(a), two
distinct shear transfer mechanisms were identified for the
plastic hinge region and the regions elsewhere. Elastic, or
standard shear, struts are formed in the wall web in regions
that have not experienced significant tensile strains along
both the longitudinal (vertical) and transverse directions,
leading to a parallel shear cracking pattern (at an angle s).
In other words, this region is stressed mainly under in-plane
shear stress while the effect of elastic flexural strain is not
significant. By comparison, plastic flexural strains force the
struts inside the plastic hinge region to realign so that they
all converge in the flexural compression toe. This can be
understood by considering that the flexural crack at the base
of the wall prohibits shear force transfer into the footing at
any location except for the flexural compression toe, and that
the struts should fan upward until they are able to carry the
full elastic shear force. These fanning struts are considered
as inelastic or flexure-shear struts.
Based on the noted force transfer mechanisms, Hines and
Seible13 proposed a web crushing capacity model through
equilibrium analysis of the free body diagrams of isolated
elastic (or standard) and inelastic (or flexure-shear) diagonal
struts (Fig. 1(b)). Calculating the demand of forces on the
elastic struts NDs and the compressive capacity NCs of these
struts leads to the standard shear web crushing equation
proposed by Oesterle et al.7 and Paulay and Priestley16

NCs
kfcsin s cos s
N Ds

(1)

The inelastic struts in the plastic hinge region differ from


their elastic counterparts both in demand and capacity. On
the demand side, the inelastic struts are required to transfer
inelastic shears at angles that are consistent with the spread
ACI Structural Journal/March-April 2014

of plasticity Lpr. On the capacity side, these struts become


narrower toward the compression toe from which they fan.
Following Hines and Seible,13 the critical flexure shear
crack angle fs is calculated as

A f

jd v yv + f1t w
s

fs = tan 1
2 T Tyav

jd

1
= tan L
pr

(2)

Web crushing is then estimated to occur when the capacity


NCfs of the critical strut (Eq. (3)) is equal to the demand NDfs
on this strut (Eq. (4))

NCfs = kfct w Rd

N Dfs =

(3)

T
f1 st w sin fs
cos fs

(4)

The related variables are determined from a moment-curvature analysis of the cross section and the strut geometry,
as shown in Fig. 1(b). The concrete compressive strength
softening factor k is calculated according to the modified
compression field theory17 with a simplified approach for
determining the principal tensile strain 1. It should be noted
that while the model has no explicit limit on fc, its prediction quality for HSC is uncertain because the available test
data for calibration of the model was from tests of normalstrength concrete (NSC) walls.
ACI 318-1114 does not reflect the direct dependence of web
crushing capacity on fc, but rather defines the expression in
Eq. (1) as 0.83fc (MPa) (10fc [psi]), although fc is a
quantity commonly related to concrete tensile strength and
diagonal tension failures. A maximum value of 0.69MPa
(100 psi) is adopted by the code because of the lack of test
data and practical experience with concrete strengths more
than 69 MPa (10,000 psi).
The proportional increase of web crushing capacity with
fc implied by Eq. (3) indicates the potential of achieving
a ductile force-displacement response in structural walls
even if ultimately limited by web crushing. Figure 2 shows
simulated force-displacement responses for structural walls
with heavily reinforced boundary elements12 along with
web crushing predictions. Curves for walls with concrete
compressive strengths varying from 34 to 138 MPa (5 to
20ksi) are shown. The web-crushing capacity envelopes
after Hines and Seible13 and the ACI Code14 limits are also
plotted. It can be seen that force-deformation response of
the walls is only slightly affected by the increased concrete
compressive strength. The Hines and Seible model, however,
predict the web crushing capacity to increase dramatically
with an increase of concrete strength. The ACI limit is
clearly conservative and independent of the inelastic deformations in the wall. The Hines and Seible model implies that
a 34 MPa (5 ksi) wall would fail by web crushing at the onset
ACI Structural Journal/March-April 2014

Fig. 2Analytical force-displacement response with web


crushing capacity predictions.
of yielding, while a 138 MPa (20 ksi) wall would fail in
flexure. The model, however, assumes integrity of the struts,
which are known to degrade with increase ductility demand
and cyclic loading as schematically shown by the cracking
pattern in Fig. 1(c). Nonetheless, the suggestion that web
crushing strength can be directly and consistently related to
concrete compressive strength is indicative of new possibilities for increasing shear capacities of thinned-webbed structural members with increased concrete strength.
EXPERIMENTAL PROGRAM
Test unit identification, geometry, and
reinforcement details
To verify the aforementioned hypotheses and establish
rational performance levels on the inelastic web crushing
limits for HSC structural walls, eight 1/5-scale cantilever
structural walls with design concrete compressive strengths
of 34, 69, 103, and 138 MPa (5, 10, 15, and 20 ksi) were
tested under cyclic and monotonic loading.18 The walls
had a barbell-type cross section with thin webs and heavily
confined boundary elements, and were designed to induce a
web crushing failure and not to represent a component from
a prototype structure. The test unit cross sections with reinforcement details are shown in Fig. 3. The identification
name for the walls starts with M, followed by two digits
denoting the design concrete compressive strength in kip/in.2
(1 kip/in.2 [ksi] = 6.895 MPa) and then by a letter describing
the loading protocol: C for cyclic and M for monotonic
loading. For example, test unit M10C refers to the wall with a
design concrete strength of 69 MPa (10 ksi) and subjected to
cyclic loading. All walls had an effective length of 2540mm
(100in.) for an aspect ratio (M/V) of 2.5. In all cases, the
wall web was 508 mm (20in.) deep and 76 mm (3in.) thick.
The boundary elements had a depth of 254 mm (10 in.). As
shown in Fig. 3 and Table 1, the steel reinforcement was
essentially the same for all walls, with a small variation in
the longitudinal reinforcement of the boundary elements in
WallM15C and in the transverse reinforcement spacing for
Walls M20C and M20M. The reinforcement ratios for the
web and boundary elements of the test units are given in
Table 2.

237

Table 1Test unit steel reinforcement material


properties and layout
Test
unit

M05C
M05M

M10C
M10M

M15C

M15M

M20C*
M20M

Bar

Size

Spacing,
mm

fy,
MPa

fu,
MPa

sh

Esh, MPa

M1

M25

4 bars

524

11,034

M2

M22

4 bars

448

672

0.0026

11,262

M10

127

445

692

0.0083

7759

M10

102

445

692

0.0083

7759

M10

76

459

703

0.0075

7759

M1

M25

4 bars

464

697

0.0096

8966

M2

M22

4 bars

448

672

0.0094

8966

M10

127

476

746

0.0060

8966

M10

102

476

746

0.0060

8966

M10

76

545

730

0.0050

7586

M19

12 bars

439

705

0.0078

8966

M10

127

481

759

0.0054

9655

M10

102

481

759

0.0054

9655

M10

102

503

756

0.0030

7931

M1

M25

4 bars

586

2759

M2

M22

4 bars

421

630

0.0093

8621

M10

127

478

748

0.0060

10,690

M10

102

478

748

0.0060

10,690

M10

76

510

656

0.0072

5517

M1

M25

4 bars

451

703

0.0054

9310

M2

M22

4 bars

446

699

0.0060

8966

M10

127

438

703

0.0043

8793

M10

76 , 102

438

703

0.0043

8793

M10

76

443

717

0.0037

8621

is not displayed in response.

Note: 1 MPa = 0.145 ksi.

Table 2Test unit steel reinforcement ratios


Test unit

M15C

0.0528

0.0147

0.0357

0.0183

M20C and M20M

0.0556

0.0147

0.0237

0.0244

All others

0.0556

0.0147

0.0237

0.0244

Material properties
The HSC for this research was attained with traditional
mixture constituents following examples from commercially available mixtures.18 Compressive, tensile, and flexural strengths were assessed through standard testing, and
the values at the day of testing for all walls are provided in
Table 3. Table 1 lists the properties for the steel reinforcement with reference to the nomenclature noted in Fig. 3. The
properties given in Table 1 are average values from three
tensile tests on 457 mm (18 in.) long segments for each of
the reinforcement bars.

238

Fig. 3Test unit cross sections with reinforcement details


(refer also to Table 3).
Loading protocol
The test setup (Fig. 4) was designed to load the walls
in-plane as cantilevers. The walls were loaded monotonically and cyclically with constant axial load. The axial
load for all test units was 579 kN (130 kip), corresponding
to 0.10fcAg for a reference concrete strength of 34 MPa (5
ksi). Axial load was applied using hydraulic jacks and highstrength rods reacting against the top load stub through a
spandrel beam. The horizontal load was applied with a
servo-controlled actuator connected to a load stub at the
top of the wall. Lateral stability was provided by a pair of
parallel inclined tensioned chains on both sides of the wall.
Cyclic and monotonic loading, respectively, were applied
on two test units with the same design concrete strength
to assess shear strength/stiffness degradation and inelastic
web crushing limits under different loading histories. Cyclic
tests were done according to an incrementally increasing
fully reversed cyclic pattern. Four cycles (in quarter increments) were first applied in force control until the theoretical first yield force, Fy, defined as the force at the onset of
yield of the extreme longitudinal reinforcing bar in tension
as obtained from a moment-curvature analysis. The top
displacement at the theoretical first yield force, y, determined by the average of the measured values from the positive and negative loading excursions, was used to define
the experimental elastic bending stiffness, KE = Fy/y. The
ideal yield displacement,19 y, corresponding to displacement ductility one ( = 1), was determined using the experimental stiffness at first yield and the ideal yield force, Fy,
by y = Fy/KE. The ideal yield force Fy was computed by
means of a moment-curvature analysis and corresponded to
the moment at the critical section at which either the extreme
confined concrete fibers reached c = 0.004 or the extreme
steel fiber in tension reached s = 0.015, whichever occurred
first.19 The remainder of the test was conducted in displacement control with two cycles each at the system displaceACI Structural Journal/March-April 2014

Table 3Test unit concrete material properties


fc, MPa

ft, MPa

fr, MPa

Test
unit

Design
fc, MPa

M05C

34

46.0

1.37

3.25

0.207

5.49

0.0897

M05M

34

38.9

1.23

3.55

0.0966

5.37

0.0551

M10C

69

56.4

1.86

4.50

0.331

6.57

0.221

M10M

69

84.0

1.37

5.54

0.490

7.33

0.669

M15C

103

102

1.01

5.70

0.441

9.01

0.559

M15M

103

111

4.99

6.17

0.910

0.935

1.08

M20C

138

131

3.01

6.19

0.400

11.5

0.359

M20M

138

115

2.55

5.96

0.172

10.2

0.593

Note: 1 MPa = 0.145 ksi.

Table 4Force and displacement values at


theoretical and ideal yield
Test

M05C M05M M10C M10M M15C M15M M20C M20M

Fy,
kN

578

576

583

618

586

644

576

572

y,
mm

17.6

18.5

18.4

15.8

15.7

16.2

13.5

15.2

Fy,
kN

842

836

723

745

731

816

809

803

y,
mm

25.7

26.9

22.9

19.1

19.6

20.6

19.1

21.3

Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.

shear deformations were measured on two wall segments


(one from the wall base to 1016 mm [40 in.] high, and
the second one from the noted level to the wall top) with
a pair of displacement transducers arranged in opposing
diagonal directions on one wall face. Global displacement
at the effective height of the wall (2540 mm [100 in.]) was
measured with a string potentiometer. Further details of the
instrumentation are reported elsewhere.18

Fig. 4Test setup overview.


ment ductility levels = 1, 1.5, 2, 3, 4, and 6, or until
failure of the test unit. The monotonic tests were conducted
by applying the lateral load in force control until Fy, and
then in displacement control until failure. The values of Fy
and y that defined the loading protocol are listed in Table 4.
Instrumentation
The walls were instrumented to measure segmental flexural curvatures, shear deformations, and steel reinforcement
strains. The layout of the external instrumentation is shown
in Fig. 5. Flexural section curvatures were calculated using
displacement transducers placed along the height of the
column on both sides of the boundary elements. Average
ACI Structural Journal/March-April 2014

OBSERVATIONS
Common behavior
Common behavior observed on all test units is described
herein. No cracking was observed up to 0.25Fy. At 0.5Fy,
flexural cracking in the boundary elements and diagonal
shear cracking in the webs appeared. At 0.75Fy, elastic shear
cracking developed throughout the entire web. The HSC
walls developed a denser pattern of flexural and shear cracks
than the NSC walls. With increasing displacement ductility
crack density increased and the active cracks became wider.
Overall, shear crack spacing in the HSC walls was much
smaller than for the NSC walls. Even though diagonal shear
cracking under tension does not control the capacity of the
walls, it defines the height and width of the struts, and thus
affects the strut capacities. Strut capacity is also affected by
crack width, which dictates shear slip behavior at the crack
interface. Finally, relatively larger cover concrete spalling
on the compression boundary element was observed on the
HSC walls. Nonetheless, the heavily confined boundary
elements had no problem resisting the compression force.
239

Fig. 5External instrumentation layout (units in mm).


(Note: 1 mm = 0.0394 in.)
Damage patterns and failure mechanisms
Peculiarities on damage pattern, web crushing failure,
and ductility capacity are described herein individually. All
walls failed in web crushing according to the experimental
aim. Walls M05C and M10C failed on the first excursion to
= 2, and web crushing occurred at = 1.8. In both walls,
there was a sudden loss of strength upon failure of approximately 40%. Wall M15C performed in a ductile manner up
to = 4, and failed in web crushing on the second excursion
to ductility 4. This HSC wall had a gradual loss of strength
upon failure, losing only approximately 8% of its strength
compared with the load reached during the first cycle at
ductility 4. Wall M20C performed in a ductile manner up
to its failure by web crushing on the first excursion to =
6. Web crushing started to develop at a top displacement of
66 mm (2.6 in.), and the wall strength degraded by roughly
40% when the displacement reached the target displacement
for ductility six. Wall M05M failed by elastic web crushing
at = 2.3. Walls M10M, M15M, and M20M performed in
a ductile manner up to web crushing failure at = 7, 6.5,
and 9.2, respectively.
The damage and failure patterns in the wall bottom third
region (850 mm [33 in.]) for all test units are shown in Fig.6.
Test units with lower concrete compressive strength (M05C,
M05M, and M10C) failed after only minor levels of inelastic
response (Fig. 6(a) to (c)). Tensile cracking was minimal, and
cracks fully closed upon load reversal. No crack realignment
was observed and thus these walls were limited by standard,
or elastic, web crushing. The failures were sudden and the
crushing of the concrete struts occurred along the interface
of the wall web and the compression boundary element, and
instantly propagated along the wall height. The rest of the
test units exhibited moderate to high ductile behavior before
web crushing failure. Cracking was more extensive, and
240

crack spacing was much smaller. The fanning flexure-shear


cracking pattern was formed within the plastic hinge region,
with fairly flat cracks close to the bottom and much steeper
cracks at the top. An example of this inelastic flexure-shear
failure mode is shown in Fig. 6(d) for Wall M10M.
Loading protocol had a large influence on damage and
failure patterns. This can be seen by comparing the failure
modes between the cyclic and monotonic tests for Walls M15
and M20 (Fig. 6(e) to (h)). The monotonically loaded units
developed a denser crack pattern, with multiple branching
and wedge-shaped struts. Conversely, the cyclic tested walls
had wider spaced cracks and struts with a more uniform
width. The uniform damage pattern from denser cracking
in the monotonic tests allowed for more diagonal compression load paths in the wall web. The consistent loading also
permitted the struts to remain integral, thus allowing these
walls to sustain larger inelastic deformations. In constrast,
the wider cracking pattern from cyclic loading led to larger
crack misalignment and damage to the diagonal struts, which
reduced deformation capacity.
The cracking pattern due to cyclic loading is illustrated in
Fig. 1(c) and shown in Fig. 7, which illustrates the degradation of inelastic struts for wall M20C upon reaching the
second negative displacement target (second cycle) for
=4. Figure 7(a) is a view of the bottom region of the
wall from which the fanning crack pattern inside the plastic
zone can be discerned. The close-up view of the wall web
in Fig.7(b) shows the crisscross cracking pattern from the
reversed loading cycles. Upon reloading, crack misalignment
results from shear deformations due to yielding and bond-slip
effects in the longitudinal and transverse web reinforcement.
As cracks close for the compression load path to reestablish in
the previously formed struts, the crack misalignment induces
large local stresses due to shear friction and distortion of the
struts. This causes the web cover concrete to lose its bond
to the reinforcement and spall off, as seen in the lower left
region of the wall web (Fig. 7(b)). The test units gradually
lost their load-carrying capacity as a result of the diminished
load transfer efficiency of the concrete struts. Web crushing
in the cyclically loaded walls was observed to expand over a
large area within the plastic hinge region, and crushing of the
flexure-shear struts initiated in the center of the web and then
extended to the edge of the compression boundary element
(Fig. 6(e) and (g)). The noted differences between the cyclic
and monotonic responses became more significant for higher
values of concrete compressive strength.
RESULTS
Force-deformation response
The hysteretic force-displacement response of the four
walls under cyclic loading is shown in Fig. 8. Again, failure
in all cases was due to web crushing. Recalling that the reinforcement details were essentially the same for all walls, the
test results demonstrate that increased concrete compressive
strength allowed the walls to considerably increase their
inelastic deformation capacity by delaying shear failure. Web
crushing was thus shifted from an essentially elastic level for
Wall M05C, failing after completion of two cycles at = 1.5,
to a highly stable ductile response for Wall M20C, failing after
ACI Structural Journal/March-April 2014

Fig. 6Damage and failure patterns in wall bottom third region (850 mm [33.5 in.]).
sustaining two full cycles at =4. Comparison of inelastic
deformation capacity in terms of displacement ductility is
adequate because the ideal yield displacement for all walls
did not vary greatly, as shown in Table 4. The ductile behavior
displayed by Walls M15C and M20C shows that these units
preserved the high stiffness and lateral load-carrying capacity
characteristic of structural walls, mostly provided by the web,
while benefiting from the inelastic deformation capacity of
column flexural hinges at the boundary elements. Finally,
the energy dissipation capacity, as judged by the area of the
hysteresis loops, is also notably increased solely due to the
increase of the concrete compressive strength.
Inelastic behavior characteristics
According to Eq. (2) the length of the plastic hinge region
Lpr is directly proportional to the angle of the flexural-shear
cracking, and hence the force demand on the critical inelastic
strut. The spread of plasticity can be taken as the length over
which plastic curvatures exceed the yield curvature from an
ACI Structural Journal/March-April 2014

idealized bilinear moment-curvature response.19 The curvature profiles of the M10 and M20 walls are shown in Fig.9.
For cyclic loaded test units that failed at low to medium
ductility levels, the curvature distribution along the height
was almost linear until failure. For monotonic loaded test
units that failed at a high ductility level, the plastic rotation
was mainly concentrated within 300 mm (12 in.) from the
bottom of the wall. It can be seen, however, that the spread
of plasticity is very similar at equal displacement ductility
levels for both monotonic and cyclic loaded walls. The curvature distributions provide local-level evidence of the significant inelastic flexural deformations sustained by the walls.
Furthermore, the observed fanning crack pattern (Fig.6) and
the essentially linear distribution of plastic curvatures along
the plastic hinge region (with the linearity only disturbed by
boundary effects at the footing) suggests that Eq. (4) can be
used to assess the demand on the critical inelastic strut.
Table 5 shows the separate contribution of flexure and
shear effects to the displacement at the wall top for the cycli241

Fig. 7Degradation of inelastic struts for M20C wall during ductility 4 cycling: (a) view of wall bottom third region (850 mm
[33.5 in.]); and (b) close-up view of wall web.

Fig. 8Hysteretic force-displacement response of cyclic tests. (Note: 1 MPa = 0.145 ksi; 1 mm = 0.0394 in.)
cally tested units at the first positive peak of each ductility
level. The flexure-induced displacement f at the top of
the wall was calculated by adding the contribution of individual sectional rotations. The top wall displacement due to
shear s was calculated by summing the shear deformations
measured on the two wall segments (Fig. 5). From the data
in Table 5 it can be confirmed that the shear displacements
are linearly related to the flexural displacements.12 For the
eight tested walls,18 the average ratio of shear to flexural
displacements was 0.23, with a standard deviation of 0.04.
Figure 10 shows the average shear stress versus shear strain
hysteretic response of Wall M20C in the bottom and top wall
segments. The shear deformations were mainly concentrated
in the bottom wall segment (1016 mm [40 in.] from the base),
which is where the plastic hinge region develops. It can be
seen that the average shear stresses considerably exceeded
the ACI limits. This observation applies to all walls because
they had similar levels of lateral load resistance.
To provide further insight and quantitative information on
the performance of the tested walls, Fig. 11 provides a brief
overview of average strain demands at mid-depth on the wall
242

web. Shown in Fig. 11(c) are longitudinal strain profiles for the
M20 walls, which have a linear variation along the height (with
disturbance near the footing). The strains were calculated using
the displacement transducers along both sides of the boundary
elements (Fig. 5). The profiles are consistent with the moment
gradient on the wall. It can be further observed that the strain
profiles for both M20 walls are essentially the same at equal
displacement ductility demands. This was consistent for the
other three wall sets, supporting the evidence of equal flexural
demands on monotonic and cyclic tests.
Principal strains in the wall web were estimated from consideration of a whole wall segment (web and boundary elements)
through Mohrs circle with the measured longitudinal strains,
the measured average shear strains (Fig. 5), and neglecting the
transverse strains (due to the presence of the heavily reinforced
boundary elements). Principal strain values calculated this way
at the web mid-depth are shown in Fig. 11(b) against displacement ductility for all walls. Except for some deviations for
the M05 walls, the average principal strains were equal for all
walls at the same displacement ductility level.

ACI Structural Journal/March-April 2014

Fig. 9Average curvature strain profiles for: (a) M10 walls; and (b) M20 walls.

Fig. 10Average shear stress-strain response of: (a) bottom; and (b) top segments of Wall M20C.
Monotonic versus cyclic loading
A comparison of the force-displacement envelopes of
the cyclic and monotonic tests is shown in Fig. 12. It is
clear that higher fc resulted in higher inelastic deformation
capacity. The force-deformation response of the walls, up
to their respective deformation limit, is considered to have
been essentially the same, with minor differences due to: a)
variations in fc; b) longitudinal reinforcement differences
for Wall M15C (Fig. 3); c) earlier spalling in the compression toe for Wall M20C; and d) reduction of the effective
concrete compressive strength due to more severe cracking
for the cyclically loaded walls.
The increase in deformation capacity, however, was not
directly proportional to fc, particularly for the cyclic loaded
walls. The monotonic and cyclic deformation capacities of
the M05 walls were essentially the same (Fig. 12(a)). The
response was similar because both walls failed close to
the elastic range, and only minimal cycling was done on
WallM05C. The responses of the other walls show that
while increased concrete strength leads to larger deformation capacity, cyclic loading curtails this improvement.
The deformation capacity reduction of the cyclically tested
walls is attributed to the damage of the flexure-shear struts
from cyclic loading, which reduces their capacity to transfer
load from the tension to the compression boundary element.
This effect is best seen by observing the responses for the M15
walls in Fig. 12(c). The effect was not as clearly captured for
the M10 walls (Fig. 12(b)) because fc for M10C was lower
than that for M10M. Nonetheless, given the response of Wall
M15C, it can be expected that if fc for Wall M10C had been
closer to the design target, its deformation capacity would
have been increased, and the cyclic and monotonic envelopes
would have been similar to those obtained for the M15 walls.
Comparison of the M20 wall response envelopes indicates a
ACI Structural Journal/March-April 2014

Table 5Flexure and shear components of wall


top displacement at first positive peak of each
ductility level
Ductility
= 1.0
= 1.5
= 2.0
= 3.0
= 4

Displacement

M05C

M10C

M15C

M20C

f, mm

18.5

18.0

14.5

14.5

s, mm

5.08

6.10

3.53

3.30

f, mm

28.4

27.7

22.6

23.1

s, mm

7.62

7.87

4.83

4.57

f, mm

30.5

30.7

s, mm

6.60

6.60

f, mm

47.0

47.0

s, mm

9.40

9.91

f, mm

62.5

62.0

s, mm

13.2

15.5

Note: 1 mm = 0.0394 in.

significant effect of cyclic loading on the deformation limit


of Wall M20C despite the larger concrete strength in M20C
compared with that in M20M. It is thus hypothesized that the
reduced deformation capacity in the cyclically tested HSC
walls is due to the negative convergence of an increased
stress intensity field at crack misalignment and a reduced
cracking bridge zone from the higher strength concrete. This
can be understood upon considering that HSC experiences
dramatic strength degradation in the postpeak response, and
the effect of further strength increase is not appreciable. For
structural walls, the later effect would indicate a curtailing
effect on the increased inelastic deformation gains on web
crushing capacity for increasing values of fc.
The reduced gain in deformation capacity of Wall M20C
compared with that of Wall M15C would seem to indicate
243

Fig. 11Overview of average strain demands on wall webs:


(a) longitudinal strain profiles in M20 walls; and (b) principal strains versus ductility.
that increased concrete strength does not lead to additional
deformation capacity beyond a certain point. Wall M20C,
however, failed after completion of two cycles at = 4, while
M15C failed during the first loading branch of the second
cycle at = 4. Evaluation of the dissipated hysteretic energy
(area inside the hysteresis loop) shows that all walls provided
essentially the same level of specific energy dissipation for a
given displacement ductility level.18 This can be qualitatively
seen by observing the hysteretic responses in Fig. 8. Because
the walls had the same reinforcement details, this is to be
expected. Thus, the increased shear capacity of the HSC walls
improved their hysteretic energy dissipation capability. Based
on this evaluation, Wall M20C had 27% higher energy dissipation capacity than Wall M15C.18 The sum of the dissipated
energy for all ductility levels shows that increased concrete
strength increased the inelastic energy dissipation capacity of
the walls by delaying web crushing failure.
Considering the strut resistance mechanism proposed by Hines
and Seible13 (Fig. 1(b)), the similitude in principal strain values
(Fig. 11(b)) indicates that the force demand in the walls at equal
displacement ductility levels was the same. It is clear, however,
that the walls had different deformation and web crushing capacities. The difference is then attributed to the concrete strength (as
reflected in Eq. (3)) and the loading pattern. This was explored
by estimating the concrete softening factor k using experimentally derived values for Lpr and the shear force at web crushing
failure. Equation (1) was used for Walls M05C, M05M, and
M10C because they are considered to have failed by elastic web

244

crushing, while the rest of the walls, which failed by inelastic


web crushing, followed Eq. (3). Figure 13(a) plots the calculated
values of k versus normalized shear distortions in the plastic
hinge region. The maximum (condition before web crushing)
average shear distortion within the plastic hinge region, m, was
normalized by the strain at peak stress in compression co, which
was estimated from the model by Tasemir et al.20 The figure also
shows the concrete strength reduction factor relation proposed
by Collins21 in terms of shear strains, which is essentially the
constitutive model for cracked concrete in compression in the
modified compression field theory (MCFT). This relation has
been shown to relate well to test data in which the compression
stresses act along uniform parallel struts across the section,9 or
elastic shear. Results from this program, however, deviate from
the noted model in interesting ways. The data in Fig. 13(a) shows
how the monotonic and cyclic tests follow different degradation
trends for the softening factor k, with lower values and a faster
decay with increasing shear distortion for the cyclic walls.
The experimental k values are plotted against concrete
strength in Fig. 13(b), where again the data is clearly segregated in terms of the loading protocol. The cyclic and monotonic wall data was fitted with exponential functions only
for the purpose of illustrating the data trends. It can be seen
that the M05 data points are close to each other because
both failed in elastic web crushing. For increasing concrete
strengths, the softening factor decays faster for the cyclic
walls. It is of interest to note that the softening factor (and
thus, the inelastic web crushing capacity) is affected both
by the cyclic loading and the increase in concrete strength,
or, stated differently, cyclic loading had an increased detrimental effect on the effective capacity of the shear resisting
struts with increasing concrete compressive strength.
Web crushing capacity models
Table 6 compares the web crushing capacities with different
predictive models. It can be noted that ACI shear provisions
considerably underestimate the web crushing strength. At the
same time, the prediction quality of the model by Hines and
Seible13 on the cyclic tests deteriorates with increasing concrete
strength. Thus, the experimental program revealed that rational
web crushing models like the one by Hines and Seible need
further considerations to be applicable to HSC structural walls.
A modified version of the Hines and Seible model, as well as
the finite element implementation of the MCFT in modeling
the behavior of the HSC walls was proposed by Liu,22 and will
be reported in a future paper by the authors.
CONCLUSIONS
Eight cantilever walls were tested with design concrete
compressive strengths of 34, 69, 103, and 138 MPa (5, 10,
15, and 20 ksi) under cyclic and monotonic loading to study
the effects of HSC and damage accumulation on the inelastic
web-crushing capacity of structural walls. The following
conclusions are offered specifically referring to structural
walls with well-confined boundary elements:
1. The experiments clearly demonstrated the differences
between elastic and inelastic web crushing behavior and
their dependence on concrete compressive strength;

ACI Structural Journal/March-April 2014

Fig. 12Comparison of force-displacement envelopes for cyclic and monotonic tests: (a) M05; (b) M10; (c) M15; and (d)M20.
(Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.; 1 ksi = 6.895 MPa.)

Fig. 13Average compression softening factor k versus: (a) normalized maximum shear distortion in plastic region; and
(b)concrete compressive strength.
Table 6Comparison of web crushing capacities with models
ACI 31814

Experiment

Hines and Seible13

Test unit

u, mm

Fu, kN

u, mm

Fu, kN

Difference, %

u, mm

Fu, kN

Difference, %

M05C

45.0

803

8.64

342

81

48.5

821

M05M

45.0

855

8.13

322

82

26.9

725

40

M10C

42.7

751

8.38

387

80

64.3

751

51

M10M

130

900

9.91

478

92

101

853

22

M15C

78.7

819

10.2

497

87

128

889

62

M15M

133

934

11.7

542

91

160

966

20

M20C

76.5

815

14.0

589

82

M20M

189

923

13.2

553

93

Flexure
196

992

Notes: 1 kN = 0.225 kip; 1 mm = 0.0394 in.

2. Increase in concrete compressive strength enhances the


ductility and hysteretic energy capacity of structural walls
by preventing web crushing shear failures;
3. Web crushing strength increases in proportion to fc as long
as struts remain undamaged. Hence, transformation from an
elastic web crushing failure to an inelastic web crushing failure
can be achieved simply by increasing the concrete strength;
4. Damage to struts caused by cyclic loading and inelastic
deformations limits web crushing strength independently of fc.
ACI Structural Journal/March-April 2014

While this conclusion may be generally well recognized for


walls limited by both elastic and inelastic web crushing, experimental evidence from this research, based on corresponding
monotonic and cyclic tests, showed that degradation from
cyclic loading increases with increasing concrete strength;
5. Concrete compressive strength does not linearly
increase web crushing strength as implied by rational web
crushing models; rather, the relationship is nonlinear, with a
decreasing limit as concrete strength increases. This obser245

vation is not well-described by current web crushing models.


Therefore, it is advised that these models consider the limits
demonstrated by the reported tests when considering HSC to
improve the web crushing performance of walls;
6. The web crushing capacity of structural walls with
well-confined boundary elements was found to be well in
excess of levels acceptable in current practice for a wide
range of concrete compressive strengths; and
7. Rational assessment of web crushing limits can open
up new possibilities for acceptable ductile failure modes on
reinforced concrete structural walls.
AUTHOR BIOS

ACI member Rigoberto Burgueo is an Associate Professor of structural


engineering and Director of the Civil Infrastructure Laboratory at Michigan
State University, East Lansing, MI. He received his BS, MS, and PhD from
the University of California, San Diego, La Jolla, CA. He is a member of ACI
Committees 341, Earthquake-Resistant Concrete Bridges, and 440, Fiber-Reinforced Polymer Reinforcement. His research interests include nano-engineered structural materials, composite materials and structures, multi-scale
modeling, and seismic performance of reinforced concrete structures.
ACI member Xuejian Liu is a former Graduate Research Assistant at
Michigan State University, where he received his PhD in civil engineering.
He is a member of Joint ACI-ASCE Committee 445, Shear and Torsion.
His research interests include the seismic behavior of reinforced concrete
structures and fiber-reinforced concrete.
ACI member Eric M. Hines is a Principal at LeMessurier Consultants,
Inc., Cambridge, MA, and is a Professor of Practice at Tufts University,
Medford, MA. He received his PhD in structural engineering from the
University of California, San Diego. He is a member of ACI Committee 341,
Earthquake-Resistant Concrete Bridges. His research interests include the
seismic performance of low-ductility structural systems in moderate seismic
regions and inelastic behavior of reinforced concrete structures.

Av
d
Esh
Fu
Fy
Fy
f1
fc
fu
fy
fyv
jd
k
Lpr
NCfs
NCs
NDfs
NDs
R
S
T
Tyav
tw
u
y
T
1
sh
y

fs
s
h
l
n
s

246

=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=

NOTATION

area of transverse steel at given level


incremental angle for calculating critical strut area
elastic modulus at onset of strain hardening
shear force at ultimate
ideal yield shear force
first yield shear force
principal tensile stress
unconfined concrete compressive cylinder strength
ultimate steel stress
steel yield stress
transverse steel yield stress
distance between flexural tension and compression force resultants
concrete compression strength softening factor
plastic hinge region length
flexure-shear strut compression capacity
standard shear strut compression capacity
flexure-shear strut compression demand
standard shear strut compression demand
radius for critical compression strut fan
transverse reinforcement vertical spacing
flexural tensile force resultant
effective average flexural tensile yield force resultant
structural wall thickness
test unit flexural top displacement
ideal yield lateral displacement
incremental tensile flexural force
principal tensile strain
steel strain at onset of strain hardening
ideal yield curvature
displacement ductility
flexure-shear crack angle measured from longitudinal axis
shear crack angle measured from longitudinal axis
structural wall transverse reinforcement ratio
boundary element longitudinal reinforcement ratio
structural wall longitudinal reinforcement ratio
boundary element transverse reinforcement ratio

ACKNOWLEDGMENTS

The research described in this paper was carried out under funding from
the National Science Foundation under Grant No. CMS-0530634. The
authors thank the staff and students of MSUs Civil Infrastructure Laboratory, where the reported work was conducted.

REFERENCES

1. Leonhardt, F., and Walther, R., The Stuttgart Shear Tests 1961,
Transaction No. 111, Cement and Concrete Association, London, UK, 1961,
134 pp.
2. Mattock, A. H., and Kaar, P., Precast-Prestressed Concrete Bridges,
4, Shear Tests of Continuous Girders, Journal, Portland Cement Association Research and Development Labs, V. 3, No. 1, Jan. 1961, pp. 19-46.
3. Robinson, J. R., Essais a leffort trenchant de pouters a ame mince en
bton arm, Annales des Ponts et Chausses, Mar.-Apr. 1961, pp. 226-255.
4. Placas, A., and Reagan, P. E., Shear Failure of Reinforced Concrete
Beams, ACI Journal, V. 68, No. 10, Oct. 1971, pp. 763-773.
5. Wang, T. Y.; Bertero, V. V.; and Popov, E. P., Hysteretic Behavior
of Reinforced Concrete Framed Walls, Earthquake Engineering Research
Center Report 75/23, University of California, Berkeley, Berkeley, CA,
Dec. 1975, 367pp.
6. Oesterle, R. G.; Fiorato, A. E.; Johal, L. S.; Carpenter, J. E.; Russell,
H. G.; and Corley, W. G., Earthquake Resistant Structural WallsTests
of Isolated Walls, NSF Report GI-43880, Portland Cement Association,
Skokie, IL, 1976, 315 pp.
7. Oesterle, R. G.; Ariztizabal-Ochoa, J. D.; Fiorato, A. E.; Russell,
H.G.; and Corley, W. G., Earthquake Resistant Structural WallsTests
of Isolated Walls, Phase II, NSF Report ENV77-15333, Portland Cement
Association, Skokie, IL, 1979, 331 pp.
8. Vallenas, J. M.; Bertero, V. V.; and Popov, E. P., Hysteretic
Behavior of Reinforced Concrete Structural Walls, Earthquake Engineering Research Center Report 79/20, University of California, Berkeley,
Berkeley, CA, 1979, 234 pp.
9. Oesterle, R. G.; Fiorato, A. E.; and Corley, W. G., Web Crushing
of Reinforced Concrete Structural Walls, ACI Journal, V. 81, No. 3,
May-June 1984, pp. 231-241.
10. Hines, E. M.; Seible, F.; and Priestley, M. J. N., Seismic Performance of Hollow Rectangular Reinforced Concrete Piers with Highly-Confined Corner ElementsPhase I: Flexural Tests, and Phase II: Shear Tests,
Structural Systems Research Project Report 1999/15, University of California, San Diego, La Jolla, CA, 1999, 266 pp.
11. Hines, E. M.; Dazio, A.; and Seible, F., Seismic Performance of
Hollow Rectangular Reinforced Concrete Piers with Highly-Confined Corner
ElementsPhase III: Web Crushing Tests, Structural Systems Research Project
Report 2001/27, University of California, San Diego, La Jolla, CA, 2001, 239 pp.
12. Hines, E. M.; Restrepo, J. I.; and Seible, F., Force-Displacement
Characterization of Well Confined Bridge Piers, ACI Structural Journal,
V. 101, No. 4, July-Aug. 2004, pp. 537-548.
13. Hines, E. M., and Seible, F., Web Crushing of Hollow Rectangular Bridge
Piers, ACI Structural Journal, V. 101, No. 4, July-Aug. 2004, pp.569-579.
14. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-11) and Commentary, American Concrete Institute,
Farmington Hills, MI, 2011, 503 pp.
15. ACI Committee 318, Building Code Requirements for Reinforced
Concrete (ACI 318-63), American Concrete Institute, Farmington Hills,
MI, 1963, 144 pp.
16. Paulay, T., and Priestley, M. J. N., Seismic Design of Reinforced
Concrete and Masonry Buildings, Wiley Interscience, New York, 1992, 768 pp.
17. Vecchio, F. J., and Collins, M. P., The Modified Compression-Field
Theory for Reinforced Concrete Elements Subjected to Shear, ACI
Journal, V. 83, No. 2, Mar.-Apr. 1986, pp. 219-231.
18. Liu, X.; Burgueo, R.; Egleston, E.; and Hines, E. M., Inelastic Web
Crushing Performance Limits of High-Strength-Concrete Structural Wall
Single wall Test Program, Report No. CEE-RR2009/03, Michigan State
University, East Lansing, MI, 2009, 281 pp.
19. Priestley, M. J. N.; Seible, F.; and Calvi, G. M., Seismic Design and
Retrofit of Bridges, John Wiley & Sons, Inc., New York, 1996, 686 pp.
20. Liu, X., Inelastic Web Crushing Performance Limits of HighStrength-Concrete Structural Walls, PhD dissertation, Department of Civil
and Environmental Engineering, Michigan State University, East Lansing,
MI, 2010, 215 pp.
21. Tasdemir, M. A.; Tasdemir, C.; Akyuz, S.; Jefferson, A. D.; Lydon,
F. D.; and Barr, B. I. G., Evaluation of Strains at Peak Stresses in Concrete:
A Three-Phase Composite Model Approach, Cement and Concrete
Composites, V. 20, 1998, pp. 301-318.
22. Collins, M. P., Toward a Rational Theory for RC Members in
Shear, Proceedings, ASCE, V. 104, No. ST4, Apr. 1978, pp. 649-666.

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S21

Response of Precast Prestressed Concrete Circular Tanks


Retaining Heated Liquids
by Michael J. Minehane and Brian D. ORourke
The present study investigated the influence of heated water
storage, upward to 95C (171F), on precast prestressed concrete
circular tanks. Modern design standards for concrete liquidretaining structures require that thermal effects be considered
for the serviceability limit state and the ultimate limit state when
deemed significant. Most recognized standards, however, do not
provide guidance for the analysis of such effects. Research in this
area is also limited and almost exclusively concerned with ambient
thermal conditions, with a maximum temperature change of 30C
(54F) in any instance.
A finite element study incorporating thermomechanical coupling
investigated the magnitude of stresses associated with thermal
storage. A linear eigenvalue analysis examined the ultimate limit
state of buckling for restrained tank walls due to the thermallyinduced combined axial compression and bending. Consequent
design implications were established and recommendations made
for accommodating thermal loading.
Keywords: buckling; elevated temperature; finite element analysis; material properties; prestressed concrete; reservoirs; thermal loading; thermal
storage.

INTRODUCTION
The significance of thermal effects on concrete reservoir
walls for ambient conditions is long established. An early
study by Priestley1 determined that temperature gradients of
30C (54F) through the wall thickness can exist in warm
climates when the effects of solar radiation are considered.
Priestley1 demonstrated that the resulting tensile stresses
were large enough to overcome the residual compression,
and cracking would inevitably occur. Ghali and Elliott2
developed closed-form solutions for the thermal analysis
of elastic tank walls with varying base restraint and that
are free at the top. Through numerical examples, it was
shown that a gradient of 30C (54F) through the wall
thickness was sufficient to cause cracking. This supported
Priestleys1 proposal that the design should be based on a
serviceability criterion of limiting crack widths rather than
a limiting tensile stress. Although modern design standards
require that thermal effects are considered for the serviceability limit state, few provide guidance for the analysis of
such effects. Pioneering design codes with regard to this
are NZS31063 and AS3735,4 which provide design tables,
originally derived by Priestley,1 to calculate hoop forces and
vertical moments for tank walls free at the top and either
free-sliding, pinned, or fixed at the base.
The studies reviewed were exclusively applicable to tank
walls free at the top. As thermal storage tanks require a roof,
the associated radial restraint at the top of the wall alters
the internal force distribution. Moreover, the magnitude of
the internal forces resulting from the thermal expansion of
ACI Structural Journal/March-April 2014

the tank walls will be shown to be prohibitive from a design


perspective, unless provisions are made for radial displacement during service.
RESEARCH SIGNIFICANCE
This paper investigates the feasibility and implications
of thermal storage using cylindrical concrete reservoirs,
for which there is currently a paucity of information. The
research has practical applications in the oil, gas, and nuclear
containment industries, in addition to thermal storage for
district heating and related schemes. Although particular
reference is made throughout to precast prestressed concrete
storage tanks, the research is also applicable to partially
prestressed and reinforced concrete reservoirs.
INFLUENCE OF ELEVATED TEMPERATURES ON
MATERIAL PROPERTIES
Mechanical properties
EN 1992-1-2,5 EN 1992-3,6 and FIB Bulletin 55: Model
Code 20107 each define the reductions in the mechanical
properties of both the concrete and steel reinforcement for
elevated temperatures. For the temperature range under
consideration for the current study, the associated strength
reductions are insignificant. It is reasonable to suggest that
any minor reduction in the strength and stiffness of concrete
may be discounted when the effect of long-term thermal
exposure is considered. Mears8 tested concrete specimens
subjected to a constant temperature of 65C (149F) for
5000 days and observed that the long-term exposure had, in
fact, the effect of increasing the compressive strength and
the modulus of elasticity of concrete. The same trend was
also recorded by Komendant et al.,9 who tested concrete at
71C (160F) for 270 days, and Nasser and Lohtia,10 who
tested concrete at 121C (250F) for 200 days.
It would appear, however, that this trend is only valid for
temperatures below 150C (302F), as long-term exposure
to temperatures in excess of this resulted in a reduction in the
mechanical properties of concrete.10,11
Creep
Creep of concrete increases at higher temperatures. Extensive research has been carried out on the influence of
temperature on concrete creep for structures used in nuclear
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-065, doi:10.14359.51686441, was received February 23, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

247

Fig. 1Creep coefficient multipliers with increasing


temperature. (Note: F = 1.8C + 32.)
containment. It would appear from the literature that the
use of a thermal scaling factor, or creep coefficient multiplier, is appropriate in accounting for temperature effects on
creep. Figure 1 presents a comparison of creep coefficient
multipliers from guidance provided by CEB 20812 and fib
Bulletin55: Model Code 20107 in addition to experimental
studies carried out by Brown,13 Gross,14 and Nasser and
Neville.15 For a temperature of 95C (203F), the creep coefficient multipliers range from approximately 1.95 to 2.45.
An accurate evaluation of creep at elevated temperatures
is difficult to attain, as creep is sensitive to the evaporable
water in the mixture. Consequently, an accurate value of
the moisture content is desirable if a precise assessment is
to be made. The moisture content, particularly at elevated
temperatures, is sensitive to the member thickness. The
majority of the experimental results were developed for the
walls of nuclear containment structures, which are generally
members of wall thicknesses in the range of 1000to 1500mm
(39 to 59 in.). Because the walls of precast prestressed
storage tanks are much thinnertypically 150to 250 mm
(5.9 to 9.8in.)the walls would lose much more moisture
comparably, which would suggest that a lower value of a
creep coefficient multiplier may be more appropriate. The
experimental results are generally based on uniaxial tests.
As prestressed concrete circular tanks are subject to a multiaxial state of stress, a further reduction in the predicted
creep strain would be appropriate, in line with the findings
of Hannant16 and McDonald.17
Bond strength
Numerous experimental studies conclusively reveal that
the bond strength decreases with increasing temperature.
This is primarily attributed to the differing coefficients of
thermal expansion for steel and concrete. Figure 2 compares
test results from studies by Harada et al.,18 Kagami et al.,19
248

Fig. 2Residual bond strength ratio with increasing


temperature. (Note: 1F = 1.8C + 32.)
Haddad et al.,20 Chang and Tsai,21 Baant and Kaplan,22 and
Huang.23 A wide scatter in the residual bond strength ratios
(ratio of bond strength of heated specimen to that of a specimen at ambient temperature) is observed. This is primarily
due to the many variables involved, including concrete
mixture, compressive strength, exposure duration, method
of cooling, and bar size.
Stress relaxation
Owing to different coefficients of thermal expansion, steel
expands relative to concrete with increasing temperature.
Consequently, for pretensioned members, this will effectively increase the loss of prestress due to stress relaxation.
fib Bulletin 55: Model Code 20107 quantifies the increase
in loss due to stress relaxation with increasing temperature
for a duration of 30 years (Fig. 3). The significance of high
temperatures on the stress relaxation is evident, as a value
of approximately 2.5% at ambient temperature increases to
approximately 15.0% at 100C (212F).
FINITE ELEMENT ANALYSIS
Finite element analysis was carried out using commercial finite element software. The cylindrical structure was
idealized using two-dimensional (2-D) axisymmetric
models comprising quadrilateral solid field and continuum
elements. The thermomechanical transient analysis incorporated a semicoupled procedure and was time-stepped
according to predefined intervals. A semicoupled analysis
involves running the thermal and structural analyses separately and is conducted when the thermal solution is not
considerably affected by changes in geometry. The thermal
analysis, which runs first, is governed by the quasi-harmonic
transient heat conduction equation. The resulting temperature distribution for a given time step is subsequently fed to
the structural analysis for the calculation of displacements
ACI Structural Journal/March-April 2014

and consequent stresses and strains. A maximum element


size of 0.3 m (0.98 ft) was established following a mesh
convergence study. The material properties used throughout
the finite element study are given in Table 1.
Verification of modeling procedure
The modeling procedure was verified with existing results
in the literature from Priestley,1 Ghali and Elliott,2 and
Vitharana and Priestley.24 In each case, the numerical examples considered a tank wall pinned at the base, free at the top,
and subject to a 30C (54F) ambient temperature gradient.
Figure 4 shows the finite element model for the comparison
with Priestley1 for a tank of dimensions R = 15.1 m (49.5 ft),
H = 7.2 m (23.6 ft), and t = 0.2 m (7.87 in.). As is evident
from Fig. 5, good agreement for hoop forces and vertical
bending moments was observed in each instance.

Idealization and boundary conditions


To idealize a typical precast concrete tank incorporating
a roof, the wall ends were restrained from radial displacement but free to rotate. The top of the wall was also free to
displace vertically because the strain due to thermal expansion far exceeds that imposed by the self-weight of typical
roof construction comprising precast flooring on a grid of
columns. Figure 6 shows the mesh discretization for the
axisymmetric models and displays a deformed contour of
hoop forces obtained from the structural analysis.
DESIGN IMPLICATIONS
Limiting compressive stress
For a tank restrained radially at its ends, the hoop force
induced by thermal storage is compressive over the entire
height of wall and is significant in magnitude. It is necessary to check the resulting compressive stress against limits
provided in design standards and guidance. Table 2 presents compressive stress limits from standards and guidance,
including EN 1992-1-1,25 BS 8007,26 PCI,27 and NZS 3106.3
The limits are expressed in terms of the concrete cylinder
strength fck. The limit stipulated by BS 8007 is given in terms
of the cube strength fcu, but an approximate conversion is
made herein. Although compressive stresses exceeding
those given in Table 2 may not lead to failure, nonlinearity
associated with creep at higher stress-strength ratios would
need to be taken into account.
Table 1Material properties used in finite element
study

Fig. 3Increase in stress relaxation with increasing


temperature, after fib Bulletin 55: Model Code 2010.7 (Note:
1F = 1.8C + 32.)

Material property

Value

Modulus of elasticity, MPa (psi)

33,000 (4.78 106)

Poissons ratio

0.2

Coefficient of thermal expansion, /K (/F)

10 106 (5.55 106)

Thermal conductivity, W/mK


(Btu.in/h.ft2F)

1.5 (10.4)

Specific heat, J/kgK (Btu/lbF)

900 (0.21)

Convective heat transfer coefficient, W/m2K (Btu/h ft.2F)

8.33 (1.47)

Fig. 4Deformed contour of stresses and associated internal force distribution for comparison with Priestley1 (in MPa). (Note:
1 MPa = 145 psi.)
ACI Structural Journal/March-April 2014

249

Figure 7 gives hoop compressive stresses for various


temperatures for a tank with dimensions D = 30.4 m (99.7 ft),
H = 7.0 m (23 ft), and t = 0.2 m (7.9 in.). A concrete cylinder
strength of 40 MPa (5800 psi) and a modulus of elasticity
of 33 GPa (4744 ksi) is assumed. These section dimensions
and material properties are used for each numerical example
Table 2Limiting compressive stress at service
from modern design standards/guidance
Design standard/guidance

Limiting compressive stress at service

EN 1992-1-1

0.45fck

BS 8007

0.41fck (0.33fcu)

PCI

0.45fck

NZS 3106

0.40fck

presented in this study. As thermal storage tanks commonly


incorporate external insulation, the resulting temperature
distribution involves predominantly a constant temperature
across the concrete section with a small temperature differential. As such, for simplicity, the small differential that
may be present is ignored herein. Figure 7 shows that an
average temperature of approximately 50C (90F) across
the concrete section produces localized hoop compressive
stresses at the wall ends that exceed each of the limits given
in Table 2.
Circumferential post-tensioning
As the thermally induced hoop forces over the wall
height are compressive, the circumferential post-tensioning
requirements remain unchanged. The hoop tension arising
from hydrostatic loading when the liquid is not heated is the

Fig. 5Verification of modeling procedure with Priestley,1 Ghali and Elliott,2 and Vitharana and Priestley.24 (Note: 1 m =
3.28ft; 1 MPa = 145 psi; 1 kN/m = 5.71 kip/in.; 1 kNm/m = 0.225 kip-ft/ft.)

Fig. 6Finite element model showing deformed contour of hoop stresses (in MPa). (Note: 1 MPa = 145 psi.)
250

ACI Structural Journal/March-April 2014

Fig. 8Compressive hoop forces with increasing temperature. (Note: 1 kN/m = 5.71 kip/in.; 1F = 1.8C + 32.)
Fig. 7Compressive hoop stresses with increasing temperature. (Note: 1 MPa = 145 psi; 1F = 1.8C + 32.)
critical loading condition that the circumferential post-tensioning is designed to cater for. Figure 8 includes the tensile
force distribution resulting from hydrostatic loading derived
using the beam-on-elastic foundation analogy.
Vertical prestressing
The vertical bending moment distribution arising from
hydrostatic and thermally induced loading is given in
Fig. 9. To establish an approximate upper limit on the
vertical moments, a cracking moment was calculated from
the following equation

I
M cr = ( ft + fmax )
t / 2

(1)

where I is the second moment of area; t is the wall thickness;


ft is the concrete tensile strength; and fmax is the concentric
precompression stress required to eliminate tensile stresses
while also satisfying maximum compressive stress limits at
the extreme fiber.
Figure 10 applies the cracking moment to the previously
derived vertical bending moments. The moments represent
total moments, that is, the hydrostatic moments subtracted
from the thermally induced moments, because the two loadings are coexistent.
It is apparent that an average temperature across the
concrete section of approximately 40 to 50C (72 to 90F)
produces vertical moments that exceed the cracking moment.
This temperature limit is approximately consistent with that
for the compressive stress requirement.

ACI Structural Journal/March-April 2014

Fig. 9Vertical bending moments with increasing temperature. (Note: 1 kNm/m = 0.225 kip-ft/ft; 1F = 1.8C + 32).
Influence of creep on thermal response
The magnitude of strain-induced effects such as thermal,
creep, and shrinkage stresses at a given time is directly
proportional to the concrete modulus of elasticity at that
time. As such, realistic values of the modulus of elasticity
251

Fig. 10Total vertical moments with increasing temperature.


(Note: 1 kNm/m = 0.225 kip-ft/ft; 1F = 1.8C + 32.)
are desirable if an accurate evaluation is to be made of the
structural response. A common approach employed by
modern design standards involves the use of an age-adjusted
modulus of elasticity that includes a creep component. Figure
11 shows the influence of creep on the concrete modulus
of elasticity using the age-adjusted or effective modulus of
elasticity defined in EN 1992-1-125 as follows
Ec ,eff =

Ecm
1 + f ( , to )

(2)

Inputs to the EN 1992-1-125 creep model were concrete


modulus of elasticity = 33 GPa (4744 ksi); wall thickness
t = 200 mm (7.87 in.); relative humidity = 80%; concrete
compressive cylinder strength = 40 MPa (5800 psi); age
at loading = 28 days; normal cement class. Figure 11 also
includes a thermally adjusted effective modulus of elasticity
using a creep coefficient multiplier of 1.5.
Influence of cracking on the thermal response
Estimating thermal stresses based on elastic section properties is only valid for uncracked sections, as is the case
for prestressed tanks where decompression is satisfied.
For reinforced concrete or partially prestressed tank walls,
cracking has the effect of considerably relaxing the thermal
response. NZS 31063 and AS 37354 account for this using
load-reduction factors that allow for the reduced section
stiffness that accompanies cracking. The factors include

252

Fig. 11Influence of creep on concrete modulus of elasticity in accordance with EN 1992-1-1.25 (Note: 1 GPa =
143.7ksi.)
tension-stiffening effects and, depending on the wall thickness and reinforcement ratio, can result in load-reduction
factors exceeding 0.5. An experimental study by Vitharana
et al.28 investigated the moment-curvature response of reinforced concrete wall elements subject to applied and thermal
loading. The study concluded that the ACI 31829 Branson
formulation and the CEB-FIP MC7830 formulation provided
upper and lower bounds, respectively, for the experimentally observed moment-curvature responses. Vitharana et
al.28 also proposed a modified Branson equation that showed
good agreement with test results for wall elements subject to
a simultaneous axial force and flexural moment.
BUCKLING ANALYSIS
It has been established that thermal loading subjects
restrained tank walls to significant combined axial compression and bending. Because precast prestressed concrete tanks
are essentially shell structures, buckling stability should to
be addressed. For relatively stiff structures, linear eigenvalue buckling analysis is a technique that can be applied to
approximate the maximum load that can be sustained prior
to structural instability or collapse. The underlying assumptions of a linear eigenvalue buckling analysis are that the
linear stiffness matrix remains unchanged prior to buckling and the stress stiffness matrix is a multiple of its initial
value. Accordingly, provided the prebuckling displacements
have an insignificant influence on the structural response, the
technique can be used effectively to predict the load at which
a structure becomes unstable.
ACI Structural Journal/March-April 2014

Fig. 12Buckled shape of cylindrical shell restrained at its ends and subject to thermally induced combined axial compression
and bending.
Commercial finite element software was used to carry out
the linear eigenvalue buckling analysis. The three-dimensional (3-D) models comprised thick shell elements. The
thermally induced compressive hoop forces and vertical
bending moments, derived from the axisymmetric modeling,
were simulated using a combination of internal stress-strain
loading and externally applied radial pressure loading. The
applied loading includes a partial safety factor of 1.55 for
persistent thermal actions in accordance with EN 199031 and
EN 1991-1-5.32 For plate or shell structures, it is prudent to
include an initial geometric imperfection, as the buckling
load is often sensitive to any deviation from the true geometry. Bradshaw33 made efforts to measure concrete cylindrical shells in the field and concluded that imperfections
were observed to be as large as the shell thickness. Therefore, for the current study, an initial geometric imperfection
of the order of magnitude of the shell thickness was adopted
and was represented as out-of-roundness.
The mode of buckling obtained from the finite element
analysis is given in Fig. 12, with the same mode observed for
all tank sizes. The buckled shape displays the characteristic
sinusoidal buckle waves consistent with Koiters34 classical
linearized shell buckling theory. The mode shape is sinusoidal both axially and circumferentially.
Figure 13 presents the eigenvalues extracted from the
finite element study for various average temperatures and
H2/Dt ratios. The eigenvalues, l, are ratios of the buckling
load to the applied load. An eigenvalue equal to unity indicates that structural instability or buckling has occurred.
The lowest eigenvalue extracted from the buckling analysis
was 2.52. Thus, the ultimate limit state of buckling was not
reached for the temperature range considered.
FREE-SLIDING CONDITION
Theoretically, for a free-sliding condition, an average
temperature across the concrete section does not induce
any additional stresses. For a gradient experienced across
the wall thickness, however, associated hoop and vertical
bending stresses develop. Figure 14 is an example for a freesliding wall subject to a temperature distribution resulting
from the storage of heated liquids. The inside and outside
temperatures are taken as 95 and 80C (171 and 144F),
ACI Structural Journal/March-April 2014

Fig. 13Eigenvalues from finite element linear buckling


analysis. (Note: 1F = 1.8C + 32.)
respectively. This arrangement is slightly conservative, as
the use of sufficient external insulation would generally
result in a temperature difference between the inside and
outside faces of less than 10C (18F).
Comparing the results observed in Fig. 14 with those in
Fig. 8 and 9, it is evident that the magnitude and significance
of the internal forces are far less for a free-sliding wall. A
noteworthy observation is the hoop tension developed over
the majority of the wall height which, although not excessive in magnitude, would need to be summed to the hydrostatic hoop tension when calculating circumferential posttensioning requirements.
CONCLUSIONS AND RECOMMENDATIONS
The following conclusions and recommendations may be
drawn from the current study:

253

Brian D. ORourke is a Lecturer and Researcher in the Department of


Civil, Structural and Environmental Engineering at Cork Institute of Technology. His research interests include structural behavior and materials
technology. He is a Chartered Engineer and member of Engineers Ireland.

NOTATION

D = diameter
Ec,eff = effective or age-adjusted concrete modulus of
elasticity
Ecm = secant concrete modulus of elasticity
fck = concrete compressive cylinder strength
fcu = concrete compressive cube strength
fmax = maximum concentric precompression
stress required to eliminate tensile stresses
ft = concrete tensile strength
H = wall height
I = second moment of area
Mcr = cracking moment
R = radius
T = wall thickness
x = height from base of wall
f(,t0) = final creep coefficient
l = eigenvalue

REFERENCES

Fig. 14Internal forces resulting from thermal storage


for a free-sliding wall. (Note: 1 kNm/m = 0.225 kip-ft/ft;
1kN/m= 5.71 kip/in.)
1. For the most part, the temperature under consideration
for the present study does not have a significant adverse
effect on the material properties. The most important factors
that require consideration are creep of the concrete, bond
strength, and stress relaxation for pretensioned and nonpretensioned reinforcement. Where material properties form
inputs for analysis and design, any associated reduction
should be accounted for, particularly if unfavorable.
2. For a tank wall restrained radially at its ends, the internal
forces resulting from the storage of heated liquids have been
shown to be significant. As such, a temperature exceeding
approximately 50C (90F) across the concrete section
appears to be prohibitive based on compressive stress limits
and vertical prestressing constraints. Consequently, it is
recommended that internal insulation be provided to prevent
temperatures from exceeding this.
3. A linear eigenvalue buckling analysis has revealed that
the ultimate limit state of buckling for a wall with restrained
ends was not reached for the temperature range considered.
A minimum eigenvalue of 2.52 was observed.
4. Where provisions are made for radial displacements at
the wall ends during service, the 95C (171F) maximum
temperature does not induce excessive stresses. Complications may arise, however, surrounding possible leakage at
the joints. Accordingly, it is recommended that a polymer
liner be included, thereby eliminating concerns regarding
liquid-tightness.
AUTHOR BIOS

ACI member Michael J. Minehane is a Structural Design Engineer at RPS


Group Ltd., Cork, Republic of Ireland. He received his BEng and MEng
from Cork Institute of Technology, Republic of Ireland, in 2010 and 2011,
respectively. His research interests include prestressed concrete, strutand-tie modeling, and finite element analysis.

254

1. Priestley, M. J., Ambient Thermal Stresses in Circular Prestressed


Concrete Tanks, ACI Journal, V. 73, No. 10, Oct. 1976, pp. 553-560.
2. Ghali, E., and Elliott, E., Serviceability of Circular Prestressed
Concrete Tanks, ACI Structural Journal, V. 89, No. 3, May-June 1992,
pp. 345-355.
3. NZS 3106, Code of Practice for Concrete Structures for Retaining
Liquid, Standards Association of New Zealand, 2009, 83 pp.
4. AS 3735, Concrete Structures for Retaining LiquidsCommentary
(Supplement to AS 3735-2001), Standards Australia, 2001, 65pp.
5. EN 1992-1-2, General RulesStructural Fire Design, Brussels,
Belgium, 2004.
6. EN 1992-3, Design of Concrete Structures, Part 3: Liquid Retaining
and Containment Structures, Brussels, Belgium, 2006.
7. Fdration Internationale du Bton (fib), Model Code 2010: Volume
1 First Complete Draft, fib Bulletin 55, Lausanne, Switzerland, 2010,
317pp.
8. Mears, A. P., Long Term Tests on the Effect of Moderate Heating
on the Compressive Strength and Dynamic Modulus of Elasticity of
Concrete, Concrete for Nuclear Reactors, SP-34, C. E. Kesler, ed., American Concrete Institute, Farmington Hills, MI, 1972, pp. 355-375.
9. Komendant, J.; Nicolayeff, V.; Polivka, M.; and Pirtz, D., Effect
of Temperature, Stress Level, and Age at Loading on Creep of Sealed
Concrete, Douglas McHenry International Symposium on Concrete and
Concrete Structures, SP-55, B. Bresler, ed., American Concrete Institute,
Farmington Hills, MI, 1978, pp. 55-81.
10. Nasser, K. W., and Lohtia, R. P., Mass Concrete Properties at High
Temperatures, ACI Journal, V. 68, No. 3, Mar. 1971, pp. 180-186.
11. Carette, G. G., and Malhotra, V. M., Performance of Dolostone
and Limestone Concretes at Sustained High Temperatures, Temperature
Effects on Concrete (ASTM STP 858), T. R. Naik, ed., ASTM International,
West Conshohocken, PA, 1985, pp. 38-67.
12. Comit Euro-International du Bton (CEB), Fire Design of Concrete
Structures: in Accordance with CEB/FIP Model Code 90, Bulletin DInformation 21, Lausanne, Switzerland, 1991, 120 pp.
13. Brown, R. D., Properties of Concrete in Reactor Vessels, Conference on Prestressed Concrete Pressure Vessels, Westminster, Mar. 1968,
Paper 13, pp. 131-151.
14. Gross, H., High-Temperature Creep of Concrete, Nuclear Engineering and Design, V. 32, No. 1, Apr. 1975, pp. 129-147.
15. Nasser, K. W., and Neville, A. M., Creep of Concrete at Elevated
Temperatures, ACI Journal, V. 62, No. 12, Dec. 1965, pp. 1567-1579.
16. Hannant, D. J., Strain Behaviour of Concrete Up to 95C under
Compressive Stresses, Conference on Prestressed Concrete Pressure
Vessels, Westminster, Mar. 1968, pp. 177-192.
17. McDonald, J. E., Creep of Concrete under Various Temperature,
Moisture and Loading Conditions, Douglas McHenry International
Symposium on Concrete and Concrete Structures, SP-55, B. Bresler, ed.,
American Concrete Institute, Farmington Hills, MI, 1978, pp. 31-53.
18. Harada, T.; Takeda, J.; Yamane, S.; and Furumura, F., Elasticity
and Thermal Properties of Concrete Subjected to Elevated Temperatures, Concrete for Nuclear Reactors, SP-34, C. E. Kesler, ed., American
Concrete Institute, Farmington Hills, MI, 1972, pp. 377-406.

ACI Structural Journal/March-April 2014

19. Kagami, H.; Okuno, T.; and Yamane, S., Properties of Concrete
Exposed to Sustained Elevated Temperatures, Third International Conference on Structural Mechanics in Reactor Technology, Paper H1/5, London,
UK, 1975, pp. 1-10.
20. Haddad, R. J.; Al-Saleh, R. J.; and Al-Akhras, N. M., Effect of
Elevated Temperature on Bond between Steel Reinforcement and Fibre
Reinforced Concrete, Fire Safety Journal, V. 43, 2008, pp. 334-343.
21. Chiang, C., and Tsai, C., Time-Temperature Analysis of Bond
Strength of a Rebar after Fire Exposure, Cement and Concrete Research,
V. 33, 2003, pp. 1651-1654.
22. Baant, Z. P., and Kaplan, M. F., Concrete at High Temperature:
Material Properties and Mathematical Models, Longman Group Limited,
England, 1996, 424 pp.
23. Huang, Z., Modeling the Bond between Concrete and Reinforcing
Steel in a Fire, Engineering Structures, V. 32, 2010, pp. 3660-3669.
24. Vitharana, N. D., and Priestley, M. J., Significance of
Temperature-Induced Loadings on Concrete Cylindrical Reservoir Walls, ACI Structural Journal, V. 96, No. 5, July-Aug. 1999,
pp. 737-749.
25. EN 1992-1-1, Design of Concrete Structures, Part 1: General Rules
and Rules for Buildings, Brussels, Belgium, 2004.

ACI Structural Journal/March-April 2014

26. BS 8007, Code of Practice for Design of Concrete Structures for


Retaining Aqueous Liquids, BSI, London, UK, 1987.
27. PCI Committee on Precast Prestressed Concrete Storage Tanks,
Recommended Practice for Precast Prestressed Concrete Storage Tanks,
PCI Journal, V. 32, No. 4, 1987, pp. 80-125.
28. Vitharana, N. D.; Priestley, M. J.; and Dean, J. A., Behaviour of
Reinforced Concrete Reservoir Wall Elements under Applied and Thermally-Induced Loadings, ACI Structural Journal, V. 95, No. 3, May-June
1998, pp. 238-248.
29. ACI Committee 318, Building Code Requirements for Reinforced
Concrete (ACI 318-89) and Commentary, American Concrete Institute,
Farmington Hills, MI, 1989, 353 pp.
30. Comit Euro-International du Bton/Fdration Internationale du
Bton, Model Code for Concrete Structures, third edition, Paris, 1978,
348 pp.
31. EN 1990, Basis of Structural Design, Brussels, Belgium, 2005.
32. EN 1991-1-5, Actions on StructuresPart 1-5: General Actions
Thermal Actions, Brussels, Belgium, 2008.
33. Bradshaw, R. R., Some Aspects of Concrete Shell Buckling, ACI
Journal, V. 60, No. 3, Mar. 1963, pp. 313-328.
34. Koiter, W. T., On the Stability of Elastic Equilibrium, PhD thesis,
Technological University of Delft, the Netherlands, 1945.

255

NOTES:

256

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S22

Bond Strength of Spliced Fiber-Reinforced Polymer


Reinforcement
by Ali Cihan Pay, Erdem Canbay, and Robert J. Frosch
To provide increased insight regarding the bond behavior of
fiber-reinforced polymer (FRP) bars, 41 glass FRP, carbon FRP,
and steel reinforced concrete beams with unconfined tension lap
splices were tested. The test results are analyzed to evaluate the
influence of splice length, surface deformation, modulus of elasticity, axial rigidity, and bar casting position on bond strength.
Furthermore, the test results are compared with the current design
expression recommended by ACI Committee 440 to evaluate its
applicability. This comparison clearly indicates that the current
design expression is inadequate, and that a new design equation
is needed. More importantly, however, this research sheds light on
the importance of the axial rigidity of the reinforcement on bond
strength. Test results demonstrate that bond strength is linearly
related to the axial rigidity of the reinforcement. This finding
has future implications regarding the development of improved
design expressions and allowing for an improved understanding
of bondstrength.
Keywords: bond strength; development length; fiber-reinforced polymer
(FRP) reinforcement; reinforced concrete; splice length.

INTRODUCTION
Fiber-reinforced polymer (FRP) reinforcement can
provide an alternative solution for structures susceptible to
corrosion and where low electric conductivity or magnetic
transparency is required. FRP bars, however, are anisotropic,
have different physical and mechanical properties than that
of steel reinforcement, and remain linear-elastic until failure.
The modulus of elasticity of glass and aramid FRP bars are
approximately one-fifth that of steel. Although carbon FRP
(CFRP) bars have a higher modulus than glass FRP (GFRP)
bars, their modulus is approximately two-thirds that of steel
reinforcing bars. Consequently, design procedures used for
steel reinforced members are not necessarily applicable for
FRP reinforced structures.
Because the physical and mechanical properties of FRP
bars are different from those of steel reinforcement, especially the surface deformation and the modulus of elasticity
of the reinforcement, the bond behavior of FRP reinforced
concrete specimens is expected to be quite different than
that of steel reinforced specimens. An early study on the
first generation of FRP reinforcement by Fish (1992) made
this fact apparent. The difference in bond strength on
modern FRP reinforcement is clearly evident from the tests
completed by Mosley et al. (2008), which found that glass
(GFRP) and aramid (AFRP) bars achieved approximately
50% of the stress developed by steel reinforcement for the
same bar size and splice length. Based on these tests, among
others, provisions for the development and splices of FRP
reinforcement were developed (Wambeke and Shield 2006)
ACI Structural Journal/March-April 2014

and incorporated into ACI 440.1R-06 (ACI Committee 440


2006). While 240 specimens were included in the study by
Wambeke and Shield, the majority of the tests had relatively
short development lengths, with l/db less than 30, and only
20splice tests (14 confined splices by Tighiouart et al. [1999]
and six unconfined splices by Mosley et al. [2008]) were
included. Splice tests are the preferred test method for the
determination of development lengths, but only a few were
available at the time. Splice tests are preferred because they
provide a realistic stress-state in the vicinity of the bars
(ACI Committee 408 2003). It is for this reason that they are
used as the basis of the ACI 318-11 (ACI Committee 318
2011) development length expressions.
While knowledge regarding the bond strength of FRP
reinforcement is developing, there is a scarcity of data
available from splice tests, particularly unconfined splices
that are commonly used in FRP applications. In addition, a
number of questions that were highlighted by Mosley et al.
(2008) remain unanswered. In particular, the study recommended evaluating the effect of longer splice lengths for
FRP reinforcement, as the maximum splice investigated was
18 in. (457 mm) for a No. 5 (15.9 mm) bar (l/db = 28.8).
The previous research indicated that bond strength is proportional to the square root of the development length, but it is
not clear if this trend continues as the development length
increases. Furthermore, as only No. 5 bars were tested, it is
not clear if the findings are appropriate for other bar sizes.
This study also found that the surface deformation of the
FRP does not significantly affect bond strength. Different
deformations are now available that can shed further light
on this subject. Finally, the study found that bond strength
is related to the modulus of elasticity of the reinforcement based on tests of steel (E = 29,000 ksi [200 GPa]),
GFRP (E 5500 ksi [37.9GPa]), and AFRP (E 6800 ksi
[46.8GPa]). Carbon fiber bars are now available with E
that falls between that of steel and GFRP (E 20,000 ksi
[138 GPa]), which may provide additional understanding
regarding this behavior.
RESEARCH SIGNIFICANCE
The objective of this research study is to provide additional experimental data from splice tests to improve understanding of the bond strength between FRP reinforcement
and concrete. Of particular interest is the influence of a
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-069.R1, doi:10.14359.51686519, was received June 19, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

257

Table 1Specimen details

Fig. 1Typical test specimen.

Fig. 2Cross section detail at splice region.


number of variables, including splice length, bar size,
reinforcement modulus of elasticity, and reinforcement
deformation type. A secondary objective is to evaluate the
applicability of current design expressions for the bond
strength of FRP reinforcement to determine if the current
expressions can be reliably used for extended splice lengths
and bar sizes considering the limited range for which the
expressions weredeveloped.
SPECIMEN DESIGN
Five series of beams were tested to evaluate the bond
strength. The specimens were designed to provide a systematic evaluation of the primary variables and were reinforced
with either GFRP, CFRP, or steel bars. Both No. 5 and 8
(15.9 and 25.4 mm) bars were also considered, with the
exception of carbon bars, for which No. 8 bars were not
available. Splice lengths were considered from 12 in.
(305mm) up to 54 in. (1372 mm). A typical test specimen is
illustrated in Fig. 1, while dimensions for each specimen are
provided in Table 1. Cross-sectional details of the specimens
are shown in Fig. 2.
All beams were rectangular in cross section, with a total
depth of 16 in. (406 mm). Specimens were designed with a
clear spacing of 1 in. (25.4 mm) between the bars located in
the splice region and with a 1.5 in. (38.1 mm) side and top
clear cover. This limitation represents the minimum clear
spacing and minimum clear cover allowed by ACI318-11.
Three reinforcing bars were spliced at the center of the
constant moment region of the beam. The width of the specimens was controlled by the minimum cover and spacing
limitations and the size of the reinforcement; therefore, the
width of specimens was 8.75 in. (222 mm) for the specimens
reinforced with No. 5 (15.9 mm) bars, and 11 in. (279mm)
258

LS, in. (mm)

L, ft (m)

LV, ft (m)

LM, ft (m)

12, 18, 24
(305, 457, 610)

13.5 (4.11)

3 (0.91)

6 (1.83)

36 (914)

18 (5.49)

3.75 (1.14)

9 (2.74)

54 (1372)

18 (5.49)

3.75 (1.14)

8.5 (2.59)

for the specimens reinforced with No. 8 (25.4mm) bars.


Concrete cover and spacing of the reinforcing bars were
maintained constant throughout the experimental program,
and no transverse reinforcement was provided in the constant
moment region. Two No. 3 (9.53 mm) longitudinal steel
bars were provided in the compression side of the beam to
prevent collapse at failure. In addition, the shear spans were
reinforced with No. 3 (9.53 mm) steel stirrups to prevent
shear failure before bond failure.
In addition to the primary variables, several other variables were considered. While previous tests clearly indicate that the modulus of elasticity is a primary parameter,
it is not clear if this is solely due to the modulus E or due
to the axial rigidity AE of the reinforcement. Therefore, a
specimen was designed where the E of the reinforcement
remains the same using the same reinforcement material, but
the cross-sectional area A is reduced. To maintain the same
surface area and deformation pattern, a hollow bar was used
to reduce A. In addition, top and bottom-cast specimens were
constructed to evaluate the influence of the casting position
and allow for linkage between top and bottom-cast results.
In the top-cast position, the specimens qualify as top bar
if more than 12 in. (305 mm) of fresh concrete is cast below
the reinforcement.
In total, the experimental program consisted of 41 reinforced concrete beam specimens. Details of the specimens
are summarized in Table 2. A four-part notation system is
used to identify the specimens. The specimens were identified first by the descriptive label B (bond) followed by the
reinforcement type (Tables 3 and 4), the bar size, and finally
by the splice length. For example, B-HC-5-12 stands for a
bond test with No. 5 (15.9 mm) CFRP bars spliced at 12 in.
(305 mm). Bottom-cast specimens were identified by adding
the notation b to the splice length.
MATERIALS
FRP reinforcement
FRP reinforcement included glass bars (No. 5 and 8
[15.9and 25.4 mm]) and carbon bars (No. 5 [15.9 mm]) with
different surface deformations. Three types of bar surfaces
were considered to evaluate the effect of varying surface
deformations, including: 1) sand coating; 2) wrapped indentations and sand coating; and 3) fabric texture. For glass bars,
all three surface types were included. For carbon bars, only
the sand coating and fabric texture surfaces were possible
because the wrapped indentation significantly reduces the
tensile strength of the bar, which makes this surface treatment impractical. A general view of the bar surfaces is
shown in Fig. 3. As shown, the same surface type (sand
coated and fabric texture) was obtained for both types of
ACI Structural Journal/March-April 2014

Table 2Specimen details and test results


Series

II

III

IV

Bar size Ls, in. (mm) Casting position Test age, days Ptest, kip (kN) ftest, ksi (MPa) avg, psi (MPa)

Specimen

Bar type

Deformation

B-S1-8-18

Steel

Deformed

No. 8

18 (457)

Top

28

33.0 (147)

40.8 (281)

570 (3.93)

B-PG-8-18

Glass

Sand-coated

No. 8

18 (457)

Top

29

24.1 (107)

27.9 (192)

390 (2.69)

B-HG-8-18

Glass

Sand and
wrapped

No. 8

18 (457)

Top

31

20.5 (91.2)

23.7 (163)

332 (2.29)

B-HG1-5-18

Glass

Sand and
wrapped

No. 5

18 (457)

Top

32

14.2 (63.2)

40.7 (281)

357 (2.46)

B-HGO-5-18

Glass

Sand and
wrapped

No. 5

18 (457)

Top

35

11.5 (51.2)

32.9 (227)

289 (1.99)

B-PG-5-18

Glass

Sand-coated

No. 5

18 (457)

Top

36

16.5 (73.4)

47.3 (326)

415 (2.86)

B-S1-5-18

Steel

Deformed

No. 5

18 (457)

Top

37

24.1 (107)

72.2 (498)

633 (4.36)

B-HC-5-18

Carbon

Fabric texture

No. 5

18 (457)

Top

38

19.9 (88.5)

58.8 (405)

515 (3.55)

B-HC-5-12

Carbon

Fabric texture

No. 5

12 (305)

Top

28

15.1 (67.2)

44.5 (307)

585 (4.03)

B-S1-8-36

Steel

Deformed

No. 8

36 (914)

Top

31

37.1 (165)

57.2 (394)

400 (2.76)

B-PG-8-36

Glass

Sand-coated

No. 8

36 (914)

Top

32

19.9 (88.5)

28.9 (199)

202 (1.39)

B-HG-8-36

Glass

Sand and
wrapped

No. 8

36 (914)

Top

34

21.0 (93.4)

30.4 (210)

212 (1.46)

B-HG1-5-36

Glass

Sand and
wrapped

No. 5

36 (914)

Top

35

12.4 (55.2)

44.3 (305)

194 (1.34)

B-HGO-5-36

Glass

Sand and
wrapped

No. 5

36 (914)

Top

38

13.3 (59.2)

47.5 (328)

208 (1.43)

B-PG-5-36

Glass

Sand-coated

No. 5

36 (914)

Top

40

13.9 (61.8)

49.9 (344)

219 (1.51)

B-HC-5-36

Carbon

Fabric texture

No. 5

36 (914)

Top

42

22.9 (102)

84.6 (583)

371 (2.56)

B-S2-8-12

Steel

Hollow
Deformed

No. 8

12 (305)

Top

133

18.1 (80.5)

29.7 (205)

465 (3.21)

B-S1-8-12

Steel

Deformed

No. 8

12 (305)

Top

130

21.9 (97.4)

27.3 (188)

571 (3.94)

B-PG-8-12

Glass

Sand-coated

No. 8

12 (305)

Top

104

17.1 (76.1)

20.0 (138)

418 (2.88)

B-HG-8-12

Glass

Sand and
wrapped

No. 8

12 (305)

Top

106

14.0 (62.3)

16.3 (112)

341 (2.35)

B-S1-8-12b

Steel

Deformed

No. 8

12 (305)

Bottom

111

21.2 (94.3)

26.3 (181)

551 (3.80)

B-HG-8-12b

Glass

Sand and
wrapped

No. 8

12 (305)

Bottom

124

14.5 (64.5)

16.9 (117)

354 (2.44)

B-PG-8-12b

Glass

Sand-coated

No. 8

12 (305)

Bottom

126

15.8 (70.3)

18.4 (127)

385 (2.65)

B-S2-5-24

Steel

Deformed

No. 5

24 (610)

Top

129

23.5 (105)

70.9* (489)

467 (3.22)

B-HC-5-24

Carbon

Fabric texture

No. 5

24 (610)

Top

132

21.9 (97.4)

64.7 (446)

426 (2.94)

B-PC-5-24

Carbon

Sand-coated

No. 5

24 (610)

Top

139

24.1 (107)

71.8 (495)

472 (3.25)

B-HG1-5-24

Glass

Sand and
wrapped

No. 5

24 (610)

Top

142

13.6 (60.5)

39.0 (269)

256 (1.77)

B-HG2-5-24

Glass

Fabric texture

No. 5

24 (610)

Top

148

16.5 (73.4)

47.6 (328)

313 (2.16)

B-PG-5-24

Glass

Sand-coated

No. 5

24 (610)

Top

153

16.7 (74.3)

48.0 (331)

316 (2.18)

B-HG1-5-24b

Glass

Sand and
wrapped

No. 5

24 (610)

Bottom

155

14.7 (65.4)

42.2 (291)

277 (1.91)

B-PG-5-24b

Glass

Sand-coated

No. 5

24 (610)

Bottom

161

17.7 (78.7)

50.8 (350)

334 (2.30)

B-HG1-5-12

Glass

Sand and
wrapped

No. 5

12 (305)

Top

156

9.5 (42.3)

27.4 (189)

361 (2.49)

B-PG-5-12

Glass

Sand-coated

No. 5

12 (305)

Top

160

10.6 (47.2)

30.4 (210)

400 (2.76)

B-HG-8-24

Glass

Sand and
wrapped

No. 8

24 (610)

Top

161

20.8 (92.5)

24.1 (166)

253 (1.74)

B-HG-8-54

Glass

Sand and
wrapped

No. 8

54 (1372)

Top

175

22.8 (101)

33.0 (228)

154 (1.06)

B-HG1-5-54

Glass

Sand and
wrapped

No. 5

54 (1372)

Top

177

14.1 (62.7)

50.4 (347)

148 (1.02)

B-PG-5-54

Glass

Sand-coated

No. 5

54 (1372)

Top

181

14.1 (62.7)

50.6 (349)

148 (1.02)

B-HC-5-54

Carbon

Fabric texture

No. 5

54 (1372)

Top

183

22.9 (102)

85.0 (586)

249 (1.72)

B-HG1-5-12b

Glass

Sand and
wrapped

No. 5

12 (305)

Bottom

162

12.1 (53.8)

34.8 (240)

458 (3.16)

B-PG-5-12b

Glass

Sand-coated

No. 5

12 (305)

Bottom

164

13.6 (60.5)

39.0 (269)

514 (3.54)

Glass

Sand and
wrapped

No. 8

24 (610)

Bottom

169

23.0 (102)

26.7 (184)

280 (1.93)

B-HG-8-24b
Reinforcement yielded.

ACI Structural Journal/March-April 2014

259

Table 3Mechanical properties of fiber-reinforced polymer bars


Bar type

Producer

Producer 1
Glass

No. 5

Designation

Surface deformation

Er, ksi (GPa)

fu, ksi (MPa)

HGO

Sand and wrapped

5800 (40.0)

71 (490)

HG1

Sand and wrapped

6400 (44.1)

98 (676)

HG2

Fabric texture

7300 (50.3)

115 (793)

No. 8

HG

Sand and wrapped

5700 (39.3)

76 (524)

No. 5

PG

Sand

6400 (44.1)

89 (614)

No. 8

PG

Sand

6200 (42.7)

76 (524)

Producer 1

No. 5

HC

Fabric texture

18500 (127.6)

129 (889)

Producer 2

No. 5

PC

Sand

21700 (149.6)

Producer 2

Carbon

Bar size

Coating of bar peeled at anchor at 100 ksi (689 MPa).

Table 4Mechanical properties of steel bars


Bar size
No. 5

No. 8

Fig. 3Carbon, glass FRP, and steel bars.


FRP reinforcement (glass and carbon), allowing for evaluation of the influence of the material type independent of the
surface deformation.
Tensile tests on representative coupons were performed
for each type of reinforcement to determine their mechanical
properties. Coupons for FRP bars were tested considering the
requirements of ACI 440.3R-04 (ACI Committee 440 2004).
The measured modulus of elasticity and ultimate strength of
the FRP bars are provided in Table 3. Two types of sand and
wrapped bars were tested: HGO and HG1. The designation
HGO represents older GFRP bars from the same manufacturer that were previously tested in the study by Mosley et
al. (2008). Over time, the surface deformation on the bars
has changed slightly; therefore, two different configurations
of the same surface treatment were considered. In addition,
these bars allow for comparison with the earlier study.
Steel reinforcement
Deformed steel reinforcement consisted of No. 5 and 8
(15.9 and 25.4 mm) bars, meeting ASTM A615 Grade 60.
To evaluate the effect of axial rigidity, a No. 8 (25.4 mm)
hollow reinforcing bar was constructed by drilling a 0.5 in.
(12.7 mm) diameter hole through a 16 in. (406 mm) length
of the deformed bar to reduce the cross-sectional area in the
splice region, as shown in Fig. 3. Bars of each size were
obtained from the same heat to ensure consistent reinforcement material properties in each phase. Table 4 presents the
properties of the steel bars.

260

Designation

Bar type

fy, ksi
(MPa)

fu, ksi
(MPa)

S1

Deformed

75 (517)

95 (655)

S2

Deformed

60 (414)

100 (689)

S1

Deformed

76 (524)

97 (669)

S2

Hollow
deformed

76 (524)

97 (669)

Concrete
The same mixture proportion was used for all test series.
The concrete used a coarse aggregate consisting of river
gravel with a 0.75 in. (19 mm) maximum aggregate size.
Batch weights and slump for each series are provided in
Table 5. Concrete compressive and splitting tensile strengths
were obtained from the average of three 6 x 12 in. (152 x
305mm) cylinders, and are also provided in Table 5.
Concrete material tests were timed with the testing program
such that results were obtained on the first, middle, and
last day of specimen testing for each series. The concrete
strengths reported are the average over the days of testing
for each test series. As noted in Table 2, which provides the
concrete age on the day of testing, specimens in each series
were tested within a fairly short timeframe. Therefore, the
variation of concrete strengths from the average during the
duration of testing was within 3% for each series. Across all
series, concrete compressive strength varied from 4010 to
5470 psi (27.7 to 37.7 MPa), even though the same concrete
mixture was ordered.
CONSTRUCTION
Specimens in each series were cast at the same time from
the same batch of concrete. The concrete was placed in
the forms in two layers, and each layer was vibrated using
mechanical vibrators. The beams were screeded, and the
surface was finished with a magnesium float. The beams
were covered with wet burlap, and plastic sheets were placed
on top of the burlap to prevent moisture loss before final set.
For each series, cylinders were cast simultaneously with the
beams. The cylinders were consolidated, cured, and stored
in the same manner as the test specimens.

ACI Structural Journal/March-April 2014

Table 5Concrete mixture and mechanical properties


Material

Series I

Series II

Series III

Series IV

Series V

Cement Type I, lb/yd (kg/m )

425 (252)

430 (255)

429 (255)

430 (255)

428 (254)

Fine aggregate, lb/yd (kg/m )

1651 (980)

1591 (944)

1611 (956)

1550 (920)

1609 (955)

Coarse aggregate, lb/yd (kg/m )

1847 (1096)

1849 (1097)

1842 (1093)

1850 (1098)

1842 (1093)

Water, lb/yd3 (kg/m3)

163 (97)

145 (86)

196 (116)

240 (142)

184 (109)

Air, oz/yd3 (mL/m3)

1.1 (43)

1.1 (43)

Water reducer, oz/yd3 (mL/m3)

8.7 (336)

7.8 (302)

6.5 (251)

1.5 (58)

6.4 (247)

Slump, in. (mm)

4 (102)

5 (127)

5.5 (140)

3 (76)

4.5 (114)

fc, psi (MPa)

5260 (36.3)

5470 (37.7)

4010 (27.7)

4640 (32.0)

4170 (28.8)

ft, psi (MPa)

590 (4.07)

520 (3.58)

380 (2.62)

440 (3.03)

430 (2.96)

Casting position

ftest, ksi (MPa)

Ratio, top/bottom

Top

30.4 (210)

Bottom

39.0 (269)

Top

27.4 (189)

Bottom

34.8 (240)

Top

48.0 (331)

Bottom

50.8 (350)

Top

39.0 (269)

Bottom

42.2 (291)

Top

20.0 (138)

Bottom

18.4 (127)

Top

16.3 (112)

Bottom

16.9 (117)

Top

27.3 (188)

Bottom

26.3 (181)

Top

24.1 (166)

Bottom

26.7 (184)

Table 6Influence of casting position


Bar size

Splice length

Bar type
PG

12 in.
(305 mm)
HG1
No. 5
PG
24 in.
(610 mm)
HG1

PG
12 in.
(305 mm)

HG

No. 8
S1
24 in.
(610 mm)

HG

Average

TEST SETUP AND PROCEDURE


Beams were placed on two supports, and two equal,
concentrated loads were applied at the end of the cantilever
with hydraulic rams, creating a constant moment region
between the supports as shown in Fig. 1. The rams were
connected to a single hydraulic hand pump to obtain equal
pressure in each ram. Load was applied in 0.5 kip (2.22 kN)
increments for the specimens with No. 5 (15.9 mm) bars,
and 1 kip (4.45 kN) increments for the specimens with No.8
(25.4 mm) bars. At each load stage, the crack pattern was
mapped, and crack widths were measured on the beam top
surface. Cracks were mapped and measured up to a critical
load, beyond which it was considered unsafe to approach
the beam.
Displacements at the ends, supports, and midspan were
monitored with displacement transducers, while loads were
monitored using load cells. In Series I and II, three strain
gauges were placed on the reinforcing bars at the ends of the
splice region. Two were attached to the middle reinforcing
ACI Structural Journal/March-April 2014

0.78

0.79

0.94

0.92

1.09

0.96

1.04

0.90
0.93

bars, and one was attached to the outer reinforcing bar. The
strains measured with strain gauges and strains calculated
based on flexural theory agreed well; therefore, no strain
gauges were installed on the FRP reinforcing bars for the
remaining series. Strain gauges, however, were installed on
all steel reinforced specimens where there was a possibility
of yielding the reinforcement.
STRUCTURAL BEHAVIOR
Specimens with the same beam width and shear span in
a given series cracked at approximately the same load. The
stiffness of the specimens was approximately the same up to
the cracking load. Flexural cracks usually first occurred at the
support or simultaneously at the support and in the constant
moment region. As loading increased, further cracks formed
within the constant moment region, shear span, and splice
region. All specimens failed in a brittle side-splitting mode
in the splice region. Two different types of side splitting
were observed during failure. In the first type, the concrete
261

Fig. 4Splitting failure.


cover in the splice region exploded. In the second case, the
failure was not explosive, and the cover remained intact
with the reinforcement. Photographs captured at the time of
failure that illustrate the failure types are shown in Fig.4.
The splitting plane observed after failure indicated that
failure typically initiated from the splitting cracks present
on the side face. No damage to the surface deformations was
observed on any of the reinforcing bars.
To further illustrate the response, the applied load versus
end deflection curves for Series I are presented in Fig. 5.
Behavior of the specimens can be described by three distinct
stages. In the first stage, before flexural cracking, the
load-deflection curves are linear and the slopes are approximately identical, indicating that before cracking, the stiffness of the specimens is primarily controlled by the concrete.
All specimens in a given series cracked at approximately the
same load, with slightly higher loads achieved for the stiffer
bars. In the second stage, after flexural cracking, the slope
reduces; however, the response remains essentially linear up
to failure. In this stage, the flexural stiffness is a function of
the modulus of elasticity and cross-sectional area of the bars.
Bars with a lower modulus of elasticity resulted in a lower
flexural stiffness. In general, beams reinforced with steel
bars had the highest stiffness, followed by beams reinforced
with CFRP bars, and then by GFRP bars. Beams reinforced
with different types of GFRP bars have approximately the
same stiffness due to their similar moduli of elasticity. In the
final stage, all specimens failed suddenly by splitting of the
concrete in the splice region. The same observations were
made for the specimens tested in the other series (SeriesII
to V). Load versus deflection plots for all specimens are
presented in Pay (2005).
In each series, among the specimens with the same
cross-sectional dimensions and shear span length, steel
reinforced specimens reached the highest load, followed
by CFRP and then GFRP reinforced specimens. Specimens
reinforced with GFRP deflected most, followed by CFRP
and steel reinforced specimens. Based on observation, the
modulus of elasticity of the reinforcing bar was directly
proportional to the failure load, and inversely proportional
to the deflection at the time of failure.

262

Fig. 5Load-deflection curves of Series I specimens with


No. 5 bars.
BOND STRENGTH
The maximum applied load Ptest at the ends of the cantilever (Fig. 1) and computed reinforcement stress reached at
failure ftest for each specimen are provided in Table 2. The
reinforcement stress at failure was calculated using both
cracked section analysis and moment-curvature analysis.
For moment-curvature analysis, the Hognestad stress-strain
curve was used, and the tensile strength of the concrete was
neglected. Values from both analyses were approximately
the same; therefore, stresses from the crack section analysis
are presented herein. The average bond stress avg was calculated assuming that the tension force in the bar is resisted
by a uniform distribution of stress along the surface of the
splice. Nominal cross-sectional dimensions were used in all
calculations. It should be noted that the steel reinforcement
in Specimen B-S2-5-24 yielded before splitting failure;
therefore, its results will not be considered in future analyses.
The variables investigated in this study are evaluated in
the following sections. To eliminate the effect of variations
in the concrete strength, bar stresses and forces are presented
normalized by the fourth root of the concrete compressive
strength. As presented in ACI 408R-03 (ACI Committee 408
2003) and Canbay and Frosch (2005), the influence of the
compressive strength on bond strength is best represented
by the fourth root. Considering the differences in concrete
compressive strength in this study, the square root tradi-

ACI Structural Journal/March-April 2014

Fig. 6Effect of splice length on bond strength (No. 5 bars).

Fig. 7Effect of splice length on bond strength (No. 8 bars).

tionally used to represent tensile strength would produce


similarfindings.

and wrapped bars for No. 5 (15.9 mm) glass bars can be seen
in Fig. 6. For this bar size, the results of the sand-coated bars
and the wrapped and sand-coated bars were similar, with
the sand-coated bars reaching slightly higher bond stresses
than the wrapped and sand-coated bars except for the 54in.
(1372 mm) splice specimens, which failed at approximately
the same stress. In addition to these commercially available
reinforcing bars, No. 5 (15.9 mm) fabric texture glass bars
were specifically produced for this test program to evaluate
the effect of the bar surface. This bar type was tested using a
24 in. (610 mm) splice (B-HG2-5-24) and reached a normalized stress of 45.9 ksi (316 MPa), which is essentially the
same as that achieved with the companion sand-coated bar
(46.3 ksi [319 MPa] for B-PG-5-24). Therefore, GFRP bars
with a fabric surface texture that is considerably smoother
were capable of reaching stresses as high as the sand-coated
GFRP bar. In considering the No. 8 (25.4 mm) bar reinforced
specimens (Fig. 7), the same trend is apparent where the
sand-coated bars provided similar bond stresses to the sand
and wrapped bars. Slightly higher bond stresses were developed with the sand-coated bars for the shorter splices, with
approximately the same stress for the longer 36 in. (914 mm)
splice. Overall, the sand-coated GFRP bars were observed
to reach slightly higher stresses for the shorter splice lengths
among the deformation types tested in the experimental
program. Considering the minor differences in test results
and the variations expected in bond tests, however, the variations in surface deformation produced little difference in
bond strength.

Splice length
The influence of splice length on bond strength was evaluated among specimens with the same bar size and surface
type for splice lengths ranging from 12 to 54 in. (305 to
1372mm), and is presented in Fig. 6 for the No. 5 (15.9mm)
specimens and Fig. 7 for the No. 8 (25.4 mm) specimens.
Best-fit power trend lines are also provided to illustrate the
trends of the data. As shown, bar stresses reached at failure
increase as the splice length increases. The effectiveness
of increasing the splice, however, decreases as the length
increases, as evidenced by the decreasing slope. In addition, the slope of the curve, which indicates the strength
gain provided by increasing the splice length, is different
for each reinforcement type. The strength gain for the steel
and carbon bars as the splice length increases is significantly
greater than that for the glass bars. For example, doubling
the splice length of the No. 5 (15.9 mm) glass bars from
18to 36 in. (457 to 914 mm) increased the stress by only 6%,
while the same increase in splice length for the carbon bars
resulted in a 43% increase in bar stress. In previous studies,
the influence of splice length for steel (Canbay and Frosch
2005) and short FRP splices (Mosley et al. 2008) was found
to be proportional to the square root; therefore, increasing
the splice length from 18 to 36 in. (457 to 914 mm) results in
a 41% increase in bar stress. While the square root is reasonable for the carbon bars, it significantly overestimates the
increase for the glass reinforcement, and is not appropriate.
Based on these results, the effect of splice length on the
ultimate stress reached by the reinforcement appears to be
a function of the modulus of the elasticity of the reinforcement. The benefits of an increase in splice length decrease
as the modulus of elasticity is decreased. As previously
discussed, the bar stress at failure increases as the modulus
of elasticity increases.
Surface deformation
Three types of surface deformations induced on GFRP
bars were tested to evaluate the effect of surface deformation
on splice strength. A comparison of sand-coated versus sand

ACI Structural Journal/March-April 2014

Modulus of elasticity
The effect of the modulus of elasticity of the reinforcement was investigated among the specimens having the
same surface deformation and bar size. Figure 8 shows
the normalized bar stress versus modulus of elasticity
for No.5 (15.9 mm) bars with a 24 in. (610 mm) splice.
Although two different surface deformations are considered, the data points follow a linear trend as the modulus of
elasticity increases. Clearly, the modulus of elasticity of the
reinforcement has a significant influence on the bond strength
of the reinforcement, with bond strength increasing as the
modulusincreases.

263

Fig. 8Effect of modulus of elasticity on bond strength.


(Note: 1 in. = 25.4 mm.)
Axial rigidity
Axial rigidity of the bar is calculated by multiplying the
nominal cross-sectional area of the bar A and the modulus
of the elasticity of the reinforcement E. The effect of axial
rigidity for No. 5 (15.9 mm) bars is illustrated in Fig. 9,
where the normalized bar force at failure is plotted versus
the axial rigidity AE for the various splice lengths. As shown,
there is an approximately linear trend between axial rigidity
and bar force. The effect of axial rigidity for No. 8 (25.4 mm)
bars is shown in Fig. 10. Carbon bars were not available in
this bar size to enable three points to be plotted across the
horizontal axis; therefore, only two points are available for
the 18 and 36in. (457 and 914 mm) splice lengths. For the
12in. (305 mm) splice, however, a specimen (B-S2-8-12)
was constructed that contained the hollow deformed steel
reinforcing bar in addition to a specimen containing the
identical bar that was not hollowed out (B-S1-8-12). The
bars in both specimens have the same modulus of elasticity,
surface area, and deformation pattern. Due to the reduction in
cross-sectional area, however, the axial rigidity was reduced.
As shown in Fig. 10, the hollow steel bar reached a lower bar
force than that of the solid bar, which indicates the importance
of axial rigidity on splice strength. Furthermore, the influence
of axial rigidity is shown to again be approximately linear.
Considering these results, the axial rigidity of the reinforcement rather than the modulus of elasticity of the reinforcement alone is a primary factor influencing splicestrength.
Bar casting position
Eight bottom-cast specimens were tested along with
eight companion top-cast specimens to determine the effect
of casting position on the behavior of specimens with
lap-spliced reinforcement and provide connection of the
results of the top bar specimens with the large body of test
results that exist for bottom-cast lap-splice specimens. Based
on the test results (Table 6), the bond strength of top-cast
specimens is generally lower than that of the bottom-cast
specimens, as expected. It should be noted, however, that
the reinforcement in two of the top-cast specimens reached
higher bar stresses than the companion bottom-cast specimens even though the slump of the concrete in the series in
which they were cast (Series III) was the highest. Research
by Ferguson and Thompson (1962), Jirsa et al. (1982), and
264

Fig. 9Effect of axial rigidity on bond strength (No. 5 bars).


(Note: 1 in. = 25.4 mm.)

Fig. 10Effect of axial rigidity on bond strength (No. 8


bars). (Note: 1 in. = 25.4 mm.)
DeVries et al. (2001) demonstrates that the influence of
casting position on bond strength is primarily affected by
concrete slump and bleeding; therefore, this behavior was
unexpected. Furthermore, the steel reinforced specimens
(No. 8 [25.4 mm] with 12 in. [305 mm] splice) produced
essentially the same ratio as the companion FRP reinforced
specimens. Regardless, in comparing the eight companion
top and bottom-cast specimens, an average reduction in
strength of 7% was observed. The most significant reduction
(22%) was only observed for the No. 5 (15.9 mm) specimens
with a 12 in. (305 mm) splice. As the splice length increased
to 24 in. (610 mm), the ratio decreased resulting in only a
7% strength reduction. Interestingly, No. 8 (25.4 mm) specimens with a 12 in. (305 mm) splice produced essentially no
difference in strength. Based on the tests conducted herein,
the top-cast bar specimens produced only a minor reduction
in strength. Therefore, the test results from the top-cast specimens can be compared directly with existing test data from
bottom-cast test specimens. In the worst case, they will only
be slightly conservative.
DESIGN EVALUATION
To evaluate the bond strength of the FRP reinforced
specimens, the data was analyzed considering the current
ACI440.1R-06 design recommendation where the bar stress
ffe can be computed according to Eq. (1). As noted, the term
C/db should not be taken larger than 3.5. For the specimens
ACI Structural Journal/March-April 2014

Fig. 11Comparison of strength calculations of CFRP.

Fig. 12Comparison
No.5GFRP.

of

strength

calculations

of

Fig. 13Comparison
No.8GFRP.

of

strength

calculations

of

tested in the experimental program conducted herein, C/db


is always less than 3.5, and this limit does not control. The
test results of the FRP specimens were not evaluated with
ACI318-11 or ACI 408R-03 recommendations because
these expressions were derived for specimens reinforced
with steel deformed bars and, as previously discussed by
Mosley et al. (2008), are not applicable
f fe =

fc

l
C le
C
13.6 e +
3.5 (1)
+ 340 where

a
db db db
db

where C is the lesser of the cover to the center of the bar or


one-half the center-to-center spacing of the bars being developed, in.; le is embedded length of reinforcing bar, in.; and
is the top bar modification factor (1.5 for reinforcement
placed so that more than 12 in. [305 mm] of fresh concrete
is cast below the development length or splice; 1.0for
otherreinforcement).
The calculated reinforcement stresses at failure were
compared with the experimental results. The ratio of the
experimental to calculated stresses for each reinforcement
type is illustrated in Fig. 11 through 13. The equations were
evaluated with and without the bar location factor. It should
be noted that the bar location factor for FRP reinforcement
is 1.5 according to ACI 440.1R-06, whereas a factor of 1.3 is
currently used by ACI 318-11 and ACI 408R-03. The top bar
factor of 1.5 was recommended in the work by Wambeke
and Shield (2006) in considering a comparison of eight top
bar tests relative to the bulk of the data (75 splitting failures
in total). Unfortunately, no comparison tests were available
to directly evaluate the top bar effect. Based on comparisons conducted in this research program for specimens with
a depth of 16 in. (406 mm), the maximum top bar effect
developed was 1.28, with an average of 1.08. Therefore, the
1.5 factor appears overly conservative. This difference in
results may be explained considering that six of the eight
top bar tests evaluated by Wambeke and Shield (2006) were
from splice specimens, while the majority of the bottom bar
data to which they were compared were from beam end and
notched beam tests, which typically produce higher bond
strengths. Therefore, the ratio resulting from this analysis
ACI Structural Journal/March-April 2014

can be expected to be higher than from a direct comparison


of splice results. In addition, as noted previously, identical
specimens with the only variable being casting position were
not available. Therefore, other parameters potentially influenced the comparison.
As shown in Fig. 11, for the CFRP reinforced specimens,
the ratio of the experimental to calculated stress ranges from
0.8 to 1.15 when the bar location factor is not considered.
With the bar location factor, the equation provides conservative results, and ranges from 1.21 to 1.73. The conservatism
decreases, however, as the splice length increases. While
ACI 440.1R-06 does not indicate that Eq. (1) does not apply
for carbon reinforcement, the expression was developed
from the results of only GFRP bars.
Although the ACI 440.1R-06 equation was derived from
a database of GFRP bars, Fig. 12 and 13 illustrate that the
equation provides unsafe results for the GFRP reinforced
specimens in this investigation even with the inclusion
of the bar location factor ( = 1.5). The unconservatism
increases as the splice length increases from 12 to 54 in.
(305 to 1372mm). In the case of bottom-cast specimens,
the experimental results are as low as 52% of the calculated
values for the No. 8 (25.4 mm) bars. Based on comparisons
of the experimental and calculated results, even with the
265

recommended bar location factor of 1.5, the ACI 440.1R-06


expression produces significantly unconservative results for
the GFRP bar reinforced specimens tested in this experimental program. The level of unconservatism varies with the
splice length, the reinforcement location (if location factor is
included), bar size, and bar type. The ratio of experimental
to calculated results is also observed to generally decrease as
the splice length increases.
CONCLUSIONS
Based on the results of this research program, the following
conclusions were made:
1. Bond strength is a function of the modulus of elasticity
of the reinforcement. As the modulus of elasticity increases,
the bond strength linearly increases;
2. Bond strength is a function of the axial rigidity AE of the
reinforcement. As the axial rigidity increases, the bond strength
linearly increases. This relationship suggests that development
of a unified design approach for the development of reinforcement, regardless of reinforcement type, ispossible;
3. Bond strength increases with increasing splice length;
however, this relationship is nonlinear. While previous
research on relatively short splices (Mosley et al. 2008)
supported a square root relationship between bond strength
and splice length for FRP reinforcement consistent with that
observed for steel reinforcement (Canbay and Frosch 2005),
the longer splice lengths tested indicate that this relationship
is not constant, and depends on the modulus of elasticity of
the reinforcement. For GFRP reinforcement, the relationship
is significantly lower than the 0.5 power;
4. Bond strength is essentially independent of the surface
deformation for the FRP reinforcement tested. While three
significantly different deformation patterns were considered,
similar bond strengths were developed. This finding supports
earlier results (Mosley et al. 2008), and also provides support
that a common design procedure can be used for a variety of
FRP bar types. This finding, however, does not imply that
surface deformation is not required. Tests were conducted
by Pay (2005) using smooth bars. These bars failed at
extremely low stresses in a pullout mode, and illustrate that
some level of deformation is required;
5. Top bar cast specimens produced only slightly lower
bond strengths (average 7% reduction) than bottom-cast specimens. The maximum reduction was 22%, which results in a
maximum top bar factor of 1.28. In addition, similar factors
were observed for steel and FRP reinforced specimens, indicating that a different factor is not needed for FRP reinforcement. Based on this research, the 1.5 factor recommended
by ACI Committee 440 (2006) appears overly conservative.
While this research suggests a lower factor, a top bar factor of
1.3 consistent with ACI 318-11 is recommended; and
6. The ACI 440.1R-06 equation for the development of
reinforcement results in significantly unconservative results
for the test results reported herein. For No. 5 (15.9 mm) bars,
all results are unconservative if the top bar factor ( = 1.5)
is not included. For the No. 8 (25.4 mm) bars, all results
are unconservative regardless of whether the top bar factor

266

is included. These results clearly indicate that an improved


design expression for FRP reinforcement is needed.
AUTHOR BIOS

Ali Cihan Pay is a Structural Engineer at Enka Teknik, Istanbul, Turkey.


He received his BS and MS from Middle East Technical University, Ankara,
Turkey, and his PhD from Purdue University, West Lafayette, IN.
Erdem Canbay is an Associate Professor of civil engineering at Middle
East Technical University. He received his BS from Istanbul Technical
University, Istanbul, Turkey, and his MS and PhD from Middle East Technical University.
Robert J. Frosch, FACI, is a Professor of civil engineering and Associate
Dean of the College of Engineering at Purdue University. He received his
BSE from Tulane University, New Orleans, LA, and his MSE and PhD from
the University of Texas at Austin, Austin, TX. He is a member of the ACIBoard
of Direction, is Chair of ACI Subcommittee 318-D, Flexure and Axial Loads
(Structural Concrete Building Code), and is a member of ACICommittees
224, Cracking; and 318, Structural Concrete BuildingCode.

ACKNOWLEDGMENTS

This study was conducted at both the Karl H. Kettelhut Structural Engineering Laboratory and Bowen Laboratory for Large Scale Research at
Purdue University, and was made possible under the sponsorship of the
Indiana Department of Transportation (INDOT) and the Federal Highway
Administration (FHWA) through the Joint Transportation Research Program
(JTRP) Project No. SPR-2491. Their support is gratefully acknowledged.
Thanks are extended to Hughes Brothers Inc. and Pultrall Inc. for providing
the FRP reinforcing bars used in this experimental program.

REFERENCES

ACI Committee 318, 2011, Building Code Requirements for Structural


Concrete (ACI 318-11) and Commentary, American Concrete Institute,
Farmington Hills, MI, 503 pp.
ACI Committee 408, 2003, Bond and Development of Straight Reinforcing
Bars in Tension (ACI 408R-03), American Concrete Institute, Farmington
Hills, MI, 49 pp.
ACI Committee 440, 2004, Guide Test Methods for Fiber-Reinforced
Polymers (FRPs) for Reinforcing or Strengthening Concrete Structures
(ACI440.3R-04), American Concrete Institute, Farmington Hills, MI,
40pp.
ACI Committee 440, 2006, Guide for the Design and Construction of
Structural Concrete Reinforced with FRP Bars (ACI 440.1R-06), American Concrete Institute, Farmington Hills, MI, 44 pp.
ASTM A615/A615M-12, 2012, Standard Specification for Deformed
and Plain Carbon-Steel Bars for Concrete Reinforcement, ASTM International, West Conshohocken, PA, 6 pp.
Canbay, E., and Frosch, R. J., 2005, Bond Strength of Lap-Spliced
Bars, ACI Structural Journal, V. 102, No. 4, July-Aug., pp. 605-614.
DeVries, R. A.; Moehle, J. P.; and Hester, W., 1991, Lap Splice
Strength of Plain and Epoxy-Coated Reinforcements, Report No. UCB/
SEMM-91/02, University of California, Berkeley, Berkeley, CA, 93 pp.
Ferguson, P. M., and Thompson, J. N., 1962, Development Length of
High Strength Reinforcing Bars in Bond, ACI Journal, V. 59, No. 7, July,
pp. 887-922.
Fish, K. E., 1992, Development Length of Fiber-Composite Concrete
Reinforcement, masters thesis, Iowa State University, Ames, IA, 129 pp.
Jirsa, J. O.; Breen, J. E.; Luke, J. J.; and Hamad, B. S., 1982, Effect of
Casting Position on Bond, International Conference on Bond in Concrete,
Paisley College of Technology, Paisley, Scotland, pp. 300-307.
Mosley, C. P.; Tureyen, A. K.; and Frosch, R. J., 2008, Bond Strength
of Nonmetallic Reinforcing Bars, ACI Structural Journal, V. 105, No. 5,
Sept.-Oct., pp. 634-642.
Pay, A. C., 2005, Bond Behavior of Unconfined Steel and Fiber
Reinforced Polymer (FRP) Bar Splices in Concrete Beams, doctoral
dissertation, Purdue University, West Lafayette, IN, 321 pp.
Tighiouart, B.; Benmokrane, B.; and Mukhopadhyaya, P., 1999, Bond
Strength of Glass FRP Rebar Splices in Beams Under Static Loading,
Construction and Building Materials, V. 13, No. 7, pp. 383-392.
Wambeke, B., and Shield, C., 2006, Development Length of Glass
Fiber-Reinforced Polymer Bars in Concrete, ACI Structural Journal,
V.103, No. 1, Jan.-Feb., pp. 11-17.

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S23

Flexural Behavior and Strength of Reinforced Concrete


Beams with Multiple Transverse Openings
by Bengi Aykac, Sabahattin Aykac, Ilker Kalkan, Berk Dundar, and Husnu Can
Reported are the results of experiments on 10 rectangular reinforced
concrete (RC) beams with and without multiple web openings. The
effects of opening geometry, the use of longitudinal stirrups in the
posts between the openings, the use of diagonal reinforcement
around openings, and the longitudinal reinforcement ratio on the
flexural behavior of RC beams with openings were investigated.
The stirrups in the posts were shown to have a significant contribution to the ductility of an RC beam with openings if no diagonal reinforcement is used. For the same reinforcement details,
RC beams with circular openings were found to have higher load
capacities and ductilities than beams with rectangular openings.
The experiments indicated that the posts between the openings
need to be prevented from undergoing shear failure to avoid Vierendeel truss action and allow a beam to develop its ductility and
bending capacity.
Keywords: diagonal reinforcement; plastic mechanism; reinforced concrete
beam; shear failure; shear reinforcement; Vierendeel truss; web crushing
failure; web opening.

INTRODUCTION
Ducts and pipes associated with the mechanical, electrical, and sewer systems in a building are usually located
underneath the floor beams, resulting in a considerable
loss in the usable floor height. Passage of these ducts and
pipes through web openings in floor beams offers an effective way to utilize the entire floor height, providing a more
economic and compact design. Nevertheless, the presence
of opening(s) in a reinforced concrete (RC) beam reduces
its load-carrying capacity and increases its service-load
deflections. The studies on concrete beams with transverse
openings in the literature focused on providing these beams
with strengths and rigidities comparable to their solid counterparts by proper reinforcement detailing. In this way, the
negative effects of the stress concentrations around the
openings could be eliminated, the load-carrying capacities
increased, and the deflections decreased.
Different types of services, including cooling and ventilation systems, power and sewer systems, and technology
and communication services, need to be effectively located
and distributed within structures. The presence of multiple
openings in a beam is needed to accommodate several pipes
and ducts related to various services. Steel beams with
multiple web openings (cellular beams) are commonly used
for this reason. In this study, the presence of multiple openings in RC beams was considered to improve the design of
RCstructures.
In an extensive experimental study on continuous RC
beams with a large rectangular opening, Mansur et al. (1991)
established that the failure of these beams is generally related
ACI Structural Journal/March-April 2014

to Vierendeel truss action. The deformations in a beam with


an opening were shown to increase and the collapse load to
decrease as the opening is moved to a more highly stressed
portion of span. As the opening length and depth increase,
Mansur et al. (1991) found that the Vierendeel action
becomes more pronounced, and the decrease in the collapse
load increases. Mansur et al. (1992) proposed that the deflections of an RC beam with a large rectangular opening can
be approximately estimated by assigning reduced flexural
and shear rigidities to the parts of containing the opening.
Tan and Mansur (1996) proposed design guidelines for the
strength and serviceability limit states of RC beams with
large openings. Mansur (1998) identified different shear
failure modes of RC beams with web openings and developed design equations. The tests carried out by Tan et al.
(2001) on RC beams with circular openings indicated that
the use of diagonal reinforcement offers an effective method
in crack control. Mansur (1999) developed design equations for RC beams subject to torsion in addition to bending
and shear. The equations correspond to the beam failure as
a whole, termed as beam-type, and failure of the top and
bottom chords separately, termed as frame-type. Mansur et
al. (2006) concluded that flexural capacities of RC beams
with large circular openings can be closely estimated using
strut-and-tie models. Yang et al. (2006) investigated the
strength and behavior of RC deep beams with web openings, and showed that the failure of a deep RC beam is
caused by the diagonal cracks projecting from the corners
of theopening.
In all aforementioned studies, RC beams with one or
two openings were considered. The failure in these beams
is generally related to the shear because the openings are
usually located in shear spans. In a recent experimental
program (Dundar 2008; Egriboz 2008; Aykac and Yilmaz
2011), the influence of multiple openings in the span was
investigated. The presence of multiple openings was
assumed to provide a more efficient design by helping the
stress concentrations around openings to be distributed to
the entire beam length. Furthermore, the presence of openings in the central zone in addition to shear spans was
assumed to shift the failure mode of the beam from brittle
shear failure to ductile flexural failure. Attempts were made
to prevent the brittle modes of shear failure (beam-type and
frame-type), and the ductilities of the beams were increased
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. 2012-070.R2, doi:10.14359.51686442, was received October 20, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

267

Fig. 1Reinforcement details of specimens without openings. (Note: 1 mm = 0.0394 in.)


EXPERIMENTAL STUDY

Table 1Test beams

Beam

Opening

Amount of
longitudinal
reinforcement

Stirrups
in posts

Diagonal
reinforcement

Concrete
batch

RBn

No

Moderate

No

No

RBb

No

High

No

No

RRxn

Square

Moderate

No

Yes

RRxcn

Square

Moderate

Yes

Yes

RRxb

Square

High

No

Yes

RRxcb

Square

High

Yes

Yes

RCb

Circular

High

No

No

RCcb

Circular

High

Yes

No

RCxb

Circular

High

No

Yes

RCxcb

Circular

High

Yes

Yes

by proper detailing: short stirrups in the chords, and posts


and full-depth stirrups next to openings. Furthermore, RC
beams with different opening geometries were tested within
the scope of the program to establish the geometry which
affects the strength and ductility of an RC beam to a lesser
extent. This paper reports 10 experiments carried out within
the program. The influence of the use of diagonal reinforcement around the openings, the use of stirrups in the posts,
and the opening geometry are the main test parameters. A
comparison of the experimental results with the estimates
from the theoretical methods yielded valuable conclusions.
RESEARCH SIGNIFICANCE
This study investigates the effects of different shear
reinforcement schemes and the opening geometry on flexural behavior of RC beams with multiple transverse openings. RC beams with different longitudinal reinforcement
ratios were tested, and different failure modes of RC beams
with openings were investigated within the course of the
study. The experimental results were compared with estimates from different theoretical formulations in the literature
to provide background knowledge for establishing design
rules for RC beams with multiple openings. The findings of
the present study will also guide further studies in the field.
268

Test specimens
A total of 10 rectangular RC beams, each 150 mm
(5.9in.) wide, 400 mm (15.7 in.) deep, and 4.0 m (13.1 ft)
long, were tested. Four specimens had 200 x 200 mm (7.9x
7.9in.) square openings, and four specimens had 200 mm
(7.9in.) circular openings. The reinforcement details of
the beams are illustrated in Fig. 1 and 2. In terms of flexural reinforcement ratios, the beams denoted with letter n
were moderately reinforced (tension reinforcement ratio rt =
0.0078), and the beams denoted with letter b were heavily
reinforced (rt = 0.014). The letter x in the specimen names
corresponds to the presence of 10 nonprestressed cables
spiraling around openings (Fig. 3), and c corresponds to
short stirrups in posts in longitudinal direction (Table 1).
Material properties
Table 2 tabulates the compressive strength of concrete of
each specimen on the test day obtained from 150 x 300 mm
(6 x 12 in.) cylinder tests. The mean values and standard
deviations of these material tests are tabulated in Table 2
together with the number of material tests. The mean values
and standard deviations of the yield and tensile strengths of
the S420 reinforcing bars and the number of samples for
each bar size are tabulated in Table 2.
Test setup and procedure
A 200 kN (45 kip) capacity steel frame was used for the
tests. The load, applied by a hydraulic cylinder and measured
by an electronic load cell, was equally distributed to four
loading points by main and secondary spreader beams
(Fig.4). In this way, the simply supported beams were
loaded at two points, each located at a distance of 300mm
(11.8 in.) from midspan, and two points, each located at a
distance of 1200 mm (47.2 in.) from midspan. Six-point
bending was adopted instead of four-point bending to more
closely simulate the moment distribution in a beam subjected
to uniform distributed loading, which is the most common
loading condition in real practice. The midspan vertical
deflection, the support settlements, and the distortions in
ACI Structural Journal/March-April 2014

Table 2Results of material tests


Cylinder compressive strength, MPa (ksi)

Yield strength, MPa (ksi)

Tensile strength, MPa (ksi)

Material test

Mean

Standard
deviation

RBn

26.8 (3.9)

3.8 (0.6)

RBb

33.9 (4.9)

2.8 (0.4)

RRxn

27.2 (3.9)

4.5 (0.6)

No. of tests

Mean

Standard
deviation

No. of tests

Mean

Standard
deviation

No. of tests

RRxcn

26.3 (3.8)

3.5 (0.5)

RRxb

27.9 (4.0)

1.4 (0.2)

RRxcb

24.8 (3.6)

5.8 (0.8)

RCb

27.8 (4.0)

2.7 (0.4)

RCcb

29.2 (4.2)

2.0 (0.3)

RCxb

28.3 (4.1)

3.6 (0.5)

RCxcb

26.1 (3.8)

6.1 (0.9)

10 bars

476.0 (69.0)

10.2 (1.5)

695.7 (100.9)

6.8(1.0)

12 bars

550.5 (79.8)

3.6 (0.5)

646.0 (93.7)

3.3 (0.5)

Fig. 2Reinforcement details of specimens with openings. (Note: All dimensions in mm; 1 mm = 0.0394 in.)
openings were measured with the help of linear variable
displacement transducers (LVDTs). The load and deflection
measurements were recorded by a data acquisition system.
The beams were loaded up to failure, and the cracks were
marked and the crack widths measured.
FAILURE MODES AND THEORETICAL EQUATIONS
Failure modes of reinforced concrete beams
withopenings
Beam- and frame-type shear and web crushing failures
are the three common types of shear failure in RC beams
with openings. In beam-type failure (Fig. 5(a)), a single
crack extending through the entire depth results in failure.
This diagonal crack is assumed to pass through the center
of opening. The frame-type shear failure (Fig. 5(b)) takes
place when two distinct cracks form in the top and bottom
chords, and one of the chords fails due to this cracking. Web

ACI Structural Journal/March-April 2014

Fig. 3Diagonal reinforcement spiraling around circular


openings.
crushing failure (Fig. 5(c)) is caused by crushing of concrete
between the diagonal cracks.
Based on the plastic hinge method, RC beams with openings are prone to failure due to formation of a collapse mechanism that is composed of four plastic hinges. This type
269

Fig. 4Test setup of present experimental program.


of failure is denoted as Vierendeel truss action (Fig. 5(d))
because the beams behave similar to a Vierendeel panel.
Vierendeel action causes an RC beam to fail at moments
below its bending capacity. Preventing Vierendeel action
and various forms of shear failure ensures that an RC beam
with openings will reach its bending capacity and fail in a
tension-controlled flexural mode due to crushing of concrete
in the compression zone, which is denoted as flexural failure
(Fig. 5(e)).
In the present study, the plastic methods based on truss
analogy (plasticity truss and strut-and-tie methods) were not
used. In these methods, the load capacity of an RC beam
is obtained from the axial capacities of different struts and
ties, obtained from the reinforcement available in each truss
member. RC beams with openings contain numerous Band D-regions, and these beams are modeled with a larger
number of truss members compared with RC beams without
openings, causing lengthy and tedious calculations.
Theoretical equations used in analysis
Shear strength and shear forces in chord members
Following the ACI approach (ACI Committee 318 2005),
Mansur (1998) was able to develop the following formula
for beam-type failure
Vn = 0.17 fc b ( d do )

Av f yv
s

( dv do ) + Ad f yd sin a


(1)

where fc is the concrete strength; b is the beam width; d


is the effective depth; do is the depth of opening; dv is the
distance between the centroids of extreme tension and
compression reinforcement layers; s is the stirrup spacing;
Av is the area of stirrups; Ad is the cross-sectional area of
the diagonal reinforcement within the failure surface; fyv and
fyd are the yield strengths of the stirrups and diagonal reinforcement, respectively; and a is the angle of inclination of
diagonalreinforcement.
The Architectural Institute of Japan (1988) gives the
following formula for the beam-type shear failure of RC
beams with openings

270

Fig. 5Failure modes of RC beams with multiple


transverseopenings.
0.092 ku k p ( fc + 17.7) 1.61 do
1

.
0
12
+
b dv
Vn =

V d

+0.846 w f yv

(2)

where h is the beam depth; ku is a size coefficient, varying


from 0.72 to 1.0; kp (Eq. (3)) is a factor accounting for the
reinforcement ratio; M and V are the bending moment and
shear force at critical section, respectively; and rw (Eq. (4))
is the web reinforcement ratio within dv

100 As
k p = 0.82
b d

w =

0.23

(3)

Av + Ad (sin a + cos a )
b dv

(4)

where As is the area of tension reinforcement. Mansur (1998)


modified the maximum allowable shear force (Vu)max formula
of ACI 318 (ACI Committee 318 2005) for RC beams
withopenings
ACI Structural Journal/March-April 2014

Table 3Analytical and experimental ultimate load values


Ultimate load, kN (kip)

Beam

Failure mode

Test Pex

Todeschini et al. (1964) Pan

RBn

Beam-type flexural

163.8 (36.8)

164.7 (37.0)

160.5 (36.1)

0.99

1.02

59.4 (2.3)

RBb

Beam-type flexural

245.8 (55.2)

291.7 (65.6)

289.0 (65.0)

0.84

0.85

95.6 (3.8)

RRxn

Vierendeel truss

156.4 (35.2)

159.6 (35.9)

156.9 (35.3)

0.98

1.00

62.5 (2.5)

RRxcn

Beam-type flexural

169.3 (38.1)

159.6 (35.9)

156.9 (35.3)

1.06

1.08

62.5 (2.5)

RRxb

Vierendeel truss

232.2 (52.2)

288.8 (64.9)

286.8 (64.5)

0.80

0.81

94.2 (3.7)

RRxcb

Web crushing

255.1 (57.3)

288.8 (64.9)

286.8 (64.5)

0.88

0.89

94.2 (3.7)

RCb

Diagonal tension

269.3 (60.5)

288.8 (64.9)

286.8 (64.5)

0.93

0.94

94.2 (3.7)

RCcb

Beam-type flexural

272.3 (61.2)

288.8 (64.9)

286.8 (64.5)

0.94

0.95

94.2 (3.7)

RCxb

Beam-type flexural

278.5 (62.6)

289.3 (65.0)

287.2 (64.6)

0.96

0.97

93.3 (3.7)

RCxcb

Beam-type flexural

284.1 (63.9)

289.3 (65.0)

287.2 (64.6)

0.98

0.99

93.3 (3.7)

(Vu )max = 5 0.85 0.17

fc b ( d do )

(5)

For frame-type shear failure, Mansur (1998) suggested


that the shear capacities of both chords should be checked
against the shear forces calculated from the following equations, proposed by Nasser et al. (1967)

Vut (Vub ) = Vu

At ( Ab )
At + Ab

(6)

where At and Ab are areas of the top and bottom chords,


respectively; Vu is the shear force in the section; and Vut and
Vub are the shear forces at the top and bottom chords, respectively. Tan and Mansur (1996) suggested that the shear force
in the section should be distributed to the chords in accordance to their flexural rigidities rather than their cross-sectional areas.
Flexural modes of failureThe bottom and top chords in
RC beams with openings are subjected to axial and shear
forces and bending moments. Due to axial forces and moments
in the chords, the liability of an RC beam to develop a failure
mechanism composed of four hinges can be evaluated with
the help of interaction diagrams as established by Tan and
Mansur (1996) and Mansur and Tan (1999). Considering
that the hinges at the top and bottom chords are subjected to
compression and tension, respectively, and the differences
in the directions of moments at different hinges, an interaction diagram is prepared for each hinge and checked against
the forces and moments that develop in the hinges at service
loads. Tan and Mansur (1996) proposed that the axial forces
in the chords can be obtained from

ACI Committee 318 (2005) PACI Pex/Pan Pex/PACI Neutral axis depth, mm (in.)

Nb = Nt =

Mm
z

(7)

where Nt and Nb are the axial forces in the top and bottom
chords, respectively; Mm is the bending moment at the section

ACI Structural Journal/March-April 2014

of hinging; and z is the distance between the centroids of


thechords.
ANALYSIS OF TEST RESULTS
Failure modes, ultimate loads, ductilities, and
rigidities of beams
Both reference beams (RBn and RBb) underwent
tension-controlled flexural failure (Fig. 6). In both beams,
the load was preserved, while the cover concrete crushed
and later dropped suddenly resulting in failure when the top
bars buckled and concrete crushing initiated. In both beams,
no considerable diagonal cracking took place in shear spans
(Fig. 7(a)). Table 3 indicates that the experimental ultimate
load of RBn was in close agreement with the values calculated from the rectangular stress block analysis of ACI 318-05
(ACI Committee 318 2005) PACI and from the Todeschini et
al. (1964) stress-strain model Pan. The load capacity of RBb
remained below the bending capacities calculated from both
models. Furthermore, Table 4 indicates that the ultimate
shear forces Vu in both beams at failure were smaller than
their respective shear strength values Vn, implying that shear
had no influence on failure.
Table 5 tabulates the deformation ductility index (DDI)
and rigidity values of the specimens. DDI is the ratio of a
beams deflection at the instant when the applied load drops
to 85% of the ultimate load to the deflection at yielding of
tension reinforcement. DDI is an indicator of the deformability of a beam without a significant reduction in load.
The rigidity values in the table correspond to the slope of
the initial linear branch of the load-deflection curve. In RC
beams, it is quite cumbersome to determine the slope of
the moment-curvature diagram due to variation of the flexural stiffness along the span caused by the discrete flexural
cracks. Therefore, slope of the load-deflection curve was
adapted.
Four different types of failure were observed in beams
with openings. RRxn and RRxb failed due to the formation of plastic failure mechanism (Fig. 8(a) and (b)). In
both RRxn and RRxb, two hinges formed at the ends of
the top and bottom chords of the opening closest to the end
271

Fig. 6Concrete crushing and buckling of compression


bars in reference beams.
support (Fig. 8(b)). The two beams differed in the locations of the remaining two hinges, which formed at the right
ends of the top and bottom chords of the fourth opening in
RRxn (Fig.9(a)) and the fifth opening in RRxb (Fig. 9(c)).
Unlike RC beams with one or two openings, the considerable distance between the hinging locations provided that
the stresses in the mechanism were distributed to greater
portions of the beam and the reversal of curvature took place
over a longer length. The posts inside the failure mechanism were also observed to fail in shear (Fig. 8(b)), which
was primarily related to the lack of stirrups in the posts.
RRxn failed at an applied load close to its ultimate capacity
(Table3), implying that the failure of a moderately reinforced
concrete beam with multiple openings due to formation of
a mechanism does not result in significant reductions in its
load capacity. The formation of mechanism caused greater
reductions in the capacity of RRxb. Table 5 and Fig.7(b)
show that both RRxb and RRxn had ductile behaviors up to
failure, with the DDI value of RRxb only 20% smaller than
RBb, and the DDI of RRxn 40% smaller than RBn. The fact
that RRxn and RRxb exhibited ductilities and load capacities comparable to their respective reference beams originated from two reasons. First, the diagonal reinforcement
in the beams carried the shear loads in the posts even after
the failure of the posts. Second, the significant longitudinal

272

distance between the hinges prevented the excessive stress


concentrations around hinges.
Figures 9(b) and (d) illustrate the linear approximations of
the interaction diagrams of RRxn and RRxb, respectively.
Because the top chords are subjected to compression and
bending, the yield planes above the bending moment axis
are composed of two linear segments, which intersect at
the point corresponding to balance failure. The yield planes
below the moment axis correspond to the bottom chords
subjected to tension as well as bending. The points corresponding to the axial forces and moments at the hinges at
failure are also shown in the diagrams. Both figures indicate that all of the points corresponding to the hinges
remain inside the yield surfaces implying that no hinging
was expected at failure. Nevertheless, the plastic hinging
at locations shown in Fig.9(a) and (c) and failure of RRxn
and RRxb due to hinging might be induced by the excessive
shear deformations in the posts causing additional stresses.
RRxcn, RCcb, RCxb, and RCxcb failed in flexure after
yielding of tension reinforcement (Fig. 8(c)). The final
failure was caused by the crushing of cover and core concrete
and buckling of compression bars. In these beams, the shear
cracks initiated in the chords and posts at the beginning of
loading did not widen and propagate in further stages of
loading, and the flexural cracks at the central part of the
beam controlled the behavior (Fig. 7(a)). Table 3 indicates
that all of these beams failed at loads close to their respective bending capacities. RCcb, RCxb, and RCxcb exhibited
greater ductilities than their reference beam RBb, with relative DDI values greater than unity, and had rigidities close to
the rigidity of the reference (Table 5). The neutral axis depth
values given in Table 3, calculated using the Todeschini et
al. (1964) stress-strain model, indicate that the compression zone in each beam remained within the top chord up to
failure and was not affected from the openings.
RCb underwent frame-type shear failure (Fig. 8(d)) due to
severe shear cracking in the chords of the opening closest to
the left end (Fig. 7(a)). Despite the final failure being originated from shear, RCb exhibited a ductile flexural behavior
up to failure, and the longitudinal reinforcing bars in the
beam yielded before failure. This explains why the load
capacity of the beam was only 7% smaller than its calculated
capacity (Table 3), and its DDI value was almost equal to the
DDI value of the reference beam (Table 5). Table 4 indicates
that the nominal shear strengths of the chords were significantly smaller than the shear forces in the chords at failure.
The use of short stirrups (Fig. 2) in the chords could not
prevent the frame-type failure. Table 4 shows that RCcb was
also liable to frame-type shear failure considering the significant discrepancies between the nominal shear strengths
of the chords and the shear forces in the chords at failure.
RCcb, however, failed in flexure, which can be attributed to
the presence of stirrups in posts.
RRxcb failed due to web crushing in the chords above
and below the second opening (Fig. 7(a) and 8(e)). Due
to this failure, RRxcb could not reach its bending capacity
(Table3), and had a limited ductility (Table 5). The failure of
the beam was not a diagonal tension failure because both the
chords and the beam had adequate shear strengths (Table4).
ACI Structural Journal/March-April 2014

Table 4Shear force and shear strength values


Beam-type failure

Frame-type failure
Top chord

Vn, kN (kip)

Bottom chord

Beam

ACI

AIJ

Vu, kN (kip)

Vnt, kN (kip)

Vut, kN (kip)

Vnb, kN (kip)

Vub, kN (kip)

RBn

168.7 (37.9)

81.9 (20.6)

RBb

165.7 (37.2)

122.9 (27.6)

RRxn

242.8 (54.6)

141.9 (31.9)

78.2 (17.6)

86.8 (19.5)

39.1 (8.8)

86.8 (19.5)

39.1 (8.8)

RRxcn

242.8 (54.6)

141.9 (31.9)

84.7 (19.0)

86.8 (19.5)

42.3 (9.5)

86.8 (19.5)

42.3 (9.5)

RRxb

257.8 (58.0)

141.8 (31.9)

116.1 (26.1)

94.2 (21.2)

58.1 (13.1)

94.2 (21.2)

58.1 (13.1)

RRxcb

257.8 (58.0)

141.8 (31.9)

127.6 (28.7)

94.2 (21.2)

63.7 (14.3)

94.2 (21.2)

63.7 (14.3)

RCb

147.2 (33.1)

90.0 (20.2)

134.6 (30.2)

38.9 (8.7)

67.3 (15.1)

38.9 (8.7)

67.3 (15.1)

RCcb

147.2 (33.1)

90.0 (20.2)

136.2 (30.6)

38.9 (8.7)

68.1 (15.3)

38.9 (8.7)

68.1 (15.3)

RCxb

257.9 (58.0)

142.1 (31.9)

139.3 (31.3)

94.3 (21.2)

69.6 (15.6)

94.3 (21.2)

69.6 (15.6)

RCxcb

257.9 (58.0)

142.1 (31.9)

142.1 (31.9)

94.3 (21.2)

71.0 (16.0)

94.3 (21.2)

71.0 (16.0)

Fig. 7(a) Cracking patterns of specimens at failure; and (b) load-deflection curves of specimens.
Nevertheless, both the chords and the entire beam were
subjected to shear forces above their maximum allowable
shear forces calculated from Eq. (5), which caused crushing
of concrete between the diagonals. Table 6 indicates that
RRxb, RCb, RCcb, RCxb, and RCxcb were also prone to
web crushing because the shear forces at failure exceeded
the maximum allowable shear forces. None of these beams,
ACI Structural Journal/March-April 2014

however, underwent web crushing failure. The maximum


allowable shear is calculated from Eq. (5) by assuming that
the depth of each chord is constant along its length, which
is an over-conservative assumption for circular openings. In
chords above and below circular openings, the chord depth
increases from mid-length of the chord to sides. Therefore,
the maximum shear force tolerable by a chord in an RC
273

Table 5Ductilities and rigidities of beams


Deformation ductility
index (DDI)
Beam

Absolute

Rigidity

Relative

Absolute,
kN/mm (kip/in.)

Relative

RBn

17.8

1.00

10.12 (57.8)

1.00

RRxn

10.7

0.60

5.77 (32.9)

0.57

RRxcn

8.4

0.47

5.87 (33.5)

0.58

RBb

6.4

1.00

12.05 (68.8)

1.00

RRxb

5.6

0.88

8.86 (50.6)

0.74

RRxcb

1.0

0.16

8.62 (49.2)

0.72

RCb

6.1

0.95

8.87 (50.6)

0.74

RCcb

9.5

1.48

9.32 (53.2)

0.77

RCxb

7.8

1.22

10.80 (61.7)

0.90

RCxcb

8.1

1.27

11.03 (63.0)

0.92

ultimate load and rigidity values greater than those of RCcb,


while the DDI value of RCcb exceeded the value of RCxcb.
The rigidity and load capacity of an RC beam with openings can be increased by using diagonal steel if the posts are
reinforced with stirrups, but the diagonal reinforcement does
not contribute to the ductility in this case.

Fig. 8Flexural and shear failure modes of specimens with


openings.
beam with circular openings is greater than the value from
Eq. (5). This might be the reason that none of the specimens
with circular openings failed in diagonal compression. Web
crushing in the chords might have affected the failure of
RRxb, which eventually failed due to Vierendeel action.
Effect of diagonal reinforcement on beam
behavior
Figure 10 indicates the load-deflection curves of beams
with and without diagonal reinforcement. The load-deflection curves indicate that the use of diagonal reinforcement
in an RC beam increases its energy absorption capacity. The
DDI, rigidity (Table 5), and ultimate load (Table 3) values of
RCxb are higher than the values of RCb, implying that diagonal reinforcement contributes to the flexural behavior if no
stirrups are used in the posts. RCxcb, on the other hand, had
274

Effect of stirrups in posts on beam behavior


Figure 7(b) illustrates that the stirrups in the posts significantly contribute to the energy capacity if no diagonal
reinforcement is used. The DDI value of RCcb (Table 5) is
also considerably greater than the value of RCb, implying
the significant contribution of the stirrups to the ductility in
the absence of diagonal reinforcement. In beams with rectangular openings, it appears that the use of stirrups in the
posts does not contribute to the ductility and energy capacity
of the beam if the beam has diagonal reinforcement. The
DDI values of RRxcn and RRxcb were considerably smaller
than the values of RRxn and RRxb, respectively. In beams
with circular openings and diagonal reinforcement, the stirrups in the posts have almost no contribution to the ductility.
The DDI value of RCxcb is approximately equal to the value
of RCxb. Table 3 indicates that, in all cases, the use of stirrups has a positive but minor effect on the load capacity. To
summarize, the stirrups in the posts improve the behavior of a
beam when the beam does not have diagonal reinforcement.
In the presence of diagonal reinforcement, the stirrups in
the posts have little or no contribution to the flexural performance. The use of stirrups in the posts in addition to diagonal reinforcement causes the posts to be too strong, shifting
the failure to the chords. If the chords fail in brittle modes
(Vierendeel truss action, diagonal tension, or compression),
the beam exhibits limited ductility.
Effect of opening geometry on beam behavior
Figure 11 illustrates the load-deflection curves of beams
with the same reinforcement details but with different
opening geometries. Both plots indicate that RC beams
with circular openings have much greater energy capacities
than the beams with rectangular openings. The DDI and
ACI Structural Journal/March-April 2014

Table 6Adequacy of beams for shear


Beam-type failure
(entire section)

Frame-type failure
(chord members)

Beam

(Vu)max, kN
(kip)

Vu, kN (kip)

(Vut)max, kN
(kip)

Vut, kN (kip)

RBn

248.0 (55.8)

81.9 (20.6)

RBb

235.3 (52.9)

122.9 (27.6)

RRxn

105.0 (23.6)

78.2 (17.6)

55.2 (12.4)

39.1 (8.8)

RRxcn

105.0 (23.6)

84.7 (19.0)

55.2 (12.4)

42.3 (9.5)

RRxb

106.4 (23.9)

116.1 (26.1)

54.8 (12.3)

58.1 (13.1)

RRxcb

106.4 (23.9)

127.6 (28.7)

54.8 (12.3)

63.7 (14.3)

RCb

106.4 (23.9)

134.6 (30.2)

54.8 (12.3)

67.3 (15.1)

RCcb

106.4 (23.9)

136.2 (30.6)

54.8 (12.3)

68.1 (15.3)

RCxb

107.1 (24.1)

139.3 (31.3)

55.2 (12.4)

69.6 (15.6)

RCxcb

107.1 (24.1)

142.1 (31.9)

55.2 (12.4)

71.0 (16.0)

rigidity values of RCxb and RCxcb are significantly higher


than the values of RRxb and RRxcb, respectively (Table 5).
Furthermore, the ultimate loads carried by RCxb and RCxcb
considerably exceeded the ultimate loads of RRxb and
RRxcb, respectively (Table 3). The less favorable behavior
of RC beams with rectangular openings is mainly due to the
stress concentrations in the corners. In beams with rectangular openings, the shear cracks in the posts were observed
to project from these sharp corners (Fig. 12), and these
cracks caused reductions in the rigidities and load capacities. Secondly, the smaller areas of the circular openings
compared with the rectangular ones resulted in less reductions in the beam capacities.
Although rectangular openings have more adverse effects
on the beam behavior compared with circular openings,
provision of rectangular openings in RC beams might be
unavoidable, considering that the air-conditioning ducts in
buildings are usually rectangular. RC beams with rectangular openings should be more carefully designed because
they are more prone to Vierendeel action and shear failure
of the chords.
SUMMARY AND CONCLUSIONS
Two reference beams, four RC beams with multiple
circular, and four beams with multiple rectangular openings were tested to determine the flexural performance of
RC beams with openings. Three of the beams had moderate
amounts of flexural reinforcement, while the remaining
beams were heavily reinforced. Each beam with openings
had longitudinal bars and full-depth stirrups adjacent to
openings and short stirrups in the chords. The longitudinal
reinforcement ratio, opening geometry, and use of diagonal
reinforcement and longitudinal stirrups in the posts were the
main test parameters. The experiments and comparison with
theoretical formulations yielded the following conclusions:
The use of diagonal reinforcement contributes to the
ductility and load capacity if the posts are not reinforced
with stirrups in longitudinal direction. Similarly, the

ACI Structural Journal/March-April 2014

Fig. 9Hinging locations and interaction diagrams of


Beams RRxn and RRxb. (Note: 1 kN = 0.225 kip; 1 kN.m =
0.738 kip-ft.)

stirrups in the posts increase the ductility and capacity


of RC with openings in the absence of diagonal steel.
The simultaneous use of diagonal reinforcement and
stirrups in the posts has minor or no contribution to
ductilities and load-carrying capacities of RC beams
with openings. The presence of diagonal reinforcement
and stirrups in the posts causes the posts to be overly
strong, which leads to failure of the chords rather than
the posts.
For the same reinforcement details, RC beams with
circular openings have higher ductilities and load
capacities compared with the beams with rectangular
openings. The experiments indicated that the stress
concentrations at corners of rectangular openings result
in cracking, which leads to the reductions in the flexural
rigidities without exhibiting full ductility.
In RC beams with multiple openings, the considerable distance between the hinging locations causes the
stresses in the failure mechanism to be distributed to
greater portions of the beam compared with the beams
with one or two openings. Therefore, RC beams with
multiple openings exhibit a more ductile behavior
even if they fail due to Vierendeel action. The posts
275

Fig. 10Contribution of diagonal reinforcement to beam


ductility in presence and absence of stirrups in posts.

remaining inside the failure mechanism in Vierendeel


action were observed to fail in shear. Diagonal reinforcement around the openings was found to carry the
forces in the posts and prevent the complete failure of
an RC beam, even after the failure of the posts. The
large shear deformations in the posts were shown to
cause Vierendeel action.
The failure of an RC beam with openings due to Vierendeel action causes greater reductions in the load-carrying capacity as the longitudinal reinforcement ratio of
the beam increases.
The use of short stirrups in the chords is not an adequate
measure for prevention of frame-type shear failure. The
use of diagonal reinforcement and stirrups in the posts
limits the extent of shear cracking in the chords, and
prevents the frame-type shear failure.
RC beams with openings are more liable to web
crushing (diagonal compression) failure if the chords
and beams have high amounts of shear reinforcement
and small widths.

DESIGN RECOMMENDATIONS FOR REINFORCED


CONCRETE BEAMS WITH WEB OPENINGS
The experimental and analytical results obtained within
the scope of the present study indicated that the load capaci276

Fig. 11Contribution of circular opening geometry on beam


ductility.

Fig. 12Shear cracks projecting from corners of rectangular openings.


ties and ductilities of RC beams with multiple openings can
be increased by proper strengthening of the chords and posts.
When the chords are weak in shear, an RC beam with openings is prone to diagonal tension or compression failures
of the chords. The use of short stirrups proved to be effective for preventing frame-type shear failure. Furthermore,
the chords should be designed to not exceed the maximum
allowable shear force to prevent web crushing failure. RC
beams with rectangular openings are more prone to different
forms of shear failure compared with beams with circular
ACI Structural Journal/March-April 2014

openings. Special attention should be given to shear design


of beams with rectangular openings.
The shear forces are distributed to the chords in accordance to
their cross-sectional areas. The most efficient design is achieved
when the openings are placed at mid-depth of the member.
When the centers of openings are offset from mid-depth, one of
the chords of each opening is weaker than the other, and is more
vulnerable to different forms of shearfailure.
The presence of several openings proved to be effective
in improving the behavior of a beam. Unlike RC beams
with one or two openings, the stresses inside failure mechanism were observed to be distributed to greater portions of
the beam. The posts inside the failure mechanism should
be designed to resist the excessive deformations in the
mechanism. The use of longitudinal stirrups in the posts
or diagonal reinforcement was shown to effectively delay
shear failure of the posts and increase the ductility and load
capacity. The present experiments indicated that the use of
longitudinal stirrups in the posts with a volumetric ratio of
0.016 provided an RC beam with circular openings with a
load capacity and DDI 11 and 48% greater than its reference.
Similarly, the use of diagonal reinforcement with a volumetric ratio of 0.017 resulted in an increase of 13 and 22%
in the load capacity and DDI of an RC beam with circular
openings compared with its reference. RC beams with
rectangular openings reached load capacities close to their
references in the presence of diagonal reinforcement, with a
volumetric ratio of 0.013. Nevertheless, the DDI value of the
moderately reinforced RC beam with rectangular openings
remained approximately 40% of its reference in this case.
It was found that the simultaneous use of longitudinal stirrups and diagonal steel causes the posts to be overly strong,
causing the chords to fail before the posts. In the present
study, the beams strengthened with both diagonal reinforcement and longitudinal stirrups in the posts exhibited ductilities more than 50% smaller than the beams with only diagonal reinforcement.
AUTHOR BIOS

Bengi Aykac is a Lecturer in the Civil Engineering Department at Gazi


University, Ankara, Turkey, where she received her BS, MS, and PhD. Her
research interests include behavior of concrete beams strengthened by
jacketing techniques, concrete beams strengthened with steel plates, and
crack repair by epoxy injection.
Sabahattin Aykac is an Assistant Professor in the Civil Engineering
Department at Gazi University, where he received his BS, MS, and PhD in
1989, 1993, and 2000, respectively. His research interests include strengthening and repair of concrete beams, earthquake behavior of strengthened
concrete beams, and concrete beams with openings.
Ilker Kalkan is an Assistant Professor in the Department of Civil Engineering at Kirikkale University, Kirikkale, Turkey. He received his BS from
Middle East Technical University, Ankara, Turkey, in 2004, and his MS
and PhD from the Georgia Institute of Technology, Atlanta, GA, in 2006
and 2009, respectively. His research interests include structural stability,
fiber-reinforced polymer concrete, and strengthening of concrete beams.
Berk Dundar is a Civil Engineer at Aydiner Construction Company,
Turkey. He received his BS from Dokuz Eylul University, Izmir, Turkey,
in 2005, and his MS from Gazi University in 2008. His research interests
include the behavior of concrete beams with openings.
Husnu Can is a Professor in the Civil Engineering Department at Gazi
University, Ankara, Turkey. He received his BS from Ankara Higher Tech-

ACI Structural Journal/March-April 2014

nology Institute in 1967; his MS from Tulane University, New Orleans, LA,
in 1977; and his PhD from Ankara University, Ankara, Turkey, in 1983.
His research interests include behavior of strengthened reinforced concrete
beams under repetitive loading and the behavior of strengthened reinforced
concrete columns.

ACKNOWLEDGMENTS

This paper represents a condensation of the thesis prepared at Gazi


University, Ankara, Turkey, by B. Dundar under the supervision of H. Can
and S. Aykac toward the degree of Master of Science.

Ab
Ad
At
Av
do
dv

=
=
=
=
=
=

fyd
fyv
z
a

=
=
=
=

r t

NOTATION

cross-sectional area of bottom chord


cross-sectional area of diagonal reinforcement within failure plane
cross-sectional area of top chord
cross-sectional area of stirrups
diameter of circular opening or depth of rectangular opening
distance between centers of extreme tension and compression
steel layers
yield strength of diagonal reinforcement
yield strength of stirrups
distance between centroids of top and bottom chords
angle of inclination of diagonal reinforcement with longitudinal
beam axis
tension reinforcement ratio

REFERENCES

ACI Committee 318, 2005, Building Code Requirements for Structural


Concrete (ACI 318-05) and Commentary, American Concrete Institute,
Farmington Hills, MI, 430 pp.
Architectural Institute of Japan (AIJ), 1988, Standard for the Structural
Calculation of Reinforced Concrete Structures, Architectural Institute of
Japan, Tokyo, Japan, 207 pp.
Aykac, S., and Yilmaz, M. C., 2011, Behaviour and Strength of RC
Beams with Regular Triangular or Circular Web Openings, Journal of
Faculty of Engineering and Architecture of Gazi University, V. 26, No. 3,
pp. 711-718. (in Turkish)
Dundar, B., 2008, Behaviour and Strength of Reinforced Concrete
Beams with Regular Openings, MSc thesis, Gazi University, Ankara,
Turkey, pp. 16-22. (in Turkish)
Egriboz, Y. E., 2008, Behaviour and Strength of R/C Beams with
Regular Rectangular or Circular Web Openings, MSc thesis, Gazi University, Ankara, Turkey, pp. 27-35. (in Turkish)
Mansur, M. A., 1998, Effect of Openings on the Behaviour and Strength
of R/C Beams in Shear, Cement and Concrete Composites, V. 20, No. 6,
pp. 477-486.
Mansur, M. A., 1999, Design of Reinforced Concrete Beams with
Small Openings under Combined Loading, ACI Structural Journal, V. 96,
No. 5, Sept.-Oct., pp. 675-681.
Mansur, M. A.; Huang, L. M.; Tan, K. H.; and Lee, S. L., 1992, Deflections of Reinforced Concrete Beams with Web Openings, ACI Structural
Journal, V. 89, No. 4, July-Aug., pp. 391-397.
Mansur, M. A.; Lee, Y. F.; Tan, K. H.; and Lee, S. L., 1991, Tests on
RC Continuous Beams with Openings, Journal of Structural Engineering,
ASCE, V. 117, No. 6, pp. 1593-1606.
Mansur, M. A., and Tan, K. H., 1999, Concrete Beams with Openings:
Analysis and Design, CRC Press, Boca Raton, FL, 224pp.
Mansur, M. A.; Tan, K. H.; and Weng, W., 2006, Analysis of Concrete
Beams with Circular Web Openings Using Strut-and-Tie Models, Malaysian Journal of Civil Engineering, V. 18, No. 2, pp. 89-98.
Nasser, K. W.; Acavalos, A.; and Daniel, H. R., 1967, Behavior and
Design of Large Openings in Reinforced Concrete Beams, ACI Journal,
V. 64, No. 1, Jan., pp. 25-33.
Tan, K. H., and Mansur, M. A., 1996, Design Procedure for Reinforced
Concrete Beams with Large Web Openings, ACI Structural Journal,
V.93, No. 4, July-Aug., pp. 404-411.
Tan, K. H.; Mansur, M. A.; and Wei, W., 2001, Design of Reinforced
Concrete Beams with Circular Openings, ACI Structural Journal, V. 98,
No. 3, May-June, pp. 407-415.
Todeschini, C. E.; Bianchini, A. C.; and Kesler, C. E., 1964, Behavior
of Concrete Columns Reinforced with High Strength Steels, ACI Journal,
V. 61, No. 6, June, pp. 701-716.
Yang, K. H.; Eun, H. C.; and Chung, H. S., 2006, The Influence of Web
Openings on the Structural Behavior of Reinforced High-Strength Concrete
Deep Beams, Engineering Structures, V. 28, No. 13, pp. 1825-1834.

277

NOTES:

278

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S24

Experimental Assessment of Inadequately Detailed


Reinforced Concrete Wall Components
by Adane Gebreyohaness, Charles Clifton, John Butterworth, and Jason Ingham
An experimental study undertaken to assess the seismic behavior of
reinforced concrete (RC) walls constructed prior to the introduction of seismic design requirements in the New Zealand Standard
Model Building By-law is presented. The geometric characteristics
and material properties of the test specimens were replicated from
those of an existing building. The primary test variables considered were wall thickness, magnitude of applied axial compressive
load, and aspect ratio. In addition, the influence of the longitudinal reinforcing bar splices, which are positioned in locations
that are not permitted by current design standards, on the seismic
performance of the walls was investigated. The response of the test
specimens was dominated by rocking, after yielding of the longitudinal reinforcing bars located adjacent to the boundaries of the
walls occurred. The peak strength and the stiffness of the test specimens dropped rapidly and significantly after low-level drift cycles.
Overall, the test specimens exhibited poor ductility and limited
energy dissipation capacity. Provisions for required tension splice
lengths of plain round bars in ASCE/SEI 41-06 were found to be
excessively conservative.
Keywords: existing buildings; lap splice; lightly reinforced; walls.

INTRODUCTION
Reinforced concrete (RC) buildings constructed prior to
the 1970s in New Zealand and other seismically active countries were not designed and detailed to undergo a ductile
mode of failure.1,2 Poor performance of this class of buildings has been observed in earthquakes occurring in many
countries around the world, such as Chile in 1985,3 Turkey
in 1999,4 Chile in 2010,5 and New Zealand in 2011.6 During
the recent 2010/2011 Canterbury (New Zealand) earthquake sequence, this class of buildings performed poorly
and in some cases collapsed catastrophically. Christchurch
City Council Building Safety Evaluation statistics indicate
that 60% of the pre-1970 RC buildings in Christchurch
were deemed suitable for restricted access only or were
unsafe after the 6.3 magnitude Christchurch earthquake on
February 22, 2011, which claimed 181 lives.7 The majority
of the fatalities were attributed to the collapse of two RC
buildingsthe PGC building and the CTV building, which
were constructed in 1963 and 1986, respectively.8 The PGC
building had RC stair/lift core walls that had similar reinforcing bar configurations to the walls discussed herein, and
the poor performance of the singly and lightly reinforced
walls led to the collapse of the building.9
Previous research attention has primarily been directed
toward the structural components of pre-1970 RC frame
buildings, principally inadequately detailed columns and
beam-column joints, owing to the clear seismic risk in
this class of buildings that these structural components
ACI Structural Journal/March-April 2014

pose. Although there have been a number of experimental


studies10-13 where the seismic performance of existing walls
was assessed, studies that evaluated the performance of
lightly and singly reinforced walls with detailing similar
to those found in the PGC building are limited. Test specimens considered in experimental studies reported in the
literature were doubly reinforced,11-13 used higher longitudinal reinforcing bar ratios than those found in many older
buildings,10-13 or used deformed bars instead of plain round
bars.10,11,13 In addition, few of the test specimens investigated previously had spliced longitudinal reinforcement.12
Therefore, it was concluded that a study was necessary to
assess the seismic performance of singly and lightly RC
walls incorporating inadequate longitudinal bar splices.
The experimental study presented herein was based on the
walls of a case study building located in Wellington, New
Zealand. This street corner building was constructed in 1928
before the publication of NZSS 95,14 which introduced seismic-resistant design requirements in New Zealand for the
first time. The building has internal, one-way, moment-resisting, riveted steel frames and RC walls located at the
perimeter as the lateral-force-resisting systems. The walls
have limited openings on sides adjacent to neighboring
buildings and have larger and more regular openings on the
street frontages.
The case study building was assessed15 in accordance with
the nonlinear dynamic procedure detailed by ASCE/SEI,2,16
using a suite of seven earthquake records17 relevant to the
seismicity of the building site, to determine the performance
of the building during a likely earthquake. From the assessment, it was found that the capacity of the building will be
exceeded in moderate level earthquakes, which were defined
as being one-third as strong as the design level earthquakes
relevant to the building site.18 The walls were identified to
be the primary lateral-force-resisting components of the
building, and those walls located at the street frontages were
found to be the most critical walls of the building.
To obtain an understanding of the seismic performance
of walls similar to those found in the case study building,
an experimental program was undertaken on replicas of
typical segments of the most critical walls. The configuration of the reinforcing bars within the walls was determined from the original structural drawings and construction specifications of the case study building. The walls are
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-071.R2, doi:10.14359.51686520, was received August 13, 2012,
and reviewed under Institute publication policies. Copyright 2014, American
Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published ten months from this journals date if the
discussion is received within four months of the papers print publication.

279

singly reinforced with widely spaced plain round bars, and


the quantity of longitudinal and transverse reinforcement
within these walls is less than what is required by current
standards19,20 to induce a ductile response. The longitudinal
reinforcing bars are spliced just above the floor levels. The
spliced reinforcing bars, which would be required by current
design standards20 to be hooked, lack proper end anchorages. In addition, the splice lengths are too short according
to current design standards19,20 to initiate yielding of spliced
reinforcing bars before slip occurs. There is also a lack of
transverse confinement reinforcement to contain concrete in
compression zones and to prevent longitudinal reinforcing
bars from buckling. Similar to the wall piers, the coupling
beams of the case study building, which would typically
receive diagonal reinforcement if constructed to current
standards,21 are reinforced with longitudinal and transverse
reinforcement to resist flexure and shear, respectively.
RESEARCH SIGNIFICANCE
Because RC buildings constructed before the 1970s
comprise a considerable portion of vulnerable RC structures that pose significant seismic risk, as-built performance
assessment and seismic retrofitting, preferably without loss
of heritage attributes to the buildings, are crucial steps toward
ensuring better performance in future earthquakes. Current
recommendations2,3,16 for the assessment and improvement
of existing RC walls are based on modern design codes,
which address new walls and are not necessarily applicable to walls designed and constructed following outdated
methods. The experimental study presented herein intends
to contribute to a more realistic seismic performance assessment of existing nonconforming RC walls.
EXPERIMENTAL PROGRAM
Twelve full-scale wall components were constructed
in-place and tested in the Civil Test Hall of the University of
Auckland as part of a research program planned to assess the
seismic performance of existing buildings in New Zealand.
The test matrix is presented in Table 1.
Description of test specimens
The geometric characteristics and reinforcing bar configurations of the test specimens were determined based on
the original structural drawings and construction specifications of the case study building. The walls in the bottom
five stories of the case study building are 230 mm (9 in.)
thick, and the walls in the upper four stories are 150 mm
(6 in.) thick. Both types of walls are provided with a single
layer of reinforcement consisting of 10 mm (3/8 in.) diameter plain round reinforcing bars placed at 305 mm (12 in.)
centers spacing, in both the horizontal and vertical directions
and located at the midthickness plane of the walls. In addition, two 12 mm (1/2 in.) diameter boundary bars are placed
around all openings. The longitudinal bars are spliced just
above the floor levels with splice lengths of 305 and 457 mm
(12 and 18 in.) for the 10 and 12 mm (3/8 and 1/2 in.) diameter bars, respectively, with no transverse reinforcing bar
enclosing the lap. In Fig. 1, the dimensions and reinforcing
bar arrangements of the test specimens are presented.
280

Table 1Test matrix


Test
specimen

lw, mm

h, mm

t, mm

M/Vlw

Compressive
axial load, kN

WPS1

1300

1750

150

1.35

200

WPS2

1300

1750

230

1. 35

300

WPS3

1300

2400

150

0.92

WPS4

1300

2400

230

0.92

WPS5

1300

2400

150

0.92

WPS6

1300

2400

230

0.92

WPS7

1300

2400

150

0.92

200

WPS8

1300

2400

230

0.92

300

WPS9

1300

2400

150

0.92

200

WPS10

1300

2400

230

0.92

300

WSS1

1000

1600

150

0.8

WSS2

1000

1600

230

0.8

Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.

Ten of the test specimens (WPS1 to WPS10) were replicated wall piers and the remaining two test specimens
(WSS1 and WSS2) were replicated coupling beams. Eight
of the wall piers (WPS3 to WPS10) and both of the coupling
beams had boundary reinforcement, as shown in Fig. 1. In
four of the wall piers (WPS5 to WPS8), the longitudinal
reinforcing bars were spliced near the base of the wall piers,
whereas in the remaining test specimens, the longitudinal
reinforcing bars were continuous and were anchored outside
the wall piers.
During construction of the test specimens, a concrete
compressive strength of 21 MPa (3.0 ksi) was specified to reflect the average strength determined from core
samples extracted from the case study building. In addition,
Grade300 (fy = 300 MPa [43.5 ksi]) plain reinforcing bars
were planned to be used, as this grade is close to the strength
of the nominal Grade 240/250 reinforcing bars used during
the era of construction of the building. Some of the reinforcing bars used, however, were incorrectly supplied as
Grade 500 (fy = 500 MPa [72.5 ksi]). This error was identified only after the testing program was completed. Generally, Grade 500 bars have a shorter yield plateau, rupture at
lower strain levels, require longer development lengths, and
are more likely to buckle than Grade 300 bars. However,
reinforcement grade did not significantly affect the behavior
of the test specimens, apart from the apparent influence on
wall strength. Material properties of the test specimens are
summarized in Table 2.
Test setup and instrumentation
The first two wall piers (WPS1 and WPS2) were tested as
vertical cantilevers (refer to Fig. 2(a)), while the remaining
eight wall piers (WPS3 to WPS10) and the coupling beams
(WSS1 and WSS2) were subjected to double bending using
a steel loading beam mounted on and anchored to the top
RC blocks (refer to Fig. 2(b)). As shown in Fig. 2(c), the
coupling beams were rotated and tested in a vertical orientation. The double bending loading condition was represenACI Structural Journal/March-April 2014

Fig. 1Test specimen geometries and reinforcement details.


ACI Structural Journal/March-April 2014

281

Table 2Material properties


Longitudinal and transverse reinforcement

Longitudinal reinforcement
splices

Boundary reinforcement

Test
specimen

fc
MPa

Al, At,
mm

fy, fyt,
MPa

fult,
MPa

l, %

t, %

Ab, mm

fy, MPa

fult, MPa

Splice

lb, 10, mm

lb, 12,
mm

WPS1

18.4

10

530

667

0.20

0.18

No

WPS2

20.9

10

300

429

0.13

0.12

No

WPS3

19.6

10

351

488

0.20

0.17

412

388

555

No

WPS4

16.2

10

351

488

0.13

0.11

412

305

436

No

WPS5

29.4

10

348

487

0.20

0.17

412

516

662

Yes

305

457

WPS6

24.8

10

348

487

0.13

0.11

412

516

662

Yes

305

457

WPS7

21.3

10

344

456

0.20

0.17

412

305

438

Yes

305

457

WPS8

22.5

10

344

456

0.13

0.11

412

305

438

Yes

305

457

WPS9

20.2

10

490

631

0.20

0.17

412

301

433

No

WPS10

19.3

10

490

631

0.13

0.11

412

301

433

No

WSS1

18.7

10

351

472

0.21

0.20

412

321

426

No

WSS2

21.4

10

351

472

0.14

0.13

412

321

426

No

tative, as closely as possible, of the fixed-fixed sway support


condition of wall components in multi-story buildings.
Compressive axial load was applied to the wall piers using
four high-strength bars that were positioned parallel to the
wall centerline and anchored to the strong floor (refer to
Fig.2(a) and (b)).
Typical test specimen instrumentation is shown schematically in Fig. 3. Load cells, denoted as LC1 to LC3, were
employed to measure the magnitude of forces applied on the
test specimens. The lateral displacement of the test specimens was measured using potentiometer TP1 and the readings were corrected to provide the horizontal displacement
of the top of the test specimens. Portal gauges PG1-PG6,
PG7-PG14, and PG15-PG24 were used to measure rocking,
flexural, and shear deformations, respectively. Relative
sliding displacements that could have occurred during the
tests at the foundation block-specimen, specimen-top block,
and strong floor foundation block interfaces were also monitored using portal gauges PG25, PG26, and PG27, respectively. Ten strain gauges per test specimen were attached to
the transverse and longitudinal reinforcement. Strain gauges
were glued at the midheight of the starter bars to four pairs
of spliced bars (one on the starter bar and another one on the
main bar) to monitor the performance of the splices located
within four of the test specimens (WPS5 to WPS8).
Testing procedure
The test specimens were subjected to quasi-static cyclic
loading, with the loading regime based on the ACI-recommended22 loading sequence for assessing the performance of new RC structural components. Potentiometer
TP1 (refer to Fig. 3) was employed for the loading regime
displacementcontrol.
OBSERVED RESPONSE
The inelastic response of the test specimens was dominated by rocking (refer to Fig. 4). All of the test specimens
282

exhibited wide cracks localized at the bottom and at the


top of the walls (refer to Fig. 5) accompanied by rupture
of longitudinal reinforcement located adjacent to the wall
boundaries. No significant flexural and shear deformations
were observed or recorded within the body of the walls.
The magnitude of the sliding deformations observed at the
wall-foundation block interfaces was insignificant when
compared to the magnitude of wall rocking deformations.
No sliding took place at the wall-top block interfaces. There
was visible slipping of the spliced bars along the provided
splice length, especially during testing of those specimens
that were not subjected to axial load. Fracturing of all of the
continuous longitudinal bars was observed during testing,
while none of the spliced bars were fractured.
The test specimens that were subjected to double bending
responded as intended, with crack development being
anti-symmetric about the midheight of the walls at drift
cycles of less than 1%. When the testing progressed to drift
cycles of greater than 1%, the top sections of the test specimens underwent significantly larger rotations than did the
bottom sections, as the top of the test specimens were not
fully restrained against rotation. This support condition
resulted in an upward shifting of the inflection point, which
in turn led to larger bending moments being developed at
the base of the walls, causing the cracks at the bottom of the
walls to become wider and the cracks at the top of the walls
to close up. During testing of WPS5 and WPS6, the cracks at
the top of the walls became fully closed with the longitudinal
reinforcement that had previously yielded in tension hidden
in the cracks. Strain gauge readings indicated that the bars
at those locations had already yielded in tension during early
cycles of loading. At drifts of greater than 1%, damage was
localized at the base of the walls and the walls were rocking
on their foundation blocks.
Significant spalling of concrete was observed at all wall
corners of the thinner test specimens (WPS3, WPS5, WPS7,
and WPS9) (refer to Fig. 5). The thicker test specimens
ACI Structural Journal/March-April 2014

Fig. 3Typical wall instrumentation.

Fig. 2Test setup.


having axial load (WPS2, WPS8, and WPS10) experienced
relatively limited spalling that was located at the bottom
wall corners only. Spalling of concrete during testing of the
coupling beam specimens (WSS1 and WSS2) was limited
to the top wall corners only. The spalling of concrete in
the compression zones, which was later exacerbated by
excessive compression at the wall toes as the displacement
demand increased, was initiated by buckling of longitudinal reinforcement. Buckling of longitudinal reinforcement
was observed during all tests except for those specimens
subjected to no axial load and being 230 mm (9 in.) thick
(WPS4 and WPS6). These two specimens also exhibited
no spalling of concrete (refer to Fig. 5 for the response of
WPS4). Reinforcement grade had no discernible influence
on reinforcing bar buckling and concrete spalling.

ACI Structural Journal/March-April 2014

Fig. 4Typical components of the lateral displacement of


wall.
RESULTS AND DISCUSSIONS
The lateral-force-carrying capacity of the test specimens
was limited by their flexural strength, with the experimentally obtained peak strengths and corresponding drift values
presented in Table 3. The test specimens with continuous
longitudinal reinforcing bars developed 99 to 108% of their
calculated flexural strength and the test specimens with
spliced longitudinal reinforcing bars developed 97 to 102%
of their computed flexural strength. The calculated flexural
strengths were determined following routine section analysis
procedures and by considering all the reinforcement within
the cross section of the test specimens to contribute to the
strength. The calculations used experimentally determined
reinforcing bar yield strengths and concrete compressive strengths. The strengths of the test specimens with
283

Fig. 5Typical state of test specimens at end of test.


lap-spliced longitudinal reinforcing bars were computed
assuming that the splices would develop the yield strength
of the spliced bars.
The strength of the test specimens degraded quickly after
low drift cycles, as shown in Fig. 6, due to rupture of the
longitudinal reinforcing bars located adjacent to the wall
boundaries. For test specimens subjected to applied axial
compressive load, the lateral-force-carrying capacity of the
test specimens was not completely lost after the longitudinal
reinforcing bars had ruptured, due to the flexural strength
attributable to the applied axial load. However, these test
284

specimens had to maintain their axial-load-carrying capacity


to retain the residual strength to higher drift demands. For
example, because of loss of axial-load-carrying capacity,
WPS9 could not achieve the same level of ductility and
residual strength as was exhibited by all other wall pier
specimens. Slipping of the spliced bars along the provided
splice length improved the performance of test specimens
having spliced longitudinal bars, especially those specimens
that were not subjected to compressive axial load (refer to
Fig. 6).

ACI Structural Journal/March-April 2014

Table 3Measured and calculated strengths of test specimens


Test specimen

VTest, kN

Drift at VTest, %

Vn,SF*, kN

Vn,S, kN

Vn,F, kN

VTest/Vn,SF

VTest/Vn,S

VTest/Vn,F

WPS1

162

0.96

616

395

151

0.26

0.41

1.07

WPS2

174

0.66

647

513

167

0.27

0.34

1.04

WPS3

159

0.43

404

389

147

0.39

0.41

1.08

WPS4

149

0.20

404

455

148

0.37

0.33

1.01

WPS5

199

0.35

536

424

195

0.37

0.47

1.02

WPS6

194

0.44

536

508

197

0.36

0.38

0.98

WPS7

231

0.36

662

391

233

0.35

0.59

0.99

WPS8

271

0.5

802

492

278

0.34

0.55

0.97

WPS9

260

0.36

740

478

259

0.35

0.54

1.00

WPS10

308

0.21

880

564

310

0.35

0.55

0.99

WSS1

160

0.09

361

320

149

0.44

0.50

1.07

WSS2

150

0.04

361

417

151

0.42

0.36

0.99

Average

0.36

0.45

1.02

Standard deviation

0.05

0.09

0.04

Nominal shear friction capacity according to ACI 318-11.

Nominal shear capacity according to ACI 318-11.

Nominal flexural capacity according to ACI 318-11 and assuming splices would develop yield strength of spliced bars.

Note: 1 kN = 0.225 kip.

The boundary and distributed longitudinal reinforcement


provided to the walls were below the limits that are specified
by both ACI 318 and NZS 3101 (refer to Table 2). ACI318
stipulates the minimum longitudinal reinforcement ratio
for earthquake force-resisting RC walls to be 0.25% if Vu
exceeds 0.83Acvfc (MPa) (10Acvfc [psi]), and the tested
walls satisfy this condition. In addition to the minimum
ratio, ACI318 requires at least two 16 mm (5/8in.) diameter
bars to be provided around all openings. The requirement
of NZS3101 is dependent on concrete strength and reinforcement yield strength. For the walls tested, the minimum
required reinforcement ratio was, on average, 0.3%.
NZS3101 also requires additional reinforcement with yield
strength equal to or greater than 600 N/mm (3426lb/in.) of wall
thickness to be provided around all openings. This requirement is approximately equivalent to two 12 mm (1/2in.)
and three 12 mm (1/2 in.) diameter Grade 500 bars for the
150and 230 mm (6 and 9 in.) thick walls, respectively.
Due to the additional two 12 mm (1/2 in.) diameter
boundary reinforcing bars, the wall pier specimens that were
tested in double bending (WPS3 to WPS10) achieved more
strength and energy dissipation when compared to the wall
pier specimens having no boundary reinforcement (WPS1
and WPS2). However, even with these additional bars, the
quantity of longitudinal reinforcement provided resulted in
a level of applied shear force that was insufficient to induce
a shear mode of failure. Similar findings were previously
reported in studies12,13 conducted on the behavior of existing
RC walls detailed following pre-1970s detailing techniques,
but having relatively more longitudinal reinforcement than
used in the test specimens discussed herein.

ACI Structural Journal/March-April 2014

Flexural failure is the principal failure mode for existing


RC walls constructed in New Zealand before the amendment of NZSS 9514 in 1955, mainly because in NZSS 95
the contribution of concrete to the shear strength of RC
structural components is underestimated.23 Consequently, in
the absence of boundary frame elements, walls of this era
are generally expected to have sufficient shear strength to
develop flexural overstrength. However, due to the provision
of a low quantity of reinforcement and a lack of transverse
confinement reinforcement, this type of wall has low flexural
strength and exhibits a low ductility capacity.
The test specimens dissipated a significant amount of
energy through yielding of longitudinal reinforcing bars
during cycles to drifts of less than 1%. However, the
energy dissipation capacity of the test specimens reduced
considerably as the drift demands increased, due to a lack
of any dissipative mechanism except sliding friction at the
wall-foundation block interfaces. The bar slip that occurred
over the splice lengths did not significantly alter the energy
dissipation capacity for those test specimens that incorporated lap splices.
After 1% drift demand, the stiffness of the test specimens had typically degraded significantly to less than 10%
of the initial stiffnesses, with continuing degradation until
reaching approximately 1% of the initial stiffnesses at drift
demands in excess of 2.5%. The rapid and significant loss of
stiffness observed during testing indicates that, after large
earthquakes, this type of wall becomes too soft to develop
significant ongoing resistance and therefore will undergo
larger displacements when subjected to small lateral forces
associated with either a long-duration event or aftershocks
having significant intensity at the site.

285

Fig. 6Lateral-force, top-displacement responses of test specimens.


COMPARISON OF RESULTS WITH ASCE 41-06
PROVISIONS
Section C6.7.1 of ASCE/SEI 41-06,16 Seismic Rehabilitation of Existing Buildings, categorizes the response of RC
walls based on aspect ratio. According to ASCE/SEI 41-06,
walls with an aspect ratio of less than 1.5 are considered to
be squat and their response is assumed to be controlled by
shear. Conversely, walls with an aspect ratio of greater than
3 are assumed to be slender and their response is considered to be controlled by flexure. The response of walls with
intermediate aspect ratios is considered to be controlled by
both flexure and shear. As the walls presented herein had

286

aspect ratios ranging between 1.35 and 1.85, the response


of the walls was supposed to be controlled by shear and by
both shear and flexure, respectively, according to ASCE/SEI
41-06.
Wall strength
ASCE/SEI 41-06 refers to Chapter 21 of ACI 318-11 to
determine the shear strength of existing walls. The nominal
shear strength of walls is given in ACI 318-11, Eq. (21-7), as

Vn = Acv (a c fc + t f yt )

(1)

ACI Structural Journal/March-April 2014

This nominal shear strength is limited to 0.083Acvfc


(MPa) (Acvfc [psi]). The coefficient c is 0.25 for hw/lw
1.5, is 0.17 for hw/lw 2.0, and varies linearly between 0.25
and 0.17 for intermediate hw/lw values. When determining
the yield and nominal shear strengths, ASCE/SEI 41-06
limits the strength of the transverse reinforcing bars fyt to the
specified yield strength.
The nominal shear-friction strength of walls across a
sliding plane perpendicular to shear-friction reinforcing bars
is given by ACI 318-11, Eq. (11-25), as

Vn = (Avf fy + N*) (2)

The nominal shear-friction strength given in Eq. (2) is


limited to the smaller of 0.2Ac fc (N) (0.2Ac fc [lb]) and 5.5Ac
(N) (800Ac [lb]).
For the flexural strength of existing walls, ASCE/SEI41-06
refers to the basic principles outlined in Chapter 10 of ACI
318-11, but requires the use of expected yield strengths of the
longitudinal reinforcing bars instead of specified minimum
yield strengths. When determining flexural yield strengths of
walls with no boundary members, ASCESEI 41-06 requires
considering only the longitudinal reinforcing bars within the
outer 25% of the wall cross section, but when determining
the nominal flexural strengths, the contributions of all longitudinal reinforcing bars within the wall component cross
section need to be considered.
As shown in Table 3, the lateral-force-carrying capacity of
the test specimens was limited by their flexural strength and,
thus, categorizing the response of RC walls by ASCE/SEI
41-06 based on aspect ratio only is found to be misleading.
The results also show that the plane sections remain plane
hypothesis, which was used during the calculation of the
flexural strengths, provided strengths that agree well with
those determined experimentally, even for squat walls with
an aspect ratio of 1.35. In addition, prior to the peak strength
of the test specimens being attained, the strain gauge readings from longitudinal reinforcement of most test specimens
generally varied linearly across the section.
Required length of splices
Current provisions for the required length of splices are
based on studies conducted to determine bond-slip relationships between isolated reinforcing bars and the surrounding
concrete. Transfer of force between starter and main reinforcing bars over the provided splice length involves a
different force transfer mechanism than that occurring
between an isolated reinforcing bar and the surrounding
concrete, but it is widely accepted that the required length
of splices is the same as the required development lengths
of single embedded reinforcing bars.24 Accordingly, ASCE/
SEI 41-06 refers to the provisions for required length of
tension splices of ACI 318-11, which are based on requirement for tension development length. ASCE/SEI 41-06
specifies that the required splice length of plain round reinforcing bars to be taken as twice that required for deformed
reinforcing bars. According to ACI 318-11, Eq. (12-1), the
required length of Class B deformed reinforcing bar splices

ACI Structural Journal/March-April 2014

Table 4Ratios of maximum stress to yield stress


for spliced bars
Test
specimen

10 reinforcing bars
*

12 reinforcing bars

fs/fy

fs/fy

fs/fy*

fs/fy

WPS5

0.51

0.80

0.44

0.73

WPS6

0.51

0.80

0.44

0.73

WPS7

0.51

0.80

0.77

1.00

WPS8

0.51

0.80

0.77

1.00

Ratio according to Eq. (5).

Ratio according to Eq. (6).

subjected to tension, which is required to be at least 300 mm


(12 in.), is

fy
ld = 1.3
1.1l fc

t e s l
db
cb + K tr
d

b

(3)

The length of deformed reinforcing bar splices subjected


to compression can be determined from Section 12.16.1 of
ACI 318-11 as

f y 420
0.071 f y db ,
ld =

(0.13 f y 24)db , f y > 420

(4)

This development length is required to be at least 300 mm


(12 in.) and is required to be increased by one-third for a
concrete compressive strength of less than 21 MPa (3.0 ksi).
When the splice length of reinforcing bars within an
existing wall is found to be inadequate, ASCE/SEI 41-06
stipulates the maximum stress that can be developed within
the spliced bars to be determined as follows

fs = (lb/ld)fy fy (5)

Ratios of the maximum stresses, which can be developed


by spliced plain round bars according to Eq. (5) to the corresponding yield strengths of the spliced bars, are presented in
Table 4. Although the provided splice lengths were significantly less than those required by Eq. (5) to develop the yield
strength of the spliced bars, the walls were able to develop
97to 102% of their computed flexural strength (refer to
Table 3), which was determined assuming that the lap splices
would develop the yield strength of the spliced bars. Peak
strengths reached by the test specimens with lap-spliced
reinforcing bars were underestimated by an average of 41%
(refer to Table 5) by predictions made using Eq. (5). Similar
findings were previously reported in studies24,25 undertaken to investigate the behavior of columns with short lap
splices. The lateral force capacity of columns investigated
by Cho and Pincheira24 was underestimated using Eq. (5)
by an average of 28%. Similarly, columns tested by Melek
and Wallace25 were reported to have achieved 97 to 103%
287

Table 5Measured and calculated strengths of test specimens with spliced longitudinal reinforcement
Test specimen

VTest, kN

Vn,F-A/S*, kN

Vn,F-PS, kN

Vn,F, kN

VTest/Vn,F-A/S

VTest/Vn,F-PS

VTest/Vn,F

WPS5

199

91

129

195

2.14

1.51

1.02

WPS6

194

91

144

197

2.16

1.37

0.98

WPS7

231

181

208

233

1.29

1.12

0.99

WPS8

271

234

262

278

1.19

1.06

0.97

Average

1.70

1.27

0.99

Standard deviation

0.46

0.18

0.02

Nominal flexural capacity according to ASCE/SEI 41-06.

Nominal flexural capacity according to proposed supplement26 to ASCE 41-06.

Nominal flexural capacity according to ACI 318-11 and assuming splices would develop yield strength of spliced bars.

Note: 1 kN = 0.225 kip.

Fig. 7Normalized maximum bond stresses that developed between spliced bars and surrounding concrete.
of their yield strengths, which were calculated assuming
that the lap splices would develop the yield strengths of the
spliced bars, but the provided splice lengths were approximately 67% of that required by ASCE/SEI 41-06.
Because Eq. (5) was found to be excessively conservative, the provided splice lengths used in the study reported
herein were also compared with those implied by a proposed
supplement26 to ASCE/SEI 41-06. The proposed equation,
which is a modified version of Eq. (5) and based on the work
of Cho and Pincheira,24 is

l
fs = 1.25 b
ld

u=

f s db
4ld

(7)

Using Eq. (3) and (7), the ASCE/SEI 41-06 implied


maximum bond stress that was expected to develop between
a plain round bar subjected to tension and the surrounding
concrete was determined as

0.67

fy fy

(6)

In most cases, the provided splice lengths were shorter


than required by Eq. (6) (refer to Table 4) to develop the full
strength of the spliced bars. Predictions made using Eq.(6)
underestimated the peak strengths by an average of 21%.
As discussed previously, both Eq. (5) and Eq. (6) predict
slip to occur at force levels less than the yield strength of the
spliced bars. However, during testing, all of the monitored
splices developed tensile stresses that were greater than the
experimentally determined yield strength of the spliced bars.

288

The maximum stresses that were measured during testing


were converted to maximum bond stresses as follows

u=

1 1 1.1 fc cb + K tr

2 4 t e s l db

(8)

Maximum bond stresses developed between the spliced


bars and the surrounding concrete are compared in Fig. 7,
with maximum bond stress values implied by ASCE/SEI
41-06 and the proposed supplement to ASCE/SEI 41-06.
The ASCE/SEI 41-06 implied average maximum bond
stress of 0.29fc (MPa) (3.49fc [psi]) is significantly less
than the measured average bond stress of 0.57fc (MPa)
(7.95fc [psi]). The average maximum bond strength
implied by the proposed supplement to ASCE/SEI 41-06,
ACI Structural Journal/March-April 2014

which was dependent on the provided splice lengths, was


0.46fc (MPa) (5.54fc [psi]).
The performance of the splices reported herein was better
than had been expected, principally because of the excessively conservative requirement of current design codes19,20
and assessment recommendations16 for the required splice
length of plain round bars. For example, ACI 318 requires
the length of splices to be increased by one-third if all of
the longitudinal bars are spliced at the same location. This
requirement is not based on strength criteria,27 but is primarily
intended to encourage designers to stagger bar splices. This
requirement alone results in underestimating the capacity of
spliced reinforcing bars in existing structures by 23%.
In some cases encountered herein, the compression splice
length requirements of ACI 318 governed the required
length of splices. The compression splice length requirements of ACI 318 have remained essentially the same since
the 1963 version of the code and appear to be conservative.
Required tension splice lengths are expected to govern the
required length of splices, as the formation of transverse
tension cracks around a splice that is subjected to tension
reduces bond strength and, thus, increases the length that is
required to allow the splice to transfer the desired magnitude
of stress. Chun et al.28 have recently reported a better performance of compression splices than tension splices, which
was principally attributed to end bearing of reinforcing bars
when subjected to compression. Similarly, Cairns29 has
found the equations contained in ACI 318-11 to be conservative, and has proposed that the length of compression splice
for deformed bars be taken as 30% shorter than that required
by the equations.
In addition to the aforementioned excessively conservative
requirements of ACI 318 for deformed bars, ASCE41-06
recommends taking the required splice length of plain round
bars as twice that required for deformed bars, which results
in significantly underestimating the capacity of plain round
splices located in existing structures.
The good performance of the splices reported herein
was not considered to be influenced by the thick concrete
cover that the splices were provided with. ACI 318 employs
parameters for concrete cover/reinforcement spacing cb
and transverse reinforcement index Ktr to account for
the confinement term (cb + Ktr)/db, and in ACI 318, it is
assumed that an increase in the value of this confinement
term, above the maximum allowed 2.5, is not likely to
increase anchorage capacity and is not likely to prevent a
pullout failure, which is a failure mode typically exhibited
by plain round bars. Similarly, Eligehausen et al.30 reported
that an increase in concrete cover, reinforcement spacing, or
transverse confinement can prevent concrete splitting failures only, which is a failure mode typically sustained when
using spliced deformed bars. The maximum allowed value
of (cb+ Ktr)/db = 2.5 was employed when calculating the
ASCE/SEI 41-06 implied bond strength of the plain round
bars discussed herein.
CONCLUSIONS
Twelve test specimens replicated from wall segments of
an existing building were experimentally tested to assess the
ACI Structural Journal/March-April 2014

seismic behavior of RC walls constructed before the introduction of seismic design requirements in the New Zealand
Standard Model Building By-law.14 The results of the experimental tests were evaluated and compared with current
assessment provisions.
Based on the study presented herein, the following conclusions are drawn:
1. The lateral-force-carrying capacity of lightly reinforced
existing walls is limited by their flexural strength. Owing to
the low quantity of reinforcing bars provided, yielding of
the longitudinal reinforcing bars dictates the strength of this
wall type.
2. The strength and the stiffness of this type of wall
degrade rapidly and significantly. The walls have limited
energy dissipation capacity, principally due to the provision
of few longitudinal reinforcing bars. In addition, the walls
suffer from a lack of transverse confinement reinforcement
to contain concrete in compression zones and to prevent
longitudinal reinforcing bars from buckling.
3. During testing, the wall pier specimens having no axial
load exhibited cracks that were wide during low-level drift
cycles, but the cracks, which were located near the supports,
closed up and appeared inconspicuous after the tests were
completed, with longitudinal reinforcement that had previously yielded in tension hidden in the cracks. This type of
crack could easily be overlooked and the walls may appear
intact during post-earthquake inspections, even if the stiffness and the strength of the walls deteriorated significantly.
4. From peak stresses measured during testing, it was
shown that the provisions contained in ASCE/SEI 41-06
significantly underestimate the maximum stresses that can
be developed by plain round reinforcing bar lap splices. The
relatively recent recommendations of a proposed supplement to ASCE/SEI 41-06 also underestimate the maximum
stresses that can be developed, but to a lesser extent.
5. During testing, the lap splices were able to develop bond
stresses that were significantly higher than the maximum
possible bond stresses implied by ASCE/SEI 41-06. Further
research is recommended, as the provisions of ASCE/SEI
41-06 for required splice lengths of plain round reinforcing
bars are based on studies conducted for deformed bars. The
provisions are excessively conservative and potentially lead
to unnecessary or expensive seismic retrofitting solutions.
The provisions also lead to the potential for overlooking the
danger of existing walls failing in shear during an earthquake,
before the actual strength of the tension splices isexceeded.
ACKNOWLEDGMENTS

The authors would like to gratefully acknowledge the financial support


provided by the New Zealand Foundation for Research, Science and Technology (FRST) through Grant UOAX0411.

AUTHOR BIOS

Adane Gebreyohaness is a Structural Engineer at Beca Limited in New


Zealand. His research interests include the seismic assessment, strengthening, and design of reinforced concrete and steel structures.
Charles Clifton is an Associate Professor at the University of Auckland,
Auckland, New Zealand. His research interests include the performance
of steel and composite steel/concrete buildings in severe earthquakes and
severe fires.

289

John Butterworth is an Associate Professor at the University of Auckland.


His research interests include nonlinear structural and solid mechanics;
stability; buckling behavior of thin-walled sections; structural dynamics;
earthquake engineering; base isolation using rolling, sliding, and rocking
mechanisms; passive control of structure response (especially by energy
dissipating joints); experimental dynamics; pounding of bridges and buildings; and assessment and retrofit of steel structures.
ACI member Jason Ingham is an Associate Professor and Deputy Head
(Research) of the Department of Civil and Environmental Engineering at
the University of Auckland. His research interests include seismic assessment, retrofit and design of reinforced and prestressed concrete structures,
and sustainable concrete technology.

Ab
Ac
Acv
Al
A t
Avf
cb

=
=
=
=
=
=
=

db =
fc =
fs
=
fult =
fy
=
fyt =
hw =
Ktr =
l
=
lb
=
ld
=
lw =
M/Vlw =
N* =
u
=
Vn =
VTest =
Vu =

=

=
l =
t

e
s
t

=
=
=

NOTATION

area of boundary reinforcement


area of concrete section resisting shear transfer
shear area
area of longitudinal reinforcing bar
area of transverse reinforcing bar
area of shear friction reinforcement
smaller of: (a) distance from center of bar or wire to nearest
concrete surface; and (b) one-half the center-to-center spacing
of bars or wires being developed
nominal bar diameter
specified compressive strength of concrete
splice strength
ultimate strength of reinforcement
yield strength of reinforcement
yield strength of shear reinforcement
height of wall
transverse reinforcement index
length of splice
provided splice length
required splice length according to ACI 318-11
horizontal length of wall
shear span-to-depth ratio
design axial load
bond stress
nominal shear strength
peak strength of test specimen
factored shear force
modification factor related to density of concrete
coefficient of friction
ratio of area of distributed longitudinal reinforcement to gross
concrete area perpendicular to that reinforcement
ratio of area of distributed transverse reinforcement to gross
concrete area perpendicular to that reinforcement
modification factor based on reinforcement coating
modification factor based on reinforcement size
modification factor based on reinforcement location

REFERENCES

1. NZSEE, Assessment and Improvement of the Structural Performance


of Buildings in Earthquake: Recommendations of a NZSEE Study Group
on Earthquake Risk Buildings, New Zealand Society for Earthquake Engineering Inc., Wellington, New Zealand, 2012, 343 pp.
2. FEMA, Pre-standard and Commentary for the Seismic Rehabilitation of Buildings (FEMA 356), Federal Emergency Management Agency,
Washington, DC, 2000, 519 pp.
3. Wood, S. L.; Wight, J. K.; and Moehle, J. P., The 1985 Chile EarthquakeObservations on Earthquake-Resistant Construction in Vina Del
Mar, Structural Research Series No. 532, University of Illinois at Urbana-Champaign, Champaign, IL, 1987, 192 pp.
4. Sezen, H.; Whittaker, A. S.; Elwood, K. J.; and Mosalam, K. M.,
Performance of Reinforced Concrete Buildings during the August 17, 1999
Kocaeli, Turkey Earthquake, and Seismic Design and Construction Practise
in Turkey, Engineering Structures, V. 25, No. 1, 2003, pp. 103-114.
5. EERI, EERI Special Earthquake Report: The Mw 8.8 Chile Earthquake of February 27, 2010, Earthquake Engineering Research Institute,
Oakland, CA, 2010, 20 pp.
6. Kam, W. Y., and Pampanin, S., The Seismic Performance of RC
Buildings in the 22 February 2011 Christchurch Earthquake, Structural
Concrete, V. 12, No. 4, 2011, pp. 223-233.

290

7. New Zealand Police, List of Deceased, http://www.police.govt.nz/


list-deceased. (last accessed Jan. 12, 2012)
8. IPENZ, Christchurch EarthquakeAn Overview, http://www.ipenz.
org.nz/ipenz/forms/pdfs/ChChFactSheets-Overview.pdf. (last accessed
Aug. 25, 2011)
9. Beca Carter Hollings & Ferner Ltd (Beca), Investigation into the
Collapse of the Pyne Gould Corporation Building on 22nd February 2011,
Prepared for Department of Building and Housing (DBH), 2011, 51 pp.
10. Orakcal, K.; Massone, L. M.; and Wallace, J. W., Shear Strength
of Lightly Reinforced Wall Piers and Spandrels, ACI Structural Journal,
V.106, No. 4, July-Aug. 2009, pp. 455-465.
11. Kuang, J. S., and Ho, Y. B., Seismic Behavior and Ductility of
Squat Reinforced Concrete Shear Walls with Nonseismic Detailing, ACI
Structural Journal, V. 105, No. 2, Mar.-Apr. 2008, pp. 225-231.
12. Ireland, M.; Pampanin, S.; and Bull, D. K., Experimental Investigations of a Selective Weakening Approach for the Seismic Retrofit of RC
Structural Walls, NZSEE Conference, Palmerston North, New Zealand,
Mar. 30-Apr. 1, 2007, 8 pp.
13. Greifenhagen, C., and Lestuzzi, P., Static Cyclic Tests on Lightly
Reinforced Concrete Shear Walls, Engineering Structures, V. 27, No. 11,
2005, pp. 1703-1712.
14. NZ Standards Institute, NZSS 95: New Zealand Standard Model
Building By-Law, Sections I to X, NZ Standards Institute, Wellington,
New Zealand, 1935, 39 pp.
15. Gebreyohaness, A. S.; Clifton, G. C.; and Butterworth, J. W.,
Assessment of Soil-Foundation-Structure Interaction Effects on the
Seismic Performance of an Old Dual Wall-Frame Building, 14th ECEE
Conference, Ohrid, Macedonia, Aug. 30-Sept. 4, 2010, 8 pp.
16. ASCE/SEI 41, Seismic Rehabilitation of Existing Buildings,
American Society of Civil Engineers, Reston, VA, 2007, 428 pp.
17. Oyarzo-Vera, C.; McVerry, G.; and 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, V. 28, No. 2, 2012,
pp. 1-22.
18. DBH, Building Act 2004, Department of Building and Housing,
Wellington, New Zealand, 2004, 353 pp.
19. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-11) and Commentary, American Concrete Institute,
Farmington Hills, MI, 2011, 503 pp.
20. NZS 3101:2006, Concrete Structures Standard: Part 1The
Design of Concrete Structures, Standards New Zealand, Wellington, New
Zealand, 2006, 696 pp.
21. Paulay, T., Seismic Response of Structural Walls: Recent Developments, Canadian Journal of Civil Engineering, V. 28, No. 6, 2001,
pp.922-937.
22. ACI Innovation Task Group 1 and Collaborators, Acceptance
Criteria for Moment Frames Based on Structural Testing (ACI T1.1-01),
American Concrete Institute, Farmington Hills, MI, 2001, 10 pp.
23. Brunsdon, D. R., Seismic Performance Characteristics of Buildings
Constructed between 1936 and 1975, University of Canterbury, Christchurch, New Zealand, 1984.
24. Cho, J. Y., and Pincheira, J. A., Inelastic Analysis of Reinforced
Concrete Columns with Short Lap Splices Subjected to Reversed Cyclic
Loads, ACI Structural Journal, V. 103, No. 2, Mar.-Apr. 2006, pp.
280-290.
25. Melek, M., and Wallace, J. W., Cyclic Behavior of Columns with
Short Lap Splices, ACI Structural Journal, V. 101, No. 6, Nov.-Dec. 2004,
pp. 802-811.
26. Elwood, K. J.; Matamoros, A. B.; Wallace, J. W.; Lehman, D. E.;
Heintz, J. A.; Mitchell, A. D.; Moore, M. A.; Valley, M. T.; Lowes, L. N.;
Comartin, C. D.; and Moehle, J. P., Update to ASCE/SEI 41 Concrete
Provisions, Earthquake Spectra, V. 23, No. 3, 2007, pp. 493-523.
27. ACI Committee 408, Bond and Development of Straight Reinforcement in Tension (ACI 408R-03), American Concrete Institute, Farmington
Hills, MI, 2003, 49 pp.
28. Chun, S. C.; Lee, S. H.; and Oh, B., Compression Lap Splice in
Unconfined Concrete of 40 and 60 MPa (5800 and 8700 psi) Compressive Strengths, ACI Structural Journal, V. 107, No. 2, Mar.-Apr. 2010,
pp. 170-178.
29. Cairns, J., Strength of Compression Splices: A Reevaluation of Test
Data, ACI Journal, V. 82, No. 4, July-Aug. 1985, pp. 510-516.
30. Eligehausen, R.; Popov, E. P.; and Bertero, V. V., Local Bond
Stress-Slip Relationships of Deformed Bars under Generalized Excitations, University of California, Berkeley, Berkeley, CA, 1983, 180 pp.

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S25

Behavior of Epoxy-Injected Diagonally Cracked Full-Scale


Reinforced Concrete Girders
by Matthew T. Smith, Daniel A. Howell, Mary Ann T. Triska, and Christopher Higgins
Many cast-in place reinforced concrete deck-girder (RCDG)
bridges in the national inventory contain diagonal cracks typically
associated with shear-moment interaction. Many transportation
agencies use epoxy injection as a prophylactic to seal cracks.
The shear-dominated performance of girders treated with epoxy
injection is not well understood and the existing data regarding
performance of members injected with epoxy have been gathered
from reduced-scale test specimens. To provide new data on the
shear performance of girders with epoxy-injected diagonal cracks,
five full-scale RCDG specimens were constructed to reflect 1950s
proportions and details. The specimens were loaded to produce
diagonal cracks, injected with epoxy under varying degrees of
axial tension and service loading, and tested to failure. The test
data indicated that epoxy injection resulted in minimal increased
shear capacity. However, the epoxy-injected specimens exhibited
increased load magnitudes prior to crack re-initiation and reduced
stirrup stresses at serviceability levels compared to the cracked
condition prior to repair.
Keywords: bridges; cracking; epoxy; full-scale testing; reinforced concrete;
repair; shear.

INTRODUCTION AND BACKGROUND


Large numbers of cast-in-place reinforced concrete deckgirder (RCDG) bridges remain in the national inventory and
are reaching the end of their originally intended design lives.
Many of these bridges are exhibiting varying degrees of
diagonal-tension cracking in the girders and supporting bent
caps. Diagonal cracks can be attributed to many sources,
including insufficient reinforcing and poor flexural detailing,
increasing service load magnitude and volume, and temperature or shrinkage strains. These cracks are generally of
concern to engineers due to the nonductile nature of shear
failure and limited reserve shear strength. Further, diagonal cracks can expose the embedded reinforcing steel to
chlorides, moisture, and oxygen, which promote corrosion,
thereby weakening the structure. Several methods exist that
can seal diagonal cracks and possibly restore or increase
capacity, such as externally bonded steel and carbon fiber
materials, applied ferrocement, cement grouting, and epoxy
injection. Epoxy, in particular, has been widely used for the
rehabilitation of concrete structures.
The first record of epoxy material testing for highway
applications was performed by the California Division of
Highways Materials and Research Department (Rooney
1963). The application was focused on highway maintenance patching of damaged roadways and securing reflective traffic markers. An earlier study using small-sized plain
concrete prisms demonstrated that the bond strength of
epoxy permitted rupture in the concrete and not in the adhe-

ACI Structural Journal/March-April 2014

sive bond layer irrespective of the loading conditions (shear,


flexure, or tension) (Tremper 1960).
As epoxy materials became more widely used, researchers
began to investigate the use of epoxy injection for reinforced
concrete members. Early tests to evaluate the performance of
epoxy-injected reinforced concrete beams were performed
by Chung (1975) with 125 x 200 mm (5 x 8 in.) beam specimens on a clear span of 2754 mm (9 ft). Longitudinal and
transverse reinforcement were provided, and each specimen
was first loaded to failure and then all major cracks exceeding
0.08 mm (0.003 in.) were injected with epoxy resin. Chung
observed that epoxy restored the flexural capacity of the
failed specimens. Another study conducted by Popov and
Bertero (1975) examined the behavior of full-scale and halfscale cantilever reinforced concrete beam-column specimens
injected with epoxy and exposed to reversed cyclic loading.
The steel detailing and member proportions were typical of
large-sized, short-span cantilevers and beam-column subassemblages for buildings. Epoxy injection improved the
original shear strength of the specimens, but the specimens
exhibited reduced overall stiffness following repair when
compared with the uncracked condition. At locations where
severe reinforcing bar-concrete bond degradation occurred,
the epoxy did not perform as well. Additional research by
Chung (1981) noted that epoxy injection of small 200 x
300x 2000 mm (7.9 x 11.8 x 78.7in.) reinforced concrete
beams was not an effective means to restore or improve reinforcing bar-concrete bond strength.
Basunbul et al. (1990) compared several flexural repair
methods for reinforced concrete beams, including epoxy
injection. Nine epoxy-injected specimens measuring 150x
150 mm (5.9 x 5.9 in.) in cross section and 1250 mm
(49.2 in.) in length, with longitudinal and transverse reinforcement, were loaded to induce varying degrees of flexural
damage. Vertical cracks were injected with epoxy resin and
allowed to cure before each specimen was loaded to failure.
The loads required to reinitiate cracking were observed to
be approximately 20% higher for the injected specimens
compared to the original member response.
In more recent years, the effects of environment and
fatigue have been included in studies of epoxy-injected specimens. Abu-Tair et al. (1991) investigated concrete beams
reinforced with transverse and longitudinal steel measuring
205 x 140 mm (8 x 5.5 in.) in cross section with a 2300 mm
ACI Structural Journal, Vol. 111, No. 2, March-April 2014.
MS No. S-2012-083, doi:10.14359.51686521, was received March 6, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

291

(90.6 in.) span. Seven specimens were loaded to failure in


flexure and then injected with epoxy resin. Several of the
specimens were loaded monotonically to failure, while
others were fatigue loaded at varying magnitudes. Additionally, three of the specimens (two monotonically and
one fatigue-loaded) were soaked in 38C (100F) water to
investigate the effect of water absorption on durability and
post-repair specimen performance. The results of the study
indicated that epoxy injection restored the original strength
and stiffness of the beams regardless of the loading conditions, while the four months of water immersion had insignificant effects on specimen durability, strength, or stiffness.
The preceding limited research on the structural performance of epoxy-injected reinforced concrete specimens has
focused on flexural response of reduced-sized specimens.
No data are available for the shear response of epoxy-injected reinforced concrete girders, and few researchers have
investigated loading conditions on the curing and bonding of
epoxy resin. Furthermore, reduced-sized specimens may not
accurately replicate strain fields and behavior of large reinforced concrete members. Other important issues to consider
are in-service loading responses and localized behavioral
effects, incorporation of service-induced diagonal cracks,
and the effects of temperature and shrinkage strains on
epoxy-injected member performance.
RESEARCH SIGNIFICANCE
Epoxy injection is widely used for remediation of cracked
RC bridges and other structures, but the efficacy of the
method on shear performance has not been established. An
experimental program was conducted using realistic fullscale bridge girders constructed with mid-twentieth century
design methods, details, and materials. Diagonal cracks were
produced under quasi-static loading. The girders were then
epoxy-injected and tested under different loading conditions
to determine the effects of epoxy injection on structural
performance. Research results improve the understanding
of the behavior of epoxy-injected diagonally cracked RC
girders and help engineers make better decisions regarding
rehabilitation alternatives.

Fig. 1Specimen reinforcing details and typical instrumentation placement.

EXPERIMENTAL PROGRAM
Test specimens
Five laboratory specimens were constructed and tested
to characterize the behavior and capacity of 1950s vintage
reinforced concrete deck girders with diagonal cracks after
being injected with epoxy resin. Previous work by Higgins et
al. (2004) identified standard details, materials, and proportions used in 1950s vintage bridge construction. Specimens
in the current study used an inverted-T (IT) configuration
to place the deck in flexural tension. This arrangement is
representative of negative moment in high-shear locations
near continuous supports such as piers and bent caps. Each
specimen had the following geometry: 1219 mm (48 in.)
overall height, a stem width of 356 mm (14 in.), and a flange
of 152 mm (6 in.) thick by 914 mm (36 in.) wide. Member
proportions and reinforcing steel are illustrated in Fig. 1. The

292

Fig. 2Specimen naming convention.


specimen naming convention used in the study is illustrated
in Fig. 2.
Longitudinal reinforcing steel consisted of ASTM A615/
A615M-05a Grade 420 No. 36 (Grade 60, No. 11) bars,
while transverse reinforcement consisted of Grade 300 No.
13 (Gr. 40, No. 4). Intermediate grade steel, with a yield
stress of 300 MPa (43.5 ksi), was typically used in 1950s
construction. However, this grade is not readily available
for large-diameter reinforcing bars. Therefore, Grade 240
(Grade 60) flexural bars were used but the area was reduced
ACI Structural Journal/March-April 2014

Table 1Concrete and steel material properties


Reinforcing steel
Concrete
Specimen

fc at failure, MPa (psi)

No. 13, Grade 300


fct at initial
loading, MPa (psi)

28.5 (4130)

3.2 (460) at
28-day

2-EC

36.2 (5250)

2.8 (410)

3-ED

28.3 (4104)

2.6 (377)

4-EL

29.5 (4279)

2.6 (377)

5-EA

35.4 (5141)

2.8 (410)

1-C

No. 36, Grade 420

fy, MPa (ksi)

fu, MPa (ksi)

fy, MPa (ksi)

fu, MPa (ksi)

350 (50.7)

544 (78.9)

477 (69.2)

712 (103)

492 (71.4)

741 (107)

484 (70.2)

728 (106)

473 (68.6)

694 (101)

492 (71.3)

741 (108)

357 (51.8)

570 (82.7)

Note: Values are for failure test unless otherwise indicated.

to produce the same tension resultant as the original designs


that used Grade 300 (Grade 40) steel. This results in a
smaller area of dowel steel in the specimens compared to the
original designs. Tension tests were performed according to
ASTM E8-04 to determine the reinforcing steel properties,
which are summarized in Table 1.
Concrete was provided by a local ready-mixed concrete
supplier. The concrete mixture design was based on 1950s
AASHO Class A concrete (Higgins et al. 2003). Specified compressive strength was 21 MPa (3000 psi), which is
comparable to the specified design strength in the original
1950s bridges. Actual concrete compressive strengths were
determined from 152 x 305 mm (6 x 12 in.) cylinders tested
for 28-day and day-of-test strengths in accordance with
ASTM C39M/C 39M-05 and ASTM C617-05. Day-of-test
concrete cylinder strengths for each specimen are shown in
Table 1. Split cylinder tests in accordance with ASTM C496/
C496M-04e1 were conducted the same day each specimen
was precracked, as reported in Table 1.
Instrumentation
Internal and external sensors were positioned on the specimens to record the local and global member responses.
Strain gauges were placed at midheight on the stirrups
located within the critical shear section near midspan. Additional strain gauges were mounted to the flexural reinforcement at midspan in the flexural-tension region. Diagonal
displacement sensors were placed at three locations along
the shear span, as shown in Fig. 1. Midspan displacement
and support settlements were also measured with additional
displacement sensors. The actual centerline displacement
presented in subsequent figures was calculated by removing
the support deformations from the overall centerline deformation during each load cycle. Typical instrumentation is
illustrated in Fig. 1.
Testing methodology
A simply-supported four-point loading configuration
was used with a span length of 6604 mm (260 in.) from the
centerline of supports. Force was applied from a hydraulic
actuator at a quasi-static rate of 8.9 kN/s (2.0 kip/s) and
was measured by a 2224 kN (500 kip) capacity load-cell. A
spreader beam distributed the applied actuator force to 102
mm (4 in.) wide plates spaced 610 mm (24 in.) symmetriACI Structural Journal/March-April 2014

cally about the midspan. Load was applied in incremental


steps followed by unloading, and repeating cycles until
failure. Load magnitudes increased each cycle by 222 kN
(50 kip). At each load peak, the load was reduced by 10%
to minimize creep effects while visible cracks were marked
and recorded.
Three tests were performed on each specimen: precrack,
baseline, and failure (with the exception of the control specimen, 1-C, which was loaded to failure in a single test). An
initial loading sequence, or precrack test, was performed
to produce diagonal cracks similar to those observed from
field inspections of RCDG bridges and of sufficient size for
epoxy injection. A target diagonal crack range of 0.65 to
1.25 mm (0.025 to 0.05 in.) was selected based on the earlier
work of Higgins et al. (2004). When diagonal cracks reached
a suitable size, the precrack loading cycle was terminated
and a baseline test was performed to establish a reference
for the specimens in the cracked condition for comparison to
the post-injection response. The baseline test of the cracked
specimens used the same loading sequence as the precrack
test described previously. In the failure test, specimens were
loaded to failure using the same load steps as the previous
two tests.
Injection and curing procedures
The epoxy resin selected is a commercial, two-part, ultralow-viscosity liquid epoxy. The specified material tensile
strength is 55.2 MPa (8000 psi). The surface sealant used
is a commerical, two-part, 100% solids epoxy. These two
materials are commonly used, are preapproved for use by
transportation agencies in several states, and are representative of similar epoxy materials. All injection materials were
provided by local suppliers. Additional installation guidance
was provided by qualified contractors to establish a repair
protocol that satisfied the installation recommendations of
the manufacturer. The procedure that was established is
summarized as follows.
The concrete surfaces around the diagonal cracks were
cleaned with a wire brush and vacuumed to remove loose
particles and dirt. The crack perimeter was sealed with the
surface epoxy and injection ports were surface-mounted
every 356 mm (14 in.), or roughly equal to the width of the
girder web. The surface epoxy cured for 24 hours before the
injection process was initiated. To allow for the release of
293

Table 2Specimen experimental summary


Specimen

VINITIAL, kN (kip)

VAPP, kN (kip)

VDL, kN (kip)

VR2K, kN (kip)

VEXP/VR2K

Failure mode

1-C

N/A

902 (203)

16.9 (3.80)

919 (207)

952 (214)

0.97

Shear-compression

2-EC

723 (162)

983 (221)

20.3 (4.56)

1001 (225)

983 (221)

1.02

Shear-compression

3-ED

778 (175)

992 (223)

17.2 (3.87)

1009 (227)

965 (217)

1.05

Shear-compression

4-EL

778 (175)

1046 (235)

16.7 (3.75)

1063 (239)

947 (213)

1.12

Shear-compression

5-EA

778 (175)

1112 (250)

18.4 (4.14)

1130 (254)

943 (212)

1.20

Shear-compression

entrapped air, diagonal cracks were injected starting from


the lowest port working up the crack. Window ports were
placed on the backside of the beam to serve as a visual aid
for assurance of epoxy penetration through the beam web.
A specialized injection machine, commonly used by local
contractors, was needed to mix the two-part liquid epoxy
in the proportions recommended by the manufacturer and
to deliver the mixture into the beam under low pressure.
Each port was injected to a maximum pressure of 690 kPa
(100psi). As liquid epoxy began to seep from the next
higher injection port, the lower port was capped and the
injection nipple was moved to the next position. Near the
top of each diagonal crack, the injection pressure climbed
more quickly to 690 kPa (100 psi), and would take longer to
dissipate, signaling that there was little available void space
to pump additional resin. When the maximum pressure was
maintained, the final port was capped. After injection, all
specimens were allowed to cure for at least 7 days. A heated
plastic enclosure was placed around the specimen to ensure
that temperatures were maintained above 4.5C (40F), as
recommended by the epoxy manufacturers. Thermocouples
outfitted with data loggers were placed into a small void cast
into the end of the specimens and on the exterior to record
temperatures throughout the curing cycle.
Specimen variables
To simulate the effects of different in-service stress conditions, each specimen was subjected to a distinct loading
scenario during the injection and curing phases. Specimen
2-EC was injected and cured with no applied loads other
than specimen self-weight. Simulated superstructure dead
load was applied to Specimen 3-ED before epoxy injection. A total load of 356 kN (80 kip) was applied to induce a
service level dead load shear of 178 kN (40 kip). This shear
magnitude is representative of an interior girder for a typical
1950s vintage three-span continuous RCDG bridge having
15.2 m (50 ft) spans and a uniform dead load of 23.3 kN/m/
girder (1.6 kip/ft/girder).
Varying live load stress was applied to Specimen 4-EL in
addition to the superstructure dead load. The live loading
was representative of average shear magnitudes produced by
ambient traffic and a fully loaded AASHTO Type 3-3 unit
truck having five axles and a gross vehicular weight of 356
kN (80 kip) moving across the bridge described for Specimen 3-ED. Realistic shear distribution factors developed
by Potisuk and Higgins (2007) from field studies were used
294

VEXP, kN (kip)

Mean

1.07

Coefficient of variation

0.08

to determine the live load magnitude. Force was applied at


0.3 Hz with an amplitude of 160 kN (36 kip) and a mean
of 463 kN (104 kip). This loading amplitude represents the
maximum girder shear caused by the dead load of the bridge
and the truck live load with impact, as well as a minimum
force resulting from hogging due to live load moving onto
an adjacent span. The loading rate represents the truck traveling over the prototype bridge at approximately 33.8 kph
(21 mph), which was controlled by the hydraulic loading
system in the laboratory.
The fifth specimen, 5-EA, had simulated locked-in drying
and thermal shrinkage strains induced by applying a uniform
tension load to the specimen. The axial load was applied
to a level of approximately 890 kN (200 kip) before the
initial precrack transverse loading cycles began. The axial
load was held at a constant magnitude of 645 kN (145 kip)
during the injection and curing phases, and then returned to
890 kN (200 kip) for the post-injection failure loading. For
additional detail on the axial force application and loading
protocol, refer to Smith (2007).
EXPERIMENTAL RESULTS
The performance of the epoxy-injected specimens was
evaluated through the shear-midspan deflection and sheardiagonal displacement responses, flexural and shear reinforcement strains, and crack deformations. The data
collected from the three phases of testing were compared to
assess the local and global responses before and after epoxy
injection. Results were also compared with an otherwise
similar un-injected specimen. All of the specimens exhibited
shear-compression failures and the applied shear at failure
for each specimen is summarized in Table 2.
Shear-midspan displacement response
Applied shear-midspan displacement responses are shown
in Fig. 3 and demonstrate the overall specimen behavior of
the initial, baseline, and post-injection tests. The post-injection response of each injected specimen showed decreased
residual deformations and greater stiffness during the first
two or three load steps as compared to the baseline response.
As the applied shear magnitudes increased, the specimens
began to soften due to the development of new cracks, often
adjacent to the repaired diagonal cracks. Specimens 3-ED
and 4-EL were similar, especially in the service load range
indicated in Fig. 3, and both had greater stiffness than Specimen 2-EC.
ACI Structural Journal/March-April 2014

Fig. 3Applied shear-midspan displacement response.


shear decreased, the specimen shortened along the axis of
the axial loading apparatus and the axial tension increased
again. Like Specimen 4-EL, Specimen 5-EA had decreased
residual deformations for many of the load cycles and did
not begin to soften until near failure.

Fig. 4Applied shear-axial load variability of Specimen


5-EA.
The axially loaded specimen exhibited a unique shear-midspan displacement response. The curve has a slender S-shape
resulting from the specimen stiffening and then softening
during each load cycle. The axial load as a function of the
applied shear for Specimen 5-EA is displayed in Fig. 4. As
the applied shear increased, the specimen length increased
at the level of the axial apparatus, thereby reducing the
hydraulic pressure in the axial load actuators, and thus
reducing the externally applied axial tension. As the applied
ACI Structural Journal/March-April 2014

Shear-diagonal displacement response


The post-injection diagonal displacement data are shown
in Fig. 5. All of the post-injection tests for the epoxy specimens had greater stiffness and smaller permanent deformations during the initial load steps than the control specimen.
Specimens 3-ED and 4-EL were stiffer than Specimen 2-EC,
with Specimen 4-EL performing slightly better than 3-ED.
Specimen 5-EA showed significantly improved stiffness and
reduced permanent deformations after unloading following
epoxy injection. For all the specimens except 4-EL and
5-EA, the north and south diagonal deformations were
essentially identical, showing similar stiffness and exhibiting increasing diagonal deformation at approximately the
same shear magnitude. For the remaining two specimens,
however, one end of the specimen produced a substantially
larger diagonal deformation than the other side.
Diagonal cracking behavior
The orientation and location of the diagonal cracks
produced during precrack and post-injection loading
295

Fig. 5Control and post-injection shear-diagonal displacement response recorded near centerline of specimens.
sequences are shown in Fig. 6. The locations of diagonal
cracks that were epoxy-injected are also shown in Fig. 6.
Diagonal cracks that were injected did not reopen during
post-injection tests. Instead, new cracks formed adjacent to
the injected cracks and propagated at similar angles. Non-repaired cracks propagated along the original paths.
The applied loads required to reinitiate diagonal cracking
are shown in Fig. 7. The load required to reinitiate diagonal
cracking was determined from the applied shear magnitude
at the moment the stirrup strain showed an abrupt increase.
The pre- and post-injection diagonal cracking shears were
compared with the precrack diagonal cracking shear on the
abscissa and the post-injection cracking shear serving on
the ordinate. Solid symbols represent stirrup strain gauges
located near injected diagonal cracks, while hollow symbols
represent stirrup strain gauges located away from injected
diagonal cracks, as described in the next section. Injected
diagonal cracks required higher applied shear than the original specimen to produce strains in the stirrups due to new
diagonal cracking. Data above the reference line show that
larger shear loads were required, while points below the line
required smaller loads to propagate or reinitiate diagonal
cracking. Non-injected cracks typically behaved similarly
to the baseline tests, where stirrup strains began increasing
immediately upon application of applied shear.

296

Reinforcement strains
The largest relative influences of epoxy injection were
seen in the individual stirrup strains, but these effects were
highly influenced by the proximity of injected diagonal
cracks to the embedded strain gauge locations. Strain gauges
located near diagonal cracks that were injected had lower
strains after injection at similar shear magnitude. Strain
gauges located between diagonal cracks or far from cracks
that were not injected displayed relatively little change. This
behavior was observed for all injected specimens. Diagonal
cracks were considered to be near the stirrup strain gauge
if the vertical distance that the crack crossed the stirrup with
the strain gauge was within the AASHTO-LRFD (AASHTO
2005) calculated development length of the Grade 300 No.13
(Grade 40, No. 4) stirrup (203 mm [8 in.]). An example of
strains measured for a stirrup located near a diagonal crack
and a stirrup located at a distance greater than the development length from a diagonal crack is shown in Fig. 8.
The strain behavior depicted in Fig. 9 shows the baseline
and post-injection stirrup strains at the maximum service
load range. In the figure, the baseline strains serving as the
abscissa are plotted against the post-injection strains on
the ordinate. Solid symbols represent stirrup strain gauges
located near injected diagonal cracks, while hollow symbols
represent stirrup strain gauges located away from injected
diagonal cracks, as defined previously. The dashed referACI Structural Journal/March-April 2014

Fig. 6Crack pattern locations on east face of specimens. Figure includes precracking, epoxy-injected cracks, post-injection
cracks, and final failure crack.
ence line marks the boundary between improved and unimproved behavior. Points above the line had higher strains
at the same service load after injection, whereas the points
below the line had lower strains after injection. Most stirrup
strains near injected diagonal cracks showed significantly
reduced strains after injection, whereas uninjected regions
were generally unaffected.

ACI Structural Journal/March-April 2014

Interaction of epoxy curing process and cyclic live


load
Data were collected continuously for Specimen 4-EL
during curing of the epoxy with cyclic service-level live
load being applied. An example of the diagonal deformation
throughout the first 3 days of curing is shown in Fig. 10. The
deformation range of a representative displacement sensor
crossing the injected diagonal crack is shown in Fig. 11.
As seen in Fig. 10, within the first several hours of curing,
297

the average diagonal crack deformation reduced from 1.39


to 1.27 mm (0.0547 to 0.0500 in.) and remained relatively
constant for the remainder of the curing process. However,
the diagonal deformation range continued to reduce over
the period of about 3 days, during which the deformation
range decreased by nearly 75% (Fig. 11). An example of the
stirrup strain throughout the curing period and the internal
and external temperature recordings is shown in Fig. 12. The

stirrup is located in the same section as the example diagonal


deformation shown in Fig. 11, and the strain range decreased
by over 50% within the first 18 hours, while little additional effects were observed for the remainder of the curing
process. The curing time reported by the epoxy manufacturer is 7 days at 4C (40F) and 2 days at 25C (77F). The
average curing temperature for Specimen 4-EL was 10C
(50F), which correlates to a required curing time of approximately 6 days. The stirrup strain was also observed to fluctuate with the external temperature with significant time lag
between surface temperature and strain changes (Fig. 12).
DISCUSSION
The results of this study indicate that epoxy injection affected the structural behavior of the RC specimens
in several ways. Overall, the most dramatic effects were
observed for Specimens 3-ED, 4-EL, and 5-EA, as described
in the following discussion.

Fig. 7Applied shear at diagonal cracking initiation before


and after injection.

Shear capacity of specimens


In this study, a computer program call Response 2000
(R2K) (Bentz 2000) was used to estimate the strength of the
specimens (neglecting any influence of the epoxy injection)
as well as the strength of the control specimen. In a previous
study, R2K, which uses the Modified Compression Field
Theory (MCFT) (Vecchio and Collins 1986), was used to
predict shear capacity for a series of 31 similar full-size RC
specimens within 0.98 of the actual capacity with a coeffi-

Fig. 8Applied shear-stirrup strain behavior for Specimen 3-ED. Strain gauges located near: (a) injected diagonal cracks;
and (b) uninjected diagonal cracks.
298

ACI Structural Journal/March-April 2014

Fig. 11Diagonal deformation range during curing of


Specimen 4-EL. (Note: Zero was taken as a reference value.)
Fig. 9Stirrup strains at service level shear before and after
injection. Shear magnitude taken as 311 kN (70 kip).

Fig. 10Diagonal deformations reduced during curing of


Specimen 4-EL.
cient of variation under 8% (Higgins et al. 2004). The R2K
predicted member capacities are shown in Table 2. The
experimental shear strength is the applied actuator force at
failure combined with the self-weight of the specimen acting
at the failed section. Except for two specimens, the epoxyinjected specimens exhibited slightly larger shear capacities
than predicted, ranging from 1.02 to 1.12. However, all but
4-EL and 5-EA fall within a standard deviation (68% prediction interval) of the expected shear strength ignoring the
effects of epoxy injection. This is within the expected variability of shear testing results. Specimens 4-EL and 5-EA
exhibited shear strengths significantly above the predicted
unaltered shear capacity. The shear strength of Specimen
4-EL was above the 95% prediction interval, and Specimen
5-EA was above the 99% prediction interval for the expected
shear strength without epoxy injection and indicates the
increase is not likely attributed to the inherent variability of
shear strength testing of large reinforced concrete girders of
the type studied herein. These two specimens received the
least amount of epoxy injection in comparison to the other
specimens, because they only exhibited three major diagonal
crack systems injected on each specimen. Axial load was
accounted for in the R2K capacity prediction of Specimen
5-EA but not 4-EL, and the differences in shear capacity for
each of the specimens is described subsequently.
ACI Structural Journal/March-April 2014

Fig. 12Stirrup strain during curing of Specimen 4-EL.


(Note: C = [F 1.8] + 32.)
Interaction of epoxy repair and superimposed
dead load
Specimen 3-ED exhibited a higher capacity than the
predicted baseline capacity. Specimen 3-ED also achieved
higher load prior to reinitiation of nonlinear response
compared to the epoxy-injected control specimen, 2-EC. The
dead load serves to keep the diagonal cracks open, allowing
for more penetration of the epoxy at the crack tips. It further
allows the epoxy to only carry superimposed live loads and
leaves dead load stresses locked into the reinforcement and
concrete. This tends to further delay crack reinitiation. This
was observed by the reduced stirrup steel demand at injected
diagonal cracks for otherwise similar load levels.
Cured epoxy characteristics following cyclic live
load
The live load magnitudes, rates, and curing conditions
considered in this program for Specimen 4-EL did not
reduce the effectiveness of the epoxy injection compared to
the specimen with dead load alone and, in fact, resulted in
higher observed shear capacity. The cyclic loading acted as
an internal pumping mechanism that enabled the epoxy to
enter and fill finer cracks than in either Specimen 3-ED or
Specimen 2-EC. It was observed during the injection process
of Specimen 4-EL that the epoxy pump pressures built and
dissipated in-phase with the actuator loading cycle, which is
299

Fig. 13Photographs of cores taken from Specimens 3-ED


and 4-EL: (a) core taken from Specimen 3-ED; and (b) core
taken from Specimen 4-EL with small voids.
consistent with the epoxy being pushed out of the diagonal
cracks as the cracks closed upon unloading.
Concrete cores measuring 102 mm (4 in.) in diameter
were taken from epoxy-injected diagonal cracks for both
Specimens 3-ED and 4-EL after failure. Both cores showed
that the epoxy was well distributed through the cracks and
even filled hairline subcracks within the cored region. The
core taken from Specimen 4-EL had small visible voids,
which were evidence of bubble formations likely caused by
the cyclic loading noted previously. Examples of the porous
epoxy matrix observed in Specimen 4-EL compared to the
solid epoxy matrix observed in Specimen 3-ED are shown
in Fig. 13. The development of bubbles within the epoxy did
not diminish the performance of Specimen 4-EL compared
to Specimen 3-ED. It is important to note that the cyclic live
loading was representative of loads moving across a typical
15.2 m (50 ft) span continuous bridge at an approximate
speed of 33.8 kph (21 mph). Additionally, due to setting of
the epoxy at a larger average crack width (dead load plus
average live load range), as compared to Specimens 2-EC
and 3-ED, the cross section was effectively induced with
compressive stresses that must be overcome before additional cracking may occur. This post-tensioning effect also
accounts for the increased shear strength observed for Specimen 4-EL. Further research is needed to study the effects
of higher loading rates and other load magnitudes applied
simultaneously during epoxy injection, as well was possible
lower-range curing temperatures. However, based on these
observations, it may be possible to inject cracks on existing
bridges while positioning static superimposed live loads on
the bridge (such as loaded maintenance trucks) to effectively
open the cracks to their maximum service-level width. After
curing the epoxy and removing the live load, compressive
stresses would be induced in the cross section. If the live
load used during repair is above the maximum expected
service loads, this could effectively prevent or significantly
delay future cracking.
Interaction of epoxy repair and axial load
The axially loaded specimen, 5-EA, exhibited the most
dramatic change between the pre- and post-injection
response. The specimen was injected with a constant externally applied axial tension force of 645 kN (145 kip), which
coincided with the load magnitude at the end of the precrack
test. The applied axial force maintained larger diagonal and
vertical cracks during injection, allowing for increased pene300

tration of the epoxy into microcracks within the concrete


matrix, similar to Specimens 3-ED and 4-EL. More importantly, as the vertical loading was applied, the axial force
decreased during the failure test to a magnitude of 267 kN
(60 kip) at ultimate load, resulting in a net compressive force
of 378 kN (85 kip) induced into the epoxy-injected section.
Had the axial load not diminished with increasing transverse
load, the specimen may have failed at lower load levels.
R2K predicted a capacity of 853 kN (192 kip) for a similar
specimen with 890 kN (200 kip) total axial tension force.
Reducing the steel yield stress by an amount equivalent to
the 267 kN (60 kip) axial tension and applying a 378 kN (85
kip) axial compression force on the section, R2K estimated
a shear capacity of 987 kN (222 kip), which is closer to the
observed shear capacity. This situation of loading and curing
would represent a structure with shrinkage and/or temperature-induced tensile strains that are recovered after epoxy
injection. The strain recovery (release of restraints at supports
or temperature change, for example) produces a post-tensioning effect for the injected cross section. However, the
beneficial temperature effect could not be relied upon in the
field and would vary during daily and seasonal changes.
CONCLUSIONS
Five RC deck girder specimens were fabricated to reflect
the design and construction materials of the 1950s for
RCDG bridges lightly reinforced for shear. The specimen
tests were designed to study the effects of epoxy on diagonal-tension, shear-dominated cracked girders. Specimens
were precracked to similar levels observed in the field,
subjected to baseline tests in the cracked condition, injected
with epoxy resin at varying levels of applied dead and/or
live load, and then, after curing the epoxy, were tested to
failure. The results of the initial cracking, baseline, and postinjection responses were compared. Factors included in the
study were superimposed dead load, service live load plus
dead load, and externally applied axial tension during epoxy
curing. Based on the experimental observations and analytically predicted shear strengths, the following conclusions
are presented:
Most of the epoxy-injected specimens exhibited
marginal increases in shear strength compared to
well-correlated analytically predicted strengths of
unrepaired specimens. The largest capacity increases
were observed for specimens subjected to superimposed cyclic live load and externally applied static axial
tension during epoxy curing (which was due principally
to an unintended post-tensioning effect). These showed
that the epoxy injection did not effectively strengthen
the specimens, as they would have failed very close to
the observed capacity with or without epoxy injected
cracks.
Superimposed cyclic live loading during injection and
curing of Specimen 4-EL produced dynamic pressure
during injection and pumping of the epoxy within the
diagonal cracks. Fine bubbles were formed within
the epoxy matrix but did not reduce structural performance at service or ultimate states. In addition, the
curing of the epoxy at larger average diagonal crack
ACI Structural Journal/March-April 2014

widths introduced compressive stresses in the cross


section that increased the threshold load for recracking.
The compressive stresses in the stem also resulted in
increased shear capacity. Consequently, this finding
may permit strategic placement of loaded maintenance
trucks to widen cracks at certain locations during the
injection and curing process.
Initial stiffness after injection was improved and development of residual deformations was delayed prior
to recracking by epoxy injection compared with the
cracked performance prior to injection.
Epoxy injection increased the threshold load level
required to form additional diagonal cracks or extend
the existing cracks within the stem.
Injected diagonal cracks did not reopen. Instead, new
cracks formed adjacent to the original injected cracks.
Epoxy injection reduced service-level stirrup strains
compared to uninjected diagonal cracks prior to
recracking. This may reduce bond fatigue, thereby
slowing or preventing additional diagonal crack growth
and help maintain force transfer across diagonal cracks.
Careful and methodical epoxy installation procedures following industry best practices enabled epoxy
penetration through the depth of the web. Based on
the observed penetration and uniformity of the epoxy
within the diagonal cracks, it is likely that the epoxy
would restrict access of moisture and chlorides to the
embedded stirrups and thereby delay or diminish corrosion potential at the injected crack locations.
ACKNOWLEDGMENTS

This research was funded by the Oregon Department of Transportation.


The findings and conclusions are those of the authors and do not necessarily
reflect those of the project sponsors.

AUTHOR BIOS

Matthew T. Smith is an Associate Engineer at CH2M-Hill, Corvallis,


OR. He received his BS and MS from Oregon State University, Corvallis,
OR. His research interests include the design of hydraulic and building
structures.
Daniel A. Howell is a Project Manager with the Bridge Design Division of
the St. Louis County Department of Highways and Traffice, Creve Coeur,
MO. He received his BS and MS from the University of Delaware, Newark,
DE, and his PhD from Oregon State University. His research interests
include the design of bridges and other structures.
Mary Ann T. Triska is a Bridge Engineer at HDR Engineering, Portland,
OR. She received her BS from the University of Portland, Portland, OR,
and her MS from Oregon State University. Her research interests include
evaluation, rehabilitation, and sustainable design of infrastructure.
Christopher Higgins is the Slayden Construction Faculty Fellow and
Professor of Structural Engineering in the School of Civil and Construction Engineering at Oregon State University. He received his BS from
Marquette University, Milwaukee, WI; MS from the University of Texas
at Austin, Austin, TX; and PhD from Lehigh University, Bethlehem, PA.
His research interests include evaluation and rehabilitation of bridges and
other structures.

ACI Structural Journal/March-April 2014

fc
fct
fu
fy
VAPP
VDL

NOTATION
=
=
=
=
=
=

compressive strength of concrete, MPa (psi)


tensile strength of concrete, MPa (psi)
ultimate tensile stress of reinforcing steel, MPa (ksi)
yield stress of reinforcing steel, MPa (ksi)
applied shear from actuator, kN (kip)
applied shear from portion of self-weight acting at failure plane,
kN (kip)
VEXP = total applied shear, kN (kip)
VINITIAL = maximum applied shear during precracking phase of testing, kN
(kip)
VR2K = predicted shear capacity, kN (kip)

REFERENCES

AASHTO, 2005, AASHTO-LRFD Bridge Design Specification, third


edition with 2005 interims, American Association of State Highway and
Transportation Officials, Washington, DC, 654 pp.
Abu-Tair, A. I.; Rigden, S. R.; and Burley, E., 1991, The Effectiveness
of the Resin Injection Repair Method for Cracked RC Beams, The Structural Engineer, V. 69, No. 19, Oct., pp. 335-341.
ASTM A615/A615M-05a, 2005, Standard Specification for Deformed
and Plain Carbon-Steel Bars for Concrete Reinforcement, ASTM International, West Conshohocken, PA, 6 pp.
ASTM C39/C39M-05, 2005, Standard Test Method for Compressive
Strength of Cylindrical Concrete Specimens, ASTM International, West
Conshohocken, PA, 7 pp.
ASTM C496/C496M-04e1, 2004, Standard Test Method for Splitting
Tensile Strength of Cylindrical Concrete Specimens, ASTM International,
West Conshohocken, PA, 5 pp.
ASTM C617-05, 2005, Standard Practice for Capping Cylindrical
Concrete Specimens, ASTM International, West Conshohocken, PA, 6 pp.
ASTM E8-04, 2004, Standard Test Methods for Tension Testing of
Metallic Materials, ASTM International, West Conshohocken, PA, 28 pp.
Basunbul, I. A.; Gubati, A. A.; Al-Sulaimani, G. J.; and Baluch, M. H.,
1990, Repaired Reinforced Concrete Beams, ACI Materials Journal,
V.87, No. 4, July-Aug., pp. 348-354.
Bentz, E. C., 2000, Section Analysis of Reinforced Concrete Members,
PhD thesis, Department of Civil Engineering, University of Toronto,
Toronto, ON, Canada.
Chung, H. W., 1975, Epoxy-Repaired Reinforced Concrete Beams,
ACI Journal, V. 72, No. 5, May, pp. 233-234.
Chung, H. W., 1981, Epoxy Repair of Bond in Reinforced Concrete
Members, ACI Journal, V. 78, No. 1, Jan.-Feb., pp. 79-82.
Higgins, C.; Farrow III, W. C.; Potisuk, T.; Miller, T. H.; Yim, S. C.;
Holcomb, G. R.; Cramer, S. D.; Covino, B. S.; Bullard, S. J.; ZiomekMoroz, M.; and Matthes, S. A., 2003, SPR 326 Shear Capacity Assessment of Corrosion-Damaged Reinforced Concrete Beams, Oregon Department of Transportation, Salem, OR, 19 pp.
Higgins, C.; Miller, T. H.; Rosowsky, D. V.; Yim, S. C.; Potisuk, T.;
Daniels, T. K.; Nicholas, B. S.; Robelo, M. J.; Lee, A.-Y.; and Forrest,
R.W., 2004, Research Project SPR 350 SR 500-91: Assessment Methodology for Diagonally Cracked Reinforced Concrete Deck Girders, Oregon
Department of Transportation, Salem, OR, Oct., 328 pp.
Popov, E. P., and Bertero, V. V., Oct.-Dec. 1975, Repaired R/C
Members Under Cyclic Loading, Earthquake Engineering & Structural
Dynamics, V. 4, No. 2, pp. 129-144.
Potisuk, T., and Higgins, C., 2007, Field Testing and Analysis of CRC
Girder Bridges, Journal of Bridge Engineering, V. 12, No. 1, Jan.-Feb.,
pp. 53-63.
Rooney, H. A., 1963, Epoxy Resins as a Structural Repair Material, State of California Department of Public Works Division of
Highways, Jan., 15 pp. (http://www.dot.ca.gov/hq/research/researchreports/1961-1963/63-23.pdf)
Smith, M. T., 2007, Investigation of the Behavior of Diagonally Cracked
Full-Scale CRC Deck-Girders Injected with Epoxy Resin and Subjected to
Axial Tension, MS thesis, Oregon State University, Corvallis, OR. (http://
scholarsarchive.library.oregonstate.edu.)
Tremper, B., 1960, Repair of Damaged Concrete with Epoxy Resins,
ACI Journal, V. 57, No. 2, Feb., pp. 173-182.
Vecchio, F. J., and Collins, M. P., 1986, The Modified Compression
Field Theory for Reinforced Concrete Elements Subjected to Shear, ACI
Journal, V. 83, No. 2, Feb., pp. 219-231.

301

NOTES:

302

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S26

High-Performance Fiber-Reinforced Concrete Bridge


Columns under Bidirectional Cyclic Loading
by Ady Aviram, Bozidar Stojadinovic, and Gustavo J. Parra-Montesinos
An experimental and analytical study was carried out on circular
column specimens representing cantilever bridge piers constructed
with tensile strain-hardening, high-performance fiber-reinforced
concrete (HPFRC). Two column specimens with different longitudinal reinforcement details in the expected plastic hinge zone were
tested under bidirectional displacement reversals, and the results
were compared with those of a geometrically identical specimen
constructed using regular concrete and designed according to
current Caltrans bridge design specifications. The results demonstrate
the considerable benefits of using tensile strain-hardening fiberreinforced concrete in typical highway overpass bridge columns.
These benefits include: improved cyclic response; substantial
reduction in transverse reinforcement and corresponding construction benefits; and increased damage tolerance and reduction of
post-earthquake repair costs.
Keywords: bidirectional cyclic load; bond stress; bridge column; drift;
fiber-reinforced concrete; plastic hinge; shear; steel fibers.

INTRODUCTION
Numerous cast-in-place and precast reinforced concrete
structures suffered significant damage or collapse during
historical and recent earthquakes, primarily due to deficient
structural design. Code-mandated reinforcement detailing
required for critical bridge and building members to ensure
adequate seismic behavior often leads to substantial reinforcement congestion and construction difficulties. Therefore, it is not surprising that many recent research efforts
have been directed to the development and implementation of innovative materials in new structures for improved
seismic performance while simplifying the required
reinforcementdetailing.
Fiber-reinforced concretes that exhibit a tensile strain-hardening behavior are now possible with the use of relatively low
fiber-volume fractions Vf (in the range of 1.5 to 2.0%). These
tensile strain-hardening materials are typically referred to as
high-performance fiber-reinforced concrete (HPFRC). In
addition to their tensile strain capacity, which often exceeds
0.5%, HPFRCs exhibit a compression response that resembles that of well-confined concrete. Hooked and twisted
steel fibers, as well as ultra-high-molecular-weight polyethylene fibers, are among the fiber types investigated for use
in earthquake-resistant construction.1 When used in structural elements subjected to large displacement reversals,
HPFRCs enable significant deformation capacity with superior damage tolerance compared with geometrically identical, well-detailed reinforced concrete members.1-5 Further,
substantial reductions in transverse reinforcement required
for confinement and shear resistance have been possible in
elements subjected to large shear stress reversals.1 There is
ACI Structural Journal/March-April 2014

substantial experimental evidence that supports the use of


HPFRC materials to enhance structural response of elements
subjected to large deformation demands caused by ground
motions, such as bridge piers with either flexural-dominated
behavior or with strong flexure-shear interaction.
Evaluation of the use of HPFRC in bridge piers to
substantially relax transverse reinforcement requirements
while leading to increased damage tolerance, shear strength,
and energy dissipation under cyclic loading compared with
regular concrete piers was the main focus of this research.
An experimental and analytical study was carried out on
two approximately 1/4-scale column specimens built with
HPFRC and subjected to bidirectional displacement reversals. The behavior of these specimens was compared with
that of a geometrically identical conventionally reinforced
concrete (RC) column.6 Additional information about the
tests, calibrated finite element models, and the repair cost and
repair time analysis of typical highway bridges in California
constructed using HPFRC columns can be foundelsewhere.7
RESEARCH SIGNIFICANCE
Experiments were performed to characterize the cyclic
response of circular HPFRC bridge columns subjected to
bidirectional lateral displacements and to compare their
response with that of conventionally reinforced concrete
columns. This experimental study is one of the first of its
kind, and was aimed at assessing the effectiveness of fiber
reinforcement as partial replacement of transverse reinforcement used for shear resistance and confinement while
increasing column flexural ductility. The experiments
performed also allowed an evaluation of the enhanced
damage tolerance of HPFRC columns, which is important
for reducing post-earthquake repair cost and repair time
assessment, and performance-based evaluation of structural
systems using HPFRC.
TEST PROGRAM
The column specimens represented the bottom half of a
typical bridge column deforming in double curvature with
an assumed inflection point at midheight. The specimen
geometry was selected so as to represent circular column
prototypes used by Caltrans for typical highway overpass
bridges in California. The length scale of the specimens was
approximately 1/4. Columns of both HPFRC specimens,
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-092, doi:10.14359.51686522, was received March 10, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

303

Table 1Summary of specimen geometry, reinforcement, and material properties


Parameter

S1: HPFRC

S2: HPFRC

BC: Plain

Column diameter Dcol

400 mm (16 in.)

400 (16 in.)

400 mm (16 in.)

Total column height Hcol

1625 mm (64 in.)

1625 mm (64 in.)

1625 mm (64 in.)

Shear span-to-diameter ratio

Longitudinal reinforcement

12 No. 4/13M + 8 No. 4/13M


dowels

12 No. 4/13M + 8 No. 4/13M dowels

12 No. 4/13M

Longitudinal reinforcement ratio l

2% (base), 1.2% (rest)

2% (base), 1.2% (rest)

1.2%

Debonding sleeves

Dowels, L = 250 mm
(10 in.), 250 mm (10 in.) above base

Main reinforcing bar, L = 100mm (4


in.), 200 mm (8 in.) above base

Transverse reinforcement

W3.5* at 64 mm (2.5 in.)

W3.5 at 64 mm (2.5 in.)

W3.5 at 32 mm (1.25 in.)

Volumetric transverse reinforcement ratio s

0.75%

0.75%

1.5%

Specified concrete compressive strength

34.5 MPa (5.0 ksi)

34.5 MPa (5.0 ksi)

34.5 MPa (5.0 ksi)

Days from concrete casting to test date

60

49

49

Concrete compressive strength at test day

47.3 MPa (6.86 ksi)

47.1 MPa (6.83 ksi)

35.1 MPa (5.09 ksi)

W3.5: diameter = 5.3 mm (0.21 in.).

denoted as S1 and S2 in Table 1, were built using the same


HPFRC material, reinforced with a 1.5% volume fraction of
commercially available high-strength (2410 MPa [350ksi])
hooked steel fibers with a length of 30 mm (1.2 in.), and
an aspect ratio of 80. A fiber-volume fraction of 1.5%
(120 kg/m3 [200 lb/yd3]) was selected such that a tensile
strain-hardening behavior could be achieved. Ready mix
concrete was used in both columns, and the fibers were added
to the concrete truck on site. HPFRC casting was performed
using a bucket and stick vibrator. The results from the test of
a reference base column specimen (BC), designed according
to the Caltrans Seismic Design Criteria (SDC)8 and built
with regular concrete, were used for comparison purposes.6
All three specimens were tested following the same predetermined bidirectional cyclic displacement pattern.
The transverse reinforcement of both HPFRC specimens was approximately half of that required by Caltrans
Seismic Design Criteria (SDC),8 counting on the HPFRC
material to contribute significantly to shear resistance and
confinement. The longitudinal reinforcement of the columns
in Specimens S1 and S2, shown in Fig. 1, was detailed to
prevent concentration of inelastic deformation at the cold
joint between the HPFRC column and the base block, which
was constructed with regular concrete. Such cold joint is
typical for conventional reinforced concrete bridge columns
in California. Concentration of deformations at this joint is
not desirable because it can lead to a premature sliding shear
failure at the base of the column, fracture of column longitudinal reinforcement, or both. The cold joint at the base of
HPFRC columns is more vulnerable to damage localization than a similar joint in reinforced concrete construction
because of the higher flexural strength of the immediately
adjacent HPFRC column sections and the excellent bond
between HPFRC and reinforcing bars.9 Thus, the HPFRC
column-foundation interface of the test specimens was
strengthened with dowel reinforcement to force most of the
inelastic deformations to occur within the HPFRC column.

304

Two different debonding schemes, shown in Fig. 1, were


devised to minimize damage localization at the section where
the dowel reinforcement was terminated. In SpecimenS1,
the upper portion of the dowels was debonded using plastic
tubes to avoid premature damage localization that could
occur because of the termination of the bars within the
plastic hinge zone. In Specimen S2, the dowels were terminated within the plastic hinge region, and the main longitudinal bars were debonded over a length of 4 in. (10 cm) to
prevent large strain concentration and premature reinforcing
bar fracture at the section where the dowels wereterminated.
Loading pattern
The HPFRC specimens were tested using a bidirectional
circular load pattern (Fig. 2). This pattern was similar to that
used for the BC specimen.6 A cycle in this pattern comprised
circles, one clock-wise and the other counterclockwise,
while return paths were sequenced to minimize the bias in
any particular loading direction; however, this resulted in
a very demanding displacement pattern characterized by
long specimen travel along circles with constant ductility
demand. The quasi-static cyclic tests were conducted by
incrementally increasing the radius of the circular pattern.
The target displacement ductility demand for each cycle in
the displacement history is presented in Table 2. The target
ductility demand, termed nominal ductility demand , was
computed with respect to the yield displacement of the BC
column, estimated at 14.0 mm (0.55 in.) (0.86% drift), to
enable a direct comparison between the BC and HPFRC
specimens. Each post-yield primary cycle (that is, cycle to a
new maximum drift level) was followed by a small displacement cycle with amplitude equal to 1/3 of that of the primary
cycle to evaluate the column stiffness degradation throughout
the loading history. Similar displacement rates among the
different test cycles were applied, not exceeding 25 mm/min
(1.0 in./min). A gravity load equivalent to 0.1fcAg of the
BC column, where Ag is the column gross cross-sectional
area, was applied at the column top through a spreader beam
ACI Structural Journal/March-April 2014

Table 2Nominal ductility level used at each cycle of displacement history, defined with respect to BC
column yield displacement
Cycle

Nominal ductility

Displacement, mm (in.), (drift, %)

Cycle

Nominal ductility

Displacement, mm (in.), (drift, %)

NA

1.0 mm (0.04 in.) (0.07)

10

41.9 mm (1.65 in.) (2.6)

NA

2.8 mm (0.11 in.) (0.17)

11

14.0 mm (0.55 in.) (0.86)

NA

5.6 mm (0.22 in.) (0.34)

12

4.5

63.0 mm (2.48 in.) (3.9)

14.0 mm (0.55 in.) (0.86)

13

1.5

21.1 mm (0.83 in.) (1.3)

0.33

4.6 mm (0.18 in.) (0.28)

14

6.25

87.4 mm (3.44 in.) (5.4)

1.5

21.1 mm (0.83 in.) (1.3)

15

27.9 mm (1.10 in.) (1.7)

0.5

7.1 mm (0.28 in.) (0.43)

16

112 mm (4.40 in.) (6.9)

27.9 mm (1.10 in.) (1.7)

17

27.9 mm (1.10 in.) (1.7)

0.67

9.4 mm (0.37 in.) (0.58)

18

12.5

175 mm (6.88 in.) (10.7)

Fig. 1Reinforcement detailing of HPFRC specimens. (Note: Dimensions in mm; 1 mm = 0.0394 in.)
and post-tensioned rods tensioned through hydraulic jacks.
This load is typically used to represent the average dead and
service live loads carried by typical column bents of overpass bridges in California.
Test setup and instrumentation
The bidirectional quasi-static cyclic testing of the two
HPFRC column specimens was carried out. The lateral and
gravity load test setup is shown in Fig. 3. The lateral load
was applied using two horizontal servo-controlled hydraulic
actuators at an initial angle of 90 degrees with respect to
each other reacting against a rigid frame. The actuators were
attached to the column top using a steel jacket. The actuator
commands were computed to follow the circular load pattern
considering actual actuator elongations.
The HPFRC specimen instrumentation scheme comprised
internal and external instruments. Strain gauges were
installed on longitudinal and transverse reinforcement
to trace the strain history at various locations along the
column height and to correlate internal strains to observed
damage, such as bar buckling and fracture. Five levels of
linear displacement potentiometers were placed on two stiff
ACI Structural Journal/March-April 2014

Fig. 2Normalized displacement cycle: (a) plan view; and


(b) displacement history for normalized cycle.
instrumentation frames located on the sides of the specimen.
These absolute displacement measurements were used to
obtain the deflected shape of the column and relative deflections between various column sections, and to monitor the
torsion of the column top. Additional linear potentiometers were attached to instrumentation rods anchored in the
column core at various sections to measure the relative
displacements necessary for determination of flexural rota305

Fig. 3(a) Plan; and (b) elevation views of experimental setup for HPFRC column specimens.
Table 3Reinforcing steel mechanical properties
Parameter
Specification

Continuous longitudinal bar


11

Dowel reinforcement
11

Spiral reinforcement
ASTM A8212

ASTM A706

ASTM A706

Size

No. 4/13M

No. 4/13M

W3.5

Elastic modulus Es

190.3 GPa (27,600 ksi)

185.5 GPa (26,900 ksi)

198.3 GPa (28,800 ksi)

Yield stress Fy

448 MPa (65 ksi)

483 MPa (70 ksi)

638 MPa* (92 ksi)

Ultimate stress Fu

621 MPa (90 ksi)

705 MPa (102 ksi)

709 MPa (102 ksi)

Yield plateau, strain

0.25 to 1.33

Yield strength determined using 0.2% offset line parallel to corresponding elastic modulus.

tions, average curvatures, and shear deformations, as well


as average column axial deformations. Load cells were used
to monitor the forces in the two horizontal actuators and the
two post-tensioned rods used for simulation of gravity load.
Reinforcing steel
The mechanical properties of the longitudinal and spiral
reinforcement used for the HPFRC columns, obtained
through ASTM A37010 standard tension tests, are summarized in Table 3. The fracture strain was not determined in
these coupon tests because the bars were unloaded before
fracture due to laboratory safety regulations. Instead, fracture strain values specified in Caltrans SDC8 were used in
analytical models of the HPFRC cantilever columns.
Fiber-reinforced concrete
The concrete mixture used for the construction of the
fiber-reinforced concrete columns and plain concrete foundation had a specified 28-day strength of 34.5 MPa (5.0ksi).
This mixture proportion, with a proportion by weight of
0.45:1:2.3:1.86 (water:cement:fine aggregate:coarse aggregate), was adjusted from that used in previous tests6 to incorporate the steel fibers with acceptable workability. Hooked
steel fibers were added to the concrete mixture at a 1.5%
volume fraction (120 kg/m3 [200 lb/yd3]) by direct pouring
into the truck mixer and mixing for 4 minutes. Mixture
quantities for 0.765 m3 (1 yd3) of concrete are listed in
Table4. A similar mixture was successfully used in HPFRC
slab-column connection tests.13 A high-range water reducing
admixture was added during the mixing process to maintain

306

acceptable workability. The column concrete had a slump


of 140 mm (5.5 in.) following the addition of the high-range
water reducing admixture and before the addition of fibers.
Fiber-reinforced concrete 150 x 300 mm (6 x 12 in.) cylinders were cast during the construction of the specimens to
assess the development of concrete compressive strength
through ASTM C39/C39M14 tests. In addition, 150 x 150
x 600mm (6 x 6 x 24 in.) beams were tested under thirdpoint loading (457 mm [18 in.] span length) following
ASTMC1609/C1609M15 to assess the flexural performance
(and indirectly tensile performance) of the HPFRC material
used. Failure occurred inside the middle third of the spans
for all beam specimens tested. The elastic modulus for the
HPFRC material, Ec,FRC, corresponding to the secant modulus
up to 0.45fc,FRC, determined from fiber-reinforced concrete
cylinder tests, was also verified through ASTMC1609/
C1609M15 beam tests. Average compressive strength test
results for plain and fiber-reinforced concrete at various days
after casting, obtained directly from ASTM C39/C39M14
cylinder tests or from linear interpolation of results at other
days, are listed in Table 5. Measured cylinder compressive
strengths at test day are listed in Table 1. Despite having
the same specified strength, measured cylinder compressive strength at test day for the HPFRCs was approximately
20% higher than that of the regular concrete used in the
BC specimen (Table 1). The elastic modulus, and peak and
residual flexural strength at various deflection levels for the
fiber-reinforced concrete material determined according to
ASTMC39/C39M14 and ASTM C1609/C1609M,15 respectively, are also listed in Table 5.

ACI Structural Journal/March-April 2014

TEST RESULTS
The damage progress in the HPFRC and BC specimens
throughout the loading history is shown in Table 6. The
HPFRC specimens, particularly Specimen S1, exhibited
enhanced damage tolerance compared with the geometrically identical plain concrete Specimen BC. Both HPFRC
columns behaved elastically up to a drift ratio of approximately 1.3% (nominal ductility = 1.5 based on first yielding
of Specimen BC). The actual ductility of the tested HPFRC
columns a is, therefore, two-thirds that of the BC column.
The HPFRC specimens developed relatively similar damage
states as the reference BC specimen, but at higher displacement levels. For example, damage observed in Specimens
S1 and S2 at 3.9% drift ratio ( = 4.5) was smaller than that
in Specimen BC, as seen in Table 6. Large portions of the
BC column cover had already spalled off at this displaceTable 4Fiber-reinforced concrete mixture
quantities
Item

Value

w/c

0.45

Coarse aggregate weight (9.5 mm [3/8 in.]


maximum size)

602 kg/m3 (1324 lb/yd3)

Top sand weight

551 kg/m3 (1212 lb/yd3)

Blend sand weight

194 kg/m3 (426 lb/yd3)

Total fine aggregate weight

745 kg/m3 (1638 lb/yd3)

Cement

324 kg/m3 (712 lb/yd3)

High-range water-reducing admixture

621 mL/m3 (21 oz/yd3)

Water

145 kg/m3 (320 lb/yd3)

Fibers

88 kg/m3 (194 lb/yd3)

Note: Specified 28-day compressive strength: 34.5 MPa (5.0 ksi); slump: 140 mm
(5.5 in.); 1 yd3 = 0.765 m3; 1 lb = 0.45 kg.

ment level, while the HPFRC columns sustained relatively


minor damage with no spalling despite experiencing relatively large flexural cracks.
At a nominal ductility level of 6.25 (5.4% drift), the spiral
reinforcement in the S1 column fractured at an approximate
height of 150 mm (6 in.) above the column base. This spiral
fracture resulted in longitudinal bar buckling and significant
concrete cover spalling and core degradation in the subsequent displacement cycles. All the continuous longitudinal
bars in Specimen S1 underwent buckling and fractured by
the end of the cycle, corresponding to a nominal ductility
level of 12.5 (10.7% drift). The dowels in Specimen S1
did not buckle or fracture. Specimen S2, on the other hand,
experienced longitudinal bar buckling and spiral fracture at
two locations, followed by fracture of two continuous longitudinal reinforcement bars during the cycle at a nominal
ductility demand level of 6.25 (5.4% drift). The BC specimen
did not experience spiral fracture, bar buckling, or longitudinal reinforcing bar fracture because the spiral spacing was
half that provided in the HPFRC specimens, and the column
was only cycled up to a nominal ductility level of 4.5 (3.9%
drift) because it was subsequently tested under monotonically increased axial load.6 The maximum displacement
ductility level attained during the test of Specimens S1 and
S2 without substantial loss of gravity load-carrying capacity
was 12.5 and 6.25, respectively (10.7 and 5.4% drift ratio).
Tests of Specimens S1 and S2 were terminated after several
of the longitudinal bars fractured, compromising the safety
of the test setup.
The length of the plastic deformation zone Lp in Specimen
S1 was approximately 450 mm (18 in.) at the end of the test,
which is slightly larger than the column diameter. Inelastic
deformations in Specimen S1 first developed at approximately 350 mm (14 in.) from the base, and propagated
upwards as well as down towards the base of the column.

Table 5Average values of HPFRC mechanical properties

Parameter

11 days

28 days

49 days

60 days

Regular concrete compressive strength fc used for


BCspecimen

28.3 MPa* (4.10 ksi)

34.0 MPa (4.93 ksi)

35.0 MPa (5.07 ksi)

35.2 MPa* (5.10 ksi)

Regular concrete compressive


strength fc used for HPFRC
specimens

33.0 MPa (4.78 ksi)

41.9 MPa (6.07 ksi)

42.1 MPa (6.11 ksi)

41.9 MPa (6.08 ksi)

HPFRC compressive strength


fcFRC

37.0 MPa (5.37 ksi)

46.7 MPa (6.77 ksi)

47.1 MPa* (6.83 ksi)

47.3 MPa* (6.86 ksi)

HPFRC elastic modulus Ec-FRC

22.9 GPa* (3323 ksi)

28.9 GPa* (4190 ksi)

29.2 GPa* (4227 ksi)

29.3 GPa* (4246 ksi)

HPFRC first peak flexural


strength f1

5.44 MPa (0.79 ksi)

6.67 MPa (0.97 ksi)

6.37 MPa (0.92 ksi)

HPFRC absolute peak flexural


strength f

6.69 MPa (0.97 ksi)

7.10 MPa (1.03 ksi)

7.58 MPa (1.10 ksi)

HPFRC residual flexural


strength at L/600 deflection
fres,L/600

6.37 MPa (0.92 ksi)

6.49 MPa (0.94 ksi)

7.29 MPa (1.06 ksi)

HPFRC residual flexural


strength at L/150 deflection
fres,L/150

4.69 MPa (0.68 ksi)

5.58 MPa (0.81 ksi)

5.38 MPa (0.78 ksi)

Obtained indirectly from linear interpolation of test results performed on other days.

ACI Structural Journal/March-April 2014

307

Table 6Comparison of damage progress in HPFRC and BC specimens


S1: HPFRC

S2: HPFRC

BC: Plain concrete

State of column specimens at nominal ductility demand of 1.5 (1.3% drift)

State of column specimens at nominal ductility demand of 2.0 (1.7% drift)

State of column specimens at nominal ductility demand of 3.0 (2.6% drift)

State of column specimens at nominal ductility demand of 4.5 (3.9% drift)

State of column specimens at nominal ductility demand of 6.25 (5.4% drift)

Test terminated.

State of column specimens at nominal ductility demand of 8.0 (6.9% drift)

Test terminated.

State of column specimens at nominal ductility demand of 12.5 (10.7% drift)

308

ACI Structural Journal/March-April 2014

Fig. 4Force-deformation response of HPFRC and BC specimens: (a) total shear stress versus total drift; and (b) shear stress
ratio versus total drift.
The S2 column, on the other hand, developed a primary
crack at a height of 250 mm (10 in.) above the foundation at
the cutoff point of the dowel bars, resulting in a concentration of deformation and damage in that region with limited
spreading of the plastic deformation zone towards the base
of the column. The extent of the BC column plastic deformation zone was estimated at 300 mm (12 in.), equivalent to
3/4 of the column diameter.6
Force-displacement response
The force-deformation response of the HPFRC and BC
specimens is shown in Fig. 4. Only the primary cycles corresponding to the nominal ductility level of 1 and higher are
shown. The deformation axis is expressed in terms of resultant drift ratio (displacement of the column top divided by
the distance between the top of the column base to the actuator axes). The force axis is expressed in terms of resultant
shear stress (Fig. 4(a)) and shear stress ratio (Fig. 4(b)). The
resultant shear stress was calculated as the resultant shear
force applied by the two actuators divided by the effective
area in shear, defined by Caltrans SDC8 as 0.8 times the gross
area of the column. The shear stress ratio was obtained by
dividing the total shear force in the column by the assumed
nominal shear strength provided by the transverse steel reinforcement at yielding Vs, calculated according to Caltrans
SDC8 as follows

Vs =

Av f yh D
s

(1)

where Av = (/2)Ab is the area of shear reinforcement,


Ab is the area of the spiral reinforcement, fyh is the corresponding yield strength, D is the cross-sectional dimension
of confined concrete core measured between the centerline
of the spiral, and s is the spiral pitch.
Figure 4 shows a three-fold increase in shear stress ratio
demand in the HPFRC specimens compared with the BC
specimen due to: 1) half as much transverse reinforcement in
the HPRFC columns than in the BC column; and 2) dowels
in the plastic hinge region of the HPRFC specimens that
added flexural resistance and moved up the plastic hinge
region, increasing the shear demand at flexural yielding.
From Fig. 4, it can be seen that even though the shear
demand on the HPFRC specimens increased significantly
ACI Structural Journal/March-April 2014

and the amount of transverse reinforcement was reduced by


half compared with that in the BC specimen, the HPFRC
specimens maintained a stable hysteretic behavior governed
by flexure throughout the entire loading history. The degradation of lateral resistance with the progression of damage
in the HPFRC specimens, which initiated at a nominal
displacement ductility of 6.25 (5.4% drift), was governed by
the loss of flexural capacity (concrete crushing, bar buckling,
and finally bar fracture), and not by shear-related cracking or
damage. Sliding of the column along the cold joint at the
column-foundation interface or significant shear distortions
of the column plastic hinge zone were not observed.
A conventional shear force versus nominal displacement
ductility envelope, derived by plotting the peak lateral force
at each nominal displacement ductility level imposed, is
shown in Fig. 5(a). This plot was computed using the peak
resultant force of the specimens in the x- and y-directions,
to reconcile the differences in the response produced by the
circular load pattern. Both HPFRC specimens remained in
the elastic response range up to a nominal ductility demand
of 1.5 (based on first yielding of Specimen BC). The peak
applied force remained generally constant between nominal
ductility levels of 1 and 4.5 for the BC column, and between
1.5 and 6.25 for the HPFRC columns. The S1 column was
slightly stronger than the S2 column.
The secant lateral stiffness of the specimens, computed
using the resultant peak force and corresponding displacement value attained during each new primary cycle, is
shown in Fig. 5(b). The cracked elastic stiffness of the two
HPFRC specimens was comparable, and higher than that of
Specimen BC. However, the stiffness degradation rate for
Specimen BC was somewhat lower than that for the HPFRC
specimens up to nominal displacement ductility of 4.5. This
is likely due to the use of twice as much transverse reinforcement (and tighter spiral spacing) in Specimen BC.
Column plastic hinge deformations
Measurements of cross section rotation relative to the
column base about the x- and y-axes were obtained at four
levels along the height of the column (0.375Dcol, 0.75Dcol,
1.125Dcol, and 4.5Dcol above the column base). From these
rotations, average curvatures for each segment between two
adjacent rotation measurements were calculated to evaluate the spread of inelastic deformations along the column

309

Fig. 5(a) Shear force-displacement envelope; and (b) lateral stiffness degradation versus nominal displacement ductility (or
drift) demand.

Fig. 6Curvature profiles from experiments for: (a) S1; (b) S2; and (c) BC.
height. Profiles of these curvatures for all three specimens
are shown in Fig. 6.
The significant difference in the length of the plastic
deformation zones among these three specimens, discussed
previously and evident in Fig. 6, shows that it is possible
to design and detail the longitudinal reinforcement of an
HPFRC column to achieve a highly desirable spreading of
plastic deformation. In particular, the addition of dowel reinforcement elevated the center of the plastic hinge zone from
the column base, thus providing more space for its spreading.
Debonding of the dowel reinforcement (thus avoiding termination of dowels within the plastic hinge) in the column of
Specimen S1 allowed for very effective spreading of bar
yielding along the column height and formation of several
flexural cracks in the plastic hinge region. The less successful
detail used in Specimen S2, on the other hand, shows that
there are significant unexplored opportunities to develop
improved designs to ensure adequate spread of yielding in
310

HPFRC plastic hinges. It should be noted, however, that


increased curvature demands were imposed on the HPRFC
specimens for a given displacement level, compared with
those imposed on the BC specimen, due to the upward shift
of the plastic deformation region.
Bond stress
In both HPFRC specimens, the bonded region of dowel
reinforcement started at 250 mm (10 in.) above the foundation. Strain gauge measurements along the dowel reinforcement recorded peak strain values exceeding the steel yield
strain at a height of 100 mm (4 in.) above the foundation or
150 mm (6 in.) from the bar termination point, resulting in
a length Ld as small as 150 mm (6 in.) or 12 bar diameters
required to develop the yield strength of the No. 4 dowel
bars. The peak average bond stress up to first yielding of
the reinforcement in the HPFRC specimens was determined
based on force equilibrium between the resultant force from
ACI Structural Journal/March-April 2014

Fig. 7Curvature profiles for calibrated plastic hinge model of Column S1: (a) formulation; and (b) results for different
displacement ductility demands of HPFRC columns.
an average uniform bond stress ub acting on the surface of
the dowel reinforcement along a length measured from the
bar termination point to the section at which first yielding
occurred, and the yield strength of the bar. It is important
to note that dowel bar yielding occurred after substantial cracking occurred around the cover of the HPFRC
specimens. The resulting peak average bond stress ub was
10.2MPa (1.5 ksi), which can be rewritten in terms of the
unconfined HPFRC compressive strength results obtained
from cylinder tests, fc,FRC = 47.3 MPa (6.86 ksi). The degradation of bond stress with increasing bar inelastic strains
could not be evaluated due to lack of data. Thus, the peak
bond stress for reinforcing bar strains not exceeding the yield
strain was approximately ub,max = 1.5 fc, FRC , MPa (ub,max =
18 fc, FRC , psi). This value is significantly higher than bond
strengths reported in the literature for regular concrete16 of
1.0fc, MPa (12fc, psi) and 0.5fc, MPa (6fc, psi) for
deformed bars at slip values smaller and larger than the slip
measured at bar yield strain, respectively. These measurements thus indicate that reinforcement development lengths
in HPFRC columns could be shorter than those in conventional concrete columns. Similar conclusions were established in recent studies.9 Conservative development lengths
for conventional concrete, however, should be used for
HPFRC until more test data on the subject become available.
CALIBRATED HPFRC COLUMN MODELS
An idealized curvature profile for columns with reinforcement detailing similar to that used in Specimen S1 is
shown in Fig. 7. Based on the curvature distribution shown
in Fig.6(a), the length of the plastic hinge zone Lp was set at
400 mm (16 in.), equal to the column diameter. The middle
of the plastic hinge was centered at approximately the middle
of the debonded region of the column dowel bars, by setting
the distance between the column base and the bottom of the
plastic hinge zone, hLP, to 150 mm (6 in.). The yield curvatures y,Top and y,Bottom were defined at a height of 560 and
150 mm (22 and 6 in.) above the column base (top and bottom
ends of the plastic hinge zone), respectively. These yield
curvatures were computed using moment-curvature analyses
of the corresponding cross sections with different longitudinal reinforcement details. For displacements beyond first
ACI Structural Journal/March-April 2014

flexural yielding, a uniform curvature over the plastic hinge


length is assumed, dependent on the actual ductility level of
the HPFRC column a. Curvature over the length hLP is also
assumed constant, as shown in Fig. 7, and equal to 1/4 of the
curvature over Lp. Additional information about the material
models and parameters used for calibration of the HPFRC
plastic hinge model can be found elsewhere.6,7 A difference
of less than 10% was obtained in the computation of lateral
displacements, u, when using the experimentally (Fig. 6(a))
and analytically (Fig. 7) obtained curvatures. In the idealized curvature profile shown in Fig.7(a), the term HTot is the
height of the column measured from the base block top to
the actuator centerline.
SUMMARY AND CONCLUSIONS
Experimental findings obtained from tests of strain-hardening, high-performance fiber-reinforced concrete (HRFRC)
bridge columns under highly demanding bidirectional cyclic
displacements are presented. Two approximately 1/4-scale
circular cantilever HPFRC column specimens (S1 and S2)
were tested, and the results were compared with those of a
geometrically identical baseline regular concrete specimen,
denoted as Specimen BC. The HPFRC column specimens
were constructed using a ready mix concrete with a 1.5%
volume fraction of high-strength hooked steel fibers. The
plastic hinge region of the HPFRC specimens was detailed
using dowels in combination with bar debonding, either of
the dowel ends or of the main longitudinal reinforcement at
the end of the dowels to prevent concentration of damage
at the column base and increase the spread of yielding. The
transverse reinforcement spacing in the HPFRC specimens
was twice that of the BC specimen.
The HPFRC specimen with long dowels, Specimen S1,
developed an extended plastic deformation zone with an
approximate length of one column diameter, improved
ductile behavior, high-damage tolerance, and high energy
dissipation compared with the regular concrete Specimen
BC. This specimen was cycled up to a 10.7% drift ratio
while sustaining the applied gravity load. The HPFRC specimen with short dowels and debonding of main longitudinal
reinforcement, Specimen S2, exhibited less deformation
capacity compared with Specimen S1. The reason for such
311

behavior was the development of a single major crack at the


cutoff point of the dowel bars, resulting in a concentration of
rotation and damage at that location and limited spreading of
the plastic hinge zone towards the base of the column.
Even though the HPFRC specimens were subjected to
larger shear force demands and the amount of transverse
reinforcement was half that of the conventional Specimen
BC, the HPFRC specimens exhibited a stable hysteretic
behavior governed by flexure up to large drift demands with
negligible shear-related damage. Such desirable structural
response of the tested HPFRC columns indicates that HPFRC
has great potential for use in flexural elements subjected
to large shear reversals. The results of these unique tests
demonstrate several other advantages of HPFRC and call
for additional research on the design and detailing of plastic
hinge regions in HPFRC flexural members to best achieve
improved seismic performance and reduced post-earthquake
repair costs of bridge structures. Furthermore, the use of
HPFRC is expected to simplify the construction of critical
regions in bridges by allowing for a substantial increase
in transverse reinforcement spacing compared with that
required in regular concrete construction without compromising seismicperformance.
While the additional shear strength provided by the steel
fibers is helpful in ensuring a flexural-dominated behavior,
the adverse consequences of increased hoop spacing, particularly reduction of lateral support of longitudinal reinforcing bars, should be carefully considered during design
of HPFRC elements. The efficiency of HPFRC cover to
provide support to the longitudinal bars is as yet unknown,
and should not be relied upon until results from research on
this topic becomeavailable.
AUTHOR BIOS

ACI member Ady Aviram is a Structural Engineer with Simpson Gumpertz


& Heger, Inc., in San Francisco, CA. She received her BS in civil engineering from the University of Costa Rica, San Pedro, Costa Rica, and her
MEng and PhD in structural engineering from the University of California,
Berkeley, Berkeley, CA. Her research interests include performance-based
earthquake-resistant design of steel and reinforced-concrete structures,
base-plate connections, bridge modeling and analysis, structural reliability,
fiber-reinforced concrete, and blast-resistant design of wall systems.
Bozidar Stojadinovic, FACI, is a Professor and Chair of structural dynamics
and earthquake engineering at the Swiss Federal Institute of Technology
(ETH), Zrich, Switzerland. He received his Dipl. Ing. from the University
of Belgrade, Belgrade, Serbia; his MS from Carnegie Mellon University,
Pittsburgh, PA; and his PhD from the University of California, Berkeley. He
is a member of ACI Committees 335, Composite and Hybrid Structures; 341,
Earthquake-Resistant Concrete Bridges; 349, Concrete Nuclear Structures;
and 374, Performance-Based Seismic Design of Concrete Buildings. His
research interests include probabilistic performance-based seismic design
of composite and reinforced concrete structures.

312

Gustavo J. Parra-Montesinos, FACI, is the C. K. Wang Professor of


Structural Engineering at the University of Wisconsin-Madison, Madison,
WI. He is Chair of ACI Committee 335, Composite and Hybrid Structures;
and a member of ACI Committees 318, Structural Concrete Building Code,
and 544, Fiber-Reinforced Concrete, and Joint ACI-ASCE Committee 352,
Joints and Connections in Monolithic Concrete Structures. His research
interests include seismic behavior and design of reinforced concrete,
fiber-reinforced concrete, and hybrid steel-concrete structures.

REFERENCES

1. Parra-Montesinos, G. J., High-Performance Fiber-Reinforced


Cement Composites: An Alternative for Seismic Design of Structures, ACI
Structural Journal, V. 102, No. 5, Sept.-Oct. 2005, pp. 668-675.
2. Billington, S. L, and Yoon, J. K., Cyclic Response of Unbonded
Post-Tensioned Precast Columns with Ductile Fiber-Reinforced Concrete,
Journal of Bridge Engineering, ASCE, V. 9, No. 4, July-Aug. 2004,
pp.353-363.
3. Harajli, M. H., and Rteil, A. M., Effect of Confinement Using
Fiber-Reinforced Polymer or Fiber-Reinforced Concrete on Seismic
Performance of Gravity Load-Designed Columns, ACI Structural Journal,
V. 101, No. 1, Jan.-Feb. 2004, pp. 47-56.
4. Canbolat, B. A.; Parra-Montesinos, G. J.; and Wight, J. K., Experimental Study on Seismic Behavior of High-Performance Fiber-Reinforced
Cement Composite Coupling Beams, ACI Structural Journal, V. 102,
No.1, Jan.-Feb. 2005, pp. 159-166.
5. Parra-Montesinos, G. J., and Chompreda, P., Deformation Capacity
and Shear Strength of Fiber-Reinforced Cement Composite Flexural
Members Subjected to Displacement Reversals, Journal of Structural
Engineering, ASCE, V. 133, No. 3, 2007, pp. 421-431.
6. Terzic, V., and Stojadinovic, B., Post-Earthquake Traffic Capacity of
Modern Bridges in California, PEER Report 2010/103, Pacific Earthquake
Engineering Research Center, Berkeley, CA, 2010, 218 pp.
7. Aviram, A.; Stojadinovic, B.; Parra-Montesinos, G. J.; and Mackie,
K. R., Structural Response and Cost Characterization of Bridge Construction Using Seismic Performance Enhancement Strategies, PEER Report
2010/01, Pacific Earthquake Engineering Research Center, Berkeley, CA,
2009, 363 pp.
8. Caltrans, Seismic Design Criteria 1.3, Report, California Department of Transportation, Sacramento, CA, 2004, 101 pp.
9. Chao, S.-H.; Naaman, A. E.; and Parra-Montesinos, G. J., Bond
Behavior of Reinforcing Bars in Tensile Strain-Hardening Fiber Reinforced
Cement Composites, ACI Structural Journal, V. 106, No. 6, Nov.-Dec.
2009, pp. 897-906.
10. ASTM A370, Standard Test Methods and Definitions for Mechanical Testing of Steel Products, ASTM International, West Conshohocken,
PA, 2012, 48 pp.
11. ASTM A706/A706M, Standard Specification for Low-Alloy Steel
Deformed and Plain Bars for Concrete Reinforcement, ASTM International, West Conshohocken, PA, 2009, 6 pp.
12. ASTM A82, Standard Specification for Steel Wire, Plain, for
Concrete Reinforcement, ASTM International, West Conshohocken, PA,
2007, 4 pp.
13. Cheng, M.-Y.; Parra-Montesinos, G. J.; and Shield, C. K., Shear
Strength and Drift Capacity of Fiber Reinforced Concrete Slab-Column
Connections Subjected to Bi-Axial Displacements, Journal of Structural
Engineering, ASCE, V. 136, No. 9, 2010, pp. 1078-1088.
14. ASTM C39/C39M, Standard Test Method for Compressive
Strength of Cylindrical Concrete Specimens, ASTM International, West
Conshohocken, PA, 2012, 7 pp.
15. ASTM C1609, Standard Test Method for Flexural Performance
of Fiber-Reinforced Concrete (Using Beam With Third-Point Loading),
ASTM International, West Conshohocken, PA, 2012, 9 pp.
16. Eligehausen, R.; Popov, E. P.; and Bertero, V. V., Local Bond
Stress-Slip Relationships of Deformed Bars under Generalized Excitations, Report 82/23, Earthquake Engineering Research Center, University of California, Berkeley, Berkeley, CA, 1983, 169 pp.

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S27

Analysis of Early-Age Thermal and Shrinkage Stresses in


Reinforced Concrete Walls
by Barbara Klemczak and Agnieszka Knoppik-Wrbel
The issues related to thermal and shrinkage stresses arising in
reinforced concrete walls at the stage of their construction are
discussed. The cause of these stresses is inhomogeneous volume
changes associated with temperature rise caused by the exothermic
hydration process of cement, as well as moisture exchange with
the environment. The induced thermal-shrinkage stresses can
reach significant levels, and in some cases, cracks appear in
structuralelements.
The article presents the results of a numerical analysis of a reinforced concrete wall cast against an old set foundation, subjected
to early-age thermal and shrinkage deformations. Development of
stresses and character of cracks is briefly described. The presented
analysis focuses on evaluation of the contribution of self-induced
and restraint stresses to the total stresses induced in the wall. The
contribution and development of thermal and shrinkage stresses is
investigated. Walls with different dimensions are considered.
Keywords: cracking; early-age concrete; reinforced concrete wall; thermalshrinkage stresses.

INTRODUCTION
The cause of thermal and shrinkage stresses arising
in early-age concrete are the volume changes due to the
temperature and moisture variations during the hardening
process. The variations of concrete temperature during
curing are the result of the exothermic nature of the chemical
reaction between cement and water. In structural elements
with thin sections, the generated heat dissipates quickly, and
causes no problem. In thicker sections, the internal temperature can reach a significant level. Furthermore, due to the
poor thermal conductivity of concrete, high temperature
gradients may occur between the interior and the surface of
thick structural elements. Concrete curing is also accompanied by moisture exchange with the environment in conditions of variable temperatures. The loss of water through
evaporation at the surface of the element results in shrinkage,
which is classified as an external drying shrinkage. There
is also internal drying resulting from the reduction in material volume as water is consumed by hydration, which is
classified as autogenous shrinkage. Additionally, the chemical shrinkage is also distinguished, which occurs because
the volume of hydration products is less than the original
volume of cement and water.
The volume changes due to the temperature and moisture
variations have consequences in arising stresses in a concrete
element. Two natures of these stresses can be distinguished:
self-induced stresses and restraint stresses.
The self-induced stresses are related to internal restraints
of the structure, resulting from nonuniform volume changes
in a cross section. In internally restrained elements, during
ACI Structural Journal/March-April 2014

the phase of temperature increase, tensile stresses originate


in the surface layers of the element, and compressive stresses
are observed inside the element. An inversion of stress body
occurs during the cooling phase: inside, tensile stresses
are observed; in the surface layers, compressive stresses
are observed. Considerable self-induced stresses can be
expected, for example, within thick foundation slabs, thick
walls, dams, and in each element with interior temperatures
considerably greater than surface temperatures.
A concrete element can also be externally restrained. For
example, such a restraint exists along the contact surface of
mature concrete against which a new concrete element has
been cast. Because of the limited possibility of deformations of the structure, restraint stresses occur in that case.
The restraint stresses are often observed in medium-thick
elements, such as a wall cast against an old set concrete.
It should also be noted that the stresses resulting from an
external restraint of a structure add to the effects of an
internal restraint.
The problem of thermal-shrinkage stressesand consequently, in some cases, cracking of early-age concrete
structuresis well known in massive concrete elements.1-3
Nevertheless, high thermal-shrinkage stresses and cracks are
also observed in newly constructed medium-thick concrete
elements, such as reinforced concrete (RC) walls cast against
an old set foundation (tank walls or abutments4,5) or walls cast
in stages (massive container walls6). These cracks, which are
particularly deep or thorough cracks, may adversely affect
the serviceability, lifespan, or even bearing capacity of a
concrete structure. Cracking in tank walls endangers their
tightness.5 The problem of thermal-moisture cracking is also
dangerous in the case of nuclear containments executed in
stages, where a previously cast layer of mature concrete
restrains younger concrete layers.6 The formation of cracks
promotes the leakage of radioactive elements into the environment during the service life of the containment. Early-age
thermal-shrinkage cracking may even influence thin-walled
structures, such as shell roof covers, which reduces their
capacity, and may lead to structural collapse as a result of
unforeseen overpressure.7
RESEARCH SIGNIFICANCE
Considerable early-age thermal-shrinkage stresses and
consequent cracks at the construction stage are frequently
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-096.R2, doi:10.14359.51686523, was received December 14,
2012, and reviewed under Institute publication policies. Copyright 2014, American
Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published ten months from this journals date if the
discussion is received within four months of the papers print publication.

313

Fig. 1(a) Temperature; (b) thermal stresses; (c) moisture content; and (d) shrinkage stress development in time for externally
restrained concrete wall.10
observed in RC walls. Correct prediction and counteracting
negative effects of these stresses is of great importance to
ensure the desired service life and function of structures. It
is particularly important in structures such as tanks, which
require a solid concrete that prevents water leakage. This
article studies distribution of the discussed stresses in RC
walls with different dimensions. The contribution of self-induced and restrained stresses to the total induced stresses is
also investigated.
DEVELOPMENT OF THERMAL AND SHRINKAGE
STRESSES
Two main phases can be distinguished when observing
a temperature change in time during the concrete curing
process (Fig. 1(a)): a phase of concrete temperature increase
(self-heating), and a cooling phase of the element down to
the temperature of the surrounding air. In the first phase,
the wall extends being opposed by the foundation, which
results in the formation of compressive stresses (Fig. 1(b)),
usually within the first 1 to 3 days. As soon as the maximum
self-heating temperature is reached, the wall begins to cool
down, which takes another few days, restrained by a cooled
foundation. This leads to development of tensile stresses in
the wall (Fig. 1(b)). In case of moisture migration, a monotonic moisture removal from the wall is observed (Fig. 1(c)).
The resulting stresses are tensile stresses in the whole curing
process (Fig. 1(d)).
It should be noted that the graphs in Fig. 1 are shown only
for illustration of the phenomena that arise in the discussed
RC walls. In fact, the values of generated temperature and
the loss of moisture will be different in each point of the wall
according to the temperature and moisture exchange with
environment. Similarly, the values of generated stresses vary
in particular areas of the wall due to the different thermal
and shrinkage strains, as well as due to the different level of
the restraint of the wall in the foundation, which is changing
314

with the height of the wall. Nevertheless, the main character


of temperature, moisture, and stress development presented
in Fig. 1 is kept in most areas of the wall. The detailed
analysis of the temperature, moisture, and stress distribution in the walls with different dimensions are presented in
following sections of the article.
Shrinkage stresses resulting from moisture migration
reach relatively low values, so the character of total thermal-shrinkage stresses is analogical to thermal stresses;
shrinkage stresses increase total tensile stress values in the
second phase. Figure 2 presents total thermal-shrinkage
stress maps with resulting deformations in the right half
of the 70 cm (27.56 in.) thick wall in a heating phase after
1.3 days (Fig. 2(a)) and a cooling phase after 18.3 days
(Fig. 2(b)).
The described tensile stresses occurring in the cooling
stage often reach considerable values, and can lead to
cracking of the wall. A typical pattern of cracking due to an
edge restraint of a wall is shown in Fig. 3, assuming that the
base is rigid. Without restraint, the section would contract
along the line of the base; thus, with a restraint, a horizontal
force develops along the construction joint. This leads to
vertical cracking at the midspan area, but splayed cracking
towards the ends of the section where a vertical tensile force
is required to balance the tendency of the horizontal force
to warp the wall. In addition, a horizontal crack may occur
at the construction joint at the ends of the walls due to this
warping restraint.
Generally, it is thought that a basic cracking pattern is
independent of the amount of reinforcement provided.4,5,8
When sufficient reinforcement is provided, the widths of
the cracks are controlled, although secondary cracks may be
induced. The extent and size of cracking will then depend
on the amount and distribution of reinforcement provided.
The cracks can reach 1/3, 1/2, or even 2/3 of the height of
the wall, depending on the length and height of the wall, and
ACI Structural Journal/March-April 2014

Fig. 2Distribution of thermal-shrinkage stresses in right half of reinforced concrete wall in: (a) heating phase; and (b)
cooling phase.

Fig. 3Image, spacing, and height of cracks in reinforced


concrete wall.4
are usually spaced every 1.5 to 3.0 m (4.9 to 9.8 ft).4 The
cracks start at the wall-foundation interface and widen up to
the wk,max value, and then decrease in width. The maximum
width of the crack wk,max is approximately 0.3 to 0.5 mm
(12 to 20 mil) (in walls with a low horizontal reinforcement
ratio), and is localized at some level above the construction joint (at 1/3 of the height of the crack, according to
Reference 4; at a height equal to the thickness of the wall,
according to References 9 and 10). Another interesting property of the cracks is their distribution: the greatest height of
the crack can be observed in the middle of the wall length,
and it declines towards free edges of the wall or towards the
expansion joints, as presented in Fig. 3.
PREDICTION OF THERMAL AND SHRINKAGE
STRESSES
Prediction and control of thermal and shrinkage stresses,
along with possible cracking in early-age RC walls, is a
complicated problem due to the complex nature of interacting phenomena and a large number of contributing
factors.11-14 One of the important factors is the geometry of
the RC wall (length, height, and the length-height ratio), as
well as a degree of its restraint in a foundation. Additionally, the crucial factor is the temperature development in
the concrete member. The complex variables that affect the
rate of temperature rise, the maximum temperature, and the
temperature gradients over sections of the wall are:
Thermal properties of early-age concrete, such as the
rate of heat evolution, the total amount of heat, specific
heat, and thermal conductivity, are strongly dependent
on the amount and properties of concrete components,
especially the amount and type of cement;
ACI Structural Journal/March-April 2014

Conditions during concreting and curing of concrete,


such as the initial temperature of concrete, type of formwork, and use of insulation or pipe cooling;
Technology of concreting, such as segmental concreting;
Environmental conditions, such as ambient temperature, temperature of neighboring elements, wind, and
humidity; and
Dimensions of wall, especially the width of the wall.
Furthermore, deformations originating from the material
from which a wall is made and occurring in the erection
stage are essential loads. The problems arise from the material itself as concrete is subjected to transformations caused
by cement hydration and its mechanical properties develop
as its maturity progresses. Therefore, first the nonlinear and
non-stationary temperature and moisture fields and corresponding strains should be determined in the analyzed wall
considering the real technological and material conditions.
The stresses are calculated in the second step of the analysis.
METHOD OF ANALYSIS
The applied original numerical model can be classified as a
phenomenological model. The influence of mechanical fields
on the temperature and moisture fields was neglected, but
thermal-moisture fields were modeled using coupled equations of thermodiffusion. Therefore, the complex analysis of
a structure consists of three steps. The first step is related to
determination of temperature and moisture development. In
the second step, thermal-shrinkage strains are calculated, and
these results are used as an input for computation of stress in
the last step. With respect to the engineering application of
the theoretical model, computer codes were also developed.
Details of the model are given in Appendix A,* and a full
description of the model and computer programs, TEMWIL
and MAFEM_VEVP, can be found in References 15 through
17. Some results of the validation of the model are presented
in Appendix B,* where the cracking image obtained in the
numerical analysis is compared with the cracking observed
in the real tank walls. For presentation of results, an opensource application PARAVIEW was adopted.
*
The Appendix is available at www.concrete.org/publications in PDF format,
appended to the online version of the published paper. It is also available in hard
copy from ACI headquarters for a fee equal to the cost of reproduction plus handling
at the time of the request.

315

Table 1Thermal and moisture coefficients


Thermal fields
Coefficient of thermal conductivity

, W/mK (Btu/sftF)

2.52 (4.04 104)

Specific heat

cb, kJ/kgK (Btu/lbF)

0.95 (0.227)

Density of concrete

, kg/m (lb/ft )

2400 (149.80)

Coefficient of thermal diffusion

TT, m /s (ft /s)

11.1 107 (1.19 105)

Coefficient representing influence of


moisture concentration on heat transfer

TW, m2K/s (ft2F/s)

9.375 105 (1.81 103)

p, W/m2K (Btu/ft2sF)

6.00 (29.41 105) surface without protection, without


considering wind
3.58 (17.56 105) surface with plywood
5.80 (28.45 105) surface with foil
0.81 (3.97 105) bottom surface: soil

Thermal transfer coefficient

Heat of hydration*

According to equation Q (T , t ) = Q e

ate0 ,5

Q = kJ/kg (218.90 Btu/lb); a = 513.62te0.17

Moisture fields
Coefficient of water-cement
proportionality

K, m3/J (ft3/Btu)

0.3 109 (1.12 105)

Coefficient of moisture diffusion

WW, m2/s (ft2/s)

0.6 109 (6.46 109)

Thermal coefficient of moisture diffusion

WT, m2/sK (ft2/sF)

2 1011 (1.20 1010)

p, m/s (ft/s)

2.78 108 (91.21 109) surface without protection


0.18 108 (5.90 109) surface with plywood
0.10 108 (3.28 109) surface with foil
0.12 108 (3.93 109) bottom surface: soil

Moisture transfer coefficient

Approximation made on basis of experimental results of heat of hydration.

Fig. 4Dimensions of analyzed walls with finite element


mesh.
ANALYSIS OF STRESSES: RESULTS AND
DISCUSSION
Assumption of geometrical, material, and
technological data
The wall of 4 m (13.12 ft) height was analyzed for the
length of 10, 15, or 20 m (32.81, 49.21, or 65.62 ft) and
two thicknesses: 40 and 70 cm (15.75 and 27.56 in.). The
length-height ratios of these walls were, respectively, 2.5,
3.75, and 5. Six examples of combinations of these cases
were considered. The analyzed walls were supported on a
4m (13.12ft) wide and 70 cm (27.56 in.) deep strip foundation of the same length. The wall and the foundation were
assumed to be reinforced with a near-surface reinforcing
net of 16bars (0.63 in.). The wall was reinforced at both
surfaces with horizontal spacing of 20 cm (7.87 in.) and
vertical spacing of 15 cm (5.91 in.). The foundation was
reinforced with 20 x 20cm (7.87 x 7.87 in.) spacing at the
top and bottom surface. Due to a double symmetry of the
wall, the model for finite element analysis was created for
1/4 of the walls. A uniform mesh was prepared and densified
316

at the free edges of the wall and within the contact surface
between the wall and the foundation. A final geometry of the
wall with a mesh of finite elements for one exemplary wall
is presented in Fig. 4.
It was assumed that the analyzed wall was made of the
following concrete mixture: cement CEM I 42.5R 375 kg/m3
(23.41 lb/ft3), water 170 L/m3 (10.61 lb/ft3), and aggregate
(granite) 1868 kg/m3 (116.60 lb/ft3). Thermal and moisture
coefficients necessary for calculations were set in Table 1.
The development of mechanical properties in time was
assumed according to CEB-FIP MC90.18 The final values
for 28-day concrete were assumed as follows: compressive strength fcm = 35 MPa (5.08 ksi), tensile strength fctm =
3.0MPa (0.44 ksi), and modulus of elasticity Ecm = 32.0 GPa
(4.64 Mpsi). It was also assumed that the foundation was
erected earlier and had hardened, so the material properties
were taken as for 28-day concrete, with the same final values
as the wall. Environmental and technological conditions
were taken as: ambient temperature 20C (68F), initial
temperature of fresh concrete mixture 20C (68F), wooden
formwork of 1.8 cm (0.71 in) plywood on the side surfaces,
and foil protection of the top surface. It was also assumed
that formwork was removed 28 days after concrete casting.
Thermal and shrinkage stresses
First, the temperature and moisture development in
time was determined. Figure 5 presents a juxtaposition of
temperature and moisture content development diagrams for
two areas in the walls (Fig. 4) with different dimensions.
Although the character of both temperature and moisture
content are independent of the dimensions of the wall, their
magnitudes depend directly on these dimensions. Only the
ACI Structural Journal/March-April 2014

thickness, not the length, of the wall is influenced by the


temperature and moisture content development. The greatest
temperatures are reached in the thicker walls, while the
moisture removal rates are almost the same; only slightly
greater loss of moisture was observed in thinner walls.
For the known thermal-moisture fields and strains, the
stress state can be determined. The following cases were
analyzed, and the results of stress distribution at the height of
the wall in its internal midspan cross section were presented:
1. Thermal stresses with assumed uniform distribution of
temperature in the wall (development of temperature in time
was only considered) (Fig. 6(a) and (c));
2. Shrinkage stresses with assumed uniform moisture
content distribution in the wall (moisture content change in
time was only considered) (Fig. 6(b) and (d));
3. Thermal stresses with assumed real distribution of
temperature in the wall (development of temperature in time
was also considered) (Fig. 7(a), (c), and (e));
4. Shrinkage stresses with assumed real distribution of
moisture content in the wall (moisture content change in
time was also considered) (Fig. 7(b), (d), and (f)); and
5. Coupled thermal and shrinkage stresses with assumed
real distribution of both temperature and moisture content
(Fig. 8(a)).
The diagrams of nonuniform temperature distribution at
the height of the wall (Fig. 7(a)) are presented at the moment
the maximum hardening temperature is reached; in case of
uniform distribution of temperature, the values of temperature from the interior of the wall are applied the entire wall
(Fig. 6(a)). Similarly, the moisture distribution is presented at
the moment the maximum hardening temperature is reached.
Figure 6 presents distribution of thermal (Fig. 6(c)) and
shrinkage (Fig. 6(d)) stresses in the midspan cross section
of the interior of the wall under the assumption of uniform
temperature and moisture content distribution in the wall.
Such an assumption is very common in the analysis of
medium-thick externally restrained structures, especially
when analytic methods of thermal-shrinkage stresses determination are used, but also in numerical analyses in which
a model is reduced to a two-dimensional problem. It is
believed that such an approach provides a good approximation because both the temperature and moisture content
differences within the body of the wall are relatively small.
The resultant stress distribution at the height of the section
is approximately linear with the maximum values of stresses
at the joint between the wall and the foundation. On such
a simple example, it can be observed that both the length
and the thickness of the wall influence the resulting stresses.
The thickness of the wall determines the maximum value
of stress; it should be noted that greater values of thermal
stresses occur in thicker walls, which results from higher
exerted temperatures; shrinkage stresses are greater in
thinner walls, which is caused by a higher rate of water
removal. The length of the wall (linear restraint) determines
the distribution of stress at the height of the wall. For the
analyzed walls characterized by length-height ratio (L/H)
2.5, tensile stresses occur at the whole height of the wall; in
high wallsthat is, the walls with L/H < 2.5compression
of top areas may be expected.
ACI Structural Journal/March-April 2014

Fig. 5Temperature distribution: (a) in interior and (b) on


surface of wall; and moisture content development: (c) in
interior and (d) on surface of wall.
Nevertheless, heat and moisture are transported within the
element and to the surrounding environment in the process
of concrete curing. Therefore, the values of temperatures
317

Fig. 6Case of uniform distribution of: (a) temperature; (b) moisture; (c) thermal; and (d) shrinkage stress distribution at
height of wall in midspan cross section after 18.3 days.
and moisture content vary in different zones of the wall, as
do the resulting stresses. Figure 7 shows thermal (Fig. 7(c))
and shrinkage (Fig. 7(d)) stress distribution at the height of
the wall in its interior taking into account the real (nonuniform) temperature and moisture content distribution. It is
also noted that temperature and moisture content difference
at the thickness of the wall leads to stress diversification in
the internal and near-surface areas (Fig. 7(e) and (f)). It can
be observed that the length of the wall influences the occurring stresses as much as its thickness in such a way that the
length determines the character of stress distribution, while
the thickness determines the maximum values of stresses.
It should be emphasized that considering real distribution
of temperature and moisture the maximum stresses, and
consequently the highest cracking risk, is observed at some
distance above the joint, which complies with observations
in References 4, 9, and 10.
The observation diagrams in Fig. 8 were prepared to
present coupled thermal-shrinkage stress distribution in
318

the wall. The character of total thermal-shrinkage stresses


results mainly from the character of their thermal component (Fig. 7(b) versus Fig. 8(a)); shrinkage stresses only add
to the final value. Thus, the aforementioned conclusions
remain valid. The location of the maximum stresses varied;
generally, it was the closest to the construction joint for the
thinnest and the shortest walls (0.4 m = 1.31 ft), and was
elevated as the thickness and the length of the wall increased
(even up to 1.2 m [3.94 ft]). Thus, the results comply with
observations in Reference 4, while the observations from
References 9 and 10 seem more accurate for high walls
(walls with a low L/H ratio). In all analyzed cases, because
the formwork was detained for the whole process, greater
total stresses were observed in the interior of the wall, which
explains the occurrence of first cracks in the interior of
the wall.5
It should be noted that Fig. 6 presents stress distribution
at the height of the wall in the midspan cross section after
18.3 days under the assumption of uniform distribution
ACI Structural Journal/March-April 2014

Fig. 7Stress distribution at height of wall in midspan cross section after 18.3 days under assumption of real (nonuniform)
distribution of: (a) temperature and (b) moisture content in wall; (c) thermal and (d) shrinkage stresses in the interior of the
wall; and (e) thermal and (f) shrinkage stress in interior and on surface of 20 m (65.6 ft) long wall.
of temperature in the wall, while in Fig. 7 and 8, the real,
nonuniform distribution of temperature is taken into account.
It accounts for the visible differences in the obtained stress
distribution, especially in thermal stress near the joint.
In this case, the self-induced stress arises in the wall due
ACI Structural Journal/March-April 2014

to the nonuniform distribution of temperature in the wall.


The importance of the self-induced stresses is discussed in
a following section. It is interesting that the same differences in the stress distribution can be obtained with the use

319

Fig. 8Coupled thermal-shrinkage stress distribution at height of wall in midspan cross section after 18.3 days under assumption of real (nonuniform) temperature and moisture content distribution in wall: (a) total stress in interior of wall; and (b)
stresses in interior and on surface of 20 m (65.6 ft) long wall.

Fig. 9Thermal-shrinkage stress development in time: (a)


self-induced stresses; and (b) total stresses in 20 m (65.6 ft)
long wall.
of analytical methods when the nonuniform distribution of
temperature at the height of the wall is taken.19
Self-induced versus restraint stresses
In the next stage, the computations were made to evaluate
the share of self-induced stresses in total stresses exerted

320

in the wall. Total stresses exerted in the wall supported on


a stiff foundation due to the thermal-shrinkage effects are
a sum of self-induced and restraint stresses. To assess the
contribution of self-induced stresses in the analyzed walls,
the impact of restraint in a form of foundation was minimized by reduction of the foundations stiffness to EF =
100MPa (14.5 ksi) (flexible foundation was assumed).
Diagrams in Fig. 9 present development of self-induced
(Fig. 9(a)) and total (Fig. 9(b)) stresses in time for the location in which the maximum value of stress was observed for
the 20 m (65.6 ft) long walls of 70 and 40 cm (2.3 and 1.3ft)
thickness, assuming that both walls were detained in the
formwork. Stress development was presented for one length
of the wall for better visibility, as the values are similar.
The resulting self-induced stresses reach relatively low
values compared with the total stresses. Moreover, their
character is different and is closer to the behavior of typical
massive concrete structures. In the first phase, the interior
of the wall is subjected to compression, while surface layers
are tensioned; in the second phase, stress body inversion is
observed. This behavior is especially visible in thicker walls;
temperature and moisture content differences at the thickness of the thinner walls are smaller, so the resulting stresses
are of lower value. It is worth noting that the generation of
self-induced stresses is the cause of total stress difference in
different zones of the wall.
Figure 10 presents a comparison of the diagrams of
self-induced (Fig. 10(a)) and total stress (Fig. 10 (b)) distribution in the midspan cross section of the wall in both the
heating (after 1.2 days) and cooling phase (after 18.3 days).
The character of the observed stresses is similar in each
wall, so the diagrams are presented on the example of one
wall (L_20,d_0.7) only. Massive concrete-like behavior can
be observed in unrestrained walls, while the signs of total
stresses are the same at the whole height of the wall.

ACI Structural Journal/March-April 2014

Fig. 10(a) Self-induced and (b) total stress distribution at height of 20 m (65.6 ft) long, 70 cm (2.3 ft) thick wall in midspan
cross section in heating (ph_I) and cooling (ph_II) phase.

Fig. 11Stress distribution in midspan cross section of 20 m


(65.6 ft) long, 70 cm (2.3 ft) thick wall after 18.3 days from
which formwork was removed after 3 and 7 days.
Influence of time of formwork removal on stress
distribution
Finally, an analysis was performed that investigated the
influence of early formwork removal (3 and 7 days after
concrete casting). Diagrams in Fig. 11 show stress development in time, while Fig. 12 shows stress distribution after
18.3 days in the midspan cross section for the walls from
which the formwork was removed after 3 and 7 days. Because
of the similar character, diagrams were also presented for
one exemplary wall (L_20,d_0.7). If the wall is kept in the
formwork long enough for the concrete to cool completely,
the heat concentration in the interior of the wall leads to
higher stress and possible first crack development in the
internal parts of the wall (Fig. 9(b) and 10(b)). When formwork is removed in early phases of concrete curing, greater
stresses are observed on the surface of the wall as a result
of rapid cooling of the wall surface, which may lead to first
crack formation in the near-surface areas (Fig. 11 and 12). It
should be noted that in both cases, cracks may extend to the
entire wall thickness. It is important to note that formwork
removal accelerates moisture loss near the surface, which
is why a significant increase of tensile stresses is observed
in the vicinity of the top surface if it is no longer protected.
ACI Structural Journal/March-April 2014

Fig. 12Stress distribution in midspan cross section of 20 m


(65.6 ft) long, 70 cm (2.3 ft) thick wall after 18.3 days from
which formwork was removed after 3 and 7 days.
CONCLUSIONS
The problem with high temperatures arising during the
hardening of concrete has been known since the 1930s,
when dams were first built in the United States. Much effort
has been focused on the creation of efficient methods for
mitigation of the negative effects of concrete curing in
massive structures; this problem is well known in concrete
elements with considerable thickness. Nevertheless, formation of cracks is also observed in medium-thick concrete
elements, such as RC walls cast against an old set foundation. Such walls can also be sensitive to early-age cracking
of thermal and shrinkage origins. Control of thermal and
shrinkage cracking in early-age concrete is of great importance to ensure a desired service life and function of structures. It is a complicated problem due to the complex nature
321

of interacting phenomena and a large number of contributing


factors. Important factors include the dimensions of structural elements and the length-height ratio, which directly
affect the level of a wall restraint in a foundation.
The contribution of self-induced and restrained stresses
to total stresses induced in the wall with different dimensions was investigated. The results obtained with the use of
the original numerical model were discussed. Very limited
analysis of the stress distribution in the walls could be found
in the literature concerning early-age concrete. This paper
is an attempt to fill the vacancy in this field, in particular by
providing information about development of stresses and the
importance of self-induced stresses in externally restrained
structures. Knowledge about the distribution of stresses is
necessary in many practical cases, for example, in the evaluation of crack risk of early-age concrete structures. The
results of the analysis can be summarized as follows:
1. Thermal stresses play a predominant role in the total
thermal-shrinkage stress development;
2. The total thermal-shrinkage stresses arising in an RC
wall result mainly from restraint stresses generated by
limited possibility of the wall deformation; the share of
self-induced stresses increases with the increasing thickness
of the wall, but even in relatively thin elements, it has an
influence on total stress distribution in the wall; and
3. Three-dimensional numerical analysis results explain
the following phenomena observed in externally restrained
elements:
The greatest thermal-shrinkage stress does not occur
at the interface between the wall and the restraint but
at some level above the restraint joint. That fact results
from nonuniform distribution of temperature and moisture within the element which concentrate in its central
parts. In this case, the self-induced stress is also considered. The first crack can consistently be observed at
some level above the restraint joint; and
When the wall is detained in formwork until it cools
down, stresses develop towards the interior of the wall;
thus, internal cracking may initially develop. When
formwork is removed in early phases of concrete curing,
an increased cooling rate leads to greater stresses in
the surface zones, and first cracks can develop on the
surface of the wall.
AUTHOR BIOS

Barbara Klemczak is an Associate Professor in the Department of Civil


Engineering at the Silesian University of Technology, Gliwice, Poland.
She received her PhD and DSc from the Silesian University of Technology
in the field of numerical modeling of early-age massive concrete. Her
research interests include nonlinear analysis of reinforced concrete structures, particularly numerical modeling of thermal and shrinkage effects in
concrete structures at early ages.

ACKNOWLEDGMENTS

This paper was done as a part of a research Project N N506 043440 entitled, Numerical Prediction of Cracking Risk and Methods of Its Reduction
in Massive Concrete Structures, funded by the Polish National Science
Centre. The co-author of the paper, Agnieszka Knoppik-Wrbel, is a
scholar under the Project SWIFT, co-financed by the European Union under
the European Social Fund.

REFERENCES

1. ACI Committee 207, Report on Thermal and Volume Change Effects


on Cracking of Mass Concrete (ACI 207.2R-07), American Concrete Institute, Farmington Hills, MI, 2007, 28 pp.
2. Branco, F. A.; Mendes, P. A.; and Mirambell, E., Heat of Hydration Effects in Concrete Structures, ACI Materials Journal, V. 89, No. 2,
Mar.-Apr. 1992, pp. 139-145.
3. De Schutter, G., Fundamental Study of Early Age Concrete Behaviour
as a Basis for Durable Concrete Structures, Materials and Structures,
V. 35, Jan.-Feb. 2002, pp. 15-21.
4. Flaga, K., and Furtak, K., Problem of Thermal and Shrinkage
Cracking in Tanks Vertical Walls and Retaining Walls Near Their Contact
with Solid Foundation Slabs, ArchitectureCivil EngineeringEnvironment, V. 2, No. 2, June 2009, pp. 23-30.
5. Zych, M., Analiza pracy scian zbiornikw zelbetowych we wczesnym
okresie dojrzewania betonu w aspekcie ich wodoszczelnosci (Analysis of
Work of RC Tank Walls in Early Ages of Concrete Curing in the View of
Their Water Tightness), PhD thesis, Faculty of Civil Engineering, Cracow
Technical University, 2011. (in Polish)
6. Benboudjema, F., and Torrenti, J. M., Early-Age Behaviour of
Concrete Nuclear Containments, Nuclear Engineering and Design, V.
238, No. 10, Oct. 2008, pp. 2495-2506.
7. Estrada, C. F.; Godoy, L. A.; and Prato, T., Thermo-Mechanical
Behaviour of a Thin Concrete Shell during Its Early Age, Thin-Walled
Structures, V. 44, No. 5, May 2006, pp. 483-495.
8. De Borst, R., and Van den Boogaard, A. H., Finite Element Modeling
of Deformation and Cracking in Early-Age Concrete, Journal of Engineering Mechanics, ASCE, V. 120, No. 12, Dec. 1994, pp. 2519-2534.
9. Nilsson, M., Restraint Factors and Partial Coefficients for Crack Risk
Analyses of Early Age Concrete Structures, PhD thesis, Department of
Civil and Mining Engineering, Lule University of Technology, Sweden,
2003.
10. Larson, M., Thermal Crack Estimation in Early Age Concrete.
Models and Methods for Practical Application, PhD thesis, Department of
Civil and Mining Engineering, Lule University of Technology, Sweden,
2003.
11. ACI Committee 207, Guide to Mass Concrete (ACI 207.1R-05),
American Concrete Institute, Farmington Hills, MI, 2005, 30 pp.
12. RILEM TC 119-TCE, Recommendations of TC 119-TCE: Avoidance of Thermal Cracking in Concrete at Early Ages, Materials and Structures, V. 30, Oct. 1997, pp. 451-464.
13. RILEM Report 25, Early Age Cracking in Cementitous Systems,
Final Report of RILEM Technical Committee TC 181-EAS, 2002.
14. Mihashi, H., and Leite, J. P., State-of-the-Art Report on Control
Cracking in Early Age Concrete, Journal of Advanced Concrete Technology, V. 2, No. 2, June 2004, pp. 141-154.
15. Majewski, S., MWW3Elasto-Plastic Model for Concrete,
Archives of Civil Engineering, V. 50, No. 1, 2004, pp. 11-43.
16. Klemczak, B., Adapting of the Willam-Warnke Failure Criteria for
Young Concrete, Archives of Civil Engineering, V. 53, No. 2, June 2007,
pp. 323-339.
17. Klemczak, B., Prediction of Coupled Heat and Moisture Transfer in
Early-Age Massive Concrete Structures, Numerical Heat Transfer. Part A:
Applications, V. 60, No. 3, Sept. 2011, pp. 212-233.
18. Comit Euro-International du Bton, CEB-FIP Model Code 1990,
Thomas Telford, London, UK, 1991, 437 pp.
19. Klemczak, B., and Knoppik-Wrbel, A., Comparison of Analytical
Methods for Estimation of Early-Age Thermal-Shrinkage Stresses in RC
Walls, Archives of Civil Engineering, V. 59, No. 1, 2013, pp. 97-117.

Agnieszka Knoppik-Wrbel is a PhD Student in the Department of Civil


Engineering at the Silesian University of Technology. Her research interests
include cracking risk in early-age externally restrained concrete structures.

322

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S28

Effects of Casting Position and Bar Shape on Bond of


Plain Bars
by Montserrat Sekulovic MacLean and Lisa R. Feldman
Twenty-five lap splice specimens were reinforced with plain round
or square longitudinal bars in the top or bottom position to evaluate the effects of casting position and bar shape on bond. All specimens failed in bond, and the bond of square bars may be evaluated
by calculating their equivalent round diameter. Top cast factors of
0.3 and 0.6 for round and square bars, respectively, reasonably
capture the reductions in bond resistance. Maximum load predictions based on the CEB-FIP draft Model Code 2010 provisions for
bond are overly conservative for all combinations of bar shape
and casting position, whereas CEB-FIP Model Code 1990 provisions for bond reasonably and conservatively capture the behavior
of specimens with bottom-cast round bars, but do not appear to
capture the behavior of specimens with bottom- or top-cast square
bars or top-cast round bars.
Keywords: bar shape; bond; casting position; lap splice; plain reinforcement; stress.

INTRODUCTION
Recently published works highlight case studies of
historic reinforced concrete structures with plain reinforcement,1 and reviews of structural inventories include many
structures with reinforcing bar details and types that do not
meet current requirements.2 Many concrete structures with
plain reinforcement have reached an age where they require
remediation,3,4 and so it is crucial for forensic engineers to
have an understanding of their behavior and capacity. Of
particular note, ACI Committee 562, organized in 2004, has
a goal of developing a code and commentary for the evaluation, repair, and rehabilitation of existing concrete structures, and will likely need to include provisions for the bond
evaluation of plain reinforcement.
Plain reinforcement does not possess lugs or other surface
deformations, and cannot transfer bond forces by mechanical interlock. Instead, bond is transferred through adhesion between the concrete and the reinforcement before slip
occurs, and by wedging of small particles that break free
from the concrete upon slip.5 Moreover, both plain square
and round bars have been used to reinforce concrete structures,6 though only round bars were included in Abrams
historic study.5 The extent of void formations beneath
top-cast round and square bars due to the upward migration of water and mortar that occurs during concrete placement operations, and differences in concrete consolidation
around the two bar shapes, might cause different relationships between the required lap splice length and the depth
of concrete cast under the bar. Plain bars may also be more
affected by casting position than deformed bars because the
adhesion component of bond is a more dominant factor in
the transfer of bond force for plain bars.
ACI Structural Journal/March-April 2014

This paper presents the design and results of an experimental program to evaluate the effects of casting position
and bar shape on plain steel bars longitudinally cast in lap
splice specimens.
RESEARCH SIGNIFICANCE
Both round and square plain steel reinforcing bars are
regularly encountered in historical structures, and criteria for
assessing their bond strength are necessary. The results of
an experimental investigation of lap splice specimens reinforced with plain bars are presented to evaluate the effects
of bar shape and casting position on bond behavior. Such
knowledge will aid in the development of bond provisions
for the evaluation of concrete structures reinforced with
plain steel bars.
EXPERIMENTAL INVESTIGATION
The description of the specimens, materials, and test setup
are similar to those described by Hassan and Feldman,7 but
are briefly described herein for comprehensiveness. Figure
1 shows the cross sections, elevation, and plan view for
the 25 specimens in this study. Ten of the specimens, as
identified in Table 1, were originally reported by Hassan
and Feldman.7 All specimens had identical cross-sectional
dimensions and span lengths. Figure 1(a) and (b) show the
cross section of specimens with the round or square longitudinal reinforcement cast in the bottom and top positions,
respectively. These cross sections show that the cover was
held constant at 50 mm (2 in.) regardless of the size of the
longitudinal reinforcing bars used in the various specimens.
Earlier works2,8 suggested that the bond strength of plain
reinforcement is independent of concrete cover because
these bars lack mechanical interlock with the surrounding
concrete; thus, the likelihood of a splitting failure is reduced.
Specimens cast with top reinforcement were inverted before
testing such that Fig. 1(c) shows the elevation of all specimens as tested, including the span length, loading, and
reinforcing steel arrangement. The shear span-depth ratio,
a/d, was approximately equal to 3.94 for all specimens.
Figure1(d) shows a plan view of the specimens and illustrates the arrangement of the spliced longitudinal bars.
All specimens were designed to fail in bond, and had lap
splice lengths Ls ranging from 12.8 to 32.1 times the longitudinal bar diameter for round bars or the side face dimenACI Structural Journal, Vol. 111, No. 2, March-April 2014.
MS No. S-2012-097.R1, doi:10.14359.51686524, was received August 3, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

323

Table 1Actual and predicted failure loads


Predicted normalized maximum load Pmax/fc, kN/MPa (lb/psi)

Specimen
identification*

Splice length
as a function of
bar size
(Ls/db)

Concrete
compressive
strength fc,
MPa (psi)

Bar surface
roughness Ry,
m
( 103 in.)

Maximum
normalized
load, Pmax/fc,
kN/MPa
(lb/psi)

19l-305

16.1

17.4 (2520)

9.54 (0.376)

8.50 (159)

18.0 (337)

29.1 (544)

4.98 (93.0)

0.96 (17.9)

21.6

17.4 (2520)

9.67 (0.381)

9.14 (171)

18.0 (337)

29.1 (544)

7.72 (144)

2.33 (43.5)

18.7 (2710)

9.86 (0.388)

9.58 (179)

17.5 (327)

28.4 (530)

10.5 (196)

3.71 (69.2)

21.0 (3040)

9.44 (0.372)

17.8 (332)

16.7 (311)

27.1 (507)

13.9 (259)

5.36 (100)

23.7 (3440)

8.88 (0.350)

16.2 (302)

28.1 (524)

45.0 (840)

12.5 (233)

4.12 (76.9)

24.0 (3480)

8.43 (0.332)

18.4 (343)

27.7 (518)

44.5 (831)

16.1 (300)

5.74 (107)

22.8 (3310)

8.71 (0.343)

20.6 (384)

28.5 (532)

45.6 (851)

19.8 (370)

7.52 (140)

19l-410

19l-510

19l-610

25l-410

25l-510

25l-610

26.8
32.1
16.4
20.4
24.4

Neglecting
strain
hardening of
reinforcement

Including strain
hardening of
reinforcement

CEB-FIP
Model Code
1990

Draft CEB-FIP
Model Code
2010

25l-410

16.4

27.1 (3930)

9.19 (0.362)

6.55 (122)

28.1 (524)

42.2 (788)

8.39 (157)

0.91 (17.0)

25l-510

20.4

28.0 (4060)

9.09 (0.358)

4.69 (87.5)

27.7 (517)

41.7 (778)

11.1 (207)

1.76 (32.8)

25l-610

24.4

35.8 (5190)

9.21 (0.362)

7.07 (132)

25.1 (468)

38.0 (710)

14.8 (276)

2.91 (54.3)

32l-410

12.8

19.8 (2870)

9.92 (0.390)

15.6 (291)

44.5 (827)

63.2 (1180)

14.1 (263)

4.43 (82.7)

32l-610

19.1

19.8 (2870)

9.72 (0.383)

25.1 (468)

44.3 (827)

63.2 (1180)

22.1 (412)

7.94 (148)

32l-810

25.3

15.8 (2290)

10.1 (0.398)

31.8 (594)

46.9 (876)

64.0 (1190)

28.0 (523)

10.9 (203)

16.4

25.5 (3700)

8.79 (0.346)

16.1 (300)

38.3 (714)

55.3 (1030)

14.8 (276)

4.85 (90.5)

25.0 (3620)

8.83 (0.348)

20.0 (373)

38.3 (714)

55.2 (1031)

18.6 (347)

6.50 (121)

28.1 (4080)

8.86 (0.349)

26.8 (500)

36.6 (684)

53.4 (996)

22.9 (427)

8.24 (154)

33.0 (4790)

9.06 (0.357)

8.97 (167)

31.3 (585)

50.2 (937)

10.5 (196)

1.47 (27.4)

33.5 (4860)

9.17 (0.361)

11.2 (209)

31.1 (581)

49.9 (932)

13.6 (254)

2.37 (44.2)

33.0 (4790)

9.16 (0.361)

12.2 (228)

31.3 (585)

50.2 (937)

16.6 (310)

3.26 (60.8)

25.5 (3700)

9.36 (0.368)

17.4 (325)

50.6 (944)

73.7 (1380)

17.3 (323)

5.69 (106)

25.5 (3700)

9.24 (0.364)

20.1 (375)

50.8 (948)

74.0 (1380)

26.6 (496)

9.63 (180)

26.9 (3900)

9.17 (0.361)

28.3 (528)

49.5 (924)

72.9 (1360)

35.6 (664)

13.4 (250)

27.5 (4000)

9.29 (0.366)

12.6 (235)

49.4 (922)

73.0 (1360)

11.6 (216)

1.66 (31.0)

9.52 (0.375)

14.3 (267)

50.4 (941)

73.8 (1380)

17.9 (334)

3.54 (66.1)

9.34 (0.368)

16.2 (302)

50.4 (941)

73.8 (1380)

24.5 (457)

5.50 (103)

25n-410
25n-510
25n-610
25n-410
25n-510
25n-610
32n-410
32n-610
32n-810
32n-410

20.4
24.4
16.4
20.4
24.4
12.8
19.1
25.3
12.8

32n-610

19.1

26.2 (3800)

32n-810

25.3

26.2 (3800)

First number in specimen identification represents nominal diameter for round bars or side face dimension for square bars. Solid circle (l) or square (n) identifies shape of longitudinal reinforcement. Number following hyphen denotes lap splice length, in millimeters, with an up arrow () showing that longitudinal bars were cast in top position, and down
arrow () showing that bars were cast in bottom position.

Originally reported by Hassan and Feldman (2012).

Specimens cast with Concrete Mixture Proportion 1.

Specimens cast with Concrete Mixture Proportion 2.

sion for square bars. Specimen dimensions and lap splice


lengths were selected to match those reported by Idun and
Darwin9 for tests of specimens with deformed bars. A direct
comparison of the specimens with plain and deformed bars
is reported elsewhere.7
Figure 2 shows that a vertical load was applied using a
spreader beam with a self-weight that exerted a load P of
1.77 kN on the specimens to establish the four-point loading
arrangement. Loading was applied at a rate of 0.0015 mm/s
(0.0006 in./s) to failure.
Concrete
The concrete had a target compressive strength of 20 MPa
(2900 psi). General purpose (Type GU) portland cement
was used without admixtures. Table 1 indicates the mixture
324

proportion used to cast each specimen. Mixture Proportion


1 consisted of a crushed limestone and granite coarse aggregate blend and a silica sand fine aggregate. The mixture
proportion per cubic meter (cubic yard) of concrete was:
250kg (421 lb) cement, 1100 kg (1854 lb) sand, 1100 kg
(1854 lb) crushed coarse aggregate, and 140 L (28.3 gal.)
water. Mixture Proportion 2 consisted of a carbonate,
gneiss, and granite coarse aggregate blend and a washed
silica sand fine aggregate. The mixture proportion per cubic
meter (cubic yard) of concrete was: 270 kg (455 lb) cement,
993kg (1674 lb) sand, 1039 kg (1751 lb) crushed coarse
aggregate, and 145L (29.3 gal.) water. The maximum size
of the coarse aggregate used in both mixture proportions
was 20mm (0.8in.), and all aggregates conformed to CAN/
CSAA23.1-09.10 It should be noted that the effect of mixture
ACI Structural Journal/March-April 2014

Fig. 1Splice specimen geometry: (a) cross section for specimens with bottom-cast longitudinal reinforcement; (b) cross
section for specimens with top-cast reinforcement; (c) elevation; and (d) plan view. (Note: Dimensions are given in mm [in.].)

Fig. 2Test setup.


ACI Structural Journal/March-April 2014

325

Table 2Longitudinal reinforcing steel properties


Bar identification

Static yield strength fys,


MPa (ksi)

Dynamic yield strength fyd,


MPa (ksi)

Ultimate strength fu,


MPa (ksi)

Modulus of elasticity Es,


GPa (ksi)

19l

326 (47.3)

355 (51.5)

520 (75.4)

203 (29,400)

25l

322 (46.7)

346 (50.2)

534 (77.4)

196 (28,400)

25l

340 (49.3)

364 (52.8)

522 (75.7)

243 (35,200)

32l

318 (46.1)

348 (50.5)

504 (73.1)

204 (29,600)

25n

357 (51.8)

381 (55.2)

544 (78.9)

192 (27,800)

25n

325 (47.1)

349 (50.6)

542 (78.6)

207 (30,000)

32n and 32n

320 (46.4)

343 (49.7)

527 (76.4)

196 (28,400)

proportion on bond strength was not within the scope of the


current investigation. Rather, the change in mixture proportions resulted from a change in ready-mix suppliers. Bleed
water measurements were not conducted as part of this
investigation.
Table 1 shows the concrete compressive strength of the
specimens at the time of testing as established from the
results of companion concrete cylinders stored under the
same conditions and tested on the same day as the corresponding splice specimen. Specimens were moist cured
using wet burlap and plastic sheets for 7 days following
casting, and were then stored in the laboratory until testing.
Reinforcement
All principal longitudinal reinforcement was hot-rolled
CSA G40.21 300W steel. Figure 1(c) shows that the bars
had 180-degree hooks at the ends adjacent to the beam
supports to ensure that the bond failure occurred within the
lap splice length. The material properties were established
from coupons obtained from surplus bar lengths and tested
in accordance with ASTM A370.11 Table 2 shows the static
yield strength fys calculated in accordance with Rao et al.,12
dynamic yield strengths fyd, ultimate strength fu, and modulus
of elasticity Es for all longitudinal bar sizes used.
The longitudinal reinforcing bars were sandblasted using
220-grit aluminum oxide, a nozzle distance of 125 mm
(5 in.), and a blast pressure of 698 kPa (100 psi) to increase
the surface roughness and make them more representative of
historical bars.8 The surface roughness of each bar was characterized by the maximum height of profile Ry, established
as the distance between the highest peak and the deepest
valley on the bar surface.13 Table 1 shows the average Ry
values for the longitudinal reinforcing bars in each specimen
based on a total of 30 roughness measurements on each
bar using a surface roughness tester and a single 0.25 mm
(0.01in.) stroke.
The shear reinforcement consisted of 12.7 mm (0.5 in.)
diameter hot-rolled CSA G40.21 300W plain steel bars
spaced at 200 mm (8 in.) on center within the shear spans,
and 250 mm (10 in.) on center within the constant moment
region outside of the lap splices (Fig. 1(d)). Two additional
stirrups were placed in the splice region one-quarter of the
splice length, but not exceeding 150 mm (6 in.), from the
ends of the splice to prevent prying action of the longitudinal
reinforcement. The specimens had considerably more shear
reinforcement than strictly necessary to ensure that failure
326

would be governed by bond between the longitudinal reinforcement and the surrounding concrete.
EXPERIMENTAL RESULTS
Table 1 shows the observed maximum loads attained
by the specimens and those predicted assuming yielding
of the reinforcement, both neglecting and including strain
hardening, and predicted loads based on average bond
stress provisions for plain reinforcement included in the
CEB-FIP Model Code 199014 and the CEB-FIP draft Model
Code 2010.15 The predicted loads have been reduced by the
weight of the spreader beam and the specimen self-weight to
allow for a direct comparison with the maximum loads that
were recorded during testing. All reported loads have been
normalized by the square root of the concrete compressive
strength given that a previous work8 showed that it is valid
for plain reinforcement and is consistent with familiar equations for deformed bars.
The specimens are identified by mark numbers that include
two numbers and associated symbols separated by a hyphen.
The first number represents the nominal diameter for round
bars or the nominal side face dimensions for square bars db,
in millimeters, used to longitudinally reinforce the specimens. A solid circle (l) or square (n) following this number
identifies the shape of the longitudinal reinforcement. The
number following the hyphen denotes the lap splice length
Ls in millimeters, with an up arrow () showing that the
longitudinal bars were cast in the top position (Fig. 1(b)),
or a down arrow () showing that the bars were cast in the
bottom position (Fig. 1(a)).
Table 1 shows that the maximum normalized load reported
for Specimen 25l-510 was only 72% that of Specimen 25l-410, a specimen cast from the same batch of
concrete. This result was considered suspect because specimens with longer splice lengths should be able to resist higher
loads when all other variables are held constant. Removal
of the concrete surrounding the longitudinal reinforcement
was completed for Specimen 25l-510 following testing;
however, no voids were identified that would have impaired
the bond between the reinforcement and the surrounding
concrete. This specimen therefore could not be identified as
a physical outlier, and is included in the regression analysis
as will be presented in a subsequent section.
Table 1 shows that all but Specimen 19l-610 failed at
loads well below those predicted using the flexural resistance procedures in ACI 318-1116 with resistance factors set
ACI Structural Journal/March-April 2014

equal to unity, and thus suggest that these specimens failed


in bond. The load-deflection behavior and crack patterns for
all specimens is consistent with that described by Hassan
and Feldman,7 including that of Specimen 19l-610, which,
based upon its load-deflection behavior, failed in bond even
though it attained a maximum load of 106% that predicted
assuming yielding of the reinforcement.
Comparison of test results with predictions based
on European code provisions
Both the CEB-FIP Model Code 199014 and the draft Model
Code 201015 provide equations for the average bond stress
for plain reinforcement, and thus allow for a prediction of
the maximum normalized load resisted by the specimens in
the study, as reported in Table 1. The design average bond
stress uave for reinforcing bars is specified in the CEB-FIP
Model Code 199014 as

uave = 12 3 fctd

(1)

where 1 is a factor that accounts for the reinforcing type,


and is equal to 1.0 for plain bars; 2 considers the casting
position of the reinforcement, and is equal to 1.0 and 0.7
for bottom and top cast reinforcement, respectively; 3
considers the bar size, and is equal to 1.0 for bars with a
diameter db of 32 mm (1.25 in.) or less and (132 db)/100 for
db > 32 mm (1.25 in.); and fctd is the design tensile strength
of the concrete in MPa.
Bond provisions have changed markedly in the CEB-FIP
draft Model Code 2010.15 The basic average bond strength
uave, which can be used to assess plain reinforcement, is
uave = 12 3 4 ( fck 20 )

0.5

(2)

where 1 is equal to 0.9 for plain bars; 2 is equal to 1.0


and 0.5 for plain bars cast in the bottom and top positions,
respectively; 3 is equal to 1.0 for db 20 mm (0.79 in.)
and (20/db)0.3 for larger bar sizes; 4 accounts for the characteristic yield stress of the reinforcement and is equal to 1.2
for reinforcement with a yield stress equal to, and presumably less than, 400 MPa (58.0 ksi); fck is the characteristic
value of the cylinder compressive strength of the concrete, in
MPa; and c is a partial factor for the concrete compressive
strength set equal to unity for the case of using Eq. (2) as a
predictive, rather than design, equation.
Neither edition of the CEB-FIP Model Code14,15 specifically provides for square reinforcing bars. An analysis of
archival test results of pullout specimens reinforced with
historical round and square bars whose deformation patterns
did not conform to ASTM A305-4917 was performed by
Howell and Higgins18 and showed that the simplified ACI
development length equations19 provided a lower bound for
both bar shapes, where square bars were evaluated by calculating their equivalent round diameter db,EQ

db, EQ =

4 db 2

ACI Structural Journal/March-April 2014

(3)

The bar size factor 3 in Eq. (1) and (2) was therefore
calculated assuming db = db,EQ for the case of specimens
longitudinally reinforced with square bars.
The predicted maximum normalized loads for all specimens tested were then calculated assuming a linear strain
distribution along the height of the specimen and the stress
versus strain distribution for the concrete as obtained from
companion specimens tested in conjunction with each
splice specimen. The neutral axis location was established
from cracked transformed section properties because all
specimens failed in bond, and, with the exception of Specimen19l-610, at loads well below those predicted based
on yielding of the longitudinal reinforcement.
Figure 3(a) shows the ratio of the test-to-predicted
maximum loads based on the CEB-FIP Model Code 199014
provisions, Pmax/(Pmax)CEB,1990, for all specimens. Values of
Pmax/(Pmax)CEB,1990 > 1.0 suggest that the predicted values
are conservative, whereas values of Pmax/(Pmax)CEB,1990<1
suggest that the same values are unconservative. The average
Pmax/(Pmax)CEB,1990 for all specimens with bottom-cast round
bars is equal to 1.2, and so suggests that the CEB-FIP Model
Code 199014 provisions reasonably and conservatively
capture the bond behavior of these bars. The same code
provisions, however, do not appear to capture the behavior of
specimens with bottom- (average Pmax/(Pmax)CEB,1990 = 0.98)
or top-cast square bars (average Pmax/(Pmax)CEB,1990 = 0.83),
or specimens reinforced with top-cast round bars (average
Pmax/(Pmax)CEB,1990 = 0.56).
Figure 3(b) shows that the ratio of the test-to-predicted
loads based on the CEB-FIP draft Model Code 201015
provisions, Pmax/(Pmax)CEB,2010, for all specimens. All values
of Pmax/(Pmax)CEB,2010 exceed 2.0, which suggests that the
CEB-FIP draft Model Code 201015 provisions are overly
conservative. Figure 3(b) does suggest that these provisions
are also more conservative when estimating the capacity of
specimens cast with the longitudinal reinforcement in the
top position, and that the provisions tend to be more conservative for specimens with shorter lap splice lengths for both
bar shapes and casting positions.
Top casting effects
Figure 4 shows the ratio of the maximum normalized
loads for each pair of specimens for which the same size,
shape, and lap splice length were provided for longitudinal
reinforcement, but with this reinforcement cast in the top
position for one specimen, and cast in the bottom position
for the other specimen. Figure 4 shows that all specimens
with top cast reinforcement failed at loads well below those
for bottom cast reinforcement, with a resulting average
ratio of the normalized maximum loads equal to 0.51. The
results of this limited investigation suggest that square reinforcement, with an average ratio of 0.60, is less sensitive to
casting position than round bars, with a resulting average
ratio of 0.33. Furthermore, the larger square reinforcing bar
size, db = 32 mm (1.25 in.), used to longitudinally reinforce
the specimens, appears to be less sensitive to casting position than the smaller square bars with db = 25 mm (1 in.).
Current American16 and Canadian20 code provisions for
reinforced concrete require a 30% increase in development
327

Fig. 4Ratio of normalized maximum loads for pairs of


specimens with longitudinal reinforcement in top versus
bottom cast position.
below the bottom face of the bar only and will affect a much
smaller portion of the overall perimeter.

Fig. 3Test-to-predicted ratio of maximum normalized


loads: (a) predicted loads based on CEB-FIP 1990 code
provisions; and (b) predicted loads based on CEB-FIP 2010
draft code provisions.
length for deformed bars cast in the top position (that is,
when more than 300 mm [12 in.] of fresh concrete is placed
below the reinforcement), whereas the CEB-FIP Model
Code 199014 provides a multiplier of 0.7 to the average bond
stress in such cases. The modifiers in all three of these codes
appear to be unconservative for plain bars in light of the
results discussed in the previous paragraph. Furthermore, a
previous investigation conducted by Chana21 also concludes
that the bond of plain bars is more affected by casting position than that of deformed bars. This conclusion appears
justified upon consideration of the mechanics of bond. The
adhesion between the concrete and the reinforcement is a
more dominant factor in the transfer of bond forces for plain
reinforcement, because they cannot transfer these forces by
mechanical interlock.22
In contrast, the CEB-FIB draft Model Code 201015
provides a multiplier of 0.5 specifically for plain bars cast
in the top position, and appears to be more reasonable when
compared with the results in the current investigation. The
test results presented herein, however, also show that square
bars are less sensitive to casting position than round bars
due to differences in the shape of voids that form beneath
these two bar shapes. For round bars, the concrete tends to
settle such that a void forms under the bottom half of the
perimeter23; whereas for square bars, assuming construction
allows them to be placed perfectly square within the reinforcing cage as shown in Fig. 1(a) and (b), the void will form

328

Effect of bar shape as assessed using predictive


equations for maximum normalized load
A regression analysis of the 25 specimens yields
the following empirical equation for the normalized
maximumload
Pmax

fc

= 9.38 10 5 Ls db Ry +

Ls
(0.24 + 0.15kb 0.50kc ) (4)
db

where Ls is the lap splice length, in mm; db is the longitudinal bar size, in mm, reported as the measured bar diameter for round bars or the measured side face dimension
for square bars; Ry is the surface roughness of the longitudinal reinforcement, in m; kb is an indicator variable for
the shape of the longitudinal bars, and is equal to zero for
round bars and 1 for square bars; and kc is an indicator variable for the casting position of the longitudinal reinforcement, and is equal to 0 if the bars are cast in the bottom
position (Fig. 1(a)), and 1 if the bars were cast in the top
position (Fig. 1(b)). The root mean square error for Eq. (4) is
2.91kN/MPa (54.3 lb/psi).
Using an equivalent round diameter db,EQ, as described
by Eq. (3), results in the following predictive equation that
allows for the elimination of the indicator variable kb
Pmax

fc

= 3.36 10 4 Ls Ry db, EQ 0.5 (2.12 kc )

(5)

where db,EQ is the diameter for round bars and the equivalent round diameter of square bars. The resulting root-meansquare error of 3.01 kN/MPa (56.2 lb/psi) for Eq.(5)
is similar to that reported for Eq. (4), and suggests that
using the equivalent round diameter for plain square bars
isreasonable.
Results of a previous investigation8 have shown that the
average surface roughness, Ry = 9.26 m, for the 25 specimens included in the regression analysis is a lower bound for
ACI Structural Journal/March-April 2014

Fig. 5Comparison of recorded normalized maximum loads to those predicted empirically using Eq. (6) for: (a) specimens
cast with round longitudinal bars; and (b) specimens cast with square longitudinal bars.
that of historical bars. Furthermore, the analysis presented in
the previous section suggests that multipliers of 0.3 and 0.6
for round and square longitudinal bars, respectively, reasonably capture the reduction in bond resistance provided by
bars in the top cast position. The following simplified predictive normalized maximum load equation therefore results

Pmax

fc

= 6.63 10 3 Ls t db, EQ 0.5

(6)

where t is a factor used to modify the lap splice length


based on the location of the longitudinal reinforcement, and
is equal to 1.0 for both round and square bars cast in the
bottom position, 0.3 for round bars cast in the top position,
and 0.6 for square bars cast in the top position. The resulting
root-mean-square error for Eq. (6) is 2.44 kN/MPa
(45.5 lb/psi).
Figure 5 shows the fit of Eq. (6) with the experimental test
data, with Fig. 5(a) showing all data for specimens longitudinally reinforced with round bars, and Fig. 5(b) showing
all data for specimens longitudinally reinforced with square
bars. It should be noted that a linear and proportional relationship, with Pmax/fc = 0 for Ls = 0, is the best fit. This
finding differs from the linear, but not proportional, relationship reported for deformed bars16 and prestressing strands.24
An alternate method of evaluating the goodness of fit of
Eq. (6) to the test data is presented in Fig. 6, which shows
the predicted normalized load calculated in accordance
with Eq.(6) versus the recorded maximum normalized
load for all 25 specimens in the test database. Also shown
is the proportional line, which represents the theoretical
case that the predicted maximum normalized load exactly
equals that recorded during specimen testing. Specimen
markers falling above this line represent cases for which the
maximum normalized load, predicted in accordance with
Eq. (6), exceeds the maximum normalized load recorded
during specimen testing. Similarly, markers falling below
the proportional line represent cases for which Eq. (6) underestimates the maximum normalized load recorded during
specimentesting.
A review of the data presented in Fig. 6 shows that the
mean and standard deviation obtained for specimens longitudinally reinforced with square bars cast in the bottom position were 1.04 and 0.154, respectively. Similarly, specimens
ACI Structural Journal/March-April 2014

Fig. 6Predicted versus recorded maximum normalized


loads.
cast with square longitudinal bars cast in the top position
attained a mean predicted-recorded maximum normalized
load ratio of 1.02, and a standard deviation of 0.151. Results
for specimens longitudinally reinforced with round bars in
the bottom position had a mean predicted-recorded load
ratio of 0.983, and a standard deviation of 0.159. In contrast,
specimens longitudinally reinforced with round bars cast in
the top position attained a mean value and standard deviation for the predicted-recorded applied load ratios of 1.22
and 0.340, respectively. The higher standard deviation
obtained for specimens cast with round longitudinal bars in
the top position is likely due to the limited number of such
specimens in the test database and the sensitivity of round
bars to casting position, as outlined in the previous section.
SUMMARY AND CONCLUSIONS
Twenty-five lap splice specimens with a shear span-depth
ratio approximately equal to 3.94 were reinforced with plain
round or square longitudinal steel bars in the top or bottom
position to evaluate the effects of casting position and bar
shape on bond. Specimens were 305 mm (12 in.) wide by
410 mm (16 in.) tall, with a span length of 4570 mm (15 ft),
and were subjected to four-point loading. Lap splice lengths
ranged from 12.8 to 32.1 times the longitudinal bar diameter or side face dimension for the case of round and square
longitudinal bars, respectively. The following significant
observations and conclusions were noted:
329

All specimens failed due to bond loss between the longitudinal reinforcement and the surrounding concrete;
Square longitudinal bars may be evaluated by calculating their equivalent round diameter, based on equal
cross-sectional areas of the actual square reinforcing bar
and the equivalent round bar;
Predictions of the maximum applied load based on
CEB-FIP Model Code 1990 provisions for bond reasonably and conservatively capture the behavior of specimens
with bottom cast round bars, but do not appear to capture
the behavior of specimens with bottom- or top-cast square
bars or specimens with top-cast round bars;
Maximum load predictions based on the CEB-FIP draft
Model Code 2010 provisions for bond are overly conservative for all combinations of bar shapes and casting positions;
Square bars appear to be less sensitive to casting position than round bars. Top cast factors of 0.3 and 0.6 for
round and square bars, respectively, reasonably capture
the reductions in bond resistance based on the range of
parameters evaluated in this study; and
A regression analysis of the specimens shows that a
linear and proportional relationship for maximum load
as a function of lap splice length, casting position, and
equivalent diameter provides a best fit for the test data.
ACKNOWLEDGMENTS

Financial support was provided by a Natural Science and Engineering


Council of Canada Discovery Grant for the second author and by scholarship support for the first author from the University of Saskatchewan.

AUTHOR BIOS

Montserrat Sekulovic MacLean is a Junior Engineer with KTA Structural


Engineers Ltd., Calgary, AB, Canada. She received her MSc in the Department of Civil and Geological Engineering at the University of Saskatchewan, Saskatoon, SK, Canada.
Lisa R. Feldman, FACI, is an Associate Professor in the Department
of Civil and Geological Engineering at the University of Saskatchewan.
She is a member of ACI Subcommittee 318-R, Code Reorganization, and
Chair of Joint ACI-ASCE Committee 408, Development and Splicing of
DeformedBars.

330

=
=
=

NOTATION

shear span
effective depth of reinforced splice specimens
diameter for round bars or side face dimension for
square bars
db,EQ
=
equivalent diameter for square bars
E s
=
modulus of elasticity of reinforcement
fc
=
concrete compressive strength
fck
= characteristic value of cylinder compressive
strength of concrete
fctd
=
design value of concrete tensile strength
fu
=
ultimate strength of reinforcement
fyd
=
dynamic yield strength of reinforcement
fys
=
static yield strength of reinforcement
kb
=
indicator variable for bar shape
kc
=
indicator variable for casting position
Ls
=
spliced length of longitudinal reinforcing bars
P
=
applied load
Pmax
=
maximum applied load
(Pmax)CEB,1990
= maximum applied load predicted using CEB-FIP
Model Code 1990 provisions
(Pmax)CEB,2010
=
maximum applied load predicted using CEB-FIP
draft Model Code 2010 provisions
R y
=
bar surface roughness
uave
=
average bond stress
c
=
partial factor for concrete compressive strength
1
= factor to describe reinforcing type
a
d
db

2
3
4
t
l, n
,

=
=
=
=
=
=

factor to account for bond conditions


factor to account for bar size
factor to account for characteristic yield strength of reinforcement
modification factor for bar shape and casting position
symbols identifying bar shape
symbols identifying casting position of reinforcement

REFERENCES

1. Feldman, L. R.; MacFarlane, D. C.; Kroman, J. A.; and Bartlett, F.


M., Construction Staging of the Centre Street Bridge Rehabilitation to
Accommodate Emergency Vehicle Traffic, 31st Annual Conference of the
Canadian Society for Civil Engineering, 2003, 10 pp. (CD-ROM)
2. Baldwin, M. I., and Clark, L. A., The Assessment of Reinforcing
Bars with Inadequate Anchorage, Magazine of Concrete Research, V. 47,
No. 171, June 1995, pp. 95-102.
3. Hassanain, M. A., and Loov, R. E., Cost Optimization of Concrete
Bridge Infrastructure, Canadian Journal of Civil Engineering, V. 30,
No.5, Oct. 2003, pp. 841-849.
4. Mirza, M. S., and Haider, M., The State of Infrastructure in Canada:
Implications for Infrastructure Planning and Policy, McGill University,
Montreal, QC, Canada, 2003, 53 pp.
5. Abrams, D. A., Tests of Bond Between Concrete and Steel, University of Illinois Bulletin No. 71, University of Illinois at Urbana-Champaign,
Urbana, IL, 1913, 240 pp.
6. Feldman, L. R., and Bartlett, F. M., Design of a Testing Program for
Bond of Plain Reinforcement, 5th International PhD Symposium in Civil
Engineering, Delft, the Netherlands, 2004, pp. 145-153.
7. Hassan, M. N., and Feldman, L. R., Behavior of Lap-Spliced Plain
Steel Bars, ACI Structural Journal, V. 109, No. 2, Mar.-Apr. 2012,
pp. 235-243.
8. Feldman, L. R., and Bartlett, F. M., Bond Strength Variability in
Pullout Specimens with Plain Reinforcement, ACI Structural Journal,
V.102, No. 6, Nov.-Dec. 2005, pp. 860-867.
9. Idun, E. K., and Darwin, D., Improving the Development Characteristics of Steel Reinforcing Bars, SM Report No. 41, University of Kansas
Center for Research, Lawrence, KS, 1995, 267 pp.
10. CAN/CSA-A23, 1/A23.2-09, Concrete Materials and Concrete
Construction/Test Methods and Standard Practices for Concrete, Canadian
Standards Association, Mississauga, ON, Canada, 2009, 582 pp.
11. ASTM A370-97a, Standard Test Methods and Definitions for
Mechanical Testing of Steel Products, ASTM International, West Conshohocken, PA, 1997, 52 pp.
12. Rao, N. R. M.; Lohrmann, M.; and Tall, L., Effects of Strain Rate
on the Yield Stress of Structural Steels, ASTM Journal of Materials, V. 1,
No. 1, May 1966, pp. 241-262.
13. Mitutoyo, SJ-201 Surface Roughness Tester Users Manual No.
99MBB0796A, Mitutoyo Corporation, Kanagawa, Japan, 2006, 190 pp.
14. CEB-FIP, CEB-FIP Model Code (1990), Comit Euro-Internationale du Bton (CEB), Thomas Telford Ltd., London, UK, 1993, 437 pp.
15. CEB-FIP, fib Bulletin 55: Model Code 2010, First Complete Draft
Volume 1, Comit Euro-Internationale du Bton (CEB), International Federation for Structural Concrete (fib), Lausanne, Switzerland, 2010, 292 pp.
16. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-11) and Commentary, American Concrete Institute,
Farmington Hills, MI, 2011, 503 pp.
17. ASTM A305-49, Minimum Requirements for the Deformations of
Deformed Bars for Concrete Reinforcement, ASTM International, West
Conshohocken, PA, 1949, 3 pp.
18. Howell, D. A., and Higgins, C., Bond and Development of
Deformed Square Reinforcing Bars, ACI Structural Journal, V. 104,
No.3, May-June 2007, pp. 333-343.
19. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-05) and Commentary (ACI 318R-05), American
Concrete Institute, Farmington Hills, MI, 2005, 430 pp.
20. CAN/CSA-A23, 3-04, Design of Concrete Structures, Canadian
Standards Association, Mississauga, ON, Canada, 2004, 258 pp.
21. Chana, P. A., A Test Method to Establish a Realistic Bond Stress,
Magazine of Concrete Research, V. 24, No. 151, 1990, pp. 83-90.
22. Feldman, L. R., and Bartlett, F. M., Bond Stresses Along Plain Steel
Reinforcing Bars in Pullout Specimens, ACI Structural Journal, V. 104,
No. 6, Nov.-Dec. 2007, pp. 685-692.
23. Sylev, T. A., The Effect of Fibres on the Variation of Bond Between
Steel Reinforcement and Concrete with Casting Position, Construction &
Building Materials, V. 25, No. 4, Apr. 2011, pp. 1736-1746.
24. Barnes, R. W.; Burns, N. H.; and Kreger, M. E., Development
Length of 0.6-Inch Prestressing Strands in Standard I-Shaped Pretensioned
Concrete Beams, Research Report 1388-1, Center for Transportation
Research, University of Texas at Austin, Austin, TX, 2000, 338 pp.

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S29

Performance of Glass Fiber-Reinforced Polymer-Doweled


Jointed Plain Concrete Pavement under Static and
CyclicLoadings
by Brahim Benmokrane, Ehab A. Ahmed, Mathieu Montaigu, and Denis Thebeau
Glass fiber-reinforced polymer (GFRP) dowel bars are a
non-corrodible and maintenance-free alternative that will potentially reduce the life-cycle cost of jointed plain-concrete pavement
(JPCP), especially in harsh environmental conditions. This paper
investigates the performance of GFRP dowels in JPCP under
static and cyclic loads. In addition, it compares their behavior
with that of commonly used epoxy-coated steel dowels. GFRP and
epoxy-coated steel dowels were employed in fabricating a total
of six JPCP prototypes (slab-joint). The test prototypes measured
2440mm long x 610 mm wide x 254 mm deep (96 x 24 x 10 in.). The
slabs were cast with a butted joint; each test prototype contained
two dowel bars. The test parameters included: 1) dowel-bar type
(GFRP and epoxy-coated steel); 2) dowel-bar diameter (34.9 and
38.1 mm [1.38 and 1.50 in.] for GFRP; 28.6 mm [1.13 in.] for
epoxy-coated steel); and 3) loading scheme (static and cyclic).
The test results revealed that both 34.9 and 38.1 mm (1.38 and
1.50 in.) GFRP dowels showed crack patterns and failure modes
similar to those of the epoxy-coated steel dowels. The 34.9 and
38.1 mm (1.38and 1.50 in.) GFRP dowels and 28.6 mm (1.13 in.)
epoxy-coated steel dowels were not affected by 1,000,000 cycles
between 10 and 50 kN (2.25 and 11.24 kip). In addition, both the
34.9 and 38.1 mm (1.38 and 1.50 in.) GFRP dowels showed higher
joint effectiveness than that of the 28.6 mm (1.13 in.) epoxy-coated
steel dowels. This paper also discusses the results of a field application in which the GFRP dowels were implemented in a new
concrete-pavement highway in Mirabel, QC, Canada.
Keywords: cyclic; design; dowel; fiber-reinforced polymers; field application; joint; joint effectiveness; pavement; static.

INTRODUCTION
Jointed plain-concrete pavements (JPCPs) are commonly
constructed with contraction joints to accommodate slab
movements due to temperature and moisture variations
(Ioannides and Korovesis 1992). The joints may either be
longitudinal, parallel to traffic, or transverse, perpendicular
to traffic. Transverse joints are placed at regular intervals,
creating discontinuities in the pavement and forming a series
of slabs. Load transfer within a series of concrete slabs takes
place across these joints. Thus, an effective load-transfer
device should be present to transfer the loads between adjacent slabs (Porter 2005).
Dowel bars are installed at the transverse joints of the
concrete slabs to reduce the deflections and stresses at
the joints while transferring the traffic load from one slab
to the adjacent slab (Westergaard 1928). In addition, the
dowel-bar system works well with both narrow and wide
joints (Maitra et al. 2009). The load-transfer efficiency of
a joint is assessed by joint effectiveness E, as specified by
the American Concrete Pavement Association (ACPA 1991).
ACI Structural Journal/March-April 2014

Joint effectiveness E based on the measured deflections of


loaded and unloaded sides of the joint is given by Eq. (1)

E = 2 u ( l + u ) 100 (1)

where E is the joint effectiveness; u is the deflection of the


unloaded slab; and l is the deflection of the loaded slab.
When the deflections of the slabs on both sides of the joint
are equal, the joint is considered 100% effective. On the
other hand, if the unloaded side of the joint experiences
no deflection, the joint is considered 0% effective. ACPA
(1991) recommends an adequate effectiveness of 75% in
pavement joints.
Load-transfer efficiency LTE (AASHTO 1993) is another
quantitative measure for assessing joint efficiency. Equation(2) provides deflection-based LTE

LTE = u l 100 (2)

ACPA (1991) considers corrosion of steel dowels as one


of the main reasons of premature concrete failure, resulting
from corrosion initiated by the chloride from deicing salts
infiltrating the joints. Because the smooth dowel bars are
assumed to be frictionless and permit free relative movement
of slabs due to temperature changes, the corrosion causes
the steel surface to exfoliate, so that the dowels lock in the
joint. Locking, in turn, exerts excessive tensile loads on the
surrounding concrete with attendant stress concentration on
the concrete and greater movement around the joint, which
hasten the rate of joint failure (FRP Dowel Bar Team 1998).
Furthermore, corrosion of the steel dowels may restrain free
movement under temperature variations. The restriction of
free movement of slabs with respect to the dowels produces
high tensile stresses, which, in turn, result in mid-slab
cracking (William and Shoukry 2001).
A common method to make steel dowels more durable
is coating them with fusion-bonded epoxy. Unfortunately, epoxy-coated steel dowels exhibit the same performance problems normally associated with surface coating,
including voids and damage during transportation and
handling. The concentrated corrosive mechanisms at defects
have led, in some cases, to more rapid failure than the same
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-098, doi:10.14359.51686525, was received March 16, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

331

steel reinforcement without an epoxy coating (FRP Dowel


Bar Team 1998).
Glass fiber-reinforced polymer (GFRP) dowels do
not corrode, and are maintenance-free. Consequently,
employing GFRP dowels in JPCP as an alternative to the
commonly used epoxy-coated dowels will eliminate the
potential of corrosion and extend the service life of jointed
concrete pavements. A few studies have been conducted to
investigate the feasibility of GFRP dowel bars as an alternative to steel in jointed concrete pavements (Porter et al.
1996; Davis and Porter 1998; Porter et al. 2001; Eddie et
al. 2001; Smith 2001). These investigations supported the
use of GFRP dowels as a potential solution to the corrosion
issues of steel dowels in JPCP.
Unlike the United States, which has the Federal Highway
Administration (FHWA), Canada has no single agency
responsible for funding pavement construction and rehabilitation or for setting pavement design standards. Pavement
design for Canadas primary highway network comes under
provincial jurisdiction, while the federal government retains
responsibility for national-park roadways. Each agency is
free to use whatever design procedure it chooses for JPCP
design and rehabilitation (Canadian Strategic Highway
Research Program 2002). The Ministry of Transportation
of Quebec (MTQ) has attempted to overcome corrosion-associated problems when steel dowels are used. More than
350km (217.5 miles) of JPCP has been built in Quebec since
1994. The method commonly employed for more durable
pavement has been using epoxy-coated steel dowels. These
dowels, however, have evidenced performance problems
normally associated with surface coating, including voids,
corrosion, and damage during transportation and handling.
Through an extended collaborative project between the
MTQ and the University of Sherbrooke, new GFRP dowels
were developed, and their long-term durability performance
was assessed (Montaigu et al. 2013). In this investigation,
the GFRP dowels were conditioned in harsh environments
at high temperatures for different duration periods. The findings of this investigation demonstrated the high stability of
vinylester-based GFRP dowels in concrete environment.
This paper, however, investigated the performance of these
newly developed vinylester-based GFRP dowels in JPCP
under static and cyclic loadings in laboratory conditions.
In addition, it compared their structural performance with
that of epoxy-coated steel dowels in JPCP. This paper also
discusses the results of a field application in which the GFRP
dowels tested were installed in a new concrete-pavement
highway (Hwy 15 North) in Mirabel, QC, Canada.
RESEARCH SIGNIFICANCE
GFRP dowels appeared as a possible cost-effective solution for JPCP. Few studies have investigated the feasibility
of using GFRP dowels in pavements. The reported study
investigated the structural performance and effectiveness
of the newly developed vinylester-based GFRP dowel bars
(34.9 and 38.1 mm [1.38 and 1.50 in.]) under static and
cyclic loadings, and in a field application. In addition, it
compared the behavior of JPCP with either GFRP or epoxycoated steel dowels in transverse joints.
332

Fig. 1Details of test prototypes and GFRP dowels.


EXPERIMENTAL PROGRAM
GFRP dowel characterization
The research involved GFRP dowels measuring 34.9and
38.1 mm (1.38 and 1.50 in.) in diameter (Fig. 1) and epoxycoated steel reference dowels 28.6 mm (1.13 in.) in diameter. The GFRP dowels were made of continuous E-glass
fibers in vinylester resin using the pultrusion process.
The fiber content was 80.6% by weight. The physical and
mechanical properties of the GFRP dowels were determined using the appropriate test methods provided in
CSAS807 (Canadian Standards Association 2010) and
ACI 440.6 (ACICommittee 440 2008). Mechanical characterization included testing representative specimens of the
GFRP dowel bars to determine transverse shear strength
(ACI440.3R [ACI Committee 440 2004], Test Method B.4),
interlaminate shear (short-beam test) (ASTM D4475), flexural strength, and flexural modulus of elasticity (stiffness)
(ASTM D4476). Table 1 presents the physical and mechanical properties of the GFRP dowels obtained from testing.
Test prototypes
A total of six JPCP prototypes were constructed and tested.
The prototypes included two reinforced with GFRP dowels
34.9 mm (1.38 in.) in diameter, two reinforced with GFRP
dowels 38.1 mm (1.50 in.) in diameter, and two reinforced
with epoxy-coated steel dowels 28.6 mm (1.13in.) in diameter. For each dowel type and diameter, one prototype was
tested under monotonic load (Phase I), while the second one
was tested using a cyclic-load scheme (Phase II). The test
prototypes were 2440 mm long x 610 mm wide x 254mm
deep (96 x 24 x 10 in.). The slabs were cast with a 19 mm
(0.75 in.) butt joint, and each test prototype had two dowel
bars spaced 305 (12 in.) mm apart. Figure 1 shows the
dimensions and details of the pavement prototypes.
The length of the specimens was selected based on
finite-element modeling (Maitra et al. 2009), which simulated the experimental study conducted by the U.S. Naval
Civil Engineering Research and Evaluation Laboratory
(Keeton and Bishop 1957). This study revealed that the
vertical shear force in a dowel beyond a distance of approximately 1200 mm (4 ft) from the center of the load was insignificant. Thus, 2440 mm (8 ft) was selected as the total length
for the jointed slab prototypes for this study. The geometry

ACI Structural Journal/March-April 2014

Table 1Physical and mechanical properties of GFRP dowel bars


Physical properties
GFRP dowel diameter, mm

34.9

Mechanical properties
GFRP dowel diameter, mm

34.9

38.1

Fiber type

Glass E-type

38.1

Transverse shear strength, MPa

184 2

173 3

Resin type

Vinyl ester resin

Short beam shear strength, MPa

61 0

54 2

Fiber content, %

80.7

80.6

Four-point flexural strength, MPa

1210 50

1077 61

Cure ratio, %

100

100

Flexural modulus of elasticity, GPa

50.3 0.5

51.6 0.8

Tg, oC

124

123

Moisture uptake, %

0.06

0.07

Notes: 1 mm = 0.0394 in.; 1 MPa = 0.145 ksi; 1 GPa = 145 ksi; C = 5/9(F 32).

and dimensions of the slab prototypes were consistent with


jointed slab prototypes tested elsewhere (Eddie et al. 2001).
The prototypes were fabricated with a MTQ Type III-A
concrete with a target 28-day compressive strength of
35 MPa (5.1 ksi), as specified in Standard 3101 for MTQ
normal-density mass concrete (MTQ 2009a). The concrete
mixture contained 380 kg/m3 (23.7 lb/ft3) of GUb-SF
cement, and had a water-cement ratio (w/c) of 0.42, with
high-range water-reducing admixture to maintain a mixture
slump of 120 30 mm (4.7 1.18 in.) (MTQ 2009a). The
pavement prototypes were cast at the structural laboratory
using three concrete batches. The average concrete strength
of the concrete batches was 48.0 3.5 MPa (7.0 0.5 ksi)
based on testing of three concrete cylinders (150 x 300 mm
[5.9 x 11.8 in.]) from each batch.

Table 2Loading schemes of test prototypes

Subgrade base layer


The granular base consisted of three 100 mm (4 in.) thick
layers of limestone aggregate compacted using a 90 kg
(198 lb) vibrating plate. The granular mixture was prepared
according to AASHTO specifications (Class A). The granular subgrade mixture consisted of 50% sand (0 to 5 mm
[0to 0.2 in.]), 20% 10 mm (0.4 in.) crushed rock (5 to 14 mm
[0.2 to 0.6 in.]), and 30% 20 mm (0.8 in.) crushed rock (14 to
28 mm [0.6 to 1.1 in.]). Aggregates were dampened before
placing to maximize the compaction efficiency. Once the
base was completed, a thin layer of sand was applied to the
final surface to provide contact between the concrete surface
and subgrade. The base was extended by 300 mm (12 in.) on
all sides to allow for load distribution and prevent failure of
the base-layer container. The overall dimensions of the base
layer were 1.52 m wide x 3.35 m long x 0.30 m deep (5 x 11x
1 ft). Upon completing the base, the base modulus (stiffness)
was measured using a Briaude Compacting Device (BCD),
and was 110 MPa/m (4.9 ksi/ft).

Table 2 summarizes the loading schemes, while Fig.2


shows the test setup.
For static testing (Phase I), the monotonic load was applied
with a stroke-controlled rate of 0.01 mm/sec (0.02in./min)
to allow for progressive contact and loading. The load was
applied using a 1000 kN (225 kip) hydraulic actuator on one
side of the joint over a loading plate of 306 mm (12 in.) in
diameter. The prototypes were loaded up to 200 kN (45kip),
then the load was released. Thereafter, the prototypes were
loaded again at the same rate until failure. E and LTE were
calculated at an applied load of 40 kN (9 kip) (service load,
which is equal to one half the equivalent axle load) from
the deflection measurements of two linear variable differential transducers (LVDTs) on both joint sides (loaded
andunloaded).
For the fatigue testing (Phase II), the prototypes were
tested up to 1 million cycles. The load followed a sinusoidal
waveform that varied from 10 to 50 kN (2.25 to 11.24 kip).
The minimum load (10 kN [2.25 kip]) was required to maintain contact between the slab and loading plate and to minimize the impact on the subgrade. The maximum load (50kN
[11.24 kip]) was set to achieve the service load and keep
40kN (9 kip) as the cyclic test amplitude, which is equal to
one-half the equivalent axle load (service load). It should
be mentioned that this loading scheme closely represents
field conditions under which load is applied and removed as
a vehicle approaches the joint or moves away from it. The
load was applied with the same hydraulic actuator (1000kN
[225 kip]) with a load-controlled scheme. The loading

Testing loads and procedures


The JPCP prototypes were tested under two different
loading conditions: static (Phase I) and cyclic (Phase II).
During Phase I, the prototypes were monotonically loaded
to 200 kN (45 kip) to induce cracks at the joints. Thereafter,
the load was released, and the prototypes were loaded again
up to failure. During Phase II, the prototypes were subjected
to 1 million cycles ranging from 10 to 50 kN (2.25 and
11.24kip), followed by monotonic loading up to failure.
ACI Structural Journal/March-April 2014

Phase

Number and prototypes

Loading scheme

One 34.9 mm (1.38 in.)


GFRP
I
Static

One 34.9 mm (1.50 in.)


GFRP

Monotonic to 200 kN (45 kip),


unloading, monotonic reloading
to failure.

One 28.6 mm (1.13 in.)


epoxy-coated steel
One 34.9 mm (1.38 in.)
GFRP
II
Cyclic

One 34.9 mm (1.50 in.)


GFRP
One 28.6 mm (1.13 in.)
epoxy-coated steel

One million cycles between 10


and 50 kN (2.24 to 11.24 kip) at
15 Hz. Thereafter, static testing
until failure.

333

Fig. 2Test setup: (a) overall view; and (b) loading plate and linear variable displacement transducers.
Table 3Summary of test results
Cracking and failure loads of jointed pavement prototypes
Phase

Load at

Steel, 28.6 mm

GFRP, 34.9 mm

GFRP, 38.1 mm

I
Static

Cracking, kN (kip)

140.7 (31.6)

100.0 (22.5)

124.8 (28.1)

Failure, kN (kip)

506.6 (113.9)

460.0 (103.4)

478.0 (107.5)

II
Cyclic

Cracking, kN (kip)

250 (56.2)

178 (40.0)

145 (32.6)

Failure, kN (kip)

622 (139.8)

526 (118.2)

413 (92.8)

Joint effectiveness and load-transfer efficiency at service load, 40 kN (9 kip)


Steel, 28.6 mm

GFRP, 34.9 mm

GFRP, 38.1 mm

Phase

Load type

E, %

LTE, %

E, %

LTE, %

E, %

I
Static

First loading

86

75

89

81

95

Reloading

65

45

64

47

74

II
Cyclic

First loading

95

90

95

90

96

After cycling

92

85

93

87

95

Note: 1 mm = 0.0394 in.

and unloading was applied at a frequency of 15 Hz. This


frequency is equivalent to the time that a vehicle needs to
cross the joint, assuming a speed of 65 to 80 kph (37.3to
49.7 mph) (MTQ 2009b). Because the test prototypes did
not fail after 1 million cycles, the prototypes were retested
under monotonic static load until failure. Before cycling, as
well as after predetermined sets of cycles (1; 1000; 10,000;
100,000; 500,000; and 1,000,000), the load cycling was
interrupted, and a monotonic loading test up to 40 kN (9 kip)
(service load) was conducted to assess joint performance.
TEST RESULTS
Static testing (Phase I)
Cracking and failureWhen the JPCP prototypes were
submitted to 200 kN (45 kip), cracks appeared in the loaded
slabs. The unloaded slabs, however, did not evidence any
cracks during the test. Table 3 gives the cracking and failure
loads of the test prototypes. Figure 3 shows the cracking
patterns and failure modes of the three prototypes tested in
334

Phase I: 28.6 mm (1.13 in.) steel dowels, 34.9 mm (1.38 in.)


GFRP dowels, and 38.1 mm (1.50 in.) GFRP dowels.
In the case of the 28.6 mm (1.13 in.) diameter steel dowels,
the first crack appeared in the loaded slab at 140.7kN
(31.6kip) on one side, and at 197.0 kN (44.3 kip) on the
other. The crack appeared at the level of dowel bars, and
continued to the surface. The second crack appeared under a
load of approximately 380 kN (85.4 kip). The slab prototype
failed at 506.6 kN (113.9 kip) by shear failure of the loaded
slab beyond the dowel bars. The GFRP-doweled prototypes
showed crack patterns and failure modes similar to the
steel-doweled prototype. The cracking loads were 100and
124.8 kN (22.4 and 28.1 kip) for both 34.9 and 38.1mm
(1.38 and 1.50 in.) diameter GFRP dowels, respectively.
These loads represent 71 and 89% of the cracking load of
the steel-doweled prototype. On the other hand, the failure
loads for the 34.9 and 38.1 mm (1.38 and 1.50 in.) diameter
GFRP dowels were 460 and 478 kN (103.4 and 107.4 kip),
respectively, which represent 91 and 94% of the failure load
of the steel-doweled prototype.
ACI Structural Journal/March-April 2014

Fig. 3Cracking at failure of Phase I prototypes: (a) 28.6 mm steel dowels; (b) 34.9 mm GFRP dowels; and (c) 38.1 mm GFRP
dowels. (Note: 1 mm = 0.0394 in.)

Fig. 4Results of Phase I prototypes (static testing up to 200 kN): (a) joint effectiveness; (b) load-transfer efficiency; and
(c)relative deflection. (Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.)
Joint effectiveness E and load-transfer efficiency LTE
Eand LTE were calculated using the deflection measurements recorded by the two LVDTs placed on the unloaded
and loaded sides of the joint. Table 3 gives the calculated E
and LTE at service load (40 kN [9 kip]) for the tested JPCP
prototypes with steel and GFRP dowels. All tested prototypes showed E and LTE higher than 75 and 60%, respectively, which meets ACPA (1991) requirements.
Both GFRP dowel diameters (34.9 and 38.1 mm [1.38and
1.50 in.]) displayed E and LTE higher than the 28.6 mm
diameter (1.13 in.) steel dowels. The 34.9 mm (1.38 in.)
diameter GFRP dowels showed E of 89% and LTE of 81%,
while the 38.1 mm (1.50 in.) diameter GFRP dowels showed
E of 95% and LTE of 92%. The values in Table 3 reveal that
using 34.9 mm (1.38 in.) diameter GFRP dowels instead of
28.6 mm diameter (1.13 in.) steel dowels increased E and
LTE by 9 and 8%, respectively. On the other hand, replacing
28.6 mm diameter (1.13 in.) steel dowels with 38.1 mm
diameter (1.50 in.) GFRP dowels increased E and LTE by
10 and 23%, respectively. In addition, E and LTE values
revealed that the behavior of the jointed pavement with
34.9 mm (1.38 in.) diameter GFRP dowels was almost the
same as that with 28.6 mm (1.13 in.) diameter steel dowels.
Furthermore, Table 3 shows that reloading the specimens
after cracking during the initial loading phase (200 kN
[45kip]) yielded very low E and LTE because of the cracks.
The 38.1 mm (1.50 in.) diameter GFRP dowels evidenced
joint effectiveness E of 74%, which is very close to 75%, as
provided for by ACPA (1991). Therefore, JPCP stability and
performance is dependent on the slabs remaining uncracked
to achieve efficient joints.
E and LTE were plotted against applied load in Fig. 4(a)
and (b). It should be mentioned that LTE corresponding to
E = 75% was 60%. Figure 4(a) and (b) demonstrate that,
after an initial loading interval till about 50 kN (11.24 kip), E
and LTE stabilized. Besides, there was no significant differACI Structural Journal/March-April 2014

ence between the two prototypes with 35.9 mm (1.38in.)


GFRP dowels and 28.6 mm (1.13 in.) steel dowels after
50kN (11.24 kip). The differences between the two prototypes under 50 kN (11.24 kip) may be related to the better
compaction of the subgrade base after testing the first prototype, which had 28.6 mm (1.13 in.) steel dowels. Furthermore, increasing the GFRP dowels to 38.1 mm (1.50 in.)
increased E and LTE. Figure 4(a) also shows that cracking
during the first loading did not affect E, because the load was
not released until it reached 200 kN (45 kip).
Relative deflectionFigure 4(c) provides the relative
deflection of loaded and unloaded slabs of three prototypes
with 28.6 mm (1.13 in.) epoxy-coated steel dowels, 34.9mm
(1.38 in.) GFRP dowels, and 38.1 mm (1.50 in.) GFRP
dowels. The relative deflection of the 28.6 mm (1.13in.) steel
dowels was between those of the 34.9 and 38.1 mm (1.38 and
1.50 in.) GFRP dowels. Figure 4(c) shows immediate deflection at the beginning of the test ranging from 0.2 to 0.45 mm
(0.01 to 0.02 in.). This immediate deflection occurred because
the specimens were not cast directly on the subgrade base
and, when the load was applied to the pavement prototypes,
the immediate deflection occurred until complete contact
between the concrete surface and the subgrade base layer was
achieved. The steel-doweled pavement prototype showed the
highest immediate deflection (0.45 mm [0.02 in.]) because
it was the first tested, and may have been affected by the
compressibility of the subgrade layer. The immediate deflection was less in the case of the GFRP-doweled pavement
prototypes (0.20mm [0.01in.]). This immediate deflection
increase might not occur in field applications in which the
pavement is cast directly on thesubgrade.
At service load (40 kN [9 kip]), the 28.6 mm (1.13 in.)
steel dowels showed a relative deflection of 0.58 mm
(0.02in.). The 34.9 and 38.1 mm (1.38 and 1.50 in.) GFRP
dowels evidenced relative deflections of 0.82 and 0.36 mm
(0.03 and 0.01 in.), respectively. The relative deflections
335

Fig. 5Cracking at failure of Phase II prototypes: (a) 28.6 mm steel dowels; (b) 34.9 mm GFRP dowels; and (c) 38.1 mm
GFRP dowels. (Note: 1 mm = 0.0394 in.)
of the 34.9 and 38.1 mm (1.38 and 1.50 in.) GFRP dowels
were 1.41 and 0.62 times that of the 28.6 mm (1.13 in.) steel
dowels, respectively. After the service load, the difference
between the relative deflections of the 28.6 mm (1.13 in.)
steel dowel and 34.9 mm (1.38 in.) GFRP dowels decreased.
At approximately 70 kN (15.7 kip), the difference between
the two prototypes was less than 0.1 mm (0.004 in.). On the
other hand, the relative deflection of the 38.1 mm (1.50 in.)
GFRP dowels was very small compared with that of the
28.6mm (1.13 in.) epoxy-coated steel dowels.
Cyclic testing (Phase II)
Cracking and failureThe three prototypes with steel
and GFRP dowels did not experience any cracking after
1,000,000 cycles at 40 kN (9 kip) (between 10 and 50 kN
[2.25 and 11.24 kip]). Thus, using durable dowel bars in
jointed pavements will yield efficient joints with extended
service life under service load. This confirms the findings
of the static testing (Phase I): JPCP efficiency will not be
altered as long as the concrete does not crack.
After 1,000,000 cycles, the three prototypes were retested
under monotonic load up to failure. Table 3 lists the cracking
and failure loads. The cracking load of the prototypes in
Phase II was higher than those in the Phase I prototypes. The
cyclic testing of the three prototypes affected the subgrade
base, and resulted in very high compaction. That, in turn,
affected the cracking loads of the test prototypes. Similarly,
the failure loads in Phase II were also higher than those in
Phase I, except for the prototype with 38.1 mm (1.50 in.)
GFRP dowels, which showed 413 kN (92.9 kip) compared
with 478 kN (107.4 kip) in Phase I (static testing).
Table 3 shows that the prototype with 28.6 mm (1.13in.)
epoxy-coated dowels experienced the highest cracking
load among the tested prototypes. The prototypes with
34.9 and 38.1 mm (1.38 and 1.50 in.) GFRP dowels had
cracking loads of 71 and 58% of that of the prototype with
steel dowels. As with Phase I prototypes, the first cracks
appeared at the joints at the level of the dowel bars, and then
extended to the surface. Compared with Phase I, Phase II
prototypes showed more cracks at failure, except in the case
of the 38.1 mm (1.50 in.) GFRP-doweled prototype, which
showed one major crack. The cracks led to the splitting of
concrete around the dowels before failure, which occurred
due to shear failure of the loaded slabs past dowel length.
The 28.6 mm (1.13 in.) steel-doweled prototype failed at
622kN (139.8 kip). The prototypes with 34.9 and 38.1mm
336

(1.38and 1.50 in.) GFRP dowels showed failure loads


equal to 85 and 66% of the failure load of the steel-doweled
prototype. Figure 5 shows the cracking patterns and failure
modes of the three prototypes tested in Phase II: 28.6 mm
(1.13 in.) steel dowels, 34.9 mm (1.38 in.) GFRP dowels,
and 38.1mm (1.50 in.) GFRP dowels. In addition, Table3
also shows that the prototype with 38.1 mm (1.50 in.)
GFRP dowels yielded lower cracking load and failure load
compared with that of the prototype with 34.9 mm (1.38 in.)
GFRP dowels. The normal variance in the concrete strength
between the different batches may have had an effect on the
lower cracking and failure loads.
Joint effectiveness E and load-transfer efficiency LTE
Before cycling, as well as after predetermined sets of cycles
(1; 1000; 10,000; 100,000; 500,000; and 1,000,000), the
load cycling was interrupted, and a monotonic loading test
up to 40 kN (9 kip) (service load) was conducted to assess
joint performance. Table 3 lists the calculated E and LTE at
service load (40 kN [9 kip]) for the JPCP prototypes with
steel and GFRP dowels. All tested prototypes showed E and
LTE higher than 75 and 60%, respectively, which meets
ACPA (1991) requirements. Table 3 shows that E and LTE
in Phase II were higher than in Phase I. Once again, this is
related to the excessive compaction resulting from the cyclic
testing of the prototypes. The continued cycling resulted
in further compaction of the base layer, which increased
jointeffectiveness.
E and LTE were plotted against applied load for a selected
number of cycles in Fig. 6. The 38.1 mm (1.50 in.) GFRP
dowels showed the highest E and LTE, which were 96and
92%, respectively. Figure 6 also reveals no significant
difference in E and LTE after 1,000,000 cycles. There was
no significant difference between the two prototypes with
34.9 mm (1.38 in.) GFRP dowels and 28.6 mm (1.13 in.)
steel dowels (E and LTE equal 95 and 90%, respectively).
DESIGN OF GFRP DOWELS
Because the properties of GFRP dowel are different from
those of steel, directly replacing steel dowels with GFRP
dowels of the same diameter and at the same spacing is
not valid. The required diameter and spacing can be determined by equating the relative deflection of a joint doweled
with steel to that of a joint doweled with GFRP (Davis and
Porter 1998). The effective design steps for GFRP-doweled
JPCP are: 1) determine the load transferred by the critical
dowel; 2) determine the relative deflection for a joint with
ACI Structural Journal/March-April 2014

Fig. 6Results of Phase II prototypes (cyclic loading) at 40 kN: (a) joint effectiveness; (b) load-transfer efficiency. (Note:
1mm = 0.0394 in; 1 kN = 0.225 kip.)
steel dowels; 3) determine the relative deflection for a joint
with GFRP dowels; 4) determine the required diameter and
spacing for GFRP dowels; and 5) check the bearing stresses.
This design procedure is summarized as follows.
When a load is applied to the edge of a slab, a portion of the
load is transferred to the adjacent slab through the dowels by
shear. Tabatabie et al. (1979) suggested that only the dowels
located within a distance of lr from the load point contributes
to transferring the load (Fig. 7(a)), where lr is the radius of
relative stiffness as defined in Eq. (3) by Westergaard (1925)

lr = 4 Ec h3 12 1 2 k (3)

where Ec is the modulus of elasticity of the pavement


concrete; h is the pavement thickness; is the Poissons
ratio of the pavement concrete; and k is the modulus of
subgrade reaction.
Corresponding to the contribution of the dowels located in
the lr distance, the load transferred by the critical dowel Pt is
given by Eq. (4)

Pt =

Design Load Transfer ( Pd )


Number of Effective Dowels

(4)

Yoder and Witczak (1975) reported that a 5 to 10% reduction in load transfer occurred due to the formation of voids
beneath the dowels at the joint face. Accordingly, a design
load transfer of 45% of the applied wheel load is recommended. Thus, the design load transfer Pd is calculated from
Eq. (5) as a function of the applied wheel load Pw

Pd = 0.45Pw (5)

Considering the schematic shown in Fig. 7(b), the relative deflection between the jointed slabs is calculated from
Eq.(6), neglecting the deflection due to the slope and flexure
along the joint width

= 2yo + (6)

where yo is the dowel deflection relative to or within the


concrete at the face; and is the shear deflection of the
dowel across the joint.

ACI Structural Journal/March-April 2014

Fig. 7Load-transfer distribution and relative deflection


between jointed slabs.
According to Friberg (1938), yo is calculated as in Eq.(7).
This equation is derived assuming the dowel bars have a
semi-infinite length. Albertson (1992), however, showed
that this equation can be applied to dowel bars with a L
value greater than 2 with a minor error, where L is the length
of dowel bar on one side of the slab

yo =

Pt
(2 + b z ) such that L > 2.0
4 b3 E I

(7a)

b = 4 Kb 4 EI (7b)

where yo is the dowel deflection relative to or within concrete


at the face; Pt is the load transferred by the critical dowel;
is the relative stiffness of the dowel bar encased in the
concrete; L is the length of dowel bar on one side of the
slab; E is the flexural modulus of elasticity; I is the moment
of inertia; b is the dowel diameter; and K is the modulus of
dowel support or reaction.
It should be mentioned that K is an important parameter in the Friberg (1938) design equation. K is determined
empirically because of the difficulty in establishing it theoretically (Friberg 1938). Yoder and Witczak (1975) found
337

that K ranges between 81.43 and 407.17 N/mm3 (3 105


and 1.5 106 lb/in.3). For the analytical calculations herein,
a value of 407.17 N/mm3 (1.5 106 lb/in.3) has been used
as suggested to simulate the worst-case scenario (Yoder and
Wiczak 1975).
The shear deflection of the dowel across the joint is then
calculated as shown in Eq. (8)

= lPt z AG (8)

where is the dowels shear deflection across the joint; l


is the shear shape factor; Pt is the load transferred by the
critical dowel; z is the joint width; A is the area of dowel bar;
and G is the shear modulus, which is assumed as 3.3 GPa
(478.6 ksi) based on the literature.
The bearing stress developed due to slab deformation
was calculated as shown in Eq. (9), which should be less
than the permissible bearing stress underneath the pavement
provided by ACI Committee 325 (1956)

b = Kyo < fb = 1 3 ( 4 b ) fc (9)

where b is the developed bearing stress; K is the dowels


modulus of support; yo is dowel deflection as the concrete
face; fb is the permissible bearing stress; b is the dowel diameter; and fc is the concrete compressive strength.
Considering the geometries and material properties of the
dowels used in this investigation, a comparable design was
conducted to determine the diameter of GFRP dowels that
could replace 28.6 mm (1.13 in.) diameter epoxy-coated steel
dowels. The design included the three diameters of GFRP
dowels used herein: 31.8, 34.9, and 38.1 mm (1.25, 1.38,
and 1.50 in.). Provided in Table 4, it shows that replacing
28.6mm (1.13 in.) epoxy-coated steel dowels with 31.8mm
(1.25in.) GFRP dowels increased the bearing stress and relative deflection by 23 and 88%, respectively. Using 34.9mm
(1.38 in.) GFRP dowels yielded almost the same bearing
stress as that of 28.6 mm (1.13 in.) epoxy-coated dowels,
while the relative deflection was 58% higher. On the other
hand, using 38.1 mm (1.50 in.) GFRP dowels yielded 29%
lower bearing stress than the 28.6 mm (1.13in.) epoxy-coated
steel dowels, while the relative deflection was 33% higher.
Considering the bearing stress as reference, the 34.9mm
(1.38 in.) GFRP dowels may be considered as a direct alternative to the 28.6mm (1.13 in.) epoxy-coated dowels when
the materials and conditions in this investigation prevail.
FIELD APPLICATION
Based on the results of this research project and the characteristics provided in Table 1, the MTQ has issued new materials
specifications for GFRP dowels as minimum requirements for
their use in JPCP in the province of Quebec (Montaigu et al.
2013): 1) the fiber type should be (E-glass) or (ECR-glass);
2) vinylester resin; 3) glass-fiber content (ASTM D3171,
Procedure G) 75%; 4) moisture absorption (ASTM D570)
0.15%; 5) glass transition temperature (ASTM D3418)
110C (230F); 6) cure ratio (CSA S807) 98%; 7) shortbeam shear strength (ASTM D4475) 59MPa (8.6ksi); and

338

8) transverse shear strength (ACI 440.3R [ACICommittee


440 2004], Test Method B.4) 170 MPa (24.7 ksi).
To take a step forward toward employing GFRP dowels
in field applications, MTQ has implemented the use of
GFRP dowels in a demonstration section on Hwy 15 North
in Mirabel, QC, Canada (just to the north of HW 50). The
traffic volume on this highway is approximately 100,000
vehicles per day, with trucks approaching 6%. The experimental section is in the right lane and the center of a length
of 500 m (1640ft) divided into two parts: 250 m (820 ft) for
38.1 mm (1.50 in.) GFRP dowels, and 250 m (820 ft) for
34.9 mm (1.38in.) steel dowels. Figure 8 shows the GFRP
dowel arrangement during the construction of the highway.
Both the GFRP and steel dowels were 450 mm (18 in.) long
and spaced at 300mm (12in.). The pavement thickness was
260 mm (10.2 in.), while the length of the jointed slabs was
set at 5000 mm (197 in.). The joints at the dowel locations
were sawn within a few hours of casting and allowed to crack
due to thermal contraction and shrinkage. The JPCP was cast
using MTQ Type IIIA concrete with a 28-day compressive
strength of 35 MPa (5.1 ksi) according to MTQ requirements
(MTQ 2009a).
As part of an ongoing MTQ effort, the performance of
the experimental section with the GFRP and steel dowels is
being assessed with a falling-weight deflectometer (FWD)
(ASTM D4694) to locate and quantify weak areas that can
be corrected before premature damage occurs. The FWD
is a nondestructive field test that involves applying impact
loads to the pavement surface and monitoring the pavement
deflection response through a series of velocity transducers
placed on the pavement at specified distances from the load,
as illustrated in the FWD test diagram in Fig. 9(a). To simulate moving-axle loads, the FWD load is applied by releasing
a mass from a selected height to impact the road surface. The
mass and drop height of the load determine the peak contact
pressure applied to the road surface.
The FWD test was conducted using the MTQ testing
vehicle (Fig. 9(b)). The falling weight is lifted to the desired
height and dropped on an anvil fitted with rubber shock
absorbers, inducing an impact of 40 kN (9 kip). The anvil
transmits the force of impact to the pavement through a plate
placed on the road surface. A series of sensors simultaneously measure pavement vertical movement resulting from
a shock wave. This movement is very subtle, because it
lasts only 30 ms, and is measured in microns. The specifications of the MTQ testing apparatus are load-plate diameter:
300mm (12 in.); falling weight: 250 kg (551 lb); impact
strength: 40 kN (9 kip); and location of sensors: 0 to 1.8 m
(0 to 5.9 ft) from the center of the plate (seven to 15 sensors).
In 2011, 20 FWD tests were conducted on each section
(steel and GFRP). The maximum, minimum, and average
LTE for the GFRP-doweled sections were 85.2, 93.6, and
88.6 2.6%, respectively. The maximum, minimum, and
average LTE for the steel-doweled sections were 85.3, 91.6,
and 87.9 1.8%, respectively. In addition, core samples
were extracted from the experimental sections during the
FWD testing, yielding an average compressive strength of
49.6 MPa (7.2 ksi). These results indicate that the steel- and
GFRP-doweled slabs had almost the same performance with
ACI Structural Journal/March-April 2014

Table 4Design for alternative GFRP dowel bar diameter


Design parameter

Reference
steel

Alternative 1
GFRP

Alternative 2
GFRP

Alternative 3
GFRP

Dowel diameter b, mm

28.6

31.8

34.9

38.1

Dowel area A, mm2

642.42

794.23

956.62

1140.09

Moment of inertia I, mm4

32,842

50,197

72,824

103,436

200,000

52,600

50,300

51,600

Dowel shear modulus G, MPa

78,000

3300

3300

3300

Dowel length L, mm

457

457

457

457

Dowel spacing s, mm

305

305

305

305

Concrete strength fc, MPa

48

48

48

48

Concretes modulus of elasticity Ec,


MPa

32,909

32,909

32,909

32,909

Concretes Poissons ratio

0.2

0.2

0.2

0.2

Pavement thickness h, mm

254

254

254

254

110

110

110

110

407.17

407.17

407.17

407.17

19

19

19

19

808

808

808

808

Number of dowels

1.87

1.87

1.87

1.87

Design load transfer = 0.45Pw, kN

18

18

18

18

Load transferred by critical dowel Pt

9.63

9.63

9.63

9.63

0.026

0.033

0.031

0.029

5.90

7.60

7.17

6.67

Pt
(2 + b z ) , mm
4 b3 E I

0.053

0.065

0.055

0.046

l Pt z
, mm
AG

0.004

0.078

0.064

0.054

= 2yo + , mm

0.110

0.208

0.175

0.147

b = Kyo, MPa

21.63

26.51

22.47

18.84

63.54

63.49

63.44

63.39

b GFRP/b steel

1.23

1.04

0.71

GFRP/ steel

1.88

1.58

1.33

Geometry and material properties

Dowel modulus of elasticity E, MPa


*

Modulus of subgrade reaction k, MPa/m


Modulus of dowel support K, N/mm

Joint width z, mm

Design

lr =

b=

Ec h 3

12 1 2 k

, mm

Kb
, mm1
4E I

L > 2.0 (ok)


yo =

fb =

(4 b) fc , MPa
3

Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip; 1 MPa = 0.145 ksi; 1 mm2 = 0.00155 in.2; 1 N/mm3 = 0.2714 kip/in.3; 1 MPa/m = 0.0442 ksi/ft; 1 mm1 = 25.4 in.1

an average LTE of 87.9 1.8% and 88.6 2.6% for the


steel- and GFRP-doweled slabs, respectively. The LTE of
both the steel- and GFRP-doweled slabs were higher than

ACI Structural Journal/March-April 2014

the 75% specified by ACPA (1991). This indicates that the


GFRP dowels tested performed efficiently in transferring
loads in JPCP.

339

SUMMARY AND CONCLUSIONS


Through an extensive collaboration project between
the Ministry of Transportation of Quebec (MTQ) and the
University of Sherbrooke, new GFRP dowels were developed, and their long-term durability characteristics were
assessed (Montaigu et al. 2013). This study, however,
investigated the performance of the newly developed GFRP
dowels (34.9 and 38.1 mm [1.38 and 1.50 in.]) in JPCP
under static and cyclic loadings. In addition, it compared
their structural performance with that of epoxy-coated steel
dowels in JPCP. Based on the test results presented and
discussed herein, the following conclusions concerning the
tested GFRP dowels can be drawn:
1. The GFRP dowel bars provided suitable and effective alternatives to the common epoxy-coated steel dowels
to overcome corrosion problems and related deterioration
under static and cyclic loading conditions;
2. Under the static testing in Phase I, the pavement prototype with 28.6 mm (1.13 in.) steel dowels showed its first
crack at 140.7 kN (31.6 kip), while the prototypes with 34.9
and 38.1 mm (1.38 and 1.50 in.) GFRP dowels showed their
first cracks at 100 and 124.8 kN (22.4 and 28.1 kip), respectively. The cracking loads of the prototypes with 34.9 and

Fig. 8GFRP dowel arrangement during construction


(Hwy 15, Mirabel, QC, Canada).

38.1 mm (1.38 and 1.50 in.) GFRP dowels were 2.5 and 3.1
times the service load (40 kN [9 kip]), respectively;
3. Under the static testing in Phase I, the pavement prototype with 28.6 mm (1.13 in.) steel dowels and those with
34.9 and 38.1 mm (1.38 and 1.50 in.) GFRP dowels showed
similar crack patterns and modes of failure. The capacities of
the three prototypes were 506.6, 460.0, and 478.0 kN (113.9,
103.4, and 107.5 kip), respectively;
4. Under the cyclic testing in Phase II, the prototypes
resisted 1 million cycles at service load without cracking.
Thus, providing durable dowel bars capable of withstanding
the environmental conditions and deterioration will yield
pavements with extended service lives;
5. Phase II produced higher cracking and ultimate loads
compared with Phase I. The results were similar for joint
effectiveness E and load-transfer efficiency LTE. This can
be accounted for by excessive compaction resulting from
prototype cyclic testing, which tended to increase joint
effectiveness;
6. The pavement prototypes with GFRP dowels (34.9 and
38.1 mm [1.38 and 1.50 in.]) showed joint effectiveness E
and load-transfer efficiency LTE similar to or higher than
that of the prototypes with steel dowels (28.6 mm [1.13 in.]);
7. Achieving stability and good performance in the
GFRP-doweled JPCP requires that slabs remain uncracked
to enable achieving efficient joints and maintain efficient
load transfer;
8. Considering the tested material and setup configuration, the 34.9 mm (1.38 in.) diameter GFRP dowels behaved
very similarly to the 28.6 mm (1.13 in.) diameter steel
dowels. Thus, the 34.9 mm (1.38 in.) GFRP dowels may be
viable alternatives to the 28.6 mm (1.13 in.) epoxy-coated
steel dowels subjected to the same loading and boundary
conditions. The test design also indicates that the 34.9 mm
(1.38in.) GFRP dowels could be direct alternatives to the
28.6 mm epoxy-coated steel dowels; and
9. The field application showed similar LTE values for of
the 38.1 mm (1.50 in.) GFRP dowels and 34.9 mm (1.38in.)
steel dowels in real service conditions. The steel- and
GFRP-doweled slabs had average LTEs of 87.9 1.8% and

Fig. 9In-place falling-weight deflectometer (FWD) test (Hwy 15, Mirabel, QC, Canada): (a) schematic; and (b) in-place test.
340

ACI Structural Journal/March-April 2014

88.6 2.6%, respectively, which were higher than the 75%


specified by ACPA (1991).
ACKNOWLEDGMENTS

The authors wish to acknowledge the financial support of Quebecs


Ministry of Transport (MTQ), the Natural Sciences and Engineering
Research Council of Canada (NSERC), and Quebecs Fonds qubcois de
la recherche sur la nature et les technologies (FQRNT). The authors are
grateful to Pultrall Inc. for providing the GFRP dowel bars and RocTest
Inc. for providing the Briaude Compacting Device (BCD). Special thanks
to the technical staff at the Department of Civil Engineering for their help
in fabricating and testing the specimens.

AUTHOR BIOS

Brahim Benmokrane, FACI, is Tier-1 Canada Research Chair Professor


in Advanced Composite Materials for Civil Structures and an NSERC
Research Chair Professor in Fiber-Reinforced Polymer Reinforcement
for Concrete Infrastructure in the Department of Civil Engineering at the
University of Sherbrooke, Sherbrooke, QC, Canada. He is a member of ACI
Committee 440, Fiber-Reinforced Polymer Reinforcement.
ACI member Ehab A. Ahmed is a Postdoctoral Fellow in the Department of Civil Engineering at the University of Sherbrooke, and Lecturer at
Menoufiya University, Shebin El-Kom, Minoufiya, Egypt. He received his
PhD in civil engineering from the University of Sherbrooke. His research
interests include structural analysis, design and testing, and structural
health monitoring of fiber-reinforced polymer concrete structures.
Mathieu Montaigu is an MSc Student in the Department of Civil Engineering at the University of Sherbrooke. His research interests include
concrete durability and performance evaluation of glass fiber-reinforced
polymer dowel bars for concrete pavement.
Denis Thebeau is a Professional Engineer with the Ministry of Transportation of Quebec (Pavement Division), Quebec City, QC, Canada. His
research interests include the design, construction, rehabilitation, and
long-term monitoring of the performance of concrete pavements.

REFERENCES

AASHTO, 1993, Guide for Design of Pavement Structures, American


Association of State Highway and Transportation Officials, Washington,
DC, 624 pp.
ACI Committee 325, 1956, Structural Design Considerations for Pavement Joints, ACI Journal, V. 53, No. 7, July, pp. 1-29.
ACI Committee 440, 2008, Metric Specification for Carbon and Glass
Fiber-Reinforced Polymer Bar Materials for Concrete Reinforcement (ACI
440.6M-08), American Concrete Institute, Farmington Hills, MI, 10 pp.
ACI Committee 440, 2004, Guide Test Methods for Fiber-Reinforced
Polymers (FRPs) for Reinforcing or Strengthening Concrete Structures (ACI
440.3R-04), American Concrete Institute, Farmington Hills, MI, 39pp.
American Concrete Pavement Association (ACPA), 1991, Design and
Construction of Joints for Concrete Highways, ACPA, Skokie, IL, 24 pp.
Albertson, M. D., 1992, Fibercomposite and Steel Pavement Dowels,
masters thesis, Iowa State University, Ames, IA, 268 pp.
ASTM D3171, 2011, Standard Test Methods for Constituent Content of
Composite Materials, ASTM International, West Conshohocken, PA, 10 pp.
ASTM D3418-12e1, 2012, Standard Test Method for Transition
Temperatures and Enthalpies of Fusion and Crystallization of Polymers by
Differential Scanning Calorimetry, ASTM International, West Conshohocken, PA, 7 pp.
ASTM D4475, 2008, Standard Test Method for Apparent Horizontal
Shear Strength of Pultruded Reinforced Plastic Rods by the Short Beam
Method, ASTM International, West Conshohocken, PA, 4 pp.
ASTM D4476, 2009, Standard Test Method for Flexural Properties
of Fiber Reinforced Pultruded Plastic Rods, ASTM International, West
Conshohocken, PA, 5 pp.
ASTM D4694, 2009, Standard Test Method for Deflections with a
Falling-Weight-Type Impulse Load Device, ASTM International, West
Conshohocken, PA, 3 pp.

ACI Structural Journal/March-April 2014

ASTM D570, 2010, Standard Test Method for Water Absorption of


Plastics, ASTM International, West Conshohocken, PA, 4 pp.
CAN/CSA S807-10, 2010, Specification for Fibre-Reinforced Polymers, Canadian Standards Association (CSA), Rexdale, ON, Canada, 27pp.
Canadian Strategic Highway Research Program (C-SHRP), 2002, Pavement Structural Design Practices Across Canada, C-SHRP Technical Brief
No. 23, Ottawa, ON, Canada, 10 pp.
Davis, D., and Porter, M. L., 1998, Evaluation of Glass Fiber Reinforced
Polymer Dowels as Load Transfer Devices in Highway Pavement Slabs,
Proceedings of Transportation Conference, Ames, IA, pp. 78-81.
Eddie, D.; Shalaby, A.; and Rizkalla, S., 2001, Glass Fiber-Reinforced
Polymer Dowels for Concrete Pavements, ACI Structural Journal, V. 98,
No. 2, Mar.-Apr., pp. 201-206.
Friberg, B. F., 1938, Load and Deflection Characteristics of Dowels in
Transverse Joints of Concrete Pavements, Proceedings of Highway Research
Board No. 18, National Research Council, Washington, DC, pp.140-154.
FRP Dowel Bar Team (DBT), 1998, Fiber-Reinforced Polymer (FRP)
Composite Dowel Bars15 Years of Durability Study, Market Development Alliance, Composites Institute, Harrison, NY, 18 pp.
Ioannides, A. M., and Korovesis, G. T., 1992, Analysis and Design of
Doweled Slab-on-Grade Pavement Systems, Journal of Transportation
Engineering, V. 118, No. 6, pp. 745-768.
Keeton, J. R., and Bishop, J. A., 1957, Load Transfer Characteristics of
a Dowelled Joint Subjected to Aircraft Wheel Loads, Proceedings of the
Highway Research Board, No. 36, Transportation Research Board, Washington, DC, pp. 190-198.
Maitra, S. R.; Reddy, K. S.; and Ramachandra, L. S., 2009, Load
Transfer Characteristics of Dowel Bar System in Jointed Concrete Pavement, Journal of Transportation Engineering, V. 135, No. 11, pp. 813-821.
Ministry of Transportation of Qubec (MTQ), 2009a, Btons de masse
volumique normale, Norme 3101, Ministre des Transports du Qubec,
Qubec, Canada, 8 pp.
Ministry of Transportation of Qubec (MTQ), 2009b, Bulletin Innovation Transport, No. 34, Ministre des Transports du Qubec, Quebec,
Canada, Jan., 42 pp.
Montaigu, M.; Robert, M.; Ahmed, E.; and Benmokrane, B., 2013,
Laboratory Characterization and Evaluation of the Durability Performance of New Polyester and Vinylester E-Glass GFRP Dowels for Jointed
Concrete Pavement, Journal of Composites for Construction, ASCE,
V.17, No. 2, pp. 176-187.
Porter, M. L., 2005, Structural Dowel Bar Alternative and Gaps of
Knowledge, Proceedings of the Mid-Continent Transportation Research
Symposium, Ames, IA, Aug., 13 pp.
Porter, M. L.; Davis, D.; Guinn, R.; Lundy, A.; and Rohner. J., 2001,
Investigation of Glass Fiber Composite Dowel Bars for Highway Pavement Slabs, Final Report TR-408, submitted to Highway Division of the
Iowa Department of Transportation and Iowa Highway Research Board,
Iowa State University, Engineering Research Institute, Ames, IA, 168 pp.
Porter, M. L.; Hughes, B. W.; and Barnes, B. A., 1996, Fiber Composite
Dowels in Highway Pavements, Proceedings of the Semisequicentennial
Transportation Conference, May, Iowa State University, Ames, IA, 6 pp.
Smith, K. D., 2001, Alternative Dowel Bars for Load Transfer in Jointed
Concrete Pavements, FHWA Technical Report Draft, Washington, DC, 16 pp.
Tabatabie, A. M.; Barenberg, E. J.; and Smith, R. E., 1979, Longitudinal Joint Systems in Slip-formed Rigid Pavements: V. II-Analysis
of Load Transfer Systems for Concrete Pavements, Report No. DOT/
FAARD-79/4, Federal Aviation Administration, U.S. Department of
Transportation, Washington, DC, 193 pp.
Westergaard, H. M., 1925, Computation of Stresses in Concrete
Roads, Proceedings of the 5th Annual Meeting of the Highway Research
Board, Washington, DC, pp. 91-112.
Westergaard, H. M., 1928, Spacing of Dowels, Proceedings of the
Highway Research Board, No. 8, Transportation Research Board, Washington DC, pp. 154-158.
William, G. W., and Shoukry, S. N., 2001, 3D Finite Element Analysis
of Temperature Induced Stresses in Dowel Jointed Concrete Pavements,
International Journal of Geomechanics, V. 1, No. 3, pp. 291-307.
Yoder, E. J., and Witczak, M. W., 1975, Principles of Pavement Design,
second edition, John Wiley & Sons, Inc., New York, 736 pp.

341

NOTES:

342

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S30

Nonlinear Static Analysis of Flat Slab Floors with Grid


Model
by Dario Coronelli and Guglielmo Corti
A grid model has been set up for the nonlinear response of a flat slab
subjected to gravity and lateral cyclic loading. The model requires
the definition of the grid geometry and properties of point hinges
in beam finite elements, and modelling the nonlinear response
in bending, torsion, and shear. The simulation is carried out for
experimental tests on a floor under gravity and lateral biaxial
cyclic loading of increasing amplitude. Pushover analyses have
been performed under gravity and horizontal loads in the two principal directions. Predictions are shown of the global response and
the connections of different column shapes and slab reinforcement
with the strength, drift capacity, and failure modes. The accuracy is
different in the two directions of loading due to the damage of the
test slab for biaxial cyclic loading. The results show the potential
of the model for design and analysis of existing flat slab structures.
Keywords: flat slabs; grid models; punching shear; slab-column connections.

INTRODUCTION
Reinforced concrete slabs, supported by columns and
walls, are among the most common structures for floors, and
their diffusion is continuously increasing all over the world.
Their use is advantageous for speed in construction, the
possibility to realize flat intrados without beams, and their
low thickness in relation to the span.
One of the leading aspects in slab design is to guarantee
adequate punching shear resistance of slab-column connections when the structure is subjected to both gravity and
seismic loads. A wide variety of literature is available on
this topic, with code provisions accompanied by several
constructions in seismic zones.1
A grid model was previously developed and validated for
the nonlinear behavior of flat slab-column connections.2 The
grid is composed by linear beam finite elements; the inelastic
response of the structure is concentrated in nonlinear point
hinges. A model2 has been proposed for static pushover
analyses that permit evaluation of flexural, torsional, and
shear internal forces and moments in the whole plate structure, with particular attention to the state of slab-column
joints. The description of their nonlinear behavior allows
evaluation of the whole slab structural response up to failure.
In particular, it is possible to assess the safety of connections with respect to punching and structural deformability,
in terms of interstory drift ratio under lateral loads.
The aim of this paper is to show the efficiency of the model2
for flat slab floors under gravity and horizontal loads. The
grid model2 was used to simulate the experimental tests on
a scaled model of a flat-plate structure.3 A rather demanding
case study, including structural irregularity and the effects of
biaxial loading, was chosen.

ACI Structural Journal/March-April 2014

The test slab consists of a rectangular floor with 16slabcolumn joints under gravity and biaxial cyclic quasi-static
horizontal loading increased up to failure. Due to the
different column cross sections and slab reinforcement, the
structure is symmetric in one direction and nonsymmetric in
the orthogonal direction.
The experimental response showed significant effects of
the biaxial loading.3 The maximum connection moments
were reached at 4% drift in the north-south (N-S) direction,
whereas in the east-west (E-W) direction, for most connections, the maximum moments were recorded at 2% drift and
punching failures occurred for drifts at approximately 3%.
Following this, the test was stopped. Biaxial loading effects
have also been documented in tests on connections,4 with
a reduction of drift capacity in both directions with respect
to the case of uniaxial loading. This effect has also been
measured in connections with rectangular cross sections of
the column.5,6
The first section of the paper briefly describes the experimental tests.3 The second section presents the setup of the
grid model; in particular, the definition of the grid geometry, nonlinear hinges, and loads applied. In the third section,
results of nonlinear analyses under gravity and lateral loads
are shown and compared with experimental results.3 The
conclusions assess the adequacy of the modeling of the
global response and of the individual connections, with
attention given to prediction of the ultimate load and drift
capacity, and the failure modes.
RESEARCH SIGNIFICANCE
A grid model developed for slab-column connections is
validated for the pushover analysis of a flat slab floor tested
experimentally under gravity and cyclic lateral loading. The
performance for a nonsymmetrical geometry is studied,
with four different types of column cross sections arranged
nonsymmetrically with respect to one principal direction of
the plan; the reinforcement layout varies accordingly. The
approximation of the slab behavior with biaxial load cycles
of increasing amplitude up to failure is investigated, with a
strong damage accumulation effect. The results corroborate
the potential of the model for design and analysis of existing
flat slab structures.

ACI Structural Journal, V. 111, No. 2, March-April 2014.


MS No. S-2012-103.R2, doi:10.14359.516686526, was received April, 8 2013, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

343

Fig. 1Geometry of test slab, plan view, in. (cm); positions of slab column connections (letters A through D and
numbers 1 through 4).3
EXPERIMENTAL RESULTS
The tests regard the scaled model of a flat-plate structure
subjected to gravity and lateral loads designed following
ACI 318-83.3 The prototype slab represents a portion of a
typical flat-plate floor of an intermediate story of a multistory office building. The slab has three bays in each direction, and a 203 mm (8 in.) thickness. The bay widths are
6.86and 4.57 m (22.5 and 15 ft) for long and short directions, respectively. The story height is 3.05 m (10 ft).
The scaled model (Fig. 1) used for the experimental
study has dimensions equal to 40% of those of the prototype. The length of each bay is 2.74 and 1.83 m (108 and
72 in.) for the long and short directions, respectively. The
slab is 81 mm (3.2 in.) thick. Columns extend above and
below the slab. The column stubs above the slab were 0.3m
(12 in.) long, and their purpose was to anchor the column
longitudinal reinforcement and to provide continuity of the
columns themselves through the floor; the inferior columns
stubs were 1.46 m (4.8 ft) long with pinned connections at
the extremities.
Four different column cross-sectional shapes were used
to collect data related to the effects of column rectangularity on the structural response: rectangular columns with
an aspect ratio of 2:1 in the east half of the floor and square
columns in the west half. With this layout, the structure is
symmetric about the floor centerline along the long direction
and nonsymmetric in the orthogonal direction.
The concrete had a mean cylinder compressive strength
of 21.8 MPa (3160 psi). The self-weight of the slab was
1.9 kN/m2 (40 lb/ft2), corresponding to 23.6 kN/m3
(150 lb/ft3). Slab longitudinal reinforcement (Appendix A*)
was made of No. 2 deformed bars with cross-sectional area
Al = 32 mm2 (0.05 in.2) ( = 6.4 mm [0.25 in.]), with fy =
462.6MPa (67.1 ksi). The layout of the reinforcement reflects
the changes in the cross sections of the columns (AppendixA).
No transverse reinforcement was used for the slab.
*
Appendixes are available at www.concrete.org/publications in PDF format,
appended to the online version of the published paper. It is also available in hard copy
from ACI headquarters for a fee equal to the cost of reproduction plus handling at the
time of the request.

344

Fig. 2Experimental crack pattern of top and edge of slab


following test at 4% drift in E-W direction3 and representation of the numerical deformations and position of the
nonlinear hinges activated for W loading direction.
Testing of the model slab included gravity load tests,
construction load tests, and lateral load tests up to failure
(Appendix B). A uniform gravity loading of 3.73 kN/m2
(78 lb/ft2) was provided by weights placed on the slab,
with a total vertical uniform load with the self-weight of
the structure equal to 5.65 kN/m2 (118 lb/ft2). Construction
loads were then applied in sequence on each panel of the
model slab; their approximated value is equal to 2.63 kN/m2
(55 lb/ft2), then removed for all subsequent tests. The total
vertical uniform load with the self-weight of 5.65 kN/m2
(118 lb/ft2) was used for the lateral load testing described in
the following, resulting in a gravity shear ratio Vg/Vc close
to 0.2 (Table B2). The lateral loads were applied to the test
slab in two principal directions (N-S and E-W) using four
reversible hydraulic actuators supported on a reaction frame.
Increasing lateral drifts from 1/400 to 1/25 were imposed,
with lateral drift first imposed in the S direction, then
reversed to the N direction, and finally returned to zero drift.
The cycle was repeated once. The same sequence was then
immediately applied in E-W direction. These alternating
loadings in the two orthogonal directions were then repeated,
doubling the lateral drift (Fig. B1). No torsional effects
leading to rotation of the floor were observed in the tests.
The experimental response to the biaxial horizontal loading
showed relevant differences in the two orthogonal directions
as a consequence of the loading history.3 The connections
in the N-S direction of loading reached their strength at 4%
drift. Loading in the E-W direction then took place, and
the expected connection strengths were not developed. The
ACI Structural Journal/March-April 2014

recorded maximum moments for most connections were


those of the previous load cycle at 2% drift, followed by
severe strength deterioration and punching failures around
3% drift. The crack pattern of the entire slab at the end of
these tests is shown in Fig. 2.
NUMERICAL MODEL
In the following, the description of the setup of the grid
model for the particular study case previously explained and
the comparison between test results and numerical outputs
will be presented. The grid model2 was developed using a
commercial finite element software7 to perform nonlinear
static analyses under displacement control. The slab is represented by a grid of linear beam finite elements, fixed at joints,
arranged in two orthogonal directions; the columns are
modeled with two beam elements, one above and one below
the slab level, fixed to the plate. The beam finite elements
have been defined as beam-column elements, thus including
the effects of flexural, torsional, and shear deformations.
Each joint has six degrees of freedom. The in-plane shear
due to the lateral loads is modeled by the flexural stiffness of
the elements around the axis perpendicular to the slab plane,
together with the in-plane shear and axial stiffness.
Grid geometryTo design the grid geometry, guidelines
given by CIRIA Report 1108 have been considered as a
reference point, resulting in the choices for the grid detailed
in the following. The grid spacing must be sufficiently close
near to the columns to obtain a good approximation of the
load effects in the slab, since concentration of internal forces
and moments exists in these zones; the elements can be more
widely spaced elsewhere. For the size of the grid spacing,
generally a width equal to c + d is adequate,2 where c is the
column dimension, and d is the effective height of the slab.
It should be noted that c + d is the width of the shear critical section of connection, as considered in the grid model2
according to the definition of ACI 3189; further details are
given in the following section.
Observing the layout of the test slab (Fig. 1), it is evident
that the structure is asymmetrical in regards to the column
sections; four different column sections were used by
authors of the experimental test.3 For beam elements placed
on the axes connecting the connections, a width equal
to the major c + d along the axes themselves was chosen
(Appendix B). The nearby elements have the same width,
with some adjustments to have the sum of the widths of all
cross sections equal to the slab dimensions in the plan. The
depth of the beam elements modeling the slab is equal to the
slab thickness.
The portions of grid corresponding to the column crosssection dimensions are modeled by four beam elements
positioned along the four column semi-axes. A high stiffness is specified to these elements, to model the support given
to the slab by the column. In the analyses, the elements had
the column width and a depth equal to three times the slab
thickness.
As already specified, columns are modeled with two beam
elements: one below the slab grid with a length of 1.260 m
(4.13 ft), and one above with a length of 0.345 m (1.13 ft).
The columns stubs below the slab are pinned at their bases.
ACI Structural Journal/March-April 2014

Fig. 3Grid model geometry: (a) overall view of grid and


columns; and (b) detail of model at connection of slab
andcolumn.
Columns sections are taken with the same dimensions of
those effectively used in the test slab.
LoadsThe structure is subjected to gravity and lateral
loads. From the total uniformly distributed gravity load per
unit of surface (comprising self-weight), the total load per
unit length acting in the two principal directions is calculated
for each element of the grid. With regard to the imposed
displacements, these have been applied in the same positions
of the experimental test compatibly to the refinement of the
grid. A displacement control analysis is carried out.
Nonlinear hingesEach grid element is composed of
an elastic part and nonlinear hinges. The cracked stiffness,
depending on the quantity of reinforcement, is used for the
elastic part of all elements to consider the effect of shrinkage
and construction stresses,3 causing cracking in the slab
previous to the action of lateral loads. The hinges are points
in which the nonlinear properties of the elements themselves are lumped, and are characterized by relations that
link bending moment, shear, and torsion with the inelastic
curvature, shear distortion, and twist angle, respectively.
345

Fig. 4Shear nonlinear hinge: relation of shear force and


inelastic strain.
The elements (Fig. 3(b)) are made of a linear elastic beam
element with a nonlinear hinge at each end for flexure. Each
element has one hinge for shear and one for torsion, both
at the center of the element. The elements are connected
together at their end nodes in the intersection points of the
grid. The simplified phenomenological approach adopted
to define the hinge properties is summarized herein; more
details can be found in the references.2
Point hinges were assigned to frame elements lying on
the center lines connecting joints and on the immediately
adjacent axes. This choice was made to make the model
more computationally light. The level of internal forces was
checked a posteriori, indicating that the cracking moments
were not reached in the internal part of the slab panels.
Three types of hinges are defined according to the grid
model2: flexural, shear, and torsional. Their properties
depend on the geometry of each elements cross section, on
concrete properties, and on longitudinal bottom and top reinforcement of the slab.
BendingFlexural nonlinear hinge properties are
computed analytically for each beam element using a
sectional model10 with nonlinear constitutive relations
based on perfect bond and plane section assumptions; the
moment-curvature relations thus obtained are then approximated with trilinear relationships. Normalized diagrams
with respect to ultimate capacity Mu and associated ultimate
curvature u are provided as input to the model, together
with the values of Mu and u. The hinge length is defined
equal to d, based on test results.2
ShearThe shear force-distortion relation is input as
a normalized diagram with respect to the ultimate shear
capacity Vu and to the associated ultimate distortion u
(Fig.4). The shear capacity of each side of the critical
perimeter was computed as Vu = vn(c + d)d according to
ACI 318-05,9 with the strength vn = 1/3(fc)1/2 MPa [4(fc)1/2
psi] for elements without shear reinforcement. Experiments
show that this can vary with the slab thickness, reinforcement ratio, and the effects of cyclic loading.11 For this
case, the slab had a low thickness that would increase vn;
the cyclic loading could cause strength deterioration. Hence
vn = 1/3(fc)1/2 [4(fc)1/2psi] was used for the analyses. The
values of ultimate shear distortion u for slabs without shear
reinforcement were obtained by a linear interpolation of
346

the optimal values for modeling the experimental punching


in tests carried out on slab-column connections,2 with u =
0.0092[pfy/(1/3fc0.5)] 0.011, where p is the longitudinal
reinforcement ratio and fc and fy are the strength of materials. The expression in customary units is given by u =
0.00229[pfy/(fc0.5)] 0.011, where fc and fy are expressed
in psi.
A trilinear relation was defined for the shear force-strain
relation (Fig. 4); the behavior is obtained with the sectional
shear model,10 normalized with respect to the maximum
shear force and the correspondent strain. The normalized
diagrams with the associated shear capacity Vu and the corresponding ultimate shear distortion u are used as input to the
model. The first point of the trilinear diagram corresponds
to cracking. The second point (1, V1/Vu) is determined by
the intersection of the portion in the cracked stage of the
curve determined by the sectional model10 with the level of
80% of the strength. The third point corresponds to the shear
strength, followed by the last point at the end of the softening branch determined using the shear sectional model.10
The third point corresponds to the shear strength, followed
by the last point at the end of the softening branch determined using the shear sectional model.10 The hinge length is
taken equal to 2d.2 Shear nonlinear hinges were applied only
to the grid elements framing into the columns; this choice is
based on the fact that shear forces in other elements of the
grid are much lower.
TorsionThe definition of torsion nonlinear hinges in the
grid model2 relates torque and twist; normalized diagrams
with respect to the ultimate torsional capacity Mtu and the
corresponding ultimate twist u are provided. Mtu0 is the
torsional capacity without interaction effects, and Mtu
considers the interaction of the torsional strength with the
bending moment and shear acting in the connection due
to gravity loads. These are determined according to the
model for slab-column connections proposed by Park and
Choi,12 where Mtu0 is the product of the polar moment of
inertia at the sides [J = 2(c + d)d3/12 + 2d(c + d)3/12] by
a strength vus = 5.0vn. To consider the interaction with the
bending moment and shear acting in the connection due to
gravity loads, a reduced torsional capacity Mtu is calculated
by reducing the strength to vus,red = vn[5.0 2.5(n/fc)]
Vg/(c + d)/d, where n is the maximum flexural stress in the
concrete at the sides of the connection and Vg/(c + d)/d is the
gravity shear stress. With regard to the ultimate twist angle
u0, values measured experimentally for specimens without
transverse steel13 are used2; a value of 0.02 is adequate.14
When the interaction with shear and bending moment determines a capacity reduction from Mtu0 to Mtu, the twist angle
u0 is reduced to u, proportionally to Mtu/Mtu0.
For the normalized torque and twist relation, a trilinear
diagram is used; the behavior is obtained with the torsional
model of Collins and Mitchell,15 normalized with respect to
the maximum torque and the correspondent twist. The piecewise linear diagram is obtained by connecting the points at
cracking, yielding, and ultimate (Fig. 5).
Torsional nonlinear hinges computed by the aforementioned method are used in elements framing into the
columns. For grid elements that are not at the interface with
ACI Structural Journal/March-April 2014

Fig. 5Torsion nonlinear hinge: relation of torque and


inelastic twist angle.

Fig. 7Comparison between numerical pushover curve


and experimental lateral load-drift envelope curve for E-W
direction (experimental cycles in small figure)

Fig. 6Comparison between numerical pushover curve and


experimental lateral load-drift envelope curve for N-S direction (experimental cycles in small figure).
the columns, the model prescribes an elastic-perfectly plastic
behavior, with a torsional strength16 equal to 0.58(fc)0.5 MPa
[6.99(fc)0.5 psi].
Comparison between numerical and experimental
results
The procedure followed for the analysis of the model slab
is the following: a load-controlled nonlinear analysis is first
performed under gravity loads specified previously; then,
starting with these results, nonlinear static analyses for horizontal loads with displacement control up to failure in each
principal direction are carried out. It should be noted that
both positive (S and W) and negative (N and E) verses of
loading are considered in the analyses, with a total of four
analyses for horizontal load, each starting after the gravity
load analysis.
The experimental study3 provides data for the total lateral
load and drift applied to the structure (Fig. 6 and 7). Envelopes of peak values of the lateral loads and the corresponding
deflections in each cycle are well suited to be compared
with pushover curves of the model for the two orthogonal

ACI Structural Journal/March-April 2014

directions of loading. The experimental responses in the


two orthogonal directions are shown in Fig. 6 and 7. As a
consequence of the orientation of a part of the columns with
the strong axis in the E-W direction, the sum of the column
stiffness in the E-W direction is the double that in the N-S
direction. This has a limited impact on the global stiffness
of the structure resulting from the assemblage of columns
and slab. For the first cycles at low drift, the stiffness in the
E-W direction was only 20 to 30% higher than in the N-S
direction (Table B1). The curves up to 1% drift are rather
close in the two directions, whereas they differ both in terms
of resistance and ultimate drift. As previously mentioned,
this difference can be explained by the degradation of slabcolumn connections due to the biaxial cyclic loading.3,4
In Fig. 6 and 7, the model predicts an ultimate drift capacity
of 4% in both directions. The prediction in the N-S direction
is accurate; a slight difference shows in the N direction with
negative forces. It should be considered that experimentally, the 4% drift was reached after that the same drift was
attained in the S direction, with some damage accumulation.
In the E-W directions, differences between tests and model
response show starting from 2% drift. The ultimate drift and
load capacity are overestimated, with the model reaching
ultimate drift at 4%; the tests showed strength deterioration beyond 2% drift, followed by several punching failures
at drifts around 3% on average. The high level of damage
inflicted by the two cycles at 4% drift in the N-S direction
should be considered. This is not taken into account by
the model. A similar analytical overestimation of the load
capacity was obtained by the authors of the tests3 using the
ACI 318-83 model for the strength of the connections; the
experimental interior connection strengths were, on average,
87% of the analytical values in the E-W direction, whereas
the model was slightly conservative (experimental results
are 110% on average of the analytical) in the N-S direction.
The model approximates the response, taking into account
the global behavior of the structure with different column
347

Fig. 8Comparison between numerical and experimental moment-rotation curve of Connection C3 for: (a) N-S; and (b)E-W
directions. Positive drifts S and W. Experimental envelope up to connection strength. Experimental E-W 4% drift cycles,
including cycles beyond failure.
cross-section shapes, with the nonsymmetry highlighted in
the description of the structure.
Figure 2 shows the experimental crack pattern at failure
with the slab numerical deformed shape and the nonlinear
hinges activated in the elements for the W loading direction.
The most damaged zonesthat is, where hinges are beyond
yieldingare those around connections corresponding to
the experimental damage. A more detailed analysis of the
prediction of damage is carried out as follows.
Punching failure of a connection is reached after the
formation of a shear hinge, when the torsion capacity at
the sides of the connection is exceeded. In the analyses
presented herein, the structure reached failure with punching
in part of the connections; several others were very close
to this stage. This reflects the experimental behavior where
failure occurred with punching occurring in several connections within a short interval of time.3
Unbalanced moment versus rotation envelopes
ofconnections
The behaviors of each connection obtained numerically and experimentally3 were compared. The test report3
provides unbalanced moment-rotation envelopes for each
joint up to the drift corresponding to connection strength
for both the loading directions. Experimental cycles for the
E-W 4% drift test are also presented, including the cycles
beyond failure. For the model, the moments were obtained
by the product of the numerical base shear and the column
height3; rotations are an output of the analysis in the nodes
connecting the slab and columns. The behavior of connections depends on their position (internal, edge, and corner)
and the direction of loading (N-S and E-W).
The figures in the following show examples of the comparison of numerical and experimental moment-rotation curves
of connections, representative of three different typology
of joints, discussed previously: the interior Connection C3
(Fig. 8 and 9), the lateral Connection D3 loaded parallel to
the free edge (Fig. 10 and 11), and the corner Connection A4
(Fig. 12 and 13).

348

With regard to the interior Connection C3 and considering


the N-S loading direction (Fig. 8(a)), the numerical curve
fits very well the experimental one for positive and negative
verse of loading. For the E-W loading direction (Fig.8(b)),
a fair approximation is obtained, though some difference
between the numerical and the experimental curve shows
from approximately 1.5 to 2% drift, overestimating the
strength; this is due to the previously mentioned deterioration of the test slab subjected to the biaxial cyclic loading.
The model reproduces the connection failure modes that
were experimentally observed. Connection C3 failed for
punching shear during the final test in the E-W direction,
for both positive and negative directions of loading at a drift
of 3.1%. Numerical results effectively confirm these data,
because punching of the connection is not detected during
analyses for the N-S loading direction (Fig. 9, left), whereas
punching is predicted in the E-W loading direction (Fig. 9,
right) for both positive and negative drifts of 3.9 and 4.0%,
respectively. The difference between experimental and
numerical punching drift is ascribed to the biaxial loading.
For the lateral Connection D3, when loaded parallel to the
free edge (N-S), good correspondence is obtained with the
pushover (Fig. 10(a)) in terms of force and drift. For the E-W
cycles, the experimental failure was punching in both E and
W directions at 3 and 2% drift, respectively. Strength is overestimated in the W loading (Fig. 10(b)) at a maximum drift
very close to the experimental. Punching failure is predicted
in the E (negative) direction (Fig. 11, right), though strength
and drift capacity are overestimated.
With regard to corner connections, for A4 (Fig. 12), the
numerical curve predicts drift capacity accurately in the N-S
and W directions, but overestimates the drift capacity in the
E (negative) direction. The connection strength is overestimated in the N (negative) and E-W directions.
The tests showed that all corner connections, including
A4, survived the E-W 4% drift test, except D1 in the W direction. The numerical model correctly shows that punching is
not reached, whereas the torsion capacity is reached for both
S and W directions, corresponding to the experimental crack
patterns (Fig. 13).
ACI Structural Journal/March-April 2014

Fig. 9State of hinges around column of Connection C3 for N-S (left) and E-W (right) loading at ultimate drift. Experimental
crack pattern on top surface after punching (with line surrounding damaged zone).

Fig. 10Comparison between numerical and experimental moment-rotation envelope of Connection D3 for: (a) N-S; and
(b)E-W direction. Positive drifts S and W. Experimental envelope up to connection strength. Experimental E-W 4% drift cycles,
including cycles beyond failure.
Summing up, the model provides a good approximation
of the experimental behavior for the N-S direction, where
the biaxial loading effects were lower. In the E-W direction,
the strength and ultimate drift capacity were affected by the
ACI Structural Journal/March-April 2014

accumulation of damage with biaxial loading, and the model


provides non-conservative results. The same results were
reached for strength of the connections using ACI 318-83,

349

and showing the experimental ultimate drift capacities


reduction in the E-W direction compared with N-S.3

Fig. 11State of hinges around column of Connection D3


for N-S (left) and E-W (right) loading at ultimate drift.

CONCLUSIONS
A grid model has been set up to reproduce the nonlinear
response of a flat slab structure subjected to gravity and lateral
cyclic loading. Experimental tests carried out on a scaled
model were analyzed. The slab was tested under gravity and
lateral biaxial cyclic loading. Both principal directions were
loaded alternatively, with a sequence of cycles of increasing
amplitude. Nonlinear static analyses under gravity loads
have been performed, followed by pushover analyses under
horizontal loads in the two principal directions of the slab on
the numerical model.
The model is efficient in showing several aspects of the
response of the test slab considered:
1. The experimental global behavior is approximated
differently in the two orthogonal directions of biaxial
loading: the pushover curves in N-S directions are very close
to the experimental in terms of path, maximum load, and
maximum drift; for the E-W direction, a numerical overestimation of lateral load and drift is detected that is due to
the experimental degradation of the test slab due to biaxial
cyclic loading.
2. The model captures the ultimate drift capacity of the
test slab in both N and S directions, the first loaded up to 4%
drift. The drift capacity in the E and W direction is overestimated by the model, predicting a maximum drift close to
4% in both E and W directions against experimental values
close to 2%.
3. Experimental moment-rotation curves of internal
connection are well approximated by numerical curves for
the N-S loading direction; for the E-W direction, an overestimation of load and drift capacity is detected. A similar result
is obtained for lateral connections loaded parallel to the free
edge. The difference is attributed to the slab degradation
observed experimentally, due to the damage accumulation
with biaxial loading.
4. For lateral connections loaded perpendicular to the
free edge and corner connections, the numerical analysis is
less accurate for both the N-S and E-W directions. In the

Fig. 12Comparison between the numerical and experimental moment-rotation envelope of joints A4 for: (a) N-S, and
(b)E-W direction. Positive drifts S and W. Experimental envelope up to connection strength. Experimental E-W 4% drift
cycles, including cycles beyond failure.
350

ACI Structural Journal/March-April 2014

Guglielmo Corti is a Civil Engineer. He received his MSc in civil engineering from Politecnico di Milano in 2010. His research interests include
design of reinforced concrete structures.

c
d
J
Mtu
Mtuo
Mu
Vc
Vg

=
=
=
=
=
=
=
=

Vu =
vc =
vn =
vus =
vus,red =
u
n
u
u
uo

=
=
=
=
=

NOTATION

column side
slab average effective depth
polar moment of inertia at sides of connection
reduced torsion capacity (with interaction effects)
torsion capacity without interaction effects
flexural capacity
punching shear capacity of flat slab-column connection
shear force transferred between slab and column under gravity
loads
shear force at capacity of critical sections
concrete shear strength, ACI 318-059
shear strength, ACI 318-059
eccentric shear strength at the sides of connection
reduced eccentric shear strength, by interaction with flexure and
shear
ultimate curvature
maximum flexural stress in concrete, compression
inelastic shear strain corresponding to shear capacity
reduced twist angle at maximum torque (with interaction effects)
twist angle at maximum torque

REFERENCES

Fig. 13State of hinges around column of Connection A4


for N-S (left) and E-W (right) loading at ultimate drift.
Experimental crack patterns (side and top).
latter case, the accumulation of damage with the biaxial
loading affects the results. Improvements are needed for the
parameters in the model for the strength of these types of
connections.
5. The model reproduces the experimental failure modes
of connections. The tests considered herein showed the
effects of biaxial loading as a reduction of ultimate drift
and capacity at punching of the slab-column connections,
with respect to those for a one-directional loading. These
punching failures are predicted by the model in correspondence of ultimate moment and drift values higher than the
experimental. The developments of the research should
improve the model to consider the biaxial loading effect.
These results show the performance of the model in a
rather demanding case study, including some structural
irregularity and the effects of cyclic loading in two orthogonal directions. The results for biaxial loading indicate the
need for further research developments.
AUTHOR BIOS

Dario Coronelli is an Assistant Professor in the Department of Civil and


Environmental Engineering at Politecnico di Milano, Milan, Italy, where
he received his PhD in 1998. His research interests include the structural
effects of corrosion and seismic design of reinforced structures.

ACI Structural Journal/March-April 2014

1. Joint ACI-ASCE Committee 421, Seismic Design of Punching Shear


Reinforcement in Flat Plates (ACI 421.2R-07), American Concrete Institute, Farmington Hills, MI, 2007, 24 pp.
2. Coronelli, D., A Grid Model for Flat Slab Structures, ACI Structural
Journal, V. 107, No. 6, Nov.-Dec. 2010, pp. 645-665.
3. Hwang, S. J., and Moehle, J. P., An Experimental Study of FlatPlate Structures under Vertical and Lateral Loads, Technical Report UCB/
EERC-93/03, Earthquake Engineering Research Center, University of California at Berkeley, Berkeley, CA, Feb. 1993, 278 pp.
4. Pan, A. D., and Moehle, J. P., Lateral Displacement Ductility of
Reinforced Concrete Flat Plates, ACI Structural Journal, V. 86, No. 3,
May-June 1989, pp. 250-258.
5. Tan, Y., and Teng, S., Interior Slab-Rectangular Column Connections under Biaxial Lateral Loadings, SP-232, M. A. Polak, ed., American
Concrete Institute, Farmington Hills, MI, 2005, pp. 147-174.
6. Anggadjaja, E., and Teng, S., Edge-Column Slab Connections
under Gravity and Lateral loading, ACI Structural Journal, V. 105, No. 5,
Sept.-Oct. 2008, pp. 541-551.
7. CSI, CSI Analysis Reference ManualSAP 2000 Advanced Research
v.10, Computers and Structures Inc., Oct. 2005, 433 pp.
8. Whittle, R. T., Design of Reinforced Concrete Flat Slabs to BS
8110, Report 110, CIRIA, 1994.
9. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-05) and Commentary (318R-05), American Concrete
Institute, Farmington Hills, MI, 2005, 430 pp.
10. Bentz, E. C., Sectional Analysis of RC Members, University of
Toronto, Toronto, ON, Canada, 2000, 198 pp.
11. Dilger, W. H., Flat Slab-Column Connections, Progress in Structural Engineering and Materials, V. 2, 2000, pp. 386-399.
12. Park, H., and Choi, K., Improved Strength Model for Interior Flat
Plate-Column Connections Subjected to Unbalanced Moment, Journal of
Structural Engineering, ASCE, V. 132, No. 5, May 2006, pp. 694-704.
13. Kanoh, Y., and Yoshizaki, S., Strength of Slab-Column Connections Transferring Shear and Moment, ACI Journal, V. 76, No. 3, Mar.
1979, pp. 461-468.
14. Tian, Y.; Jirsa, J. O.; and Bayrak, O., Nonlinear Modeling of SlabColumn Connections under Cyclic Loading, ACI Structural Journal,
V. 106, No. 1, Jan.-Feb. 2009, pp. 30-38.
15. Collins, M. P., and Mitchell, D., Shear and Torsion Design of
Prestresses and Non-Prestressed Concrete Beams, PCI Journal, Sept.-Oct.
1980, pp. 32-100.
16. Park, R., and Paulay, T., Reinforced Concrete Structures, John
Wiley& Sons, Inc., New York, 1975, 769 pp.

351

NOTES:

352

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S31

Effect of Steel Stirrups on Shear Resistance Gain Due


to Externally Bonded Fiber-Reinforced Polymer Strips
andSheets
by Amir Mofidi and Omar Chaallal
This paper presents the results of an experimental investigation of reinforced concrete (RC) T-beams strengthened in shear
with externally bonded (EB) carbon fiber-reinforced polymer
(CFRP) strips and sheets. The main objective of this study was to
gain insight, by varying the test parameters, into the interaction
between the internal transverse steel reinforcement and the externally bonded CFRP used for strengthening of RC beams in shear.
The test parameters of this study were: 1) the CFRP ratio (that
is, the spacing of the CFRP strips); 2) the presence or absence of
transverse steel; 3)the transverse steel ratio (that is, the spacing
of the stirrups); and 4) the use of CFRP strips versus CFRP sheets.
In total, 10 tests were performed on 4520 mm (14 ft, 10 in.) long
T-beams. The study showed that the presence of internal transverse
steel reinforcement resulted in a significant decrease in the gain
due to CFRP for all strengthened specimens. It also showed that
the steel yielded before failure for all test specimens with transverse steel. Finally, the presence of CFRP did not result in a
significant decrease in transverse steel strain. It can be concluded
that the contribution of transverse steel to shear resistance is not
affected by the addition of EB CFRP. These results are in good
agreement with the assumptions made by existing codes and design
guidelines, which are based on the yielding of transverse steel at
ultimate strain for RC beams strengthened in shear with EB CFRP.
Keywords: carbon fiber-reinforced polymer (CFRP); CFRP sheet;
reinforced concrete beam; shear strengthening; strip; transverse
steelreinforcement.

INTRODUCTION
In recent years, shear strengthening of reinforced concrete
(RC) beams with externally bonded (EB) fiber-reinforced
polymer (FRP) material has attracted attention and has been
studied by several researchers (Uji 1992; Chaallal et al.
1998; Triantafillou 1998; Khalifa et al. 1998; Bousselham
and Chaallal 2004; Chaallal et al. 2011; Mofidi and Chaallal
2011a,b). These experimental and analytical studies have
provided valuable insights and results. Several questions,
however, still linger in the area of shear strengthening of
RC beams with FRP composites (Bousselham and Chaallal
2004; Mofidi and Chaallal 2011a).
For instance, a comparison between the experimental shear
resistance due to FRP and the values predicted by the analytical models used in existing codes and guidelines shows that
major aspects of shear strengthening with FRP material are
still not captured by the predictive models used in the codes
and guidelines (Bousselham and Chaallal 2008; Mofidi and
Chaallal 2011a). This is mainly because the calculated shear
contribution of FRP according to the codes and guidelines
does not account for the effect of certain parameters that
have experimentally been found to have a major influence
ACI Structural Journal/March-April 2014

on the contribution of FRP to shear resistance. The effect of


internal transverse steel reinforcement is one of these influential parameters. The adverse effect of transverse steel on
the effectiveness of externally bonded FRP used for shear
strengthening and retrofit of RC beams is well established
(Bousselham and Chaallal 2008; Chaallal et al. 2011; Mofidi
and Chaallal 2011a,b). In contrast, the effect of adding EB
FRP for shear retrofit on the performance of internal steel
stirrups has not been thoroughly documented. Moreover,
in most modern codes and guidelines, the contribution of
steel stirrups to the shear resistance Vs at the ultimate state is
calculated on the premise that the steel stirrups have yielded.
The premature debonding failure observed in RC beams
strengthened in shear with FRP, however, has prompted
legitimate questions and concerns as to whether the assumption that the steel stirrups yield before failure holds true
(Chen et al. 2010).
RESEARCH SIGNIFICANCE
Researchers have not yet reached a commonly accepted
agreement on the effect of transverse steel in RC beams retrofitted with FRP. Therefore, this effect is not yet considered
in the design codes and guidelines, including ACI 440.2R-08
(ACI Committee 440 2008). Consequently, current design
codes and guidelines may overestimate the shear resistance
of RC beams with transverse steel that are strengthened
using EB FRP sheets and strips. Such uncertainties in shear
strengthening of RC beams using EB FRP have provided the
key impetus for conducting the current research study, the
objective of which was to gain insight into the interaction
between internal transverse steel reinforcement and externally bonded FRP strips and sheets used for shear strengthening of RC beams. This insight has been achieved based on
results obtained from an experimental program carried out
on full-size T-beam specimens, described as follows.
EXPERIMENTAL PROGRAM
The experimental program (Table 1) involved 10 tests
performed on full-scale RC T-beams. The control specimens, which were not strengthened with carbon FRP
(CFRP), were labeled NF (for No FRP), whereas the specimens retrofitted with a layer of EB CFRP sheet were labeled
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-104.R2, doi:10.14359.51686527, was received November 12,
2012, and reviewed under Institute publication policies. Copyright 2014, American
Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published ten months from this journals date if the
discussion is received within four months of the papers print publication.

353

Table 1Experimental results


Total shear
resistance,
kN

Resistance
due to
concrete,
kN

Resistance due
to steel, kN

Resistance due
to CFRP, kN

Gain due to
CFRP, %

Deflection at
loading point,
mm

Specimen

FRP type

wf /sf

Load at
rupture,
kN

NR-NF

122.7

81.2

81.2

0.0

0.0

2.6

NR-ST-LF

Strips

87.5/175

203.1

134.5

81.2

0.0

53.3

66

6.2

NR-ST-HF

Strips

87.5/125

227.3

150.6

81.2

0.0

69.3

85

7.2

NR-SH

Sheet

181.2

120.0

81.2

0.0

38.7

48

4.2

HR-NF

350.6

232.2

81.2

151.0

0.0

11.9

HR-ST-LF

Strips

87.5/175

372.5

246.7

81.2

151.0

14.5

15.9

HR-ST-HF

Strips

87.5/125

383.4

253.9

81.2

151.0

21.7

15.7

HR-SH

Sheet

378.3

250.6

81.2

151.0

18.4

15.2

MR-NF

294.0

194.7

96.2

98.5

11.2

MR-SH

Sheet

335.2

222.0

96.2

98.5

27.3

14

11.3

Notes: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.

Fig. 1Test setup configuration: (a) cross section of tested RC beams; and (b) side view of loading configuration. (Note:1mm=
0.0394 in.)
SH (for SHeet), and the specimens strengthened with FRP
strips (strip width = 87.5 mm = 3-7/16 in.) were labeled ST
(for STrips). Specimens strengthened with narrowly spaced
FRP strips (spacing equal to 125 mm [5 in.]) were labeled
HF (for Heavily strengthened with FRP), whereas the specimens strengthened with widely spaced strips (spacing
equal to 175 mm [6 7/8 in.]) were labeled LF (for Lightly
strengthened with FRP). Series NR (Not Reinforced with
transverse steel) consisted of specimens with no internal
transverse steel reinforcement (that is, no stirrups). Series
HR (Heavily Reinforced with transverse steel) and MR
(Moderately Reinforced with transverse steel) contained
specimens with internal transverse steel stirrups spaced
at s = d/2 and s =3d/4, respectively, where d = 350 mm
(13-3/4in.) represents the effective depth of the cross section
of the beam. Therefore, for instance, Specimen NR-ST-HF
featured a beam with no transverse steel retrofitted using
CFRP strips spaced at 125mm (5 in.). The specimen details
are provided in Table1, together with the identification
codes used hereafter.
Description of specimens
The T-beams were 4520 mm (14 ft, 10 in.) long, and their
T-sections had overall dimensions of 508 x 406 mm (20 x
16 in.). The width of the web and the thickness of the flange
were 152 and 102 mm (6 and 4 in.), respectively (Fig. 1(a)
and (b)). It should be noted that the web of the strengthened
beams is chamfered at the outer corners. The longitudinal
steel reinforcement consisted of four 25M bars (diameter
354

25.2 mm [1 in.], area 500 mm2 [0.78 in.2]) laid in two layers
at the bottom, and six 10M bars (diameter 10.3 mm [ 0.4 in.],
area 100 mm2 [0.16 in.2]) laid in one layer at the top. The
bottom bars were anchored at the support with 90-degree
hooks to prevent premature anchorage failure. The internal
steel stirrups (where applicable) were 8 mm (5/16 in.) in
diameter (area 50 mm2 [0.08 in.2]).
To apply the EB FRP sheets and strips to the RC specimens, the following steps were implemented: 1) the area
of the specimens where the CFRP sheets and strips was to
be epoxy-bonded was sand-blasted to remove any surface
cement paste and to round off the beam edges; 2) the specimen corners were chamfered to provide a radius of 12.7mm
(0.5 in.) to avoid stress concentration in the FRP sheets
during the tests; 3) residues were removed using compressed
air; and 4) layers of U-shaped CFRP sheets and strips were
glued to the bottom and lateral faces of the RC beam using a
two-component epoxy resin.
Materials
A commercially available concrete delivered to the
structural laboratory by a local supplier was used in this
project. The average 28-day concrete compressive strength
was 25MPa (3626 psi), which is very close to the average
compressive strength of 27 MPa (3916 psi) obtained
during the tests. It should be noted that the specimens of
the MR series were cast using a different concrete batch,
the compressive strength of which was 35 MPa (5076 psi).

ACI Structural Journal/March-April 2014

location as the transverse steel. The load was applied using


a 2000kN (449kip) capacity MTS hydraulic jack. All tests
were performed under displacement-control conditions at a
rate of 2 mm/min (approximately 3/16 in./min).
ANALYSIS OF RESULTS
All the specimens failed in shear. The control specimens
failed due to diagonal tension failure of the concrete cross
section. The specimens strengthened with CFRP failed by
premature FRP debonding followed by diagonal tension
failure (Fig. 4(a) through (d)). Local CFRP fracture was
observed in few specimens (NR-ST-LF and NR-ST-HF);
this local failure is attributed to stress concentration at the
web corners.
Fig. 2Side view of strengthened specimen using FRP sheet
U-jacket with crack gauges on CFRP strips.
The scatter between the results of compression tests on the
cylinder specimens was insignificant.
The longitudinal steel reinforcement consisted of
25Mbars (modulus of elasticity 187 GPa [27,122,057 psi],
and yield stress 500 MPa [72,519 psi]), and the transverse
steel reinforcement consisted of deformed 8 mm (13/16 in.)
bars (modulus of elasticity 206 GPa [29,877,774 psi] and
yield stress 650 MPa [94,275 psi]).
The composite material was a unidirectional carbon-fiber
fabric epoxy-bonded over the test zone in a U-shape around
the web (Fig. 2). The dry CFRP sheet had an ultimate tensile
strength of 3450 MPa (500,380 psi), an elastic modulus of
230 GPa (33,358,697 psi), and an ultimate strain of 1.5%,
as reported by the manufacturer. The thickness of the CFRP
fabric used was 0.11 mm.
Test setup
All 10 tests were conducted in three-point load flexure
(Fig. 1(b)). This loading configuration was chosen because it
enabled two tests to be performed on each specimen. Specifically, while one end zone was being tested, the other end
zone was overhung and unstressed. The load was applied at
a distance a = 3d from the nearest support, a configuration
which was representative of a slender beam.
Instrumentation
The measuring equipment used in this research study was
carefully designed to meet the objectives of this study. The
vertical displacement was measured at the position under the
applied load and at midspan, using a linear variable differential transformer (LVDT) with a 150 mm (5-7/8 in.) stroke.
Different types of strain gauges were installed on the longitudinal reinforcement, on the steel stirrups, and embedded in
the concrete to measure the strains experienced by the various
materials as the loading increased and to monitor thereby
the yielding of the steel. The strain gauges on the stirrups
were installed along the anticipated plane of the major shear
crack. Displacement sensors, also known as crack gauges,
were used to measure the strains experienced by the CFRP
strips and sheets (Fig. 2 and 3). These gauges were fixed
vertically onto the lateral faces of the specimens at the same
ACI Structural Journal/March-April 2014

Deflection response
Figure 5 compares the deflection response for RC beams
without transverse steel reinforcement. It reveals that the
NR-SH and NR-ST-HF specimens exhibited slightly greater
overall stiffness than the other beams. Specimen NR-ST-HF
exhibited the highest deflection at the loading point and a
higher maximum load at failure than the other specimens
(Fig. 5). The beams strengthened with FRP strips exhibited
more deformability than the beams strengthened with FRP
sheets. This occurred mainly because in RC beams strengthened with FRP strips, local FRP debonding did not result
in a complete debonding failure. Each local strip-debonding
event resulted in a drop in the load-carrying capacity of the
beam (Fig. 5), but the load continued to increase as the cracks
propagated in the RC beams web, engaging the unloaded
CFRP strips in their path. In specimens strengthened with
FRP strips, and unlike beams strengthened with FRP sheets,
local debonding of FRP cannot propagate from one FRP
strip to the next. Therefore, using FRP strips results in a
more progressive type of failure, and a sudden and brittle
failure is prevented.
Figure 6 shows the load versus maximum deflection
curves for RC beams with transverse steel reinforcement.
It reveals that each of the specimens in Series HR and MR
exhibited an overall stiffness relatively similar to that of the
other beams. The maximum load at failure and the maximum
deflection attained at the loading point for each specimen
are provided in Table 1. Specimen HR-ST-HF reached the
highest maximum load at failure. Meanwhile, the HR-ST-LF
and HR-SH specimens exhibited a slightly higher deflection at the loading point than the other strengthened and
unstrengthened specimens (Table 1). It should be mentioned
that in Table 1, the shear contributions of concrete and steel
were calculated based on the measured experimental results
for the control beams.
Strain analysis
This part of the study investigated the behavior of CFRP
and transverse steel during loading of the specimens. As
mentioned previously, extensive instrumentation for strain
monitoring was carefully planned and implemented to
provide the data needed to gain a better understanding of
the effect of transverse steel on the contribution of FRP to

355

Fig. 3Location of strain gauges on transverse steel reinforcement and FRP sheets/strips. Positions of strain gauges on FRP
sheets/strips are identical in specimens strengthened with similar FRP strengthening configurations.
the shear resistance of RC beams retrofitted in shear with
EB FRP.
CFRP strainThe distribution of the maximum strains
attained in the CFRP is shown in Fig. 7 for all strengthened
test specimens. It should be noted that these strain values are
the maximum measured values, but not necessarily the absolute maximum values, experienced by the CFRP U-jackets.
The two values may differ in cases where the strain gauges
did not intercept the main cracks. From Fig. 7, the following
observations can be made:
1. All curves presented in these figures show that the
CFRP did not contribute to load-carrying capacity in the
initial stage of loading.
2. For specimens of Series NR, the measured strains were
greater for specimens strengthened with FRP strips than for
similar beams strengthened with FRP sheets.
3. For the beams strengthened with a layer of FRP sheet
(NR-SH, MR-SH, and HR-SH), the maximum strain in the
CFRP increased as the amount of transverse reinforcement
was increased. In fact, for Specimen HR-SH, the maximum
strain attained by the FRP sheet was approximately 48%
of the ultimate strain value, whereas the corresponding
356

maximum strain values for Specimens NR-SH and MR-SH


were 17 and 27%, respectively, of the ultimate strain.
4. For specimens strengthened with FRP strips in Series
NR (no steel stirrups), the maximum measured FRP strain
values were approximately equal. This is also true for the
specimens strengthened with FRP strips in the HR series.
5. For all specimens strengthened with FRP strips in both
Series NR and HR, the maximum FRP strain was greater than
5000 . It may be of interest to note that ACI 440.2R-08
(ACI Committee 440 2008) limits the maximum design FRP
strain value to 4000 . On the basis of the results achieved
in this study, this limit appears to be conservative for RC
beams strengthened with FRP strips.
Transverse steel reinforcement strainFigure 8 shows the
measured strain in the transverse steel reinforcement for the
test specimens with internal steel reinforcement. The vertical
line identifies the strains corresponding to the yielding of the
transverse steel, as obtained by tests (y = 3250 ). From
Fig. 8, the following observations can bemade:
1. Like the CFRP, the steel stirrups did not contribute to
load-carrying capacity in the initial stage of loading. The

ACI Structural Journal/March-April 2014

Fig. 4Failure mode and multiline cracking pattern of strengthened specimens: (a) NR-ST-HF; (b) NR-SH; (c) HR-ST-LF;
and (d) HR-ST-HF after failure.

Fig. 5Load versus deflection at load point: Series NR.


(Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.)

Fig. 6Load versus deflection at load point: Series HR and


MR. (Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.)

transverse steel contribution to shear resistance started after


the formation of diagonal cracking initiated.
2. All monitored stirrups were significantly strained. This
is also reflected by the cracking pattern observed in the
beams with transverse steel (Fig. 4(a) through (d)).
3. Yielding of transverse steel was observed in all cases.
This observation is in good agreement with existing code
specifications and guidelines, which assume that the trans-

verse steel yields at ultimate strain for RC beams strengthened in shear with EB FRP.
Figure 8 shows that addition of EB FRP did not result in a
decrease of transverse steel strain. For all the specimens with
transverse steel, the steel yielded well before the RC beam
reached ultimate failure. Therefore, it can be concluded that
at the ultimate state the contribution of internal steel stirrups to shear resistance was not affected by the addition of

ACI Structural Journal/March-April 2014

357

Fig. 7Load versus strain in FRP for all strengthened specimens. (Note: 1 kN = 0.225 kip.)

Fig. 8Load versus strain in transverse steel: Series HR


and MR. (Note: 1 kN = 0.225 kip.)

externally bonded FRP. It follows that the shear contribution of internal steel reinforcement Vs should be calculated
using the same formula for both FRP-strengthened and
unstrengthened RC beams, which confirms the assumptions
of the design guidelines (ACI 440.2R-08; CSA S806-02;
Oehlers et al. 2008).
These results are not in agreement with those based on
finite-element simulations reported by Chen et al. (2010,
2012); these researchers found that for RC beams strengthened in shear with FRP, the internal steel stirrups did not
reach the yield point. Based on their finite-element model,
they concluded that the yield strength of the internal steel
stirrups in such strengthened RC beams cannot be fully used.
The models by Chen et al. (2010, 2012) were originally
generated based on a single crack failure pattern assumption. Single crack failure pattern was adopted by most design
models to simplify the calculation and design of strengthening FRP sheets and strips. Experimental observations,
however, clearly show that for RC beams strengthened with
EB FRP, the cracking pattern on the FRP-concrete interface
is rather distributed (Mofidi and Chaallal 2011a). Eventually,
the distributed cracks at the concrete cover merge together at
the concrete core to form one major shear crack at ultimate.
Therefore, it is believed that assuming a single crack pattern
is overly simplistic when considering such a precise finite
element modeling tool. Considering the fact that the crack
width plays a governing role in the Chen et al. (2010, 2012)
models, the discrepancies between the results produced by
Chen et al. (2010, 2012) models and the experimental results
are to be expected.

The experimental contribution of transverse steel Vs is


calculated as the sum of the contributions corresponding
to the stirrups crossing the plane of rupture using the
followingequation

Shear resistance under increasing load


In accordance with most codes and standard guidelines,
the nominal shear resistance Vn can be expressed as follows
Vn = Vc + Vs + Vf

358

(1)

Vs = As Es e s ,i (2)

where As is the section area of one stirrup; Es is the elastic


modulus of the transverse steel; and s,I (y) is the measured
strain in stirrup i in the failure zone, where y is the yield
strain of the stirrups.
The experimental contribution of FRP Vf can be calculated
as follows

V f = 2 E f t f (wi e f ,i ) (3)

where Ef is the elastic modulus of the CFRP; tf is the thickness of the CFRP; f,i is the measured strain in the CFRP
corresponding to instrumented section i in the failure zone;
and wi is the tributary width of the strengthening FRP strips
intercepted by the major shear crack, where the CFRP strain
f,i is assumed constant. The CFRP strengthening width
represents the portion of the CFRP that effectively contributes to shear resistance.
Figure 9 shows the experimental evolution under
increasing load of the contributions to the shear resistance
of the two components (Vs and Vf) for Specimens HR-NF,
HR-ST-LF, HR-ST-HF, and HR-SH. The specimens shown
in this figure had the same degree of transverse steel reinforcement (highly reinforced), but were strengthened using
different amounts of externally bonded FRP strips and sheet.
The behavior of the transverse steel under increasing load
exhibited a similar pattern for both the unstrengthened beam
(HR-NF) and the beams retrofitted with different amounts
of EB FRP strips and sheet (HR-ST-LF, HR-ST-HF, and
HR-SH). This result indicates that strengthening of RC
beams with EB FRP does not alter the behavior of internal
ACI Structural Journal/March-April 2014

Table 2Coefficient of determination (R2) between


values of Vf as calculated by each of the models
and experimental values of Vf

Specimen

Vfexp

Vf cal
by ACI
440.2R
(2008)

NR-ST-LF

53.3

20.5

32.5

43.7

NR-ST-HF

69.3

28.6

39.8

37.9

NR-SH

38.7

40.9

45.9

35.5

HR-ST-LF

14.5

20.5

32.5

7.3

HR-ST-HF

21.7

28.6

39.8

8.3

HR-SH

18.4

40.9

45.9

10.2

27.3

44.1

50.7

14.4

0.04

0.03

0.81

MR-SH

Fig. 9Transverse steel and FRP contributions under


increasing load for Specimens HR-NF, HR-ST-LF,
HR-ST-HF, and HR-SH. (Note: 1 mm = 0.0394 in.; 1 kN=
0.225 kip.)

Fig. 10Transverse steel and FRP contributions under


increasing load for Specimens HR-NF, MR-SH, and HR-SH.
(Note: 1 kN = 0.225 kip.)
transverse steel. It also reveals that addition of EB FRP
does not attenuate the shear contribution of the transverse
steelreinforcement.
Figure 10 shows the experimental progression of the
shear contributions of the two components (Vs and Vf) under
increasing load for Specimens HR-NF, MR-SH, and HR-SH.
The strengthened specimens illustrated in this figure were
both retrofitted with one layer of CFRP sheet, but reinforced
with different amounts of transverse steel reinforcement.
The behavior of the FRP under increasing load followed
different patterns for the strengthened beams depending
on the amount of internal transverse steel reinforcement
(MR-SH and HR-SH). This clearly shows that the behavior
of EB FRP depends on the amount of transverse steel used in

ACI Structural Journal/March-April 2014

Vf cal
by HB 305
(2008)

Vf cal
by Mofidi
and
Chaallal
(2011a)

Notes: Vfexp is experimental shear resistance due to FRP; Vfcal is calculated shear
resistance due to FRP (not factored); 1 kN = 0.225 kip.

RC beams. This result confirms that increasing the amount


of transverse steel leads to a reduction in the contribution of
the FRP during loading and at the ultimate state.
COMPARISON OF TEST RESULTS WITH
SHEARDESIGN EQUATIONS
The shear resistance due to CFRP as obtained by tests
Vfexp is compared in Table 2 to the nominal shear resistance
Vfcal predicted by ACI 440.2R (2008), HB 305 (Oehlers et al.
2008), and Mofidi and Chaallal (2011a).
Mofidi and Chaallal (2011a) proposed a model for calculating the contribution of FRP to shear resistance, taking into
consideration the attenuating effect of transverse steel as
well as of the cracking pattern on the EB FRP contribution in
shear. Based on their study, it was determined that the presence of transverse steel favors the formation of a multi-line
shear-cracking pattern in the RC beam, which decreases the
anchorage length of the FRP fibers and hence the available
effective width of FRP wfe and bonding area between the
FRP and the concrete. In the calculation of wfe, it is assumed
that the cracking pattern of the RC beam becomes more
propagated with the increase in the amount of internal steel
and external FRP shear reinforcement as measured by their
respective rigidities. On the other hand, the cracking pattern
influences the anchorage length of the fibers. As the cracking
pattern becomes more propagated, fewer fibers will provide
the minimum effective anchorage length. Therefore, the
effective width, that is the width of the fibers long enough
to attain the effective anchorage length, decreases. Using a
computational analysis based on the available test results
in the literature (Mofidi and Chaallal 2011a), the effective
width is defined as a function of the sum of the rigidities
of transverse steel reinforcement and that of transverse FRP
sheets (Eq. (9) and (10)).

w fe =

0.6
f E f + s E s

df

for U-jacket (4)

359

w fe =

0.43
d f for side bonded
f E f + s E s

(5)

With wfe defined, the cracking modification factor can then


be calculated as kc = wfe/dfthat is
kc =

kc =

w fe
0.6
=
df
f E f + s E s

w fe
0.43
=
df
f E f + s E s

for U-jackets

(6)

for side bonded

(7)

The effect of kL for beams with an anchorage length less


than the effective length and that of kf counting for the wf /sf
ratio of the FRP strips are considered in the equation for
effective strain, as follows

k k k L
fc
e fe = c L f eff e = 0.31kc kL k f
e fu (8)
tf Ef
tf Ef

It should be noted that kL and kf can be calculated using


Eq. (9) and (11), as follows

if l 1
1
kL =

(
2
l
)
l
if l < 1

l=

Lmax
Le

(9)

where

Lmax

df
for U-jackets

sin b
=

df

for side plates

2 sin b

kf =

wf 1
1+

sf 2

wf 1

1
sf 2

(10)

(11)

In addition, Vf can be calculated as a function of fe using


the following equation that accounts for the effect of the
cracking angle (the cracking angle can be taken equal to
45 degrees for simplicity)
Vf =

2t f w f e fe E f (cot + cot a ) sin a d f


sf
(12)

360

= f E f e fe b d f (cot + cot a ) sin a

Fig. 11FRP contribution: values as calculated by ACI


440.2R-08, HB 305-08, and Mofidi and Chaallal (2011a)
models versus experimental values. (Note: 1 kN = 0.225 kip.)
The Mofidi and Chaallal (2011a) model justifies premature FRP debonding in RC beams with internal transverse
steel reinforcement compared with beams with no transverse
steel and explains the superior gain achieved due to FRP in
beams with few or no steel stirrups compared with beams
with moderate to high amounts of internal transverse steel.
To test the correlation between the experimental results
and the predicted results by the models, the best fit of the
nominal predicted results versus the experimental results
was considered. The following assumptions were made
when calculating the experimental results: 1) the shear resistance due to concrete was assumed constant for beams with
or without transverse steel reinforcement; 2) the shear resistance due to concrete was assumed constant for both retrofitted and unstrengthened specimens; and 3) the contribution
of the transverse steel was assumed constant for both retrofitted and unstrengthened specimens.
For specimens with no transverse steel strengthened using
FRP strips (NR-ST-LF and NR-ST-HF), all three models
underestimated the shear resistance due to FRP. This effect
was more significant when using ACI 440.2R-08 (ACI
Committee 440 2008), where, for example, for NR-ST-HF,
the shear resistance predicted was 28 kN (6.3 kip), compared
with 69.3 kN (15.6 kip) obtained by test. On one hand, for
the specimen with no transverse steel strengthened using an
FRP sheet (NR-SH), the ACI 440.2R-08 (ACI Committee
440 2008) and HB 305-08 (Oehlers et al. 2008) models
slightly overestimated the shear resistance due to FRP.
On the other hand, the Mofidi and Chaallal (2011a) model
slightly underestimated the FRP contribution to shear resistance (Table 2). For all strengthened specimens with transverse steel reinforcement, the ACI 440.2R-08 and HB 305-08
models provided unconservative predictions and therefore
overestimated results (Table 2). In contrast, results produced
by Mofidi and Chaallal (2011a) for specimens strengthened
with transverse steel reinforcement correlate fairly well with
ACI Structural Journal/March-April 2014

the test results. Figure 11 shows that the Mofidi and Chaallal
(2011a) model predicted the experimental shear contribution
of FRP (R2 = 0.81) with a high level of accuracy. The ACI
440.2R-08 and HB 305-08 models produced low coefficients
of determination (0.04 and 0.03, respectively). In general,
current design guidelines models (including ACI 440.2R-08
and HB 305-08) fail to consider the effect of the transverse
steel (Mofidi and Chaallal 2011a,b). Therefore, they may
predict conservative results for beams without transverse
steel reinforcement. In contrast, current design guidelines
models may overestimate the shear contribution of FRP
for the specimens with transverse steel reinforcement and
hence, the shear resistance. Such unconservative results are
exemplified in the results predicted by ACI 440.2R-08 and
HB 305-08 for Specimens HR-ST-LF, HR-ST-HF, HR-SH,
MR-SH in the current study. The presence of the transverse
steel has a significant effect on the shear resistance of RC
beams strengthened with FRP, and therefore, should ultimately be considered in design models.
CONCLUSIONS
This paper presents the results of an experimental investigation involving 10 tests on RC T-beams strengthened
in shear with EB FRP strips and sheets. The effects of
the following parameters were studied: 1) the CFRP ratio
(that is, the spacing of the CFRP strips); 2) the presence or
absence of transverse steel; 3) the transverse steel ratio (that
is, the spacing between the stirrups); and 4) the use of CFRP
strips versus CFRP sheets. The following conclusions can
be drawn:
1. The addition of internal transverse steel reinforcement
resulted in a significant decrease in the gain due to FRP for
all the strengthened specimens.
2. For all test specimens with transverse steel reinforcement, the steel yielded before the specimen failed. The
presence of externally bonded FRP for shear retrofit did
not cause a significant decrease in transverse steel strain.
Overall, the contribution of steel stirrups to shear resistance
was not adversely affected by the addition of FRP.
3. Comparison of the resistance predicted by the
ACI440.2R-08 (ACI Committee 440 2008), HB 305-08
(Oehlers et al. 2008), and Mofidi and Chaallal (2011a)
models with test results showed that the guidelines failed to
capture the influence of transverse steel on the shear contribution of FRP. The model proposed by Mofidi and Chaallal
(2011a) showed a better correlation with experimental
results than the guidelines mentioned.
4. The maximum measured strain values in CFRP strips,
and hence the gain in shear strength due to CFRP strips,
were significantly greater than for CFRP continuous sheets.
In addition, the maximum deflection was slightly greater for
beams retrofitted with CFRP strips than for beams strengthened with continuous CFRP sheets.
5. In all the specimens strengthened with FRP strips, the
maximum attained FRP strain was greater than 5000 . It
follows that the ACI 440.2R-08 limit for maximum FRP

ACI Structural Journal/March-April 2014

strain (that is, 4000 ) seems conservative for RC beams


strengthened in shear with FRP strips.
AUTHOR BIOS

Amir Mofidi is a Postdoctoral Fellow in the Department of Civil Engineering and Applied Mechanics of McGill University, Montreal, QC,
Canada. He received his PhD from University of Quebec, cole de Technologie Suprieure, Montreal, QC, Canada. His research interests include
the use of fiber-reinforced polymer composites for strengthening and retrofitting of concrete structures.
Omar Chaallal, FACI, is a Professor of construction engineering,
University of Quebec, cole de Technologie Suprieure. He is a member
of ACI Committee 440, Fiber-Reinforced Polymer Reinforcement. His
research interests include experimental and analytical research on the use
of fiber-reinforced polymer composites for reinforcement and repair of
concrete structures.

ACKNOWLEDGMENTS

The financial support of the National Science and Engineering Research


Council of Canada (NSERC), the Fonds qubcois de la recherche sur la
nature et les technologies (FQRNT), and the Ministre des Transports du
Qubec (MTQ) is gratefully acknowledged. The efficient collaboration of
J. Lescelleur (Senior Technician) and J. M. Rios (Technician) at TS in
conducting the tests is acknowledged.

REFERENCES

ACI Committee 440, 2008, Guide for the Design and Construction of
Externally Bonded FRP Systems for Strengthening Concrete Structures
(440.2R-08), American Concrete Institute, Farmington Hills, MI, 76 pp.
Bousselham, A., and Chaallal, O., 2004, Shear-Strengthening Reinforced Concrete Beams with Fiber-Reinforced Polymer: Assessment of
Influencing Parameters and Required Research, ACI Structural Journal,
V. 101, No. 2, Mar.-Apr., pp. 219-227.
Bousselham, A., and Chaallal, O., 2008, Mechanisms of Shear Resistance of Concrete Beams Strengthened in Shear with Externally Bonded
FRP, Journal of Composites for Construction, ASCE, V. 12, No. 5,
pp.499-512.
Chaallal, O.; Mofidi, A.; Benmokrane, B.; and Neale, K., 2011,
Embedded Through-Section FRP Rod Method for Shear Strengthening
of RC Beams: Performance and Comparison with Existing Techniques,
Journal of Composites for Construction, ASCE, V. 15, No. 3, pp. 374-383.
Chaallal, O.; Nollet, M. J.; and Perraton, D., 1998, Strengthening
of Reinforced Concrete Beams with Externally Bonded Fiber-Reinforced-Plastic Plates: Design Guidelines for Shear and Flexure, Canadian
Journal of Civil Engineering, V. 25, No. 4, pp. 692-704.
Chen, G. M.; Teng, J. G.; and Chen, J. F., 2012, Shear Strength Model
for FRP-Strengthened RC Beams with Adverse FRP-Steel Interaction,
Journal of Composites for Construction, ASCE, DOI: 10.1061/(ASCE)
CC.1943-5614.0000313.
Chen, G. M.; Teng, J. G.; Chen, J. F.; and Rosenboom, O. A., 2010,
Interaction between Steel Stirrups and Shear-Strengthening FRP Strips in
RC Beams, Journal of Composites for Construction, ASCE, V. 14, No. 5,
pp. 498-509.
Khalifa, A.; Gold, W. J.; Nanni, A.; and Aziz, A., 1998, Contribution
of Externally Bonded FRP to Shear Capacity of RC Flexural Members,
Journal of Composites for Construction, V. 2, No. 4, pp. 195-203.
Mofidi, A., and Chaallal, O., 2011a, Shear Strengthening of RC Beams
with Epoxy-Bonded FRPInfluencing Factors and Conceptual Debonding
Model, Journal of Composites for Construction, ASCE, V. 15, No. 1,
pp.62-74.
Mofidi, A., and Chaallal, O., 2011b, Shear Strengthening of RC Beams
with Externally Bonded FRP Composites: Effect of Strip-Width to StripSpacing Ratio, Journal of Composites for Construction, ASCE, V. 15,
No.5, pp. 732-742.
Oehlers, D. J.; Seracino, R.; and Smith, S. T., 2008, Design Guideline
for RC Structures Retrofitted with FRP and Metal Plates: Beams and Slabs,
HB 305-2008, Standards Australia, Sydney, Australia, 73 pp.
Triantafillou, T. C., 1998, Shear Strengthening of Reinforced Concrete
Beams Using Epoxy-Bonded FRP Composites, ACI Structural Journal,
V.95, No. 2, Mar.-Apr. pp. 107-115.
Uji, K., 1992, Improving Shear Capacity of Existing Reinforced
Concrete Members by Applying Carbon Fiber Sheets, Transactions of the
Japan Concrete Institute, V. 14, pp. 253-266.

361

NOTES:

362

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S32

Punching of Reinforced Concrete Flat Slabs with DoubleHeaded Shear Reinforcement


by Maurcio P. Ferreira, Guilherme S. Melo, Paul E. Regan, and Robert L. Vollum
Twelve slabs, 11 of which contained double-headed studs as
shear reinforcement, were tested supported by central column
and loaded concentrically. Their behavior is described in terms of
deflections, rotations, strains of the concrete close to the column,
strains of the flexural reinforcement across the slab width, and
strains of the studs. All failures were by punching, in most cases
within the shear reinforced region. The treatments of punching
resistance in ACI318, Eurocode 2 (EC2), and the critical shear
crack theory (CSCT) are described, and their predictions are
compared with the results of the present tests and 39 others from
the literature. The accuracy of predictions improves from ACI 318
to EC2 to CSCTthat is, with increasing complexity. However, the
CSCT assumptions about behavior are not well supported by the
experimentalobservations.
Keywords: codes; flat slabs; punching; shear studs.

INTRODUCTION
There is no generally accepted theoretical treatment of
punching, and design is based on empirical methods given in
codes of practice. While there is similarity between them in
terms of general approach, there are considerable differences
in their assumptions and the resulting equations, which leads
to uncertainties about their reliability.
A further cause of uncertainty is the wide variety of types
of shear reinforcement, such as stirrups of various forms,
bent-up bars, welded fabric, and stud systems. Comparisons
of design equations with the results of tests using different
types of shear reinforcement can result in a wide scatter,
while comparisons of slabs with only one type are often
limited by the restricted data available.
This paper presents the results of tests1 of slabs with
double-headed studs as shear reinforcement, followed by
a short review of the design methods of ACI 318,2 Eurocode2 (EC2),3 and the critical shear crack theory (CSCT) of
Muttoni et al.,4,5 which is the basis of the punching clauses
of the fib Model Code 2010 draft.6 The results of the present
tests and of others on slabs with double-headed shear reinforcement are then compared with the three design methods.
RESEARCH SIGNIFICANCE
There are considerable differences between the design
methods for punching in ACI 318, EC2, and the CSCT.
The primary objective of the experimental study described
in this paper was to assess the realism of the assumptions
underlying these design methods. The principal variables in
the test series were the sizes and spacings of the studs, and
the size and shape of the columns. Extensive measurements
were made of slab rotations and strains in the concrete, and
flexural and shear reinforcement. Comparisons between
ACI Structural Journal/March-April 2014

Fig. 1Test arrangements. (Note: Dimensions in mm;


1mm= 0.0394 in.)
experimental and calculated strengths for the present tests
and others are presented to evaluate the accuracies of
themethods.
EXPERIMENTAL PROGRAM
Twelve tests were made at the University of Brasilia. The
specimens were square slabs 2.5 x 2.5 m (8.2 x 8.2 ft) on plan
and 180 mm (7.1 in.) thick supported centrally by circular or
square columns (Type C and S slabs, respectively). Equal
downward loads were applied at eight points close to the
slab edges, as shown in Fig.1.
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-119, doi:10.14359.51686535, was received April 4, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

363

The main variables were the shape and size of the column,
the amount and distribution of the shear reinforcement, and
some details of the main reinforcement.
The concrete was made with ordinary portland cement,
natural sand, and crushed limestone aggregate with a
maximum size of 9.5 mm (3/8 in.). The concrete strength
was determined from 100 x 200 mm (4 x 8 in.) control cylinders that were tested at the same time as the slabs.
The arrangement of flexural reinforcement was basically
the same in all but two of the specimens (Slabs C5 and C6).
The general arrangement of the upper tension reinforcement
was 16 mm (No. 5) bars with fy = 540 MPa (78 ksi) and
Es = 213 GPa (30,893 ksi) at spacings of 100 mm (4 in.)
in the outer layer and 90 mm (3.54 in.) in the inner layer,
providing almost equal flexural resistances in two directions.
The bottom reinforcement was 8 mm (0.315 in.) bars positioned directly below alternate top bars. At the edges, each
top bar was lapped with a 12.5 mm (No. 4) hair-pin shaped
bar with 500 mm (20 in.) horizontal legs. Only minor adjustments to this arrangement were needed to avoid clashes with
shear reinforcement.
In Slab C5, the tension reinforcement in the central parts
of the widths was increased to 20 mm (No. 6) bars with fy =
544 MPa (79 ksi) and Es = 208 GPa (30,168 ksi) and that in
the outer parts was decreased, to obtain a higher reinforcement ratio near the column without significantly altering the
flexural capacity. The details of this slab are shown in Fig.2.
As the failures of some of slabs appeared to be influenced
by crushing of the soffit near the column, Slab C6 was
provided with compression reinforcement comprised of four
16.0 mm (No. 5) bars through the column in each direction,
and 12.5 mm (No. 4) bars below all the top bars in the rest
of the width.
The shear reinforcement was double-headed studs made
of deformed 10 mm (No. 3) bars with fyw = 535 MPa (78 ksi)
and Es = 211 GPa (30,603 ksi), or 12.5 mm (No. 4) bars with
fyw = 518 MPa (75 ksi) and Es = 204 GPa (29,588 ksi). The
heads, with diameters three times the bar size, were welded
to the shanks, and the completed studs were spot-welded to
nonstructural carrier rails, which were 10 mm (3/8 in.) wide
and 3.2 mm (1/8 in.) thick. The shear reinforcement was
positioned from above, with the carrier rails sitting on the
upper tension bars either directly or via cross rails.
Tests of studs, in which the loading was applied via the
heads, showed that the welds between the heads and shanks
were able to develop the full strengths of the bars with
ductile failures away from the welds.
In all but one of the slabs, the lines of studs ran outward
from the columns along equally spaced radial lines (radial
arrangement). The exception was Slab C4, where a cruciform arrangement was used. Typical details are shown in
Fig. 3. Table 1 summarizes the characteristics of all slabs.
TEST RESULTS
Deflections and rotations
Deflections of the top surfaces of the slabs were measured
along their centerlines by dial gauges mounted from frames
spanning over the slabs and supported on the laboratory
364

Fig. 2Flexural reinforcement of Slab C5. (Note: Dimensions


in mm; 1 mm = 0.0394 in.)
floor. An example of the deflected profiles is shown in
Fig.4, where it can be seen that segments of the slab rotated
about axes at or very close to the column face, and the top
surfaces remained more or less straight on radial lines. The
displacements of the slab very close to the column, visible in
Fig.4, were likely due mostly to movements in the support
of the column, and were not deflections of the slab relative
to thecolumn.
Figure 5 shows envelopes of the experimental loadrotation relationships, and the theoretical ones according
to CSCT. The experimental rotations plotted are the averages of values determined from deflections, measured on
the centerline of the slabs at distances 274 and 1049 mm
(10.8and 41.3 in.) from the slab center in the North, South,
East, and West directions. The CSCT values have been
calculated from Eq. (13) and (14). The correlations between
experimental and calculated results are good.
Strains of concrete
Strain gauges were used to measure the strains of the
bottom surfaces of the concrete close to the columns. The
general responses were similar to those observed by others.
At low loads, the compression strains in both directions
were similar, and increased with increasing load. As loading
continued, the radial strains stabilized and then decreased,
sometimes becoming tensile before failure. The tangential
compressions increased at progressively higher rates.
At circular columns, the maximum compression strains
were from 2.5 to 2.8, where the gauges were 20 mm
(0.8in.) from the column faces, and 3.1 to 3.2, where the
gauges were 40 mm (1.6 in.) from the faces. In the slabs
with square columns, the maximum compressions recorded
were lower, but this was probably due to their being on the
slab centerlines, while the greatest strains were likely at the
corners of the columns.
ACI Structural Journal/March-April 2014

Table 1Characteristics of test slabs


Shear reinforcement
Slab No.

Column size , mm (in.)

d, mm (in.)

,%

fc, MPa (ksi)

Studs

so, mm (in.)

sr, mm (in.)

C1

270 (10.6)

143 (5.6)

1.48

47.8 (6.9)

10 10.0 x 6

70 (2.8)

100 (3.9)

C2

360 (14.2)

140 (5.5)

1.52

46.9 (6.8)

10 10.0 x 6

70 (2.8)

100 (3.9)

C3

450 (17.7)

142 (5.6)

1.49

48.9 (7.1)

10 10.0 x 6

70 (2.8)

100 (3.9)

C4

360 (14.2)

140 (5.5)

1.52

47.9 (6.9)

12 10.0 x 6

70 (2.8)

100 (3.9)

C5

360 (14.2)

140 (5.5)

2.00

49.7 (7.2)

10 10.0 x 6

70 (2.8)

100 (3.9)

C6

360 (14.2)

143 (5.6)

1.48

48.6 (7.0)

10 10.0 x 6

70 (2.8)

100 (3.9)

C7

360 (14.2)

144 (5.7)

1.47

49.0 (7.1)

10 10.0 x 7

55 (2.2)

80 (3.1)

C8

360 (14.2)

144 (5.7)

1.47

48.1 (7.0)

12 10.0 x 6

70 (2.8)

100 (3.9)

S1

300 (11.8)

145 (5.7)

1.46

48.3 (7.0)

12 10.0 x 2

70 (2.8)

100 (3.9)

S2

300 (11.8)

143 (5.6)

1.48

49.4 (7.2)

12 10.0 x 4

70 (2.8)

100 (3.9)

S5

300 (11.8)

143 (5.6)

1.48

50.5 (7.3)

S7

300 (11.8)

143 (5.6)

1.48

48.9 (7.1)

12 12.5 x 4

70 (2.8)

100 (3.9)

Diameter in Series C, side length in Series S.

Calculated as

Number of studs per perimeter, stud size (mm) by number of perimeters.

x y . In all slabs except C5, reinforcement distributed uniformly across widths. For C5, 2.00% is ratio in central (c + 6d), and 1.56% is ratio for full width.

Fig. 3Shear reinforcement. (Note: Dimensions in mm;


1mm = 0.0394 in.)
The difference in maximum strains at distances of 20 and
40 mm (0.8 and 1.6 in.) points to restraint from the columns,
and the strains 40 mm (1.6 in.) from the columns were
probably high enough to indicate distress of the concrete
due to tangential stresses, except in the one slab without
shearreinforcement.
ACI Structural Journal/March-April 2014

Fig. 4Load-displacement of Slab C3. (Note: 1 mm =


0.0394 in.)
365

Fig. 5Load-rotation behavior of tested slabs. (Note:


1kN= 0.2248 kip.)
Figure 6 shows the developments of strains measured in
Slabs C2 and C6, the former having only nominal bottom
steel, and the latter being the slab with considerable compression reinforcement. In C6, the maximum strain of 3.2 was
reached at a load equal to the failure load of C2. Thereafter, the strain in C6 decreased as the load was increased to
failure. This suggests that the 12% higher ultimate strength
was achieved with the compression reinforcement locally
taking over the function of the failing concrete.
Strains of flexural reinforcement
Strains of the flexural tension reinforcement were
measured by pairs of strain gauges at opposite ends of diameters of the upper bars, at a section just outside the column.
The resulting profiles of tangential strains for Slabs C1
through 4 and C8 are shown in Fig. 7. Strains beyond yield
were recorded in considerable parts of the slab widths, but
the yielding never reached the slab edge, that is, a yield line
was never developed.
Strains of shear reinforcement
Strains were measured at the midheight of the shear
studs in four lines of the shear reinforcement in all slabs.
366

Fig. 6Strains of concrete at soffits of Slabs C2 and C6.


(Note: 1kN= 0.2248 kip.)
The strains were measured in the inner three rings of shear
reinforcement in the slabs with circular columns, and at all
perimeters for the slabs with square columns. The average
stresses (strain Es) in the shear reinforcement are summarized in Table 2. Typical profiles of stud stresses along radial
lines are shown in Fig. 8. The stud stresses summarized in
Table 2 are thought to be reasonably close to the maximum
stresses in the studs because the studs were short. The difference between the measured and maximum stress depends on
the product of the bond stress and the distance from the shear
crack to the midheight of the studs. Allowing for bond along
lengths between cracks and strain gauges, it appears that
the first perimeter of shear reinforcement is likely to have
yielded in all the tests except C6, C7, and S7.
Ultimate loads and modes of failure
All of the slabs failed by punching, and Table 2 gives the
ultimate loads and summarizes data from relevant strain
measurements at or close to failure. The slabs with shear
reinforcement failed inside the shear reinforced areas in all
cases but S1, where there were only two perimeters of studs,
and S7, where the diameter of the studs was 12.5 mm (No. 4).
ACI Structural Journal/March-April 2014

Fig. 7Strains of flexural reinforcement. (Note: 1 mm =


0.0394 in.; 1 kN = 0.2248 kip.)

ACI Structural Journal/March-April 2014

Fig. 8Average stud stresses at Perimeters 1 to 3. (Note:


1kN = 0.2248 kip; 1 MPa = 0.1450 kip.)
367

Specimens C2, C5, and C6 (Table 1) were similar apart


from the detailing of the flexural reinforcement. Comparison
of the shear strengths of C2 and C5 shows that the punching
resistance was increased by approximately 15% by concentrating 60% of the flexural reinforcement into a 1 m (39 in.)
wide band centered on the column whilst maintaining the
same flexural capacity across the slab width as in the other
tests. The shear strength of C6 was increased by approximately 12% relative to C2 through the provision of additional compression steel.

The specimens were saw-cut half width in two orthogonal


directions to reveal the failure surfaces. For the inside failures, most of the surfaces ran from the soffit at the column
face to reach the level of the top steel at the second perimeter of studs, but there were exceptions. In both sections of
SlabC1 and one of Slab C2, the failure surfaces crossed the
inner studs very close to their upper heads. In Slabs C5, C6,
and S2, they reached to the level of the top bars at the third or
fourth perimeter of studs. In C7, there were multiple cracks
reaching the main steel from the third to the fifth perimeter
in one section, while in the perpendicular direction, the
surface ran just above the lower heads of the studs out to the
fourth layer and reached the top steel at the sixth layer. In the
outside failures, the surface was entirely outside the studs in
S1, but did cross them just above their lower heads in S7. In
some slabs, most notably C6, there was spalling of the slab
around the column that commenced before failure.
METHODS OF CALCULATION
All three approaches considered herein take the punching
strength of a slab with shear reinforcement as the least of
VR,cs, VR,out, and VR,max, but not less that VR,c, where VR,c is the
resistance of an otherwise similar slab without shear reinforcement; VR,cs is the combined resistance of the concrete
and shear reinforcement; VR,out is the resistance from the
concrete alone just outside the shear reinforcement; and
VR,max is the maximum resistance possible for a given column
size, slab effective depth and concrete strength.
These resistances correspond to failures of the types
shown in Fig. 9. The calculations are made for perimeters at
specified distances from supports: uo is the perimeter at the
outline of the support; u1 is the perimeter used in the calculation of VR,c and VR,cs; and uout is the perimeter used in the
calculation of VR,out.

Fig. 9Types of punching failure.


Table 2Summary of test results

Average stud stresses, MPa (ksi)


Slab No.

c,max ,

ry , mm (in.)

Vu, kN (kip)

Failure mode

C1

2.66

450 (17.7)

535 (77.6)

317 (46.0)

137 (19.9)

858 (192.9)

In

C2

2.81

550 (21.7)

530 (76.9)

235 (34.1)

121 (17.5)

956 (214.9)

In

C3

2.54

625 (24.6)

511 (74.1)

362 (52.5)

189 (27.4)

1077 (242.1)

In

C4

||

770 (30.3)

535 (77.6)

461 (66.8)

297 (43.1)

1122 (252.2)

In

2.28

C5

3.24

490 (19.3)

504 (73.1)

264 (38.3)

160 (23.2)

1117 (251.1)

In

C6

3.20

750 (29.5)

479 (69.5)

421 (61.0)

474 (68.7)

1078 (242.3)

In

C7

3.14

540 (21.3)

386 (56.0)

419 (60.8)

167 (24.2)

1110 (249.5)

In

C8

3.14

660 (26.0)

535 (77.6)

436 (63.2)

179 (26.0)

1059 (238.1)

In

S1

2.37

560 (22.0)

535 (77.6)

473 (68.6)

1021 (229.5)

Out

S2

2.15

570 (22.4)

535 (77.6)

514 (74.5)

216 (31.3)

1127 (253.4)

In

S5

1.47

130 (5.1)

779 (175.1)

S7

2.67

600 (23.6)

238 (34.5)

285 (41.3)

137 (19.9)

1197 (269.1)

Out

c,max is maximum tangential strain of concrete (measured 20 mm from columns in C1 to C4, S1 and S2, and 40 mm from columns in C5 to C8 and S7). For slab Type S, strains
measured on centerlines.
*

ry is radius in which tangential strain > y.

Averages of Ess fyw in Perimeters 1, 2, and 3.

Ultimate shear force including self-weights of slabs and loading system.

||

Measured at 0.85Vu.

368

ACI Structural Journal/March-April 2014

Fig. 12Control perimeters: CSCT.


Fig. 10Detailing and control perimeters: ACI 318.

VR,cs = 0.75VR,c + VR,s (2)


VR,s =

d
Asw f yw with fyw 414 MPa (60,000 psi) (3)
sr
VR,out = 1
6

VR,max = 2
3
VR,max = 1
2

fc uout d (4)

fc u1 d if sr 0.5d (5a)

fc u1 d if 0.5d sr 0.75d (5b)

fc is limited to 69 MPa (10,000 psi) for calculation purposes.


Fig. 11Detailing and control perimeters: EC2.

EC2-04

The locations and lengths of u1 and uout vary with the


method of calculation.
The symbols used for spacings of shear reinforcement are
as follows: so is the distance from column to inner studs; sr
is the radial spacing of studs; and st is the tangential spacing
of studs at a perimeter. The effective depth d is taken as
the average for orthogonal directions, d = (dx + dy)/2. The
expressions for punching resistances are given below in SI
units (N and mm) without any explicit safety factors. Those
from ACI 318 are for nominal resistances, and the others
are for characteristic resistances. The perimeters u1 and uout
and the detailing requirements, in relation to the spacings of
shear reinforcement, are illustrated by Fig. 10, 11, and 12 for
ACI 318, EC2, and the CSCT, respectively.

k = 1 + 200 / d 2

VR.cs = 0.75VR,c + VR,s (8)

ACI 318-08
As double-headed studs are not considered explicitly, the
equations used herein are those for studs with heads at their
top ends and bottom anchorages provided by welds to structural rails

fc

u d (12)
VR,max = 0.3 fc 1
250 o

VR,c = 1
3

fc u1 d (1)

ACI Structural Journal/March-April 2014

VR,c = 0.18k(100fc)1/3u1d (6)

VR,s = 1.5

(7)

d
Asw f yw,ef (9)
sr

fyw,ef = 1.15(250 + 0.25d) fyw 600 MPa (87,000 psi) (10)


VR,out = 0.18k(100fc)1/3uout,efd (11)

is the ratio of flexural reinforcement calculated as x y ,


where x and y are the ratios in orthogonal directions determined for widths equal to those of the column plus 3d to

369

VR,s = Aswsi() Aswfyw (16)

VR,cs = VR,c + VR,s (17)


VR,out =

0.75uout dv fc
1 + 15d / (16 + dg )

(18)

VR,max = 3VR,c (19)

Fig. 13Punching strengths according to CSCT: Slab C1.


(Note: 1 kN = 0.2248 kip.)
either side. 0.02 for calculation purposes, and the scope
of EC2 is limited to fc 90 MPa (13,000 psi).
Critical shear crack theory
In the CSCT, punching resistances are related to the rotation of the slab, outside a critical crack. Half of this rotation is assumed to occur in the critical shear crack and, as the
slab rotates, the concrete component of shear resistance at
the crack is assumed to decrease, while the component from
the shear reinforcement increases up to yield. The rotation
is related to the ratio V/Vflex, where V is the acting shear, and
Vflex is the shear force corresponding to the flexural capacity,
calculated by yield-line theory. Values of VR,c, VR,cs, VR,out,
and VR,max can be determined as shown in Fig. 13, by plotting
the resistances against and finding their intersections with
Eq. (13)
3

r fy V 2
= 1.5 s

d Es V flex (13)

where rs is the distance from the column center to the line


of radial contraflexure; and is in radians. For typical
punching test specimens, rs is the distance from the column
center to the slab edge. The flexural failure load Vflex is
approximatedby5
rs
Vflex = 2mR r r (14)
q c
where mR is the moment resistance per unit length of yield
line; rq is the radius at which loading is applied; and rc is the
radius of the column and for square columns can be taken as
2c/, where c is the side length of the column.
In the CSCT average method given in Reference 5
0.75u1 d fc

370

VR,c =

1 + 15d / (16 + dg )

(15)

where si is the stress in the i-th perimeter of shear reinforcement which is related to the width of the critical shear crack,
where it crosses the shear reinforcement.5 The summations
Asw and Aswsi() are for all the shear reinforcement
within a distance d from the column.
The CSCT average method is intended to give approximately mean strengths. In it, the stresses in studs at different
distances (d) from the column are calculated assuming that
the width of the critical shear crack increases linearly, from
zero at the slab soffit to the width corresponding to a slab
rotation and a crack opening angle of 0.5, at the level
of the tension reinforcement. The stress in a stud is then
obtained by equating the vertical component of the crack
opening to the elongation of the stud for a given stress at
the crack.
COMPARISONS OF TESTS AND CALCULATIONS
General
Experimental strengths from the present tests and from
others reported in the literature have been compared with
resistances calculated by the three methods described previously. The shear reinforcement in the tests by Regan,7 Regan
and Samadian,8 Beutel,9 and Birkle10 was double-headed
studs made from either deformed or plain round bars. In the
tests by Gomes and Regan,11 it was slices of steel I-beams
with the flanges acting as anchorages. The shear reinforcement was positioned radially unless noted as ACI type in
Table A1 in Appendix A.
The calculations of punching resistances were made using
the expressions given previously, with their limits generally
respected. Exceptions to this were as follows.
For ACI 318 and EC2, the limits on so/d and sr/d were
given a little tolerance. Values of sr/d up to 0.8 were treated
as acceptable, and for EC2 the lower limit so/d < 0.3 was
waived with values going down to 0.24. EC2 does not
envisage the use of plain round shear reinforcement, but this
has been ignored, and lower limits on d for the use of shear
reinforcement were ignored. (The least effective depth in the
tests used was 124 mm [4.9 in.] in six slabs by Birkle.10)
The CSCT shear strengths were calculated using slab
rotations calculated with Eq. (13), in which Vflex was calculated with Eq. (14). The stresses in the shear reinforcement
were calculated in accordance with the recommendations in
(5). The resulting slab rotations were slightly greater than
the measured slab rotations, as illustrated in Fig. 5. The
predicted shear strengths typically increase by less than 5%
for the slabs tested in this program if measured rotations are
used instead of calculated rotations. For the CSCT, there
ACI Structural Journal/March-April 2014

Table 3Comparisons with test results for slabs


without shear reinforcement
Vu/Vcalc
Slab No. d, mm (in.) , %

ACI 318

EC2

CSCT

EC2

Table 4Statistics of Vu /Vcalc for slabs with shear


reinforcement
No. of
tests

ACI 318
Mean

EC2

COV

143 (5.6)

1.48

1.30

Gomes and Regan


1
1A

159 (6.3)
159 (6.3)

1.27
1.27

1.16
1.20

1.20

1.24

1.10

0.89
0.92

1.30
1.12
0.88

1.11
0.94
0.78

1.10
1.02
0.86

0.96
0.93
0.78

Mean

1.16

0.99

1.04

0.93

Coefficient of variation

0.13

0.15

0.12

0.11

Birkle10

1.53
1.29
1.10

Mean

COV

1.47

0.148

1.56

0.100

are only a few cases in which there were two perimeters of


shear reinforcement within a distance d from the support,
but there are some where a second layer was not much
further out (all of the slabs by Gomes and Regan11 where the
distances varied from 1.0d to 1.04d and Slabs 2, 3, 9, and 12
by Birkle10 where the distances were from 1.09d to 1.18d).
The second perimeter has been included in Asw, where the
distance was less than 1.05d.
Details of the individual slabs and the results obtained
are given in Appendix A, while Tables 3 and 4 summarize
the results of the comparisons for slabs without and with
shearreinforcement.
Although there are only six results in Table 3, it is noteworthy that, for all the methods of calculation, Vu/Vcalc
decreases with increasing effective depth. This is not
surprising for ACI 318, which has no size factor, or for EC2,
where the size factor is constant for d 200 mm (7.9 in.).
It is surprising for the CSCT, which includes a size factor
taking account of the effective depth and the maximum size
of aggregate. The best correlation in the table is that for
EC2*, which is the same as EC2, but without the limit on
k=1 + (200 / d ). With this modification, however, the
mean Vu/Vcalc is low, and the coefficient of 0.18 in Eq. (7)
would need to be reduced.
Table 4 summarizes the results of the comparisons with
the 45 slabs with shear reinforcement, and all three methods
of calculation are broadly satisfactory.
The coefficient of variation of Vu/Vcalc decreases with
increasing complexity in the method. The ACI method is
the simplest and gives a coefficient of variation of 0.162.
The EC2 method is slightly more complex, and reduces the
coefficient by 0.026, while the CSCT is considerably more
complicated, but gives a further reduction of 0.015.
There are no unsafe predictions from ACI 318, but there
are four from EC2 and the CSCT, with the lowest values
of Vu/Vcalc being 0.88 for EC2, and 0.90 for the CSCT. An
overall reduction of Vcalc by 4% would make each of these
methods safe in the sense of limiting the probability of an
unsafe prediction to 5%, assuming a statistically normal
distribution of Vu/Vcalc.

1.06
Beutel

1.72

0.093

0.160

1.23

0.108

0.087

1.05

0.102

0.087

1.16

0.101

0.081

1.28

0.083

0.105

1.06

0.050

0.136

1.16

0.121

1.34

Gomes and Regan11


10

1.76

0.134

1.28
Birkle

1.35

0.185

EC2 calculations as for EC2, but with no upper limit on k = 1+(200/d).

ACI Structural Journal/March-April 2014

1.20

Regan7 and Regan and Samadian8


1.00
1.03

124 (4.9)
190 (7.5)
260 (10.2)

11

11

0.96
1.01

1
7
10

COV

Ferreira1

Ferreira1
S5

Mean

CSCT

10

1.09
Total

45

1.56

0.162

1.19

ACI 318 and EC2 are basically empirical, but the CSCT
claims a rational basis. Unfortunately, its modeling of slab
deformations is incorrect. The rotation is predicted satisfactorily by Eq. (13), but, as can be seen from Fig. 4, it is not
divided equally into movements at the column face and in a
shear crack. In addition, the surfaces at which failure occurs
are not at 45 degrees to the slab plane, but have variable
geometries. Refer to the section entitled, Ultimate loads
and modes of failure.
In nine tests by Ferreira1 and five by Birkle,10 EC2 predicts
outside failures for slabs that actually failed in the shear-reinforced zones. Its predictions of failure modes in the other
series are generally good. The main cause of the problem
seems to be the overestimation of VR,cs. For the slabs by
Ferreira,1 the mean Vu/VR,cs is 0.98, and the coefficient of
variation is 0.061. For Birkles tests,10 the corresponding
figures are 0.88 and 0.101, but would be improved if the
four slabs with so/d less than 0.3 were excluded. The situation could be improved by either a reduction of VR,c or by
interpreting the codes expression for the design value of the
stud stress (fywd,ef) as not requiring a safety factor so long as
fywd,ef is less than fyw/1.15that is, by taking fyw,ef as (250 +
0.25d) fyw.
The EC2 predictions of VR,out for slabs with radial arrangements of shear reinforcement are generally satisfactory,
though perhaps over-conservative for the slabs by Gomes
and Regan.11 In these slabs, the 0.64d widths of the I-beam
flanges reduced the clear tangential spacing of the shear
reinforcement. This could be allowed for, and would make
Vu/VR,out for these tests similar to those for other series.
The strength of Ferreiras1 Slab C4 with an ACI cross
arrangement of studs which failed inside is predicted very
conservatively with Vu/VR,out = 1.69. For Birkles10 slabs with
the ACI layout, which failed outside, the strengths are well
predicted with Vu/VR,out = 1.21. Slab C4 was unrealistic in
relation to EC2 design because it had six perimeters of studs,
while the same strength would be calculated for a slab with
two perimeters of studs. The performance of C4 is in marked
371

Fig. 14Influence of slab rotation on Vu/Vcalc CSCT.


contrast to that of slabs by Mokhtar et al.,12 with up to eight
perimeters of studs on stud rails. Their strengths are quite
well predicted by EC2.
Influence of slab rotation on shear resistance
provided by concrete
Unlike ACI 318 and EC2, the CSCT predicts that the shear
resistance provided by concrete reduces with slab rotation,
which is assumed to be proportional to (V/Vflex)1.5. This has
significant implications for design because Vu/Vflex may be
close to one for practical slabs. Consequently, the CSCT can
require significantly greater areas of shear reinforcement
than EC2. The influence of slab rotation on Vu/Vcalc for the
CSCT is illustrated by Fig. 14, where the ratio is plotted
against the normalized rotation d/(16 + dg), with calculated by Eq. (13). There is a clear tendency for the CSCT
to become more conservative with increasing slab rotation,
which suggests that Vu is either independent of , or that
the CSCT overestimates the influence of rotation. In the
case of outside failure, this is to be expected as the rotation
develops close to the column and not within a crack outside
the shearreinforcement.
Muttoni4 plots Vu/u1d fc against d/(16 + dg) for 99 tests
of slabs without shear reinforcement, and shows that experimental strengths are close to the predictions of Eq. (15),
which may appear to contradict the preceding paragraph.
This, however, is not the case. Figure 15 shows Vu/u1d fc
plotted against d/(16 + dg) for slabs similar to Ferreiras1
C2, but without shear reinforcement, and with from 0.4
to 4.0%. The values of Vu have been calculated by EC2 and
the CSCT, u1 is the CSCT control perimeter, and is the
rotation calculated by Eq. (13). It can be seen that the effect
shown in Muttonis figure can be accounted for by the EC2
relationship between VR,c and 1/3 without involving in the
calculations.
Failure surface and locations of shear
reinforcement
ACI 318 and the CSCT assume punching surfaces to be
inclined at 45 degrees, while EC2 assumes an inclination of
26.6 degrees. These are simplifications of a reality in which
the angle increases with increasing shear reinforcement

372

Fig. 15Influence of slab rotation on shear strength.


(refer to Carvalho et al.13). Reasonable results can, however,
be obtained in most instances with fixed angles, provided the
expressions for VR,c and VR,s are constructed appropriately.
This seems to be the case with ACI 318 and EC2, with the
former considering d/sr perimeters of studs acting at fyw, and
the latter assuming 1.5d/sr perimeters acting at fyw,eff, which
is typically approximately 0.7fyw for test slabs. The situation
is more complex in the CSCT, as its numbers of perimeters
depend on the exact distances of studs from the column.
There are cases where the use of different failure surfaces
has a significant effect. Slabs 2, 3, 10, and 11 by Gomes
and Regan11 are an example. In these slabs so = sr 0.5d,
Asw = 226 mm2 (0.36 in.2) in Slabs 2 and 10, and 325 mm2
(0.5in.2) in Slabs 3 and 11. Slabs 2 and 3 had two perimeters of shear reinforcement, while Slabs 10 and 11 had three.
Thus, for ACI 318, two perimeters are taken into account for
all the slabs, while in EC2, two perimeters are active in Slabs
2 and 3, but three are active in Slabs 10 and 11. All four
slabs failed inside their shear reinforced zones. The EC2
ratios Vu/VR,cs are 1.26 and 1.21 for Slabs 2 and 3, and 1.28
and 1.31 for Slabs 10 and 11. The ACI ratios are 1.39 and
1.28 for Slabs 2 and 3, and 1.58 and 1.61 for Slabs 10 and
11, showing that the extra perimeter of shear reinforcement
had an effect.
The same slabs illustrate a problem with the CSCTs
considering the active shear reinforcement to be exactly
that within a distance d from a column rather than using an
expression in d/sr. Because of variations of effective depths,
(so + sr) = 1.05d in three cases instead of the intended 1.0d.
This discrepancy has been ignored in the calculations for
Table A1, but if the CSCT were applied strictly the ratios
Vu/VR,cs, which are already unusually high for Slabs 10 and
11, would be significantly increased.
CONCLUSIONS
Comparisons have been made between the punching
strengths of 45 slabs with and six slabs without shear reinforcement and those predicted by ACI 318, EC2, and the CSCT.
ACI 318 is the simplest method, and gives only one unsafe
prediction, which is for a slab without shear reinforcement.
Its mean value of Vu/Vcalc for slabs with shear reinforcement
is rather high, and the coefficient of variation is 0.162. The
ACI Structural Journal/March-April 2014

most apparent weakness is the lack of any treatment of the


depth effect in the shear resistance from the concrete.
EC2 is only slightly more complicated, but reduces the
coefficient of variation of Vu/Vcalc by 0.026. The mean is
also reduced, and there are four unsafe predictions for slabs
with and four for slabs without shear reinforcement. The
simplest way to obtain a characteristic level of safety would
be to reduce the constant in the expression for the concrete
component of resistance and extend the range of slab depths
affected by the depth factor.
The CSCT is considerably more complex, and reduces
the mean and coefficient of variation of Vu/Vcalc, by a further
0.015. There are unsafe predictions for four slabs with and
one without shear reinforcement.
The CSCT goes further than the other two approaches in
attempting to model the slab behavior. Although its expression for total slab rotation seems good, the assumption that
half of this rotation occurs in the critical shear crack is incorrect, as nearly all of it is at the column face. The assumption
that all critical cracks are at 45 degrees to the slab plane is
also incorrect.
The relationship assumed between the concrete component of punching resistance and slab rotation is not
confirmed by the test data, and the determination of the area
of active shear reinforcement as that crossed by a particular
45 degree surface seems less satisfactory than considering
d/sr perimeters.
AUTHOR BIOS

ACI member Maurcio P. Ferreira is a Lecturer at the Federal University of Para, Belem, Brazil. He received his PhD from the University of
Braslia, Braslia, Brazil, in 2010. His research interests include ultimate
shear design, strut and tie, and nonlinear finite element modeling.
ACI member Guilherme S. Melo is an Associate Professor at the University of Brasilia, where he was Head of the Department of Civil and Environmental Engineering. He is a member of ACI Committees 440, Fiber-Reinforced Polymer Reinforcement; and Joint ACI-ASCE Committee 445, Shear
and Torsion. His research interests include punching and post-punching of
flat plates, the use of fiber-reinforced plastic (FRP) in concrete structures,
and strengthening and rehabilitation of structures.
ACI member Paul E. Regan is a Professor Emeritus at the University of
Westminster, London, UK, where he was Head of Architecture and Engineering. He was Chair of the European Concrete Committee (CEB) commission on member design. His research interests include member design in
both reinforced and prestressed concrete, with particular emphasis on
problems of punching, shear, and torsion.
ACI member Robert L. Vollum is a Reader in concrete structures at Imperial College London, London, UK, where he also received his MSc and PhD.
His research interests include design for shear, strut-and-tie modeling, and
design for the serviceability limit states of deflection and cracking.

ACKNOWLEDGMENTS

The authors are grateful to the Brazilian Research Funding Agencies


CAPES (Higher Education Co-ordination Agency) and CNPq (National
Council for Scientific and Technological Development) for their support
throughout this research and to RFA-Tech for their permission to include
test results from Reference 7.

Asw =

NOTATION

area of shear reinforcement in one perimeter

ACI Structural Journal/March-April 2014

side length of square column or diameter of circular column


mean effective depth
maximum size of aggregate
depth from tension reinforcement to compression zone
anchorage of shear reinforcement
Es = modulus of elasticity of reinforcement
fc
= cylinder compression strength of concrete
fy
= yield stress of flexural reinforcement
so = distance from column face to first perimeter of shear
reinforcement
sr
= radial spacing of shear reinforcement
st
= tangential spacing of shear reinforcement
st,max = maximum value of st (general in outer perimeter of shear
reinforcement)
u0 = length of column perimeter
u1 = length of control perimeter for calculation of VR,c and VR,cs
uout = length of control perimeter for calculation of VR,out
uout,ef = effective value of uout for calculations by EC2, where st,max > 2d
V
= applied shear force
Vcalc = calculated punching resistance
Vflex = flexural strength of slab calculated by yield-line theory
VR,c = punching resistance of slab without shear reinforcement
VR,cs = punching resistance within shear reinforced zone
VR,max = maximum punching resistance for given column size, slab effective depth, and concrete strength
VR,out = punching resistance outside shear reinforced zone
VR,s = contribution of shear reinforcement to punching resistance VR,cs
Vu = experimental punching strength

= ratio of flexural reinforcement = x y (calculated for width
of column plus 3d to either side in EC2)
= rotation of part of slab outside critical shear crack
c
d
dg
dv

=
=
=
=

REFERENCES

1. Ferreira, M. P., Puno em Lajes Lisas de Concreto Armado com


Armaduras de Cisalhamento e Momentos Desbalanceados, PhD thesis,
Universidade de Braslia, Braslia, Brazil, 2010, 275 pp. (in Portuguese)
available at http://repositorio.bce.unb.br/handle/10482/8965.
2. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-08) and Commentary, American Concrete Institute,
Farmington Hills, MI, 2008, 473 pp.
3. Eurocode 2, Design of Concrete Structures Part 1-1: General Rules
and Rules for Buildings, CEN, EN 1992-1-1, Brussels, Belgium, 2004,
225 pp.
4. Muttoni, A., Punching Shear Strength of Reinforced Concrete Slabs
without Transverse Reinforcement, ACI Structural Journal, V. 105, No. 4,
July-Aug. 2008, pp. 440-450.
5. Fernadez-Ruiz, M., and Muttoni, A., Applications of the Critical
Shear Crack Theory to Punching of R/C Slabs with Transverse Reinforcement, ACI Structural Journal, V. 106, No. 4, July-Aug. 2009, pp. 485-494.
6. Fdration internationale du bton, fib Model Code 2010, First
complete draftV. 2, Bulletin 56, fib, Lausanne, Switzerland, Apr. 2010,
288 pp.
7. Regan, P. E., unpublished tests for RFA-TECH at Cambridge University, 2009.
8. Regan, P. E., and Samadian, F., Shear Reinforcement against
Punching in Reinforced Concrete Flat Slabs, The Structural Engineer,
V.79, No. 10, May 2001, pp. 24-31.
9. Beutel, R., Punching of Flat Slabs with Shear Reinforcement at Inner
Columns, Rheinisch-Westflischen Technischen Hochschule Aachen,
Aachen, Germany, 2002, 267 pp. (in German)
10. Birkle, G., Punching of Flat Slabs: The Influence of Slab Thickness
and Stud Layout, PhD thesis, Department of Civil Engineering, University
of Calgary, Calgary, AB, Canada, Mar. 2004, 152 pp.
11. Gomes, R., and Regan, P. E., Punching Strength of Slabs Reinforced for Shear with Offcuts of Rolled Steel I-Section Beams, Magazine
of Concrete Research, V. 51, No. 2, 1999, pp. 121-129.
12. Mokhtar, A. S.; Ghali, A.; and Dilger, W., Stud Shear Reinforcement for Flat Concrete Plates, ACI Journal, V. 82, No. 5, Sept.-Oct. 1985,
pp. 676-683.
13. Carvalho, A. L.; Melo, G. S.; Gomes, R. B.; and Regan, P. E.,
Punching Shear in Post-Tensioned Flat Slabs with Stud Rail Shear
Reinforcement, ACI Structural Journal, V. 108, No. 5, Sept.-Oct. 2011,
pp.523-531.

373

APPENDIX A
Table A1Comparison between theoretical and experimental results

Birkle10

Gomes and Regan11

Beutel9

Regan and
Samadian8

Regan7

Ferreira1

Author

Slab
No.
C1

Column
size, mm
270
C

d,
mm
143

, %
1.48

fy,
MPa
540

fc,
MPa
48

Shear reinforcement
fyw,
s o,
sr,
Studs
MPa mm
mm
10 f10.0 x 6 535
70
100

stmax,
mm
436

Vu,
kN
858

Failure
mode
In

Vu/
Vflex
0.72

Vu/Vcalc and critical strength


CSCT
ACI 318-08
EC2-04
average
1.34 Max 0.96 Out 1.07
In

C2

360

140

1.52

540

47

10 f10.0 x 6

535

70

100

464

956

In

0.78

1.27

Max

1.11

Out

1.12

In

C3

450

142

1.49

540

49

10 f10.0 x 6

535

70

100

491

1077

In

0.82

1.21

Out

1.20

Out

1.15

In
Out

C4*

360

140

1.52

540

48

12 f10.0 x 6

535

70

100

900

1122

In

0.92

1.47

Max

1.69

Out

1.50

C5

360

140

2.00

544

50

10 f10.0 x 6

535

70

100

464

1118

In

0.88

1.44

Max

1.16

Out

1.29

In

C6

360

143

1.48

540

49

10 f10.0 x 6

535

70

100

464

1078

In

0.86

1.36

Max

1.19

Out

1.24

In
Out

C7

360

144

1.47

540

49

10 f10.0 x 7

535

55

80

442

1110

In

0.88

1.39

Max

1.21

Out

1.09

C8

360

144

1.47

540

48

12 f10.0 x 6

535

70

100

388

1059

In

0.84

1.34

Max

1.03

Out

1.14

In

S1

300

145

1.46

540

48

12 f10.0 x 2

535

70

100

177

1022

Out

0.80

1.71

In

1.36

Out

1.37

Out

S2

300

143

1.48

540

49

12 f10.0 x 4

535

70

100

280

1128

In

0.89

1.77

Out

1.12

Out

1.24

Out

S5

300

143

1.48

540

50

779

0.61

1.30

1.20

1.24

S7

300

143

1.48

540

49

12 f12.5 x 4

518

70

100

280

1197

Out

0.94

1.88

Out

1.19

Out

1.32

Out

300

150

1.45

550

33

10 f10.0 x 4

550

80

120

390

881

0.79

1.45

Out

1.02

Out

1.06

In

300

150

1.76

550

30

12 f10.0 x 6

550

60

100

390

1141

0.88

1.71

Out

1.13

Out

1.11

Out

300

150

1.76

550

26

10 f12.0 x 5

550

60

120

455

1038

0.83

1.73

Out

1.22

Out

1.09

Out

240

160

1.65

550

62

12 f12.0 x 5

550

80

120

352

1268

0.88

1.61

Max

0.88

Out

1.12

In

240

150

1.75

550

42

12 f10.0 x 5

550

75

120

349

1074

0.83

1.81

Max

1.04

In

1.24

In

R3

200

160

1.26

670

33

8 f12.0 x 4

442

80

120

413

850

Out

0.63

1.44

Out

1.04

Out

0.90

Out

R4

200

160

1.26

670

39

8 f12.0 x 6

442

80

80

444

950

Out

0.69

1.39

Out

1.10

Out

0.93

Out

A1

200

160

1.64

570

37

8 f10.0 x 6

519

80

80

444

1000

Out

0.67

1.50

Out

1.08

Out

0.98

Out

A2

200

160

1.64

570

43

8 f10.0 x 4

519

80

120

413

950

In

0.62

1.42

Out

1.03

In

1.00

In

Z1

200

250

0.80

890

25

12 f14.0 x 5

580

100

200

518

1323

Max

0.41

1.50

Max

1.26

Max

0.96

In

Z2

200

250

0.80

890

26

12 f14.0 x 5

580

88

200

511

1442

Max

0.44

1.59

Max

1.30

Max

1.08

In

Z3

200

250

0.80

890

24

12 f14.0 x 5

580

95

188

487

1616

Max

0.50

1.86

Max

1.57

Max

1.20

In

Z4

200

250

0.80

890

32

12 f14.0 x 5

580

88

175

459

1646

Max

0.49

1.66

Max

1.27

Max

1.18

In

Z5

263

250

1.25

562

28

12 f16.0 x 5

544

94

188

505

2024

Max

0.41

1.90

Max

1.31

Max

1.28

In

Z6

200

250

1.25

562

37

12 f16.0 x 5

544

94

188

489

1954

Max

0.39

1.81

Max

1.31

Max

1.23

In
P

200

159

1.27

680

40

560

0.40

1.16

0.94

1.00

1a

200

159

1.27

680

41

587

0.41

1.20

0.98

1.03

2*

200

153

1.32

680

34

8 f6.0 x 2

430

80

80

255

693

In

0.53

1.64

In

1.26

In

1.12

Out

3*

200

158

1.27

670

39

8 f6.9 x 2

430

80

80

255

773

In

0.57

1.64

In

1.21

In

1.20

Out

4*

200

159

1.27

670

32

8 f8.0 x 3

430

80

80

368

853

Out

0.64

1.98

In

1.27

Out

1.26

Out

5*

200

159

1.27

670

35

8 f10.0 x 4

430

80

80

481

853

Out

0.63

1.77

Out

1.24

Out

1.13

Out

200

159

1.27

670

37

8 f10.0 x 4

430

80

80

323

1040

Out

0.76

2.07

Out

1.23

Out

1.34

Out

200

159

1.27

670

34

8 f12.0 x 5

430

80

80

385

1120

Out

0.83

2.02

Out

1.38

Out

1.38

Out

200

159

1.27

670

34

8 f12.0 x 6

430

80

80

447

1200

Out

0.89

1.90

Out

1.48

Out

1.38

Out

200

159

1.27

670

40

8 f12.2 x 9

430

80

80

425

1227

0.89

1.31

Out

1.09

Out

1.26

Max

10

200

154

1.31

670

35

8 f6.0 x 5

430

80

80

385

800

In

0.61

1.58

In

1.28

In

1.33

In

11

200

154

1.31

670

35

8 f6.9 x 5

430

80

80

385

907

In

0.70

1.68

Out

1.31

In

1.42

In

250

124

1.53

488

36

483

0.56

1.30

1.11

1.10

2*

250

124

1.53

488

29

8 f9.5 x 6

393

45

90

721

574

In

0.68

1.24

Out

1.19

Out

1.08

Out

250

124

1.53

488

32

8 9.5 x 6

393

45

90

495

572

In

0.67

1.10

Out

1.12

Out

1.02

In

4*

250

124

1.53

488

38

8 9.5 x 5

465

30

60

403

636

Out

0.73

1.67

In

1.21

Out

1.09

Out

5*

250

124

1.53

488

36

8 9.5 x 5

465

30

60

403

624

Out

0.72

1.67

In

1.21

Out

1.09

Out

250

124

1.53

488

33

8 9.5 x 5

465

30

60

330

615

Out

0.72

1.67

Out

1.18

Out

1.04

Out

300

190

1.29

531

35

825

0.49

1.12

0.94

1.02

8*

300

190

1.29

531

35

8 9.5 x 5

460

50

100

658

1050

In

0.62

1.29

Out

0.98

Out

0.97

In

9*

300

190

1.29

531

35

8 9.5 x 6

460

75

150

1188

1091

In

0.64

1.28

In

1.06

In

1.15

In

10*

350

260

1.10

524

31

1046

0.40

0.88

0.78

0.86

11*

350

260

1.10

524

30

8 12.7 x 5

409

65

130

856

1620

In

0.63

1.24

Out

1.00

Out

1.02

In

12*
350
S 260 1.10 524
34
8 12.7 x 6
409
95
195
1541 1520
In
0.58 1.03
In
0.90 Out 1.08
In
ACI stud layout.
Notes: Vu includes self-weight; Vflex is approximate yield-line capacity from Eq. (14).
Shear reinforcement: In References 1 and 7: deformed studs, 3 heads, so as given for all lines; in Reference 8, slabs R, plain studs, 2.5 heads, Slabs A deformed studs, 2.5
heads, so as given for orthogonal lines, so = 40 mm for diagonal lines; in Reference 9, deformed studs, 3 heads, so as given for all lines; in Reference 11, I-beam slices, flange
breath 102 mm, web breath 4.7 mm. values in the table are equivalent diameters giving the same areas as the actual web sections. so as given for orthogonal lines, so = 40 mm for
diagonal lines; in Reference 10, plain studs with 3.2 heads, so as given for all lines. Birkles Slabs 5 and 6 had 7 perimeters of studs. The outer two, with sr = d, have been ignored.
Aggregate (maximum size and type): In Reference 1, 9.5 mm crushed limestone. In References, 7, 8, 9, and 11, 20 mm gravel. In Reference 10, Slabs 1-614 mm; Slabs
7-1220 mm, type unknown.
Failure modes: P is punching of slabs without shear reinforcement, In = failure inside shear reinforced zone (VR,cs), Out = failure outside shear reinforced zone (VR,out); Max =
inclined compression failure of concrete close to column (VR,max); in Reference 7 and Slab 9 of Reference 10, the concrete soffit around the column crushed and spalled due to
tangential compression, the spalling extended and at failure there was inclined cracking starting at the end of the spalled area.
1 mm = 0.03937 in.; 1 kN = 0.225 kip; 1 MPa = 145 psi.
*

374

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S33

Behavior of Concentrically Loaded Fiber-Reinforced


Polymer Reinforced Concrete Columns with Varying
Reinforcement Types and Ratios
by Hany Tobbi, Ahmed Sabry Farghaly, and Brahim Benmokrane
Fiber-reinforced polymer (FRP) materials have proven their effectiveness as an alternative reinforcement for concrete structures in
severe environmental conditions. Many studies have investigated
the flexural and shear behaviors of FRP-reinforced concrete beams
and slabs. Limited research, however, has gone into investigating
the behavior of internally reinforced FRP concrete columns. This
paper reports the experimental investigation of the compressive
performance of concrete columns reinforced longitudinally with
FRP or steel bars and with FRP as transverse reinforcement.
Twenty concrete columns measuring 350 x 350 x 1400 mm (13.8x
13.8 x 55.1 in.) were constructed and tested under concentric
compressive load. The parametric study included variables such as
transverse reinforcement configuration, material type and spacing,
longitudinal reinforcement ratio, and confining volumetric stiffness. Results showed that FRP bars have contribution as longitudinal reinforcement for concrete columns subjected to concentric
compression and that the combination of FRP transverse reinforcement and steel longitudinal bars offers acceptable strength
and ductility behavior.
Keywords: column; concentric compression; confinement volumetric
stiffness; failure mechanism; fiber-reinforced polymer (FRP); steel;
volumetricratio.

INTRODUCTION
The deterioration of infrastructure owing to corrosion of
steel reinforcement is one of the major challenges facing the
construction industry. The use of reinforcement with fiberreinforced polymer (FRP) composite materials in concrete
structures subjected to severe environmental exposure has
been growing to overcome the common problems caused by
corrosion of steel reinforcement (ACI Committee 440 2007;
Federation Internationale de Bton 2007). Recent advances
in polymer technology have led to the development of the
latest generation FRP reinforcing bars (ACI Committee 440
2007). These corrosion-resistant bars have shown promise
as a way to further protect bridges and public infrastructure
from the devastating effects of corrosion. With standards
ACI 440.6M (ACI Committee 440 2008) and CSA S807
(2010) and bars being produced of the highest quality, FRP
bars are emerging as a realistic and cost-effective reinforcement alternative to traditional steel for concrete structures
under severe environmental conditions. Steel bars cannot,
however, simply be replaced with FRP bars due to various
differences in the mechanical and bond properties of the
two materials (Nanni 1993; ISIS Canada Research Network
2007) and the greater variation of material properties for
FRP reinforcing products.

ACI Structural Journal/March-April 2014

Columns are one of several structural elements that may


be exposed to severe environmental effects. The response of
FRP bars in compression is affected by different modes of
failure (transverse tensile failure, buckled FRP bar, or shear
failure). General acceptance of FRP bars by practitioners
requires that appropriate design guidelines for using FRP bars
in compression members be established. Due to the lack of
experimental data, the current ACI 440.1R (ACICommittee
440 2006) design guidelines do not recommend the use of
FRP bars as longitudinal reinforcement in compression
members, while the CSA S806 (2012) code states that the
compressive contribution of FRP longitudinal reinforcement
is negligible. Moreover, confined concrete behaves differently from unconfined concrete. Several studies clarified
the importance of the lateral reinforcement as a confining
system to the performance in terms of capacity and ultimate
axial strain of axially loaded concrete members (Richart et
al. 1928, 1929; Sheikh and Uzumeri 1980, 1982; Sheikh
1982; Saatcioglu and Razvi 1992; Mander et al. 1988 a,b;
Teng et al. 2002; Harries and Kharel 2003).
RESEARCH SIGNIFICANCE
As the use of FRP reinforcement in concrete structures
grows, appropriate design guidelines for axially concentric loaded concrete columns should be established. In
this regard, laboratory investigations should be conducted
to expand understanding of the compressive behavior of
concrete columns internally reinforced with FRP, particularly given the lack of data about this application, but also to
highlight the most important parameters affecting compressive performance of FRP reinforced columns. This study
investigated concrete columns reinforced longitudinally
with glass FRP (GFRP), carbon FRP (CFRP), and steel bars
and GFRP and CFRP transverse reinforcement subjected to
concentric loading. The experimental study yielded a better
understanding of the mechanical behavior of FRP reinforced
columns. The set of specimens is presented to enrich the
literature regarding the use of FRP as internal reinforcement
for compressive members in preparation for developing
design models.

ACI Structural Journal, V. 111, No. 2, March-April 2014.


MS No. S-2012-123.R1, doi:10.14359.51686528, was received November 9, 2012,
and reviewed under Institute publication policies. Copyright 2014, American
Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published ten months from this journals date if the
discussion is received within four months of the papers print publication.

375

Fig. 1(a) and (b) C-shaped; and (c) closed transverse reinforcement.

Fig. 2Transverse reinforcement configuration: (a) 1; (b) 2; and (c) 3.


EXPERIMENTAL PROGRAM
The experimental study comprised 20 concrete columns
measuring 350 x 350 x 1400 mm (13.8 x 13.8 x 55.1 in.)
subjected to concentric compressive loading. These dimensions are representative of the columns commonly found
in concrete structures. One column was kept un-reinforced
(plain concrete) while the remaining 19 were internally reinforced with FRP and steel according to different parameters.
All used transverse reinforcements were GFRP or CFRP,
while the longitudinal reinforcement was GFRP, CFRP, or
steel bars.
Studied parameters included the shape of transverse reinforcement (C-shaped parts assembly or closed ties, as shown
in Fig. 1), longitudinal reinforcement ratio, longitudinal
reinforcement material (GFRP, CFRP, or steel), FRP-transverse reinforcement material (GFRP or CFRP), the diameter
of the transverse reinforcement (No. 9.5 and 12.7 mm [No. 3
and No. 4]), transverse reinforcement spacing, and confining
volumetric stiffness. Volumetric ratio is an important parameter for confinement efficiency (Sheikh and Uzumeri 1980,
1982; Sheikh 1982; Saatcioglu and Razvi 1992; Mander et
al. 1988 a,b; Watson et al. 1994; Cusson and Paultre 1994;
Saatcioglu et al. 1995) for passive confinement with internal
reinforcement. Volumetric ratio v is defined as the ratio
of the volume of transverse confining reinforcement to the
volume of confined concrete core. FRP mechanical proprieties vary depending on fiber material and fiber content. For
consistency, therefore, volumetric ratio should be multiplied by the modulus of elasticity of FRP confining material
(vEf), which is the so-called confining volumetric stiffness. The same confining volumetric stiffness (v Ef) can
be obtained by changing at least two of the following parameters: transverse reinforcements configuration, spacing,
material, or diameter.
The GFRP transverse reinforcement diameter was
12.7mm (No. 4). Two transverse reinforcement shapes
376

were used: the first is assembled from C-shaped parts as


shown in Fig. 1(a) and (b), and the second is closed form,
as shown in Fig. 1(c). The first shape was the first available product for bent bars; therefore, it has been chosen to
build up the transverse reinforcement of the FRP reinforced
columns. The second shape, however, has the benefit of
eliminating the discontinuity of the C-shaped and reducing
the assembly labor time. Generally speaking, both can be
used in construction. The closed transverse reinforcements
were cut from continuous square spirals, with an overlap
equal to one side length. For each transverse reinforcement
shape, three configurations labeled 1, 2, and 3, as shown
in Fig. 2 (a), (b), and (c), were investigated in Tobbi et al.
(2012), which revealed that the effect of Configuration 2 is
not different from Configuration 3; therefore, only Configurations 1 and 3 will be addressed in this study. In the case of
C-shaped transverse reinforcements in Configuration 1, the
cross hairpins in two consecutive layers were staggered to
preserve overall symmetry. In the case of the closed transverse reinforcements in Configuration 1, cross hairpins were
staggered C-shaped legs. In Configuration 3, the cross ties
were closed rectangular ties with one side overlap. All CFRP
transverse reinforcements were closed with different diameters (No. 9.5 and 12.7 mm [No. 3 and No. 4]) to investigate
their effect on the compressive performance of the columns.
The test matrix is listed in Table 1.
To fulfill the objectives of the parametric study, eight
columns were exclusively reinforced with FRP: seven with
GFRP, both longitudinal bars and transverse reinforcements
(Configurations 1 and 3), and one with CFRP, also with both
longitudinal bars and transverse reinforcements. Eleven
columns were reinforced longitudinally with steel bars and
FRP transverse reinforcements (Configurations 1 and 3):
four reinforced transversally with GFRP, and seven with
CFRP. One column was unreinforced (plain concrete).

ACI Structural Journal/March-April 2014

Table 1Test matrix

Columns reinforced with steel longitudinal bars


and FRP tie and cross-ties

Columns entirely reinforced with


FRP bars

Group

fc, Pa
(ksi)

Longitudinal reinforcement

Transverse reinforcement

Ties
spacing,
mm (in.)

v x Ef,
GPa

Designation

Material

Designation

Material

Ties
configuration

P-0-00-0

G-1c-120-1.9

8 No. 6 (19 mm)

120 (4.7)

0.96

12 No. 5 (15.9 mm)

120 (4.7)

1.28

12 No. 5 (15.9 mm)

80 (3.2)

1.92

Specimen

G-3c-120-1.9

33 (4.8)

G-3c-80-1.9
G-1-120-1.9

8 No. 6 (19 mm)

120 (4.7)

1.18

G-3-120-1.9

12 No. 5 (15.9 mm)

120 (4.7)

1.58

G-1-120-1.0

4 No. 4 + 4 No. 5
(12.7; 15.9 mm)

120 (4.7)

1.18

G-1-120-0.8

35 (5.1)

2 x 8 No. 4 (12.7 mm)*

G-1-120-1.0S

4 M15 + 4 M10

C-1-120-1.0S

4 M15 + 4 M10

No. 3C-1-67-1.0S

4 M15 + 4 M10

No. 3C-3-80-1.0S

12 M10

C-1-80-1.0S

4 M15 + 4 M10

C-1-60-1.0S

4 M15 + 4 M10

C-3-80-1.0S

27 (3.9)

No. 4 (12.7 mm)

GFRP

8 No. 4 (12.7 mm)

No. 3C-1-67-1.6

C-3-120-1.0S

GFRP

CFRP

No. 3 (9.5 mm)


No. 4 (12.7 mm)

No. 3 (9.5 mm)

Steel

G-1-80-1.0S

4 M15 + 4 M10

G-3-120-1.0S

12 M10

G-3-80-1.0S

12 M10

120 (4.7)

1.18

67 (2.6)

2.87

GFRP

120 (4.7)

1.18

CFRP

120 (4.7)

3.04

67 (2.6)

2.87

80 (3.2)

3.20

80 (3.2)

4.56

60 (2.4)

6.08

120 (3.2)

4.05

80 (3.2)

6.08

80 (3.2)

1.78

120 (4.7)

1.58

80 (3.2)

2.37

CFRP

CFRP

12 M10
12 M10

1
CFRP

No. 4 (12.7 mm)

GFRP

Bundled bars.

M15 placed at corners.

Notes: P is plain concrete; G is GFRP; C is CFRP; (1; 3) = Stirrup configuration; c is C shaped legs assembly; (120; 80; 67; 60) is stirrup spacing, mm; (0.8; 1.0; 1.9) is longitudinal reinforcement ratio; S is steel longitudinal bars.

Materials
The columns were cast vertically using normalweight
ready mixed concrete with a target 28-day concrete compressive strength of 30 MPa (4.4 ksi). The columns were cured
for 7 days, after which the specimens were left in the laboratory at ambient temperature for at least three more weeks
before testing. The concrete compressive strength used for
analysis was based on the average values of tests performed
on at least five 150 x 300 mm (6 x 12 in.) cylinders for each
concrete batch under displacement control standard rate of
0.01 mm/s (3.9 104 in./s) (Table 1). Grade 60 steel reinforcing bars were used as longitudinal reinforcement for
specific specimens. Table 2 provided the tensile properties
of Grade 60 steel bars.
The longitudinal reinforcement for the exclusively FRPreinforced columns was (1) No. 12.7 mm (No. 4) straight
CFRP bars, and (2) No. 15.9 mm (No. 5) and 19.1 mm (No.6)
GFRP straight bars. The tensile properties of longitudinal FRP
and steel bars were determined by performing the B.2 test
method according to ACI 440.3R (ACI Committee 440 2004)
as reported in Table 2. Bent bars of 12.7 mm (No. 4) GFRP
and 9.5 mm (No. 3) and 12.7 mm (No. 4) CFRP were used
as transverse reinforcements. The ultimate tensile strength
ffu and modulus of elasticity Ef for the straight portions of
the transverse reinforcements were determined according
ACI Structural Journal/March-April 2014

Table 2Tensile properties of FRP and steel


longitudinal reinforcement
Af, mm2
(in.2)

Ef, GPa
(ksi)

ffu, MPa
(ksi)

12.7 (0.5)

127
(0.19)

46.3
(6715)

1040
(151)

2.25

No. 5
GFRP

15.9 (0.62)

199
(0.31)

48.2
(6990)

751
(109)

1.56

No. 6
GFRP

19.1 (0.75)

284
(0.44)

47.6
(6904)

728
(106)

1.53

No. 4
CFRP

12.7 (0.5)

127
(0.19)

137
(19,870)

1902
(276)

1.38

Steel M10

11.3 (0.44)

100
(0.15)

200
(29,000)

fy = 450
(65)

y = 0.2

Steel M15

16.0 (0.62)

200
(0.31)

200
(29,000)

fy = 460
(66)

y = 0.2

Bar type

db, mm (in.)

No. 4
GFRP

f, %

Notes: db is bar diameter; Af is cross-sectional area of bar; Ef is modulus of elasticity


of bar; ffu is ultimate tensile strength of bar; f is ultimate strain of bar.

to the ACI440.3R B.2 test method (ACI Committee 440


2004). The ultimate bent strength ffu,bend, however, was determined using the B.5 test method according to ACI 440.3R
(ACICommittee 440 2004). Table 3 provides the measured
377

Fig. 3(a) Loading machine; and (b) instrumentation.


Table 3Tensile properties of FRP transverse
reinforcement
Straight portion
Ef, GPa (ksi)

f, %

ffu,bend,
MPa (ksi)

ffu,bend/
ffu

No. 3 CFRP 1327 (192) 126 (18,275)

1.05

614 (89)

0.46

No. 4 CFRP 1372 (198) 133 (19,290)

1.03

700 (101)

0.51

No. 4 GFRP

962 (139)

52 (7542)

1.85

500 (72)

0.52

C-shaped
No. 4 GFRP

640 (92)

44 (6382)

1.45

400 (58)

0.62

Bar type

ffu, MPa
(ksi)

Bend portion

Notes: Ef is modulus of elasticity of bar; ffu is ultimate tensile strength of bar; f is


ultimate strain of bar; ffu,bend is ultimate tensile strength of bar bend.

Fig. 4Stress-strain relationship for both plain concrete


cylinder and P-0-00-0 column. (Note: 1 MPa = 0.145 ksi.)
tensile strength and modulus of elasticity for the straight and
bent portions. The GFRP and CFRP longitudinal bars were
pultruded products (Pultrall, Inc. 2009). Transverse reinforcements were fabricated with the bend process (Pultrall,
Inc. 2009). Sand coating was used on the surface of the
longitudinal and transverse FRP bars to improve the bond to
concrete, as in standard industry practice.
Instrumentation and testing procedures
Reinforcement strain was measured with electrical strain
gauges adhered to the bars at midheight of the column. A
set of ties in each specimen was instrumented with strain
gauges placed at the middle and in the corner of the outer tie
and the cross hairpins. The test specimens were loaded by a
rigid MTS high force load frame (Fig. 3(a)) with a maximum
compressive capacity of 11,400 kN (2,560,000lbf) having
the load controlled up to 2200 kN (495,000 lbf) with a rate
of 2.5 kN/s (562 lb/s). Thereafter, displacement control
was used to apply the load until failure with the rate of
0.002mm/s (7.87 105 in./s). The axial displacement of the
378

column specimens was recorded using four linear variable


differential transducers (LVDTs) located at the midheight
of each side of the specimens, as shown in Fig.3(b). The
top and bottom ends of the specimens were capped with a
thin layer of high-strength mortar to ensure that the bearing
surfaces were parallel and the load was distributed uniformly
during testing. To ensure the failure would occur in the
instrumented region, the ends of each specimen were further
confined with bolted steel plates made from 13 mm (0.5 in.)
thick steel plates (Fig. 3(b)).
EXPERIMENTAL RESULTS AND DISCUSSION
Overall behavior
The unreinforced plain concrete column (P-0-00-0) was
the first to be tested. The stress-strain behavior until peak
was similar to the concrete cylinder, while peak stress was
slightly lower as shown in Fig. 4. Post-peak behavior was
completely different. Concrete cylinders exhibit a considerable softening branch meaning gradual damage, while
P-0-00-0 failure was brittle; total loss of sustained load
occurred just after reaching peak stress. This difference in
post-peak behavior was due to the higher energy accumuACI Structural Journal/March-April 2014

Fig. 5Total cross-section-based stress-strain curves for all tested columns.


lated by the P-0-00-0 column compared to the small concrete
cylinder. The post-peak behavior of plain concrete subjected
to axial load is an important parameter for determination of
cover contribution of the reinforced columns subjected to
the same type of load. The stress-strain curve for confined
concrete core of all reinforced columns is then obtained.
Considering the total cross-sectional area for concrete
columns with FRP transverse reinforcements (Fig. 5), the
stress-strain curve was divided into three phases. The first
phase corresponds to the behavior until the peak stress,
which was similar to plain concrete column, implying that
transverse reinforcement had no effect on this phase. The
concrete cover was visually free of cracks (Fig. 6(a)), yet
the peak stress varied depending on the longitudinal reinforcement material and ratio (Fig. 5(b) and (c)). The second
phase was very short, and was characterized by a rapid drop
in bearing capacity. This started once the peak stress was
reached, and finished with passive-confinement activation
(strain increase in the transverse reinforcements). In this
phase, cracks began growing in the concrete cover, as shown
in Fig. 6(b), leading to gradual spalling that reduced the
load-resisting cross-sectional area and resulted in strength
degradation. The third phase is characterized by activation
ACI Structural Journal/March-April 2014

of transverse reinforcing associated with complete spalling


of concrete cover, and ended with total failure of the column
(Fig. 6(c)). This phase is clearly governed by transverse
reinforcement and longitudinal bar material (Fig. 5). The
following sections contain detailed discussion.
Confined concrete core behavior
Analyzing compressive behavior of internally confined
concrete columns considering the total cross-sectional area
from the starting of the elastic phase until failure is not
accurate because the stress calculation does not take into
account the degradation of concrete cover contribution
after cracking. Nevertheless, when concrete cover is spalled
off, the confined concrete core remained uncracked until a
certain level, depending on the confinement effect. Therefore, studying the effect of confinement accurately requires
considering the confined concrete core only. Nevertheless,
the concrete cover strength should be subtracted from the
total applied load based on the behavior of the plain concrete
column, as shown in Fig. 4. The reduced load, divided by the
concrete core area delimited by the centerlines of the outer
transverse reinforcements, presents the columns actual
stress behavior. Figure 7 shows the curves representing the
379

Fig. 6Cracking appearance of test specimens at different loading stages.

Fig. 7Effect of concrete cover (G-3c-80-1.9).


axial stress sustained by the concrete with respect to: 1) the
total load divided by the total concrete area (Path 0-A-B-C);
and 2) the total load divided by the confined concrete area
delineated by the centerline of the outer transverse reinforcements (Path 0-A-B-C). The actual response of the concrete
column, represented by the bold curve (Path0-A-B-C), is
expected to be a combination of the two calculated curves.
The response of the concrete column (bold curve) coincides
with the ascending part of the lower curve (total concrete
area) up to Point A, which corresponds to the spalling of
the concrete cover. When the concrete cover no longer
contributed to the axial strength, the response of the concrete
column coincided with the part of the higher curve (confined
concrete area) that follows Point B, when the concrete core
began to gain strength due to confinement by the transverse reinforcement. The transition between Points A and
B of the response of the concrete column was quantified by
subtracting the contribution of the concrete cover (which
decreased with increasing axial deformation) based on the
stress-strain response of the plain concrete (Fig. 4). Point C
corresponds to the ultimate strength of the tested columns.

380

Strength and failure mode


Different failure modes were observed based on reinforcement layout. The failure mode of exclusively FRPreinforced column followed this progression: 1) crushing or
buckling of the FRP longitudinal bar; and 2) transverse reinforcement rupture. Excessive buckling of the longitudinal
steel bars was the failure mode of columns reinforced longitudinally with steel bars. The failure modes were governed
by the shape, configuration, and diameter of the transverse
reinforcements, as well as longitudinal bar material.
The failure of all longitudinally and transversally FRPreinforced columns was due to longitudinal bar crushing or
buckling, as shown in Fig. 8. In general, the columns with
C-shaped GFRP transverse reinforcements experienced
brittle failure. The failure of Column G-1c-120-1.9, which
had the lowest confinement volumetric stiffness, was explosive. Column G-3c-120-1.9, which had higher confinement
volumetric stiffness, showed failure starting with the longitudinal GFRP bars crushing, followed by total concrete
crushing. In both columns, the slipping of the outer C-shaped
transverse reinforcements at the splice location occurred due
to concrete core expansion pressure, leading to degradation
of sustained load until crushing of the longitudinal bars,
followed instantaneously by concrete core crushing. Moreover, inclined shear sliding surfaces were observed, leading
to a separation of the concrete core into two wedges, causing
a sudden drop in axial strength. The failure of Column
G-3c-80-1.9 was different: no transverse reinforcement slippage was observed. This might be attributed to the smaller tie
spacing, which allowed the column to fail progressively in
successive crushing of all longitudinal GFRP bars followed
by concrete core crushing. The FRP reinforced columns with
closed transverse reinforcements failed progressively due to
successive crushing of the longitudinal bars before concrete
core crushing. No transverse reinforcements rupture was
observed, except in Column G-3-120-1.9, which was the
only FRP-reinforced column with Configuration 3 transverse reinforcements.
Reinforcing the columns longitudinally with steel bars
instead of FRP bars changed the failure mode. The longiACI Structural Journal/March-April 2014

Fig. 8Failure mode of columns reinforced longitudinally and transversely with FRP.

Fig. 9Failure mode of columns reinforced longitudinally with steel and transversely with FRP.
tudinal steel bars consistently buckled (Fig. 9(a) and (b)).
In addition, cross-tie rupture was observed for columns
with Configuration 3, (Fig. 9(c)), while failure was due to
excessive bars buckling and substantial decrease in bearing
capacity in columns with Configuration 1 transverse reinforcements. Moreover, excessive buckling of the longitudinal bars in columns induced openings in the GFRP transverse reinforcements, as shown in Fig. 9(b). Opening (albeit
minor) was also observed with transverse CFRP reinforcement at 80 and 60 mm (3.2 and 2.4 in.) spacing.
Failure due to transverse reinforcement rupture was
experienced in columns transversely reinforced with CFRP
No.9.5 mm (No. 3) with both Configurations 1 and 3
(Fig.10). Even Column No. 3C-1-67-1.6 experienced transverse reinforcement rupture after the longitudinal CFRP
bars experienced crushing.
Parametric investigation
Parametric investigation was carried out to study the
strength mechanism and performance based on stress-strain
relationship for the tested columns. The investigated parameters included transverse reinforcement shape, material,
spacing and diameter (No. 9.5 and 12.7 mm [No. 3 and
No.4]), longitudinal reinforcement ratio, longitudinal reinforcement material and confining volumetric stiffness.
To compare the strength behavior of columns cast from
different concrete batches, the stress values c were normalized to the cylinder compressive strength fc of the batch.
Therefore, the stress response c along the test for each
column was divided by the concrete compressive cylinder
ACI Structural Journal/March-April 2014

Fig. 10Transverse reinforcement rupture for No. 3 CFRP


laterally reinforced columns.
strength fc. Columns were cast in three different groups.
While the targeted concrete strength was 30 MPa (4.4 ksi),
the actual concrete strength for the three groups was 33,
35, and 27 MPa (4.8, 5.1, and 3.9 ksi). Before testing the
columns, actual cross-sectional area was measured to calculate the precise stress values.
Effect of transverse reinforcement shape
(C-shaped versus closed)
Four columns were studied to investigate the effect
of transverse reinforcement shape. Two columns, G-1c120-1.9 and G-3c-120-1.9, were transversely reinforced
with C-shaped GFRP No.12.7 mm (No. 4). The other two
columns, G-1-120-1.9 and G-3-120-1.9, were transversely
381

Fig. 11C-shaped Vs closed transverse reinforcement


normalized stress-strain relationship.
reinforced with closed GFRP No.12.7 mm (No. 4). The four
columns had identical longitudinal reinforcement.
Figure 11 shows the normalized stress-strain response.
The confined concrete strength gain (fcc/fc) is quite similar
for the columns transversely reinforced with either transverse reinforcement type. Moreover, Configuration 3
showed a higher strength gain than Configuration 1. Nevertheless, a significant difference was noted based on transverse reinforcement type. In the columns with C-shaped
transverse reinforcement (G-1c-120-1.9 and G-3c-120-1.9),
the strength decreased due to leg slippage after reaching
normalized confined concrete strength (fcc/fc). Therefore,
the normalized ultimate strength (fcu/fc) corresponding to
ultimate axial strain cu was lower than the normalized peak
strength (fcc/fc). Meanwhile, in the columns with closed
transverse reinforcement (G-1-120-1.9 and G-3-120-1.9),
no descending branch was observed. The failure of columns
occurred at the maximum normalized confined concrete
strength. In other words, fcc and fcu had the same value.
Therefore, it can be deduced that closed transverse reinforcement yield more efficient confinement than C-shaped
transverse reinforcement because of the material continuity
that eliminates slippage, increasing the lateral confinement
pressure rather than confinement degradation.
Effect of longitudinal reinforcement
The longitudinal reinforcement effect was more
pronounced in the stress-strain curves based on total
cross-sectional area because the contribution of the longitudinal reinforcement was more effective on the pre-peak
phase before the activation of the confinement effect.
Figure 12 shows that increasing the FRP longitudinalreinforcement ratio from 0.8 to 1.0 then 1.9 increased the
load at the peak before activation of confinement. The stressstrain curves for these three columns had the same trend. The
three columns failed when the longitudinal bars buckled at
nearly the same axial strain. Figure 12 also shows the effect
of longitudinal reinforcement material: Columns G-1-1201.0 and G-1-120-1.0S had GFRP and steel longitudinal reinforcement, respectively, with the same reinforcement ratio
382

Fig. 12Effect of longitudinal reinforcement on compressive behavior of columns.


(1.0%). Two main differences were observed: the peak stress
for the steel longitudinally reinforced column was higher than
that of GFRP-reinforced column, and the GFRP-reinforced
column showed stabilization of the load-carrying capacity,
represented by nearly horizontal plateau until failure, at the
post-peak phase, while the load-carrying capacity of steel
longitudinally reinforced column decreased after reaching
the peak load. The difference in post-peak behavior is due to
longitudinal reinforcement material. Indeed, the load carried
by steel bars after yielding remained constant, while the load
increased with axial strain with the elastic GFRP bars. This
behavior is more pronounced in Column G-3c-80-1.9, which
was the only column that exhibited stiffening behavior after
cracking of the cover and a reduction in the load drop that
ended with a second peak load higher than the first one, as
shown in Fig. 5(a).
The ultimate axial strain of the columns reinforced longitudinally with FRP is lower than those reinforced with steel.
Columns longitudinally reinforced with FRP, however,
reached axial compressive strains of 0.011 and 0.018 in
G-1-120-1.9 and G-3c-80-1.9, respectively, which showed
that, under good confinement conditions, the FRP bars were
able to reach high compressive strains.
The nominal compressive capacity of the FRP reinforced columns at peak, considering the gross cross-sectional area Pn, was defined as the sum of the forces carried
by the concrete and the longitudinal reinforcement. Based
on the elastic theory, the contribution of FRP longitudinal
reinforcement bars in compression at peak was calculated
according to the material proprieties given in Table 2. The
authors proposed an equation to calculate the nominal
compressive capacity of the longitudinally and transversally
FRP-reinforced columns, which given as follows
Pn = 0.85 fc (Ag Afrp) + co Efrp Afrp (SI) (1)
where Ag is gross cross-sectional area of the column; Afrp
is cross-sectional area of FRP longitudinal reinforcement;
fc is concrete compressive strength; co is concrete strain
at peak stress; and Efrp is modulus of elasticity of FRP
longitudinalreinforcement.
ACI Structural Journal/March-April 2014

Table 4Prediction of nominal compressive capacity


fc, MPa (ksi)

Ac, mm2 (in.2)

PExp, kN (lbf)

Pn, kN (lbf)

Pn/PExp

Pc, kN (lbf)

PLongi, kN (lbf)

G-1-120-0.8

127,132 (197)

3900 (876,755)

3899 (876,530)

1.00

3782 (850,227)

117 (26,303)

G-1-120-1.0

128,803 (200)

4212 (946,895)

3995 (898,111)

0.95

3832 (861,468)

163 (36,644)

Column

G-1-120-1.9

125,528 (194)

4297 (966,004)

4048 (910,027)

0.94

3734 (839,437)

314 (70,590)

G-1-120-1.0S

124,642 (193)

4272 (960,384)

4260 (957,686)

1.00

3708 (833,592)

552 (124,094)

127,688 (198)

5159 (1,159,789)

4714 (1,059,749)

0.91

3799 (854,049)

919 (206,599)

No. 3C-1-67-1.0S

125,340 (194)

4660 (1,047,610)

4281 (962,407)

0.92

3729 (838,312)

552 (124,094)

G-3-120-1.9

125,734 (195)

4615 (1,037,493)

4086 (918,569)

0.89

3741 (841,010)

345 (77,559)

C-1-120-1.0S

128,922 (200)

4584 (1,030,524)

4387 (986,237)

0.96

3835 (862,142)

552 (124,094)

No. 3C-1-67-1.6

35 (5.07)

Notes: Ac = (Ag Afrp); Ag is actual total cross-sectional area of considered column.

Fig. 13Transverse reinforcement layout effect on confined concrete stress-strain response.


Whereas for columns with steel longitudinal reinforcement, the nominal compressive capacity at the peak is given
as follows according to ACI 318 (ACI Committee 318 2008)
Pn = 0.85 fc (Ag As) + fy As (2)
where As is the cross-sectional area of longitudinal steel
reinforcement; and fy is the yielding strength of steel
reinforcement.
The strength of full-scale plain-concrete columns tested
under concentric compression load is generally lower than
the concrete compressive strength measured on standard
150x 300 mm (6 x 12 in.) cylinders. The 0.85 reduction
factor suggested by ACI 318 (ACI Committee 318 2008) is
mainly attributed to the differences between the reinforced
concrete column and the concrete cylinder regarding concrete
compressive strength, size, and shape of the element.
Table 4 compares the predicted nominal compressive
capacity at the peak to the experimental results for columns
reinforced transversely with FRP and reinforced longitudiACI Structural Journal/March-April 2014

nally with FRP or steel bars according to Eq. (2) and (1),
respectively. The results showed that nominal compressive
capacity predictions Pn were conservative and very close to
experimental results, with Pn/PExp ratios varying from 0.89
to 1.00. It is important to note that when Pn differs from
one specimen to another, the load carried by the concrete
remained similar in all the columns. In other words, the
difference in Pn is primarily due to the longitudinal reinforcement ratio and material, not to concrete strength.
Effect of transverse reinforcement
The transverse reinforcement restrains the expansion of
the concrete core in the column subjected to compressive
load and delays its failure. Accordingly, the compressive
performance of concrete columns depends strongly on the
transverse reinforcement efficiency. Figure 13 shows the
stress-strain curves of the columns reinforced with different
transverse reinforcement layouts to investigate the effect of
transverse reinforcement configuration, spacing, material,
diameter, and confining volumetric stiffness. The stress383

strain curves based on the confined concrete core showed


that, regardless the transverse reinforcement configuration or material, reducing spacing increases the nominal
confined-concrete strength (fcc/fc) and changes the behavior
after this point. With 120 mm (4.7 in.) spacing, a descending
branch followed the peak stress, while for 80 mm (3.2 in.)
spacing, the stress stabilized at a nearly horizontal plateau;
however, stress increased with the 60 mm (2.4 in.) spacing.
Configuration 3 proved to be more efficient than Configuration 1, offering higher nominal confined concrete
strength (fcc/fc) and enhancing the peak stress. In the case of
120mm (4.7 in.) spacing, the slope of the descending branch
following the peak stress was less steep for Configuration
3 than Configuration 1, resulting in higher ultimate strain
(almost double) as comparing C-3-120-1.0S and G-3-1201.0S with C-1-120-1.0S and G-1-120-1.0S, respectively
(0.039 and 0.041 versus 0.019 and 0.016, respectively).
For columns with 80 mm (3.2 in.) spacing, the stabilization
plateau was longer in Configuration 3 than Configuration1.
The ultimate strain increased for the CFRP transversely
reinforced columns from 0.026 to 0.034, corresponding to
C-1-80-1.0S and C-3-80-1.0S, respectively. The increase of
ultimate strain in the GFRP transversely reinforced columns,
however, was more than the double comparing G-1-80-1.0S
and G-3-80-1.0S (0.021 and 0.052, respectively). It is clearly
shown that the transverse reinforcement spacing and configuration determined the effectively confined concrete volume,
which increased with closer transverse reinforcement and a
better distribution of longitudinal bars around the column
concrete core. The larger the effectively confined concrete
volume, the higher the confinement efficiency. In addition,
transverse reinforcement spacing controlled the buckling of
the longitudinal bars by reducing their slenderness ratio.
The effect of transverse reinforcement material was
related to columns mode of failure, which is dependent on
the configuration. Figure 13(b), (c), and (d) showed that, in
the case of columns with Configuration 1 (C-1-120-1.0S,
G-1-120-1.0S, C-1-80-1.0S, and G-1-80-1.0S) that failed
because of excessive longitudinal bars buckling, CFRP
transverse reinforcement enabled the columns to attain
higher nominal confined concrete strength (fcc/fc) than those
reinforced with GFRP transverse reinforcement. The stressstrain curves of both materials, however, followed the same
trend. The modulus of elasticity was determinant: given
the same layout, the CFRP transverse reinforcements were
stiffer than the GFRP ones. The stiffer transverse reinforcement tended to open less, thereby better limiting the buckling of longitudinal bars. Two different spacing related cases
were observed in columns with Configuration 3. The first
relates to 120 mm (4.7 in.) spacing, which was wide enough
to enable significant bar buckling and the development of
a localized plastic hinge, as illustrated in Fig. 9(a) and (b).
The columns reinforced transversely with CFRP and GFRP
behaved identically. In the second case, the stirrups spacing
of 80 mm (3.2 in.) prevented excessive bar buckling and
the development of a localized plastic hinge. In this case,
the CFRP stirrups allowed the column to achieve higher
nominal confined concrete strength than GFRP stirrups (1.7
versus 1.6). Conversely, the larger ultimate elongation of
384

GFRP (Table 3) allowed Column G-3-80-1.0S to reach a


higher strain than C-3-80-1.0S (0.052 versus 0.034).
Regarding transverse reinforcement diameter, C-3-80-1.0S
outperformed 3C-3-80-1.0S in terms of nominal confined
concrete strength (1.7 versus 1.4) and ultimate strain (0.034
versus 0.015). In 3C-3-80-1.0S, however, the longitudinal
bars buckled outside the strain measurementzone.
Analyzing the results shown in Fig. 13 illustrates the
effect of the confining volumetric stiffness. Given the same
transverse reinforcement layout v, the GFRP volumetric
stiffness (v Ef) was far less than that of the CFRP, yet the
performances were close, which indicates that the configuration and spacing are more important parameters than
modulus of elasticity. Given the same confining volumetric
stiffness and transverse reinforcement material, Configuration 3 performed better than Configuration 1 in terms of
ultimate strain (C-1-60-1.0S and C-3-80-1.0S, as shown in
Fig. 13).
CONCLUSIONS
Failure mechanisms of axially loaded concrete columns
reinforced longitudinally with FRP or steel bars and with
FRP transverse reinforcement involving different layouts
were investigated. Based on the analytical results, the
following remarks can be made:
1. The confinement efficiency of closed FRP transverse
reinforcements cut from continuous square spiral is higher
than C-shaped type transverse reinforcements.
2. The ultimate axial strain of columns reinforced longitudinally with FRP is almost 30% lower than those reinforced
with the same volume of steel.
3. The ultimate axial compressive strain for columns reinforced longitudinally and transversally with FRP can reach
a value on the same order of magnitude as the FRP ultimate
tensile strain of the longitudinal bars under good confinement conditions.
4. The contribution of FRP longitudinal reinforcement
in concrete columns subjected to axial concentric loading
should not be neglected. A proposed equation based on
elastic theory yields good predictions compared with laboratory test data.
5. FRP transverse reinforcement configuration and
spacing are the most important parameters (compared with
confinement provided by concrete cover) affecting confining
efficiency in internally reinforced concrete columns under
axial loading.
6. In the case of large spacing with low volumetric ratio,
CFRP transverse reinforcement performed significantly
better than GFRP. Increasing the volumetric ratio while
reducing spacing will eliminate the effect of material stiffness. In such cases, the GFRP transverse reinforcements are
more cost effective.
7. Columns internally reinforced with a combination of
steel longitudinal bars and FRP transverse reinforcements
exhibit good gains in terms of compressive strength and
ultimate axial strain. Nonetheless, the use of FRP transverse
reinforcement should still improve corrosion resistance of
a column by adding an extra 10 mm (0.4 in.) of cover to
thesteel.
ACI Structural Journal/March-April 2014

8. The presented study showed the applicability of exclusively reinforcing the columns with FRP and subjected to
concentric load. Further research elaboration is necessary
to investigate the behavior of FRP reinforced columns
loaded laterally or subjected to load combination (axially
andlaterally).
AUTHOR BIOS

Hany Tobbi is a Doctoral Candidate in the Department of Civil Engineering at the University of Sherbrooke, Sherbrooke, QC, Canada. He
received his BSc from the University of Mentouri, Constantine, Algeria, and
his MSc from the University of Claude Bernard, Lyon, France. His research
interests include structural analysis, design, and testing of concrete structures reinforced with fiber-reinforced polymers.
Ahmed Sabry Farghaly is a Postdoctoral Fellow in the Department of
Civil Engineering at the University of Sherbrooke, and Associate Professor
in the Department of Civil Engineering, Assiut University, Assiut, Egypt.
His research interests include nonlinear analysis of reinforced concrete
structures, and behavior of structural concrete reinforced with fiber-reinforced polymers.
Brahim Benmokrane, FACI, is an NSERC Research Chair in FRP Reinforcement for Concrete Infrastructures and Tier-1 Canada Research Chair
Professor in Advanced Composite Materials for Civil Structures in the
Department of Civil Engineering at the University of Sherbrooke. He is a
member of ACI Committee 440, Fiber-Reinforced Polymer Reinforcement.

ACKNOWLEDGMENTS

The authors would like to express their special thanks and gratitude to the
Natural Science and Engineering Research Council of Canada (NSERC),
the Fonds qubcois de la recherche sur la nature et les technologies
(FQRNT), the Canadian Foundation for Innovation (FCI), and the technical
staff of the structural lab in the Department of Civil Engineering at the
University of Sherbrooke.

REFERENCES

ACI Committee 318, 2008, Building Code Requirements for Structural


Concrete (ACI 318-08) and Commentary, American Concrete Institute,
Farmington Hills, MI, 473 pp.
ACI Committee 440, 2004, Guide Test Methods for Fiber-Reinforced
Polymers (FRPs) for Reinforcing or Strengthening Concrete Structures
(ACI440.3R-04), American Concrete Institute, Farmington Hills, MI, 40pp.
ACI Committee 440, 2006, Guide for the Design and Construction of
Structural Concrete Reinforced with FRP Bars (ACI 440.1R-06), American Concrete Institute, Farmington Hills, MI, 44 pp.
ACI Committee 440, 2007, Report on Fiber-Reinforced Polymer (FRP)
Reinforcement Concrete Structures (ACI 440R-07), American Concrete
Institute, Farmington Hills, MI, 100 pp.
ACI Committee 440, 2008, Specification for Carbon and Glass
Fiber-Reinforced Polymer Bar Materials for Concrete Reinforcement (ACI
440.6M-08), American Concrete Institute, Farmington Hills, MI, 6 pp.
CSA S806, 2012, Design and Construction of Building Components
with Fiber-Reinforced Polymers, Canadian Standards Association, Mississauga, ON, Canada, 177 pp.

ACI Structural Journal/March-April 2014

CSA S807, 2010, Specification for Fibre-Reinforced Polymers, Canadian Standards Association, Mississauga, ON, Canada, 44 pp.
Cusson, D., and Paultre, P., 1994, High Strength Concrete Columns
Confined by Rectangular Ties, Journal of Structural Engineering, ASCE,
V. 120, No. 3, Mar., pp. 783-804.
Federation Internationale de Bton (FIB), 2007, FRP Reinforcement in
RC Structures, Task Group 9.3, Lausanne, Switzerland, 157 pp.
Harries, K. A., and Kharel, G., 2003, Experimental Investigation of the
Behavior of Variably Confined Concrete, Cement and Concrete Research,
V. 33, pp. 873-880.
ISIS Canada Research Network, 2007, Reinforcing Concrete Structures
with Fibre Reinforced Polymers, ISIS Design Manual No. 3, ISIS Canada
Research Network, 151 pp.
Mander, J. B.; Preistley, M. J. N.; and Park, R., 1988a, Theoretical
Stress-Strain Model for Confined Concrete, Journal of Structural Engineering, ASCE, V. 114, No. 8, pp. 1804-1826.
Mander, J. B.; Preistley, M. J. N.; and Park, R., 1988b, Observed
Stress-Strain Behaviour of Confined Concrete, Journal of Structural Engineering, ASCE, V. 114, No. 8, pp. 1827-1849.
Nanni, A., 1993, Flexural Behavior and Design of RC Members Using
FRP Reinforcement, Journal of Structural Engineering, ASCE, V. 119,
No. 11, pp. 3344-3359.
Pultrall, Inc., 2009, V-ROD Composite Reinforcing Rods Technical Data
Sheet, Thetford Mines, Canada, www.pultrall.com.
Richart, F. E.; Brandtzaeg, A.; and Brown, R. L., 1928, A Study of
the Failure of Concrete under Combined Compressive Stresses, Bulletin
No.185, Engineering Experimental Station, University of Illinois, Urbana,
IL, 104 pp.
Richart, F. E.; Brandtzaeg, A.; and Brown, R. L., 1929, The Failure of
Plain and Spirally Reinforced Concrete in Compression, Bulletin No. 190,
Engineering Experimental Station, University of Illinois, Urbana, IL, 74 pp.
Saatcioglu, M., and Razvi, S. R., 1992, Strength and Ductility of
Confined Concrete, Journal of Structural Engineering, ASCE, V. 118,
No. 6, pp. 1590-1607.
Saatcioglu, M.; Salamat, A. H.; and Razvi, S. R., 1995, Confined
Columns under Eccentric Loading, Journal of Structural Engineering,
ASCE, V. 121, No. 11, pp. 1547-1556.
Sheikh, S. A., 1982, A Comparative Study of Confinement Models,
ACI Journal, V. 79, No. 4, July-Aug., pp. 296-306.
Sheikh, S. A., and Uzumeri, S. M., 1980, Strength and Ductility of Tied
Concrete Columns, Journal of the Structural Division, ASCE, V. 106,
No.5, pp. 1079-1112.
Sheikh, S. A., and Uzumeri, S. M., 1982, Analytical Model for Concrete
Confinement in Tied Columns, Journal of the Structural Division, ASCE,
V. 108, No. 12, pp. 2703-2722.
Teng, J. G.; Chen, J. F.; Smith, S. T.; and Lam, L., 2002, FRP Strengthened RC Structures, John Wiley & Sons, Ltd., Hoboken, NJ, 266 pp.
Tobbi, H.; Farghaly, A. S.; and Benmokrane, B., 2012, Concrete
Columns Reinforced Longitudinally and Transversally with GFRP Bars,
ACI Structural Journal, V. 109, No. 4, July-Aug., pp. 551-558.
Watson, S.; Zahn, F. A.; and Park, R., 1994, Confining Reinforcement
for Concrete Columns, Journal of Structural Engineering, ASCE, V. 120,
No. 6, pp. 1798-1824.

385

NOTES:

386

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S34

Repair of Prestressed Concrete Beams with Damaged


Steel Tendons Using Post-Tensioned Carbon FiberReinforced Polymer Rods
by Clayton A. Burningham, Chris P. Pantelides, and Lawrence D. Reaveley
Research implementing unibody clamp anchors and a simple
mechanical stressing device to post-tension external, unbonded
carbon fiber-reinforced polymer (CFRP) rods is presented. The
experiments described in the paper concern three prestressed
concrete beams: one was used as the control beam and the other
two were damaged. Damage consisted of cracked concrete that
was removed and internal steel tendons that were cut to simulate
vehicle collision, corrosion, or both. The repair system was then
applied to the two damaged concrete beams. The CFRP repair
system performed well, increasing the ultimate strength and flexural capacity of the damaged beams to meet or exceed the strength
capacity of the control. An analytical model considering the tendon
stress at ultimate and the distribution of internal forces was developed to explore design recommendations for the use of the unibody
clamp anchors and stressing device for post-tensioning CFRP rods.
Keywords: beams; carbon fiber-reinforced polymer (CFRP); post-tensioning; prestressed concrete; repair; retrofit.

INTRODUCTION
Many bridges in the Unites States are approaching the
end of their design life, and some bridges are showing signs
of aging and damage such as corrosion of steel reinforcement, large cracks, and missing concrete cover. Damage to
concrete cover and internal steel prestressing tendons can
occur when large vehicles attempt to pass under a bridge
without adequate clearance. Vehicular impact can fracture
the concrete cover, expose the internal steel prestressing
tendons, and/or sever all or part of the outer steel prestressing
tendons. Even if the tendons are not severed, removal of the
protective concrete cover accelerates the corrosion process.
Additionally, cracking from overloading or fatigue could
facilitate corrosion of internal steel prestressing tendons.
Damage to internal steel prestressing tendons decreases flexural capacity, and bridges exhibiting these symptoms could
be in critical need of replacement, repair, or strengthening.
Typically, girder replacement is expensive, time
consuming, and disruptive; therefore, repair or retrofit is often
the preferred option. One system used for repair applications
is external post-tensioning. This repair method not only
restores flexural capacity, but can also mitigate the demands
of an increase in service load and help with serviceability
considerations such as deflection. Thus, external post-tensioning is an excellent option for repairing concrete bridge
girders with damage to internal steel prestressing tendons.
Traditionally, external post-tensioning has been implemented with high-strength steel tendons because of low
material cost, material availability, and ease of installation.
ACI Structural Journal/March-April 2014

Despite its historic use, however, exposed steel is susceptible


to corrosion, limiting its useful lifespan and requiring extensive protection from deicing salt and moisture.
The limitations of steel tendons can be overcome in
external post-tensioning applications by the use of fiber-reinforced polymer (FRP) materials. FRP materials are advantageous because of their corrosion resistance and high specific
strength. Additionally, the use of FRP materials is becoming
increasingly attractive as the price of FRP composites
decreases. Several studies have shown that post-tensioned
FRP tendons can contribute to flexural strength in new
construction or for strengthening (Abdel Aziz et al. 2005;
El-Hacha and Elbadry 2006; Tljsten and Nordin 2007);
however, few studies have shown the usefulness of post-tensioned FRP tendons in flexural repair and retrofit applications (Elrefai et al. 2007). As a result, additional research is
required to investigate the suitability of post-tensioned FRP
tendons for the repair of severe flexural damage.
Widespread use of FRP tendons in post-tensioning applications has been slow because of the difficulty in developing an effective tendon anchor. Research has produced
a unibody clamp anchor and mechanical stressing device
for use in post-tensioning carbon FRP (CFRP) rods (Burningham 2011). The clamp anchors are machined from a
single piece of steel, and the clamping force is provided by
high-strength bolts. Further research is needed to analyze
the effectiveness of the complete post-tensioning system
consisting of the CFRP rods, unibody clamp anchors, and
mechanical stressing device when applied to prestressed
concrete members.
The research in this paper is concerned with applying
CFRP rods, unibody clamp anchors, and the aforementioned
mechanical stressing device as a complete FRP strengthening system for the repair of damaged prestressed concrete
beams. In the present research, the unibody clamp anchors
were fabricated using mild steel. In actual implementation,
the anchor and stressing device might need to be manufactured using stainless steel or other corrosion-resistant steel.
The specific damage considered during this research was
damage resulting from impact with vehicles passing underneath a bridge without adequate clearance. Such impact
could result in severed internal steel prestressing tendons
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-135.R2, doi:10.14359.51686529, was received October 18, 2012,
and reviewed under Institute publication policies. Copyright 2014, American
Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published ten months from this journals date if the
discussion is received within four months of the papers print publication.

387

or removal of concrete cover and subsequent corrosion of


internal steel prestressing tendons. An illustrative example
of impact damage observed on an actual prestressed concrete
bridge girder can be seen in Fig. 1.
The paper includes details of testing methods, specimen
design and fabrication, experimental design, and analysis of
results. Additionally, the methods used for collecting data
during laboratory testing as well as details pertaining to the
performance and effectiveness of the FRP repair system
and its application are provided, with specific focus on the
performance of the CFRP tendons and their ability to aid in
the repair of damaged beams. An analytical model considering the tendon stress at ultimate and conventional beam
theory is presented to explore design recommendations for
the use of the unibody clamp anchors and stressing device.
RESEARCH SIGNIFICANCE
Previous research on using external post-tensioned CFRP
tendons for repair of damaged concrete beams has been
limited. This research presents the implementation of newly
developed unibody clamp anchors and a simple mechanical stressing device for the repair of damaged prestressed
concrete beams with post-tensioned CFRP rods. In addition,
the paper validates equations from the literature for evalu-

ating the ultimate stress of unbonded post-tensioned CFRP


rods. The CFRP repair system implemented in this research
could facilitate the acceptance of CFRP post-tensioning
systems by the construction industry.
EXPERIMENTAL INVESTIGATION
Specimen fabrication
Three prestressed concrete (PC) beam specimens were
designed and fabricated for testing. The three beams (Specimens P2, RP1, and RP3, with R indicating a specimen to
which the repair system was applied) were manufactured by
a local PCI-certified precast/prestressed concrete company.
The precast beams measured 12 in. (305 mm) wide x 20 in.
(508 mm) tall x 15 ft (4.57 m) long, and each prestressed
beam had three 1/2 in. (13 mm) seven-wire low-relaxation prestressing steel strands with an ultimate strength of
270ksi (1862 MPa). The beams were also reinforced with
two No.5 (16 mm) mild steel bars in the tension zone, and
two No.5 (16 mm) mild steel bars in the compression zone.
The prestressed beams had No. 3 (10 mm) stirrups placed at
12 in. (305 mm) on center. The beam dimensions and location of internal reinforcement are shown in Fig. 2.
Experimental design
All three beams were tested and subjected to initial damage
using a four point loading system to induce tensile cracking.
The same setup was used to test the beams to failure. A
hydraulic actuator with a 500 kip (2220 kN) inline load cell
and a steel spreader beam were used to apply a two-point
load, spaced 30 in. (762 mm) apart, to the top of the specimens, as shown in Fig. 3. The specimens were tested with
an unbraced length of 13 ft, 8 in. (4.17 m) and had a depth
of 20in. (508 mm), giving a shear span-depth ratio (a/d)
of3.35.

Fig. 1Girder damage from vehicle impact.

Material properties
The materials used in this research are typical of construction in the United States. All steel reinforcing bars used in the
fabrication of the specimens had a nominal tensile strength

Fig. 2Reinforcement layout for Specimens P2, RP1, and RP3. (Note: 1 in. = 25.4 mm; 1 ft = 0.305 m.)
388

ACI Structural Journal/March-April 2014

Fig. 4Damaged outer steel tendon in Specimen RP1 (three


wires cut) and RP3 (seven wires cut).
Fig. 3Test setup. (Note: 1 in. = 25.4 mm; 1 ft = 0.305 m.)
of 60 ksi (414 MPa). The CFRP rods used in this research
had the following properties as provided by the manufacturer: rod diameter = 3/8 in. (9.53 mm), tensile strength
= 250 ksi (1724 MPa), tensile modulus = 22,500ksi (155
GPa), and elongation at break = 1.11%. In separate tensile
tests of CFRP rods from the same lot as the ones used in
this research, carried out using unibody clamp anchors, the
average measured tensile strength was 308 ksi (2124MPa),
and the average ultimate strain measured was 1.37%. The
internal steel prestressing tendons were low relaxation
1/2in. (13 mm) diameter seven-wire strands with a nominal
ultimate strength of 270 ksi (1862 MPa). Concrete cylinder
tests performed at 7 days after casting of the steam cured
prestressed concrete beams gave an average compressive
concrete strength of 7.0 ksi (48 MPa), and at the time of
specimen testing, the concrete had an average compressive
strength of 10.0 ksi (69 MPa) based on compression tests of
4 in. (102 mm) diameter by 8 in. (204 mm) high cylinders.
Testing methods
Load testing was carried out in three phases: damage,
repair, and failure. First, loading was used to introduce tensile
cracks, which could lead to accelerated corrosion of internal
steel prestressing tendons. Additionally, Specimens RP1 and
RP3 were damaged with respect to the internal prestressing
steelto simulate impact damage. Subsequently, Specimens RP1 and RP3 were repaired with external post-tensioned CFRP rods. Finally, all three specimens were loaded
monotonically to failure.
Damage loadingDamage loading applied to the specimens consisted of downward half-cycles to induce tensile
cracking. The loading was displacement controlled to
avoid catastrophic failure and subsequent loss of the specimens. Displacement half-cycles were applied in increments of 0.0625 in. (1.59 mm), with the amplitude of each
successive half-cycle increasing by 0.0625 in. (1.59 mm).

ACI Structural Journal/March-April 2014

In addition, the rate of displacement was held constant at


0.0625 in./min (1.59 mm/min) throughout the test. All specimens were subjected to the same loading protocol, with
termination of loading dependent upon the level of cracking.
The vertical deflection at midspan was limited to 0.375 in.
(9.5 mm), and the cracks were of a hairline width mainly in
the constant moment region at a spacing of 12 in. (305 mm).
Additional damage was inflicted on Specimens RP1 and
RP3 to simulate damage to internal steel prestressing tendons
from vehicle collision, subsequent corrosion, or both. An
area approximately 8 in. (203 mm) long with a depth equal to
the concrete cover of the prestressing tendons was removed
from both Specimen RP1 and RP3 to expose an outer sevenwire steel prestressing strand within the constant moment
region. For Specimen RP1, three of the seven wires in this
strand were cutleaving two intact seven-wire strands and
one four-wire damaged strand. For Specimen RP3, all seven
wires of an outer steel prestressing strand were cut to simulate
severe damage, leaving two intact seven-wire strands. These
cuts, seen in Fig. 4, simulated partial or complete severing of
the exterior tendon on impact in an exterior girder or corrosion of an exterior tendon due to loss of concrete cover and
subsequent exposure to the elements.
FRP repairAfter simulating damage to internal steel
prestressing tendons in Specimens RP1 and RP3, the beams
were repaired with external post-tensioned CFRP rods. The
unibody clamp anchors and mechanical stressing device
seen in Fig. 5 were used to introduce the post-tensioning
force. Specimens RP1 and RP3 were repaired with two rods,
one on each side of the beam along the beam length at a
depth of 15 in. (381 mm) from the top compression fiber.
The CFRP rods were post-tensioned to a strain of approximately 0.485% measured using strain gauges attached to
the rod at midspan, producing a calculated design post-tensioning force of 12 kip (53.4 kN) in each rod. ACI 440.4R
(ACI Committee 440 2004) recommends initial jacking
stresses of 40 to 65% of the design ultimate tensile strength
of the prestressed FRP tendon and anchorage system. In
the present case, the amount of initial jacking of the CFRP
389

Fig. 5End view of stressing system showing clamp anchors


and stressing device.
rods was approximately 44% of the design ultimate tensile
strength based on the manufacturers specification.
The novel mechanical stressing device implemented
in this research consists of a slotted square HSS section
running perpendicular to the beam length at both ends of
the beam. The slots in the HSS section allow the tendons to
pass through the slots such that the unibody clamp anchors
make contact with the back side of the tube. On the stressing
end of the beam, two sleeve nuts are positioned on top, and
two on the bottom of the HSS section. The sleeve nuts run
parallel with the beam, and tendon stressing occurs when
1.0in. (25 mm) diameter bolts are screwed into the sleeve
nuts; the bolts react against the beam end, moving the HSS
section back to stress the tendons. Tightening the stressing
bolts in an alternating star pattern ensures the tendons are
stressed with controlled increments of tightening. More
details of the unibody clamp anchor and stressing device are
provided by Burningham (2011).
Loading to failureAfter repairing Specimens RP1 and
RP3 with external post-tensioned CFRP rods, the specimens
were loaded to failure, with Specimen P2 as the control
specimen. The displacement controlled loading to failure
was monotonic at a rate of 0.0625 in./min (1.59 mm/min).
The test was stopped at imminent failure of the specimen,
as measured by a 20% decrease from the maximum load or
failure of the external CFRP rods, whichever occurred first.
EXPERIMENTAL RESULTS
Instrumentation and data collection methods
Instrumentation consisted of strain gauges and linear variable differential transformers (LVDTs). The specimens were
instrumented with three LVDTs, one 54.5 in. (1.38 m) from
either end and one at midspan as shown in Fig. 3; these were
attached to the bottom of the beam to measure the deflected
shape under load. Concrete strain gauges placed at 69.5 in.
(1.77 m) from each end and at midspan on the top face of
the beams measured the concrete compressive strain. Strain
gauges were also applied to each CFRP rod at midspan as
shown in Fig. 3; they were used to measure strain at initial
prestress and throughout the tests until failure of Specimens RP1 and RP3. All measurements were collected by
an electronic data acquisition system at a sampling rate of
two data points per second. All strain gauge readings were
390

Fig. 6Control Specimen P2 at failure; grid = 4 in.


(102mm).

Fig. 7Repaired Specimen RP1 at imminent failure:


(a)front; and (b) back; grid = 4 in. (102 mm).
measured in units of microstrain, and all LVDT readings
were measured within 0.001 in. (0.025 mm).
Specimen data analysis
No anchor slippage was observed in any of the CFRP rods
during testing of the specimens to failure. The lack of slip
demonstrates that the anchors work as designed. This was
confirmed in the laboratory in CFRP rod tensile tests using
the unibody anchors according to ACI 440.3R guidelines
(ACI Committee 440 2012).
During testing, Specimen P2 (control) failed due to
concrete compressive failure at the center, after flexural
cracks had developed; crack spacing in the constant moment
region was 6 in. (152 mm) and the maximum crack width
was 0.08 in. (2 mm), as shown in Fig. 6. Specimen RP1
(repaired) failed due to concrete compression failure accompanied by flexural cracks, after rupture of the external CFRP
rods; crack spacing in the constant moment region was
6in. (152 mm), and the maximum crack width was 0.06in.
ACI Structural Journal/March-April 2014

Fig. 9Concrete compressive strain at midspan versus


applied load to failure.

Fig. 8Repaired Specimen RP3 at failure: (a) front;


(b)back; grid = 4 in. (102 mm).
(1.5mm), as shown in Fig. 7. Specimen RP3 (repaired)
failed due to concrete compression failure at the center;
flexural crack spacing in the constant moment region was
4 in. (102 mm), and the maximum crack width was 0.07in.
(1.8mm). In addition, Specimen RP3 developed several
flexural tension cracks around the area of the cut tendon, and
the top compression mild steel bars buckled subsequent to
concrete crushing, as shown in Fig. 8.
For all three specimens, the highest compressive concrete
strain was observed near midspan, as seen in Fig. 9, which
shows the change in concrete compressive strain at midspan
as a function of applied load. From Fig. 9 it can be seen
that Specimens P2 and RP3 experienced maximum concrete
strains just greater than 0.30%. Specimen RP1 also started
failing in compression, but the test was terminated prematurely at a smaller deflection than the other two tests with the
maximum concrete strain reaching 0.27%. These numerical
data correlate well with the failure mode visually observed
in Specimens P2 and RP3 and seen in Fig. 6 and 8, respectively. Figure 7 shows that Specimen RP1 experienced severe
cracking damage with significant flexural cracks reaching
the top of the beam, and imminent concrete compression
failure, similar to Specimen P2.
The change in CFRP rod strain at midspan as a function of
applied load for Specimens RP1 and RP3 is shown in Fig. 10.
An increase in strain of the CFRP rods for Specimens RP1
and RP3 was observed during testing of the specimens to
failure. The initial strain in the rods from the post-tensioning
application was 0.485%. At failure of the external CFRP
tendons, the average maximum strain in the CFRP rods was
0.750 and 0.814% for Specimens RP1 and RP3, respecACI Structural Journal/March-April 2014

Fig. 10CFRP rod strain at midspan versus applied load


to failure.
tively. Because the measured ultimate strain of the rods in
axial tension is 1.37, the most likely explanation for rupture
of the rods at lower strains is eccentric bending due to lack
of rotation of the anchor at the supports; this caused stress
concentrations at a point other than where the strain gauges
were located at midspan. One location of stress concentrations is near the clamp anchors at the beam ends due to rotation of the latter at large deflections. Figure 11 shows the two
modes of failure observed: straw broom and splitting failure;
the former failure was observed near the clamp anchors, and
the latter near midspan. In actual applications, the use of a
rotating rocker at both anchorage devices is desirable to keep
the rod from experiencing flexural bending stresses.
The CFRP repair system was successfulboth Specimens
RP1 and RP3 exhibited an increase in ultimate capacity and
benefited from application of the CFRP repair system. The
ultimate load for Specimens P2 (control), RP1 (repaired),
and RP3 (repaired) was 104, 112, and 102 kip (463, 498,
and 454 kN), respectively. For Specimen RP1, the ultimate
load corresponds to an increase of approximately 7.7% in
ultimate capacity from the use of external post-tensioned
CFRP rods. It should be remembered, however, that Specimen RP1 (repaired) had two intact seven-wire strands and
one four-wire strand (three wires were cut), whereas Specimen P2 (control) had three intact seven-wire strands. Therefore, application of the theoretical capacity of Specimen
RP1 (repaired) based on the cut wires produces an effec391

Fig. 12Applied load versus midspan deflection to failure.

Fig. 11Failure of CFRP rods: (a) straw broom; and


(b)splitting.
tive increase in ultimate capacity of 20.6% from the use of
external post-tensioned CFRP rods. In addition, although a
third of the prestressing force was removed from Specimen
RP3, the repair with the external post-tensioned CFRP rods
produced load-deflection behavior virtually identical to that
of the control specimen. This identical behavior implies
restoration of the girder and an effective increase in ultimate
capacity of 31.1% from the use of external post-tensioned
CFRP rods.
Deformability and ductilityThe similarities in the
performance of Specimens P2 (control) and RP3 (repaired)
are shown in Fig. 12. From Fig. 12, it can be seen that failure
of the CFRP rods for Specimen RP1 occurred at a deflection
about eight times greater than L/800 (0.21 in. or 5.21 mm),
the maximum allowable design deflection under service live
loads for concrete bridge construction (AASHTO 2009).
Therefore, although failure of the post-tensioned CFRP rods
was brittle, failure of the beams occurred at a deflection much
greater than any expected service load deflections. After
failure of the CFRP rods, Specimen RP1 (repaired) exhibited a residual strength of approximately 97 kip (431 kN).
This residual strength is evidence that complete catastrophic
failure of the beam did not occur because of the reserve
capacity and ductility of the original system. Additionally,
it can be concluded from Fig. 12 that the residual strength
in Specimen RP1 (repaired) after failure of the CFRP rods
suggests that at large deflections, it would fail similarly to
Specimen P2 (control)from concrete compressive failure;
the test for Specimen RP1 was terminated prematurely at a
smaller deflection.
Another measure of performance is the ductility of the
repaired beams compared with the control beam. Ductility
is provided by the mild steel present in the tension and
compression zone of all three beams, as shown in Fig. 2.
CFRP rods are brittle, and even though the amount of mild
steel was the same, the amount of prestressing steel was
less for the repaired beams; therefore, it is interesting to
compare the ductility of the three beams. One definition of
392

Fig. 13Member ductility of concrete beams prestressed by


FRP by Abdelrahman et al. (1995).
ductility for FRP reinforced beams is the method developed
by Abdelrahman et al. (1995). As shown in Fig. 13, ductility
is defined in terms of the equivalent displacement of the
uncracked section and the displacement at ultimate and is
given by the following expression

2
1

(1)

where 1 is the displacement of the uncracked section at a


load equal to the ultimate load; and 2 is the displacement at
failure. Using Eq. (1) and Fig. 12, the ductility of the control
beam Specimen P2 is obtained as 8.1, and that of the repaired
beams as 7.7 for RP3 and 7.3 for RP1. It is observed that the
repaired beams are practically as ductile as the control beam.
It should be noted that Specimen RP1 would have achieved
greater ductility if the test had not been terminated.
ANALYTICAL INVESTIGATION
Conventional beam theory can be used to predict the
ultimate load of the specimens tested in this research. The
nominal theoretical capacity of control Specimen P2 was
calculated to be 73 kip (330 kN). Compared with the actual
ultimate load of 104 kip (463 kN), the ratio of actual to theoretical ultimate load is 1.42, indicating that the theoretical
prediction is in good agreement with the actual value, and the
ACI Structural Journal/March-April 2014

Table 1Data for calculation of theoretical values at ultimate


Specimen

fp_CFRP, ksi (MPa)

cu, in. (mm)

Aps, in.2 (mm2)

a, in. (mm)

Mu, kip-ft (kN-m)

RP1 (repaired)

155 (1069)

2.51 (63.8)

0.393 (254)

1.63 (41.4)

220 (298)

PR3 (repaired)

162 (1117)

2.22 (56.4)

0.306 (197)

1.44 (36.6)

196 (266)

design is conservative. To determine the theoretical capacity


of Specimens RP1 and RP3, the stress in the CFRP rods at
ultimate should first be determined. Previous research for
unbonded steel tendons has shown that strain compatibility
can be used to analyze the tendons as if they were bonded,
and then apply a strain reduction factor to account for the
tendons being unbonded (Naaman and Alkhairi 1991). It has
been suggested that this method of using a strain reduction
factor could also be applied to FRP tendons (Naaman et al.
2002; ACI Committee 440 2004); the stress at ultimate in
unbonded FRP tendons is given as

d
f p _ CFRP = f pe _ CFRP + u ECFRP e cu CFRP 1

cu

(2)

where cu is the depth to the neutral axis at ultimate; dCFRP is


the depth to the CFRP rods, 15 in. (381 mm); ECFRP is the
modulus of elasticity of the CFRP rods, 22,500 ksi (155GPa);
fp_CFRP is the stress in the CFRP rods at ultimate; fpe_CFRP is the
effective prestress in the CFRP rods, 109 ksi (752 MPa); u
is the strain reduction factor; and cu is the failure strain of
concrete in compression, 0.003 in./in. (mm/mm).
Suggested values for the strain reduction factor depend on
the type of loading. The research presented in this paper is
best described as center point loading because the distance
between loading points was only 30 in. (762 mm) compared
with an unbraced length of 13ft, 8 in. (4.17 m). The strain
reduction factorstandardized to likely produce a conservative predicted valuefor center point loading is given as
1.5
u =
L / dCFRP

(3)

B + B2 4 AC
2A

A = 0.85 fcbb1

(4a)

B = ACFRP ECFRP e cu u f pe _ CFRP As f y Aps f ps (4b)


u =

C = ACFRP ECFRP e cu u dCFRP

In the aforementioned expressions, ACFRP is the area of


CFRP rods, 0.22 in.2 (142 mm2); Aps is the area of internal
ACI Structural Journal/March-April 2014

A
0.21
+ 0.04 CFRP int + 0.04
L / dCFRP
ACFRP tot

(5)

where ACFRP int is the area of internal CFRP tendons; and


ACFRP tot is the total area of CFRP tendons. Additionally, to
account for the change in an external tendons eccentricity,
the effective CFRP tendon depth can be found using

where L is the unbraced length of the beam, 13 ft, 8 in.


(4.17m) (Naaman et al. 2002; ACI Committee 440 2004).
The use of Eq. (3) for specimens in this research results in
a strain reduction factor of 0.137. Additionally, appropriate
values of cu can be calculated from the following equation
cu =

steel prestressing strands; As is the area of tensile mild steel


reinforcement, 0.62 in.2 (400 mm2); b is the beam width,
12in. (305 mm); fc is the compressive strength of concrete,
10.0 ksi (69 MPa); fps is the prestressing force in the internal
steel strands, 243 ksi (1.68 GPa); fy is the yield stress of
mild steel reinforcement, 60 ksi (414 MPa); and 1 is 0.65.
The use of Eq. (4) produces cu values of 2.51 and 2.22 in.
(63.8and 56.4 mm) for Specimens RP1 and RP3, respectively. The resulting CFRP rod ultimate stresses predicted
by Eq. (2) are 155 ksi (1069 MPa) for Specimen RP1, and
162 ksi (1120 MPa) for Specimen RP3, which are conservative compared with the actual values of measured ultimate
CFRP rod stress (from strain gauges on the rods) of 169 and
183 ksi (1163 and 1263 MPa), respectively. The analytical
results for ultimate stress are summarized in Table 1.
An alternative strain reduction factor for external,
unbonded steel tendons has also been developed, and it had
been recommended for FRP tendons (Aravinthan et al. 1997;
ACI Committee 440 2004). This alternative strain reduction
factor is given as

de _ CFRP = dCFRP Rd

(6)

where Rd is the depth reduction factor given as

L
S
0.19 d 1.0
Rd = 1.14 0.005

L
dCFRP

(7)

with the deviator spacing Sd equal to the span L for specimens in this research because no deviators were used. For
the case of specimens in this research, the value of Rd is
0.895. The predicted ultimate CFRP tendon stresses from
Eq. (2) using the alternative strain reduction factor from
Eq. (5) are 135 and 138 ksi (931 and 951 MPa) for Specimens RP1 and RP3, respectively. Similar to the results from
the application of the strain reduction factor of Eq. (3), these
predictions are conservative. A comparison of the applicability of the two strain reduction factors can be made when
considering the actual and predicted ultimate CFRP tendon
stresses for Specimens RP1 and RP3. The measured CFRP
tendon stresses, the predicted, and the error are shown in
Table 2. Although both strain reduction factors produced
393

Table 2Measured and predicted ultimate tendon stresses

Specimen

Measured CFRP tendon


stress at ultimate, ksi (MPa)

Predicted CFRP tendon stress at


ultimate from Eq. (3), ksi (MPa)

Error in prediction
from Eq. (3), %

Predicted CFRP tendon stress


at ultimate from Eq. (5), ksi
(MPa)

Error in prediction
from Eq. (5), %

RP1 (repaired)

169 (1163)

155 (1069)

135 (931)

18

PR3 (repaired)

183 (1263)

162 (1120)

11

138 (951)

24

conservative ultimate stress predictions, the strain reduction


factor from Eq. (3) results in the smallest error, indicating
that it works best for specimens in this research.
Conventional beam theory leads to the following equation
for ultimate moment capacity
a
a

Mu = Aps f ps d ps + As f y d

2
2

+ ACFRP f p _ CFRP dCFRP

(8)

where Mu is the ultimate moment capacity; a is the depth of


the equivalent compression stress block equal to 1cu; and d
is the depth to the mild steel reinforcement, 15 in. (381 mm).
Next, from the ultimate moment capacity, the ultimate load
Pu can be found from the following equation
Pu =

4 Mu

(L s)

(9)

where s is the spacing between load points, 30 in. (762 mm),


as shown in Fig. 3.
From Eq. (9), the theoretical ultimate capacity is 79 kip
(351 kN) for Specimen RP1 and 70 kip (311 kN) for Specimen RP3 when implementing the predicted CFRP tendon
stresses at ultimate from the use of Eq. (3). Consequently,
the corresponding ratios of actual to theoretical ultimate load
are 1.42 and 1.46 for Specimens RP1 and RP3, respectively.
Similar to the ratio of 1.42 found for control Specimen P2,
these ratios show that theoretical ultimate loads are in good
agreement with actual measured ultimate loads, and that
the design is conservative. A summary of the experimental
and theoretical ultimate loads is given in Table 3. Furthermore, the ratios of actual to theoretical ultimate load and the
percentage of error between the actual and theoretical stress
in the CFRP rods at ultimate indicate that Eq. (2) and (3)
are appropriate for predicting the stress in the CFRP rods at
ultimate when calculating the theoretical ultimate capacity
of prestressed concrete members repaired with the system
of unibody clamp anchors, mechanical stressing device, and
CFRP rods used in the current research.
CONCLUSIONS
Based on the experiments carried out in this research,
it can be concluded that Specimens RP1 and RP3 were
successfully repaired using an external post-tensioning
system consisting of CFRP rods, unibody clamp anchors,
and a mechanical stressing device. Repaired Specimens RP1
394

Table 3Experimental and theoretical


ultimateloads

Specimen

Experimental ultimate load, kip (kN)

Theoretical
ultimate load
from Eq. (3),
kip (kN)

Ratio of
experimental
to theoretical
ultimate load

P2 (control)

104 (463)

73 (325)

1.42

RP1 (repaired)

112 (498)

79 (351)

1.42

PR3 (repaired)

102 (454)

70 (311)

1.46

and RP3 showed an effective increase in ultimate strength of


20.6 and 31.1%, respectively, with respect to the damaged
condition. This increase in ultimate strength of Specimens
RP1 (repaired) and RP3 (repaired) compared with Specimen
P2 (control) demonstrate that external post-tensioned CFRP
rods are able to compensate for partial or complete removal
of a prestressing strand.
Although the repaired Specimen RP1 failed as a result
of rupture of the external post-tensioned CFRP rods, this
rupture occurred at deflections much greater than those
expected from service loads. Additionally, residual capacity
was present after CFRP rod rupture. This is significant in
that catastrophic beam failure did not occur even though the
CFRP rods failed in tension. The repaired beams were essentially as ductile as the control beam, because the mild steel
and the remaining prestressing steel dominated the ductility
of the damaged beams.
It was found that theoretical expressions from the literature may be used to predict the stress at ultimate in the CFRP
tendons used in this research as well as the ultimate capacity
of the beams, with Eq. (3) being the preferred strain reduction factor even though it was originally developed for steel
tendons. Post-tensioning CFRP rods using unibody clamp
anchors and a mechanical stressing device is a viable technique for the repair of concrete beams with severe damage
to internal steel prestressing tendons. It is recommended that
further studies be carried out to assess the system and its
feasibility for general use.
It is also recommended that further studies be carried out
to test unibody clamp anchors made of stainless steel and to
determine the suitability of unibody clamp anchors for use
as a coupling device for CFRP rods. The successful implementation of the anchors and CFRP rods in this research
suggests that the anchors could potentially be used to join
two sections of CFRP rod, facilitating post-tensioning of
longer spans that are typical of actual bridges. Although the
repair system investigated in this paper was successful, it
requires access to the end of the beam, which is not always
available in field applications. For use of the CFRP repair
ACI Structural Journal/March-April 2014

system implemented in this research, however, only 18 in.


(0.46 m) of free space is required behind the beams.
ACKNOWLEDGMENTS

The authors wish to acknowledge the financial support of the Utah


Department of Transportation and the University of Utah. The authors
would also like to acknowledge the contributions of Hanson Structural
Precast and Sika, Inc. In addition, the authors would like to thank M. Bryant
and R. Liu for their assistance in specimen fabrication and testing.

NOTATION

ACFRP = area of post-tensioned CFRP rods


ACFRP int = area of internal post-tensioned CFRP rods
ACFRP tot = total area of post-tensioned CFRP rods
Aps
= area of internal steel prestressing strands
A s
= area of tensile mild steel reinforcement
a
= depth of equivalent compression block
cu
= depth to neutral axis at ultimate
d
= depth of mild steel reinforcement
dCFRP = depth to CFRP rods
de_CFRP = effective depth to CFRP rods
ECFRP = modulus of elasticity of CFRP rods
fc
= compressive strength of concrete
fp_CFRP = stress in CFRP rods at ultimate
fpe_CFRP = effective prestress in CFRP rods
fps
= prestressing force in internal steel strands
fy
= yield stress of mild steel reinforcement
L
= unbraced length of beam
Mu
= ultimate moment capacity
Rd
= depth reduction factor
Sd
= deviator spacing
s
= spacing between load points on top of beam
1
= factor relating depth of equivalent compression block to depth
of neutral axis
1
= displacement of uncracked section at load equal to ultimate load
2
= displacement at failure
cu
= failure strain of concrete in compression
u
= strain reduction factor

AUTHOR BIOS

Clayton A. Burningham is a PhD Candidate in the Department of Civil


and Environmental Engineering at the University of Utah, Salt Lake City,
UT. He received his bachelors and MS degrees from the University of
Utah. His research interests include repair of reinforced and prestressed
concrete structures and post-tensioning carbon fiber-reinforced polymer
materials.
ACI member Chris P. Pantelides is a Professor in the Civil and Environmental Engineering Department at the University of Utah. His research
interests include seismic design and rehabilitation of reinforced concrete,

ACI Structural Journal/March-April 2014

precast and prestressed concrete buildings and bridges, and the application
of fiber-reinforced polymer composites.
Lawrence D. Reaveley is a Professor and former Department Chair of
Civil and Environmental Engineering at the University of Utah. He received
his bachelors and MS degrees from the University of Utah, and his PhD
from the University of New Mexico, Albuquerque, NM. His research interests include structural dynamics, with an emphasis on earthquake engineering and seismic rehabilitation.

REFERENCES

AASHTO, 2009, AASHTO LRFD Bridge Design Specifications,


fourth edition, American Association of State Highway and Transportation
Officials, Washington, DC, 1518 pp.
Abdel Aziz, M.; Abdel-Sayed, G.; Ghrib, F.; Grace, N.; and Madugula,
M., 2005, Analysis of Concrete Beams Prestressed and Post-Tensioned
with Externally Unbonded Carbon Fiber Reinforced Polymer Tendons,
Canadian Journal of Civil Engineering, V. 31, pp. 1138-1151.
Abdelrahman, A. A.; Tadros, G.; and Rizkalla, S. H., 1995, Test Model
for the First Canadian Smart Highway Bridge, ACI Structural Journal,
V.92, No. 4, July-Aug., pp. 451-458.
ACI Committee 440, 2004, Prestressing Concrete Structures with FRP
Tendons (ACI 440.4R-04) (Reapproved 2011), American Concrete Institute, Farmington Hills, MI, 35 pp.
ACI Committee 440, 2012, Guide Test Methods for Fiber-Reinforced
Polymer (FRP) Composites for Reinforcing or Strengthening Concrete and
Masonry Structures (ACI 440.3R-12), American Concrete Institute, Farmington Hills, MI, 23 pp.
Aravinthan, T.; Mutsuyoshi, H.; Fujioka, A.; and Hishiki, Y., 1997,
Prediction of the Ultimate Flexural Strength of Externally Prestressed PC
Beams, Transactions of the Japan Concrete Institute, V. 19, pp. 225-230.
Burningham, C., 2011, Development of a Carbon Fiber Reinforced
Polymer Prestressing System for Structural Applications, PhD dissertation, University of Utah, Salt Lake City, UT, 103 pp.
El-Hacha, R., and Elbadry, M., 2006, Strengthening Concrete Beams
with Externally Prestressed Carbon Fiber Composite Cables: Experimental
Investigation, PTI Journal, V. 4, No. 2, pp. 53-70.
Elrefai, A.; West, J.; and Soudki, K., 2007, Strengthening of RC Beams
with External Post-Tensioned CFRP Tendons, Case Histories and Use of
FRP for Prestressing Applications, SP-245, R. El-Hacha and S. H. Rizkalla,
eds., American Concrete Institute, Farmington Hills, MI, pp.123-142.
Naaman, A., and Alkhairi, F., 1991, Stress at Ultimate in Unbonded
Post-Tensioning Tendons: Part 2Proposed Methodology, ACI Structural Journal, V. 88, No. 6, Nov.-Dec., pp. 683-692.
Naaman, A.; Burns, N.; French, C.; Gamble, W.; and Mattock, A., 2002,
Stresses in Unbonded Prestressing Tendons at Ultimate: Recommendation, ACI Structural Journal, V. 99, No. 4, July-Aug., pp. 518-529.
Tljsten, B., and Nordin, H., 2007, Concrete Beams Strengthened with
External Prestressing Using External Tendons and Near-Surface-Mounted
Reinforcement, Case Histories and Use of FRP for Prestressing Applications, ACI SP-245, R. El-Hacha and S. H. Rizkalla, eds., American
Concrete Institute, Farmington Hills, MI, pp. 143-164.

395

NOTES:

396

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S35

Study of Composite Behavior of Reinforcement and


Concrete in Tension
by John P. Forth and Andrew W. Beeby
This paper aims to further the understanding of the interaction
between reinforcement in tension and the surrounding cracked
concrete. This is achieved using the elastic analysis of axisymmetric
prisms reinforced with a single central bar. As a preliminary to the
analyses, the behavior of axially reinforced prisms is described
based on previous experiments. This preliminary analysis confirms
that the elastic analysis adopted in this investigation is reasonable.
Two analytical exercises are described: the first assumes no slip,
plasticity, or internal cracking at the interface between the steel
and the concrete; the second introduces internal cracking and
debonding between ribs. The first analysis indicates that shear
deformation of the surrounding concrete accounts for a substantial
proportion of the surface crack width, and therefore that this form
of deformation cannot be ignored in crack prediction formulae. The
second analytical exercise shows that the internal cracking model
described by Goto is appropriate.
Keywords: axisymmetric tension specimens; cover; crack width calculations; cracking mechanisms; finite element modeling; shear distortions.

INTRODUCTION
The objective of this study is to gain greater understanding of the interaction of reinforcement and concrete in
tension. The analysis models used have been kept as simple
as possible; the approach has been limited to pure elastic
behavior, and an assumed internal cracking pattern, based
on the work by Goto,1 has been adopted (rather than use
a nonlinear finite analysis software based on, for instance,
a smeared cracking approach where cracks are predicted
regions of damaged material with degraded properties). This
keeps difficulties in interpretation to a minimum, though it
is recognized that concrete does not necessarily behave in a
perfectly elastic manner.
Use of elastic modeling, where the cracks being studied are
open, is not so unreasonable as might be thought by some.
Extensive data obtained by Scott and Gill2 and Beeby and
Scott3-5 suggest that much of the behavior revealed during
tension tests is close to what would be expected from an
assumption of elastic-brittle behavior for concrete in tension.
This is effectively what will be assumed in thestudy.
RESEARCH SIGNIFICANCE
In reinforced concrete, the interaction between reinforcement in tension and the surrounding concrete is still not
fully understood. Two internal failure mechanisms, pure slip
and internal cracking, form the basis of three approaches
that exist in the codes to model the tension zone behavior
under service loads. The models presented in this investigation are based only on Gotos internal cracking mechanism.
These models predict the experimental behavior of tension
ACI Structural Journal/March-April 2014

Fig. 1Load-deformation response of Specimen 10T12


(figure taken from Reference 4).
members quite effectively. The simpler model proposed
herein further confirms the concept that crack width is a
function of the shear deformation of the concrete cover.
BASIC BEHAVIOR AS REVEALED BY TESTS
The information used herein is taken from Reference 4.
Initially, strain data for Specimen 100T12 will be presented,
as this gives a convenient illustration of a number of aspects
of behavior. Figure 1 shows the load-average reinforcement
strain response for this specimen.
It should be noted that the response is not a continuous
smooth curve as is commonly plotted, but is made up of a
series of linear segments separated by a sudden increase in
strain on the occurrence of each crack. Up to a load of approximately 7.868 kip (35 kN), these linear segments, extrapolated backwards, can be seen to pass through the origin. The
behavior of the tension specimen with a given number of
cracks is thus elastic. Using the computer to produce best
fit lines for each segment enables the stiffness of the specimen to be established for each crack configuration. Figure 2
shows this compliance plotted against the number of cracks.
There is a linear relationship between stiffness and number
of cracks. This implies that the formation of each crack
reduces the stiffness of the element by a constant amount.
The final point for four cracks does not quite fit the linear
relationship. This point is obtained from the behavior immediately after formation of the fourth crack. Figure 1 shows
that, at higher loads, there are two further sudden increases
in strain. These increases were not related to the formation
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-148, doi:10.14359.51686564, was received May 10, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

397

Fig. 2Stiffness of element as function of number of cracks


for Specimen 100T12.
Fig. 4Concrete stresses at right end of Specimen T16B1.
(Note: 1 MPa = 145 psi; 1 kN = 0.225 kip; 1 mm = 0.0394in.)

Fig. 3Specimen T16B1: reinforcement strains along bar at


various loads. (Note: 1 kN = 0.225 kip; 1 mm = 0.0394in.)
of visible surface cracks, and it may be speculated that they
arise from some form of internal failure. It should be noted
that both these increases in strain occurred only at very high
levels of stress in the reinforcement (>58.02 ksi [400 MPa]).
Figures 1 and 2 show that the behavior between cracking
events is generally elastic, and this is the assumption that
will effectively be made in the finite element analyses.
The next aspect of behavior to be considered is the variation of steel stress or concrete stress with distance from a
crack. Figure 3 shows the variation of strain along a reinforcing bar for various levels of axial load. Various pieces of
information can be gleaned from this figure.
First, over a considerable part of the distance from a crack
or the free end of the specimen, the variation in strain is very
close to linear. This can possibly be seen better from Fig.4,
which shows the variation in concrete stress over the end
11.81 in. (300 mm), enlarged for two levels of load. At the
lower load level, there is a clear curve over the part of the
bar where the stress is close to that for uncracked concrete.
This is not clear for the higher load, where it would not be
unreasonable to consider the relationship to be linear over
the whole distance to midway between cracks.
Secondly, even at the low level of load, which is below
the cracking load for the specimen, the strain in the reinforcement over most of the length affected by the end of
the specimen So is considerably greater than the cracking
strain of the concrete, which can be assumed to be in the
range of 100to 150 106. This implies that some form of
internal failure occurs over the whole length So as soon as
crackingoccurs.
398

Fig. 5Relationship between So and cover. (Note: 1 mm =


0.0394 in.)
A critical factor in crack prediction theories is the definition of the transfer length. This is the distance on one side of
a crack over which the stress in the reinforcement is affected
by the crack. In References 3 through 5, this is represented
by the symbol So. In some papers, the symbol ltr is used.
There are various ways of estimating So from experimental
results. Most of these are indirect and assume a relationship, for example, between So and crack width. The work
reported in References 3 through 5, however, recorded the
variation in strain at closely spaced intervals along the reinforcing bar, permitting the direct measurement of So. Even
with these tests there are difficulties, as can be seen from the
strain variation for the 4.496 kip (20 kN) level of load shown
in Fig. 4. It is difficult to define exactly the point where the
crack no longer influences the strain. The method used to
establish a value for So is illustrated in Fig. 4, where So is
defined as the distance from the crack (or free end of the bar)
and the point where a best fit line through the strains intersects the strain in the uncracked concrete. This is clearly an
imperfect procedure, but it is consistent and seems to agree
well with the calculation of So by indirect means. A good,
straight line relationship was found between So and cover.
This is illustrated in Fig. 5, which also includes the equation
for the straight line.
An issue that has been studied by a number of researchers,
but is generally ignored by those developing theories of
ACI Structural Journal/March-April 2014

Fig. 6Schematic illustration of situation analyzed.


cracking, is the shape of the crack (that is, how the width
of a crack varies between the bar surface and the surface
of the concrete). An assumption generally seems implicit
in cracking formulae based on the classical theory that
the crack width at the concrete surface is the same as that
at the bar surface. There is now ample evidence that this
is not the case. The research evidence has been reviewed
in Reference6, and this shows that the cracks are tapered,
being much smaller near the bar surface than at the concrete
surface. Though the results are highly variable, it can be
concluded that the width at the concrete surface is at least
twice that near the bar surface.
SIMPLE ELASTIC ANALYSIS WITHOUT
INTERNALCRACKING
Initially, a simple two-dimensional analysis was performed
on an axially reinforced circular cross section prism to give
some idea of the stress and deformation conditions around a
bar. The axial symmetry greatly simplified the analysis. A
fully elastic analysis of the area surrounding a bar, assuming
no slip between the bar and the concrete, was considered.
The situation analyzed is illustrated in Fig. 6, in which ws is
the surface crack width.
Analyses were carried out using axisymmetric elements.
For the initial analysis, a 0.79 in. (20 mm) diameter f bar was
considered with 1.97 in. (50 mm) cover c. Square elements
of 0.20 in. (5 mm) were used, and a length from the free
face up to the fixed end of 5.91 in. (150 mm) was assumed.
A uniform stress of 14.51 kip/in.2 (100 MPa) was applied to
the free end of the bar. Figure 7 shows the stress distribution
in the concrete along the specimen obtained in two ways: the
stress in the concrete on the outer face (that is, the concrete
surface), and the average stress in the concrete. The average
stress was calculated by taking the force in the reinforcing
bar at each 0.20 in. (5 mm) section, deducting this from the
total applied force at the free end, and dividing this difference by the area of the concrete. In algebraic terms, if T is
the tension force applied at the free end, the bar area is As,
and the concrete area is Ac, the stress in the bar at any section
a distance x from the crack is ssx and the average stress in
the concrete is scx then, by equilibrium, because the force at
section x is T, the stress in the concrete is given by

ACI Structural Journal/March-April 2014

Fig. 7Variation in calculated stresses in concrete with


distance from free end. (Note: 1 MPa = 145 psi; 1 mm =
0.0394 in.; S*0s is surface spacing; S*0m is mean spacing.)
scx = (T Asssx)/Ac (1)
The average stress in the concrete was calculated in this
way because it corresponds to the method of concrete stress
calculation used in References 3 through 5. It is also effectively what is used in many theoretical derivations of crack
prediction formulas.
There are several interesting points which arise from
Fig.7. First, as might be expected, the surface stress is not
the same as the average stress, but is considerably lower over
almost the whole of the length analyzed. So for the surface is
thus different than So for the average stress, with the surface
value being considerably longer. Straight lines have been
drawn in passing through the origin and the point where
the calculated curves reach two-thirds of the homogeneous
stress. This is simply done to permit a simple visual comparison of the curves. It should also be noted that the surface is
actually in compression for the area closest to the crack face.
Second, the stress in the concrete does not actually reach the
stress calculated for a homogeneous section. Thus, there is
no absolutely clear definition of So, as is assumed in all theoretical equations for predicting cracking. This is not necessarily a critical point, but it may be worth remembering that
So is an effective value rather than an absolute value.
Figure 8 shows the variation in the deformed shape of the
free end over the height of the crack from the bar surface
to the specimen surface. Quantitative comparisons cannot
directly be made in this case, as the geometries of the available experimental specimens are somewhat different from
that analyzed. The deformation at the surface in Fig. 8 corresponds to a crack width of 0.0019 in. (0.047 mm).
Analyses have been carried out for different covers,
and Fig. 9 shows the calculated crack widths as a function
ofcover.
Crack width decreases with a decrease in cover. The
decrease is not linear as suggested from the experimental
data3-5; however, certain factors should be borne in mind.
The finite element analyses are elastic, and therefore, any
specimen having a geometrically similar cross section to the
one for which the crack width has been calculated will give a
crack width that can be calculated by direct scaling from the
previously calculated width. Thus, for example, the crack
399

Fig. 8Variation in calculated crack width with distance


from bar surface. (Note: 1 mm = 0.0394 in.)
width for any specimen with a value of c/f of 2.5 will lie on a
straight line joining the point for c/f = 2.5 to the origin. This
applies for any other value of c/f. Thus, all results for any
specimens with c/f in the range 1 to 2.5 will lie between the
two dashed lines drawn in Fig. 9. If a relatively random set
of tension specimens are analyzed, the result, when plotted
on a graph such as Fig. 9 will, to the engineering eye, be
accepted as giving a linear relationship between cover and
crack width with some relatively small level of scatter.
Figure 10 aims to make an approximate quantitative
comparison between calculated and experimental crack
widths. The test specimens, from Farra and Jacccoud,7 were
3.94 in. (100 mm) square and reinforced with a single axial
0.79 in. (20 mm) bar. The cover was thus 1.57 in. (40 mm),
and results from an analysis for 1.57 in. (40 mm) cover have
been used in the comparison. It should be remembered,
however, that the experimental specimens had a square
cross section, whereas this analysis considered a circular
cross section. It can be seen that the analysis underestimates
the maximum crack width by approximately 30%. This is to
be expected, as no account has been taken in this analysis
of internal failure (slip or internal cracking) which, as has
been discussed previously, occurs and reduces the stiffness
of the concrete in tension. This will be considered further in
a following section.
It seems likely that this initial simple analysis gives a
lower-bound indication of the deformation of the tensile
concrete, and hence, the estimate of the crack width. In
reality, concrete in tension is not absolutely elastic-brittle,

400

Fig. 9Variation in maximum surface crack width with


cover. (Note: 1 MPa = 145 psi; 1 mm = 0.0394 in.)

Fig. 10Maximum crack widths from Farra and Jaccoud


Specimens N-20-207 compared with finite element analysis.
(Note: 1 MPa = 145 psi; 1 mm = 0.0394 in.)
but will undergo some plastic deformation before rupture.
This will result in the actual deformation of the concrete
being greater than that calculated on the assumption of elasticity. Additionally, a short-term value has been used for Ec.
There is likely to be some creep during the test that would
result in further deformation of the concrete and steel, and
hence higher calculated crack widths. Depending on the
effect of these two factors, the calculated width could be
closer to the experimental values.
Overall, the analysis seems to have been very successful
in predicting the general qualitative behavior of axially reinforced specimens.

ACI Structural Journal/March-April 2014

Fig. 11Finite element model of axially reinforced tension


specimen with internal cracks.
ANALYSIS OF TENSION ZONE INCLUDING
INTERNAL CRACKS
As mentioned previously, some form of internal failure
takes place over the whole of the length So from a very early
load stage. There are two mechanisms that are commonly
proposed for this internal failure: slip along the steel-concrete interface, and internal cracks forming at an angle to the
axis of the bar. Slip is the most common mechanism invoked,
and has been used as the basis for many crack prediction
theories. The internal cracking mechanism was first illustrated by Goto,1 and was elaborated further by Beeby and
Scott,4 whose model which will be investigated in this paper
(it is believed that there are plenty of advocates of the slip
model who can, if they wish, carry out further modeling of
this option). It should be noted that the type of modeling
that will be attempted herein will not prove that a particular
model is the actual behavior; at best, it can simply show that
the particular model gives a reasonable simulation of reality.
Other models may exist that are as good or better. It could,
however, show that a particular model was unreasonable.
An axially symmetrical specimen was chosen for the
analysis. Initially, the analysis will be carried out on a specimen the same basic size as that used to produce the results
detailed in Fig. 7 and 8. It will now, however, also model a
number of internal cracks. The length of the model specimen
has been doubled to 11.81 in. (300 mm), partly because the
length So was expected to be greater than in the case with
no internal cracks, and partly because it was felt that the
length used in the previous analysis may have been slightly
short for the largest cover. The number of cracks, the angle
of the cracks to the bar axis, and the length of the cracks
is somewhat arbitrary, but is based on photographs from
Goto1 and Otsuka and Ozaka,8 and the analysis presented
in Beeby and Scott.4 Experimental work by Goto and others
suggest that internal cracks form at each rib on the bar. The
spacing used in the model analyzed herein is rather larger
than the rib spacing for reasons of practicality. An angle of
60degrees to the axis of the bar was chosen, although this
angle could only be approximately maintained as the basic
grid of 0.20in. (5mm) did not permit exactly 60 degrees
to be maintained for all cracks. Furthermore, the aim was
to use a linear decrease in crack height with distance from
the crack face. Again, the 0.20 in. (5 mm) grid meant that
this could only be achieved approximately. The elastic
finite element model used is illustrated in Fig. 11 and 12.
Others have carried out analyses aimed at studying internal
ACI Structural Journal/March-April 2014

Fig. 12Axisymmetric model created using OASYS-GSA


software.

Fig. 13Axisymmetric model: detail of assumed crack


pattern.
cracking (for example, Gerstle and Ingraffea9). The difference between those analyses and this analysis is that most
of the other analyses have attempted to study the development of the internal cracking as a function of applied load,
whereas in this study, to study a larger range of variables, a
crack pattern has been assumed.
The results from the analysis are shown in Fig. 13 to 18.
Figure 15 shows a number of interesting changes from the
stress results shown in Fig. 7 resulting from the element
without internal cracks. First, the relationship between the
average stress and distance from the crack is much closer to
linear. It now models more closely the experimental result
shown in Fig. 4.
The deformed shape of the free end (Fig. 16), which is
actually a tracing of the finite element analysis graphical
output (Fig. 14) with the elements and nodes removed for
clarity, is now possibly less similar to that obtained for the
specimen without internal cracks and to that of the experimental specimen, but it is still reasonable. The surface
401

Fig. 16Calculated deformation of specimen with seven


internal cracks. (Note: Dimensions in mm; 1 mm =
0.0394in.)
Fig. 14Axisymmetric model: deformed state. (Note:
1MPa = 145 psi.)

Fig. 15Concrete stresses calculated by finite element


analysis for specimen with seven internal cracks. (Note:
1MPa = 145 psi; 1 mm = 0.0394 in.)
deformation corresponds to a crack width of 0.0031 in.
(0.08 mm), 60% greater than that obtained for the specimen
without internal cracks. This agrees closely with the experimental results, as can be seen from Fig. 17, which shows the
same data as used for Fig. 10, but with the calculated line
shown for the analysis with internal cracks.
The predicted effect of cover is shown in Fig. 18, where it
can be seen that a reasonably linear relationship is predicted
between cover and surface crack width. There is, however,
some scatter in these results, possibly due to the difficulty of
modeling absolutely geometrically similar internal cracks in
the analyses for the various covers.
A further assessment of the performance of the model
can be performed by considering the work by Broms10 and
Beeby.11 Broms10 carried out a series of tests on short prisms
and measured the longitudinal extension at various stress
levels in the reinforcement. Results are presented in Reference 10 for a circular cross section specimen, 6 in. (152 mm)
in diameter and 8 in. (203 mm) long with a central 1 in.
(25mm) diameter bar. An elastic analysis has been carried
out for this specimen, and Fig. 19 shows the experimental
402

Fig. 17Maximum crack widths from Farra and Jaccoud


Specimen N-20-207 compared with finite element analysis,
with and without internal cracks. (Note: 1 mm = 0.0394 in.;
1 MPa = 145 psi.)

Fig. 18Predicted maximum crack width as function of


cover for analyses including internal cracks. (Note: 1 mm
= 0.0394 in.)
and calculated results for two levels of stress. The experimental results have been scaled from a figure in Broms
paper.10 It can be seen that, in this case, the experimental
results exceed the calculated results by approximately 20%.
The general trend of the results, however, is well reflected by
the calculations. This is not in absolute agreement, but it is
ACI Structural Journal/March-April 2014

Fig. 19Comparison of calculated and measured crack


widths for Broms Specimen T-C-5.10 (Note: 1 mm =
0.0394in.; 1 MPa = 145 psi.)

Fig. 20Comparison of calculated and measured overall


extension for specimen reported in Reference 11. (Note:
1mm = 0.0394 in.; 1 MPa = 145 psi.)

probably within the range that could be covered by judicious


adjustment of the model.
A cylindrical specimen was tested by Beeby11 where the
overall extension of a 5.91 in. (150 mm) diameter specimen
with an axial 0.87 in. (22 mm) bar at various distances from
the bar surface were measured. This has been analysed, and
the extensions scaled off the figure presented in Reference
11. Figure 20 shows the measured extensions compared with
the finite element calculations for two levels of steel stress.
Agreement between experiment and calculation is slightly
better than for Broms10 in Fig. 19, though the calculation,
again, tends to underestimate the measured results.
The introduction of the internal cracks (Fig. 12 to 14)
has, in general, resulted in an improved agreement with the
experimental behavior, and does suggest that the internal
cracking model is a reasonable model for the behavior of
tension zones. The analysis is clearly capable of further
refinement, and has only been carried far enough to demonstrate
its inherent reasonableness.

tions implicitly assume that this shear deformation is negligible. The analyses show that this is not so; the elastic shear
deformation of the concrete in the analyses reported herein
accounts for around two thirds of the crack width. Had other
material factors, such as creep or inelasticity of the concrete
in tension, been taken into account in the analyses, the shear
deformations and their contribution to the crack widths
would have been even greater. This substantial contribution
of the shear deformation of the cover concrete seems inescapable, and suggests that any approach to the prediction of
crack widths that ignore this are fundamentallyflawed.
In the second analysis, the output from the model with
internal cracks generally agreed with the experimental
data, where comparisons could be made. This suggests
that the internal cracking model of behavior can provide
a good model of cracking behavior. It does not prove that
the mechanism accommodating excess tensile strains above
those which the concrete can support in tension is internal
cracking; it shows that it is a viable alternative to the bondslip model, and should not be dismissed.
The analyses carried out are somewhat limited, and can be
considered to make a prima facie case for the reasonableness
of the internal cracking model rather than a fully developed
analytical study. Some of the more obvious limitations of the
model are given as follows.
Circular cross sections are analysed, not square or rectangular ones. Due to difficulties in manufacture, very
few circular specimens have been made and tested;
thus, it is not possible to compare the analytical results
rigorously with test results that are almost all from specimens with square or rectangular cross sections.
Location and size of internal cracks is somewhat arbitrary. As mentioned in a previous section, no attempt has
been made to refine the form of the internal cracking.
From the existing experimental evidence, the pattern
assumed seems reasonable, but it cannot be said to be
rigorously justified.
Rib pattern. By its nature, the axisymmetric analysis
assumes that the ribs are perpendicular to the bar axis
and extend round the full circumference of the bar.
This is not normally so for modern ribbed bars, where
the ribs tend to be staggered. There is also frequently

DISCUSSION
Elastic analyses
Two basic analyses have been described in this paper.
In the first, the concrete is considered to remain elastic
and uncracked, and complete bond is assumed between the
reinforcement and the concrete. In the second, a pattern of
internal cracking has been assumed, based on the findings
of Goto.1 In neither of the analyses is any form of bond-slip
relationship assumed; thus, bond-slip can have no influence
on the results obtained.
In the first analysis (without internal cracking), it was
expected that the predicted crack widths would be less than
obtained experimentally, and this proved to be the case.
Nevertheless, the analyses were not trivial, and the results
illustrate a significant point that has commonly been ignored.
If a shear stress is applied to a material, then shear strains
and displacements occur. Bond stress is simply a shear
stress, and therefore, the concrete surrounding a bar in the
region of a crack undergoes shear deformations. Though this
has not been shown to be explicitly stated, the classical theories of cracking that lie behind many crack prediction equaACI Structural Journal/March-April 2014

403

Table 1Comparison of crack widths at center


and near corner of axially reinforced tension
specimens (from Reference 12)

Mean values of w/e

5% values of w/e

Specimen

b/a

b/a

Z2

80

76

113

1.48

174

203

1.17

Z6

130

104

171

1.64

253

354

1.40

Z7

180

91

254

2.79

231

535

2.32

Z9

230

101

259

2.56

282

580

2.06

analysis of three-dimensional specimens (with the exception


of axisymmetric situations).
Issues relating to development of valid design
formulae for crack width prediction
It would be helpful for further discussion if a brief
outline is given of the development of crack theories and
codeprovisions.
The earliest developed theory of cracking assumed that
the widths of cracks accommodated slip between the bar
and the concrete. The theory ignored any contribution to
the crack width from the shear deformation of the cover
concrete. To develop equations, it required assumptions be
made about the development of the bond stresses as a function of slip. Many different assumptions were considered,
but all resulted in a basic equation of the form

Notes: B, a, and b, are in mm; 1 mm = 0.0394 in.; w/e is average crack width/average
surface strain.

w = kffcte/tr (2)

a longitudinal rib. The result of this is that, in reality,


the pattern of internal cracks may be considerably more
complicated than is modeled in this analysis.
The spacing of the internal cracks, which would be
expected to follow the spacing of the ribs, is too large
in the model.

where w is crack width (variously defined); k is a constant;


fct is the tensile strength of the concrete; e is strain (variously
defined); f is the bar diameter; r is reinforcement ratio (variously defined); and t is bond strength.
Many design provisions have been based directly on
this equation, including those in the CEB-FIP Model
Code1990.18
In 1965, Broms19 and Broms and Lutz20 published a radically different theory which assumed that the crack width
arose entirely from the shear deformation of the cover
concrete. The following formula was developed

Issues requiring further study


What is measured when crack widths are being investigated is the crack width on the surface. It is found that the
surface crack width is strongly dependent on where the
cracks are measured relative to the position of the reinforcement. If the cracks are measured at points on the surface
directly over the bar, they are found to be substantially
smaller than if they were measured, for example, close to
the corner of an axially reinforced prism. The variation is
less in situations where there are multiple bars and the crack
width over the bars is compared with that at mid-spacing.
This behavior is illustrated in Table 1, containing data from
Reference 12.
This effect seems perfectly rational for crack widths
resulting from shear deformation of the concrete; the shear
displacement will increase with increasing distance from the
bar in any direction. Because the corner of the specimens
used in Table 1 are further from the bar than a point directly
over the bar, the deformation will be greater, and the crack
width larger. This effect was recognized by Broms10 and in
the work carried out at the Cement and Concrete Association.12,13 It is implicitly included in the ACI code14 formula
and taken into account directly in the UK code.15
The analyses performed herein have been exclusively
concerned with members subjected to pure tension. There
is evidence that flexural members behave rather differently.
Studies by Beeby16 showed that, in shallow members, such
as slabs, the depth of the tension zone has a significant effect
on the crack width. This is explicitly taken into account in
the UK code,15 and is also recognized in Eurocode 2.17
These effects have not been investigated in this study
because the finite element package used did not permit the

404

wav = 2tes (3)


or
wmax = 4tes
where wav is the average crack width; wmax is the maximum
crack width; t is distance from the center of the bar to the
point on the surface where the crack width is considered; and
es is average strain of the steel.
For multiple bars, t was modified to te, an effective distance,
which is defined, for bottom cracks, as 3 (tb A), where tb is the
distance from the bottom of the beam to the center of the
lowest layer of bars, and A is the area of concrete immediately surrounding the tension reinforcement. This formula,
along with many others, was tested against the available
crack width data by Gergely and Lutz21 and shown to be the
best available at the time. The formula has formed the basis
of the ACI code14 crack width control provisions ever since.
At the same time as the work being carried out at by
Broms, Lutz and Gergely,19-21 a major series of tests were
carried out at the Cement and Concrete Association in the
UK.13 The first series of tests consisted of 105 beams, and
the results were with the publishers at the time that Broms19
paper appeared. The paper13 concluded that crack width
could be predicted by the formula

ACI Structural Journal/March-April 2014

wmax = 3.3acrem (4)

where wmax is the maximum crack width; acr is the distance


from the point where the crack width is being considered to
the surface of the nearest bar; and em is the average strain at
the level where the width is considered.
It can be seen that, while acr is slightly larger than t in
Broms19 equation and that the equation aims to predict
the maximum width rather than the average, the Cement
and Concrete Association13 and Broms19 basic formula
are almost the same. Like Broms19 theoretical approach,
the formula assumes that there is no bond failure or slip at
the bar-concrete interface, and that the crack width results
entirely from the deformation of the cover concrete. The
Cement and Concrete Association13 work was extended over
the following few years by Beeby,22-24 resulting in modifications to deal with bar spacing and the effect of the depth of
the tension zone. It was recognized that some mechanism
was necessary to accommodate the strains in the concrete
in excess of the tensile strain capacity of the concrete, and it
was proposed that the form of cracking identified by Goto1
provided that mechanism. The resulting formula was simplified somewhat, and has been used in UK codes15 since 1972.
This formula is
w = 3acrem/{1 + 2(acr c)/(h x)} (5)
where c is the cover to the face of the member where the
crack width is being considered; h is the overall depth of
the section; and x is the neutral axis depth (depth of the
compression zone).
During the same time, Ferry-Borges25 developed a
formula that combines the theoretical ideas behind the classical bond-slip model and the shear deformation models. His
formula is
sav = k1c + k2f/r (6)
wav = savem
where sav is the average crack spacing.
The term k1c is justified in Reference 25 by the following
statement: The need to consider the influence of the thickness of the cover, c, is easy to understand. In fact, even if
perfect bonding between concrete and steel existed, the
mean distance between cracks would not be zero but proportional to c.
This formula was adopted in the CEB Model Code 1978,26
and also in Eurocode 2.17
The object of this survey of approaches to crack width
calculation is to point out that there are three basic approaches
used in codes:
1. The assumption that the crack width arises purely from
slip. This is used in a number of codes and, notably, in
CEB-FIB Model Code 199018;
2. The assumption that there is no significant slip and
that the crack width arises entirely from the deformation of
the concrete around the bar. Bond and slip do not feature

ACI Structural Journal/March-April 2014

in formulae derived on this basis. Strains beyond those


supportable by concrete in tension are accommodated by a
reduction in the stiffness of the cover concrete by internal
cracking. This is the case for the UK15 and ACI codes14 and
the codes of any country that basically follow either British
or American practice; and
3. The assumption that the crack width arises from a
combination of slip and deformation of the concrete. This
is true of CEB-FIB Model Code 197824 and Eurocode 2,17
and will become the case for all countries which either adopt
Eurocode 2 or base their national codes on Eurocode 2.
From this investigation, it is apparent that a significant
proportion of the crack width is due to the deformation of the
concrete surrounding the bar. Therefore, Item 1 mentioned
previously is not tenable as a basis for a rational crack
prediction formula; Items 2 and 3 are tenable depending on
whether or not slip plays a significant role. The paper does
not aim to show which of these possibilities is closest to the
truth, merely that the concept of crack width being a function of the deformation of the cover concrete is reasonable.
CONCLUSIONS
In this paper, a number of simple elastic finite element
analyses of the concrete in tension surrounding tensile reinforcement, in members subject to pure tension, are described
and the results compared with the behavior of actual tension
specimens, as revealed by experiment. The study leads to the
following conclusions.
1. The analyses show clearly that cover should be an
important factor in any approach to the calculation of crack
widths. This effect arises from the shear distortion of the
concrete between the bar surface and the concrete surface.
2. Experimental results show clearly that there should be
some form of internal failure in the region of the bar over
the whole length over which the crack influences the stress
distribution. Two mechanisms have been proposed for this
internal failure: slip along the bar-concrete interface, and
internal cracks of the form proposed by Goto1 and elaborated by Beeby and Scott.4 In this paper, Beeby and Scotts
model is analyzed and is shown to describe the experimental
behavior of tension members effectively.
3. The results from this study and those described in
References 2 through 5 suggest the possibility of a model for
tension zone behavior under service loads which is, in principle, simpler and more all-embracing than current models.
This can be described by the following two assumptions:
a) Concrete in tension behaves in an elastic-brittle
manner;and
b) Force is transferred between ribbed reinforcing bars
and concrete by the mechanical action of the ribs. It is
assumed that there is no bond between the ribs and no slip
past theribs.
AUTHOR BIOS

John P. Forth is a Senior Lecturer in the School of Civil Engineering


at the University of Leeds, Leeds, UK. He received his BEng in civil and
structural engineering from the University of Sheffield, Sheffield, UK. He
received his PhD from the University of Leeds. His research interests
include serviceability and durability performance of reinforced concrete
and masonry structures.

405

The late Andrew W. Beeby was Emeritus Professor of Structural


Design in the School of Civil Engineering at the University of Leeds.
He received both his first degree and his PhD from London University,
London, UK. He research interests included the serviceability behavior of
concretestructures.

REFERENCES

1. Goto, Y., Cracks Formed in Concrete around Deformed Tension


Bars, ACI Journal, V. 68, No. 4, Apr. 1972, pp. 224-251.
2. Scott, R. H., and Gill, P. A. T., Short Term Distributions of Strain
and Bond Stress along Tension Reinforcement, The Structural Engineer,
V. 65B, No. 2, June 1987, pp. 29-43.
3. Beeby, A. W., and Scott, R. H., Insights into the Cracking and
Tension Stiffening Behavior of Reinforced Concrete Tension Members
Revealed by Computer Modeling, Magazine of Concrete Research, V. 56,
No. 3, Apr. 2004, pp. 179-190.
4. Beeby, A. W., and Scott, R. H., Cracking and Deformation of Axially
Reinforced Members Subjected to Pure Tension, Magazine of Concrete
Research, V. 57, No. 10, Dec. 2005, pp. 611-621.
5. Beeby, A. W., and Scott, R. H., Mechanisms of Long-Term Decay
of Tension Stiffening, Magazine of Concrete Research, V. 58, No. 5, June
2006, pp. 255-266.
6. Beeby, A. W., The Influence of the Parameter f/reff on Crack
Widths, Structural Concrete, V. 5, No. 2, 2004, pp. 71-83.
7. Farra, B., and Jaccoud, J.-P., Influence du beton et de larmature sur
la fissuration des structures en beton, Publication No. 140, Rapport des
essais de tirants sous deformation impose de court duree, Department de
Genie Civil, Ecole Polytechnique Federale de Lausanne, Nov. 1993.
8. Otsuka, K., and Ozaka, Y., Group Effects on Anchorage Strength
of Deformed Bars Embedded in Massive Concrete Block, Proceedings of
International Conference on Bond in ConcreteFrom Research to Practice, V. 1, Riga Technical University, Riga, Latvia, Oct. 1992, pp. 1.38-1.47.
9. Gerstle, W., and Ingraffea, A. R., Does Bond-Slip Exist? Concrete
International, V. 13, No. 1, Jan. 1991, pp. 44-48.
10. Broms, B., Theory of the Calculation of Crack Width and Crack
Spacing in Reinforced Concrete Members, Cement och Betong, No. 1,
1968, pp. 52-64.
11. Beeby, A. W., Concrete in the Oceans: Cracking and Corrosion,
CIRIA/UEG, Cement and Concrete Association, Department of Energy,
1978, 77 pp.
12. Beeby, A. W., A Study of Cracking in Reinforced Concrete
Members Subjected to Pure Tension, Technical Report No. 42.468,
Cement and Concrete Association, June 1972, 25 pp.

406

13. Base, G. D.; Beeby, A. W.; Read, J. B.; and Taylor, H. P. J., An
Investigation of the Crack Control Characteristics of Various Types of Bar
in Reinforced Concrete Beams, Research Report 18, Cement and Concrete
Association, London, UK, Dec. 1966, 31 pp.
14. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-11) and Commentary, American Concrete Institute,
Farmington Hills, MI, 2011, 503 pp.
15. BS8110-2, Structural Use of ConcretePart 2: Code of Practice for
Special Circumstances, British Standards Institution, BSI Milton Keynes,
London, UK, 1985, 62 pp.
16. Beeby, A. W., An Investigation of Cracking in Slabs Spanning One
Way, Technical Report TRA 433, Cement and Concrete Association, Apr.
1970.
17. Eurocode 2, Design of Concrete Structures Part 1-1: General
Rules and Rules for Buildings, CEN, EN 1992-1-1, Brussels, Belgium,
2004, 225 pp.
18. CEB-FIP, CEB-FIP Model Code (1990), Bulletin dinformation
No. 213/214, Comit Euro-Internationale du Bton (CEB), Lausanne, Switzerland, 1993, 437 pp.
19. Broms, B. B., Crack Width and Crack Spacing in Reinforced
Concrete Members, ACI Journal, V. 62, No. 10, Oct. 1965, pp. 1237-1256.
20. Broms, B. B., and Lutz, L. A., Effects of Arrangement of Reinforcement on Crack Width and Spacing of Reinforced Concrete Members, ACI
Journal, V. 62, No. 11, Nov. 1965, pp. 1395-1410.
21. Gergely, P., and Lutz, L. A., Maximum Crack Width in Reinforced Concrete Flexural Members, Causes, Mechanism, and Control of
Cracking in Concrete, SP-20, R. E. Philleo, ed., American Concrete Institute, Farmington Hills, MI, 1968, pp. 87-117.
22. Beeby, A. W., An Investigation of Cracking in the Side Faces of
Beams, Technical Report 42.466, Cement and Concrete Association, Dec.
1971, 11 pp.
23. Beeby, A. W., A Study of Cracking in Reinforced Concrete
Members Subjected to Pure Tension, Technical Report 42.468, Cement
and Concrete Association, June 1972, 25 pp.
24. Beeby, A. W., The Prediction of Crack Widths in Hardened
Concrete, The Structural Engineer, V. 57A, No. 1. Jan. 1979, pp. 9-17.
25. Ferry-Borges, J., Cracking and Deformability of Reinforced
Concrete Beams, V. 26, International Association for Bridge and Structural Engineering. Zurich, Switzerland, 1966, pp. 75-95.
26. CEB-FIP Model Code for Concrete Structures, Bulletin dinformation No. 125, Comit Euro-International du Beton, Paris, France, Apr.
1978, 460 pp.

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S36

Flexural Capacity of Fiber-Reinforced Polymer


Strengthened Unbonded Post-Tensioned Members
by Fatima El Meski and Mohamed Harajli
Experimental and analytical investigations were carried out
for evaluating the nominal moment capacity of unbonded posttensioned members when strengthened using external fiberreinforced polymer (FRP) composites. In the experimental part of
the study, 36 simply supported specimens were tested to failure. The
main test parameters included area of internal tension reinforcement, area of external FRP reinforcement, span-depth ratio of the
member, profile of the prestressing tendons, and type of concrete
structural system. In the analytical part of the study, a designoriented procedure for evaluating the nominal moment capacity
of FRP-strengthened post-tensioned members with internal or
external unbonded tendon systems is developed. The procedure
is consistent with the approach proposed in ACI Committee 440
report for reinforced concrete or bonded prestressed concrete
members, and is applicable for both simply supported and continuous members. The accuracy of the design approach was verified
by comparing it with the test results of the experimental part of this
investigation.
Keywords: fiber-reinforced polymer; flexure; prestressing; post-tensioning;
strengthening; unbonded tendons.

INTRODUCTION AND BACKGROUND LITERATURE


The flexural capacity of bonded prestressed concrete
members can be evaluated in accordance with the general
ACI Building Code1 approach and the guidelines recommended by ACI Committee 4402 by accounting for the effect
of fiber-reinforced polymer (FRP) reinforcement as follows

M n = Aps f ps d p b1c / 2 + As fs ( d b1c / 2 ) + f A f E f e f d f b1c / 2

) (1)

where

fps = F(eps) (2)

fs = Eses fy (3)

dp c
e ps = e pe + e ce + e cu
c

e ce =

(4)

2
1 Aps fse Aps fse e
+

Ec Ac
Ig

ACI Structural Journal/March-April 2014

(5)

d c
e s = e cu
c

(6)

d f c
e bi e fd
e f = e cu
c

(7)

fc
0.9e fu
nf Ef t f

(8)

e fd = 0.41

c=

Aps f ps + As fs + A f ( f f = E f e f )
a1 fc b1b

(9)

in which c is the neutral axis depth of the section at nominal


flexural strength; Aps and dp are area and depth of the
unbonded prestressing steel (PS); Af is area and df is depth
of the FRP reinforcement; and As is area of bonded ordinary
reinforcing steel (RS) at the section under consideration. For
slabs, the areas Aps, As, and Af are per unit width b of the slab
section. ce is precompression strain in concrete at the level
of the prestressed tendons; Ac and Ig are area and moment
of inertia of the gross section, respectively; e is eccentricity
of the tendons; Ec, Es, and Ef are modulus of elasticity of
concrete, steel, and FRP reinforcement, respectively; f and ff
are strain and stress in the FRP reinforcement, respectively;
fu is rupture strain of the FRP reinforcement; F is the material stress-strain relationship of the prestressing reinforcement; pe, fse, ps, and fps are effective strain, effective stress,
strain, and stress at ultimate, respectively, in the prestressing
steel (pe = fse/Eps, where Eps is the modulus of elasticity of
the prestressing steel); s, fs, and fy are strain, stress, and
yield stress, respectively, for the ordinary RS; fc is concrete
cylindrical compressive strength; cu is usable concrete
strain at compression failure (equal to 0.003 in accordance
with ACI 318-111); 1 = 0.85 corresponding to a concrete
strain in the outermost compression fiber equal to cu; 1 is
concrete strength factor defined in section 10.2.7.3 of the
ACI Building Code1; bi is initial substrate strain at which
the FRP was applied for strengthening (ACI Committee
ACI Structural Journal, Vol. 111, No. 2, March-April 2014.
MS No. S-2012-168.R2, doi:10.14359.51686565, was received January 10, 2013,
and reviewed under Institute publication policies. Copyright 2014, American
Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published ten months from this journals date if the
discussion is received within four months of the papers print publication.

407

4402), determined using elastic cracked section analysis,


considering all loads that will be on the member during the
FRP installation; fd is strain in the FRP reinforcement at
which FRP debonding failure occurs; nf and tf are number of
FRP layers and thickness per one layer, respectively; and f
is the FRP strength-reduction factor recommended by ACI
Committee 440,2 which is taken equal to 0.85 for flexure.
To calculate the nominal moment capacity of
FRP-strengthened reinforced concrete (RC) or bonded PC
members, ACI Committee 4402 recommends using a trialand-error procedure for estimating the neutral axis depth
c until the requirements of strain compatibility and force
equilibrium across the depth of the critical section are satisfied (Eq. (9)). Also, two modes of flexural failure are recognized by ACI Committee 440: 1) concrete crushingthat is,
when the strain in the outermost concrete compression fiber
reaches cu before FRP failure; and 2) FRP failure before
concrete crushing. FRP failure could occur either by FRP
rupture, cover delamination, or FRP debonding. Accordingly, ACI Committee 4402 limits, conservatively, the strain
in the FRP reinforcement at which FRP failure takes place to
the debonding strain fd (Eq. (8)). When the strain in the FRP
reaches its limiting strain fd before the concrete compression strain reaches cu, the concrete compression strain cu
in Eq. (2) through (9) should be replaced by its actual value
expressed as a function of fd as follows

c
e c = e fd + e bi

df c

(10)

The values of 1 and 1, which are now different from


their values when c = cu, can be estimated with reasonable
accuracy using the following expressions recommended by
ACI Committee 440,2 which were derived assuming a parabolic relationship for the stress-strain curve of concrete in
compression

4 e c e c
6 e c 2 e c

(11)

3e c e c (e c )2
3b1 (e c )2

(12)

b1 =

a1 =

where c is the strain corresponding to fc, which can either


be taken equal to 0.002 or calculated more accurately as
c = 1.7fc/Ec.
RESEARCH SIGNIFICANCE
An experimental study was carried out and a designoriented approach was developed for evaluating the nominal
flexural capacity Mn of unbonded PC members when
strengthened using external FRP composites. The design
approach builds on a previous model generated for evaluating the stress in unbonded tendons at ultimate in simply
supported or continuous members, and is consistent with the

408

Fig. 1 FRP-strengthened continuous unbonded member


with multi-collapse mechanisms.
ACI Committee 4402 guidelines for calculating Mn for FRP
strengthened bonded PC or RC members. The approach is
validated for simply supported members by comparing with
the test results generated in the experimental part of this
investigation.
STRESS IN UNBONDED TENDONS AT ULTIMATE
While the approach recommended by ACI Committee
4402 for estimating Mn for FRP strengthened RC or bonded
PC members is simple because it relies on commonly
adopted and familiar principles, it is not as straightforward
for PC members with unbonded tendons. In PC members
with internal or external unbonded tendons, because the
unbonded steel slips relative to the surrounding concrete,
the strain or stress fps in the prestressing steel that develops
at nominal flexural strength relies on the deformation of
the member as a whole. This makes the evaluation of the
corresponding strain and stress member, rather than section,
dependent.
Several design approaches are available for predicting
the stress in unbonded tendons at ultimate.1,3,4 A comparative evaluation of these approaches has been discussed in
detail elsewhere.5 Using plastic analysis and the concept
of collapse mechanism in continuous members, as well as
idealization of the curvature distribution along the span
lengths at nominal flexural strength as shown in Fig. 1 (for
c = cu, and Af = 0.0) and further incorporating an empirically derived expression for the equivalent plastic hinge
length for unbonded members given as5 lp = (20.7/f + 10.5)c,
Harajli6 developed the following general and yet elegant
expression for calculating the strain ps in unbonded tendons
at ultimate

dp

c
e ps = e pe + f ps N p e cu
N p (e cu e ce ) (13)
La
La

where
ACI Structural Journal/March-April 2014

20.7

Np =
+ 10.5 n p + + 10.5n p
f

(14)

in which c is the neutral axis depth at the section under consideration; ce is calculated using Eq. (5) by neglecting the effect
of the secondary moment due to prestressing in continuous
members; La is the total length of tendons between anchorages; Np (Eq. (14)) is a parameter that combines the effect
of member continuity and type of applied load; f = for a
single concentrated load, 3.0 for two-third point loads, and
6.0 for uniform load application, respectively; np+, np are the
number of positive and negative plastic hinges, respectively,
that develop in the process of forming a collapse mechanism; and ps is a stress-reduction factor taken equal to 0.7.
Note that because the live load in buildings and bridges is
seldom a concentrated load, and also because the dead load
is uniformly distributed, the multiplier (20.7/f + 10.5) of the
positive number of plastic hinge(s) np+ can always be set
equal to 14.0 to correspond to uniform load application.6
Equations (13) and (14) were developed with the perspective that the calculation of the stress in unbonded tendons
at ultimate in continuous members differs from one critical
section to another. That is, the stress at a critical section
depends on the pattern of applied load and the consequent collapse mechanism that would potentially develop
for producing the maximum factored design moment at
that section. For the hypothetical case of Fig. 1, in which
collapse mechanisms are assumed to form in all spans:
np1+ = 2.0, np2 = 2.0, and np3+ = 2.0. In actual design,
however, the numbers of plastic hinges np+ and np are
obtained from the collapse mechanisms that develop when
loading the minimum number of spans for producing
maximum moment at the section under consideration. For
instance, in simply supported members, one span is loaded,
and hence np+ = 1.0, np = 0.0, and Np = 14.0. For continuous
members, collapse mechanisms (a), (b), (c), (d), and (e) in
Fig. 2, and corresponding values of Np are recommended6
for computing the tendon stress fps and the nominal moment
capacity at the maximum positive moment section in exterior spans; negative moment section at the interior support
of a two-span member; maximum positive moment section
in interior spans; negative moment section at the first interior support; and negative moment section at the remaining
interior supports of members with more than two spans,
respectively.
PROPOSED APPROACH FOR COMPUTING MN IN
FRP STRENGTHENED UNBONDED MEMBERS
Using Eq. (13) but neglecting the precompression strain
ce due to its minor effect on the tendon stress in unbonded
members, particularly when compared with bonded
members, and considering that the flexural strength may
be controlled by FRP failure at which c cu (Fig. 1), the
following expression is recommended for computing the
strain ps in unbonded tendons of FRP-strengthened simply

ACI Structural Journal/March-April 2014

Fig. 2Loading pattern and corresponding values of continuity parameter Np for continuous members.
supported or continuous members at nominal flexural
strength

dp c
e ps = e pe + f ps N p e c
La

(15)

Recognizing that the stress in the prestressing steel


seldom exceeds yield and limiting the corresponding stress
to 0.95fpy6 allows the use of a linear relationship between the
stress and strain in the prestressing steel, that is, fps = Epsps,
leading to

f ps = fse +

f ps N p E ps e c
c
1 0.95 f py
La / d p d p

(16)

The force equilibrium across the depth of the critical


section, assuming rectangular section or rectangular section
behavior, is expressed as

Apsfps + Asfs + AfEfef = a1fcbb1c (17)

It should be noted that for unbonded members without


FRP reinforcement (that is, Af = 0.0 and 1 = 0.85), the value
of c from Eq. (17) can be integrated in Eq. (16) to produce
a direct expression for computing fps in unbonded members
at ultimate.6
Using Eq. (15), (16), and (17), the following step-by-step
procedure can be adopted for evaluating the nominal moment
capacity Mn of FRP strengthened unbonded post-tensioned
members.

409

Case IFlexural capacity controlled by concrete


crushing
Step 1Assume that the nominal flexural strength at a
critical section is controlled by concrete crushingthat is, f
calculated using Eq. (7) is less than or equal to fd (Eq.(8)).
This implies that c = cu, 1 = 0.85, and 1is as defined in
section 10.2.7.3 of the ACI Building Code.1 Replacing fps
from Eq. (16) into Eq. (17) and assuming that the RS yields,
that is, fs = fy, leads to the following quadratic equation for
calculating the neutral axis depth at nominal flexural strength
at the section under consideration
c=

B + B2 + 4 AC
2A

(18)

where

A = 0.85b1 fcb +

f ps N p Aps E ps e cu
La

(19a)

f ps N p E ps e cu d p

B = Aps fse +
+ As f y A f E f (e cu + e bi )
La

(19b)

C = AfEfecudf (19c)

in which Np is as defined previously (Fig. 2). For simply


supported members, Np = 14.0.
Step 2Check if f (Eq. (7)) is indeed fd. Also check
if the strain s in the RS is larger than the yield strain y. If
f fd while s is less than y, repeat Step 1 for recalculating more accurately the neutral axis depth using Eq. (18)
in which the coefficients A, B, and C are revised by substi

d c

tuting fs = Es e s = e cu c for fy in Eq. (17).


Step 3Calculate fps from Eq. (16) corresponding to
c = cu, and fs from Eq. (3), and hence calculate the nominal
moment capacity Mn from Eq. (1).
Case IIFlexural capacity controlled by FRP
failure
Step 4If f calculated from Step 2 is greater than fd
(Eq.(8)), then FRP failure occurs before the strain c reaches
cu. In this case, the strain f in the FRP reinforcement is equal
to fd, and hence, a trial-and-error procedure for calculating
c, as described in the next steps, becomes more appropriate.
Step 5 Using the value of f = fd, together with an initial
assumed value of c, calculate the concrete strain c at the top
concrete compression fiber from Eq. (10) and calculate the
stress in the prestressing steel (Eq. (16)) and the stress in the
RS from Eq. (3) and (6) by replacing c for cu.
Step 6Check equilibrium of forces using Eq. (17) in
which f = fd, and 1 and 1 are as calculated from Eq. (11)
and (12).

410

Step 7Repeat Steps 5 and 6 by revising the neutral axis


depth c until the requirement of force equilibrium in Eq. (17)
is satisfied within some degree of tolerance.
Step 8Calculate the nominal moment capacity Mn at the
section under consideration using Eq. (1) in which f = fd.
Step 9Check if Mn Mu, where Mu is the load-factored
applied moment at the section under consideration obtained
by loading the spans (pattern loading) for producing
maximum moment at the section under consideration.
The strength-reduction factor is taken in accordance
with the ACI Building Code1 approach using the relationship
between the net tensile strain in the tension reinforcement
and the neutral axis depth c at nominal flexural strength as
follows

= 0.90 for c/de 0.38

(20a)

= 0.65 for c/de 0.6

(20b)

c
f = 0.65 + 0.25 2.73 4.55 for
de

(20c)


0.38 c / de 0.6

where de is the equivalent depth of the tension reinforcement


(PS, RS, and FRP laminates) expressed as
de =

( Aps f ps d p + As fs d + A f E f e f d f )
Aps f ps + As fs + A f E f e f

(21)

EXPERIMENTAL STUDY
Twenty-four unbonded post-tensioned specimens were
tested to evaluate their nominal flexural strength before and
after FRP strengthening. An additional six bonded posttensioned and six RC specimens were also tested for
comparison. All 36 specimens were simply supported over
a 3.0m (9.84 ft) span. Dimensions and reinforcement layout
are given in Fig. 3. Eighteen of the specimens had a 150mm
(5.9 in.) wide by 250 mm (9.8 in.) deep cross section and
a span-depth ratio (depth to center of tension steel) of 15,
representing beam members, and the remaining 18 had
a 360 mm (14.2 in.) wide by 120 mm (4.7 in.) deep cross
section and a span-depth ratio of 35, simulating one-way
slab members. The specimen designation, along with areas
and depths of reinforcement and other pertinent details
and design properties, are summarized in Table 1. The test
parameters included, in addition to the span-depth ratio of
the member, area of internal prestressed reinforcement for
the PC specimens or area of ordinary tension reinforcement
for the RC specimens; area of external FRP reinforcement;
and tendon profile. Two tendon profiles were selected for
each set of specimens: one horizontal and one parabolic.
In the specimens designation provided in Table 1, the first
letter U stands for unbounded, B for bonded, and R for reinforced. The second letter B stands for beam, while S stands
ACI Structural Journal/March-April 2014

Fig. 3Typical dimensions and reinforcement details of the beam and slab specimens. (Note: dimensions in mm; 1 mm = 0.039 in.)
for slab. The numbers 1 and 2 following the second letter
designate two different levels of PS or RS areas or ratios.
Letters H and P designate horizontal and parabolic tendon
profile, respectively, and F1 and F2 denote two different
levels (areas or layers) of external FRP reinforcement. The
parabolic tendon profile in all beam and slab specimens had
zero eccentricity at the support.
The prestressing reinforcement consisted of seven-wire
strands having 7.9 and 9.5 mm (5/16 and 3/8 in.) diameter,
with an area of 37.5 and 52.0 mm2 (0.058 and 0.08 in.2),
and actual ultimate strength fpu of 1958 and 1978 MPa (284
and 287 ksi), respectively. Generated from coupon tests, the
actual stress-strain (fps ps) behavior of the two sizes of the
prestressing steel were best reproduced using the following
relationship7

f ps = E ps e ps Q +

1Q

1/ N
E e N
1 + ps ps
Kf py

ACI Structural Journal/March-April 2014

(22)

where fpy = 1670 MPa (242 ksi), Eps = 195,130 MPa


(28,400ksi), N = 14.84, K = 1.0, and Q = 0.0357 for the
7.9mm (5/16 in.) strands; and fpy = 1690 MPa (245 ksi), Eps
= 194,440 MPa (28,200 ksi), N = 12.1, K = 1.011, and Q =
0.0301 for the 9.5 mm (3/8 in.) strands.
All ordinary reinforcing bars (except the 6 mm [0.25in.]
bars) were deformed Grade 60 steel with actual yield strength
as given in Table 1. The FRP composite used for strengthening consisted of carbon fiber-reinforced polymer (CFRP)
flexible sheets having unidirectional carbon fabric with glass
cross-fiber for added strength and fabric stability during
installation. The design thickness, the modulus of elasticity, and the ultimate tensile strain of the fibers are 1 mm
(0.039in.), 95,800 MPa (13,895 ksi), and 1%, respectively.
Before casting the specimens, the steel cages were
instrumented with electric strain gauges and then placed in
plywood formwork ready for concrete casting. The ducts
for the post-tensioned steel consisted of galvanized flexible
tubes 20 mm (0.79 in.) in diameter. One strand was placed
in each duct. For the bonded post-tensioned slab and beam
specimens, cement-based grout was injected inside the ducts
after the tendons were stressed to provide bond between the
strands and concrete. The grout mixture was prepared using
Type I portland cement, and proportioned in accordance
411

Table 1Summary of test parameters


Prestressing steel
Concrete
system

Reinforcing steel

FRP

Beam
Unbonded post-tensioned

Av

fy,
MPa

Af,
mm2
(nf)

fc,
MPa

8 at 150

560

42

220

8 at 150

612

150 (1)

36

2 (8 mm)

220

8 at 150

612

300 (2)

36

815

2 (8 mm)

220

8 at 150

560

42

200

971

2 (8 mm)

220

8 at 150

612

150 (1)

36

1 (5/16 in.)

200

781

2 (8 mm)

220

8 at 150

612

300 (2)

37

Horizontal

2 (3/8 in.)

200

778

2 (8 mm)

220

2 (8 mm)

8 at 150

560

42

UB2-H-F1

Horizontal

2 (3/8 in.)

200

924

2 (8 mm)

220

2 (8 mm)

8 at 150

612

150 (1)

36

UB2-H-F2

Horizontal

2 (3/8 in.)

200

896

2 (8 mm)

220

2 (8 mm)

8 at 150

612

300 (2)

37

UB2-P

Parabolic

2 (3/8 in.)

200

836

2 (8 mm)

220

2 (8 mm)

8 at 150

560

42

UB2-P-F1

Parabolic

2 (3/8 in.)

200

936

2 (8 mm)

220

2 (8 mm)

8 at 150

612

150 (1)

36

UB2-P-F2

Parabolic

2 (3/8 in.)

200

923

2 (8 mm)

220

2 (8 mm)

8 at 150

612

300 (2)

37

US1-H

Horizontal

2 (5/16 in.)

85

927

2 (8 mm)

92.5

560

42

US1-H-F1

Horizontal

2 (5/16 in.)

85

917

2 (8 mm)

92.5

612

150 (1)

36

US1-H-F2

Horizontal

2 (5/16 in.)

85

964

2 (8 mm)

92.5

612

300 (1)

36

d,mm

As,
top

2 (8 mm)

220

962

2 (8 mm)

200

963

1 (5/16 in.)

200

Parabolic

1 (5/16 in.)

UB1-P-F2

Parabolic

UB2-H

Specimen
label

Tendon
profile

Aps

UB1-H

Horizontal

UB1-H-F1

dp, mm

fse,
MPa

As,
bottom

1 (5/16 in.)

200

813

Horizontal

1 (5/16 in.)

200

UB1-H-F2

Horizontal

1 (5/16 in.)

UB1-P

Parabolic

UB1-P-F1

Slab
Beam
Slab

Bonded post-tensioned

Beam
Slab

Reinforced concrete

US1-P

Parabolic

2 (5/16 in.)

85

886

2 (8 mm)

98.5

560

42

US1-P-F1

Parabolic

2 (5/16 in.)

85

949

2 (8 mm)

98.5

612

150 (1)

36

US1-P-F2

Parabolic

2 (5/16 in.)

85

971

2 (8 mm)

98.5

612

300 (1)

37

US2-H

Horizontal

3 (3/8 in.)

85

804

2 (8 mm)

92.5

560

42

US2-H-F1

Horizontal

3 (3/8 in.)

85

912

2 (8 mm)

92.5

612

150 (1)

36

US2-H-F2

Horizontal

3 (3/8 in.)

85

858

2 (8 mm)

92.5

612

300 (1)

37

US2-P

Parabolic

3 (3/8 in.)

85

831

2 (8 mm)

98.5

560

42

US2-P-F1

Parabolic

3 (3/8 in.)

85

921

2 (8 mm)

98.5

612

150 (1)

36

US2-P-F2

Parabolic

3 (3/8 in.)

85

916

2 (8 mm)

98.5

612

300 (1)

37

BB2-P

Parabolic

2 (3/8 in.)

200

884

2 (6 mm)

220

2 (6 mm)

8 at 150

37

BB2-P-F1

Parabolic

2 (3/8 in.)

200

894

2 (6 mm)

220

2 (6 mm)

8 at 150

150 (1)

37

BB2-P-F2

Parabolic

2 (3/8 in.)

200

885

2 (6 mm)

220

2 (6 mm)

8 at 150

300 (2)

37

BS2-P

Parabolic

3 (3/8 in.)

85

970

2 (8 mm)

98.5

37

BS2-P-F1

Parabolic

3 (3/8 in.)

85

915

2 (8 mm)

98.5

150 (1)

37

BS2-P-F2

Parabolic

3 (3/8 in.)

85

892

2 (8 mm)

98.5

300 (1)

37

RB2

2 (16 mm)

220

8 at 100

530

37

RB2-F1

2 (16 mm)

220

8 at 100

530

150 (1)

37

RB2-F2

2 (16 mm)

220

8 at 100

674

300 (2)

37

RS2

4 (12 mm)

100

555

37

RS2-F1

4 (12 mm)

100

555

150 (1)

37

RS2-F2

4 (12 mm)

100

624

300 (1)

37

Note: 1 in. = 25.4 mm; 1 MPa = 0.145 ksi; 1 mm2 = 0.0016 in.2.

with the ACI Building Code1 with a water-cement ratio of


0.40.
The CFRP sheets were attached to the bottom tension
face of the beam and slab specimens in accordance with
412

the manufacturers recommendations and in compliance


with the ACI Committee 4402 recommendation for securing
proper development length. No particular measures were
taken to improve bond strength between the FRP and the
ACI Structural Journal/March-April 2014

substrate. One or two layers of 150 mm (5.9 in.) wide FRP


sheets were applied for the beam specimens, while only one
layer of 150 or 300 mm (5.9 or 11.8 in.) wide sheets was
applied for the slab specimens (Table 1).
All specimens were tested in four-point bending using two
symmetrical concentrated loads spaced a distance equal to
1/6 the span length, or 500 mm (19.7 in.) apart (Fig. 3). It
should be noted that because the increase in tendon strain/
stress (above effective prestrain/prestress) of unbonded
members due to increase in applied load depends on the
overall deformation of the member or elongation of the
tendon between the anchorages, the ultimate tendon stress
and flexural capacity of unbonded members are influenced
by the geometry of applied load.8 Two-point loads spaced at
1/6 the span length were found analytically8,9 to be equivalent to uniform load application for predicting the ultimate
tendon stress and moment capacity of unbonded post-tensioned members.
To replicate actual conditions of concrete flexural
members that require strengthening or repair, all control
and strengthened specimens were first subjected to cyclic
loading consisting of six cycles before the FRP application, and another six cycles after FRP application, ranging
between a minimum load Pmin and a maximum load Pmax,
simulating dead load and dead plus live load, respectively.
The loads Pmin and Pmax were set at 30 and 70%, respectively,
of the calculated nominal load capacity of the specimens.
For the control or unstrengthened specimens, the cyclic
loading stage was followed immediately by a stage of monotonically increasing load until complete flexural failure of
the specimens. The specimens that were strengthened
using CFRP were first subjected to the same cyclic loading
protocol as the control specimens, and then unloaded to
prepare them for CFRP application. Following at least
7days of CFRP application, the strengthened specimens
were subjected to a loading protocol consisting of cyclic
loading and monotonically increasing load to failure, similar
to the controlspecimens.
Test measurements included strains and stresses in the
prestressing strands, CFRP laminates, and reinforcing bars
of the RC specimens within the constant moment region
close to midspan; applied load; and deflection. Crack
patterns were also monitored throughout the test for each
specimen. The strains were measured using electric strain
gauges, while deflection was measured using a linear voltage
differential transformer (LVDT). The test results were automatically collected and recorded using a data acquisition and
control system.
DISCUSSION OF RELEVANT TEST RESULTS
Typical photos of the specimens at the conclusion of the
test are shown in Fig. 4. Representative modes of failure
and cracking patterns at nominal strength for the unbonded
PC beam specimens in comparison with the bonded PC and
RC beam specimens are provided in Fig. 5. Flexural failure
for the various beam and slab specimens occurred either
by concrete crushing or by FRP debonding or fracturing.
Relevant experimental results at nominal flexural strength
including load capacity, deflection, stress in the prestressing
ACI Structural Journal/March-April 2014

Fig. 4Typical photos of specimens at conclusion of test.


steel, strain in the CFRP reinforcement, and mode of failure
are all summarized in Table 2. Representative variations of
deflection, FRP strain, and stress increase in the prestressing
steel above effective prestress fse with applied load are shown
in Fig. (6) through (9), respectively.
The crack patterns for the unbonded beam specimens
(Fig. 5) and the slab specimens (not shown for brevity) were
similar to those developed in the companion bonded specimens. The spreading of cracks outside the constant moment
region increased with the use of FRP reinforcement and
as the area of FRP reinforcement increased, and was most
significant for the RC specimens (Fig. 5).
It is clear from the test data summarized in Table 2 that
the use of FRP reinforcement significantly increased the
moment capacity of unbonded post-tensioned members. The
corresponding increase grew higher as the area of FRP reinforcement increased, and varied between 24 and 105% for
the beam specimens and between 21 and 126% for the slab
specimens of the current investigation. As would be expected,
however, the increases in load capacities were accompanied
with reductions in ductility or ultimate deformation capacities, which were most notable for the beam specimens. The
reductions in deformation capacity, which are attributed in
413

were significantly larger than the stresses developed in the


unbonded specimens.
Based on the experimental results and observations of
the current investigation, it was obvious that the cracking
and crack patterns, increase in load capacities, reduction in
deformation capacities, and modes of flexural failure of the
unbonded PC specimens as a result of FRP strengthening
were quite similar to those of the bonded PC and RC specimens. More test data and a detailed discussion of test results
are reported elsewhere.10

Fig. 5Typical comparison of crack pattern for unbonded


PC, bonded PC, and RC beam specimens.
part to the FRP debonding mode of flexural failure, were
almost identical for the slab or beam specimens reinforced
with Level F1 and F2 FRP reinforcement (Table 2).
The FRP strain/stress developed at flexural failure (Fig.8
and Table 2) generally decreased as the area of FRP reinforcement increased within the specimens of the same test
series, or as the area of prestressing reinforcement increased.
The overall average FRP strains at ultimate for the combined
beam and slab specimens strengthened using the two
different areas or levels (F1 and F2) of FRP reinforcement
were 6260 to 5400 for the unbonded PC, 6875 to 4970
for the bonded PC, and 7608 to 5419 for the RC specimens, respectively.
Being unbonded, the stress in the prestressing steel at
nominal flexural strength was below yield for all specimens.
The corresponding strain (Table 2) decreased as the area of
FRP reinforcement increased. On the other hand, the strains
and stresses in the prestressing steel for all companion
bonded PC beam and slab specimens exceeded yield and

414

COMPARISON OF ANALYTICAL PREDICTIONS


WITH TEST DATA
Table 2 shows comparisons of the predictions of the
proposed design-oriented approach developed in this study
for FRP-strengthened unbonded PC members together with
the predictions of the ACI Committee 4402 approach for
bonded PC and RC members against the test results generated in the experimental part of this investigation. These
include stress fps in the prestressing steel, strain f in the FRP
reinforcement, mode of flexural failure, and nominal moment
capacity Mn. In calculating the stress in the prestressing steel
and the nominal moment capacity, the reduction factors ps
and f were both set equal to 1.0. Also, because the load at
which the FRP was applied was very small (equal to the selfweight of the specimen), the initial substrate strain bi was
taken equal to zero.
Figure 10 shows predicted stress increase fps (fps=fpsfse)
and total stress fps in the prestressing steel at ultimate versus
test results for the unbonded prestressed beams and slab
specimens of the current investigation. The data are superimposed on predicted versus test results of a collection of
unbonded members (without FRP reinforcement) compiled
and reported previously.5,6 The compiled data correspond
to internally or externally post-tensioned simply supported
members, and continuous members having two or three
spans, and loaded with two-point or single-point loads,
respectively. Figure 11 shows predicted versus measured
nominal moment capacities for the 36 beam and slab specimens tested in this investigation.
Given the inevitable scatter associated with the prediction
of the stress in unbonded tendons at ultimate, it can be seen
from Fig. 10 that the proposed analytical approach predicted
the test data with reasonable accuracy. More conservative predictions can be obtained by using the stress reduction parameter ps = 0.7 proposed for design purposes. As
expected, the stress predictions for the bonded members
were more accurate than the unbonded members. The
average ratio of the test-to-predicted tendon stress were 1.10
(standard deviation [SD] = 0.12) for the unbonded specimens, and 1.02 (SD = 0.04) for the bonded specimens. It
should be noted that due to the presence of 28 mm ordinary steel bars, which are required as minimum bonded reinforcement in accordance with the ACI Building Code,1 the
control unbonded PC specimens developed well-distributed
cracks along their length as opposed to the development of
few cracks or concentration of deformation at a single crack
that normally occurs in members with an unbonded tendon
system.5 In other words, the equivalent plastic hinge length
ACI Structural Journal/March-April 2014

Table 2Summary of test results and analytic predictions


u
Pu
(kN) (mm)

FRP Strain f

fps (MPa)

Mode of failure

Exp./
Exp./
Exp. Analysis Analysis Exp. Analysis Analysis

Specimen

Exp.

Exp.

UB1-H

42.3

64

1567

1253

1.25

UB1-H-F1

66.9

31

1303

1250

1.04

7122

UB1-H-F2

86.9

33

1223

1290

0.95

UB1-P

46.8

81

1669

1255

UB1-P-F1

66.3

35

1413

1259

Nominal moment Mn (kN-m)


Exp./
Analysis Analysis

Experiment

Analysis

Exp.

Concrete crushing

Concrete crushing

26.4

20.8

1.27

7948*

0.90

FRP debonding

FRP debonding

41.8

46.5

0.90

5378

5620*

0.96

FRP debonding

FRP debonding

54.3

55.6

0.98

1.33

Concrete crushing

Concrete crushing

29.3

20.8

1.41

1.12

4556

7948*

0.57

FRP debonding

FRP debonding

41.4

46.5

0.89

UB1-P-F2

89.0

36

1102

1098

1.00

5604

5698

0.98

FRP debonding

FRP debonding

55.6

55.2

1.01

UB2-H

63.6

43

1246

1205

1.03

Concrete crushing

Concrete crushing

39.8

35.3

1.13

UB2-H-F1

80.8

26

1163

1230

0.95

6934

7949*

0.87

Partial rupture +
partial debonding

FRP debonding

50.5

60.4

0.84

UB2-H-F2 104.8

31

1122

1239

0.91

5329

5698*

0.94

FRP debonding

FRP debonding

65.5

68.9

0.95

UB2-P

75.2

88

1598

1260

1.27

Concrete crushing

Concrete crushing

47.0

36.3

1.29

UB2-P-F1

93.6

37

1339

1244

1.08

5393

7948*

0.68

FRP debonding +
concrete crushing

FRP debonding

58.5

60.6

0.97

UB2-P-F2 101.2

30

1223

1270

0.96

4285

5698*

0.75

FRP debonding

FRP debonding

63.3

69.3

0.91

62

1211

1106

1.09

US1-H

22.7

Concrete crushing

Concrete crushing

14.2

11.5

1.23

1.05

FRP rupture

FRP debonding

21.4

23.3

0.92

US1-H-F1

34.3

63

1195

1033

1.16

8309

7948

US1-H-F2

43.0

66

1152

1112

1.04

6074

6490

0.94

Partial rupture +
partial debonding

Concrete crushing

26.9

30.1

0.89

US1-P

21.3

100

1413

1066

1.33

Concrete crushing

Concrete crushing

13.3

11.6

1.15

US1-P-F1

34.5

68

1177

1065

1.11

5770

7948*

0.73

FRP debonding

FRP debonding

21.6

23.8

0.91

US1-P-F2

48.1

75

1165

1117

1.04

6280

6580

0.95

FRP rupture +
concrete crushing

Concrete crushing

30.1

30.8

0.97

US2-H

35.1

87

1227

966

1.27

Concrete crushing

Concrete crushing

21.9

16.3

1.34

US2-H-F1

42.6

62

1039

6277

6758

0.93

Concrete crushing
Concrete crushing
+ partial debonding

26.6

26.4

1.01

US2-H-F2

57.2

75

1065

972

1.10

5517

5551

0.99

Concrete crushing

Concrete crushing

35.8

31.6

1.13

US2-P

36.9

66

1105

992

1.11

Concrete crushing

Concrete crushing

23.1

17.0

1.36

US2-P-F1

47.6

68

1146

1047

1.09

7554

6731

1.12

Concrete crushing

Concrete crushing

29.8

26.9

1.11

US2-P-F2

59.8

67

1136

1029

1.10

5570

5429

1.03

Concrete crushing

Concrete crushing

37.4

32.1

1.16

BB2-P

63.4

44

1738

1737

1.00

Concrete crushing

Concrete crushing

39.6

34.2

1.16

BB2-P-F1

87.0

35

1710

1682

1.02

5906

7027

0.84

FRP rupture

Concrete crushing

54.4

53.6

1.01

BB2-P-F2 107.8

30

1680

1591

1.06

4781

5523*

0.87

FRP debonding

FRP debonding

67.4

63.1

1.07

BS2-P

33.9

71

1662

1687

0.99

Concrete crushing

Concrete crushing

21.2

21.4

0.99

BS2-P-F1

44.3

67

1702

1556

1.09

7834

6000

1.31

FRP debond. and


rupture

Concrete crushing

27.7

28.1

0.98

BS2-P-F2

59.7

70

1429

1435

1.00

5701

5045

1.13

Concrete crushing

Concrete crushing

37.3

32.2

1.16

RB2

72.5

46

Concrete crushing

Concrete crushing

45.3

42.5

1.07

RB2-F1

98.6

35

7608

6740

1.13

FRP rupture

Concrete crushing

61.6

62.0

0.99

RB2-F2

110.8

33

5003

4177

1.20

Concrete crushing

Concrete crushing

69.3

73.4

0.94

RS2

37.0

77

Concrete crushing

Concrete crushing

23.1

22.7

1.02

RS2-F1

48.5

60

6863

FRP rupture

Concrete crushing

30.3

32.2

0.94

RS2-F2

66.7

77

5834

4660

1.25

FRP rupture

Concrete crushing

41.7

36.6

1.14

Equal to debonding strain calculated using Eq. (8).


Notes: Exp. is Experiment; 1 kN = 0.224 kip; 1 mm = 0.039 in.; 1 MPa = 0.145 ksi; 1 kN-m = 8.83 k-in.)

ACI Structural Journal/March-April 2014

415

Fig. 7Representative load-deflection response of slab


specimens. (Note: 1 mm = 0.039 in.; 1 kN = 0.224 kip.)

Fig. 6Comparison between load-deflection response of


unbonded and bonded PC beam specimens. (Note: 1 mm =
0.039 in.; 1 kN = 0.224 kip.)
developed in these specimens was, admittedly, larger than
that predicted from lp = (20.7/f + 10.5)c, based on which
Eq. (13) and (16) were developed. Consequently, the ultimate strains or stresses in the prestressing steel for the
control specimens in this investigation (Table 2 and Fig. 10)
were conservatively larger than those predicted by Eq. (13)
or(16).
Also, as can be seen from Table 2, the analytical approach
predicted the experimentally measured strain in the FRP
reinforcement and associated mode of flexural failure
reasonably accurately for most unbonded specimens, as well
as for the bonded PC and RC specimens. While the results
are, to a great extent, in support of the ACI Committee
4402 guidelines for predicting modes of flexural failure in
FRP-strengthened members, the experimentally measured
FRP strains for the unbonded specimens were consistently
slightly lower than those predicted using Eq. (8). The
average ratio of test-to-predicted FRP strains is calculated at
0.9 (SD= 0.15) for the unbonded specimens, 1.1 (SD = 0.18)
for the combined bonded PC and RC specimens, and 0.96
(SD = 0.18) for all unbonded, bonded, and RC specimens.
Finally, from Fig. 11, Table 2, and the statistical data
provided, except for the control unbonded PC specimens
that developed larger than predicted moment capacities due
to the development of larger-than-predicted steel stresses,
416

Fig. 8Representative variation of FRP strain with applied


load. (Note: 1 kN = 0.224 kip.)

Fig. 9Typical variation of stress increase in unbonded


prestressing steel with applied load. (Note: 1 kN = 0.224 kip.)
ACI Structural Journal/March-April 2014

Fig. 11Prediction of nominal moment capacity of combined


specimens of current investigation. (Note: 1 kN-m = 8.83 k-in.)

Fig. 10Prediction of experimental data for unbonded


members6 including FRP strengthened unbonded specimens of
current investigation for ps = 1.0. (Note: 1 MPa = 0.145 ksi.)
as illustrated previously, the predicted nominal moment
capacities Mn of the FRP-strengthened specimens (assuming
f = 1.0) were generally in good agreement with the test
results. Despite little discrepancy, the level of accuracy
in predicting Mn for the unbonded specimens using the
proposed design approach is consistent with the level of
accuracy in predicting Mn for bonded PC and RC specimens
using the ACI Committee 4402 approach. The average ratio
of test-to-predicted results is calculated as 0.97 (SD = 0.09)
for the FRP-strengthened unbonded specimens, and 1.05
(SD = 0.09) for the combined FRP-strengthened bonded
PC and RC specimens. For the combined control and
FRP-strengthened specimens, the coefficient of correlation
R between the experimentally measured and the calculated
nominal moment capacities is equal to 0.97 for the unbonded
specimens, 0.98 for the bonded PC and RC specimens, and
0.97 for the overall unbonded PC, bonded PC, and RC
specimens. Their corresponding average ratios of test-topredicted results are equal to 1.07 (SD = 0.17), 1.04 (SD =
0.08), and 1.06 (SD = 0.15).
It should be noted that because the experimentally
measured FRP strains of the FRP-strengthened unbonded
ACI Structural Journal/March-April 2014

specimens, particularly when the mode of failure is by


FRP debonding or rupture, were slightly lower than those
predicted by Eq. (8), these specimens developed slightly
lower moment capacities than those predicted using the
proposed approach (Table 2). Consequently, although additional conservatism in predicting Mn for design purposes
can be gained by setting f = 0.85 as recommended by
ACI Committee 4402 (Eq. (1)), the FRP debonding strain
in Eq.(8) may require slight modification in its application
for unbonded members. Until further experimental evidence
is available to support such modification, however, Eq. (8)
can still be used for unbonded members without significant
loss of accuracy.
CONCLUSIONS
A design-oriented approach for calculating the nominal
moment capacity Mn of unbonded PC members when
strengthened using external FRP composites is presented.
The approach is applicable for simply supported and continuous members, and is consistent with the guidelines of
ACI Committee 4402 for evaluating Mn of bonded PC or
RC members. The accuracy of the proposed approach for
simply supported members was verified by comparing with
the results of a comprehensive test program designed and
carried out specifically for the purpose of this study.
Except for a slight discrepancy in predicting the FRP
debonding strain for the unbonded specimens that encountered FRP failure before concrete crushing, the proposed
approach for unbonded PC members, and the approach
recommended by ACI Committee 440 for bonded PC and RC
members, predicted well the test results at ultimate, including
the strain/stress in the PS, strain in the FRP reinforcement,
mode of flexural failure, and nominal flexuralstrength.
The experimental results and the proposed approach for
calculating Mn of FRP-strengthened unbonded PC members
clearly show that the use of external FRP reinforcement is
as effective in improving the nominal flexural strength of
417

unbonded PC members, as when used for strengthening


bonded PC or RC members. Consequently, in designing
the FRP system for flexural strengthening of unbonded PC
members, no special guidelines are needed beyond those
recommended in the ACI Committee 4402 report and the
design-oriented approach proposed for unbonded members
in this investigation.
AUTHOR BIOS

Fatima El Meski is a Lecturer at the Lebanese American University,


Beirut, Lebanon. She was formerly a PhD Student in the Department of
Civil and Environmental Engineering at the American University of Beirut,
Beirut, Lebanon.
ACI member Mohamed H. Harajli is a Professor of civil engineering at
the American University of Beirut. His research interests include design
and behavior of reinforced, prestressed, and fiber-reinforced concrete
members and strengthening and repair of concrete structures.

REFERENCES

1. ACI Committee 318, Building Code Requirements for Reinforced


Concrete and Commentary (ACI 318-11), American Concrete Institute,
Farmington Hills, MI, 2011, 503 pp.
2. ACI Committee 440, Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening Concrete Structures (ACI

418

440.2R-08), American Concrete Institute, Farmington Hills, MI, 2008,


76 pp.
3. AASHTO, LRFD Bridge Design Specifications, American Association of State Highway and Transportation Officials, Washington, DC,
2004, 1324 pp.
4. Naaman, A. E.; Burns, N.; French, C.; Gamble, W. L.; and Mattock, A.
H., Stresses in Unbonded Prestressing Tendons at Ultimate: Recommendation, ACI Structural Journal, V. 99, No. 4, July-Aug. 2002, pp. 518-529.
5. Harajli, M. H., On the Stress in Unbonded Tendons at Ultimate: Critical Assessment and Proposed Changes, ACI Structural Journal, V. 103,
No. 6, Nov.-Dec. 2006, pp. 803-812.
6. Harajli, M. H., Tendon Stress at Ultimate in Continuous Unbonded
Post-Tensioned Members: Proposed Modification of ACI Eq. (18-4)
and (18-5), ACI Structural Journal, V. 109, No. 2, Mar.-Apr. 2012, pp.
183-192.
7. Menegotto, M., and Pinto, P. E., Method of Analysis for Cyclically
Loaded Reinforced Concrete Plane Frames. IABSE Preliminary Report for
Symposium on Resistance and Ultimate Deformability of Structures Acted
on Well-Defined Repeated Loads, Lisbon, Portugal, 1973, pp. 15-22.
8. Harajli, M. H., and Hijazi, S., Evaluation of the Ultimate Steel Stress
in Unbonded Partially Prestressed Beams, PCI Journal, V. 36, No. 2,
Jan.-Feb. 1991, pp. 62-82.
9. Moon, J. H., and Burns, N. H., Flexural Behavior of Members with
Unbonded Tendons. II: Application, Journal of Structural Engineering,
ASCE, V. 123, No. 8, Aug. 1997, pp. 1095-1101.
10. El Meski, F., Behavior of Unbonded Post Tensioned Members
Strengthened Using External FRP Composites: Experimental Evaluation
and Analytic Modeling, PhD dissertation, Department of Civil and Environmental Engineering, American University of Beirut, Beirut, Lebanon,
2012, 283 pp.

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S37

Size Effect on Strand Bond and Concrete Strains at


Prestress Transfer
by Jos R. Mart-Vargas, Libardo A. Caro, and Pedro Serna-Ros
The size-effect phenomenon has been analyzed with pretensioned
prestressed concrete prismatic specimens at prestress transfer. An
experimental program that includes variables such as concrete
mixture design, specimen cross section size, and concrete age, has
been conducted. Several series of specimens with different embedment lengths have been made and tested using a testing technique
based on the bond behavior analysis by measuring prestressing
strand force. In addition, some specimens have been instrumented
to determine longitudinal concrete strain profiles. The tests have
provided data on concrete strains, transfer length, effective
prestressing force, bond stress, and concrete modulus of elasticity.
Relationships between measured prestressing strand forces and
effective prestressing forces obtained from concrete compressive
strains have been found. A coefficient to account for the specimen crosssection size-effect on the concrete modulus of elasticity
has been proposed. Comparisons between test results and theoretical predictions from pre-existing equations in the codes have
beenmade.
Keywords: bond; modulus of elasticity; prestress; size-effect; strain;
strand; transfer length.

INTRODUCTION
Force in a prestressing strand is transferred to concrete
by bond in the case of pretensioned prestressed concrete
members.1 At prestress transfer, the prestressing strand
tends to shorten, the concrete around the prestressing strand
shortens as the prestressing force is applied to it, and the
prestressing strand that is bonded to the concrete shortens
with it. Consequently, prestress loss due to the elastic shortening of concrete occurs in the central zone of the member.
At the member ends, the prestressing strand force varies from
zero to the effective prestressing force along the distance
defined as transfer length2 (Fig. 1).
Both bond strength and transfer length depend on several
factors, such as concrete strength at prestress transfer, initial
strand stress, concrete cover, prestress transfer method,
strand geometry, and strand surface condition,3 and bond
strength improves when a confining stress is applied.4,5
Average bond stress along the transfer length has been
characterized as being proportional to the square root of
the concrete compressive strength given the influence
of the elastic modulus of the concrete that surrounds the
prestressing strand.6
A short transfer length results in higher stresses and
risk of cracking near member ends, and a long transfer
length shortens the available member length to resist the
bending moment and shear.6 As the transfer length is an
important parameter in the design exercise,6,7 bond performance is assumed essential for an adequate response of
pretensioned prestressed concrete applications, and quality
ACI Structural Journal/March-April 2014

Fig. 1Idealized prestressing strand force diagram.


assurance procedures for bonded applications should be
used.2 However, neither codes2,8,9 nor standards10,11 specify
minimum requirements for the bond performance of
prestressing strands, and there is no consensus on the main
parameters to be considered in equations to compute transfer
length12,13 or on a standard test method for bond quality.3
By way of example, the ACI 318-112 code provisions
for transfer length are not a function of concrete strength,
while the fib Model Code 20109 includes concrete properties; moreover, several test procedures to determine bond
characteristics14,15 or to measure transfer length16,17 have
been proposed as alternatives to the two most widely used
methods18: measuring the strand end slip or determining
the longitudinal concrete surface-strain profile. At prestress
transfer, variation in strand stress along the transfer length
involves slips between the strand and the surrounding
concrete. These slips can be used to estimate transfer
length,19,20 but the results obtained from transfer length estimations vary vastly.21 After prestress transfer, the longitudinal concrete surface strain profile also follows a similar
law to that of the prestressing strand force shown in Fig. 1.
Transfer length can be determined directly from the concrete
strain profile18,22,23 by either the Slope-Intercept Method24
(by intersecting an adjusted line to the initial branch with
an adjusted line to the central branch) or the 95% Average
Maximum Strain (AMS) Method7 (by intersecting the initial
branch with the horizontal line corresponding to 95% of the
average strains in the central branch).
Based on test results obtained from a recently developed experimental methodology, the ECADA (Ensayo para
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-174.R2, doi:10.14359.51686530, was received February 19. 2013,
and reviewed under Institute publication policies. Copyright 2014, American
Concrete Institute. All rights reserved, including the making of copies unless
permission is obtained from the copyright proprietors. Pertinent discussion including
authors closure, if any, will be published ten months from this journals date if the
discussion is received within four months of the papers print publication.

419

Table 1Concrete mixture designs


Designation
3

Cement, lb/yd (kg/m )

843 (500)

624 (370)

674 (400)

Water-cement ratio (w/c)

0.3

0.45

0.5

Aggregate 7/12, lb/yd3 (kg/m3)

1674 (993)

1637 (971)

1645 (976)

Sand 0/4, lb/yd3 (kg/m3)

1468 (871)

1448 (859)

1443 (856)

1.7
Polycarboxylate ether polymers

1.4
Modified polycarboxylic ethers

0.1
Modified polycarboxylic ethers

High-range water-reducing admixture


additive

Concrete compressive
strength, ksi (MPa)

(%)
Type

12 hours

5.8 (40)

24 hours

7.5 (52)

4.6 (32)

3.5 (24)

48 hours

8.4 (58)

5.2 (36)

4.2 (29)

28 days

12.3 (85)

7.3 (50)

6.3 (43)

Note: 1 mm = 0.04 in.

Caracterizar la Adherencia mediante Destesado y Arrancamiento; in English: Test to Characterize the Bond by
Release and Pull-Out) test method,17 the research herein
analyzes the specimen cross section size-effect on strand
bond and concrete strains at prestress transfer. Three
different concrete mixture designs applicable to the precast
prestressed concrete members industry, in combination with
three different specimen cross sections, have been tested.
RESEARCH SIGNIFICANCE
This research provides information on how specimen
cross section size-effect influences strand bond behavior and
concrete strains at prestress transfer. This paper analyzes
series of tests conducted on pretensioned prestressed
concrete prismatic specimens using a testing technique
based on bond behavior analysis by measuring prestressing
strand force. The tests provide data on concrete strains,
transfer length, effective prestressing force, bond stress,
and concrete modulus of elasticity. A coefficient to account
for the specimen cross section size-effect on the concrete
modulus of elasticity is proposed. The experimental results
have been compared with predictions from ACI 318-112 and
fib Model Code 2010.9
EXPERIMENTAL RESEARCH
An experimental program was conducted and the ECADA
test method17 was used. The method allows for the analysis
of bond behavior by the sequential reproduction of the
prestress transfer and the anchorage of prestressing strands
on the same specimen. A series of specimens with different
embedment lengths is required to determine both transfer
and development lengths25,26 by means of the ECADA
test method. Its feasibility has been verified for both shorttime27,28 and long-term analyses.29,30 In this work, only
the transfer length test results and analyses at prestress
transfer were included. Complementarily, several specimens were also instrumented to obtain longitudinal concrete
surfacestrains.
Materials
Three concrete mixtures applicable for the precast
prestressed concrete industry with different compressive
420

strengths at the time of testing (fci), ranging from 24 to 55


MPa (3.5 to 8.4 ksi), were tested. For all the concretes, the
components were: cement CEM I 52.5 R,31 crushed limestone aggregate 7 to 12 mm (0.275 to 0.472 in.), washed
rolled limestone sand 0 to 4 mm (0 to 0.157 in.), and a highrange water-reducing admixture additive. The mixtures of
the tested concretes are shown in Table 1.
The prestressing strand was a low-relaxation seven-wire
steel strand specificied as UNE 36094:97 Y 1860 S7 13.010
with a guaranteed ultimate strength of 1860 MPa (270 ksi).
The main characteristics were those according to the manufacturer: diameter, 13 mm (0.5 in.); cross-sectional area,
100 mm2 (0.154 in.2); ultimate strength, 200.3 kN (45.1 kip);
yield stress at 0.2%; 189.9 kN (42.8 kip); and the modulus
of elasticity, 203,350 MPa (29,500 ksi). The prestressing
strand was used under the as-received condition (rust-free
and lubricant-free). The strand was not treated in any special
way. The strand was stored indoors and care was taken to not
drag the strand along the floor.
Testing program
The variables considered in the test program were
concrete mixture, specimen cross section, and concrete age
at prestress transfer. A series of specimens with different
embedment lengths were tested for each combination of
variables selected. Embedment lengths followed increments
of 50 mm (2 in.) in the nearness of the transfer length for
its determination and specimens with 1350 mm (53.15 in.)
were included to determine profiles of longitudinal concrete
surface strains.
Specimens were designed as M-D-T-L, where M is the
concrete mixture type (A, B, or C); D is the specimen cross
section size in mm (100, 80, or 60, for a 100 x 100 mm2
[3.94 x 3.94 in.2], 80 x 80 mm2 [3.15 x 3.15 in.2], and a 60
x 60 mm2 [2.36 x 2.36 in.2] cross section, respectively); T is
the concrete age at prestress transfer (12, 24, or 48 hours);
and L is the specimen embedment length (in mm).
The designation for a complete series included only
the parameters detailing M-D-T. In two cases (A-60-48,
A-8012), only specimens with an embedment length of
1350 mm (53.15 in.) were made.

ACI Structural Journal/March-April 2014

Table 2Test program


Age
at prestress transfer,
hours

12

24

48

Concrete (specimen embedment lengths, mm)


A

A-80-12 (1350)

A-100-12 (300, 400, 500, 550, 600, 650,


700, 750, 900, 1350)

A-60-24 (300, 450, 600, 950, 1350)

A-80-24 (300, 350, 400, 450, 500, 550,


600, 650, 700, 750, 1350)

B-80-24 (550, 600, 650, 700, 750, 800,


1350)

A-100-24 (300, 400, 450, 500, 550, 600,


650, 700, 750, 900, 1350)

B-100-24 (300, 550, 600, 650, 700, 750,


800, 850, 1350)

C-100-24 (600, 650, 700, 750, 800, 1350)

A-60-48 (1350)

B-60-48 (300, 350, 400, 450, 500, 550,


600, 650, 950, 1000, 1350)

B-80-48 (300, 450, 500, 550, 600, 650,


700, 950, 1350)

C-80-48 (550, 600, 650, 800, 850, 900,


950, 1350)

B-100-48 (300, 400, 450, 500, 550, 600,


650, 700, 750, 800, 850, 1350)

C-100-48 (550, 600, 650, 700, 750, 1350)

Note: 1 mm = 0.04 in.

Fig. 3General view of test equipment.


Fig. 2Pretensioning frame layout.
Table 2 summarizes the test program conducted and
provides the embedment lengths of tested specimens.
Test equipment and instrumentation
The ECADA test method is based on measuring and
analyzing the prestressing strand force in a series of pretensioned prestressed concrete specimens with different
embedment lengths. Specimens are made and tested using
pretensioning frames, as shown in Fig. 2. In this way, each
specimen has only one end zone with the corresponding
transfer length.
Several components were coupled at both ends of the
pretensioning frames to complete the test equipment. A
hollow hydraulic jack of 300 kN (67.5 kip) capacity with
an end-adjustable anchorage device was placed at one end
to carry out tensioning, provisional anchorage, and detensioning of prestressing strands. At the opposite end, an
Anchorage-Measurement-Access (AMA) system was
placed to guarantee the anchorage of the prestressing strand
and to simulate the specimens sectional stiffness. The AMA
system was made up of a sleeve of 120 mm (4.72 in.) in
length in the final specimen stretch (beyond the specimen
embedment length), the end frame plate and an anchorage
plate supported on the frame by two separators.
Figure 3 shows a general view of the test equipment,
which includes a pretensioning frame (blue components) of
ACI Structural Journal/March-April 2014

2000 mm (78.74 in.) in length and both end frame plates


of 320 x 320 x 50 mm (12.60 x 12.60 x 1.97 in.), and an
anchorage plate of 320 x 200 x 60 mm (12.60 x 12.60 x
2.36 in.) supported by both 250 x 120 x 20 mm (9.84 x 9.84
x 0.79 in.) welded elements 250 mm (9.84 in.) in length
(white components).
The instrumentation used included a hydraulic jack pressure transducer to control the tensioning and detensioning
operations; the AMA system included a hollow force transducer to measure the prestressing strand force; and in the
specimens with an embedment length of 1350 mm (53.15 in.),
detachable mechanical strain gauges were used to obtain the
longitudinal concrete surface strain profile at the prestressing
strand level. The gauge points were uniformly spaced at 50
mm (2 in.) intervals and were placed on two opposite specimen faces. No internal measuring devices were used in the
test specimens to not distort the bondphenomenon.
Test procedure
Each ECADA test series included a variable number of
specimens (usually 6 to 12). The step-by-step test procedure
for each specimen is described in detail in Mart-Vargas
et al.17 and may be summarized as follows (steps marked
by opt have been included only for any specimens additionally instrumented to obtain the longitudinal concrete
surfacestrains):
a) Fabrication stage:
421

1. The strand is placed in the frame with both anchorage


devices at the ends.
2. Strand tensioning is done using the hydraulic jack.
3. Provisional strand anchorage is done by the end-adjustable device.
4. The concrete is mixed, placed into the formwork in
the frame, and consolidated.
5. The specimen is cured to achieve the desired concrete
properties at the time of testing.
b) Preparation stage:
1. The formwork is removed from the frame.
2. (opt) Attaching gauge points by epoxy glue along
both the lateral sides of the specimen at the prestressing
strand level.
3. (opt) Reading the initial set of distances between
gauge points.
4. The end-adjustable strand anchorage is relieved using
the hydraulic jack.
c) Prestress transfer. Strand detensioning is produced by
unloading the hydraulic jack. The specimen is supported on
the end frame plate.
d) Stabilization period. The prestressing strand force
depends on not only the strain compatibility with the
concrete specimen but also on its action in the AMA system.
This force requires a stabilization period to guarantee its
measurement.
e) Measurement:
1. Measuring the prestressing strand force achieved (Pi)
in the AMA system.
2. (opt) Rereading the set of distances between gauge
points (after prestress transfer).
Test parameters
All the specimens were prestressed by a concentrically
located single strand at a prestress level before releasing
75% of the guaranteed ultimate strength. Specimens were
subjected to the same consolidation and curing conditions.
The prestress transfer was gradually performed in each specimen at a controlled speed of 0.80 kN/s (0.18 kip/s). A 2-hour
stabilization period after detensioning was established.
DATA PROVIDED FROM TESTS
The measured prestressing strand forces and sets of
distances between gauge points were the direct data
collected from the specimen test. Based on these direct data
from a specimen or from complete series of specimens, and
by means of back-calculations using theory of mechanics
concepts, parameters such as concrete strains, transfer
length, effective prestressing force, bond stress, and concrete
modulus of elasticity can be obtained.
Concrete strains
Concrete strains can be obtained from the changes in
distances between gauge points before and after the prestress
transfer by dividing them by gauge length. The strain change
for each gauge length was assigned to its center point. From
the specimen free end, a profile with an ascendent branch,
followed by a practically horizontal branch, was depicted

422

Fig. 4Transfer length determination according to ECADA


test method.
when these strains were plotted according to specimen
embedment length.
Transfer length
Approximation to transfer length determination can be
obtained directly from the longitudinal concrete strain profile
of a specimen with a long embedment length (1350 mm
[53.15 in.] in this research), such as ascendent branch length.
In this work, the 95% Average AMS Method7 was used.
On the other hand, for a complete series of specimens
tested with the ECADA test method, the measured Pi forces
can be ordered according to specimen embedment lengths
(Fig. 4). The obtained curve presented a bilinear tendency,
with an ascendent initial branch and horizontal branch corresponding to effective prestressing force Pe. Transfer length
Lt corresponded to the embedment length that marked the
beginning of the horizontal branchthat is, it corresponded
to the shorter specimen embedment length, where Pi = Pe.
Effective prestressing force
Beyond transfer length, compatibility of the strains between
the prestressing strand and the concrete exists: the prestressing
strand strain change p accounted for just before the prestress
transfer is equal to the concrete strain change c. Therefore,
effective prestressing strand force can be obtained from the
concrete strains according to Eq. (1)

Pes = P0 P = P0 p Ep Ap (1)

where Pes is the effective prestressing strand force obtained


from the measured specimen strains; P0 is the prestressing
strand force just before the prestress transfer; P is the
prestress loss due to elastic concrete shortening (P = p
Ep Ap); p is the prestressing strand strain change beyond
the transfer length, accounted for just before the prestress
transfer (p = c); Ep is the modulus of elasticity of the
prestressing strand; and Ap is the prestressing strand area.
In addition, effective prestressing strand force can be
measured directly from the AMA system on specimens with
an embedment length equal to or longer than the transfer
length. An ideal AMA system must have the same sectional
stiffness as the specimen,17 which depends on the concrete
ACI Structural Journal/March-April 2014

Fig. 6Concrete strain profile for Specimen B-100-48-1350.

Fig. 5End-discontinuity effect.


properties and the specimen cross section. Different AMA
system designs should be devised for different test conditions. However, it would not be feasible to design a system
for each specific test condition. For this reason, the stiffness
of the designed AMA system was greater than the specimens sectional stiffness. Consequently, the prestressing
strand force measured in the AMA system after the prestress
transfer was greater than the effective prestressing force
in the specimen. This difference of forces caused an enddiscontinuity effect and gave rise to a slight overestimation
of the real transfer length (Fig. 5).
Bond stress
Based on the equilibrium of the effective prestressing
strand force achieved, the average bond stress along the
tranfer length can be obtained according to Eq. (2)
ut =

Pe
4
3 f Lt

(2)

where ut is the average bond stress along the transfer length;


Pe is the effective prestressing force; (4/3) is the actual
seven-wire strand perimeter; is the nominal diameter of
prestressing strand; and Lt is the transfer length.
Concrete modulus of elasticity
An early concrete modulus of elasticity at prestress
transfer for each specimen can be obtained from prestress
loss due to elastic concrete shortening and the transformed
cross-section properties according to Eq. (3)

P0
1
e ci =

1 + n Eci Ac

P0
E p Ap
e ci
or Eci =
(3)
Ac

where ci is the elastic shortening strain of concrete due


to the prestress transfer; P0 is the prestressing strand force

ACI Structural Journal/March-April 2014

just before prestress transfer; n is the initial steel modular


ratio (n = Ep/Eci, where Ep is the modulus of elasticity of
the prestressing strand and Eci is the concrete modulus of
elasticity at prestress transfer); is the geometric ratio
( = Ap/Ac, where Ap is the prestressing strand area and Ac is
the net cross-sectional area of the specimen).
EXPERIMENTAL RESULTS AND DISCUSSION
Concrete strains
By way of example based on the collected test data,
Fig. 6 provides the concrete strain profile for Specimen
B-100-48-1350. The results are obtained by averaging
the readings from the two opposite specimen faces. Three
regions can be distinguished: initial branch, plateau, and
end-discontinuity. In this case, a transfer length of 600 mm
(23.6 in.) is observed directly from the curve, and another of
575 mm (22.6 in.) is determined from the 95% AMS.
Beyond the transfer length, greater concrete strains
ci = 0.00075result because of the early age of concrete
at prestress transfer. The influence of specimen cross
section size-effect on the average concrete strains beyond
the transfer length is depicted in Fig. 7. A strong influence
is observed and explained because of the combined effects
of the different concrete stress levels and the deformability behavior related to the specimens cross sections.
The concrete stress level can be obtained by dividing the
effective prestressing force, transferred according to Eq.
(1), between the specimens net cross-sectional area and
the concrete compressive strength at prestress transfer. As
Fig. 8 depicts, for the same concrete type and concrete age
at prestress transfer, the concrete stress level and concrete
strains evidently increase when the specimen cross section
decreases. Besides, the specimens with different concrete
stress levels present similar concrete strains when specimens are of the same cross section size: approximately
0.0008 for specimens with a larger cross section, between
0.0012 and 0.0016 for specimens with an intermediate cross
section, and approximately 0.0020 for specimens with a
smaller cross section. However, for the same concrete stress
level, concrete strains significantly increase when the cross
section size reduces. This implies that a significant deformability behavior relating to the specimen cross section exists.
This fact will be analyzed later by considering the concrete
modulus of elasticity.
423

Fig. 7Concrete strain versus specimen side cross section.

Fig. 9Prestressing strand forces for Series B-100-48.

Fig. 8Concrete strain versus concrete stress level.

Fig. 10Measured and predicted transfer length.

Transfer length
Figure 9 provides the transferred prestressing forces
versus the embedment lengths for the complete B-100-48
series. All the test specimens with embedment lengths
equal to or longer than 650 mm (25.6 in.) present similar Pi
values and are, therefore, equal to the effective prestressing
force Pe. In contrast, all the test specimens with embedment lengths shorter than 650 mm (25.6 in.) present lower
Pi values. Therefore, the transfer length determined by the
ECADA test method can be affirmed as 650 mm (25.6 in.)
in this case.
Figure 10 shows the transfer length obtained from
concrete strains (directly [two values are depicted when
the ascendent branch length is unclear] and by 95% AMS)
and from prestressing strand forces. Specimens have been
ordered by concrete mixture by increasing cross section size
and concrete age at prestress transfer.
As observed in Fig. 10, the transfer lengths in the specimens made with Concrete C are longer than those in the specimens made with Concrete B which, in turn, are longer than
the transfer lengths in the specimens made with Concrete
A: C-100-48/B-100-48, C-100-24/B-100-24/A-100-24,
C-80-48/B-80-48, B-80-24/A-80-24, and B-60-48/A-60-48.
Generally, transfer length values also reduce when concrete
age at prestress transfer increases: A-60-24/A-60-48,
A-80-12/A-80-24, A-100-12/A-100-24, B-80-24/B-80-48,
and C-100-24/C-100-48, except for B-100-24/B-100-48.
Besides, similar transfer lengths are obtained for the 100 x

100 mm2 (3.94 x 3.94 in.2) and the 80 x 80 mm2 (3.15 x


3.15 in.2) specimen cross sections. However, A-60-48 and
B-60-48 show the shortest transfer lengths, while A-60-24
presents a high value.
The transfer lengths predicted according to ACI 318-112
and fib Model Code 20109 from the measured parameters are
included in Fig. 10. As observed, the predictions from ACI
318-112 have similar values, as only strand parameters are
considered. However, the predictions from fib Model Code
20109 vary vastly as concrete properties are also considered. ACI 318-112 overestimates transfer length, except for
Concrete Mixture C, while fib Model Code 20109 generally
overestimates it (only case A-60-24 is underestimated).

424

Effective prestressing force


Figure 11 shows the measured prestressing strand forces
in the AMA system (P0, Pe), and the effective prestressing
forces (Pes) obtained according to Eq. (1). For each complete
series, P0 and Pe are obtained by averaging the corresponding
prestressing strand forces P0 and Pi from those specimens
with an embedment length equal to or longer than the
transfer length. Due to the end-discontinuity effect, overestimation of the effective prestressing strand force is observed;
Pe is always greater than Pes for all the specimens. Effective prestressing force increases within the same concrete
mixture when the specimen cross section and the concrete
age at prestress transfer increase. These tendencies are seen
more clearly by the specimen strains.

ACI Structural Journal/March-April 2014

Fig. 11Effective prestressing force.


Bond stress
According to Eq. (2), Fig. 12 shows the average bond
stress along the measured transfer lengths from the two
techniques, based on strains or forces. Predicted bond stress
values from ACI 318-112 and fib Model Code 20109 have
been included for comparison purposes: a constant value of
2.76 MPa (0.4 ksi)32 results from ACI 318-11,2 and 3.2fctdi
(0.464fctdi)12where fctdi is the specified tensile strength
at prestress transferresults from fib Model Code 2010.9
Besides, a corrected value has been obtained from fib Model
Code 20109 by applying a 0.75 factor based on the ratio
between nominal and actual strand perimeter ( in fib
Model Code 20109 and 4/3 in ACI 318-112 and Eq. (2),
respectively).33
As observed in Fig. 12, the predictions from fib Model
Code 20109 are of the measured values order and follow the
logical sequence for the different concrete types and within
the same concrete type by varying specimen cross-section size
and concrete age at prestress transfer, except for A-60-24
and A-60-48. The predictions from ACI 318-112 and the
corrected predictions from fib Model Code 20109 show
global tendencies in accordance with the transfer lengths
depicted in Fig. 10; but inversely, however, in this case,
transfer length increases when bond stress decreases. As
observed in Fig. 12, the bond stresses in the specimens made
with Concrete A are greater than those in the specimens
made with Concrete B which, in turn, are greater than the
bond stresses in the specimens made with Concrete C.
The specimen cross section size-effect on the average
bond stresses can be analyzed from Fig. 13: a) greater bond
stress values for larger specimen cross sections are observed
in some cases (A-24h, B-24h, and C-48h); b) similar bond
stress values are obtained for A-12h irrespectively of the
specimen cross sections; and c) bond stress values decrease
for B-48h when the specimen cross section increases. Some
authors have found these tendencies: examples of reported
increases in bond strength with increases of concrete
cover thickness (or specimen cross section) can be found
in the literature34,35 and also examples of a size effect on
bond strength, which reduces as concrete cover thickness
increases.36 Finally, Fig. 13 also shows that, for the same

ACI Structural Journal/March-April 2014

Fig. 12Average bond stress along transfer length.

Fig. 13Average bond stress versus specimen side cross


section.
cross-section size, the bond stress values rise when concrete
age at prestress transfer increases (except for Concrete B and
side 100 mm [3.94 in.]).
Concrete modulus of elasticity: cross section
size-effect
Figure 14 shows the experimental Eci value according
to Eq. (3) for each specimen with an embedment length of
1350 mm (53.15 in.). As seen, higher Eci values are obtained
when concrete compressive strength increases, when specimen cross section increases, and when concrete age at
prestress transfer increases.
The early concrete modulus of elasticity is commonly
only obtained from the concrete compressive strength at 28
days. To this end, as ACI 318-112 overestimates the concrete
modulus of elasticity for very early-age and concrete
compressive strengths above 50 MPa (7.25 ksi),37 the fib
Model Code 20109 provisions (Eq. (4) to Eq. (6)) are taken
as a reference
Ec 28

f
= 19, 400 cm
10

1/ 3

(Eq. 5.1 21)9 (4)

Ec(t) = [cc(t)]0.5 Ec28 (Eq. 5.1-56)9 (5)

425

Fig. 14Experimental concrete modulus of elasticity at


prestress transfer.

28 0.5
bcc (t ) = exp s 1 (Eq. 5.1 51)9 (6)
t

where Ec28 is the concrete modulus of elasticity at 28 days


(MPa); fcm is the concrete compressive strength at 28 days
(MPa); Ec(t) is the concrete modulus of elasticity at age t;
cc(t) is a coefficient to describe development with time;
sis a coefficient to account for the strength class of cement
(s=0.2 for Class 52.5 R); and t is concrete age (days).
To analyze the specimen cross section size-effect on the
concrete modulus of elasticity from the obtained test results,
several adjustments of the experimental Eci values (Eq. (3))
and the expected Ec(t) values at prestress transfer (Eq. (5))
are made as shown in Fig. 15. Therefore, the following equation is proposed
Eci = Ec(t) (7)
where is a coefficient to account for the cross-section
size-effect: 0.721 for 100 x 100 mm2 (3.94 x 3.94 in.2) specimen cross section, 0.612 for 80 x 80 mm2 (3.15 x 3.15 in.2),
and 0.523 for 60 x 60 mm2 (2.36 x 2.36 in.2).
The concrete modulus of elasticity for each specimen at
28 days (Ec28_s) can be obtained as follows:
1. Determination of Eci (Eq. (3)).
2. Computation of a concrete modulus of elasticity at 28
days (Ec28_s) using Eq. (5) with Ec(t) = Eci.
3. Application of the coefficient (Eq. (7)), as follows:
Ec28_s = Ec28_ref.
4. The Ec28_ref values are obtained. These values have
good agreement with the fib Model Code 20109 provisions
(Eq. (4)) for each concrete mixture, as shown in Fig. 16.
Therefore, the coefficientinitially obtained at prestress
transferis also applicable at 28 days.
5. An Ec28_m value is computed for each concrete mixture
by averaging the Ec28_ref values from the specimens made
with the same concrete mixture.
6. The Ec28_s values are obtained using Eq. (7): Ec28_s =
Ec28_m.

426

Fig. 15Comparison of concrete modulus of elasticity at


prestress transfer.

Fig. 16Comparison of concrete modulus of elasticity at 28


days.
The results of this process are presented in Table 3.
Also, the Ec28_s values can be obtained approximately
from the concrete compressive strength through Ec28 (Eq.
(4)) using Eq. (7): Ec28_s = Ec28; and an expected Eci can
be computed using Eq. (5), by substituting Ec(t) for Eci and
Ec28 for Ec28_s.
Concrete modulus of elasticity: influence of
concrete stress
A normalized concrete modulus of elasticity Eci* = Eci/
(Eci from Eq. (3) and from Eq. (7)) is obtained to avoid
the cross section size-effect. Figure 17 depicts the obtained
Eci* relating to the concrete stress level. As observed, for the
same concrete type and concrete age at prestress transfer,
the tendencies when the concrete stress level increases are
as follows: Eci* decreases for Series A-12 h and B-24 h and
Eci* increases for Series C-48 h, with Eci* practically nothing
varies for Series A-24 h and B-48 h. Besides, a general
tendency showing similar Eci* values for different concrete
stress levels (refer to the horizontal line) is observed when
considering all the series. Therefore, the concrete modulus
of elasticity seems to be practically independent of concrete
stress, and the main influence is based on the specimen cross
section size-effect.

ACI Structural Journal/March-April 2014

Table 3Concrete modulus of elasticity, MPa


Modulus of elasticity of concrete, MPa
fib Model Code 20109

Experimental
Specimen

Eci

Ec28_s

Ec28_ref

Ec28_s

Eci

A-60-24

13,790

21,180

40,498

20,848

20,707

13,481

A-60-48

16,205

21,317

40,759

20,848

20,707

15,741

A-80-12

11,221

21,458

35,062

24,230

12,670

A-80-24

15,820

24,299

39,703

24,230

15,775

A-100-12

16,718

31,971

44,342

28,741

28,546

14,927

A-100-24

18,219

27,984

38,812

28,741

28,546

18,585

B-60-48

12,296

16,175

30,927

17,190

17,350

13,189

B-80-24

13,146

20,191

32,992

20,116

20,302

13,218

Ec28_m

39,863

Ec28_s

24,396
24,396

39,592

B-80-48

16,037

21,095

34,470

20,302

15,434

B-100-24

16,002

24,578

34,088

23,698

23,918

15,572

B-100-48

17,465

22,975

31,865

23,698

23,918

18,183

C-80-48

15,366

20,213

33,027

19,320

19,307

14,677

C-100-24

14,742

22,643

31,405

22,745

14,809

C-100-48

16,594

21,828

30,275

22,745

17,291

32,868

31,569

20,116

Ec28

22,761
22,761

33,174

31,547

Note: 1 MPa = 145 psi.

Fig. 17Concrete modulus of elasticity versus concrete


stress level.

Fig. 18Comparison of elastic shortening losses.

Concrete modulus of elasticity: elastic shortening


prestress loss
Finally, to verify the applicability of the obtained coefficients to account for the cross section size-effect, predicted
prestress losses due to the elastic shortening of concrete
from the measured and theoretical parameters are estimated
using preexisting equations. According to the ACI 318-11
Commentary,2 prestress losses can be calculated in accordance with several procedures.38,39 Figure 18 shows the
measured and predicted prestress losses due to the elastic
shortening of concrete for all the specimens with an embedment length of 1350 mm (53.15 in.). The coefficient has
been applied in all cases.
As observed in Fig. 18, the predicted prestress losses
follow the tendencies of measured prestress losses. The
prestress loss ranges with values of 10% for those specimens
with a larger cross section, with values of 15 to 20% for
those with an intermediate cross section, and of 25 to 30%

for those specimens with a smaller cross section. This fact


can be explained by the different concrete stress levels and
the deformability behavior relating to cross sections of specimens. Therefore, a clear effect of specimen cross section
size on prestress loss cannot be ruled out.
Moreover, Fig. 18 shows that the tendencies of the
measured prestress losses according to the variables
concrete mixture, specimen cross section size, and concrete
age at prestress transfer are followed by the prestress losses
predicted by all the methods: prestress losses disminish in
the same concrete mixture when the specimen cross section
increases and when the concrete age at prestress transfer
increases, and prestress losses in the specimens made with
Concrete C are greater than those in the specimens made
with Concrete B, which are also greater than the prestress
losses in the specimens made with Concrete A. The PCI
CPL39 predictions made with experimental Eci (Eq. (3)) and
fib Model Code 20109 practically coincide with the measured

ACI Structural Journal/March-April 2014

427

prestress losses. However, the PCI CPL39 predictions made


with Eci according to ACI 318-112 underestimate prestress
losses as the concrete modulus of elasticity is overestimated.
For Zia et al.38 predictions, which consider gross section
properties, an overestimation trend of prestress losses for
cases with smaller specimen cross section sizes is observed.
CONCLUSIONS
Based on the results of this experimental research, the
following main conclusions can be drawn:
Transfer lengths have been determined by two techniques: the longitudinal concrete strain profile and
prestressing strand forces. Based on both techniques,
transfer length values are higher when concrete
qualitydiminishes.
The transfer lengths predicted from ACI 318-112 have
similar values, irrespectively of concrete quality, as
only strand parameters are considered. However, the
fib Model Code 20109 predictions vary considerably as
concrete properties are also considered. ACI 318-112
overestimates transfer length when concrete quality
is good, while fib Model Code 2010 overestimates
ingeneral.
A strong influence of specimen cross section size-effect on average concrete strains beyond transfer length
has been observed. This fact can be explained by the
combined effects of different concrete stress levels and
deformability behavior in relation to the cross sections
of specimens.
The effective prestressing force increases in the same
concrete mixture when specimen cross section and
concrete age at prestress transfer increase. These tendencies are seen more clearly with the specimen strains
technique rather than with the strand forces technique.
Higher bond stress values for larger specimen cross
sections have been observed in most cases. For the same
cross section size, bond stress values are greater when
the concrete age at prestress transfer increases and when
concrete quality increases.
The concrete modulus of elasticity at prestress transfer
has been obtained from the experimental data of
prestressed specimens. Higher concrete modulus of
elasticity values result from greater concrete compressive strength, larger specimen cross section, and older
concrete age at prestress transfer.
A coefficient to account for the specimen cross section
size-effect on the concrete modulus of elasticity
isproposed.
fib Model Code 2010 predictions for transfer length,
average bond stress, concrete modulus of elasticity at
prestress transfer, and elastic shortening prestress loss
have showed a best agreement to the experimental
results than the ACI 318-112 predictions.
AUTHOR BIOS

Jos R. Mart-Vargas is an Associate Professor of civil engineering at


the Universitat Politcnica de Valncia (UPV), Valncia, Spain, where he
received his MEng in civil engineering and his PhD. His research interests
include the bond behavior of reinforced and prestressed concrete structural

428

elements, fiber-reinforced concrete, durability of concrete structures, and


strut-and-tie models.
Libardo A. Caro is an Assistant Researcher and PhD Candidate in the
Department of Construction Engineering and Civil Engineering Projects at
UPV. He received his civil engineering degree from the Universidad Santo
Toms, Bogot, Colombia. His research interests include bond properties
of prestressed concrete structures and the use of advanced cement-based
materials in structural applications.
Pedro Serna-Ros is a Professor of civil engineering at UPV, where he
received his MEng in civil engineering; he received his PhD from cole
Nationale des Ponts et Chausses, Paris, France. His research interests
include self-consolidating concrete, fiber-reinforced concrete, and the bond
behavior of reinforced and prestressed concrete.

ACKNOWLEDGMENTS

Funding for this research has been provided by the Spanish Ministry of
Education and Science, the Ministry of Science and Innovation, and ERDF
(Projects BIA2006-05521 and BIA2009-12722).

REFERENCES

1. Janney, J., Nature of Bond in Pretensioned Prestressed Concrete,


ACI Journal, V. 51, No. 5, May 1954, pp. 717-737.
2. ACI Committee 318, Building Code Requirements for Structural
Concrete (ACI 318-11) and Commentary, American Concrete Institute,
Farmington Hills, MI, 2011, 473 pp.
3. fib, Bond of Reinforcement in Concrete. State-of-Art Report,
Fdration Internationale du Bton, Bulletin dInformation No. 10, Lausanne, Switzerland, 2000, 427 pp.
4. ElBatanouny, M. K.; Ziehl, P. H.; Larosche, A.; Mays, T.; and
Caicedo, J. M., Bent-Cap Confining Stress Effect on the Slip of
Prestressing Strands, ACI Structural Journal, V. 109, No. 4, July-Aug.
2012, pp. 487-496.
5. ElBatanouny, M. K., and Ziehl, P. H., Determining Slipping Stress of
Prestressing Strands in Confined Sections, ACI Structural Journal, V. 109,
No. 6, Nov.-Dec. 2012, pp. 767-776.
6. Barnes, R. W.; Grove, J. W.; and Burns, N. H., Experimental Assessment of Factors Affecting Transfer Length, ACI Structural Journal, V.
100, No. 6, Nov.-Dec. 2003, pp. 740-748.
7. Russell, B. W., and Burns, N. H., Measured Transfer Lengths of 0.5
and 0.6 in. Strands in Pretensioned Concrete, PCI Journal, V. 44, No. 5,
Sept.-Oct. 1996, pp. 44-65.
8. CEN, Eurocode 2: Design of Concrete StructuresPart 1-1: General
Rules and Rules for Buildings, European Standard EN 1992-1-1:2004:E,
Comit Europen de Normalisation, Brussels, Belgium, 2004, 225 pp.
9. fib, Model Code 2010. First complete draftVolume 1, Fib Bulletin
No. 55, International Federation for Structural Concrete, Lausanne, Switzerland, 2010, 292 pp.
10. AENOR, UNE 36094: Alambres y Cordones de Acero para Armaduras de Hormign Pretensado, Asociacin Espaola de Normalizacin y
Certificacin, Madrid, Spain, 1997, 21 pp.
11. ASTM A416/A416M-10, Standard Specification for Steel Strand,
Uncoated Seven-Wire for Prestressed Concrete, ASTM International,
West Conshohocken, PA, 2010, 5 pp.
12. Mart-Vargas, J. R.; Arbelez, C. A.; Serna-Ros, P.; Navarro-Gregori,
J.; and Pallars-Rubio, L., Analytical Model for Transfer Length Prediction of 13 mm Prestressing Strand, Structural Engineering & Mechanics,
V. 26, No. 2, 2007, pp. 211-229.
13. Mart-Vargas, J. R.; Serna, P.; Navarro-Gregori, J.; and Pallars,
L., Bond of 13 mm Prestressing Steel Strands in Pretensioned Concrete
Members, Engineering Structures, V. 41, 2012, pp. 403-412.
14. Moustafa, S., Pull-out Strength of Strand and Lifting Loops,
Technical Bulletin 74-B5, Concrete Technology Associates, Tacoma, WA,
1974, 34 pp.
15. Cousins, T. E.; Badeaux, M. H.; and Moustafa, S., Proposed Test
for Determining Bond Characteristics of Prestressing Strand, PCI Journal,
V. 37, No. 1, Jan.-Feb. 1992, pp. 66-73.
16. Peterman, R. J., A Simple Quality Assurance Test for Strand Bond,
PCI Journal, V. 54, No. 2, 2009, pp. 143-161.
17. Mart-Vargas, J. R.; Serna-Ros, P.; Fernndez-Prada, M. A.; MiguelSosa, P. F.; and Arbelez, C. A., Test Method for Determination of the
Transmission and Anchorage Lengths in Prestressed Reinforcement,
Magazine of Concrete Research, V. 58, No. 1, Feb. 2006, pp. 21-29.
18. Thorsen, N., Use of Large Tendons in Pretensioned Concrete, ACI
Journal, V. 52, No. 2, Feb. 1956, pp. 649-659.

ACI Structural Journal/March-April 2014

19. Guyon, Y., Bton Prcontrainte. tude Thorique et Exprimentale, Ed. Eyrolles, Paris, France, 1953, 711 pp.
20. Balzs, G., Transfer Length of Prestressing Strand as a Function of
Draw-in and Inicial Prestress, PCI Journal, V. 38, No. 2, Mar.-Apr. 1993,
pp. 86-93.
21. Mart-Vargas, J. R.; Arbelez, C. A.; Serna-Ros, P.; and Castro-Bugallo, C., Reliability of Transfer Length Estimation from Strand End Slip,
ACI Structural Journal, V. 104, No. 4, July-Aug. 2007, pp. 487-494.
22. Mahmoud, Z. I.; Rizkalla, S. H.; and Zaghloul, E. R., Transfer
and Development Lengths of Carbon Fiber Reinforcement Polymers
Prestressing Reinforcement, ACI Structural Journal, V. 96, No. 4,
July-Aug. 1999, pp. 594-602.
23. Caro, L. A.; Mart-Vargas, J. R.; and Serna, P., Prestress Losses
Evaluation in Prestressed Concrete Prismatic Specimens, Engineering
Structures, V. 48, 2013, pp. 704-715.
24. Deatherage, J. H.; Burdette, E.; and Chew, C. K., Development
Length and Lateral Spacing Requirements of Prestressing Strand for
Prestressed Concrete Bridge Girders, PCI Journal, V. 39, No. 1, Jan.-Feb.
1994, pp. 70-83.
25. Mart-Vargas, J. R.; Arbelez, C. A.; Serna-Ros, P.; Fernndez-Prada,
M. A.; and Miguel-Sosa, P. F., Transfer and Development Lengths of
Concentrically Prestressed Concrete, PCI Journal, V. 51, No. 5, Sept.-Oct.
2006, pp. 74-85.
26. Mart-Vargas, J. R.; Serna, P.; and Hale, W. M., Strand Bond Performance in Prestressed Concrete Accounting for Bond Slip, Engineering
Structures, V. 51, 2013, pp. 236-244.
27. Mart-Vargas, J. R.; Serna-Ros, P.; Arbelez, C. A.; and Rigueira-Victor, J. W., Bond Behaviour of Self-Compacting Concrete in Transmission and Anchorage, Materiales de Construccin, V. 56, No. 284,
2006, pp. 27-42.
28. Mart-Vargas, J. R.; Serna, P.; Navarro-Gregori, J.; and Bonet, J. L.,
Effects of Concrete Composition on Transmission Length of Prestressing
Strands, Construction & Building Materials, V. 27, 2012, pp. 350-356.
29. Mart-Vargas, J. R.; Caro, L.; and Serna, P., Experimental Technique
for Measuring the Long-Term Transfer Length in Prestressed Concrete,
Strain, V. 49, 2013, pp. 125-134.

ACI Structural Journal/March-April 2014

30. Caro, L. A.; Mart-Vargas, J. R.; and Serna, P., Time-Dependent


Evolution of Strand Transfer Length in Pretensioned Prestressed Concrete
Members, Mechanics of Time-Dependent Materials, V. 17, No. 4, Nov.
2013, pp. 501-527.
31. EN 197-1:2000, Cement. Part 1: Compositions, Specifications and
Conformity Criteria for Common Cements, Comit Europen de Normalisation, Brussels, Belgium, 2000, 30 pp.
32. Tabatabai, H., and Dickson, T., The History of the Prestressing
Strand Development Length Equation, PCI Journal, V. 38, No. 5,
Sept.-Oct. 1993, pp. 64-75.
33. Mart-Vargas, J. R., and Hale, W. M., Predicting Strand Transfer
Length in Pretensioned Concrete: Eurocode versus North American Practice, Journal of Bridge Engineering, ASCE, V. 18, No. 12, Dec. 2013, pp.
1270-1280.
34. Yerlici, V. A., and zturan, T., Factors Affecting Anchorage Bond
Strength in High-Performance Concrete, ACI Structural Journal, V. 97,
No. 3, May-June 2000, pp. 499-507.
35. Garca-Taengua, E.; Mart-Vargas, J. R.; and Serna-Ros, P., Statistical Approach to Effect of Factors Involved in Bond Performance of
Steel Fiber-Reinforced Concrete, ACI Structural Journal, V. 108, No. 4,
July-Aug. 2011, pp. 461-468.
36. Ichinose, T.; Kanayama, Y.; Inoue, Y.; and Bolander, J. E., Size
Effect on Bond Strength of Deformed Bars, Construction & Building
Materials, V. 18, 2004, pp. 549-558.
37. Khan, A. A.; Cook, W. D.; and Mitchell, D., Early Age Compressive
Stress-Strain Properties of Low-, Medium-, and High-Strength Concretes,
ACI Materials Journal, V. 92, No. 6, Nov.-Dec. 1995, pp. 617-624.
38. Zia, P.; Preston, H. K.; Scott, N. L.; and Workman, E. B., Estimating Prestress Losses, Concrete International, V. 1, No. 6, June 1979,
pp. 32-38.
39. PCI Commitee on Prestress Losses, Recommendations for Estimating Prestress Losses, PCI Journal, V. 20, No. 4, July-Aug. 1975,
pp.43-75.

429

NOTES:

430

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S38

Proposed Minimum Steel Provisions for Prestressed and


Nonprestressed Reinforced Sections
by Natassia R. Brenkus and H. R. Hamilton
The current ACI code includes two separate provisions for minimum
steel reinforcement: one for nonprestressed reinforced sections,
and another for prestressed sections. For nonprestressed reinforced
sections, the current minimum steel requirement is written in terms
of geometric and material properties of the section. For prestressed
concrete, the current provision is written in terms of the cracking
moment of the section. Prestressed concrete only cracks when the
applied flexural tensile stress exceeds both the tensile strength of the
concrete and the net compressive stress from the prestressing force
in the steel. Consequently, when bonded prestressing steel quantities are increased, the cracking moment of the section increases.
Depending on the shape of the cross section and ultimate strength
requirements, in certain instances it is possible that a section can
contain a large volume of bonded prestressing steel and yet not
meet the minimum reinforcement requirement. A parametric study
of several cross sections was performed to investigate this behavior.
This paper describes an exact and unified solution approach for
specifying minimum reinforcement for both nonprestressed and
prestressed sections. A second parametric study to validate the
proposed minimum steel provisions is also presented.
Keywords: ductile design; flexure; minimum reinforcement; prestressed
concrete.

INTRODUCTION
Reinforced and prestressed concrete members are typically designed with a minimum quantity of flexural reinforcement. Steel reinforcement ensures that the postcracking flexural capacity of the section is greater than the
cracking moment Mcr, which is typically defined as the
moment necessary to cause the maximum flexural tensile
stress to exceed the modulus of rupture of the concrete. If a
beam with insufficient reinforcement cracks, then the reinforcement will rupture immediately following crack formation; the resulting failure mode, being sudden and brittle, is
undesirable and dangerous. Consequently, minimum reinforcement requirements have life-safety implications.
Both AASHTO LRFD (AASHTO 2007) and ACI 318-11
(ACI Committee 318 2011) design specifications have
minimum reinforcement requirements. ACI 318-11 has separate provisions for nonprestressed and prestressed concrete.
The current nonprestressed provisions do not explicitly
consider cracking moment, while the prestressed provisions
require the direct calculation of cracking moment. This is
true in the AASHTO LRFD as well. The AASHTO LRFD,
however, applies to both prestressed and nonprestressed
concrete. Although both design specifications aim to ensure
that flexural members contain a sufficient quantity of reinforcement so that the post-cracking flexural capacity is at
least equal to the cracking moment of the concrete section,
the method of implementation has been somewhat different.
ACI Structural Journal/March-April 2014

Explicit consideration of the cracking moment when


determining minimum reinforcement leads to an iterative
problem, sometimes without solution. Prestressed concrete
only cracks when the applied flexural tensile stress exceeds
both the net compressive stress from the prestressing force
in the steel and the tensile strength of the concrete. Consequently, when bonded prestressing steel quantities are
increased, the cracking moment of the section increases.
In certain instances, depending on the shape of the cross
section and ultimate strength requirements, it is possible for
a section to contain a large volume of bonded prestressing
steel yet fail to meet the minimum reinforcement requirement. This issue has been raised by others (Kleymann et al.
2006; Freyermuth and Aalami 1997; Ghosh 1986).
To address this issue and the inherent problem of separate provisions for prestressed and nonprestressed concrete,
a unified approach for determining minimum reinforcement for both nonprestressed and prestressed sections was
derived, and is presented in this paper. The proposed provisions are based on minimum steel concepts for prestressed
concrete introduced by Leonhardt (1964), avoid the explicit
consideration of cracking moment, and are applicable to
both prestressed and nonprestressed concrete. The parametric studies presented in this paper compare the proposed
provisions with current provisions and detailed calculations
of moment capacity using a direct tensile concrete behavior
approach rather than the modulus of rupture. Simplified
forms of the provisions are presented for rectangular and
T-shaped members.
RESEARCH SIGNIFICANCE
ACI 318-11 has different minimum reinforcement
requirements for prestressed and nonprestressed reinforcement. Furthermore, the current method of calculating the
minimum steel requirements for prestressed members
results in a variable quantity of minimum reinforcement,
which depends on the prestress level. Such an approach can
result in an inefficient use of reinforcing steel or even, in
some circumstances, the inability to satisfy the minimum
reinforcement requirements. The derivation and parametric
studies described in this paper provide a direct and unified
method of determining minimum steel requirements for both
nonprestressed reinforced and prestressed sections.
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-175.R1, doi:10.14359.51686531, was received July 17, 2012, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

431

CURRENT PROVISIONS: HISTORICAL


PERSPECTIVE
The earliest known minimum reinforcement requirement
appeared in a 1936 ACI committee document. A minimum
area of steel satisfying Eq. (1) was required for nonprestressed concrete (ACI Committee 501 1936)

AASHTO LRFD PROVISIONS


AASHTO LRFD specifications (2007) have unified provisions that apply to sections containing either prestressed
or nonprestressed reinforcement, or both. To meet the
minimum reinforcement requirements, the flexural resistance Mr should satisfy Eq. (3) or (4)

As,min = 0.0005bd (1)

Mr 1.2Mcr (3)

where b is the width of web in I- or T-beam sections, and d


is the depth from the compression face to the center of the
longitudinal tensile reinforcement.
In 1954, the Bureau of Public Roads issued design
requirements for prestressed concrete bridges (Bureau of
Public Roads 1954) and, in 1958, general recommendations were published for the design of prestressed concrete
(ACI-ASCE Committee 323 2004). Neither of these documents, however, contained minimum reinforcement requirements for prestressed concrete.
In 1963, prestressed concrete was incorporated within
the scope of ACI 318-63 (ACI Committee 318 1963);
the committee reasoned that combined discussion of the
nonprestressed and prestressed reinforced concrete was less
confusing than their separate consideration. To this end, ACI
318-63 combined nonprestressed reinforced and prestressed
concrete under the generic term reinforced concrete; it
included, however, two separate requirements for minimum
tensile reinforcement.
During this code cycle, the minimum reinforcement provision for nonprestressed concrete was rewritten in terms of p,
the ratio of the area of the tension reinforcement to the effective area of the concrete. As a lower limit, an area of steel
satisfying 200bd/fy was required; this absolute minimum
was derived by equating the ultimate strength of the section
without reinforcement to the ultimate strength of the section
with reinforcement and solving for p (ACI Committee 318
1963). In cases where the provided area of reinforcement
was one-third greater than that required by analysis, the
minimum reinforcement requirement was considered satisfiedan exception included to ensure that the minimum
reinforcement required for large members was not excessive.
The separate prestressed concrete design chapter added
to the ACI code during this cycle introduced minimum
reinforcement requirements for prestressed concrete: the
provided area of steel was required to be adequate to develop
an ultimate load capacity greater than 1.2 times the cracking
load. The cracking load was based on a modulus of rupture
of 7.5fc psi (0.62fc MPa).
In 1995, the minimum reinforcement requirements for
nonprestressed concrete were revised to explicitly account
for concrete and reinforcement tensile strengths (ACI
Committee 318 1995).
Equation (2) gives the typical requirement

where Mcr is the cracking moment

As,min =

432

3 fc
fy

bw d 200bw d / f y (2)

Mr 1.33Mu (4)
where Mu is the factored moment of the applicable load
combination.
The cracking moment is determined using Eq. (5)

S
M cr = Sc ( fr + fcpe ) M dnc c 1 Sc fr (5)
S
nc

where fr is the modulus of rupture of the concrete; fcpe is the


compressive stress in the concrete due to effective prestress
forces (after allowance for all prestress losses) at the extreme
fiber of the section where tensile stress is caused by externally applied loads; Sc is the section modulus with respect
to the extreme tensile fiber of the composite section; Snc is
the section modulus with respect to the extreme tensile fiber
of the monolithic or noncomposite section; and Mdnc is the
total unfactored dead load moment acting on the monolithic
or noncomposite section. When monolithic or noncomposite
sections are designed to resist all loads, the designer is
directed to substitute Snc for Sc in Eq. (5) for the calculation
of Mcr.
The modulus of rupture fr used depends on the limit state
being checked and the specified concrete strength. Literature is cited in which the modulus of rupture values typically
range between 7.5fc psi (0.62fc MPa) and 11.7fcpsi
(0.97fc MPa) (ACI Committee 318 1992; Walker and
Bloem 1960; Khan et al. 1996). For determining minimum
steel requirements using the AASHTO LRFD provisions, the
cracking moment is calculated using an estimated modulus
of rupture equal to 11.7fc psi (0.97fc MPa). The AASHTO
LRFD rationale for using a higher modulus of rupture value
for minimum steel requirements is that it is a strength limit
state, so the use of the upper bound value is justified; the
20% margin provided by Eq. (3) could be lost by using a
lower modulus-of-rupture value. These provisions are valid
for specified concrete strengths up to 15,000 psi (100 MPa).
ACI PROVISIONS
ACI 318-11, Section 18.8.2, specifies the following for
minimum reinforcement:
Total amount of prestressed and nonprestressed reinforcement in members with bonded prestressed reinforcement shall be adequate to develop a factored load at least
1.2 times the cracking load computed on the basis of the
modulus of rupture fr specified in 9.5.2.3.

ACI Structural Journal/March-April 2014

The cracking load, however, is not explicitly defined in


ACI 318-11 as it is in AASHTO LRFD. In general, it is
implicitly understood that the cracking load is the force
required such that the stress in the extreme tension fiber
is equal to the modulus of rupture, which is defined as
7.5fcpsi (0.62fc MPa).
EFFECT OF PRESTRESSING ON MINIMUM STEEL
REQUIREMENTS
As described previously, the current AASHTO LRFD
and ACI 318-11 minimum steel requirements are based
on cracking moment, either implicitly or explicitly. The
consequence of this dependence is that the quantity of
minimum reinforcement for a given cross section varies
with the prestressing force, and thus, the quantity of bonded
prestressing steel.
Oladapo (1968) noted this peculiarity in a parametric
study that compared the moment capacity for varying levels
of prestress, eccentricity, and tensile strength of the concrete.
He found that, after a certain level of prestress, continuing to
increase the amount of prestressing steel moved the section
further away from meeting the minimum requirements.
ONeill and Hamilton (2009) conducted a similar parametric study on a Florida bulb tee 78, a 4 ft (1.22 m) wide
by 8 in. (20.3 cm) deep hollow core slab, and a segmental
box girder, verifying Oladapos findings. Figure 1 shows
the results of the study. The variation of the design moment
capacity Mn of each section was calculated using strain
compatibility and normalized by the cracking moment;
cracking moment was calculated using Eq. (5). The moment
capacity varies as a result of the increase in area of bonded
prestressing steel Aps from zero to an arbitrary maximum;
Aps was normalized by the area of the concrete cross section
between the flexural tension face and the center of gravity of
the gross section Act. The Florida bulb tee and hollow core
were investigated for positive bending, and the segmental
box girder was investigated for negative bending. The graph
contains two curves for each section, which correspond to
the ACI 318-11 modulus of rupture estimate of 7.5fc psi
(0.62fc MPa) and AASHTO LRFD estimate of 11.7fc psi
(0.97fc MPa).
All three sections exhibit a similar trend: as bonded
prestressing steel is added, the curves quickly rise above
the minimum steel required to reach 1.2Mcr. Because the
cracking moment increases at a slightly greater rate than
that of the moment capacity, each curve eventually declines
as more prestressing steel is added. In general, the large
quantities of prestressing steel required to cause a decrease
in the moment ratio are likely beyond typical strength and
serviceability needs. In the case of the box girder in negative bending, however, the addition of bonded prestressing
steel of approximately 1.4% beyond an Aps/Act ratio results
in a section that does not meet AASHTO LRFD minimum
reinforcement requirements. Beyond this point, unnecessary
bonded prestressing steel should be added to satisfy Eq. (3).
The precompression force in a section has less effect than
the concrete tensile strength on the sections overall strength
when the compression zone is smaller than tension zone, as
in the case of the box girder (SB). This is seen in the plot
ACI Structural Journal/March-April 2014

Fig. 1Results from parametric study of current ACI provisions (ONeill and Hamilton 2009).
where the difference between the two curves for SB in negative bending is much greater than that between the hollow
core curves or the Florida bulb-tee curves. Consequently,
as the amount of prestressing steel increases, the cracking
moment increases at a faster rate than the moment capacity
due to the overriding effect of the relatively large tension
zone in sections such as the SB.
The use of 7.5fc psi (0.62fc MPa) versus 11.7fc
psi (0.97fc MPa) was also investigated by ONeill and
Hamilton (2009) in a series of prestressed girder and pile
laboratory tests. It was found that the girder cracking stresses
ranged from approximately 6fc psi (0.5fc MPa) to 14fc
psi (1.2fc MPa). In cases with low amounts of prestressing
steel, however, the cracking stress ranged from 6.1fc psi
(0.5fc MPa) to 7.6fc psi (0.63fc MPa). As the minimum
steel requirement usually controls the steel quantity when the
section is larger than required for strength, and the amount
of prestressing steel required is similarly small, the use of
7.5fc psi (0.62fc MPa) as an estimate of the modulus
of rupture is justified. Minimum reinforcement requirements for nonprestressed sections do not change with the
selected quantity of reinforcement. The authors believe that
a more rational approach would be for nonprestressed and
prestressed concrete to have a single unique minimum reinforcement requirement for each section. This requires direct
calculation of the cracking moment using only section and
material properties, as will be shown in the following section
with the derivation of a minimum reinforcementequation.
PROPOSED MINIMUM REINFORCEMENT
PROVISIONS
Leonhardt (1964) proposed a minimum steel requirement for prestressed concrete based on providing sufficient
steel area to resist net tensile concrete stresses that occur
just before cracking. His approach takes advantage of the
prestressing steels tensile capacity between the effective
prestressed state (State I) and the ultimate state at flexural
capacity (State II) without explicit consideration of the
cracking moment to determine the minimum prestressing
steel required. It is applicable to both nonprestressed and
433

Fig. 2Concrete and reinforcement forces at State I and


State II (Leonhardt 1964).

Fig. 3Stress state of section at decompression.

prestressed concrete. A representation of the concept is


shown in Fig. 2.
The method ensures sufficient area of reinforcement such
that the tensile force associated with the incremental change
in steel stress from the effective prestress fse to the stress
associated with ultimate flexural capacity fps, combined with
the nonprestressed reinforcement yield strength fy, is greater
than the tensile force generated in the concrete just before
cracking Tcr. This relationship can be expressed as
Apsfps Tcr + Apsfse + Asfy (6)
Leonhardts proposed method made several assumptions:
the tensile zone is rectangular, the resultant of the tension
block is at the same elevation as the prestressing steel, and
the section is noncomposite. These assumptions are probably conservative in most cases; however, to accurately
model real sections without requiring excessive steel area, it
is important to account for the variation of the steel location,
the section shape, and the effects of composite construction.
The following derivation takes all of these variations into
account by resolving the forces into the internal and external
moments acting on a section. To illustrate the proposed
method, consider a rectangular section with nonprestressed
steel and prestressing steel reinforcement as shown in Fig. 3.
The left stress profile shows the section under effective
prestress after time-dependent losses have occurred; fse is the
effective prestress in the tendon, and fpe is the compressive
stress in concrete at the extreme fiber of section where tensile
stress is caused by externally applied loads. The center
stress profile shows an externally applied moment causing
the concrete tensile stress at the bottom of the section to be
exactly equal in magnitude to fpe. Superimposing the two
stress profiles results in a net stress at the extreme fiber of
zero. The applied moment required to cause this stress state
is defined as the decompression moment Mdec. While this
moment will cause an increase in prestressing steel stress,
in typical prestressed concrete members, this increase is not
considered significant.
As moment is applied beyond the decompression moment,
the net stress in the bottom of the section becomes tensile.
The point at which the tensile stress at the extreme fiber is
equal to the modulus of rupture defines cracking; this applied
moment is shown as the right stress profile and is referred to
as Mncr, or the net cracking moment. The moment applied to
434

Fig. 4External and internal forces diagram.


achieve zero stress at the extreme tensile fiber is differentiated herein from the moment subsequently applied to achieve
the rupture stress at the extreme tensile fiber. The cracking
moment Mcr is then defined as the sum of the decompression
and net cracking moment, Mdec + Mncr, and is used to determine the minimum quantity of prestressing steel necessary
to ensure a ductile failure mode. This difference in definition
is illustrated in Fig. 3. Leonhardt (1964) reasoned (reframed
herein in terms of moment rather than force) that because
the prestressing steel necessary to resist in Mdec is already
present, then additional reinforcement or prestressing steel is
needed only to ensure minimum strength to support Mncr. To
ensure that the structure remains stable after cracking, sufficient reinforcement should be present such that the moment
capacity is greater than the cumulative applied moment
Mdec + Mncr Mn. Figure 4 shows these external and internal
moments acting on a rectangular section.
The left side of Fig. 4 illustrates the stress distribution in
a concrete section just before cracking. If the stress at the
extreme fiber of the concrete is equal to the rupture strength,
then the net cracking moment can be defined as
M ncr =

fr I
(7)
yt

ACI Structural Journal/March-April 2014

which ACI currently employs as an estimate of concrete


tensile strength when determining minimum reinforcement.
Further, as discussed previously, additional investigations
by ONeill and Hamilton (2009) justify the use of this estimate for the calculation of minimum steel. The net cracking
moment described in Eq. (7) can then be written
M ncr = 7.5 fc

I
(8)
yt

The force coupled on the right side of the figure represents


the internal forces immediately after cracking, in which the
reinforcement resists the tensile force carried by the concrete
before cracking. If the reinforcement volume is sufficiently
low, then the section will reach its nominal moment capacity
immediately after cracking; low volumes of reinforcement
will ensure the section is tension-controlled. Defining the
internal moment arms of the nonprestressed and prestressed
steel reinforcement for this condition as jd and jdp, respectively, the moment capacity can then be defined as
Mn = Ts(jd) + Tps(jdp) (9)
If the extreme tension fiber stresses caused by the prestress
force and decompression moment shown in Fig. 5 are
summed, the result is the following equation

I
yt
1
M dec = Aps fse
A + e I

(10)
Ay
t final
transfer

This approach results in the self-weight moment being


carried by the composite section, which results in a conservative minimum steel requirement. The transfer term should
incorporate the section properties at the time of prestress
transfer, and the final term should incorporate the section
properties of the element after construction is complete.
For composite construction, the transfer term will typically be the noncomposite section properties, and the final
term will be the composite section properties. For noncomposite sections, these terms are simply the noncomposite
sectionproperties.
To ensure sufficient flexural capacity immediately after
cracking, the resisting moment should be greater than or
equal to the total applied moment. By Leonhardts reasoning,
the decompression moment is resisted by the already-present
prestressing steel; the 1.2 (included to ensure ductility) need
only be considered for Mncr. Further, when the safety factor
of 1.2 used by ACI 318-11 and the strength reduction factor
are included, the relationship can be written as

Mn 1.2Mncr + Mdec (11)

Using Leonhardts approach and the strength capacity of


the prestressing steel beyond the effective prestress to resist
the tensile force in the concrete, the relationship described in
Eq. (11) may be rewritten as

ACI Structural Journal/March-April 2014

Fig. 5Stress states under decompression moment.


fAps f ps jd p + fjdAs f y 1.2

7.5 fcI
yt

I
yt
1
+ Aps fse
+ e I
(12)
Ayt final A
transfer

As tension failure is the intended failure mode = 0.9.


Assuming that j = 0.9

9 fcI
I
yt
1
+ 0.8dAs f y
Aps 0.8 f ps d p fse
+ e I
yt
Ayt final A

transfer

(13)

When the member contains only prestressing steel, then


the minimum requirement for bonded prestressed reinforcement is simplified as
Aps,min =

9 fc
y
y
1
(14)
0.8 t
+e t
f d f
I final ps p se A
I transfer

When the member contains only nonprestressed steel,


then the minimum requirement is simplified as
As,min

11.25 fcI
df y yt

(15)

To illustrate the implementation of Eq. (14), consider


a typical precast bridge with a cast-in-place topping. To
calculate the minimum reinforcement using the proposed
minimum, the designer would use the moment of inertia,
area, and yt of the precast section for the transfer terms,
as these were the section properties during the prestress
transfer. For the final terms, the designer would consider the
section properties of the completed structure: the composite
moment of inertia, area, and yt, all considering the cast-inplace deck.
The proposed provisions are in agreement with the fundamental approach of providing a moment capacity that is
greater than Mcr to ensure a ductile failurea provision
that has been in the code either explicitly or implicitly
for many years, and has empirically shown itself to be a
successful design practice. The proposed provisions provide
two things: unification of the minimum reinforcement
provisions to include both prestressed and nonprestressed
concrete sections (to provide less confusion and a more
uniform level of safety) and revision of the provisions to
eliminate the explicit consideration of the cracking moment
435

for prestressed sections (as such, avoiding the problem that


arises in the case of prestressed concrete systems in which
the minimum prestressing cannot be reached.)
PARAMETRIC STUDY: COMPARISON OF
CURRENT AND PROPOSED PROVISIONS
A parametric study was conducted to compare the
minimum area of steel and the subsequent provided capacity
as calculated using the proposed provisions against those
calculated using the ACI 318-11 minimum reinforcement
provisions. A wide range of geometries was chosen for
investigation: standard precast sections (single tees, double
tees, inverted tees, and hollow core slabs), a Florida bulb tee,
a box girder, and an AASHTO girder. Sections were selected
to include geometries commonly used in construction, as
well as both top-heavy and bottom-heavy sections, as the
proposed provisions are dependent on the centroid of the
gross area relative to the centroid of the tension area. Both
nonprestressed and prestressed sections were investigated.
For each cross section, the minimum required area of steel
was calculated according to ACI 318-11 and the proposed
provisions. The cracking moment of the geometry was
calculated by assuming the concrete tensile strength based
on a modulus of rupture of 7.5fc psi (0.62fc MPa). The
ultimate design strength was calculated using strain compatibility. The ultimate flexural capacity and the theoretical
cracking moment were then compared to determine the
margin between cracking and flexural failure.
PRESTRESSED CONCRETE: NONCOMPOSITE
AND COMPOSITE SECTIONS
Comparison of the current and proposed provisions is
awkward because ACI 318-11 minimum reinforcement
provisions (hereafter referred to as ACI minimum) depend
on a moving targetthat is, the cracking moment that
changes as the amount of bonded prestressing steel changes.
On the contrary, the proposed minimum reinforcement
provision (hereafter referred to as proposed minimum) does
not depend directly on the cracking moment; it depends,
instead, on only the material and geometric properties of
the section. Defined as such, the proposed minimum is like
the AASHTO LRFD minimum reinforcement requirement,
encompassing both types of reinforced sections while basing
the minimum steel requirement on the concrete section properties rather than the cracking moment.
A wide variety of prestressed sections was investigated
in this parametric study. A table showing the ACI and
proposed minimums for all geometric sections investigated by this parametric study is included as Appendix A.
Prestressing strand was assumed to be Grade 270. Prestress
losses due to elastic shortening and time-dependent losses
were assumed to be 20% after an initial strand stress of
0.74fpu. The sections in the following discussion were
chosen as representative of the range of all standard geometries analyzed in the study and include a 12 ft (3.66 m) wide
by 30 in. (76.2cm) deep double-tee (12DT30), a 28 in. (71.1
cm) wide by 24 in. (61cm) deep inverted tee (28IT24), a 4 ft
(1.22 m) wide by 12 in. (30.5 cm) deep hollow core (12HC),
an AASHTO IV girder (AIV), a 78 in. (1.98 m) deep Florida
436

Fig. 6Prestressed noncomposite sections: proposed and


ACI minimums.
bulb tee (FBT78), and a segmental box girder (SB). The
SB was considered under negative bending (tension at the
top of section). All other sections were considered under
positive bending. The nominal moment capacity was calculated using strain compatibility, and the cracking moment
was calculated using Eq. (5). Nonprestressed steel was not
considered in these calculations. Appendix A tabulates data
for all investigated sections.
A comparison of the ACI minimum and proposed
minimum (from Eq. (14)) reinforcement quantities for
noncomposite sections is shown in Fig. 6. The sections were
also evaluated by calculating the ratio of the design strength
Mn to the cracking moment Mcr; this relation is shown in
Fig. 7. For noncomposite sections, the computed minimum
area of prestressing steel is within approximately 10% of
the area currently required by the ACI minimum. It can be
observed that for all sections, the proposed minimum area of
steel results in a Mn/Mcr ratio greater than 1, and is close to
the 1.2Mcr limit. Notably, comparison of the design moment
strength across a variety of sections indicates a nearly
constant excess of moment strength (around 17%) beyond
the cracking limit of each section, ensuring a ductile failure
mode for each section as is intended by the ACI minimum.
Use of the proposed method eliminates the issue that as a
section increases in depth, the moment capacity increases at
a faster rate than the cracking moment. As Fig. 8 illustrates, a
series of varying depth noncomposite hollow core slabs were
examined, including a 6 in. (15.2 cm) deep (6HC), an 8 in.
(20.3 cm) deep (8HC), a 10 in. (25.4 cm) deep (10HC), and
a 12 in. (30.5 cm) deep (12HC) slab. Appendix A tabulates
these data for all investigated sections. Figure 8 also shows
that as the section depth increases, the ratio of the provided
moment capacity to the cracking moment remains constant.
Furthermore, the proposed minimum provides adequate
reinforcement to ensure ductile failure mode in every case.
A deck was added to each of the noncomposite sections
evaluated previously (with the exception of the SB) to
investigate the results of the proposed minimum reinforceACI Structural Journal/March-April 2014

Fig. 7Prestressed noncomposite


capacity versus cracking moment.

sections:

moment
Fig. 9Prestressed composite sections: proposed and ACI
minimums.

Fig. 8Prestressed noncomposite hollow core slabs with


proposed minimum.
ment provisions for composite sections. Figure 9 shows the
proposed minimum reinforcement quantity normalized by
the ACI minimum. The proposed minimum was compared
against the ACI minimum calculated for both shored and
unshored construction to demonstrate the full range of
expected discrepancies between the two methods. The
sections shown in Fig. 9 were also evaluated by calculating
the ratio of the design moment strength Mn to the cracking
moment Mcr using the minimum area of steel prescribed by
the proposed method. The results are shown in Fig. 10.
For composite sections, with the exception of the inverted
tee, a 15 to 61% increase in minimum reinforcement over the
ACI minimum was found. The inverted tee is even larger,
with a 94 to 112% increase. Furthermore, the difference is
greatest when comparing the proposed minimum to ACI
minimum for an unshored condition. This is, in part, due to
the conservative assumption made in the derivation in which
the self-weight of the section is assumed to be carried by the
composite section. To evaluate the impact of this increase,

ACI Structural Journal/March-April 2014

Fig. 10Prestressed composite sections: moment capacity


versus cracking moment.
typical design volumes of prestressing steel were determined for each of the sections considered in the parametric
study. Typical design loads were taken from load tables (if
appropriate), or actual highway designs. Figure 11 shows
how the proposed minimum reinforcement compares to the
volume of reinforcement required by the design. In most
cases, the differences between the current ACI minimum
and the proposed minimum are minor. In general, as can be
observed in Fig. 11, the proposed minimum area of steel is
always less than, and usually much less than, that of a typical
design. Only in the hollow core slab case does the proposed
minimum approach the typical design value. As Fig. 11
demonstrates, the area of steel is usually controlled by either
stress or strength limits, and the proposed minimum provisions will not affect most designs.

437

Fig. 11Proposed area of steel versus typical design.


NONPRESTRESSED CONCRETE: CURRENT
VERSUS PROPOSED
The current ACI minimum for concrete reinforced with
nonprestressed steel requires a minimum area of steel, as
prescribed by Eq. (2). In general, minimum steel requirements
are usually invoked when sections are larger than required
for strength, such as for architectural or other reasons. To
evaluate the effect the proposed provisions would have in
these instances, the minimum steel requirement was allowed
to control the quantity of steel in the section. The steel quantity was then compared with the current ACI minimum.
Two standard reinforced geometries were analyzed as
nonprestressed concrete, a 10 ft (3.05 m) wide top flange by
48 in. (1.22 m) deep single-tee with an 8 in. (20.32 cm) web
(MS10ST48) and a 28 in. (71.1 cm) wide by 60 in. (1.52m)
deep inverted tee (MS28IT60), as well as three rectangular sections of varying depth: 12 x 12 in. (30.5x 30.5
cm), 12 x 36 in. (30.5 x 91.4 cm), and 12 x 72 in. (30.5x
182.9 cm). All sections were assumed to be under positive
bending (compression in the top of the section). A comparison of ACI and proposed minimums is shown in Fig. 12.
AppendixB tabulates these data for all investigatedsections.
The sections were also evaluated by calculating the ratio
of the design moment capacity Mn to the cracking moment
Mcr; these ratios are shown in Fig. 13. The nominal moment
capacity was calculated using strain compatibility, and the
cracking moment was calculated using Eq. (5).
Figure 12 highlights the conservative nature of the current
ACI minimum. In the sections selected for this parametric
study, the ACI minimum for nonprestressed members
provide between 1.25 and 2.0 safety margins, growing excessively conservative in sections with tensile zones relatively
larger than their compressive zones. Parametric studies
completed by Seguirant et al. (2010) demonstrated the variability of the safety margin provided by the ACI minimum
for nonprestressed sections from slightly under-conservative
to extremely over-conservative (up to 4.43 margin of safety).
In most cases, the ACI minimum requires more steel than
the proposed provisions. Despite the general reduction in

438

Fig. 12Nonprestressed sections: proposed and ACI


minimums.

Fig. 13Nonprestressed sections: moment capacity versus


cracking moment using proposed provision and ACI 318-11
(ACI Committee 318 2011) Eq. (10-5).
steel quantity, as inspection of the sections moment capacity
and cracking moment reveals, the proposed method still
ensures a ductile failure mode. The rectangular sections and
the inverted tee reveal the consistent level of conservatism
achieved by the proposed method with less steel. As calculated using the proposed provisions, these sections require
approximately 15 to 30% less steel than the ACI minimum,
and yet provide a uniform moment capacity near 1.6Mcr.
In fact, an inspection of all of the nonprestressed sections
reveals that the proposed method, unlike ACI, provides a
consistently conservative design with ductile failure mode
the proposed minimum requirement for each section type,
regardless of the section geometry, provides a capacity equal
to approximately 1.6Mcr. Considering this, the proposed
provisions, although requiring less steel in some cases,
provide a more reliable estimate of minimum reinforcement
required, and one that is independent of cracking moment.
ACI Structural Journal/March-April 2014

SUMMARY
The derivation and parametric studies described in this
paper provide a direct and unified method of determining
minimum steel requirements for both nonprestressed reinforced and prestressed sections. For composite prestressed
sections, the moment capacity provided by the area of steel
determined with the proposed minimum always exceeds
1.2Mcr. While the proposed method requires more steel
for composite prestressed sections than the ACI 318-11
minimum reinforcement provision, the difference is negligible when considering the steel typically provided for
design. For nonprestressed sections, the proposed method
results in a consistent moment capacity of 1.6Mcr, regardless of the section shape. Noncomposite prestressed sections
have a consistent moment capacity of 1.17Mcr. The proposed
method has several advantages for a designer; it is independent of the cracking moment and explicitly calculated.
AUTHOR BIOS

ACI member Natassia R. Brenkus is a Graduate Research Assistant at the


University of Florida, Gainesville, FL, where she received her BS in 2006
and her ME in 2012.
H. R. Hamilton, FACI, is the Byron D. Spangler Professor of Civil Engineering at the University of Florida. He is past Chair of Joint ACI-ASCE
Committee 423, Prestressed Concrete, and a member of ACI Subcommittee
318-6, Precast and Prestressed Concrete (Structural Concrete Building
Code).

ACKNOWLEDGMENTS

The authors would like to thank C. ONeill for her contributions to this
paper and the Florida Department of Transportation for providing funding
that supported this work.

=
=
=
=
=
=
=

NOTATION

area of concrete cross section


area of concrete in flexural tension
area of prestressing steel in flexural tension zone
area of nonprestressed longitudinal tension reinforcement
web width, or diameter of circular section
width of web in I- or T-beam sections
depth from compression face to center of nonprestressed tensile
reinforcement
dp = depth from compression face to center of prestressed tensile
reinforcement
e
= distance between prestressed reinforcement and centroidal axis
fc = specified compressive strength of concrete
fcpe = compressive stress in concrete due to effective prestress forces
only (after allowance for all prestress losses) at extreme fiber of
section where tensile stress is caused by externally applied loads
fpe = compressive stress in concrete due to effective prestress forces
only (after allowance for all prestress losses) at extreme fiber of
section where tensile stress is cause by externally applied loads
fps = stress in prestressing steel at nominal flexural strength
fr
= modulus of rupture of concrete
fse = effective stress in prestressing steel (after allowance for all
prestress losses)
fy
= specified yield strength of reinforcement
I
= moment of inertia of section about centroidal axis
j
= distance between compressive resultant force and tensile resultant force
Mcr = cracking moment, computed as sum of decompression and net
cracking moment, Mdec + Mncr
Mdec = applied moment causing the net stress at bottom of section to be
zero
A
Act
Aps
A s
bw
b
d

ACI Structural Journal/March-April 2014

Mdnc =

unfactored dead load moment acting on monolithic or noncomposite section


Mn = nominal flexural strength at section
Mncr = moment applied beyond decompression moment to reach
rupture stress in extreme tensile fiber; net cracking moment
Mr = flexural resistance
Mu = factored moment at section
p
= ratio of area of tension reinforcement to effective area of concrete
Sc = section modulus with respect to extreme tensile fiber of
composite section
Snc = section modulus with respect to extreme tensile fiber of monolithic or noncomposite section
Tps = tensile force in prestressed steel
Ts = tensile force in nonprestressed steel
yt = distance from centroidal axis of gross section, neglecting reinforcement, to tension face

= strength reduction factor

REFERENCES

AASHTO, 2007, AASHTO LRFD Bridge Design Specifications,


4th edition with 2007 interim revisions, American Association of State
Highway and Transportation Officials, Washington, DC, pp. 5-44 through
5-45.
ACI-ASCE Committee 323, 2004, Landmark Series: Tentative Recommendations for Prestressed Concrete, Concrete International, V. 26, No.
3, Mar., pp. 95-131.
ACI Committee 318, 1963, Building Code Requirements for Reinforced Concrete (ACI 318-63), American Concrete Institute, Farmington
Hills, MI, 144 pp.
ACI Committee 318, 1986, Building Code Requirements for Reinforced Concrete (ACI 318-83) (Revised 1986), American Concrete Institute, Farmington Hills, MI, 114 pp.
ACI Committee 318, 1992, Building Code Requirements for Structural Concrete (ACI 318-92) and Commentary (ACI 318R-92), American
Concrete Institute, Farmington Hills, MI, 347 pp.
ACI Committee 318, 1995, Building Code Requirements for Structural Concrete (ACI 318-95) and Commentary (ACI 318R-95), American
Concrete Institute, Farmington Hills, MI, 373 pp.
ACI Committee 318, 2008, Building Code Requirements for Structural
Concrete (ACI 318-08) and Commentary, American Concrete Institute,
Farmington Hills, MI, 473 pp.
ACI Committee 318, 2011, Building Code Requirements for Structural
Concrete (ACI 318-11) and Commentary, American Concrete Institute,
Farmington Hills, MI, 503 pp.
ACI Committee 501, 1936, Building Regulations for Reinforced
Concrete, ACI Journal, V. 32, No. 3, Mar., pp. 407-444.
Bureau of Public Roads, 1954, Criteria for Prestressed Concrete
Bridges, U. S. Government Printing Office, Washington, DC, 25 pp.
Freyermuth, C. L., and Aalami, B. O., 1997, Unified Minimum Flexural Reinforcement Requirements for Reinforced and Prestressed Concrete
Members, ACI Structural Journal, V. 94, No. 4, July-Aug., pp. 409-420.
Ghosh, S. K., 1986, Exceptions of Precast Prestressed Concrete
Members to Minimum Reinforcement Requirements, PCI Journal, V. 31,
No. 6, pp. 74-91.
Khan, A. A.; Cook, W. D.; and Mitchell, D., 1996, Tensile Strength of
Low, Medium, and High-Strength Concretes at Early Ages, ACI Materials
Journal, V. 93, No. 5, Sept.-Oct., pp. 487-493.
Kleymann, M.; Girgis, A.; Tadros, M. K.; and Vranek, C. J., 2006,
Open Forum Problems and SolutionsMinimum Reinforcement in Flexural Members, PCI Journal, V. 51, No. 5, Sept.-Oct., pp. 146-148.
Leonhardt, F., 1964, Prestressed Concrete Design and Construction,
Wilhelm Ernst & Sohn, Germany, 677 pp.
ONeill, C. M., and Hamilton, H. R., 2009, Determination of Service
Stresses in Pretensioned Beams, No. BD 545-78, Florida Department of
Transportation.
Oladapo, I. O., 1968, Relationship between Moment Capacity at
Flexural Cracking and at Ultimate in Prestressed Concrete Beams, ACI
Journal, V. 65, No. 10, Oct., pp. 863-875.
Seguirant, S. J.; Brice, R. B.; and Khaleghi, B., 2010, Making Sense of
Minimum Flexural Reinforcement Requirements for Reinforced Concrete
Members, PCI Journal, V. 55, No. 3, Summer, pp. 64-85.
Walker, S., and Bloem, D. L., 1960, Effects of Aggregate Size on Properties of Concrete, ACI Journal, V. 32, No. 3, Mar., pp. 283-298.

439

APPENDIX A
f = 0.9

Area, in.2

Mn, kip-ft

Mcr, kip-ft

fMn/Mcr

2.448

1067.9

629.4

1.53

0.924

411.0

309.5

1.20

ACI min

0.931

414.1

311.0

1.20

Table 18X8

2.750

532.4

506.4

0.95

0.481

214.6

165.6

1.17

0.520

230.3

171.3

1.21

0.510

47.2

30.7

1.39

0.314

30.1

23.2

1.17

ACI min

0.339

32.3

24.2

1.20

Table 6X6

0.510

70.2

48.6

1.30

e, in.
Table 8X16

12DT30

28IT24

Prop min

Prop min

10

2.73

ACI min
Table 6X6
6 in. hollow core

8 in. hollow core

10 in. hollow core

12 in. hollow core

Prop min

Prop min

1.5

0.382

53.6

41.1

1.17

ACI min

0.405

56.4

42.3

1.20

Table 4X8

0.612

110.7

77.3

1.29

0.467

85.7

65.8

1.17

Prop min

1.5

1.5

ACI min

0.499

91.2

68.3

1.20

Table 7X6

0.595

135.1

97.1

1.25

0.502

114.2

87.6

1.17

0.535

121.1

90.8

1.20

7.350

3157.8

4162.0

0.68

1.309

1284.4

990.0

1.17

ACI min

1.405

1373.1

1028.7

1.20

Typical

7.490

11403.3

6704.2

1.53

Prop min

1.5

ACI min
Typical
AASHTO Type IV

FBT78

Box girder

Prop min

Prop min

2.237

3453.8

2611.8

1.19

ACI min

2.290

3533.6

2649.7

1.20

Typical

65.290

209444.4

173166.7

1.09

45.148

147777.8

115096.7

1.16

51.100

166235.6

124574.2

1.20

Prop min

6.75

ACI min
Notes: 1 in. = 2.54 cm; 1 in.2 = 6.45 cm2; 1 kip-ft = 1.36 kN-m.

APPENDIX B
f = 0.9
MS 10ST48

MS 28IT60

MS 12X12

MS 12X36

MS 12X72

Area, in.2

Mn, kip-ft

Mcr, kip-ft

fMn/Mcr

1.2Mcr

1.310

281.1

210.8

1.20

prop min

1.838

393.3

210.8

1.68

ACI Sect 10-5 min

1.358

291.4

210.8

1.24

1.2Mcr

2.895

763.2

570.7

1.20

prop min

3.932

1024.4

570.7

1.62

ACI Sect 10-5 min

4.619

1194.4

570.7

1.88

1.2Mcr

0.310

15.2

11.4

1.20

prop min

0.416

20.7

11.4

1.63

ACI Sect 10-5 min

0.389

18.8

11.4

1.48

1.2Mcr

0.820

138.0

102.4

1.21

prop min

1.122

187.6

102.4

1.65

ACI Sect 10-5 min

1.300

215.6

102.4

1.89

1.2Mcr

1.580

544.4

409.8

1.20

prop min

2.188

750.9

409.8

1.65

ACI Sect 10-5 min

2.666

911.7

409.8

2.00

Notes: 1 in. = 2.54 cm; 1 in.2 = 6.45 cm2; 1 kip-ft = 1.36 kN-m.

440

ACI Structural Journal/March-April 2014

ACI STRUCTURAL JOURNAL

TECHNICAL PAPER

Title No. 111-S39

Lateral Strain Model for Concrete under Compression


by Ali Khajeh Samani and Mario M. Attard
The relationship between the lateral and axial strain is important
when predicting the confinement stresses within reinforced
concrete or fiber-reinforced polymer confined columns. Difficulties in measuring reliable lateral strains in triaxial compressive
experiments mean that there is a scarcity of lateral strain experimental results. Two recent lateral strain models will be compared
with available experimental results. Discussed in this paper is the
transition point in the lateral and axial strain relationship at which
the volumetric strain changes sign, and how this transition point is
related to the peak stress. A lateral strain-versus-axial strain model
is proposed based on the supposition that the concrete behaves
linear elastically in the early stages of loading. Once microcracks
form, nonlinear hardening occurs up to the peak stress. After the
peak stress, the inelastic lateral strain varies linearly with the
inelastic axial strain. The lateral-to-axial inelastic strain ratio
is shown to be a function of the lateral confinement level and the
failure mechanism. Moreover, the shear band and tensile crackinginduced size effect is also discussed from the lateral strain versus
axial strain perspective.
Keywords: confined concrete; fracture energy; lateral strain; size effect;
volumetric strain.

INTRODUCTION
There are many design formulas for predicting the
behavior of reinforced concrete column behavior. Many of
these formulas depend on estimating the lateral confinement.
The lateral ties providing confinement are usually assumed
to be at yield when a column reaches its peak load-carrying
capacity. For reinforced high-strength concrete columns,
the confinement ties are generally not at yield at the peak
column load-carrying capacity. Therefore, this implies that
formulas which rely on the confining ties to be at yield are
not applicable. To overcome this problem, the lateral expansion of the confined column core can be related to the stress
in the confining ties, which can then be used to predict the
core confinement. The reinforcing ties are assumed to be
fully bonded to the concrete, with no sliding or slippage
occurring during the loading of the column. The lateral
deformation of the concrete and the confining reinforcement
cage or fiber-reinforced polymer (FRP) wrap is then taken
as equal and related to axial strain in the column core. The
lateral strain relationship has been used in the prediction of
the confinement level for columns confined by reinforcement cages by Cusson and Paultre1 and Ahmad and Shah,2
and FRP wraps by Talaat and Mosalam3 and Lokuge et al.4
RESEARCH SIGNIFICANCE
To predict the lateral expansion of axially loaded concrete,
a lateral strain-versus-axial strain relationship is needed.
Baseline experimental data on measured lateral strain from
uniaxial and triaxial loaded concrete are limited to a few
research publications such as Candappa et al.,5 Hurlbut,6
ACI Structural Journal/March-April 2014

Imran and Pantazopoulou,7 Jamet et al.,8 Lee and Willam,9


Lu and Hsu,10 Newman,11 and Smith et al.12 The major focus
of this paper is to establish an analytical relationship between
lateral and axial strain for concrete under uniaxial compression or compression under confinement. Such a model is
important when determining the confinement and ductility
of reinforced and FRP confined concrete columns. Furthermore, specimen size effects associated with the formation of
a shear band and tensile cracking during softening are also
incorporated into the model.
LATERAL CONCRETE STRAIN UNDER
COMPRESSION
The lateral strain behavior of concrete under compression
is a key aspect of concrete behavior. Studies on the lateral
strain behavior under axial compression are conducted under
various loading regimes. The loading conditions referenced
herein are mainly uniaxial and triaxial tests conducted on
cylindrical specimens. Triaxial testing of concrete is usually
conducted in Hoek cells or triaxial cells. In a Hoek cell, the
loading path begins with a sitting or locking axial load. After
securing the specimen, the lateral confinement is enforced
up to the desired confinement level. Once the confinement
reaches the desired value, the axial load is then gradually
applied while the confining stress is maintained. Because the
lateral confining stress is applied using liquid pressure and the
axial strain is externally applied, it is very difficult to enforce a
hydrostatic load state without the use of complicated computerized devices. The triaxial cell used on concrete specimens is
similar to those used for triaxial testing of soils.
It is important to note the limitations of the various techniques used to extract or estimate the lateral strain. Lateral
strains can be measured by utilizing strain gauges on the
specimens or measuring the deformation of the circumference via a lateral ring. Strain gauges can show some inaccuracy under high confining pressures inside the test cell. Moreover, the lateral strain can be drastically different at different
locations on the specimen. The change in the volume of oil
used to apply the confinement can also be used to measure
the volumetric strain, which can then be used to estimate
an average lateral strain. For an intact specimen, the lateral
strain obtained from the volumetric changes in the specimen
correlates well with the lateral strain measured from strain
gauges and linear variable displacement transducers (LVDTs)
placed at several locations on the surface of the specimen. If
ACI Structural Journal, V. 111, No. 2, March-April 2014.
MS No. S-2012-177.R3, doi:10.14359.51686532, was received March 27, 2013, and
reviewed under Institute publication policies. Copyright 2014, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including authors
closure, if any, will be published ten months from this journals date if the discussion
is received within four months of the papers print publication.

441

Fig. 1Typical: (a) lateral strain; and (b) volumetric strain diagrams for concrete subjected to increasing levels of confinement. (Note: 1 MPa = 145 psi; 1 mm = 0.0394 in.)
localization is present, however, the lateral strains obtained
from the changes in oil volume and the strains obtained from
strain measurement devices are drastically different.
Lateral strain results are presented as either a set of axial
stress-versus-axial strain and axial stress-versus-lateral
strain or volumetric strain-versus-axial strain diagrams
(Fig.1). If volumetric strains are used to obtain the lateral
strain, the obtained lateral strain is an average strain for the
whole specimen. Consider, for example, the situation where a
specimen under confinement and axial compression is loaded
past the peak and reaches a residual axial load level after
softening. If a localized shear band has formed, the applied
axial displacements from the loading machine only impose
a sliding displacement on the specimen. This displacement
does not represent any change of specimen volume, but corresponds to sliding across the shear band. In this case, the actual
change in volume is zero, but the plastic strain or permanent
displacement across the whole specimen is not zero. If LVDTs
are used across the fractured zone, a displacement will be
measured, and a plastic or irrecoverable strain estimated.
Analytical models by Lokuge et al.4 and Binici13
Lateral strain versus axial strain models developed by
Lokuge et al.4 and Binici13 are presented as follows, and
are later compared with some experimental results. These
two models are analytical models for the total lateral strain,
and are functions of the axial strain. The model proposed by
Lokuge et al.4 is defined by

a e
vi e e1
e e0
=
(1)
e 0 e a
>
e
e
1
e
0

equal to 0.50. The strain 0 corresponding to the peak


stress is given by Eq. (3), and fc is the uniaxial compressive
strength of concrete in MPa.

f
e 0 = e c 1 + (17 0.06 fc) r (fc in MPa; 1 MP
Pa = 145 psi) (3)
fc

In the previous equation, fr is the lateral confinement


on the specimen. The strain at the peak stress for uniaxial
compression, c, is taken as 0.002 for all concrete grades.
In the model proposed by Binici,13 the secant ratio s
is used to correlate the lateral deformation with the axial
deformation. The concrete behavior in the lateral direction is
divided into three regions: the elastic region, inelastic hardening region, and softening region. The model is defined by

= s

in which, is the lateral strain, is the axial strain, and s is


the secant ratio, defined by
s = e

e ee 2
s = l ( l e ) exp

ee =

(fc in MPa, 1 MPa = 145 psi)


and is the axial strain, is the lateral strain, 1 can be
obtained by equating the two right hand side expressions in
Eq. (1), and 0 is the lateral strain at the peak axial stress

e > ee

(5)

f
f
f
fc 0.1 1 + 9.9 r 0.9 r + r

fc
fc
fc

4750 fc

l = p +

a = 0.0177fc + 1.2818; vi = 8 10 (fc) + 0.0002fc + 0.138 (2)

442

e ee

In the previous equation, e is the Poissons ratio of the


concrete, which varies between 0.15 to 0.2; e, l, and are
defined by

where
a

(4)

1
fr

f + 0.85
c

(6)

e0 ee
l p
ln
l 0

ACI Structural Journal/March-April 2014

Fig. 2Comparison of Jamet et al.8 test results with: (a) Binici13; and (b) Lokuge et al.4 models. (Note: 1 MPa = 145 psi; 1
mm = 0.0394 in.)
Herein, p is the ratio of lateral strain to the axial strain at
the peak stress and assumed to be 0.5, l is the largest secant
ratio, and 0 is the axial strain at the peak stress defined by

e 0 = 5 0.067 fc 2 + 29.9 fc + 1053

f
f
1 + 9.9 r + r 0.8 10 6
fc fc

(7)

Comparing existing models with experimental


results for confined concrete
In this section, the two analytical models described previously are compared with available experimental results. In
the following comparisons, the elastic moduli and the value
of the axial strain at the peak stress used in the analytical
models is adopted from the experimental results. The first set
of results are from Jamet et al.,8 who tested microconcrete
in triaxial and uniaxial compression. The results of Jamet
etal.8 for lateral strain were extracted from measurements
of the volumetric strain rather than from LVDTs, strain
gauges, or both. The cylindrical specimens had a diameter
of 110mm (4.3 in.) and height of 220 mm (8.7 in.). The
uniaxial compression strength was approximately 26 MPa
(3.8 ksi). The applied confinement levels were 0, 3, 10, 25,
50, and 100 MPa (0, 0.4, 1.5, 3.6, 7.3, and 14.5 ksi). Jamet
et al.8 presented results using true strain rather than engineering or simple strain values. It can be shown that for
the magnitudes of strains measured, the true strain and the
corresponding engineering strain are virtually indistinguishable. All results presented herein are based on engineering
strains rather than true strains. The lateral strain results are
compared with the models proposed by Binici13 and Lokuge
et al.4 in Fig. 2.
Hurlbut6 and Willam et al.14 tested normal strength
concrete in direct tensile and uniaxial compression, as well
as triaxial compression. The ultimate uniaxial compressive
strength was 22 MPa (3.2 ksi). The size of the cylindrical
specimens in the uniaxial compression tests had a diameter
of 76 mm (3 in.) and a height of 152 mm (6 in.), while in the
triaxial tests, the specimens were 54 mm (2.125 in.) in diameter by 108 mm (4.25 in.) in height. A modified Hoek cell
ACI Structural Journal/March-April 2014

was used to impose the desired confinement pressure on the


triaxial test specimens. The confining pressures ranged from
0.69 to 13.79 MPa (0.1 to 2 ksi). The axial displacements
were transmitted to the specimens via steel rams without
any friction-reduction measures. Hurlbut6 observed that the
unconfined compression test exhibited the largest dilatant
behavior in the post-peak regime, with the radial displacement three times greater than the applied axial displacements. The results showed a large volumetric expansion in
the post-peak region. In agreement with the observations by
Imran and Pantazopoulou,7 the results of Hurlbut6 showed
that as the level of confinement was increased, the dilatancy
rapidly decreased while exhibiting a transition from large
volumetric expansions under low confinement to an elastic
volumetric compaction for high confinement. The results of
Hurlbut6 and Willam et al.14 are compared with the models
of both Binici13 and Lokuge et al.4 in Fig. 3.
Smith et al.12 and Smith15 tested 51 cylindrical samples
with dimensions of 54 x 108 mm (2.125 x 4.25 in.), which
were tested either in an unconfined or confined state using
a Hoek cell. The compressive strength was either 34.5 or
44.2MPa (5.0 or 6.4 ksi). No friction-reducing layer was used
between the steel rams and the specimen ends. The confinement applied to the specimens varied with a maximum of
34.5 MPa (5.0 ksi). The test results on 34.5 MPa (5.0 ksi)
concrete specimens are compared with Binici13 and Lokuge
et al.4 models in Fig. 4.
The analytical models proposed by Binici13 and Lokuge et
al.4 give a reasonable fit to the trend of the results in Fig.2
to 4, but not a good match to the lateral strain quantities.
Moreover, one can see that neither of the models predicts the
initial compaction in the lateral direction under high confinement before the specimen starts to expand.
The test results of Lee and Willam,9 who tested cylindrical specimens with three different heights in uniaxial
compression, are also compared in Fig. 5. All the specimens had a diameter of 76.2 mm (3 in.). The heights were
137.2, 91.44, or 45.72 mm (5.4, 3.6, and 1.8 in.). The peak
concrete strength was 30.7 MPa (4.5 ksi) for the specimens
with heights of 137.2 and 91.44 mm (5.4 and 3.6 in.), and
32MPa (4.6 ksi) for the specimen with a height of 45.72mm
(1.8 in.). The results obtained by Lee and Willam9 are shown
in Fig. 5, and demonstrate the effects of different specimen
443

Fig. 3Comparison of Willam et al.14 test results with: (a) Binici13; and (b) Lokuge et al.4 models. (Note: 1 MPa = 145 psi;
1 mm = 0.0394 in.)

Fig. 4Comparison of Smith15 test results with: (a) Binici13; and (b) Lokuge et al.4 models. (Note: 1 MPa = 145 psi; 1 mm =
0.0394 in.)
levels and concrete strengths, and to incorporate size effect
issues.
PROPOSED LATERAL STRAIN MODEL

Fig. 5Comparison of Lee and Willam9 test results compared


with Binici13 model and Lokuge et al.4 model. (Note: 1 mm=
0.0394 in.)
height on the measured lateral strains. Figure 5 also shows
the comparison with the models of Binici13 and Lokuge et
al.4 The lateral strain size effect was not considered in the
model of Lokuge et al.4 Binici13 did consider specimen size
effect on the axial strain-versus-axial stress relationship, but
not on the lateral strain, as the model of Binici13 considers
the lateral strain as a direct function of the axial strain. A new
model for predicting the lateral strain is developed in the
following section, which attempts to give a better predictor
of the lateral strain for a broad spectrum of confinement
444

Lateral strain at peak axial stress


Kotsovos and Newman16 proposed that the point where
the volume expansion begins is at the onset of unstable fracture propagation (OUFP), and that this should be considered as the true strength indicator of concrete. The load at
OUFP is usually very close to the peak load (approximately
90%), and it is observed that deformations increase significantly between the OUFP and the peak. Based on experimental results, Nielsen17 suggested that the axial strain at
the peak stress is 2.2 times greater than the absolute value of
the lateral strain at peak stress. Imran18 studied the effect of
water-cement ratio, concrete strength, and confinement level
on the lateral strain of concrete. Figure 6 shows the relationship between the axial compressive strain at the peak stress
0, and the axial compressive strain at the instant of zero
volumetric strain. A similar figure was originally presented
by Imran.18 Herein, results from the experimental studies
of Candappa et al.,5 Jamet et al.,8 Newman,11 and Smith15
have been augmented to Imrans results. The diagonal line in
Fig.6 would correspond to a situation where the volumetric
strain is zero at the instance of the peak load. Imran and
Pantazopoulou7 concluded that the axial compressive strain
ACI Structural Journal/March-April 2014

Fig. 6Axial strain at zero volumetric change versus axial


strain at peak stress. (Note: 1 mm = 0.0394 in.)
at the instant of zero volumetric strain is approximately
equal to the axial compressive strain at which the concrete
reaches its peak compressive axial stress. The volumetric
strain v at the peak axial stress can be written as

ev

e = e0

= e 0 + 2 e 0 = 0 (8)

where 0 is the lateral strain at the peak axial stress. Hence,


the ratio of the lateral strain 0 to the axial strain at the peak
stress 0 can be taken as

0 = 0/0 = 0.5

Post-peak plastic lateral strain


Vonk19 observed that in the first stage of loading, the
average lateral deformation versus average axial deformation is governed by Poissons ratio, which results in a
decrease of the volume of the specimen. In the second stage,
macrocracks are formed, causing dilatant behavior. After a
gradual change, a more-or-less constant dilatant behavior
was observed, which indicated that the process of macrocrack formation had stopped, and the final crack pattern had
been developed. The behavior of concrete after the peak
stress can be quantified by looking at the plastic or inelastic
deformations after the peak. Figure 7 plots the plastic lateral
strain minus the plastic lateral strain at the peak stress against
the plastic axial strain minus the plastic axial strain at the
peak stress based on the test results of Candappa et al.1 The
results show an almost linear relationship. For relatively low
levels of confinement associated with the results in Fig.7,
a shear band has probably formed with the deformations in
the axial and lateral directions concentrated on the deformations within the shear band. The slope of the linear trends in
Fig. 7 are associated with differing levels of confinement.
As the confinement level increases, the lateral deformation
rate decreases. An equation for this approximate linear trend
is therefore
p p(at peak) = [p p(at peak)] (10)

ACI Structural Journal/March-April 2014

in which p is the plastic axial strain, and p is the plastic


lateral strain. The quantities p(at peak) and p(at peak) are the
axial and lateral plastic strain at the peak stress level, respectively. If it is assumed that the elastic part of the strains can
be calculated using Hookes law, then Eq. (10) becomes
f
f
f
f
e r e . e 0 r e . 0
Ec
Ec
Ec
Ec

(9)

This assumption will be used in this study.

Fig. 7Plastic lateral strain minus plastic lateral strain


at the peak stress versus the plastic axial strain minus the
plastic axial strain at the peak stress for a 40 MPa concrete
specimens tested by Candappa et al.5 (Note: 1 MPa =
145psi; 1 mm = 0.0394 in.)

f
f
f
f
= b e e . r e0 0 e . r
Ec
Ec
Ec
Ec

(11)

Using Eq. (9), the previous equation becomes

f
f
e + e 0 e 0 + e 0
Ec
Ec

f
f
= b e e0 0
Ec
Ec

(12)

In the previous equation, f is the stress level at a strain of ;


f0 is the peak stress, and Ec is the elastic modulus of concrete.
The axial strain versus axial stress model for confined
concrete proposed by Samani and Attard20 will be used for
the relationship between the axial strain and axial stress.
Experimental estimates for the parameter as a function
of the confinement ratio, obtained from the tests of Candappa
et al.,5 Hurlbut,6 Jamet et al.,8 Lu and Hsu,10 Newman,11 and
Smith15 are shown in Fig. 8. The experimental data used in
Fig. 8 are for specimens with a height-diameter or heightwidth ratio of h/D or h/w = 2. The experimental results in
Fig.8 show a large scatter, particularly for the uniaxial
compression state. The parameter varies from approximately 2.5 (for h/D or w/h = 2) at the uniaxial state, to an
upper limit of 0.5 for high confinement.
To gain an understanding of the numerical range for the
parameter , consider a simple Mohr Coulomb nonassoci445

residual level under confinement, the dilatancy angle would


be zero if a shear band forms and there is purely friction type
sliding across the shear band. When the dilatancy angle is
zero, Eq. (15) would give
b

Fig. 8Variation of parameter with confinement. (Note:


1MPa = 145 psi.)
ated plasticity model. The model has an initial failure angle
f, and a plastic strain dilatancy angle . Assuming there is
a shear failure plane with an inclination angle measured
from the horizontal, the inclination of the failure shear plane
can be related to the Mohr Coulomb failure surface by

ff
+
(13)
4 2

The plastic strain rate acting on the shear failure plane can
be decomposed vectorially into a plastic normal strain rate,
and a plastic shear strain rate. Incorporating a nonassociated
flow rule, the relationship between the normal and shear
plastic strain rates is

tan ( ) =

e np
e sp

(14)

The relationship between the plastic axial strain rate and


plastic lateral strain rate is then defined by

e p tan ( )
=
e p
tan ()

(15)

For concrete, the value of the dilatancy angle is approximately 13 degrees.21 The Mohr Coulomb failure surface
angle (before softening) is usually approximately 30 degrees,
which results in a shear plane failure angle of approximately
60 degrees. Using Eq. (15), the ratio of the plastic lateral
strain rate to the plastic axial strain rate is then roughly
0.58. The ratio of the plastic lateral strain rate to the plastic
axial strain rate approximates the definition of the parameter
defined by Eq. (10). The experimentally measured expansion rate, as reflected by the parameter shown in Fig.8, for
the uniaxial, low confinement case, or both, is much greater
than predicted by assuming a shear plane failure and a Mohr
Coulomb failure surface. The estimate based on Eq.(15)
assumed the plastic strain dilatancy angle and the friction
angle to be constant during softening, while it is known
that these parameters change. When the load reduces to a
446

e p
= 1

e p

(16)

Referring to Fig. 8, the parameter is approximately 1


when the confinement ratio is approximately 10 to 20%.
In the study by Samani and Attard,20 it was observed that
the post-peak compressive fracture energy is not constant,
but varies with the level of confinement. The compressive
fracture energy increases with increasing confinement and
reaches a limit at a confinement ratio of approximately 10
to 20%. The compressive fracture energy then decreases
for increasing confinement until the compressive fracture
energy becomes zero after the transition from brittle to hardening failure. It was suggested by Samani and Attard20 that
for uniaxial compressive and triaxial compression with low
levels of confinement, tensile cracking exists along with the
formation of a shear band during failure (Fig. 9). In Fig. 9,
hd is the damage zone height taken to be between 2 to 2.5
times the width or diameter of the specimen. With increasing
levels of confinement, eventually tensile splitting is nullified, and the compressive fracture energy increases above
the uniaxial level. As the confinement is further increased,
the mode of failure changes, and is dominated by barreling
dispersed cracking and the compressive fracture energy
decreases to zero. Applying these observations to the lateral
strain phenomenon, the presence of tensile splitting coupled
with a shear band failure under uniaxial and low confinement could explain why the measured parameter is much
greater than predicted using Eq. (15) based on a simple Mohr
Coulomb failure criterion and the assumption of a single
shear band. Tensile splitting increases the magnitude of the
relative lateral dilation. Vonk19 observed that the formation
of a larger number of splitting cracks causes larger lateral
deformations. The plotted data for the parameter in Fig.8
is categorized in Fig. 10 to reflect the failure mode. When
the confinement ratio is less than 10 to 20%, the failure and
fracture is due to tensile splitting and shear failure, with the
associated less than 1. As the confinement increases,
tensile splitting is nullified, and failure is predominately a
shear band failure with an associated of between 1 to
0.5. Increasing the lateral confinement beyond the brittle
continuous hardening transition point (confinement ratio of
approximately 30 to 50%), the concrete behavior changes
from shear failure to ductile behavior. When ductile, the
parameter is 0.5.
Figure 11 plots experimental estimates for the parameter
for uniaxial compression as a function of the aspect ratio D/h
or w/h, where D is the diameter for a cylindrical specimen,
w is the width for a prism, and h is the specimen height.
The experimental results are from the work of Lee and
Willam,9 Van Mier,22 Hurlbut,6 Lu and Hsu,10 Newman,11
and Smith.15 As mentioned previsously, Lee and Willam9
tested cylindrical specimens in uniaxial compression with
ACI Structural Journal/March-April 2014

Fig. 10Concrete behavior based on equation.

Fig. 9Shear band failure plane and tensile splitting under


uniaxial compression.
three different heights of 137.2, 91.44, and 45.72 mm (5.4,
3.6, and 1.8 in.), and a diameter of 76.2 mm (3 in.). Results
by Van Mier22 are also shown. Van Mier22 tested prisms with
a cross section of 100 x 100 mm (4 x 4 in.) and varying
heights of 50, 100, and 200 mm (2, 4, and 8 in.). Both Lee
and Willam9 and Van Mier22 measured the axial and lateral
strain for specimens with aspect ratios h/D or h/w less than
or equal to 2. The results show a large scatter, which could
be attributed to the increasing dominance of tensile cracking
for specimens with aspect ratios less than 2.
An equation for the parameter defined by Eq. (12) is
estimated using an analysis of the experimental results by
Candappa et al.,5 Hurlbut,6 Jamet et al.,8 Lee and Willam,9
Lu and Hsu,10 Newman,11 Smith et al.,12 Smith,15 and Van
Mier22 (Fig. 8 and 11). The parameter is taken as a function of the specimen aspect ratio, concrete strength, and
the confinement ratio, but independent of axial strain. The
proposed expression is written as
b=

c
b

f
a r + 1
fc

a = 65e

0.015 fc

0.5;

; b = 1.5 e

0.02 fc

D
c = 4 h hd ; c = 2
h

( fc in MPa, 1 MPa = 145 psi) (17)


h > hd

Lateral strain model


A model for the lateral strain versus axial strain is
proposed herein based on the assumption that the concrete
behavior is firstly linear elastic. Once the microcracks form,
the behavior becomes nonlinear hardening up to the peak
stress. After the peak stress, volumetric expansion starts
with plastic/inelastic lateral strain varying linearly with
ACI Structural Journal/March-April 2014

Fig. 11Variation of parameter for uniaxial compression


with varying aspect ratio. (Note: 1 MPa = 145 psi.)
the plastic/inelastic axial strain. A preliminary and primitive version of the model presented herein was presented in
Samani and Attard.23 The model described herein is a refined
version based on comprehensive research on reinforced
concrete columns, and also includes size effect issues. The
equations for the lateral strains in the three phases are

f
f
e = pr e
E
Ec

f pr e f

e =
+ A1 ( e )
Ec
Ec

e = f + b e f e f0 + e + f0
e
0
0 0
e

Ec
Ec
Ec
Ec

e e pr
e pr < e e 0

(18)

e > e0

in which is the axial strain and is the lateral strain; e


is the elastic Poissons ratio that normally varies between
0.15 and 0.25; fpr and pr are the axial stress and strain at
the proportional limit in concrete (this is usually taken at
45% of the peak stress f0); and Ec is the elastic modulus.
0 is the axial strain at the peak stress; 0 is the ratio of the
lateral strain to the axial strain at the peak stress; is defined
by Eq.(17); and A1() is a function describing the dilation
447

during the nonlinear hardening phase up to the peak stress.


The Appendix contains the detailed derivation for A1(),
which is determined to ensure continuity between the three
phases in Eq. (18). The lateral strain axial strain model is
then given by
f pr e f

e =
Ec
Ec

e e pr
e e pr

exp a

e
e

f pr e . f
f pr e f0 o
e o e pr
pr

e =
+ 0 e 0

Ec
Ec
exp(a )
Ec
Ec

e = e f + b e f e f0 + e + e f0

0
0 0

Ec
Ec
Ec
Ec

e pr < e e 0
e > e0

The parameter is given by


h
b h
r
a=
e e pr 2 (20)
e po o

where hr is the reference cylinder height. To adjust the stress


strain model to incorporate size effects, Samani and Attard20
divided the total axial strain into its inelastic and elastic
components, such that

f w pc + e d hd f
e = e0 0 +
+ E
Ec
h

h > hd

f w pc
+ ed +
e = e0 0 +
Ec h

Ec

h hd

(21)

Herein, wpc is the localized inelastic axial displacement


due to shear band fracture; hd is the damage zone height;
and d is the additional inelastic axial strain in the damaged
zone associated with longitudinal tensile cracking given in
Samani and Attard20 by
ed =

f0 f
r (1 + k ) f0 f0 fresidual
2 kG ft

0.8

fresidual f f0 (22)

where Gft is the tensile fracture energy; r is a parameter with


the dimension of length proportional to the average distance
between successive longitudinal cracks; k is a material
constant; and fresidual is the residual axial stress level. The
value of r was estimated by Markeset and Hillerborg24 to be
approximately 1.25 mm (0.05 in.) for a maximum aggregate
size of 16 mm (0.6 in.), with r increasing with increasing
maximum aggregate size. The value of k suggested by
Markeset and Hillerborg24 was approximately 3 for normal
density concrete and 1 for lightweight aggregate concrete.
An expression for the tensile fracture energy as a function
of the uniaxial compressive proposed by Van Mier25 can
beused
448

Samani and Attard20 back-calculated an expression for the


localized inelastic displacement using the results of Attard
and Setunge26 for a reference cylinder height of hr = 200 mm
(8 in.), that is

e e pr



(19)

Gft = 0.00097fc + 0.0418 N/mm (N/mm = 5.71 lpf/in.) (23)

w pc = ( e e 0 ) hr +

( f0 f ) h
Ec

e d hr (24)

Substituting Eq. (21) into (19) for the case of > 0, one
obtains
e = e .

w pc + e d hd
f0
f
+ b.
+ 0 e 0 + e E
Ec
h

w pc
f
f
e = e . + b.
+ ed + 0 e0 + e 0
Ec
Ec

h > hd
(25)
h hd

Using Eq. (24), Eq. (25) can be rewritten to give the lateral
strain for the post-peak region as:
e = e

h ( f f ) hr
f
h h
+ b (e e0 ) r + 0
ed r d
h h
Ec
h
Ec
h

+0 e0 + e

f0
Ec

h > hd and e > e 0

f0
Ec

h hd and e > e 0

(26)

hr ( f0 f ) hr
f
hr
e d 1
e = e
+ b (e e0 ) +
h

Ec
h
Ec
h

+0 e0 + e




Figures 12 to 15 show a comparison of the proposed model
with test results presented in Candappa et al.,5 Hurlbut,6 Imran
and Pantazopoulou,7 Jamet et al.,8 and Smith.15 Generally,
the proposed model makes very good predictions giving the
correct trends, and, in most cases, reasonable estimates of
the lateral strain. These comparisons demonstrate the capability of the proposed model in modelling a wide range of
compressive strengths and confining pressures. Figures 16
and 17 show comparisons of the new analytical model with
uniaxial compression tests involving specimens of different
dimensional aspect ratios. The tests of Lee and Willam,9
who tested cylindrical specimens in uniaxial condition with
different heights of 137.2, 91.44, and 45.72 mm (5.4, 3.6,
and 1.8 in.) and a diameter of 76.2 mm (3 in.), have already
been mentioned. Figure 16 shows the comparison with Lee
and Willam9 results. Although the match is only fair, the new
model has the correct trend that shows larger lateral strains
for the specimens with the smallest height. Van Mier,27 a
pioneer in the work of strain softening and size effect issues,
also presented experimental lateral strain versus axial strain
results for prisms with a square cross section of 100 x
100mm (4 x 4 in.) and heights of 50, 100, and 200 mm (2, 4,
ACI Structural Journal/March-April 2014

Fig. 12Comparison of lateral strain versus axial strain for


Candappa et al.5 test results (fc = 100 MPa) with proposed
model. (Note: 1 MPa = 145 psi; 1 mm = 0.0394 in.)

Fig. 15Comparison of lateral strain versus axial strain for


Smith15 test results (fc = 34.5 MPa) with proposed model.
(Note: 1 MPa = 145 psi; 1 mm = 0.0394 in.)

Fig. 13Comparison of lateral strain-versus-axial strain


test results of Willam et al.14 with proposed model. (Note:
1MPa = 145 psi; 1 mm = 0.0394 in.)

Fig. 14Comparison of lateral strain versus axial strain


for Jamet et al.8 test results with proposed model. (Note:
1MPa= 145 psi; 1 mm = 0.0394 in.)
and 8 in.) under uniaxial compression. Comparisons with the
results of Van Mier27 are shown in Fig. 17. Van Mier27 tested
several specimens, and presented a range of results for each
specimen height. There is a wide scatter in the test results
indicative of the dominance of tensile splitting with speci-

ACI Structural Journal/March-April 2014

Fig. 16Comparison of lateral strain versus axial strain for


Lee and Willam9 test results with proposed model. (Note:
1MPa = 145 psi; 1 mm = 0.0394 in.)
mens having aspect ratios less than 2. The proposed model
gives quantitatively good results.
CONCLUSIONS
Two analytical models for the lateral strain versus axial
strain proposed by Binici13 and Lokuge et al.4 were reviewed
and compared with test results. The correlation between
449

Fig. 17Comparison of lateral strain versus axial strain for


Van Mier24 test results with proposed model. (Note: 1 MPa =
145 psi; 1 mm = 0.0394 in.)
the test results and the analytical model predictions was
reasonable, but not for high confinement levels. The initial
compaction under high confinement was not predicted, and
the models did not incorporate size effects associated with
varying specimen aspect ratios.
A new model for the lateral strain-versus-axial strain relationship has been proposed based on the assumption that
the concrete behavior could be classified into three phases.
It was assumed that the concrete responds to loads linearly
elastically up to a proportional limit, and then its response
changes to a nonlinear hardening up to the peak stress. From
previous studies, the point at which the volumetric strain
becomes zero was taken to be the same as the peak stress
point; therefore, the lateral strain at the peak axial stress was
half that of the axial strain at the peak axial stress. In the
post-peak phase, a linear relationship between the plastic
lateral strain and the plastic axial strain was observed. The
total lateral strain was calculated by adding the elastic and
estimated plastic components of the lateral axial. The parameters affecting the relationship between the plastic lateral
and axial strain was shown to be the confinement level and
the specimen aspect ratio.
To show the models accuracy, the proposed model predictions were compared with a vast range of results in which
the concrete strengths varied from low to high strength, and
differing levels of confinement were applied. The model
showed a realistic match with similar trends to experimental results. The model also compared reasonably well
for uniaxial specimens of different aspect ratios, although
the experimental results showed a large scatter because of
the effects of tensile splitting.
AUTHOR BIOS

Ali Khajeh Samani is currently a Post-Graduate Fellow at the School of


Civil and Environmental Engineering, the University of New South Wales,
Sydney, Australia. His research interests include modeling of confined
concrete.
Mario M. Attard is an Associate Professor at the School of Civil and Environmental Engineering, the University of New South Wales. His research
interests include buckling analysis, hyperelastic constitutive modeling,
finite strain continuum mechanics, and softening and constitutive modeling
of concrete.

450

A1()
a
b
c
D
Ec
f
f0
fc
fpr
fr
fresidual
Gft
h
hd
hr
k
r

NOTATION

= function in proposed model


= parameter in proposed model
= parameter in proposed model
= parameter in proposed model
= diameter of specimen
= elastic modulus of concrete
= stress level at a strain of
= peak stress
= uniaxial compressive strength of concrete, in MPa
= axial stress and strain at proportional limit in concrete
= lateral confinement on specimen, in MPa
= residual axial stress level
= tensile fracture energy
= height of specimen
= damage zone height
= reference cylinder height
= material constant
= parameter with dimension of length proportional to average
distance between successive longitudinal cracks
wpc
= localized inelastic axial displacement due to shear band fracture

= parameter in proposed model

= parameter defining the relationship of plastic lateral and axial
strain

= scaling parameter in Binici13

= axial strain
= strain at peak stress for uniaxial compression tests
c
e
= elastic limit strain
e
= additional inelastic axial strain in the damaged zone
p
= plastic axial strain
pr
= axial strain at proportional limit in concrete
p(at peak) = axial plastic strain at peak stress level
0
= strain corresponding to peak stress
0
= lateral strain at peak axial stress equal
1
= transition point in Lokuge et al.4

= lateral strain
p
= plastic lateral strain
p(at peak) = lateral plastic strain at peak stress level
e np
= plastic normal strain rate
e sp
= plastic shear strain rate
f
= initial failure angle
0
= ratio of lateral strain to axial strain at peak stress
e
= elastic Poissons ratio
e
= Poissons ratio of concrete varying between 0.15 to 0.2
l
= largest secant ratio
p
= ratio of lateral strain to axial strain at the peak stress
s
= secant Poissons ratio

= shear failure plane

= plastic strain dilatancy angle

REFERENCES

1. Cusson, D., and Paultre, P., Stress-Strain Model for Confined HighStrength Concrete, Journal of Structural Engineering, ASCE, V. 121,
1995, p.468.
2. Ahmad, S., and Shah, S., Stress-Strain Curves of Concrete Confined
by Spiral Reinforcement, ACI Journal, V. 79, No. 6, Nov.-Dec. 1982,
pp. 484-490.
3. Talaat, M., and Mosalam, K., Computational Modeling of Progressive Collapse in Reinforced Concrete Frame Structures, PEER Report
2007/10, Pacific Earthquake Engineering Research Center, University of
California, Berkeley, CA, 2008, 308 pp.
4. Lokuge, W.; Sanjayan, J.; and Setunge, S., Stress-Strain Model for
Laterally Confined Concrete, Journal of Materials in Civil Engineering,
V. 17, No. 6, 2005, pp. 607-616.
5. Candappa, D.; Sanjayan, J.; and Setunge, S., Complete Triaxial
Stress-Strain Curves of High-Strength Concrete, Journal of Materials in
Civil Engineering, V. 13, No. 3, 2001, pp. 209-215.
6. Hurlbut, B., Experimental and Computational Investigation of
Strain-Softening in Concrete, University of Colorado, Boulder, CO, 1985,
256 pp.
7. Imran, I., and Pantazopoulou, S., Experimental Study of Plain
Concrete under Triaxial Stress, ACI Materials Journal, V. 93, No. 6,
Nov.-Dec. 1996, pp. 589-601.
8. Jamet, P.; Millard, A.; and Nahas, G., Triaxial Behaviour of
Micro-Concrete Complete Stress-Strain Curves for Confining Pressures
Ranging from 0 to 100 MPa, RILEM-CEB International Conference:
Concrete under Multiaxial Conditions, V. 1, May 1984, pp. 133-140.

ACI Structural Journal/March-April 2014

9. Lee, Y., and Willam, K., Mechanical Properties of Concrete in


Uniaxial Compression, ACI Materials Journal, V. 94, No. 6, Nov.-Dec.
1997, pp. 457-471.
10. Lu, X., and Hsu, C., Stress-Strain Relations of High-Strength
Concrete under Triaxial Compression, Journal of Materials in Civil Engineering, V. 19, No. 3, 2007, pp. 261-268.
11. Newman, J., Concrete under Complex Stress, Developments
in Concrete TechnologyI, F. Lydon, ed., Applied Science Publishers,
London, 1979, pp. 151-219.
12. Smith, S.; Willam, K.; Gerstle, K.; and Sture, S., Concrete Over the
TopOr, is there Life After Peak? ACI Materials Journal, V. 86, No. 5,
Sept.-Oct. 1989, pp. 491-497.
13. Binici, B., An Analytical Model for Stress-Strain Behavior
of Confined Concrete, Engineering Structures, V. 27, No. 7, 2005,
pp. 1040-1051.
14. Willam, K.; Hurlbut, B.; and Sture, S., Experimental and Constitutive Aspects of Concrete Failure, Finite Element Analysis of Reinforced
Concrete Structures, (Proceedings of seminar sponsored by Japan Society
for the Promotion of Science and the U.S. National Science Foundation),
1986, pp. 226-254.
15. Smith, S. H., On Fundamental Aspects of Concrete Behavior, MS
thesis, CEAE Department, University of Colorado Boulder, Boulder, CO,
1987.
16. Kotsovos, M., and Newman, J., A Mathematical Description of the
Deformational Behavior of Concrete under Complex Loading, Magazine
of Concrete Research, V. 31, No. 107, 1979, pp. 77-90.
17. Nielsen, C., Triaxial Behavior of High-Strength Concrete and
Mortar, ACI Materials Journal, V. 95, No. 2, Mar.-Apr. 1998, pp. 144-151.
18. Imran, I., Applications of Non-Associated Plasticity in Modeling
the Mechanical Response of Concrete, PhD thesis, University of Toronto,
Toronto, ON, 1994.
19. Vonk, R., Softening of Concrete Loaded in Compression, thesis,
1992.
20. Samani, A., and Attard, M., A Stress-Strain Model for Uniaxial and
Confined Concrete under Compression, Engineering Structures, V. 41,
2012, pp. 335-349.
21. Vermeer, P. A., and De Borst, R., Non-Associated Plasticity for
Soils, Concrete and Rock, Technical University of Delft, Delft, Netherlands, 1984.
22. Van Mier, J., Strain-Softening of Concrete under Multiaxial Loading
Conditions, Dissertation, 1984, 349 pp.
23. Samani, A., and Attard, M., Lateral Behaviour of Concrete, International Conference on Earthquake and Structural Engineering (ICESE
2011), Venice, Italy, 2011, pp. 940-945.
24. Markeset, G., and Hillerborg, A., Softening of Concrete in CompressionLocalization and Size Effects, Cement and Concrete Research,
V. 25, No. 4, 1995, pp. 702-708.
25. Van Mier, J., Fracture Processes of Concrete, CRC Press, Boca
Raton, FL, 1996, 464 pp.
26. Attard, M., and Setunge, S., Stress-Strain Relationship of Confined
and Unconfined Concrete, ACI Materials Journal, V. 93, No. 5, Sept.-Oct.
1996, pp. 432-442.
27. Van Mier, J., Multiaxial Strain-Softening of Concrete, Materials
and Structures, V. 19, No. 3, 1986, pp. 179-190.

APPENDIX A
The proposed lateral strain model continuity and boundary
conditions are detailed here. To define A1() used in Eq. (18),
the following continuity conditions need to be satisfied. The
lateral strains at the transition points between the elastic phase
and the nonlinear hardening phase, and the nonlinear hardening phase and at the peak stress, need to be equal, hence:

f pr e f0

e =

+ A1 ( e 0 )
Ec
Ec

e = e0
A1 ( e 0 ) = e p0

e f0
f0
f0
pr

+
+
=

e
b
e

p
0
0
0

Ec
Ec
Ec
Ec

(A2)
The slopes at these transition points also need to be equal,
therefore
de

= e

e
d
Ec

e = e pr
de = e
de
Ec

dA1 ( e )

=0

d e e = e pr
dA1 ( e )

+
d e e = e pr

df
= e
d e e = e pr
df
d e e = e pr


(A3)
0


dA1 ( e )
de

e df
=
+

e
e
e
d
E
d
d
e = e0
e = e0
c

dA ( e )

1
=

d e e = e0

dA1 ( e )
h
0
0







e = e0
=b

d e e = e0
hr
e df
df

+ b 1

Ec d e e = e 0
Ec d e e = e 0

h
e
d

=b

=
de
hr
hr hr df
+ 1

e
h
h
E
d

e = e0







(A4)
A form for the function A1(), which allows the continuity
conditions to be satisfied, is
2

e e pr
e e pr
e e exp a e e
o
o
pr
pr
(A5)
A1 ( e ) = e p = e p0
exp(a )
with defined as
h
b h
r
a=
e e pr 2 (A6)
e p0 o

e f pr
f
e = r

Ec

Ec
e = e pr
A1 (e pr ) = 0

f
e
pr

r
e = E E + A1 (e pr )
c



(A1)

ACI Structural Journal/March-April 2014

451

NOTES:

452

ACI Structural Journal/March-April 2014

DISCUSSION
Disc. 110-S31/From the May-June 2013 ACI Structural Journal, p. 404.

Cyclic Loading Test for Beam-Column Connection with Prefabricated Reinforcing Bar Details. Paper by
Tae-Sung Eom, Jin-Aha Song, Hong-Gun Park, Hyoung-Seop Kim, and Chang-Nam Lee
Discussion by Yun Liu and Dun Wang
Lecturer, College of Architecture and Civil Engineering, Zinjiang University, Urumqi, China; ACI member, PhD candidate, Research Institute of Structural Engineering and Disaster
Reduction, School of Civil Engineering, Tongji University, Shanghai, China.

The subject of precast concrete systems has been a


promising research field since the PRESSS (Precast Seismic
Structural System Program) was initiated by the U.S. and
Japan in the 1990s.16 Numerous research activities have
been conducted and many novel structural systems, such as
hybrid frame systems, rocking frame systems, and rocking
wall systems, have been developed and applied to practical
projects. The major features of precast concrete systems are
the fast speed of construction, the high quality of precast
and prestressed concrete units, durability improvements,
and superior performance during earthquakes.17 To
improve the speed of construction and ensure the seismic
performance of precast systems, various techniques and
construction methods have been developed, such as the
welded reinforcement grid (WRG) and the SEN Steel
Concrete Construction method, as mentioned in the paper.
The discussed paper presents a prefabricated reinforcing
bar connection method for the earthquake design of beamcolumn connections using the techniques of reinforcing
bar welding, coupler splicing, and headed-bar anchorage.
Test results of four test specimens under cyclic loading are
given to elaborate on the effectiveness of this construction
method, which could be an alternative for the application
of precast concrete frame systems. Some topics in the paper
are interesting and the discussers would like to comment on
them as follows:
Welding stress, cracks, pores, and slags in the weld zone
are complex phenomena for steel bars with welding
during loading, which have an adverse effect on the
behavior of reinforcement. Therefore, as mentioned in
the paper, according to Saatcioglu and Grira1, to secure
the ductile behavior of a column under lateral loading,
the grid bar with a welded joint should have at least 4%
elongation capacity in tension. However, as observed
in Table 2 of the paper, no such information is present.
Moreover, mechanical properties of coupler splices and
headed bars should be also included in the paper.
It is indicated in the paper that axial load was not applied
to columns and that the performance of such reinforcing
bar details is not affected by the axial load of columns.
The discussers do not agree with the authors on this point.
As is well known, the behavior of the beam-column
connection is more complicated with the existence of
axial load than with no axial load, and the steel band
plates would expand around the column, which may
have an unexpectedly adverse effect on the reinforcing
bar couplers. In addition, with the slenderness ratio of
the column not exceeding 3 (2100/700 equates to 3), it
can be defined as a short column; the existence of axial
load can cause compression failure to the column and
make the behavior of the connection more complex.
Moreover, as mentioned in the recently pubished
ACI Structural Journal/March-April 2014

374.2R-13,18 gravity loads should be simulated during


testing, whether their effects are deemed important
or not, because there is an important aspect of the
application of gravity loads on a column, resulting in a
fast rate of lateral force strength degradation. Therefore,
it is best to include the axial loads effect into the test of
columns to get a more realistic behavior.
In the Test Program section of the paper, it can be
seen in Table 1 that there are more bottom reinforcing
bars than top reinforcing bars in the beam, which may
not be the case in practice. What is the consideration
of such an arrangement by the authors? A second
question regards the setup of the test. It is indicated
that the ends of the beam cannot move upward and
downward; instead, it allows the beam end to move
horizontally. Does such a test setup conform to the
true situation? What is the consideration for such a
test setup? Please clarify. Moreover, it is mentioned in
the paper that D25 bars used for the beam bottom bars
had a relatively small fracture strain5.36%which
was less than the minimum requirement specified in
the Korean Industrial Standard. Strictly speaking, it is
forbidden to use unqualified material in tests and in the
practical project. However, such steel bars are still in
use for the test specimens, which may be caused by the
sequence of the test specimens constructionthat is,
the test specimens were constructed before the material
properties test finished. Although there are requirements
for steel bars before entering into the laboratory, it is a
good choice to do material properties tests early, before
the test specimens are constructed.
As seen in Fig. 2 of the paper, diagonal bars are used
along the entire length of the beam and in the transverse
hoops in the beam, which is more than the conventional
cast-in-place specimen. Undoubtedly, such an
arrangement of shear reinforcement would change the
behavior of the columnthat is, the diagonal bars along
the length of the beam strengthen the fixing of bottom
reinforcing bars, resulting in minimizing the slip at the
joint, as described in Fig. 2. Maybe it would be better to
use a specimen with only diagonal bars in the beam to
make comparisons between behaviors of test specimens.
In the paper, techniques of reinforcing bar welding,
coupler splicing, and headed-bar anchorage are used
for the beam-column connection; however, not much
more information is presented on the headed-bar
anchorage and reinforcing bar welding. In addition,
when constructing the specimens, how are the couplers
constructed to connect the reinforcing bars from the
beam and from the connection? It can be seen in Fig. 8(b)
that reinforcing-bar slip occurred due to the loosened
thread of the couplers, which indicated that there is a
453

possibility of the couplers not working in practice.


Perhaps an alternative is to place the couplers outside
the beam-column connection to connect the reinforcing
bars of the beam, not in the connection region.
To sum up, the construction method of a prefabricated
reinforcing bar for beam-column connection by
reinforcing-bar welding, coupler splicing, and headed-bar
anchorage helps to accelerate the construction speed and is
a good choice for a precast concrete frame system. Because
the behavior of the beam-column connection is vital to the
entire structural system, much more experimental research
on such connections and structural frame systems with such
connections should be conducted to obtain enough reliable
proof for the proposed construction method, especially
focusing on welding reinforcement, which has been pointed
out by the authors. The discussers are waiting with great
interest for the discussions in the announced follow-up paper.

Fig. 15Direct tension tests on three D25 bars with tag


welding joint.

REFERENCES

16. Priestley, M. J. N., Overview of PRESSS Research Program, PCI


Journal, July-Aug. 1991, pp. 50-57.
17. fib Bulletin 27, Seismic Design of Precast Concrete Building
Structures, fib, Lausanne, Switzerland, 2003, pp. 1-2.
18. ACI Committee 374, Guide for Testing Reinforced Concrete
Structural Elements under Slowly Applied Simulated Seismic Loads (ACI
374.2R-13), American Concrete Institute, Farmington Hills, MI, 2013,
18 pp.

AUTHORS CLOSURE
The authors would like to thank the discussers for their
interest in the paper and the informative discussion. The
authors response to the five comments is presented as
follows.
1. In the proposed prefabricated reinforcing bar construction method (PRC method), tag welding is used for the
connection between the transverse D13 bars and longitudinal D25 and D22 bars. However, as shown in Fig. 5, the
amount of tag welding is relatively small, when compared
to the area of the longitudinal bars. Thus, the adverse effect
of tag welding on the longitudinal bars was expected to be
minimal. As presented in the conclusion No. 5 of the paper,
in this test, the bar tag welding did not have detrimental
effects on the structural performance of the specimens.
The authors performed direct tension tests on the D25 bars
that had a tag welding joint to a transverse D13 bar at the
center (refer to Fig. 15). The results showed that two
D25 bars were fractured away from the weld joints, but the
third specimen failed near the weld joint. Note that the elongation capacities at rupture were much greater than 4%.
A material test for the steel used in the bar couplers
was not performed. The yield and tensile strengths provided
by the manufacturer were fy = 751 MPa (109 ksi) and fu =
760 MPa (110 ksi), which were much greater than those of
the D22 and D25 bars. The couplers used in this test are
commercial products, and the mechanical properties were
proved elsewhere. However, in this test, reinforcing bar slip
occurred due to the loosened threads of the coupler in Specimens PRC1. Thus, attention should be paid to the coupler
splice.
The headed bars used in the test were different from the
conventional one specified in the design code. As shown
in Fig. 4, the beam flexural bars in the exterior joint were
anchored to the steel band plate by nuts and washers. Because
the steel band plate provides additional bearing capacity for
bar anchorage, the headed bars were expected to be safe. In
this test, failure did not occur in the headed bars.
454

Fig. 16Simplified models for beam-column connections


subjected to seismic loading.
2. As the discusser commented, when columns are
subjected to high axial load, the axial load effect on the
connection behavior may be undesirable and even critical.
Thus, the overall behaviors of the beam-column connection specimens would be more realistic if axial load was
applied to the columns. However, this test was performed
to investigate the effect of the reinforcing bar details (such
as the couplers, headed bars, band plates, and bar welding)
rather than the behavior of beam-column connections itself.
In the test specimens, the reinforcing bar details, except
the proposed details, are the same as the conventional
ones. Thus, if the proposed details do not affect the overall
behavior of the specimens, the axial load effect on the overall
behavior of the specimens should be the same as the effect
on beam-column connections with conventional reinforcing
bar details.
The steel band plates are used to connect the beam reinforcing bars to the column reinforcing bars for reinforcing
bar fabrication and erection, and to provide additional
bearing resistance for bar anchorage. Furthermore, as shown
in Fig. 3, conventional ties or hoops specified in current
design codes are used in the beam-column joints. Thus,
the use of the steel band plates in addition to the conventional ties is not likely to affect the overall behavior of the
beam-column connections. Thus, the axial load effect on the
proposed connection method is expected to be the same as
the effect on conventional beam-column connections.
3. There is a misunderstanding. As shown in Table 1 and
Fig. 3, the number of the top flexural bars was greater than
the number of the bottom bars.
The test setup was planned to simulate the lateral movements of the interior and exterior beam-column connections
under seismic loading, as presented in Fig. 16. The test
ACI Structural Journal/March-April 2014

setup has been used in many previous tests of beam-column


connections.
The material tests on the reinforcing bars used in the test
program were conducted after the cyclic load tests. Because
the D25 bars used in this test are commercial products
and the mechanical properties were originally guaranteed
by the manufacturer, material tests were not performed
before testing. However, in the material tests, the elongation at failure was proven to be unqualified. As the discussers
commented, a reinforcing bar tension test before the connection test is desirable and appropriate.
4. The diagonal D13 bars in the beam are nonstructural
elements that are used only for reinforcing bar fabrication.
Therefore, conventional vertical stirrups are required for the
shear capacity of the beam. Further, as shown in Fig. 3, the
number of the diagonal bars is significantly less than that of
the vertical stirrups. Thus, the effect of the diagonal bars on
the overall behavior of the specimens was expected to be
minimal. As shown in Fig. 7, no notable difference in the

cyclic behaviors of RC1 and PRC1 was observed (compare


the pinching in the cyclic curves presented in Fig. 7).
5. Please refer to the authors reply No. 1 regarding the bar
welding, coupler splice, and headed bar anchorage.
In the proposed method, the reinforcing bar cages of
columns and beams are fabricated in a fabrication shop
separately. In the construction field, the column bar cage
is erected first. Then beam bar cage is connected to the
column bar cage using the bar couplers, which are located
at the column face. The location of the couplers was determined considering the efficiency and economy in fabrication
and shipping. In a mockup test, the reinforcing bar fabrication of the PRC connection specimens was conducted
without difficulty. In real construction fields, however, the
PRC method may be a challenging construction method.
Although the present study focused on verifying the structural performance of the proposed beam-column connection
method, further improvement in the construction technique
is required for the practical use of the proposed method.

Disc. 110-S32/From the May-June 2013 ACI Structural Journal, p. 415.

Shear Strength of Reinforced Concrete Walls for Seismic Design of Low-Rise Housing. Paper by Julian
Carrillo and Sergio M. Alcocer
Discussion by Dun Wang and Xilin Lu
ACI member, PhD candidate, Research Institute of Structural Engineering and Disaster Reduction, School of Civil Engineering, Tongji University, Shanghi China; Professor,
Research Institute of Structural Engineering and Disaster Reduction, School of Civil Engineering, Tongji University.

It is known that various models have been proposed to


estimate the shear strength of shear walls, such as those based
on the softened truss model initially developed by Mau and
Hsu (1987) and later modified by Guta (1996), the softened
strut-and-tie model proposed by Hwang et al. (2001), and the
UCSD shear model by Kowasky and Priestley (2000), which
was later modified by Krolicki (2011). The discussed paper
presents a set of semi-empirical design equations that are
capable of estimating the shear strength of walls for low-rise
housing, based on an experimental program of 39 isolated
walls in cantilever, and discussion of previous experimental
studies by other researchers. As seen in the conclusions of
this paper, the proposed equations qualify the efficiency
factor of horizontal reinforcements by their contributions
to the shear strength of walls in one- to two-story concrete
buildings in Latin America. The discussers would like to
raise the following significant issues and suggestions:
As seen in Eq. (2), the concrete contribution to the shear
strength of walls is associated to a1 and fc, and a1 is
related to the shear span ratio from Eq. (7), all of which
are constant because the wall specimen is cast. However,
with the increase of displacement, the cracks appear and
widen in concrete, resulting in reducing the effectiveness
of the aggregate interlock shear resistance along the
crack surface. Moreover, vertical reinforcements in the
web wall may also contribute to the shear resistance,
but none of these are included in the proposed model
by the authors. The UCSD shear model seems to give
a relatively reasonable mechanism for the concrete
contribution, which takes into account the effective shear
area, the degradation of the shear resistance of concrete,
the volumetric ratio of longitudinal reinforcement, and
the effect of the shear-span ratio.
ACI Structural Journal/March-April 2014

As noted by the authors, several existing proposed


models for estimating the shear strength of concrete
walls have limitations, which are described in the
paper. Special attention should be paid to the fact that
most of these proposed shear models are based on
their individual experimental research and have their
own limitations to particular scopes, with different
influencing factors, loading protocols, and dataprocessing methods. Therefore, there is a need for a
unified experimental research activity on shear strength
of walls to be launched worldwide due to the complex
mechanism of shear phenomena. In addition, research
on the shear strength of walls should also focus on a
model based on theoretic analysis and then be verified
by experimental results, such as the softened truss model
theory proposed by Mau and Hsu (1987) and modified
by Guta (1996), or the softened strut-and-tie model by
Hwang et al. (2001); it should not be based merely on
experimental results.
The paper includes wall specimens with web
reinforcement made of welded-wire mesh. As
mentioned by the authors, displacement ductility of
such a wall type may be limited by the small elongation
capacity of cold-drawn wire reinforcement; it can be
seen from Table 2 that the elongation is only 1.4% to
1.9%. The discussers wonder whether it is allowed by
design codes to use such cold-drawn wire reinforcement
with low elongation for web reinforcement of walls,
especially for seismic design. In addition, Saatcioglu
and Grira (1999) recommended that, to secure the
ductile behavior of a column under lateral loading, the
grid bar with a welded joint should have at least a 4%
elongation capacity in tension. Moreover, it can be seen
455

from Table 2 that the ratio of ultimate strength to yield


strength for welded-wire mesh used in the paper is, at
most, 1.16, which does not seem to have enough margin
of strength for seismic design.
It is mentioned in the paper that wall reinforcement
of specimens was placed in a single layer at wall midthickness due to the thickness of 100 mm (4 in.) for
one- and two-story buildings. However, no information
on the diameter of welded-wire mesh is provided in the
paper, and it is not clear whether the concrete cover
is too large to have a durability problem in practical
projects with only one layer of reinforcement. That is to
say, it is prone to cracks or shrinkage with single-layer
welded-wire mesh. In addition, in practice with a thicker
wall section, there would be more than one layer of
reinforcement in the wall where tie bars for connecting
the layers of reinforcement would be essential. So the
contribution of tie bars to the shear strength of the wall
should also be taken into account.
In Eq. (1) and (2), Aw is the area of wall concrete section
used to calculate the shear strength. In Eq. (6) and (8),
lwtw is used. It is not mentioned whether the two indexes,
Aw and lwtw, are equal to each other. What is the value for
Aw if they are not the same? Maybe the calculation of
Aw should be clarified because there are different values
used in different models, such as those by Kassem
(2010) and Krolicki (2011).
As seen in Eq. (2), the web steel contribution to
the shear strength of walls is limited to horizontal
reinforcement and is associated with the yield strength
of reinforcement, fyh, the volumetric ratios rh and hh, as
well as Aw, which has nothing to do with the longitudinal
steel reinforcement. However, as referred to by the
authors in the paper, according to Barda et al. (1977),
for a wall with hw/tw < 1 and with boundary elements,
horizontal reinforcement becomes less effective as
compared to vertical reinforcement, particularly for
walls with M/Vlw < 0.5, which indicates that vertical
reinforcement contributes to the shear strength of the
wall. Both of the cases hw/tw < 1 and M/Vlw < 0.5are
all included in the scope of the proposed shear-strength
equations by the authors; that is, the proposed equations
best predict peak shear strength of walls with M/Vlw
ratios less than or equal to 2.0. In addition, what is the
mechanism of the web horizontal reinforcement for the
contribution to shear resistance? Because it is calculated
as the products of fyh, rh, hh, and Aw, which is not similar
to that mentioned in Krolicki (2011)that is, VS =
rttwhcrfy (as seen in detail in the reference). Moreover,
the factor hh used for considering the yielding extent
of horizontal web reinforcement is constant (0.8 for
deformed bars and 0.7 for welded-wire mesh), which
does not reflect the effects of other influences such as
the imposed lateral drift, volumetric ratio of horizontal
reinforcement, axial loading, and yielding strength.
An assumption that the contribution of vertical axial
stress to the shear strength of squat walls is unimportant
is made in the paper by the authors. Therefore, there
is no contribution of vertical axial loading effect in
Eq. (2). However, as a matter of fact, axial stresses
always exist whether it is increased or decreased by
vertical acceleration or coupling between walls. In
addition, whether it is conservative, and to what extent,
is not clear without the contribution of vertical axial

456

load. Therefore, it is better to make out the condition for


not considering the axial loading effect by quantitative
values. Moreover, as mentioned in the newly published
ACI 374.2R-13 (ACI Committee 374 2013), gravity
loads should be simulated during testing, whether their
effects are deemed important or not, because it would
result in a fast rate of lateral force strength degradation
with the existence of vertical loadings. Therefore, the
contribution of vertical axial loads to the shear strength
of walls should be included in Eq. (2), just like those
proposed by Krolicki (2011).
Lastly, a clear definition for the squat wall needs to be
made, and consensus should be reached on issues such
as the influencing factors, the expression equation, and
the mechanisms of each contribution to shear strength.
After that, comparisons between the previous proposal
models can be made to get a conclusive judgment for
the shear strength of the shear wall. Maybe there is
still a long way to go to determine the mechanisms of
each contribution to the shear strength of walls based
on theoretic analysis, not merely on the experimental
research, whether the shear wall is a squat wall or
flexural wall.
REFERENCES

ACI Committee 374, 2013, Guide for Testing Reinforced Concrete


Structural Elements under Slowly Applied Simulated Seismic Loads (ACI
374.2R-13), American Concrete Institute, Farmington Hills, MI, 18 pp.
Gupta, A., 1996, Behavior of High Strength Concrete Structural Walls,
PhD thesis, Curtin University of Technology, Perth, Australia.
Hwang, S. J.; Fang, W. H., Lee, H. J.; and Yu, H. W., 2001, Analytical
Model for Predicting Shear Strength of Squat Walls, Journal of Structural
Engineering, ASCE, V. 127, No. 1, pp. 43-50.
Kassem, W., and Elsheikh, A., 2010, Estimation of Shear Strength of
Structural Shear Walls, Journal of Structural Engineering, ASCE, V. 136,
No. 10, Oct., pp. 1215-1224.
Kowalsky, M. J., and Priestley, M. J. N., 2000, Improved Analytical
Model for Shear Strength of Circular Reinforced Concrete Columns in
Seismic Regions, ACI Structural Journal, V. 97, No. 3, May-June, pp. 388396.
Krolicki, J.; Maffei, J. G.; and Calvi, M., 2011, Shear Strength of
Reinforced Concrete Walls Subjected to Cyclic Loading, Journal of
Earthquake Engineering, V. 15, No. S1, pp. 30-71.
Mau, S. T., and Hsu, T. C., 1987, Shear Behavior of Reinforced Concrete
Framed Wall Panels with Vertical Loads, ACI Structural Journal, V. 84,
No. 3, May-June, pp. 228-234.
Saatcioglu, M., and Grira, M., 1999, Confinement of Reinforced
Concrete Columns with Welded Reinforcement Grids, ACI Structural
Journal, V. 96, No. 1, Jan.-Feb., pp. 29-39.

AUTHORS CLOSURE
The authors acknowledge the interest of discussers in
the paper. Indeed, shear strength of concrete members is
still an issue where the development of a unified approach
is needed. Studies, both experimental and analytical,
contribute to providing data, evidence, and reflections on the
phenomenon. Our research was aimed at developing simple
design tools that could be used in practice. The following are
comments based on discussers items:
1. The model developed in our research recognizes that
factors hh,v, a1, and a2 do depend on wall ductility or wall
drift. The model discussed in the paper is applicable to drifts
associated with peak strength (pp. 418 and 422). The model
developed also recognizes the contribution of vertical web
reinforcement to wall shear strength (pp. 420-422). In fact,
strains measured in the vertical web bars or steel wires during
tests were mainly associated to the uniform distribution of
inclined cracks on the wall web. It was confirmed that, as
specified by ACI 318,1 a minimum amount of web vertical
ACI Structural Journal/March-April 2014

steel should be computed using Eq. (5). On the other hand,


the accuracy and variables included in the UCSD shear
model are not within the scope of the paper.
2. The authors of the paper agree with the discussers
that there is a need for a unified approach for experimental
research on shear strength of walls to be launched worldwide
due to the lack of a consistent trend of the contribution of
horizontal and vertical web reinforcement to wall peak
shear strength (p. 418). As the discussers indicate, other
methodologies for assessing shear behavior have been
developed, such as truss models and softened strut-and-tie
models. Such models were not discussed in the paper and,
therefore, should be studied in future research.
3. The authors agree with the discussers that steel
reinforcement to be used in seismic design applications must
exhibit minimum elongation capacity. The purpose of using
the welded-wire mesh available was to assess its adequacy as
reinforcement for seismic design applications (p. 415). The
authors have recommended that welded-wire mesh with low
elongation capacity should not be used for shear resistance.
4. Space limitations in the paper hindered the possibility
of including details of the experimental program such as
the diameter of the bars and wires. Discussers are invited
to revisit the recommended references (Carrillo and
Alcocer 2012, 2013). Regarding the use of one curtain of
reinforcement within the wall web, Section 21.9.2.2 of ACI
318-11 also allows its use when the factored shear force is
lower than a predefined value, as is the common case in lowrise concrete housing.
5. The calculation of Aw is clearly specified after Eq. (2)
as the gross area of wall concrete bounded by wall thickness
and wall length (Aw = twlw).
6. As indicated above, the model reported in the paper
considers the contribution of vertical web reinforcement to

wall shear strength. An equation to compute the vertical web


steel ratio is included. The equation to calculate the wall
shear strength explicitly includes the horizontal web steel
ratio only. However, for this equation to be applicable, the
vertical web steel ratio should be placed in the wall and
computed using the proposed Eq. (5). Differences between
the proposed model and Krolicki model are not within
the scope of the paper and, therefore, should be studied
in future research. In the case of the so-called efficiency
factor of horizontal web steel ratio, as argued in Note 1,
the model acknowledges that hh depends on wall drift. The
model proposed in the paper is applicable to calculation of
maximum load-carrying capacity and, thus, the proposed
equation is applicable to drifts associated with peak strength.
In addition, as discussed on p. 421 of the paper, values of hh
should be used when rhfyh is lower than 1.25 MPa (181 psi).
7. The authors agree with the discussers that axial stress
is important to accurately determine wall shear strength
and deformation capacities. As discussed on p. 423 of the
paper, for these box-type low-rise structures, axial stresses
on walls typically have small values. Therefore, assuming
sv 0 is conservative for low-to-medium seismic hazard
areas. A detailed analysis is warranted for structures located
in areas of high seismic hazard.
8. The authors agree with the discussers that more work
needs to be done to better differentiate shear-dominated and
flexural-governed walls, as well as on the development of
a comprehensive and unified theory on shear strength and
deformation capacities. These issues and the comparisons
between the previous proposal models are not within the
scope of the paper and should be the topic of a new study.

Disc. 110-S39/From the May-June 2013 ACI Structural Journal, p. 491.

Performance of AASHTO-Type Bridge Model Prestressed with Carbon Fiber-Reinforced Polymer


Reinforcement. Paper by Nabil Grace, Kenichi Ushijima, Vasant Matsagar, and Chenglin Wu
Discussion by Jos R. Mart-Vargas
Associate Professor, ICITECH, Institute of Concrete Science and Technology, Universitat Politcnica de Valncia, Valencia, Spain.

The discussed paper presents an interesting experimental


study on the design philosophy, construction techniques
employed, and flexural performance of both an AASHTO
I-beam and a bridge model reinforced and prestressed with
carbon fiber composite cable (CFCC) strands.
The authors should be complimented for producing a
detailed paper with comprehensive information. This is
acknowledged by the discusser, who would like to offer the
following comments and questions for their consideration
and response:
1. A scale factor of 1:3.6 was used for both the AASHTOType control I-beam with a span of 12,141 mm (39 ft 10 in.)
and 502 mm (19.75 in.) deep and the 2500 mm (98.75 in.)
width bridge model made up of five such beams. Regarding
the cross section of the beam, it seems that the scale factor
was taken in relation to the AASHTO Type V or Type VI
I-beam when based on the 203 mm (8 in.) width of the bottom
flange; 3.6 203 = 731 mm (28.77 in.) is next to 712 mm
(28 in.), which is the actual bottom flange of AASHTO
Type V and Type VI I-beam. In addition, and according to
Gerges and Gergess,14 if the 502 mm (19.75 in.) depth or
ACI Structural Journal/March-April 2014

the 12,141 mm (39 ft 10 in.) span length are considered,


the beam is related to the AASHTO Type VI I-beam, which
is 1830 mm (72 in.) depth and is often used for span lengths
ranging from 36 to 45 m (118 ft 1.32 in. to 147 ft 7.65 in.)
from the corresponding width and span length obtained as
3.6 502 mm = 1807 mm (71.14 in.) and 3.6 12.141 m
= 43.7 m (143 ft 4.77 in.), respectively. On the other hand,
the 95 mm (3.75 in.) width of the beam web results in 3.6
95 mm = 342 mm (13.5 in.), which seems to be a broad width
to be related to the 205 mm (8.1 in.) width of the AASHTO
Type V and Type VI I-beams. What were the criteria applied
to choose the scale factor of 1:3.6 and the width of the beam
web? It is worth noting that different effects on length, cross
section, and second area moment can result in an identical
scale factor.
2. Beam spacing in a bridge depends on several factors such
as span length, concrete strength, loads, and environment,
among others. What was the criterion used to establish a
beam spacing of 502 mm (19.75 in.) in the bridge model?
Regarding the average 28-day concrete compressive strength
in the six AASHTO I-beams of 44.82 MPa (6500 psi),
457

502 mm beam spacing is approximately 1.13 to 1.45


obtained as (3.6 502)/1600 and (3.6 502)/1250times
the maximum beam spacing for 43.7 m (143 ft 4.5 in.) span
length when taking as a reference the charts provided by
Gerges and Gergess14 for optimizing girder size and spacing.
3. Based on the allowable concrete stresses at prestress
transfer, the maximum number of 15.2 mm (0.6 in.)
diameter 7-wire steel strands that can be accommodated in
an AASHTO Type VI I-beam with a 43.7 m (143 ft 4.5 in.)
span length is 45 to 55.14 The authors used 10 longitudinal
CFCC strands per beam: seven non-prestressed and three
prestressed at 30% of their average breaking load. It seems
the concrete stresses after prestress transfer were too small
to be representative of pretensioned concrete structures
when compared with current practice (prestressing steel
strands prestressed at 75%). This can explain the fewer
prestress losses obtained as compared with conventional
steel prestressed members.15,16
4. Apart from the prestressing force being small according
to the number of CFCC prestressed strands and the used
prestress level, a shorter transfer length is obtained. Transfer
length can be measured17,18 and/or predicted from different
equations based on equilibrium of forces19,20 or strand
slips.21,22 As transfer length provisions differ for distinct
codes and researchers,19,23 and no consensus has been
reached as to the main parameters to be considered,24-26 it
would be interesting to detail the transfer length provisions
used by the authors to design the prestressed beams. Do the
authors have any information on measured transfer length?
5. The concrete stresses at prestress transfer can be
assumed with a linear distribution beyond the dispersion
length,27 which is longer than transfer length and depends
on beam depth and tendon position, among other factors.
According to the scale factor and the beam depth, what
were both the dispersion and transfer lengths in relation to
the beam length? In addition, transfer length remains within
the development length of prestressing strands.12,28,29 It
is worth remarking that the development length equation
in AASHTO1 includes a 1.6 multiplier factor that is used
to avoid bond failure caused by inadequately developed
lengths in structural members whose depth is greater
than 610 mm (24 in.) because of high shear effects. The
tested beams were 502 mm in depth (19.75 in.)lesser
than 610 mm (24 in.)but the loading conditions regarding
strand development length would differ in the full-scale
member, as the resulting depth of 3.6 502 mm = 1807 mm
(71.14 in.) is greater than 610 mm (24 in.).
REFERENCES

14. Gerges, N., and Gergess, A. N., Implication of Increased Live Loads
on the Design of Precast Concrete Bridge Girders, PCI Journal, V. 57,
No. 4, Fall 2012, pp. 78-95.
15. Caro, L. A.; Mart-Vargas, J. R.; and Serna, P., Prestress Losses
Evaluation in Prestressed Concrete Prismatic Specimens, Engineering
Structures, V. 48, 2013, pp. 704-715.
16. Mart-Vargas, J. R.; Serna-Ros, P.; Arbelez, C. A.; and RigueiraVictor, J. W., Bond Behaviour of Self-Compacting Concrete in
Transmission and Anchorage, Materiales de Construccin, V. 56, No. 284,
2006, pp. 27-42.
17. Mart-Vargas, J. R.; Serna-Ros, P.; Fernndez-Prada, M. A.; MiguelSosa, P. F.; and Arbelez, C. A., Test Method for Determination of the
Transmission and Anchorage Lengths in Prestressed Reinforcement,
Magazine of Concrete Research, V. 58, No. 1, 2006, pp. 21-29.
18. Mart-Vargas, J. R.; Caro, L.; and Serna, P., Experimental Technique
for Measuring the Long-Term Transfer Length in Prestressed Concrete,
Strain, V. 49, 2013, pp. 125-134.
19. Mart-Vargas, J. R.; Arbelez, C. A.; Serna-Ros, P.; Navarro-Gregori,
J.; and Pallars-Rubio, L., Analytical Model for Transfer Length Prediction

458

of 13 mm Prestressing Strand, Structural Engineering and Mechanics,


V. 26, No. 2, 2007, pp. 211-229.
20. Mart-Vargas, J. R.; Serna, P.; Navarro-Gregori, J.; and Pallars,
L., Bond of 13 mm Prestressing Steel Strands in Pretensioned Concrete
Members, Engineering Structures, V. 41, 2012, pp. 403-412.
21. Mart-Vargas, J. R.; Arbelez, C. A.; Serna-Ros, P; and CastroBugallo, C., Reliability of Transfer Length Estimation from Strand End
Slip, ACI Structural Journal, V. 104, No. 4, July-Aug. 2007, pp. 487-494.
22. Mart-Vargas, J. R.; Serna, P.; and Hale W. M., Strand Bond
Performance in Prestressed Concrete Accounting for Bond Slip,
Engineering Structures, V. 51, 2013, pp. 236-244.
23. Mart-Vargas, J. R., and Hale, W. M., Predicting Strand Transfer
Length in Pretensioned Concrete: Eurocode versus North American
Practice, Journal of Bridge Engineering, ASCE, 2012, DOI:10.1061/
(ASCE)BE.1943-5592.0000456.
24. Mart-Vargas, J. R.; Serna, P.; Navarro-Gregori, J.; and Bonet, J. L.,
Effects of Concrete Composition on Transmission Length of Prestressing
Strands, Construction and Buiding Materials, V. 27, 2012, pp. 350-356.
25. Caro, L. A.; Mart-Vargas, J. R.; and Serna, P., Time-Dependent
Evolution of Strand Transfer Length in Pretensioned Prestressed Concrete
Members, Mechanics of Time-Dependent Materials, V. 17, No. 4, Nov.
2013, pp. 501-527, DOI: 10.1007/s11043-012-9200-2.
26. Mart-Vargas, J. R.; Ferri, F. J.; and Yepes, V., Prediction of the
Transfer Length of Prestressing Strands with Neural Networks, Computers
and Concrete, V. 2, No. 2, 2013, pp. 187-209.
27. CEN, Eurocode 2: Design of Concrete Structures Part 1-1:
General Rules and Rules for Buildings, European standard EN 1992-11:2004:E, Comit Europen de Normalisation, Brussels, Belgium, 2004.
28. Mart-Vargas, J. R.; Arbelez, C. A.; Serna-Ros, P.; Fernndez-Prada,
M. A.; and Miguel-Sosa, P. F., Transfer and Development Lengths of
Concentrically Prestressed Concrete, PCI Journal, V. 51, No. 5, Sept.-Oct.
2006, pp. 74-85.
29. Mart-Vargas, J. R.; Garca-Taengua, E.; and Serna, P., Influence
of Concrete Composition on Anchorage Bond Behavior of Prestressing
Reinforcement, Construction and Buiding Materials, V. 48, 2013,
pp. 1156-1164.

AUTHORS CLOSURE
The authors would like to thank the discusser for his
interest, insightful discussion, and thoughtful observations
on the presented research paper. The beam cross section of
the bridge model was inspired by the Taylor Bridge built
over Assiniboine River in the Parish of Headingley, Manitoba, Canada. The five-span AASHTO-type bridge with
fiber-reinforced polymer (FRP) was built in 1997 and has
a span length of 31.25 m (102.53 ft) with a depth of 1.8 m
(5.9 ft) and a 200 mm (7.87 in.) thick composite deck.3 The
prestressed beams for the Taylor Bridge are spaced at 1.8 m
(5.9 ft) with a bottom flange width of 660 mm (25.98 in.). The
span-depth ratio of the tested control beam and bridge model
beams were based on the AASHTO1 Table 2.5.2.6.3.1 traditional minimum depths. For a simple span length of 12.141 m
(39.833 ft), the composite depth required for a prestressed
concrete precast I-beam is 0.045 times the spans length
that is, 0.045 12.5 m = 0.563 m (1.847 ft). The modeled
beam cross section had an approximate scale factor of
1/3.6 for an AASHTO Type IV beam. The true scale model
of the AASHTO Type IV beam has narrower web and depth
and it would have posed difficulties in placing concrete,
vibrating, and compacting; therefore, the web width and
depth were adjusted for the beam models based on the
constructibility in the laboratory setting.30 A similar adjustment was made in bottom flange width to accommodate
minimum spacing of the carbon fiber-reinforced polymer
(CFRP) reinforcement and clear cover requirements. All
such modifications done in the AASHTO Type IV beam for
suiting laboratory setup are clear from the dimensions given
in Fig. 1 to 3. Because of these modifications, the presented
beam cross section is not exactly a true scaled version of
the AASHTO Type IV beam in terms of geometric properties; nevertheless, the cross section does closely represent it.
ACI Structural Journal/March-April 2014

Also, it is worth mentioning that the State Departments of


Transportation (DOTs) do modify, to certain extent, the standard AASHTO type prestressing beam shapes depending
on the local needs (Michigan 1800 girder, Wisconsin
type 70, PennDOT 28 beams, and so on).31 Similarly,
the scaled spacing of the bridge model was based on the
AASHTO1 Table 4.6.2.2.2.b.1 limitations to validate load
distribution factors.
In the traditional design approach of the prestressed
concrete bridges, the emphasis is given more on cost optimization, and hence the designers try to use the full capacity
of the prestressed concrete beams. The mentioned reinforcement ratio of 0.46% is based on the strain compatibility
design approach for bonded prestressed and nonprestressed
CFRP tendons arranged vertically.6 It may be noted that,
unlike the balanced ratio for steel, the balanced ratio for
FRP bars/tendons does not signify yielding; rather, it signifies failure/rupture of the bars. Also, substantial research
has been accomplished on the use of steel wire strands for
the prestressing operations; therefore, the process has been
industrialized with a higher level of confidence and duly
specified guidelines. However, no such guidelines have
yet been evolved on the use of CFRP reinforcements as
prestressing strands while conducting this research. Therefore, suitably, the prestressing force level was chosen as 30%
of the average measured breaking load of strands.
Grace32 and Grace et al.33 investigated transfer length of the
CFRP reinforcement used in prestressed concrete box beams

ACI Structural Journal/March-April 2014

and developed new equations to predict transfer length.


However, for the investigation reported in the manuscript, the
authors emphasized the flexural behavior of the AASHTO
Type IV beams prestressed with the CFRPtendons.
The significance of the presented experimental study
has been to explain the design philosophy as per the ACI
440.4R-047 design guidelines and the Unified Design
Approach,6 the construction techniques employed, and
the flexural performance of an AASHTO I-beam and
bridge model reinforced and prestressed with the CFRP
strands. Moreover, the test results reported herein also
aid in validation of the results of analytical and numerical
studies subsequently undertaken in the future, with lucidly
provided geometric and material details along with various
responsequantities.
REFERENCES

30. Wu, C., Performance of Concrete I-Beam Bridge Prestressed with


CFCC Reinforcement, masters thesis, Lawrence Technological University
(LTU), Southfield, MI, 2009.
31. Pennsylvania Department of Transportation (PennDOT), Bridge
Design Drawings, http://www.dot.state.pa.us/Internet/BQADStandards.
nsf/home.
32. Grace, N. F., Transfer Length of CFRP/CFCC Strands for Double-T
Girders. PCI Journal, V. 45, No. 5, 2000, pp. 110-126.
33. Grace, N. F.; Enomoto, T.; Abdel-Mohti, A.; Tokal, Y.; and
Puravankara, S., Flexural Behavior of Precast Concrete Box Beams PostTensioned with Unbonded, Carbon-Fiber-Composite Cables, PCI Journal,
V. 53, No. 4, 2008, pp. 62-82.

459

IN ACI MATERIALS JOURNAL


The American Concrete Institute also publishes the ACI Materials
Journal. This section presents brief synopses of papers appearing
in the current issue.

PDF versions of these papers are available for download at


the ACI website, www.concrete.org, for a nominal fee.

From the March-April 2014 issue

111-M11Practical Approach for Assessing Lightweight


Aggregate Potential for Concrete Performance
by Daniel Moreno, Patricia Martinez, and Mauricio Lopez
The properties and amount of lightweight aggregates used in a concrete
mixture can significantly influence its mechanical properties and density.
Nevertheless, such influence cannot be accurately described and used in
practical application without an extensive experimental work. A practical
evaluation method for assessing the influence of lightweight aggregate on
concrete properties is required to anticipate the performance of concrete in
advance and choose the most suitable lightweight aggregate for a certain
structural application.
In this paper, existing models are reviewed, generalized, and validated
to obtain a methodology for assessing the potential of the lightweight
aggregates to provide a specified concrete density, modulus of elasticity,
and compressive strength. A practical evaluation methodology is proposed
and validated with four different lightweight aggregates, obtaining correlations between measured and estimated density; modulus of elasticity; and
compressive strengths of 93.4, 84.8, and 91.7%, respectively. Therefore,
this methodology allows practical and reliable comparison and selection of
lightweight aggregates based on only one trial mixture.
111-M12Axisymmetric Fiber Orientation Distribution of Short
Straight Fiber in Fiber-Reinforced Concrete
by Jun Xia and Kevin Mackie
The anisotropic orientation distribution of short fibers in fiber-reinforced
concrete and the impact on mechanical properties have been established in
past research. In this paper, the cast-flow induced anisotropic fiber distribution was categorized as axisymmetric with respect to the cement paste
flow direction. The probabilistic spatial orientation for fibers is introduced
using the beta distribution and uniformity parameters. Either theoretical or
approximate equations for the orientation factor and the probability density
function of the crossing angle were derived for any arbitrary cut plane. The
derived orientation factor equation can be used to quantify the degree of
anisotropy via image analysis. This process is easier and more accurate
because instead of detailed orientation information for every single fiber,
only total fiber counts on the cut sections are needed. The probabilistic
macromechanical properties, such as ultimate tensile strength, are estimated
based on the selected micromechanical model that defines the single fibermatrix interactions.
111-M13Mixture Design and Testing of Fiber-Reinforced
SelfConsolidating Concrete
by Kamal H. Khayat, Fodhil Kassimi, and Parviz Ghoddousi
An extensive testing program was undertaken to evaluate the applicability of a mixture-proportioning method proposed for shrinkage control
in fiber-reinforced concrete (FRC) in proportioning fiber-reinforced selfconsolidating concrete (FR-SCC). The study also proposed test methods
to evaluate workability of FR-SCC. The investigated fibers included
polypropylene, steel, and hybrid fibers of different properties with fiber
lengths of 5 to 50 mm (0.20 to 1.97 in.). Fiber volume Vf ranged between
0.25 and 0.75%. The study also aimed to determine the impact of fiber
type and addition on key properties of the fresh and hardened concrete.
Hardened properties included compressive, splitting tensile, and average
residual strengths.
Test results indicate that the proposed methodology to maintain constant
mortar thickness over coarse aggregate and fibers can reduce any significant drop in workability of FR-SCC, resulting from an increase in fiber

460

factor. In general, a Vf of 0.5% is found to be an upper limit for the production of SCC. A greater Vf can hinder the self-consolidatingcharacteristics.
For the assessment of the passing ability of FR-SCC, a modified J-ring
setup containing six or eight bars instead of 16 bars is proposed. The
passing ability of FR-SCC can be expressed as the ratio of diameter:height
at the center of the J-ring test. The passing ability can also be evaluated
using the L-box with a single bar instead of three bars.
A superworkable concrete (SWC) requiring low consolidation energy
can still be produced with a Vf of 0.75% when a viscosity-modifying
admixture is incorporated to prevent segregation and blockage. For the
tested fiber types, the average residual strength (ARS) in flexure is shown
to increase with Vf. Steel fibers exhibited the highest ARS value.
111-M14Effect of Misalignment on Pulloff Test Results:
Numerical and Experimental Assessments
by Luc Courard, Benot Bissonnette, Andrzej Garbacz, Alexander
Vaysburd, Kurt von Fay, Grzegorj Moczulski, and Maxim Morency
The successful application of a concrete repair system is often evaluated through pulloff testing. For such in-place quality control (QC) testing,
the inherent risk of misalignment might affect the recorded value and
eventually make a difference in the acceptance of the work. The issue of
eccentricity in pulloff testing has been ignored in field practice because it
is seen as an academic issue. This paper presents the results of a project
intended to quantify the effect of misalignment on pulloff tensile strength
evaluation and provide a basis for improving QC specifications if necessary. The test program consisted first of an analytical evaluation of the
problem through two-dimensional finite element modeling simulations and,
in a second phase, in laboratory experiments in which the test variables
were the misalignment angle (0, 2, and 4 degrees) and the coring depth
(15 and 30 mm [0.6 and 1.2 in.]). It was found that calculations provide
a conservative but realistic lower bound limit for evaluating the influence
of misalignment upon pulloff test results: a 2-degree misalignment can
be expected to yield a pulloff strength reduction of 7 to 9%, respectively,
for 15 and 30 mm (0.6 and 1.2 in.) coring depths, and the corresponding
decrease resulting from a 4-degree misalignment reaches between 13 and
16%. From a practical standpoint, the results generated in this study indicate
that when specifying a pulloff strength limit in the field, the value should
be increased (probable order of magnitude: 15%) to take into account the
potential reduction due to testing misalignment.
111-M15Strength and Microstructure of Mortar Containing
Nanosilica at High Temperature
by Rahel Kh. Ibrahim, R. Hamid, and M. R. Taha
The effect of high temperature on the mechanical properties and microstructure of nanosilica-incorporated mortars has been studied. Results show
that the incorporation of nanosilica increases both compressive and flexural strengths significantly at both ambient and after a 2-hour exposure to
752F (400C) temperatures; the strengths increase with the increase of
nanosilica content. A significant decrease in strength was recorded for all
control and nanosilica-incorporated mortar specimens after a 2-hour exposure to 1292F (700C) heat; however, nanosilica-incorporated specimens
show higher residual strength than those without nanosilica. Microstructural analysis shows that nanosilica reduces the calcium hydroxide crystals
to produce more calcium silicate hydrate, the process that contributes to
the strength and the residual strength of the material. In addition, the material exhibits a stable structure state up to 842F (450C), while exposure to
higher temperatures results in a decomposition of hydration products.

ACI Structural Journal/March-April 2014

111-M16Prediction of Strength, Permeability, and Hydraulic


Diffusivity of Ordinary Portland Cement Paste
by B. Kondraivendhan and B. Bhattacharjee
In this investigation, the compressive strength, permeability, and
hydraulic diffusivity of ordinary portland cement (OPC) paste have been
predicted from the knowledge of mixture factors, such as water-cement
ratio (w/c) and curing ages. The relationships for pore size distribution
(PSD) parameters, such as mean distribution radius r0.5, dispersion coefficient d, and permeable porosity P, are functions of the w/c and curing
ages, and are readily available in the literature. By using these relationships, the compressive strength, permeability, and hydraulic diffusivity can
be estimated from the w/c and curing age information. It is observed that the
predicted compressive strength closely matches the reported experimental
compressive strength. It is also observed that the estimated permeability
data also closely matches the reported experimental permeability data.
The predicted hydraulic diffusivity follows a similar trend, as reported in
theliterature.
111-M17Cracking Tendency of Lightweight Aggregate Bridge
Deck Concrete
by Benjamin E. Byard, Anton K. Schindler, and Robert W. Barnes
Early-age cracking in bridge decks is a severe problem that may reduce
the functional life of the structure. In this project, the effect of using lightweight aggregate on the cracking tendency of bridge deck concrete was
evaluated using testing frames that restrain movement due to volume
change effects from placement to cracking. Expanded shale, clay, and slate
lightweight coarse and fine aggregates were used to produce internal curing,
sand-lightweight, and all-lightweight concretes to compare their behavior
relative to a normalweight concrete in bridge deck applications.
Specimens were tested under temperature conditions that simulate
summer and fall placement scenarios. Increasing the amount of pre-wetted
lightweight aggregate in the concrete systematically decreased the density,
modulus of elasticity, and coefficient of thermal expansion of the concrete.
When compared to a normalweight concrete, the use of lightweight aggregates in concrete effectively delays the occurrence of early-age cracking in
bridge deck applications.
111-M18Multi-Aggregate Approach for Modeling Interfacial
Transition Zone in Concrete
by Hongyan Ma and Zongjin Li
Interfacial transition zone (ITZ) has long been of particular interest in
concrete technology. The limited sensitivity of experimental techniques
makes it attractive to study ITZ using computer simulations. In this paper,
a multi-aggregate approach is proposed to simulate the formation of ITZ
in concrete. In light of a modified status-oriented computer model for

ACI Structural Journal/March-April 2014

simulating cement hydration, the evolution of ITZ is also simulated in this


approach. Through simulations, the influences of several factors related
to concrete mixture proportion on ITZ are investigated. It is found that
ITZ thickness, as defined by the overall average porosity, can be reduced
by using finer aggregate, increasing aggregate volume fraction, reducing
water-cement ratio (w/c), or making the binder system finer. Following
hydration, the ITZ thickness decreases continuously, but the difference of
porosity between ITZ and bulk paste keeps almost constant at mature ages.
111-M19C4AF ReactivityChemistry and Hydration of
Industrial Cement
by Hugh Wang, Delia De Leon, and Hamid Farzam
The study described in this paper involved a close examination of
one of the least-researched cement phases, 4CaOAl2O3Fe2O3 (C4AF
or ferrite). The two ferrite materials used in this program were: 1)
extracted from an industrial clinker free of 3CaOAl2O3 (C3A or aluminate); and 2) a laboratory-synthesized C4AF compound. Both ferrite
materials were initially characterized using optical microscopy, X-ray
diffraction (XRD), and Raman spectroscopy. The hydration kinetics and
reactivity of both materials were studied with a calorimetric technique.
Similar to aluminate, ferrite also demonstrated a strong early hydration rate. An adequate amount of sulfate (SO3) was needed to regulate
the ferrite hydration rate. Based on this work, an equivalent aluminate
content, C3Aeq, is proposed. The hydration of C3Aeq needs to be properly
controlled with the appropriate amount of sulfate. The concept, based
on equivalent aluminate content for cement sulfate optimization, ensures
proper early hydration behavior from both C3A and C4AF and avoids
potential cement-admixture incompatibility problems, especially for
cements containing little or no C3A content.
111-M20Tailoring Hybrid Strain-Hardening Cementitious
Composites
by Alessandro P. Fantilli, Hirozo Mihashi, and Tomoya Nishiwaki
A class of fiber-reinforced concrete, commonly called strain-hardening cementitious composite (SHCC), can show very ductile behavior
under tension. In the post-cracking stage, several cracks develop before
complete failure, which occurs when tensile strains finally localize in one
of the formed cracks. To predict the mechanical performances of monofiber
SHCC, a cohesive model has been proposed. Such a model is used herein
to tailor hybrid SHCC, made with long and short fibers. By combining
uniaxial tensile tests and the theoretical results of the model, the critical
value of the fiber-volume fraction can be evaluated. It should be considered
as the minimum amount of long fibers that can lead to the formation of
multiple cracking and strain hardening under tensile actions. The aim of the
present research is to reduce such volume as much as possible, to improve
the workability, and to reduce the final cost of SHCC.

461

REVIEWERS IN 2013
In 2013, the individuals listed on these pages served as technical reviewers of papers offered for publication
in ACI periodicals. A special thank you to them for their voluntary assistance in helping ACI maintain
the high quality of its publication program.
azniewska-Piekarczyk, Beata
Silesian University of Technology
Rybnik, Poland
Aamidala, Hari Shankar
Parsons Brinckerhoff
Herndon, VA
Abdalla, Hany
College of Technological Studies
Shuwaikh, Kuwait
Abdel-Fattah, Hisham
University of Sharjah
Sharjah, United Arab Emirates
Abdelgader, Hakim
Tripoli University
Tripoli, Libyan Arab Jamahiriya
Abdelrahman, Amr
Heliopolis, Egypt
Abo-Shadi, Nagi
Robert Englekirk, Inc.
Santa Ana, CA
Abou-Zeid, Mohamed
American University in Cairo
Cairo, Egypt
Abu Yosef, Ali
University of Texas at Austin
Austin, TX
Achillopoulou, Dimitra
Democritus University of Thrace
Xanthi, Greece
Acun, Bora
University of Houston
Houston, TX
Adebar, Perry
University of British Columbia
Vancouver, BC, Canada
Adhikary, Bimal
Austin, TX
Agarwal, Pankaj
Indian Institute of Technology Roorkee
Roorkee, Uttarakhand, India
Aggelis, Dimitrios
University of Ioannina
Ioannina, Ioannina, Greece
Agustiningtyas, Rudi
Ministry of Public Works
Bandung, Indonesia
Ahmadi, Jamal
University of Science and Technology
Tehran, Tehran, Islamic Republic of Iran
Ahmadi Nedushan, Behrooz
Yazd University
Yazd, Yazd, Islamic Republic of Iran
Ahmed, Ayub
Birla Institute of Technology and Science, Pilani
Pilani, Rajasthan, India

462

Ahmed, Ehab
University of Sherbrooke
Sherbrooke, QC, Canada
Aidoo, John
Rose-Hulman Institute of Technology
Terre Haute, IN
Aire, Carlos
National Autonomous University of Mexico
Mexico City, DF, Mexico
Akakin, Tumer
Turkish Ready Mixed Concrete Association
Istanbul, Turkey
Akbari, Reza
University of Tehran
Tehran, Tehran, Islamic Republic of Iran
Akcay, Burcu
Kocaeli University
Kocaeli, Turkey
Akiyama, Mitsuyoshi
Waseda University
Tokyo, Japan
Akkaya, Yilmaz
Istanbul Technical University
Maslak, Istanbul, Turkey
Alam, A.K.M. Jahangir
Dhaka, Bangladesh
Alam, M. Shahria
The University of British Columbia
Kelowna, BC, Canada
Alam, Mahbub
Stamford University Bangladesh
Dhaka, Bangladesh
Al-Attar, Tareq
University of Technology
Baghdad, Iraq
Al-Azzawi, Adel
Nahrain University
Baghdad, Iraq
Albahttiti, Mohammed
Kansas State University
Manhattan, KS
Albuquerque, Albria
Federal Center of Technological Education of Mato Grosso
Cuiab, Mato Grosso, Brazil
Al-Chaar, Ghassan
U.S. Army Engineer Research and Development Center
Champaign, IL
Alcocer, Sergio
Institute of Engineering, UNAM
Mexico City, DF, Mexico
Aldajah, Saud
United Arab Emirates University
Al-Ain, United Arab Emirates
Aldea, Corina-Maria
St. Catharines, ON, Canada

ACI Structural Journal/March-April 2014


Alexander, Scott
UMA Engineering, Ltd.
Edmonton, AB, Canada
Ali, Nisreen
Universiti Putra Malaysia
Serdang, Selangor, Malaysia
Ali, Samia
University of Engineering and Technology, Lahore
Lahore, Punjab, Pakistan
Aljewifi, Hana
The University of Cergy-Pontoise
Cergy-Pontosie Cedex, France
Alkhairi, Fadi
Arabtech Jardaneh
Amman, Jordan
Alkhrdaji, Tarek
Structural Group
Hanover, MD
Allahdadi, Hamidreza
Bangalore, India
Al-Manaseer, Akthem
San Jose State University
San Jose, CA
Al-Martini, Samer
University of Western Ontario
London, ON, Canada
Al-Qaisy, Wissam
Safe Mix Ready Concrete
Sharjah, United Arab Emirates
Alqam, Maha
The University of Jordan
Amman, Jordan
Alsiwat, Jaber
Saudi Consulting Services
Riyadh, Saudi Arabia
Aly, Aly Mousaad
Louisiana State University
Baton Rouge, LA
Amani Dashlejeh, Asghar
Tehran, Tehran, Islamic Republic of Iran
Anania, Laura
University of Catania
Catania, Italy
Andersson, Ronny
Hollviken, Sweden
Andrade, Jairo
Pontifical Catholic University of Rio Grande do Sul
Porto Alegre, RS, Brazil
Andriolo, Francisco
Andriolo Ito Engenharia SC Ltda
So Carlos, So Paulo, Brazil
Ansari, Abdul Aziz
Quaid-e-Awam Engineering University
Nawabshah, Sindh, Pakistan
Aqel, Mohammad
King Saud University
Riyadh, Saudi Arabia
Aragn, Sergio
Holcim (Costa Rica)
San Rafael, Alajuela, Costa Rica

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Aravinthan, Thiru
University of Southern Queensland
Toowoomba, Queensland, Australia
Arisoy, Bengi
Ege University
Izmir, Turkey
Aristizabal-Ochoa, Jose
National University
Medellin, Antioquia, Colombia
Armwood, Catherine
University of Nebraska
Omaha, NE
Asaad, Diler
Gaziantep University
Gaziantep, Turkey
Asamoto, Shingo
Saitama University
Saitama, Saitama, Japan
Aslani, Farhad
University of Technology, Sydney
Sydney, New South Wales, Australia
Assaad, Joseph
Notre Dame University
Beirut, Lebanon
Atamturktur, Sez
Clemson University
Clemson, SC
Athanasopoulou, Adamantia
AKMI Metropolitan College
Xalandri, Greece
Attaalla, Sayed
ADR Engineering, Inc.
Mission Hills, CA
Aviram, Ady
Simpson Gumpertz & Heger, Inc.
San Francisco, CA
Awati, Mahesh
B.L.D.E.A.s College of Engineering & Technology, Bijapur
Bijapur, Karnataka, India
Awida, Tarek
KEO International Consultants
Kuwait
Ayano, Toshiki
Okayama University
Okayama, Japan
Aydin, Abdulkadir Cuneyt
Ataturk University
Erzurum, Turkey
Aydin, Ertug
European University of Lefke
Nicosia, Turkey
Aydn, Serdar
Dokuz Eylul University
Izmir, Turkey
Aykac, Sabahattin
Gazi University, Faculty of Engineering and Architecture
Ankara, Turkey
Azad, Abul
King Fahd University of Petroleum & Minerals
Dhahran, Saudi Arabia

463

REVIEWERS IN 2013
Aziz, Omar
University of Salahaddin
Erbil, Iraq
Babafemi, Adewumi
Obafemi Awolowo University
Ile-ife, Nigeria
Bacinskas, Darius
Vilnius Gediminas Technical University
Vilnius, Lithuania
Badger, Christian
Bates Engineering, Inc.
Lakewood, CO
Bae, Sungjin
Bechtel Corporation
Frederick, MD
Bai, Shaoliang
Chongqing University
Chonqqing, China
Bai, Yongtao
Kyoto University
Kyoto, Japan
Bakhshi, Mehdi
Arizona State University
Tempe, AZ
Balaguru, P.
Rutgers, the State University of New Jersey
Piscataway, NJ
Balouch, Sana
University of Dundee
Dundee, UK
Banibayat, Pouya
New York, NY
Banic, Davor
Civil Engineering Institute of Croatia
Zagreb, Croatia
Baran, Eray
Atilim University
Ankara, Turkey
Barbosa, Maria
Federal University of Juiz de Fora
Juiz De Fora, Brazil
Barroso de Aguiar, Jose
University of Minho
Guimaraes, Portugal
Bartos, Peter
University of the West of Scotland - Paisley
Paisley, Scotland, UK
Bashandy, Alaa
Menofiya University
Shibin El-Kom, Egypt
Batson, Gordon
Clarkson University
Potsdam, NY
Bayrak, Oguzhan
University of Texas at Austin
Austin, TX
Bayraktar, Alemdar
Trabzon, Turkey

464

Bayuaji, Ridho
Universiti Teknologi PETRONAS
Perak, Malaysia
Beddar, Miloud
Msila University
Msila, Algeria
Bediako, Mark
CSIR - Building and Road Research Institute
Kumasi, Ghana
Bedirhanoglu, Idris
Dicle University
Diyarbakir, Turkey
Behnoud, Ali
Iran University of Science and Technology
Tehran, Islamic Republic of Iran
Belleri, Andrea
University of Bergamo
Dalmine, Italy
Benboudjema, Farid
Ecole normale suprieure de Cachan
Cachan, France
Benliang, Liang
Shanghai, China
Bennett, Richard
The University of Tennessee
Knoxville, TN
Bentz, Dale
National Institute of Standards and Technology
Gaithersburg, MD
Berry, Michael
Montana State University
Bozeman, MT
Beygi, Morteza
Mazandaran University
Babol, Mazandaran, Islamic Republic of Iran
Bharati, Raj
National Institute of Technology Calicut
Calicut, Kerala, India
Bhargava, Kapilesh
Bhabha Atomic Research Centre
Mumbai, Maharashtra, India
Bhattacharjee, Bishwajit
Indian Institute Of Technology, Delhi
New Delhi, India
Bhatti, Abdul
National University of Sciences and Technology
Islamabad, Pakistan
Bilcik, Juraj
Slovak University of Technology in Bratislava
Bratislava, Slovakia
Bilek, Vlastimil
ZPSV a.s.
Brno, Czech Republic
Bilir, Turhan
Blent Ecevit University
Zonguldak, Turkey
Billah, Abu Hena
The University of British Columbia
Kelowna, BC, Canada

ACI Structural Journal/March-April 2014


Bimschas, Martin
Regensdorf, Switzerland
Bindiganavile, Vivek
University of Alberta
Edmonton, AB, Canada
Biolzi, Luigi
Politecnico di Milano
Milan, Italy
Birely, Anna
Texas A&M University
College Station, TX
Birkle, Gerd
Stantec Consulting Ltd.
Calgary, AB, Canada
Bisschop, Jan
University of Oslo
Oslo, Norway
Blair, Bruce
Lafarge North America
Herndon, VA
Bobko, Christopher
North Carolina State University
Raleigh, NC
Bochicchio, Victor
Hamon Custodis
Somerville, NJ
Bondar, Dali
Tehran, Islamic Republic of Iran
Bondy, Kenneth
Consulting Structural Engineer
West Hills, CA
Boshoff, William
Stellenbosch University
Stellenbosch, Western Cape, South Africa
Boulfiza, Moh
University of Saskatchewan
Saskatoon, SK, Canada
Bousias, Stathis
University of Patras
Patras, Greece
Bradberry, Timothy
TxDOT Bridge Division
Austin, TX
Braestrup, Mikael
Ramboll Hannemann and Hojlund A/S
Virum, Denmark
Brand, Alexander
University of Illinois at Urbana-Champaign
Urbana, IL
Brena, Sergio
University of Massachusetts
Amherst, MA
Brewe, Jared
CTLGroup
Skokie, IL
Bui, Van
BASF Admixtures, Inc.
Cleveland, OH

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Burak, Burcu
Orta Dogu Teknik Universitesi
Ankara, Turkey
Burdette, Edwin
University Of Tennessee
Knoxville, TN
Byard, Benjamin
University of Tennessee at Chattanooga
Chattanooga, TN
Calixto, Jos
Universidade Federal de Minas Gerais (UFMG)
Belo Horizonte, Brazil
Campione, Giuseppe
Universita Palermo
Palermo, Italy
Canbay, Erdem
Middle East Technical University
Ankara, Cankaya, Turkey
Cano Barrita, Prisciliano
Instituto Politcnico Nacional/CIIDIR Unidad Oaxaca
Oaxaca, Oaxaca, Mexico
Canpolat, Fethullah
Yildiz Technical University
Istanbul, Turkey
Capozucca, Roberto
Faculty of Engineering
Ancona, Italy
Carino, Nicholas
Chagrin Falls, OH
Carreira, Domingo
Chicago, IL
Carrillo, Julian
Universidad Militar Nueva Granada (UMNG)
Bogot, D.C., Colombia
Carvalho, Alessandra
Pontifical Catholic University of Gois (PUC-Gois)
Goinia, Gois, Brazil
Castles, Bryan
Western Technologies, Inc.
Phoenix, AZ
Castro, Javier
Pontificia Universidad Catolica de Chile
Santiago, Chile
Catoia, Bruna
Federal University of So Carlos (UFSCar)
So Carlos, So Paulo, Brazil
Cattaneo, Sara
Politecnico di Milano
Milan, Italy
Cedolin, Luigi
Politecnico di Milano
Milano, Italy
Cetisli, Fatih
Firat University
Elazig, Turkey
Chai, Hwa Kian
Tobishima Corporation
Noda, Chiba, Japan

465

REVIEWERS IN 2013
Chakraborty, Arun
Bengal Engineering and Science University
Howrah, West Bengal, India
Chang, Jeremy
Holmes Fire & Safety
Christchurch, New Zealand
Chastre, Carlos
FCT/UNL
Lisbon, Portugal
Chaudhry, Asif
Geoscience Advance Research Laboratories
Islamabad, Pakistan
Chawla, Komal
Indian Institute of Technology Kanpur
Kota, India
Chen, Chun-Tao
National Taiwan University of Science and Technology
Taipei, Taiwan
Chen, Hua-Peng
The University of Greenwich
Chatham, Kent, UK
Chen, Qi
Boral Materials Technology
San Antonio, TX
Chen, Shiming
Tongji University
Shanghai, China
Chen, Wei
Wuhan University of Technology
Wuhan, HuBei, China
Chen, Xi
Shanghai, China
Cheng, Min-Yuan
National Taiwan University of Science and Technology
Taipei, Taiwan
Chi, Maochieh
WuFeng University
Chiayi County, Taiwan
Chiaia, Bernardino
Politecnico di Torino
Torino, Piedmont, Italy
Chiang, Chih-Hung
Chaoyang University of Technology
Wufong, Taichung, Taiwan
Chindaprasirt, Prinya
Khon Kaen University
Khon Kaen, Thailand
Chiorino, Mario
Politecnico di Torino
Torino, Italy
Cho, Chang-Geun
Chosun University
Gwangju, Republic of Korea
Cho, Jae-Yeol
Seoul National University
Seoul, Republic of Korea
Cho, Soon-Ho
Gwangju University
Gwangju, Republic of Korea

466

Choi, Bong-Seob
Chungwoon University
Hongseong-Gun, Republic of Korea
Choi, Eunsoo
Hongik University
Seoul, Republic of Korea
Choi, Kyoung-Kyu
Soongsil University
Seoul, Republic of Korea
Choi, Sejin
University of California, Berkeley
Albany, CA
Chompreda, Praveen
Mahidol University
Nakornpathom, Thailand
Choong, Kokkeong
Universiti Sains Malaysia
Pulau Pinang, Seberang Perai Selatan, Malaysia
Chorzepa, Migeum
Park Ridge, IL
Chowdhury, Subrato
Ultra Tech Cement, Ltd.
Mumbai, Maharashtra, India
Chun, Sung-Chul
Mokpo National University
Mooan-gun, Republic of Korea
Chung, Deborah
State University of New York
Buffalo, NY
Chung, Jae
University of Florida
Gainesville, FL
Chung, Lan
Dankook University
Seoul, Republic of Korea
Claisse, Peter
Coventry University
Coventry, UK
Cleary, Douglas
Rowan University
Glassboro, NJ
Clendenen, Joseph
Pleasant Hill, IL
Cobo, Alfonso
Polytechnic University of Madrid
Madrid, Spain
Coelho, Jano
AltoQi Informatica
Florianopolis, Santa Catarina, Brazil
Colombo, Matteo
Politecnico di Milano
Lecco, Italy
Conley, Christopher
United States Military Academy
West Point, NY
Cordova, Carlos
La Paz, Bolivia
Coronelli, Dario
Politecnico di Milano
Milano, Italy

ACI Structural Journal/March-April 2014


Corral, Ramn
Universidad Autnoma de Sinaloa
Los Mochis, Sinaloa, Mexico
Correal, Juan
Arcon Structural Engineers, Inc.
Rancho Santa Margarita, CA
Cortes, Douglas
New Mexico State University
Las Cruces, NM
Criswell, Marvin
Colorado State University
Fort Collins, CO
Cueto, Jorge
Universidad de La Salle
Bogota, Colombia
D. L., Venkatesh Babu
Kumaraguru College of Technology
Coimbatore, Tamil Nadu, India
Dang, Canh
University of Arkansas
Fayetteville, AR
Das, Sreekanta
University of Windsor
Windsor, ON, Canada
Daye, Marwan
Intima International Co., Ltd.
Al-Khobar, Saudi Arabia
de Brito, Jorge
IST/TUL
Lisbon, Portugal
de Frutos, Jose
Universidad Politecnica de Madrid
Madrid, Spain
De Rooij, Mario
TNO
Delft, the Netherlands
Deb, Arghya
Indian Institute of Technology, Kharagpur
Kharagpur, West Bengal, India
Degtyarev, Vitaliy
Columbia, SC
Dehn, Frank
University of Leipzig
Leipzig, Germany
Delalibera, Rodrigo
University of So Paulo
So Carlos, So Paulo, Brazil
Delatte, Norbert
Cleveland State University
Broadview Heights, OH
Demir, Ali
Celal Bayar University
Manisa, Turkey
Demir, Serhat
Blacksea Technical University
Trabzon, Turkey
Den Uijl, Joop
Delft University of Technology
Delft, the Netherlands

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Deng, Yaohua
Iowa State University
Ames, IA
Detwiler, Rachel
Braun Intertec Corporation
Minneapolis, MN
Devries, Richard
Milwaukee School of Engineering
Milwaukee, WI
Dhanasekar, Manicka
Queensland University of Technology
Brisbane, Queensland, Australia
Dhonde, Hemant
University of Houston
Houston, TX
Di Ludovico, Marco
University of Naples Federico II
Naples, Italy
Dias, WPS
University of Moratuwa
Moratuwa, Sri Lanka
Diaz Loya, Eleazar
Louisiana Tech University
Ruston, LA
Dilger, Walter
University of Calgary
Calgary, AB, Canada
Ding, Yining
Dalian, China
Diniz, Sofia Maria
Universidade Federal de Minas Gerais (UFMG)
Be lo Horizonte, Brazil
Do, Jeongyun
Kunsan National University
Kunsan, Jeonbuk, Republic of Korea
Dodd, Larry
Parsons Brinckerhoff
Orange, CA
Dogan, Unal
Istanbul Technical University
Istanbul, Turkey
Dolan, Charles
University of Wyoming
Laramie, WY
Dongell, Jonathan
Pebble Technologies
Scottsdale, AZ
Dongxu, Li
Nanjing University of Technology
Nanjing, Jiangsu, China
Dontchev, Dimitar
University of Chemical Technology and Metallurgy
Sofia, Bulgaria
Dragunsky, Boris
Universal Construction Testing, Ltd.
Highland Park, IL
Du, Jinsheng
Beijing Jiao Tong University
Beijing, China

467

REVIEWERS IN 2013
Du, Lianxiang
The University of Alabama at Birmingham
Birmingham, AL
Du, Yingang
Anglia Ruskin University
Chelmsford, UK
Dubey, Ashish
United States Gypsum Corp
Libertyville, IL
Dutta, Anjan
Indian Institute of Technology Guwahati
Guwahati, Assam, India
Dutton, John
Edmonton, AB, Canada
Eid, Rami
University of Sherbrooke
Sherbrooke, QC, Canada
El Ragaby, Amr
University of Manitoba
Winnipeg, MB, Canada
Elamin, Anwar
University of Nyala
Khartoum, Sudan
El-Ariss, Bilal
United Arab Emirates University
Al Ain, United Arab Emirates
Elbahar, Mohamed
Ken Okamoto & Associates: Structural Engineering
Rancho Santa Margarita, CA
ElBatanouny, Mohamed
University of South Carolina
West Columbia, SC
Eldarwish, Aly
Alexandria, Egypt
El-Dash, Karim
College of Technological Studies
Kuwait, Kuwait
El-Dieb, Amr
Ain Shams University
Abbasia, Cairo, Egypt
El-Hawary, Moetaz
Kuwait Institute for Scientific Research
Safat, Kuwait
Elkady, Hala
National Research Centre (NRC)
Giza, Egypt
El-Maaddawy, Tamer
United Arab Emirates University
Al-Ain, Abu Dhabi, United Arab Emirates
Elmenshawi, Abdelsamie
University of Calgary
Calgary, AB, Canada
El-Metwally, Salah
University of Hawaii at Manoa
Honolulu, HI
Elnady, Mohamed
Mississauga, ON, Canada
El-Refaie, Sameh
El-Gama City, Mataria, Cairo, Egypt

468

El-Salakawy, Ehab
University of Manitoba
Winnipeg, MB, Canada
Elsayed, Tarek
Cairo, Egypt
Emamy Farvashany, Firooz
Perthpolis Pty, Ltd.
Perth, Western Australia, Australia
Eom, Tae-Sung
Catholic University of Daegu
Kyeongsan-si, Republic of Korea
Erdem, T.
Izmir Institute of Technology
Izmir, Turkey
Ergn, Ali
Afyon Kocatepe University
Afyonkarahsar, Turkey
Evangelista, Lus
Instituto Superior de Engenharia de Lisboa
Lisbon, Portugal
Fafitis, Apostolos
Arizona State University
Tempe, AZ
Fanella, David
Klein and Hoffman
Chicago, IL
Fantilli, Alessandro
Politecnico di Torino
Torino, Italy
Faraji, Mahdi
Tehran, Islamic Republic of Iran
Fardis, Michael
Patras, Greece
Farghaly, Ahmed
University of Sherbrooke
Sherbrooke, QC, Canada
Faria, Duarte
Faculdade de Cincias e Tecnologia
Caparica-Lisbon, Portugal
Farrow, William
Lebanon, NJ
Farzam, Masood
Structural Engineering
Tabriz, Islamic Republic of Iran
Fenollera, Maria
Universidade de Vigo
Vigo, Spain
Ferguson, Bruce
University of Georgia
Athens, GA
Fernndez Montes, David
Madrid, Spain
Fernndez Ruiz, Miguel
Ecole Polytechnique Federale De Lausanne
Lausanne, Vaud, Switzerland
Ferrara, Liberato
Politecnico di Milano
Milan, Italy

ACI Structural Journal/March-April 2014


Ferrier, E.
Universit Claude Bernard Lyon 1
Villerubanne, France
Folino, Paula
University of Buenos Aires
Buenos Aires, Argentina
Foster, Stephen
University of New South Wales
Sydney, New South Wales, Australia
Fouad, Fouad
University of Alabama at Birmingham
Birmingham, AL
Fradua, Martin
Feld, Kaminetzky & Cohen, P.C.
Jericho, NY
Francois, Buyle-Bodin
University of Lille
Villeneuve dAscq, France
Fuchs, Werner
University of Stuttgart
Stuttgart, Germany
Furlong, Richard
Austin, TX
G., Dhinakaran
Sastra University
Thanjavur, India
Gabrijel, Ivan
University of Zagreb
Zagreb, Croatia
Galati, Nestore
Structural Group, Inc.
Elkridge, MD
Gallegos Mejia, Luis
Fundacion Padre Arrupe de El Salvador
Soyapango, San Salvador, El Salvador
Gamble, William
University of Illinois at Urbana-Champaign
Urbana, IL
Ganesan, N.
National Institute of Technology
Calicut, India
Garboczi, Edward
National Institute of Standards and Technology
Gaithersburg, MD
Garcez, Estela
Universidade Federal de Pelotas (UFPel)
Pelotas, RS, Brazil
Garcia-Taengua, Emilio
Universidad Politecnica de Valencia
Valencia, Spain
Gardoni, Paolo
University of Illinois at Urbana-Champaign
Urbana, IL
Gayarre, Fernando
Gijon, Spain
Gayed, Ramez
University of Calgary
Calgary, AB, Canada

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Gesund, Hans
University of Kentucky
Lexington, KY
Ghafari, Nima
Laval University
Quebec, QC, Canada
Giaccio, Craig
AECOM
Melbourne, Victoria, Australia
Giancaspro, James
University of Miami
Coral Gables, FL
Gilbert, Raymond
The University of New South Wales
Sydney, New South Wales, Australia
Girgin, Canan
Yildiz Technical University
Istanbul, Turkey
Gisario, Annamaria
Sapienza Universit di Roma
Rome, Italy
Gnaedinger, John
Con-Cure Corporation
Chesterfield, MO
Goel, Rajeev
CSIR - Central Road Research Institute
Delhi, India
Goel, Savita
Whitlock Dalrymple Poston & Associates
New York, NY
Gke, H. Sleyman
Gazi University
Ankara, Turkey
Gongxun, Wang
Hunan University of Science and Technology
Xiangtan, China
Gonzales Garcia, Luis Alberto
Lagging SA
Lima, Peru
Gonzlez-Valle, Enrique
Madrid, Spain
Grandic, Davor
University of Rijeka
Rijeka, Croatia
Grattan-Bellew, P.
Materials & Petrographic Research G-B Inc.
Ottawa, ON, Canada
Greene, Thomas
W. R. Grace & Co.
Houston, TX
Gribniak, Viktor
Vilnius Gediminas Technical University
Vilnius, Lithuania
Gu, Xiang-Lin
Tongji University
Shanghai, China
Guadagnini, Maurizio
The University of Shefifeld
Sheffield, UK

469

REVIEWERS IN 2013
Guan, Garfield
University of Cambridge
Cambridge, UK
Guimaraes, Giuseppe
Pontificia Universidade Catlica do Rio de Janeiro
Rio de Janeiro, Brazil
Gneyisi, Erhan
Gaziantep University
Gaziantep, Turkey
Guo, Guohui
Overland Park, KS
Guo, Liping
Southeast University
Nanjing, Jiangsu Province, China
Guo, Zixiong
Huaqiao University
Quanzhou, Fujian, China
Gupta, Ajay
M.B.M. Engineering College
Jodhpur, Rajasthan, India
Gupta, Pawan
Post-Tensioning Institute
Phoenix, AZ
Haach, Vladimir
University of So Paulo
So Carlos, So Paulo, Brazil
Habbaba, Ahmad
Technische Universitt Mnchen
Garching, Germany
Haddad, Gilbert
SNC-Lavalin, Inc.
St. Laurent, QC, Canada
Haddadin, Laith
United Nations
New York, NY
Hadi, Muhammad
University of Wollongong
Wollongong, New South Wales, Australia
Hadje-Ghaffari, Hossain
John A. Martin & Associates
Los Angeles, CA
Hagenberger, Michael
Valparaiso University
Valparaiso, IN
Hamid, Roszilah
Universiti Kebangsaan Malaysia
Bangi, Selangor, Malaysia
Hamilton, Trey
University of Florida
Gainesville, FL
Hansen, Will
Brighton, MI
Harajli, Mohamed
American University of Beirut
Beirut, Lebanon
Harbec, David
University of Sherbrooke
Sherbrooke, QC, Canada

470

Harris, Devin
University of Virginia
Charlottesville, VA
Harris, G Terry
Green Cove Springs, FL
Harris, Nathan
Menlo Park, CA
Hashemi, Shervin
Seoul National University
Seoul, Republic of Korea
Hassan, Assem
Toronto, ON, Canada
Hassan, Maan
University of Technology
Baghdad, Iraq
Hassan, Wael
University of California, Berkeley
Berkeley, CA
He, Xiaobing
Chongqing Jiaotong University
Chongqing, China
He, Zhiqi
Southeast University
Nanjing, Jiangsu, China
Heinzmann, Daniel
Lucerne University of Applied Sciences and Arts
Horw, Switzerland
Henry, Richard
University of Auckland
Auckland, New Zealand
Himawan, Aris
Singapore, Singapore
Hindi, Riyadh
Saint Louis University
St. Louis, MO
Hoehler, Matthew
Encinitas, CA
Holschemacher, Klaus
HTWK Leipzig
Leipzig, Germany
Hong, Sung-Gul
Seoul National University
Seoul, Republic of Korea
Hooton, R. Doug
University of Toronto
Toronto, ON, Canada
Hossain, Khandaker
Ryerson University
Toronto, ON, Canada
Hossain, Tanvir
Louisiana State University
Baton Rouge, LA
Hosseini, Ardalan
Isfahan University of Technology (IUT)
Isfahan, Islamic Republic of Iran
Hoult, Neil
Toronto, ON, Canada
Hover, Kenneth
Cornell University
Ithaca, NY

ACI Structural Journal/March-April 2014


Hrynyk, Trevor
University of Toronto
Toronto, ON, Canada
Hsu, Thomas
University of Houston
Houston, TX
Hu, Jiong
Texas State University-San Marcos
San Marcos, TX
Hu, Nan
Tsinghua University
Beijing, China
Huang, Chang-Wei
Chung Yuan Christian University
Chung Li, Taiwan
Huang, Jianwei
Southern Illinois University Edwardsville
Edwardsville, IL
Huang, Xiaobao
GM-WFG/GM-N American Project Center
Warren, MI
Huang, Yishuo
Chaoyang University of Technology
Wufeng, Taichung, Taiwan
Huang, Zhaohui
Brunel University
London, UK
Hubbell, David
Toronto, ON, Canada
Hueste, Mary Beth
Texas A&M University
College Station, TX
Hulsey, J.
University of Alaska
Fairbanks, AK
Hung, Chung-Chan
University of Michigan
Ann Arbor, MI
Husain, Mohamed
Zagazig University
Zagazig, Egypt
Husem, Metin
Karadeniz Technical University
Trabzon, Turkey
Hussain, Raja
King Saud University
Riyadh, Saudi Arabia
Hussein, Amgad
Memorial University of Newfoundland
St. Johns, NL, Canada
Hwang, Chao-Lung
National Taiwan University of Science and Technology
Taipei, Taiwan
Ibell, Tim
University of Bath
Bath, UK
Ibrahim, Amer
Baquba, Iraq

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Ibrahim, Hisham
Buckland and Taylor, Ltd.
North Vancouver, BC, Canada
Ichinose, Toshikatsu
Nagoya Institute of Technology
Nagoya, Japan
Ideker, Jason
Oregon State University
Corvallis, OR
Ince, Ragip
Firat University
Elazig, Turkey
Ipek, Sleyman
Gaziantep University
Gaziantep, Turkey
Irassar, Edgardo
Universidad Nacional del Centro de la Provincia de Buenos
Aires (UNCPBA)
Olavarria, Buenos Aires, Argentina
Islam, Md.
Chittagong University of Engineering & Technology (CUET)
Chittagong, Bangladesh
Issa, Mohsen
University of Illinois at Chicago
Chicago, IL
Izquierdo-Encarnacin, Jose
Porticus
San Juan, PR
Jaari, Asaad
Dera, Dubai, United Arab Emirates
Jaeger, Thomas
Baenziger Partner AG
Chur, Switzerland
Jain, Mohit
Nirma University
Ahmedabad, Gujarat, India
Jain, Shashank
Delhi Technological University (DTU)
New Delhi, India
Jamshidi, Masoud
Building and Housing Research Center (BHRC)
Tehran, Islamic Republic of Iran
Jang, Seung Yup
Korea Railroad Research Institute
Uiwang, Gyongggi-do, Republic of Korea
Jansen, Daniel
California Polytechnic State University
San Luis Obispo, CA
Jau, Wen-Chen
National Chiao Tung University (NCTU)
Hsinchu, Taiwan
Jawaheri Zadeh, Hany
Miami, FL
Jayapalan, Amal
Georgia Institute of Technology
Atlanta, GA
Jeng, Chyuan-Hwan
National Chi Nan University-Taiwan
Puli, Nantou, Taiwan

471

REVIEWERS IN 2013
Jensen, Elin
Lawrence Technological University
Southfield, MI
Jeon, Se-Jin
Ajou University
Suwon-si, Gyeonggi-do, Republic of Korea
Jiang, Jiabiao
W. R. Grace (Singapore) Pte. Ltd.
Singapore
Jirsa, James
University of Texas at Austin
Austin, TX
Jozic, Draan
Universty of Split
Split, Croatia
Kabele, Petr
Czech Technical University in Prague
Praha 6, Czech Republic
Kagaya, Makoto
Akita, Japan
Kalkan, Ilker
Krkkale University
Krkkale, Turkey
Kan, Yu-Cheng
Chaoyang University of Technology
Taichung County, Taiwan
Kanakubo, Toshiyuki
University of Tsukuba
Tsukuba, Japan
Kandasami, Siva
Bristol, UK
Kang, Thomas
Seoul National University
Seoul, Republic of Korea
Kankam, Charles
Kwame Nkrumah University of Science & Technology
Kumasi, Ghana
Kansara, Kunal
University of Bath
Bath, UK
Kantarao, Velidandi
Central Road Research Institute
New Delhi, Delhi, India
Karahan, Okan
Erciyes University
Erciyes, Turkey
Karayannis, Christos
Democritus University of Thrace
Xanthi, Greece
Karbasi Arani, Kamyar
University of Naples Federico II
Napoli, Italy
Kawamura, Mitsunori
Kanazawa, Ishikawa, Japan
Kazemi, Mohammad
Sharif University of Technology
Tehran, Islamic Republic of Iran
Kazemi, Sadegh
University of Alberta
Edmonton, AB, Canada

472

Kenai, Said
Universit de Blida
Blida, Algeria
Kevern, John
University of Missouri-Kansas City
Kansas City, MO
Khan, Mohammad
King Saud University
Riyadh, Saudi Arabia
Kheder, Ghazi
University of Al Mustansiriya
Baghdad, Iraq
Khennane, Amar
Australian Defense Force Academy, University of New South
Wales (AFDA, UNSW)
Canberra, Australian Capital Territory, Australia
Khuntia, Madh
DuKane Precast Inc.
Naperville, IL
Kianoush, M. Reza
Ryerson University
Toronto, ON, Canada
Kilic, Sami
Bogazici University
Istanbul, Turkey
Kim, Jae Hong
UNIST
Ulsan, Republic of Korea
Kim, Jang Hoon
Ajou University
Suwon, Republic of Korea
Kim, Yail
University of Colorado Denver
Denver, CO
Kirgiz, Mehmet
Hacettepe University
Ankara, Turkey
Kishi, Norimitsu
Muroran Institute of Technology
Muroran, Japan
Kisicek, Tomislav
University of Zagreb
Zagreb, Croatia
Klemencic, Ronald
Magnusson Klemencic Associates
Seattle, WA
Ko, Lesley Suz-Chung
Holcim Group Support, Ltd.
Holderbank, AG, Switzerland
Koehler, Eric
University of Texas at Austin
Austin, TX
Koenders, Eddy A. B.
Delft University of Technology
Delft, the Netherlands
Kotsovos, Gerasimos
National Technical University of Athens
Athens, Greece
Kotsovos, Michael
Athens, Greece

ACI Structural Journal/March-April 2014


Krem, Slamah
University of Waterloo
Waterloo, ON, Canada
Krstulovic-Opara, Neven
ExxonMobil Development Company
Houston, TX
Kumar, Pardeep
University of California, Berkeley
Berkeley, CA
Kumar, Rakesh
Central Road Research Institute
Delhi, India
Kumar, Vinod
Steel Authority of India Limited
Ranchi, Jharkhand, India
Kunieda, Minoru
Nagoya University
Nagoya, Japan
Kunnath, Sashi
University of California, Davis
Davis, CA
Kupwade-Patil, Kunal
Massachusetts Institute of Technology
Cambridge, MA
Kurtis, Kimberly
Georgia Institute of Technology
Atlanta, GA
Kusbiantoro, Andri
Universiti Malaysia Pahang
Gambang, Pahang, Malaysia
Kwan, Albert
The University of Hong Kong
Hong Kong, China
Kwan, Wai Hoe
Universiti Sains Malaysia
Gelugor, Penang, Malaysia
La Tegola, Antonio
University of Lecce
Lecce, Italy
LaFave, James
University of Illinois at Urbana-Champaign
Champaign, IL
Lai, Jianzhong
Nanjing University of Science and Technology
Nanjing, Jiangsu, China
Larbi, Kacimi
University of Sciences and Technology of Oran
Oran, Oran, Algeria
Laskar, Aminul
National Institute of Technology
Silchar, Assam, India
Laterza, Michelangelo
University of Basilicata
Potenza, Italy
Law, David
RMIT University
Melbourne, Victoria, Australia
Lawler, John
Wiss, Janney, Elstner Associates, Inc.
Northbrook, IL

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Lee, Chadon
Chung-Ang University
Ansung, Kyungki-do, Republic of Korea
Lee, Chung-Sheng
University of California, San Diego
La Jolla, CA
Lee, Deuck Hang
University of Seoul
Seoul, Republic of Korea
Lee, Douglas
Douglas D. Lee & Associates
Fort Worth, TX
Lee, Heui Hwang
Arup
San Francisco, CA
Lee, Hung-Jen
National Yunlin University of Science and Technology
Douliu, Yunlin, Taiwan
Lee, Jaeman
Kyoto University
Kyoto, Japan
Lee, Jung-Yoon
Sung Kyun Kwan University
Suwon, Republic of Korea
Lee, Nam Ho
SNC-Lavalin Nuclear Inc.
Oakville, ON, Canada
Lee, Seong-Cheol
KEPCO International Graduate School (KINGS)
Ulsan, Republic of Korea
Lee, Seung-Chang
Samsung C&T Corporation
Seoul, Republic of Korea
Lee, Yoon-Si
Bradley University
Peoria, IL
Lee, Young Hak
Seoul, Republic of Korea
Leiva Fernndez, Carlos
University of Seville
Seville, Andalucia, Spain
Leo Braxtan, Nicole
New York, NY
Lequesne, Remy
University of Kansas
Madison, KS
Leutbecher, Torsten
Universitt Kassel
Kassel, Germany
Li, Fumin
China University of Mining and Technology
Xuzhou, Jiangsu, China
Li, Wei
Wenzhou University
Wenzhou, Zhejiang, China
Lignola, Gian Piero
University of Naples
Naples, Italy

473

REVIEWERS IN 2013
Lima, Maria Cristina
Federal University of Uberlndia
Uberlandia, Minas Gerais, Brazil
Lin, Wei-Ting
Institute of Nuclear Energy Research
Taoyuan, Taiwan
Lin, Yiching
National Chung-Hsing Univ
Taichung, Taiwan
Liu, Shuhua
Wuhan University
Wuhan, HuBei, China
Liu, Yanbo
Florida Department of Transportation-State Materials Office
Gainesville, FL
Lizarazo Marriaga, Juan
Coventry University
Coventry, UK
Londhe, Rajesh
Government College of Engineering Aurangabad
Aurangabad, Maharashtra, India
Long, Adrian
Queens University
Belfast, Ireland
Loo, Yew-Chaye
Gold Coast, Australia
Loper, James
Jacobs Facilities, Inc.
Arlington, VA
Lopes, Anne
Furnas Centrais Eltricas S/A
Aparecida De Goiania, Goias, Brazil
Lopes, Sergio
University of Coimbra
Coimbra, Portugal
Lpez-Almansa, Francisco
Technical University of Catalonia
Barcelona, Spain
Lounis, Zoubir
National Research Council
Ottawa, ON, Canada
Lubell, Adam
Read Jones Christoffersen, Ltd.
Vancouver, BC, Canada
Ludovit, Nad
Alfa 04
Kosice, Slovakia
Luo, Baifu
Harbin, China
Lushnikova, Nataliya
National University of Water Management and Nature

ResourcesUse
Rivne, Ukraine
Ma, Zhongguo
University of Tennessee
Knoxville, TN
MacDougall, Colin
Kingston, ON, Canada

474

Machida, Atsuhiko
Saitama University
Saitama, Japan
Macht, Jrgen
Kirchdorf, Austria
Maekawa, Koichi
University of Tokyo
Tokyo, Japan
Magliulo, Gennaro
University of Naples Federico II
Naples, Italy
Magureanu, Cornelia
Technical University of Cluj Napoca
Cluj Napoca, Cluj, Romania
Mahboub, Kamyar
University of Kentucky
Lexington, KY
Mahfouz, Ibrahim
Cairo, Egypt
Malik, Adnan
University of New South Wales
Sydney, Australia
Mancio, Mauricio
University of California, Berkeley
Berkeley, CA
Mander, John
Texas A&M University
College Station, TX
Manso, Juan
University of Burgos
Burgos, Castilla-Len, Spain
Marikunte, Shashi
Southern Illinois University
Carbondale, IL
Martinelli, Enzo
University of Salerno
Fisciano, Italy
Mart-Vargas, Jos
Universitat Politcnica de Valncia
Valencia, Valencia, Spain
Maruyama, Ippei
Nagoya University
Nagoya, Aichi, Japan
Maslehuddin, Mohammed
King Fahd University of Petroleum and Minerals
Dhahran, Saudi Arabia
Matamoros, Adolfo
University of Kansas
Lawrence, KS
Mathew, George
Cochin University of Science and Technology
Cochin, Kerala, India
Matsagar, Vasant
Lawrence Technological University
Southfield, MI
Matta, Fabio
University of South Carolina
Columbia, SC

ACI Structural Journal/March-April 2014


Maximos, Hany
Pharos University in Alexandria
Alexandria, Egypt
Mbessa, Michel
University of Yaound I - ENSP
Yaound, Center, Cameroon
McCabe, Steven
Lawrence, KS
McCall, W.
Concrete Engineering Consultants
Charlotte, NC
McCarter, John
Heriot Watt University
Edinburgh, UK
McDonald, David
USG Corporation
Libertyville, IL
McLeod, Heather
Kansas Department of Transportation
Topeka, KS
Meda, Alberto
University of Bergamo
Bergamo, Italy
Medallah, Khaled
Saudi Aramco IKPMS
Al Khobar, Saudi Arabia
Meddah, Seddik
Kingston University London
Kingston, London, UK
Mehanny, Sameh
Cairo University
Cairo, Egypt
Meinheit, Donald
Wiss, Janney, Elstner Associates, Inc.
Chicago, IL
Meininger, Richard
Turner-Fairbank Highway Research Center-FHWA
Columbia, MD
Mejia, Luis Gonzalo
LGM & Cia
Medellin, Colombia
Melchers, Robert
The University of Newcastle
Newcastle, New South Wales, Australia
Melo, Jos
University of Aveiro
Aveiro, Portugal
Meng, Tao
Institution of Building Materials
Hangzhou, Zhejiang, China
Menon, Devdas
Indian Institute of Technology
Chennai, Tamilnadu, India
Mermerdas , Kasm
Hasan Kalyoncu University
Gaziantep, Turkey
Meshgin, Pania
University of Colorado Boulder
Boulder, CO

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Mezhov, Alexander
Moscow State University of Civil Engineering
Moscow, Russian Federation
Milestone, Neil
Callaghan Innovation
Lower Hutt, New Zealand
Minelli, Fausto
University of Brescia
Brescia, Brescia, Italy
Mishra, Laxmi
Motilal Nehru National Institute of Technology Allahabad
Allahabad, Uttar Pradesh, India
Mlynarczyk, Alexandar
Wiss, Janney, Elstner Associates, Inc.
Princeton Junction, NJ
Mohamad, Gihad
University of Extremo Sul Catarinense - UNESC
Alegrete, Rio Grande do Sul, Brazil
Mohamed, Ashraf
Alexandria University
Alexandria, Egypt
Mohamed, Nayera
Assiut University
Assiut, Egypt
Mohammed, Tarek
University of Asia Pacific
Dhaka, Bangladesh
Montejo, Luis
North Carolina State University
Raleigh, NC
Moradian, Masoud
Oklahoma State University
Stillwater, OK
Moreno Jnior, Armando
Unicamp
Campinas, So Paulo, Brazil
Moretti, Marina
University of Thessaly
Athens, Greece
Moriconi, Giacomo
Technical University of Marche
Ancona, Italy
Morley, Christopher
Cambridge University
Cambridge, UK
Moser, Robert
US Army Engineer Research and Development Center
Vicksburg, MS
Mostofinejad, Davood
Isfahan University of Technology
Isfahan, Isfahan, Islamic Republic of Iran
Motaref, Sarira
University of Connecticut
Storrs, CT
Mubin, Sajjad
University of Engineering and Technology
Lahore, Punjab, Pakistan
Mulaveesala, Ravibabu
Indian Institute of Information Technology
Jabalpur, India

475

REVIEWERS IN 2013
Munoz, Jose
Federal Highway Administration
McLean, VA
Muttoni, Aurelio
Swiss Federal Institute of Technology
Lausanne, Switzerland
Nabavi, Esrafil
Rezvanshahr, Guilan, Islamic Republic of Iran
Nafie, Amr
Cairo, Egypt
Naish, David
California State University, Fullerton
Fullerton, CA
Najimi, Meysam
University of Nevada, Las Vegas
Las Vegas, NV
Nakamura, Hikaru
Nagoya University
Nagoya, Aichi, Japan
Narayanan, Pannirselvam
VIT University
Vellore, Tamilnadu, India
Narayanan, Subramanian
Gaithersburg, MD
Nassif, Hani
Rutgers, The State University of New Jersey
Piscataway, NJ
Negrutiu, Camelia
Technical University of Cluj Napoca
Cluj Napoca, Cluj, Romania
Neves, Lus
University of Coimbra
Coimbra, Portugal
Ng, Ernesto
Maveang, S.A.
Panama, Panama
Ng, Pui Lam
The University of Hong Kong
Hong Kong
Nichols, John
Texas A&M University
College Station, TX
Nimityongskul, Pichai
Asian Institute of Technology
Pathumthani, Thailand
Nkinamubanzi, Pierre-Claver
Institute for Research in Construction
Ottawa, ON, Canada
Nokken, Michelle
Concordia University
Montreal, QC, Canada
Noor, Munaz
Bangladesh University of Engineering and Technology
Dhaka, Bangladesh
Noshiravani, Talayeh
cole polytechnique fdrale de Lausanne (EPFL)
Lausanne, Switzerland
Novak, Lawrence
Portland Cement Association
Skokie, IL

476

Nowak, Andrzej
University of Nebraska-Lincoln
Lincoln, NE
Nwaubani, Sunny Onyebuchi
Anglia Ruskin University
Chelmsford, UK
OConnor, Arthur
O C Engineering
Reno, NV
Offenberg, Matthew
Rinker Materials
Orlando, FL
Oh, Byung
Seoul National University
Seoul, Republic of Korea
Okeil, Ayman
Louisiana State University
Baton Rouge, LA
Olanitori, Lekan
The Federal University of Technology Akure
Akure, Ondo State, Nigeria
Orakcal, Kutay
Bogazici University
Istanbul, Turkey
Orr, John
University of Bath
Bath, UK
Orta, Luis
Instituto Tecnolgico y de Estudios Superiores de Monterrey
(ITESM)
Zapopan, Jalisco, Mexico
Ortega, J.
University of Alacant
Alacant, Alicante, Spain
Otieno, Mike
University of Cape Town
Cape Town, Western Cape, South Africa
Otsuki, Nobuaki
Tokyo Institute of Technology
Tokyo, Japan
Ousalem, Hassane
Takenaka Corporation - Research and Development Institute
Inzai, Chiba, Japan
Ozbay, Erdogan
Iskenderun, Hatay, Turkey
Ozden, Sevket
Kocaeli Universitesi
Kocaeli, Turkey
Ozturan, Turan
Bogazici University
Istanbul, Turkey
Ozturk, Ali
Dokuz Eylul University
Izmir, Buca, Turkey
P S, Ambily
Council of Scientific & Industrial Research (CSIR)
Chennai, Tamilnadu, India
Pacheco, Alexandre
Universidade Federal do Rio Grande do Sul (UFRGS)
Porto Alegre, Rio Grande do Sul, Brazil

ACI Structural Journal/March-April 2014


Page, Adrian
University of Newcastle
Newcastle, New South Wales, Australia
Palmisano, Fabrizio
Politecnico di Bari
Bari, Italy
Pan, Wang Fook
SEGi University
Petaling Jaya, Selangor, Malaysia
Pantazopoulou, Stavroula
Demokritus University of Thrace
Xanthi, Greece
Pape, Torill
University of Newcastle
Callaghan, New South Wales, Australia
Parghi, Anant
The University of British Columbia
Kelowna, BC, Canada
Park, Honggun
Seoul National University
Seoul, Republic of Korea
Parsekian, Guilherme
Federal University of So Carlos
So Carlos, So Paulo, Brazil
Patel, Raj
CeraTech, Inc.
Baltimore, MD
Pauletta, Margherita
University of Udine
Tavagnacco, Udine, Italy
Paulotto, Carlo
Acciona S.A.
Alcobendas, Spain
Pavlikova, Milena
CTU in Prague
Prague, Czech Republic
Pellegrino, Carlo
University of Padova
Padova, Italy
Peng, Cao
Harbin Institute of Technology
Harbin, Heilongjiang, China
Peng, Jianxin
Institute of Bridge Engineering
Changsha, Hunan, China
Perez, Gustavo
Universidad Nacional de Tucumn
Yerba Buena, Tucumn, Argentina
Perez Caldentey, Alejandro
Universidad Politcnica de Madrid
Madrid, Madrid, Spain
Persson, Bertil
Bara, Sweden
Pessiki, Stephen
Lehigh University
Bethlehem, PA
Phillippi, Don
Diamond Pacific
Rancho Cucamonga, CA

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Pirayeh Gar, Shobeir
Houston, TX
Polak, Marianna
University of Waterloo
Waterloo, ON, Canada
Potter, William
Florida Department of Transportation
Tallahassee, FL
Pourazin, Khashaiar
International Institute of Earthquake Engineering

Seismology (IIEES)
Tehran, Islamic Republic of Iran
Prakash, M. N.
J.N.N. College of Engineering
Shimoga, Karnataka, India
Prasittisopin, Lapyote
Oregon State University
Corvallis, OR
Proske, Tilo
Technische Universitt Darmstadt
Darmstadt, Germany
Prota, Andrea
University of Naples
Naples, Italy
Provis, John
University of Melbourne
Victoria, Australia
Prusinski, Jan
Slag Cement Association
Sugar Land, TX
Pujol, Santiago
Berkeley, CA
Puthenpurayil Thankappan, Santhosh
Granite Construction Company
Abu Dhabi, United Arab Emirates
Putra Jaya, Ramadhansyah
Universiti Teknologi Malaysia
Skudai, Malaysia
Qasrawi, Hisham
The Hashemite University
Zarqa, Jordan
Qiu, Bin
Xian University of Architecture & Technology
Xian, China
Rafi, Muhammad
NED University of Engineering and Technology
Karachi, Sindh, Pakistan
Ragueneau, Frederic
Ecole normale suprieure de Cachan (ENS Cachan)
Cachan, France
Rahal, Khaldoun
Kuwait University
Safat, Kuwait
Rajamane, N. P.
SRM University
Kattankulathur, Tamil Nadu, India
Ramamurthy, K.
Indian Institute of Technology Madras
Chennai, Tamilnadu, India

and

477

REVIEWERS IN 2013
Ramaswamy, Ananth
Indian Institute of Science
Bangalore, Karnataka, India
Ramos, Antnio
Faculdade de Cincias e Tecnologia
Monte de Caparica, Portugal
Rangan, Vijaya
Curtin University of Technology
Perth, Western Australia, Australia
Rao, Hanchate
Jawaharlal Nehru Technological University

Engineering
Anantapur, India
Rao, Sarella
National Institute of Technology
Warangal, Andhra Pradesh, India
Raoof, Mohammed
Loughborough University
Loughborough, UK
Rautenberg, Jeffrey
Wiss, Janney, Elstner Associates, Inc.
Emeryville, CA
Ray, Indrajit
Purdue University Calumet
Hammond, IN
Razaqpur, A. Ghani
McMaster University
Hamilton, ON, Canada
Regan, Paul
Trigram
London, UK
Reiterman, Roy
Roy H. Reiterman, P.E., and Associates
Troy, MI
Ren, Xiaodan
Shanghai, China
Richardson, James
University of Alabama
Tuscaloosa, AL
Riding, Kyle
Kansas State University
Manhattan, KS
Rinaldi, Zila
University of Rome Tor Vergata
Rome, Italy
Rizk, Emad
Memorial University of Newfoundland
St. Johns, NL, Canada
Rizwan, Syed Ali
University of Engineering and Technology
Lahore, Punjab, Pakistan
Roberts, Lawrence
Grace Construction Products
Cambridge, MA
Robery, Peter
Halcrow Group Ltd.
Solihull, West Midlands, UK

478

College

of

Rodrigues, Conrado
Federal Centre for Technological Education in Minas Gerais
(CEFET-MG)
Belo Horizonte, Minas Gerais, Brazil
Rodrigues, Publio
LPE Engenharia e Consultoria
So Paulo, Brazil
Rodriguez, Mario
National University of Mexico
Mexico City, DF, Mexico
Roh, Hwasung
Chonbuk National University
Jeonju, Republic of Korea
Rosenboom, Owen
Hong Kong Polytechnic University
Hung Hom, Kowloon, Hong Kong
Rteil, Ahmad
University of British Columbia
Kelowna, BC, Canada
Russell, Henry
Henry G. Russell, Inc.
Glenview, IL
S. A., Jaffer Sathik
CSIR-Structural Engineering Research Center
Chennai, Tamilnadu, India
Saatci, Selcuk
Izmir Institute of Technology
Izmir, Turkey
Sabouni, Faisal
Architectural Consulting Group
Abu Dhabi, United Arab Emirates
Sadeghi Pouya, Homayoon
Coventry University
Coventry, UK
Saedi, Houman
Tabiat Modares University
Tehran, Islamic Republic of Iran
Safan, Mohamed
Menoufia University
Shebeen El-Koom, Menoufia, Egypt
Safi, Brahim
University of Boumerdes
Boumerdes, Algeria
Sagaseta, Juan
University of Surrey
Guildford, Surrey, UK
Sagues, Alberto
University of South Florida
Tampa, FL
Sahamitmongkol, Raktipong
CONTEC; Sirindhorn International Institute of Technology,

Thammasat University; MTEC
Pathumthani, Thailand
Sahmaran, Mustafa
Gazi University
Ankara, Turkey
Sahoo, Dipak
Cochin University of Science and Technology
Cochin, Kerala, India

ACI Structural Journal/March-April 2014


Saito, Shigehiko
University of Yamanashi
Kofu, Japan
Sajedi, Fathollah
Universiti Malaya
Kuala Lumpur, Selangor, Malaysia
Saka, Mehmet
Middle East Technical University
Ankara, Turkey
Sakai, Etsuo
Tokyo Institute of Technology
Ichikawa-shi, Japan
Saleem, Muhammad
Florida International University
Miami, FL
Salem, Hamed
Cairo University
Giza, Egypt
Sallam, Hossam El-Din
Zagazig University
Zagazig, Sharkia, Egypt
Snchez, Isidro
University of Alicante
Alicante, Alicante, Spain
Sanchez, Leandro
So Paulo, Brazil
Sant, Gaurav
University of California, Los Angeles
Los Angeles, CA
Saqan, Elias
American University in Dubai
Dubai, United Arab Emirates
Sarker, Prabir
Curtin University of Technology
Bentley, Western Australia, Australia
Sato, Ryoichi
Hiroshima University
Higashi-Hiroshima, Japan
Sato, Yuichi
Kyoto University
Kyoto, Japan
Scanlon, Andrew
The Pennsylvania State University
University Park, PA
Schindler, Anton
Auburn University
Auburn, AL
Schwetz, Paulete
Universidade Federal do Rio Grande do Sul
Porto Alegre, Rio Grande do Sul, Brazil
Semaan, Hassnaa
Ottawa Hills, OH
Sener, Siddik
Gazi University
Ankara, Turkey
Sengul, Ozkan
Istanbul Technical University
Istanbul, Turkey

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Sennour, Larbi
Consulting Engineers Group
San Antonio, TX
Serna-Ros, Pedro
Universitat Politcnica de Valncia
Valencia, Spain
Serrano, Miguel
University of Oviedo
Gijon, Spain
Shafigh, Payam
Kuala Lumpur, Malaysia
Shafiq, Nasir
Universiti Teknologi Petronas
Tronoh, Perak, Malaysia
Shah, Santosh
Dharmsinh Desai University
Nadiad, Gujarat, India
Shah, Surendra
Northwestern University
Evanston, IL
Shahnewaz, Md
University of British Columbia Okanagan
Kelowna, BC, Canada
Shannag, M. Jamal
King Saud University
Riyadh, Saudi Arabia
Shao, Yixin
McGill University
Montreal, QC, Canada
Shariq, Mohd
Civil Engineering
Aligarh, Uttar Pradesh, India
Sharma, Akanshu
Bhabha Atomic Research Centre
Mumbai, Maharashtra, India
Shayan, Ahmad
ARRB Group
Vermont South, Victoria, Australia
Shehata, Medhat
Ryerson University
Toronto, ON, Canada
Sheikh, Shamim
University of Toronto
Toronto, ON, Canada
Sherman, Matthew
Simpson Gumpertz & Heger
Melrose, MA
Sherwood, Edward
Carleton University
Ottawa, ON, Canada
Shi, Xianming
Bozeman, MT
Shi, Xudong
Tsinghua University
Beijing, China
Shield, Carol
University of Minnesota
Minneapolis, MN

479

REVIEWERS IN 2013
Shing, Pui-Shum
University of California, San Diego
La Jolla, CA
Shivali, Ram
Central Soil and Materials Research Station
New Delhi, India
Shrive, Nigel
University of Calgary
Calgary, AB, Canada
Shukla, Abhilash
Jaypee University of Information Technology
Waknaghat, Himachal Pradesh, India
Sideris, Kosmas
Democritus University of Thrace
Xanthi, Greece
Sigrist, Viktor
Hamburg University of Technology
Hamburg, Germany
Sihotang, Fransiscus
National Taiwan University of Science and Technology

(NTUST)
Taipei City, Taiwan
Silfwerbrand, Johan
Swedish Cement and Concrete Research Institute
Stockholm, Sweden
Singh, Harvinder
Guru Nanak Dev Engineering College
Ludhiana, Punjab, India
Sivey, Paul
Sivey Enterprises
Hilliard, OH
Skazlic, Marijan
University of Zagreb
Zagreb, Croatia
Smith, Scott
Southern Cross University
Lismore, New South Wales, Australia
Smyl, Danny
United States Marine Corps
Quantico, VA
Sneed, Lesley
Missouri University of Science and Technology
Rolla, MO
So, Hyoung-Seok
Seonam University
Namwon, Republic of Korea
Soejoso, Mia
Hiroshima University
Saijo, Japan
Soltani, Amir
Purdue University Calumet
Hammond, IN
Soltanzadeh, Fatemeh
Engineering and Technology
Aligarh, Uttar Pradesh, India
Sonebi, Mohammed
Queens University Belfast
Belfast, UK

480

Song, Kai
Building Materials
Dalian, China
Sossou, Gnida
Kwame Nkrumah University of Science and Technology
(KNUST)
Kumasi, Ghana
Souza, Rafael
Universidade Estadual de Maring
Maring, Paran, Brazil
Sylev, Altug
Yeditepe University
Istanbul, Turkey
Sozen, Mete
Purdue University
West Lafayette, IN
Spadea, Giuseppe
University of Calabria
Arcavacata di Rende, Cosenza, Italy
Spinella, Nino
University of Messina
Messina, Italy
Spyridis, Panagiotis
Institute for Structural Engineering
Vienna, Austria
Sreekala, R.
Structural Engineering Research Centre
Chennai, Tamilnadu, India
Stanton, John
University of Washington
Seattle, WA
Stein, Boris
Twining Laboratories
Long Beach, CA
Steuck, Kyle
University of Washington
Seattle, WA
Sudhahar, Sridevi
United Institute of Technology
Coimbatore, Tamil Nadu, India
Sujjavanich, Suvimol
Kasetsart University
Bangkok, Thailand
Sullivan, Patrick
Sullivan and Associates
Rickmansworth, UK
Sun, Shaoyun
University of Illinois at Urbana-Champaign
Urbana, IL
Ta, Binh
University of Civil Engineering
Melbourne, Victoria, Australia
Tabatabai, Habib
University of Wisconsin-Milwaukee
Milwaukee, WI
Tadayon, Mohammad Hosein
University of Tehran
Tehran, Islamic Republic of Iran
Tadayon, Mohsen
Iranian Concrete Institute
Tehran, Islamic Republic of Iran

ACI Structural Journal/March-April 2014


Tadros, Maher
e.construct.USA, LLC
Omaha, NE
Tae, Ghi ho
Leader Industrial Co.
Seoul, Republic of Korea
Taghaddos, Hosein
PCL Industrial Management Inc.
Edmonton, AB, Canada
Tahmasebinia, Faham
University of Wollongong
Wollongong, New South Wales, Australia
Takahashi, Susumu
Nagoya Institute of Technology
Nagoya, Japan
Tan, Kefeng
Southwest University of Science and Technology
Sichuan, China
Tan, Kiang Hwee
National University of Singapore
Singapore, Singapore
Tanacan, Leyla
Istanbul, Yesilkoy, Turkey
Tanesi, Jussara
Federal Highway Administration-SaLUT
Vienna, VA
Tang, Chao-Wei
Cheng-Shiu University
Niaosong District, Kaohsiung City, Taiwan
Tangtermsirikul, Somnuk
Sirindhorn International Institute of Technology
Patumthani, Thailand
Tank, Tejenadr
Pandit Deendayal Petroleum University
Gandhinagar, Gujarat, India
Tankut, Tugrul
Middle East Technical University
Ankara, Turkey
Tanner, Jennifer
University of Wyoming
Laramie, WY
Tantary, Manzoor
Indian Institute of Technology Roorkee
Roorkee, Uttarakhand, India
Tapan, Mcip
Yuzuncu Yil University
Van, Turkey
Tasdemir, Mehmet
Istanbul Technical University
Istanbul, Turkey
Tassios, Theodosios
Athens, Greece
Tastani, S.P
Demokritus University of Thrace
Xanthi, Greece
Tavares, Maria
State University of Rio de Janeiro (UERJ)
Rio de Janeiro, Brazil

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013
Tavio
Sepuluh Nopember Institute of Technology (ITS)
Surabaya, East Java, Indonesia
Tegos, Ioannis
Salonica, Greece
Tehrani, Fariborz
California State University, Fresno
Fresno, CA
Thermou, Georgia
Aristotle University of Thessaloniki
Thessaloniki, Greece
Thiagarajan, Ganesh
University of Missouri-Kansas City
Kansas City, MO
Thokchom, Suresh
Manipur Institute of Technology
Imphal, India
Thomas, Adam
Europoles GmbH & Co.
Neumarkt, Germany
Thompson, Phillip
Palm Desert, CA
Thorne, A.
Center of Engineering Materials and Structures
Guilford, Surrey, UK
Tian, Ying
University of Nevada, Las Vegas
Las Vegas, NV
Tiberti, Giuseppe
University of Brescia
Brescia, Italy
Tito, Jorge
University of Houston-Downtown
Houston, TX
Tixier, Raphael
Western Technologies Inc.
Phoenix, AZ
Tjhin, Tjen
Buckland & Taylor Ltd
North Vancouver, BC, Canada
Tobolski, Matthew
San Diego, CA
Tokgoz, Serkan
Mersin University
Mersin, Turkey
Tolentino, Evandro
Centro Federal de Educao Tecnolgica de Minas Gerais
Timteo, Minas Gerais, Brazil
Topu, Ilker
Eskisehir Osmangazi University
Eskisehir, Turkey
Torrenti, Jean-Michel
Chevilly Larue, France
Torres-Acosta, Andres
Universidad Marista de Quertaro
Quertaro, Mexico
Tosun, Kamile
Dokuz Eyll University
Izmir, Turkey

481

REVIEWERS IN 2013
Toufigh, Vahab
University of Arizona
Tucson, AZ
Trautwein, Leandro
Federal University of ABC
So Paulo, Brazil
Triantafillou, Thanasis
University of Patras
Patras, Greece
Tsonos, Alexander
Aristotle University of Thessaloniki
Thessaloniki, Greece
Tsubaki, Tatsuya
Yokohama National University
Yokohama, Japan
Tuchscherer, Robin
Northern Arizona University
Flagstaff, AZ
Tureyen, Ahmet
Wiss, Janney, Elstner Associates, Inc.
Birmingham, MI
Turgut, Paki
Harran University
Sanliurfa, Turkey
Turk, A. Murat
Istanbul Kultur University
Istanbul, Turkey
Tutikian, Bernardo
Unisinos
Porto Alegre, Rio Grande do Sul, Brazil
Unterweger, Andreas
Institute for Structural Engineering
Vienna, Austria
Uygunoglu, Tayfun
Afyon Kocatepe University
Afyonkarahisar, Turkey
Vakilly, Sedigheh
Isfahan, Islamic Republic of Iran
Varum, Humberto
University of Aveiro
Aveiro, Portugal
Vaz Rodrigues, Rui
cole Polytechnique Fdrale de Lausanne (EPFL)
Lausanne, VD, Switzerland
Vazquez-Herrero, Cristina
Universidade da Corua
La Corua, Spain
Veen, Cornelis
Delft University of Technology
Delft, the Netherlands
Velzquez Rodrguez, Sergio
Universidad Panamericana
Zapopan, Jalisco, Mexico
Velu, Saraswathy
Central Electro Chemical Research Institute
Karaikudi, Tamil Nadu, India
Vichit-Vadakan, Wilasa
CTLGroup
Skokie, IL

482

Vimonsatit, Vanissorn
Curtin University
Perth, Western Australia, Australia
Vintzileou, Elizabeth
National Technical University of Athens
Athens, Greece
Vitaliani, Renato
University of Padua
Padua, Italy
Viviani, Marco
Haute Ecole dIngnierie et de Gestion du Canton de Vaud
(HEIG-VD)
Yverdon les Bains, Switzerland
Vogel, Thomas
Institute of Structural Engineering
Zurich, Switzerland
Vollum, Robert
Imperial College London
London, UK
Volz, Jeffery
The Pennsylvania State University
University Park, PA
Vosooghi, Ashkan
AECOM
Sacramento, CA
Wagh, Prabhanjan
College of Engineering, Pune
Satara, Maharashtra, India
Waldron, Christopher
University of Alabama at Birmingham
Birmingham, AL
Wan, David
Old Castle Precast, Inc.
South Bethlehem, NY
Wang, Chang-Qing
Tongji University
Shanghai, China
Wang, Chong
Brisbane, Queensland, Australia
Wang, Huanzi
San Jose, CA
Wang, Junyan
National University of Singapore
Singapore
Wang, Kejin
Iowa State University
Ames, IA
Wehbe, Nadim
South Dakota State University
Brookings, SD
Wei, Ya
University of Michigan
Ann Arbor, MI
Wei-Jian, Yi
Changsha, China
Weiss, Charles
U.S. Army Corps of Engineers Engineer Research and

Development Center
Vicksburg, MS

ACI Structural Journal/March-April 2014

REVIEWERS IN 2013

Weiss, Jason
Purdue University
West Lafayette, IN
Wen, Ziyun
South China University of Technology
Guangzhou, Guangdong, China
Werner, Anne
Southern Illinois University Edwardsville
Edwardsville, IL
Weyers, Richard
Blacksburg, VA
Wheat, Harovel
University of Texas at Austin
Austin, TX
Wheeler, Andrew
University Of Western Sydney
Sydney, New South Wales, Australia
Wilson, William
Universit de Sherbrooke
Sherbrooke, QC, Canada
Windisch, Andor
Karlsfeld, Germany
Won, Moon
Texas Tech University
Lubbock, TX
Wong, Hong
Imperial College London
London, UK
Wong, Sook-Fun
Nanyang Technological University
Singapore
Wood, Richard
University of California, San Diego
La Jolla, CA
Woyciechowski, Piotr
Warsaw University of Technology
Warsaw, Poland
Wu, Chenglin
Missouri University of Science and Technology
Rolla, MO
Wu, Hui
Beijing, China
Wu, Hwai-Chung
Wayne State University
Detroit, MI
Xia, Zuming
Grand Prairie, TX
Xiang, Tianyu
Chengdu, Sichuan, China
Xiangguo, Wu
Harbin Institute of Technology
Harbin, Heilongjiang, China
Xiao, Feipeng
Clemson University
Clemson, SC
Xiao, Yan
Hunan University
Changsha, Hunan, China
Xingyi, Zhu
Hangzhou, China

Xin-hua, Cai
Wuhan University
Wuhan, HuBei, China
Xu, Aimin
ARRB Group
Melbourne, Victoria, Australia
Xuan, D.X.
Delft University of Technology
Delft, the Netherlands
Yahia, Ammar
Universit de Sherbrooke
Sherbrooke, QC, Canada
Yakoub, Haisam
Ottawa, ON, Canada
Yan, Libo
University of Auckland
Auckland, New Zealand
Yang, KuoChen
National Kaohsiung First University of Science and Technology
Kaohsiung, Taiwan
Yang, Xinbao
Olathe, KS
Yang, Yanan
Pitt & Sherry
South Melbourne, Victoria, Australia
Yang, Zhifu
Middle Tennessee State University
Murfreesboro, TN
Yassein, Mohamed
Doha, Qatar
Yatagan, Serkan
Istanbul Technical University
Istanbul, Turkey
Yazc, Semsi
Ege University
Izmir, Turkey
Yehia, Sherif
American University of Sharjah
Sharjah, United Arab Emirates
Yen, Peter
Bechtel National, Inc.
San Francisco, CA
Yerramala, Amarnath
Dundee University
Dundee, Scotland, UK
Yigiter, Huseyin
Dokuz Eyll University
Izmir, Turkey
Yildirim, Hakki
Istanbul, Turkey
Yoon, Young-Soo
Korea University
Seoul, Republic of Korea
Yost, Joseph
Villanova University
Villanova, PA
Youkhanna, Kanaan
University of Dohuk
Duhok, Iraq

ACI Structural Journal/March-April 2014

483

REVIEWERS IN 2013
Young-sun, Kim
Tokyo University of Science
Noda-Shi, Chiba, Japan
Youssef, Maged
University of Western Ontario
London, ON, Canada
Yu, Baolin
Michigan State University
East Lansing, MI
Yuan, Jiqiu
PSI; Federal Highway Administration Turner-Fairbank Highway
Research Center
McLean, VA
Yksel, Isa
Bursa Technical University
Bursa, Turkey
Zahedi, Farshad
Babol Noshirvani University of Technology
Babol, Mazandaran, Islamic Republic of Iran
Zaki, Adel
SNC-Lavalin
Montreal, QC, Canada
Zanuy, Carlos
Universidad Politcnica de Madrid
Madrid, Spain
Zatar, Wael
West Virginia University Institute of Technology
Montgomery, WV
Zerbino, Raul
La Plata, Argentina
Zeris, Christos
National Technical University of Athens
Zografou, Greece
Zhang, Jieying
National Research Council Canada
Ottawa, ON, Canada

484

Zhang, Jun
Tsinghua University
Beijing, China
Zhang, Peng
Karlsruhe Institute of Technology (KIT)
Karlsruhe, Germany
Zhang, Wei Ping
Tongji University
Shanghai, China
Zhang, Xiaogang
Shenzhen University
Shenzhen, Guangdong, China
Zhang, Xiaoxin
Universidad de Castilla-La Mancha
Ciudad Real, Spain
Zhang, Yamei
Southeast University
Nanjing, China
Zheng, Herbert
Gammon Construction Limited
Hong Kong
Zheng, Jianjun
Zhejiang University of Technology
Hangzhou, China
Zhou, Wei
Harbin Institute of Technology
Harbin, China
Zhu, Han
TianJin University
TianJin, China
Ziehl, Paul
University of South Carolina
Columbia, SC
Zilch, Konrad
Technische Universitt Mnchen
Munich, Germany

ACI Structural Journal/March-April 2014

2014 ACI Membership Application


American Concrete Institute 38800 Country Club Drive Farmington Hills, MI 48331 USA
Phone: (248) 848-3800 Fax: (248) 848-3801 Web: www.concrete.org

Please print or type all information requested below:


Mr. Mrs. Ms. Dr.

Gender: Female Male

FIRST NAME ______________________________________ MIDDLE INITIAL ___________ LAST NAME (SURNAME) ______________________________________________________

EMPLOYER _________________________________________________________________ CORPORATE TITLE __________________________________________________________

E-MAIL ADDRESS ____________________________________________________________ BIRTHDATE _________________________________________________________________

ADDRESS _______________________________________________________________________________________________________________________________________________

CITY _____________________________________________ STATE ___________________ ZIP __________________ COUNTRY ___________________________________________

TELEPHONE ________________________________________________________________ FAX ________________________________________________________________________

Categories of membership

Please select the desired category of membership and submit the appropriate dues described below.

ORGANIZATIONAL $1007 PLUS APPLICABLE SHIPPING FEES


Includes one set of the Manual of Concrete Practice (MCP), a wall plaque, and a subscription to
Concrete International, Concrete Repair Bulletin, and both ACI Materials and Structural Journals.
MANUAL OF CONCRETE PRACTICE: Select a format and include applicable shipping fee:
Seven Volume Set U.S. add $30 shipping fee; Canada and Mexico add $130 shipping fee;
and all others add $220 shipping fee.
CD-ROM no additional shipping costs
Online Subscription no additional shipping costs
INDIVIDUAL $226/year INDICATE SUBSCRIPTION PREFERENCE
Individuals 28 years old or above residing worldwide.
YOUNG PROFESSIONAL $126/year INDICATE SUBSCRIPTION PREFERENCE
Individuals under the age of 28 who do not qualify for student membership.
E-STUDENT FREE Join at www.students.concrete.org.
STUDENT (U.S. AND CANADA) $41/year INDICATE SUBSCRIPTION PREFERENCE
STUDENT (Outside U.S. and Canada) $81/year INDICATE SUBSCRIPTION PREFERENCE
Individuals under the age of 28 who are registered full-time students at an educational institution.
Full-time students age 28 and above may be granted Student Membership when the request is
endorsed by the students advisor.
SUSTAINING MEMBERSHIP ACI Sustaining Members receive all membership benefits of
Organizational Members plus a free copy of every new ACI publication and increased corporate
exposure, positioning them as a leader in the concrete industry, and much more. For complete details
or to join, visit www.concrete.org/sustainingmembership or call (248) 848-3800.

Profile Information (Select all that apply.)


Markets
Occupations

Design
Materials
Production
Construction
Testing
Repair
Owner

Management
Consultant
Engineer
Architect
Contractor
Technical Specialist
Quality Control

Inspector
Craftsman
Sales & Marketing
Association
Employee
Government
Employee

Your ACI membership includes a hard copy and


online subscription to Concrete International, and
your choice of one of the following:
ACI Materials Journal, hard copy
ACI Materials Journal, online
ACI Structural Journal, hard copy
ACI Structural Journal, online
Concrete Repair Bulletin, hard copy
Please allow up to two months for delivery outside the U.S.
To add subscriptions: please mark above and include an
additional $70 ($75 for individuals outside U.S. & Canada,
$41 for students outside U.S. & Canada, and $34 for U.S.
& Canada students) for hard copy subscriptions or $29 for
online subscriptions.

Optional Online Subscriptions


Bundle your ACI membership with these optional
online subscriptions for additional ACI resources:
Symposium Papers Subscription, $97
Manual of Concrete Practice, $409

Researcher
Educator
Student
Other __________

Payment

Fees

Included Subscriptions

CHECK NUMBER

Membership dues

CHECK payable to American Concrete Institute

Additional subscriptions

CREDIT CARD Visa MasterCard American Express

Organizational seven volume set


shipping fees

For Journals, Canadian residents


add 5% GST (#126213149RT)

ACCOUNT NUMBER

EXPIRATION DATE

NAME (AS APPEARS ON CREDIT CARD)

TOTAL (U.S. Funds Only)


SIGNATURE

Code: CI

74 JANUARY 2014 Concrete international


ACI Structural Journal/March-April 2014485

NOTES:

486

ACI Structural Journal/March-April 2014

CALL FOR ACTION


ACI Invites You To...

Do you have EXPERTISE in any of these areas?


BIM
Chimneys
Circular Concrete Structures Prestressed by Wrapping
with Wire and Strand
Circular Concrete Structures Prestressed with
Circumferential Tendons
Concrete Properties
Demolition
Deterioration of Concrete in Hydraulic Structures
Electronic Data Exchange
Insulating Concrete Forms, Design, and Construction
Nuclear Reactors, Concrete Components
Pedestal Water Towers
Pipe, Cast-in-Place
Strengthening of Concrete Members
Sustainability

Then become a REVIEWER for the


ACI Structural Journal or the ACI Materials Journal.
How to become a Reviewer:

1. Go to: http://mc.manuscriptcentral.com/aci;
2. Click on Create Account in the upper right-hand corner; and
3. Enter your E-mail/Name, Address, User ID and Password, and
Area(s) of Expertise.

Did you know that the database for MANUSCRIPT


CENTRAL, our manuscript submission program,
is separate from the ACI membership database?
How to update your user account:
1. Go to http://mc.manuscriptcentral.com/aci;
2. Log in with your current User ID & Password; and
3. Update your E-mail/Name, Address, User ID and Password,
and Area(s) of Expertise.

QUESTIONS?

E-mail any questions to Journals.Manuscripts@concrete.org.

One Click... One Entire Journal


NOTES:

Introducing

a new feature on ACIs website!

Now, with one click of the mouse, subscribers of the


ACI Structural and Materials Journals digital editions can
download an Adobe Acrobat PDF of an unabridged
issue at www.concrete.org/Publications/
ACIStructuralJournal.aspx and www.concrete.org/
Publications/ACIMaterialsJournal.aspx.
488

ACI Structural Journal/March-April 2014

ACI
STRUCTURAL
J O U R N A L
J O U R N

This journal and a companion periodical,


ACI Materials Journal, continue the publishing
tradition the Institute started in 1904.
Information published in ACI Materials Journal
includes: properties of materials used in
concrete; research on materials and concrete;
properties, use, and handling of concrete; and
related ACI standards and committee reports.

You might also like