SJ March 2022 V. 119 No. 2

V. 119, NO.
2
MARCH 2022
ACI
STRUCTURAL J O U R N A L
A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE

CONTENTS
STATEMENT OF OWNERSHIP
Editorial Board
Robert J. Frosch, Editor-in-Chief
ACI Structural Journal
Purdue University
Catherine French MARCH 2022, V. 119, No. 2
University of Minnesota
Michael Kreger a journal of the american concrete institute
University of Alabama an international technical society
David Sanders
Iowa State University
James K. Wight 3 Effect of Noncontact Lap Splices in Reinforced Concrete Beams, by
University of Michigan Hyeon-Jong Hwang, Fan Yang, and Gao Ma
Board of Direction 19 Maximum Shear Strength of Reinforced Concrete Beams, by Shyh-

President
Jiann Hwang, Yu-Hsuan Yang, and Yi-An Li
Cary S. Kopczynski
31 Long-Term Behavior of Glass Fiber-Reinforced Polymer-Reinforced
Vice Presidents Columns under Multi-Axial Loading, by Yail J. Kim
Antonio Nanni
Charles K. Nmai 45
In-Plane Seismic Response of Autoclaved Aerated Concrete
Block Masonry-Infilled Reinforced Concrete Frame Building, by
Directors
Scott M. Anderson Nikhil P. Zade, Avadhoot Bhosale, Pradip Sarkar, and Robin Davis
Michael C. Brown
Anthony R. DeCarlo Jr. 61 Simplified Strut-and-Tie Model for Shear Strength Prediction of Rein-
Walter H. Flood IV forced Concrete Low-Rise Walls, by Jimmy Chandra and Susanto Teng
John W. Gajda
G. Terry Harris 75 Modeling of Alkali-Silica Reaction-Affected Shear-Critical Reinforced
Maria G. Juenger
Kamal H. Khayat Concrete Structures, by Anca C. Ferche and Frank J. Vecchio
Michael E. Kreger
Kimberly E. Kurtis 89 Nonlinear Modeling Parameters of Reinforced Concrete Coupling
Ishita Manjrekar Beams, by Tae-Sung Eom, Seung-Jae Lee, Su-Min Kang, and
W. Jason Weiss Hong-Gun Park
Past President Board Members
Jeffrey W. Coleman 103 Strength Reduction Factors for ACI 318 Strut-and-Tie Method for
David A. Lange Deep Beams, by Victor Aguilar, Robert W. Barnes, and Andrzej Nowak
Randall W. Poston
113 Reliability Assessment and Strength Reduction Factor Calibra-
Executive Vice President tion for Screw Anchors Concrete Breakout, by Ahmad N. Tarawneh,
Ron Burg Eman F. Saleh, and Sereen A. Majdalaweyh
Staff 123 Fatigue Performance of Recycled Aggregate Concrete Beams with
Publisher Corroded Steel Reinforcement, by Guoqing Dong, Jin Wu, and Xing Zhao
John C. Glumb
Managing Director, Engineering and 139 Local Bond-Slip Behavior of Reinforcing Bars in High-Performance
Professional Development Steel Fiber-Reinforced Concrete Beams, by Rita-Elizabeth Saikali,
Michael L. Tholen S. J. Pantazopoulou, and D. Palermo
Engineers 155 Uncertainty Modeling and Analysis of Punching Shear Resistance of

Katie A. Amelio Flat Slabs, by Ming-Yue Jiang, Qing-Xuan Shi, and Wang-Hu Zhao
H. R. Trey Hamilton
Robert M. Howell
Khaled Nahlawi
165 Vehicle Collision with Reinforced Concrete Columns Wrapped with
Matthew R. Senecal Fiber-Reinforced Polymer Composites, by Tao Liu, Lin Chen, Jinjun Xu,
Sureka Sumanasooriya Cristoforo Demartino, and Thomas H.-K. Kang
Gregory M. Zeisler
Jerzy Z. Zemajtis
Managing Editor
Lauren E. Mentz
Associate Editor Contents continued on next page
Kelly Dudley
Discussion is welcomed for all materials published in this issue and will appear ten months from
this journal’s date if the discussion is received within four months of the paper’s print publication.
Editors Discussion of material received after specified dates will be considered individually for publication or
Erin N. Azzopardi private response. ACI Standards published in ACI Journals for public comment have discussion due
Kaitlyn J. Dobberteen dates printed with the Standard.
Tiesha Elam ACI Structural Journal
Hannah E. Genig Copyright © 2022 American Concrete Institute. Printed in the United States of America.
Angela R. Matthews The ACI Structural Journal (ISSN 0889-3241) is published bimonthly by the American Concrete Institute. Publication
Kelli R. Slayden office: 38800 Country Club Drive, Farmington Hills, MI 48331. Periodicals postage paid at Farmington, MI, and at
additional mailing offices. Subscription rates: $189 per year, 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: +1.248.848.3700.
Facsimile (FAX): +1.248.848.3701. Website: http://www.concrete.org.
ACI Structural Journal/March 20221

Contributions to
CONTENTS ACI Structural Journal
The ACI Structural Journal is an open
181 Effect of Salt-Frost Cycles and Impact Loads on Residual Axial Capacity forum on concrete technology and papers
of Reinforced Concrete Column, by Guoqing Dong, Jin Wu, and Xing Zhao related to this field are always welcome.
All material submitted for possible publi-
195 Nonlinear Modeling of Reinforced Concrete Columns under Cyclic cation must meet the requirements of
Loading, by Tae-Sung Eom, Seung-Jae Lee, Chul-Goo Kim, and the “American Concrete Institute Publi-
Hong-Gun Park cation Policy” and “Author Guidelines
and Submission Procedures.” Prospective
209 Shear Strength of Flanged Squat Walls with 690 MPa Reinforcing Bars, authors should request a copy of the Policy
by Ju-Hyung Kim and Hong-Gun Park and Guidelines from ACI or visit ACI’s
website at www.concrete.org prior to
submitting contributions.
221 Side-Face Blowout Strength of Two-Layer Headed Bars Embedded in
Exterior Beam-Column Joints, by Hye-Jung Sim and Sung-Chul Chun Papers reporting research must include
a statement indicating the significance of
233 Cyclic Behavior of Ultra-High-Performance Concrete Shear Walls with the research.
Different Axial-Load Ratios, by Rui Hu, Zhi Fang, and Baodan Xu The Institute reserves the right to return,
without review, contributions not meeting
247 Unified Strength Model for Spiral-Steel Confined Concrete Columns, by the requirements of the Publication Policy.
Muhammad Junaid Munir, Syed Minhaj Saleem Kazmi, and Yu-Fei Wu All materials conforming to the Policy
requirements will be reviewed for editorial
257 Experiments on Partially Coupled Concrete Wall System with Perfo- quality and technical content, and every
rated Steel Beam, by Woo-Young Lim, Thomas H.-K. Kang, Donghyuk Jung, effort will be made to put all acceptable
and Sung-Gul Hong papers into the information channel.
However, potentially good papers may be
271 Glass Fiber-Reinforced Polymer-Reinforced Beam-Column Connec- returned to authors when it is not possible
to publish them in a reasonable time.
tions with Novel Connection Details, by Muhammad Safdar,
M. Neaz Sheikh, and Muhammad N. S. Hadi
Discussion
287 Effect of Slenderness Ratio on Glass Fiber-Reinforced Polymer- All technical material appearing in the
Reinforced High-Strength Concrete Columns, by Mu’taz Almomani, ACI Structural Journal may be discussed.
Karam Mahmoud, and Ehab F. El-Salakawy If the deadline indicated on the contents
page is observed, discussion can appear
in the designated issue. Discussion should
301
Moment-Curvature-Deformation Response of Post-Tensioned be complete and ready for publication,
Ultra-High-Performance Concrete Beams, by Mehmet Dogu and including finished, reproducible illustra-
Fatmir Menkulasi tions. Discussion must be confined to the
scope of the paper and meet the ACI Publi-
317 A Multi-Gene Genetic Programming Model for Predicting Shear Strength cation Policy.
of Steel Fiber Concrete Beams, by Mohamed K. Ismail, Ahmed Yosri, and Follow the style of the current issue.
Wael El-Dakhakhni Be brief—1800 words of double spaced,
typewritten copy, including illustrations
329 2021 Reviewers List and tables, is maximum. Count illustra-
tions 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 refer-
ences 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 discus-
sions. Closures are published together
with the discussions.
Discuss the paper, not some new or
outside work on the same subject. Use
ON COVER: 119-S37, p. 170, Fig. 6—Damage comparison of RC pier between references wherever possible instead of
simulation result and post-accident result. repeating available information.
Discussion offered for publication should
offer some benefit to the general reader.
Discussion which does not meet this
Permission is granted by the American Concrete Institute for libraries and other users registered with the Copyright Clearance requirement will be returned or referred to
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 the author for private reply.
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. Send manuscripts to:
The Institute is not responsible for statements or opinions expressed in its publications. Institute publications are not able to, http://mc.manuscriptcentral.com/aci
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 Institute’s Publication Policy by individuals Send discussions to:
expert in the subject area of the papers.
Journals.Manuscripts@concrete.org
2 ACI Structural Journal/March 2022

ACI STRUCTURAL JOURNAL TECHNICAL PAPER
Title No. 119-S25
Effect of Noncontact Lap Splices in Reinforced Concrete

Beams
by Hyeon-Jong Hwang, Fan Yang, and Gao Ma
Current design codes specify the bar development length empiri-

cally based on contact lap splice test results. The bar splice length
of noncontact lap splices is not clearly specified due to limited test
results. In the present study, the bond performance of noncontact
lap splices was investigated. Lap splice tests were performed on
24 beam specimens under uniformly distributed moment or nonuni-
formly distributed moment with shear force. The test parameters
were spacing of spliced bars, reinforcing bar diameter, transverse
reinforcement, and moment distribution. The test results showed
that the tensile stress of bar splices increased as the spacing of
spliced bars increased in the specimens with stirrups. In the spec-
imens under nonuniformly distributed moment with shear force,
the bar splices exhibited better bond strength. The test results were
compared with the predictions of current design codes.
Keywords: lap splice; moment distribution; noncontact lap splice; rein-

forced concrete beam; reinforcing bar stress. Fig. 1—Effective lap splice length.
and spaced splices was insignificant. Chinn et al.3 evaluated
INTRODUCTION
the bond strength of noncontact splice bars with a spacing
In reinforced concrete (RC) structures, lap splices are
of 19.0 to 25.4 mm (0.8 to 1.0 in.) in beams without stir-
necessary due to the limited length of reinforcing bars.
rups. The lap splice performance of noncontact splice bars
The stress of the lap slice bars is transferred by bond stress
was comparable to that of contact splices. Goto and Otsuka4
between the concrete and reinforcing bars. The bar bond
reported that diagonal cracks developed between the
stress is mainly affected by cover concrete thickness, bar
noncontact spliced bars, and the angle of cracks was affected
spacing, transverse reinforcement, and material properties
by the lap splice length and bar spacing. According to Sagan
of the reinforcing bars and concrete.
et al.,5 the diagonal cracks decreased the effective lap splice
Regarding the spacing between lap splice bars, lap splices
length and concrete tensile strength in noncontact splices
can be divided into contact lap splices and noncontact lap
(Fig. 1). However, the inner concrete between noncon-
splices. Contact lap splices have been widely used in cast-
tact spliced bars provided the reinforcing bar confinement.
in-place reinforced concrete structures, and the majority of
Sagan et al.5 tested 47 full-scale flat-plate specimens to study
existing studies on lap splices focused on contact lap splices.
the behavior of noncontact lap splices subjected to mono-
In special situations, such as the connection of precast
tonic and repeated inelastic loading. The test results showed
members, noncontact lap splices are used in consideration
that the ultimate strength was not affected by the splice
of construction error. However, limited studies on noncon-
bar spacing up to at least six times the bar diameter under
tact lap splices have been performed. The bond performance
monotonic loading, but noncontact splices exhibited better
of noncontact splice bars under various design conditions
cyclic load resistance compared to contact splices. Hamad
needs to be evaluated.
and Mansour6 tested 17 slabs to evaluate the validity of ACI
Several studies have been performed to evaluate the bond
318-897 provisions for noncontact splices, and reported that
stress and development length of reinforcing bars. In the
the bar bond strength of the noncontact splices was greater
early studies, the discrepancy of bond strength between
than that of the contact splices when the clear spacing of
contact and noncontact lap splices was not revealed within
spliced bars increased to 5db. However, when the clear
the limited scope of tests. Chamberlin1,2 investigated the
spacing was larger than 7db, the bar bond strength of the
effect of spacing of spliced bars through pullout tests and
noncontact splices was less than that of the contact splices.
beam tests. The pullout test results showed that adjacent tied
splices developed better average bond stress than that of
ACI Structural Journal, V. 119, No. 2, March 2022.
spaced splices, but the discrepancy of bond strength between MS No. S-2020-083.R2, doi: 10.14359/51734374, received July 27, 2020, and
reviewed under Institute publication policies. Copyright © 2022, American Concrete
a clear spacing of 1db and 3db (where db is reinforcing bar Institute. All rights reserved, including the making of copies unless permission is
diameter) was insignificant. On the other hand, the beam test obtained from the copyright proprietors. Pertinent discussion including author’s
closure, if any, will be published ten months from this journal’s date if the discussion
results showed that the bond strength between the adjacent is received within four months of the paper’s print publication.
ACI Structural Journal/March 2022 3

Kilpatrick and Gilbert8 reported that the noncontact splices the validity of the current design codes, lap splice tests were
with small spaced distance exhibited a bond strength similar performed on 24 RC beams with various design parameters.
to that of contact splices in slabs. Nevertheless, as the splice
bar spacing increased, the contact splice showed better bond CURRENT DESIGN CODES
strength than that of the noncontact splices. In general, Table 1 shows the tension bar development length speci-
although the existing studies provided valuable test results fied in current design codes and methods. Figure 2 shows the
and analysis for the noncontact splices, each study reported definition of cover concrete thickness and bar spacing cb, cso,
conflicting results due to the different lap splice conditions and csi in ACI 318-1910 and ACI 408R-03.11 When the bar
in their studies: bond strength discrepancy between the development length or lap splice length is calculated, ACI
noncontact and contact lap splices was insignificant in the 318-1910 and Eurocode 212 specify the allowable maximum
noncontact lap splice with small clear spacing of spliced bars clear spacing between spliced bars. In ACI 318-19,10 the
(approximately 12.7 to 25.4 mm [0.5 to 1.0 in.])2,3; and the center-to-center spacing of spliced bars is limited to the
bond strength of noncontact lap spliced bars increased and lesser of one-fifth of the required lap splice length and
then decreased as the bar spacing increased from 30 to 150 152 mm (6.0 in.). Such limitation is to avoid the formation
mm (1.2 to 5.9 in.).6 Further, the number of noncontact lap of an unreinforced section leading to potential cracks. In
splice test specimens is still limited compared to contact lap Eurocode 2,12 the clear spacing of spliced bars should not
splice test results, which cannot reveal the effect of design be greater than 4db or 50 mm (1.97 in.). Otherwise, the bar
parameters on the bond strength of noncontact lap splices. development length should be increased by a length equal to
Particularly, the majority of the existing test specimens used the clear spacing where it exceeds 4db or 50 mm (1.97 in.).
the slab and beam without stirrups. Thus, further studies are fib Model Code 201013 and GB 50010-201014 do not specify
needed to accurately investigate the bond performance of the bar spacing requirements of noncontact bar splices. ACI
noncontact splices under various design parameters. 408R-0311 and the Hwang et al. model15 specify the same
In the present study, lap splice tests were performed to requirements as ACI 318-19.10 Although the specific provi-
investigate the bond strength of noncontact lapped bars in RC sions for noncontact lap splices are not prescribed in the
beams. The test parameters were reinforcing bar diameter, current design codes, only Eurocode 212 requires an extra
transverse reinforcement, clear distance between the spliced lap splice length increment for noncontact lap splices with
bars, and moment distribution. Addressing the limited inves- large spacing between reinforcing bars.
tigation of the design parameter effects, the use of trans-
verse reinforcement and various clear distances between the TEST PLAN
spliced bars of 36 to 100 mm (1.4 to 3.9 in.) were consid- Specimens
ered. Further, the effect of moment distribution on the bond Table 2 shows the design properties of 24 RC beam spec-
strength of noncontact lap splices was investigated because imens (that is, 12 beams for four-point loading tests, and
the linear moment decreased the bond requirement in lap 12 beams for three-point loading tests). Figure 3 shows the
splices.9 The structural performance, including the load- configuration and reinforcement details of the test speci-
carrying capacity, bond strength, reinforcing bar strain, and mens. The cross-section dimensions of the of the beams
failure mode, was evaluated. The current design codes were were 450 x 450 mm (17.7 x 17.7 in.) to ensure the beam
applied to the test results, including the existing test results. width for widely separated noncontact lap splices. The
The effect of noncontact bar spacing and moment distribu- length of the beams for four-point and three-point loading
tion on the bond strength of spliced bars were discussed. tests were 4000 and 2500 mm (157.5 and 98.4 in.), respec-
tively. Two D18 bars (diameter = 18 mm [0.71 in.] and
RESEARCH SIGNIFICANCE cross-sectional area = 254.5 mm2 [0.39 in.2]) or two D25
With the continuous development of precast structures bars (diameter = 25 mm [0.98 in.] and cross-sectional area =
and construction diversity, noncontact lap splices are being 491.0 mm2 [0.76 in.2]) were used for beam longitudinal bars,
widely used in the connection of structural members (for which were lap-spliced at the center of the beam. The lap
example, precast concrete members, prefabricated beam- splice length was 300 mm (11.81 in.), which was equiva-
column joints, or particular conditions where contact lap lent to 20 to 48% and 27 to 73% of the required bar splice
splices are unable to be applied). Further, construction errors lengths of ACI 318-1910 and ACI 408R-0311 calculated
cause the unintentional application of noncontact lap splices. using the actual material strength, respectively, to evaluate
Compared to contact splices, the stress-transfer mechanism the effect of test parameters on the bond strength without
between reinforcing bars is different in noncontact lap reinforcing bar yielding. The test parameters were the rein-
splices because of the distance between the spliced bars. forcing bar diameter (db = 18.0 mm [0.71 in.] or 25.0 mm
However, the available experimental range and collected [0.98 in.]), clear spacing of spliced bars (that is, contact
data of noncontact lap splices are limited. On the other hand, bars or noncontact bars), and stirrups within the lap splice
current design codes and methods rarely consider the influ- length. For concrete cover, all specimens used a fixed value
ence of noncontact splices or just limit the spacing range of of cb and cso = 33 mm (1.3 in.). The value of csi depends
lapped bars. The design methods of contact lap splices are on the reinforcing bar diameter and spacing of spliced bars,
directly applied to the design of noncontact lap splices. To ranging from 42 to 156 mm (1.65 to 6.14 in.). In four-
investigate the reliability of noncontact lap splices and verify point loading test specimens, the only constant moment is
developed in the lap splice region. To investigate the effects

Table 1—Current design codes and methods for Table 1 (cont.)—Current design codes and
tension development length methods for tension development length
(when the total area of spliced bars is greater than two times the required and less
Design than half of bars are spliced within the required lap splice length, Class A, otherwise,
codes and Development length Class B).
methods (MPa and mm) †
c = cmin + db/2; cmax = max(cb, cs); cmin = min(cb, cs); cs = min(cso, csi+ 6.4); and ϕ =
0.82 is the safety factor for structural design. In ACI 408R-03, ls = ld.
f y db t e s g
ld 300 mm ‡
α2α3 ≥ 0.7; cd = min(cb,cso,csi); l0 = max(0.45fydb/4fbd, 15db, 200 mm); K is coeffi-
1.1 f c c f K tr / db cient for arrangement of transverse bars (0 to 0.1); Atr is total cross-sectional area of
ACI
318-1910* transverse bars within the development length; As is maximum cross-sectional area
(cf + Ktr)/db ≤ 2.5
of the bar; and η2 = (132 – db)/100 is coefficient for bar diameter (≤1.0). In Eurocode
cf = min(cb, cso, csi) + 0.5db
2, ls varies between 1.0ld and 1.5ld according to the area ratio of spliced bars within
Ktr = 40Atr/(stns) 0.65ld, and a partial safety factor is applied to concrete for structural design.
f lm0 = max(0.7fydb)/4fbd, 15db, 200 mm; kd is coefficient for arrangement of transverse

§
y / 4 f c 57.4 w t e s db bars (0 to 20); αt is coefficient for transverse bar diameter (0.5 for D25 to 1.0 for
ld
1.83 cw K atr / db D50); cm = max(cso, csi); η3 = (25/db)0.3 is coefficient for bar diameter (≤1.0); and η4
ACI is coefficient for bar yield strength (0.68 to 1.2). In fib Model Code 2010, ls varies
(cw + Katr)/db ≤ 4.0 between 0.7ld and 1.0ld according to the area ratio and tensile stress of spliced bars
408R-0311† w = 0.1(cmax/cmin) + 0.9 ≤ 1.25 within the lap splice length, and a partial safety factor is applied to concrete for
K atr 6 f ctd Atr / st ns structural design.
ζl is coefficient for lap splice ratio within 1.3ls (1.2 to 1.6); ζa is coefficient for
||
td = 0.03db + 0.22
reinforcing bar properties (=1.0 for prestressed steel); γ1 = 1.10 for ribbed bars with
reinforcing bar diameter above 25 mm; γ2 = 1.25 for epoxy-coated steel bars; γ3 =
f y db l0
ld 2 3 1.10 for bars that are easily disturbed during construction; γ4 = coefficient for the
4 f bd 1.5 ratio of design longitudinal reinforcement area to the actual longitudinal reinforce-
Eurocode
ment area (1.0 for structures with seismic requirements or directly under dynamic
212‡ α2 = 0.7 ≤ 1 – 0.15(cd – db)/db ≤ 1.0
loads); γ5 is coefficient for cover concrete thickness (0.80 for 3db to 0.7 for 5db); α is
α3 = 0.7 ≤ 1 – K(∑Atr – As)/As ≤ 1.0 coefficient for reinforcing bar geometry (0.16 for round bars and 0.14 for deformed
fbd = 2.25η2[0.7(0.3)(fc′)2/3] bars); and fct is specified concrete tensile strength (0.91 MPa for C15 concrete to
2.22 MPa for C80 concrete).
f y db
ld = #
αd is coefficient related to reinforcing bar diameter (= 1.1 for D19 bars or less, 1.0
4 f mbd for D22 to D29 bars, and 0.9 for D32 bars or greater); cw + Katr is coefficient-related
transverse bars specified in ACI 408R-03; Es is elastic modulus of reinforcing bar
fib Model fmbd = (αm2 + αm3)fbd0 < 2.5fbd0 < f c′ (=200,000 MPa).
Code fbd0 = 0.35η3η4 f c′ Note: 1 in. = 25.4 mm; 1 ksi = 6.90 MPa.
201013§
m2 cd / cm cm / cd
0.15
αm3 = kd(Kmtr – αt/50) ≥ 0

Kmtr = ∑Atr/stndb ≤ 0.05
ls = ζaζllab ≥ 300 mm
GB 50010- ζa = γ1γ2γ3γ4γ5 ≥ 0.6
201014|| f d
lab y b
f ct
ld
fs 31 2 f y
db
u 1 f / s1
1.4

1 u
1.4 1 f / s1

Fig. 2—Definition of concrete cover and bar spacing.
f 1.47ld2 u l
1 0.007 d 0 of moment gradient and shear force on the bond strength of
Hwang et s1 Es d b f c f c
al. model15# bar splices, three-point loading tests were also performed.
cw K atr / db
u 0.91 d Moment gradient and shear force are simultaneously applied
2.5 to the lap splice region in the three-point loading test spec-
16 6C11 u imens. Stirrups were placed to avoid shear failure in shear
2 u
16 6C1u 2 span length (that is, the shear strength-demand ratio was 1.9
to 3.8 and 2.1 to 4.6 for three-point loading and four-point

C1 ld2 / 1 0.003 f c Es db
loading test specimens, respectively).
Regarding the specimen names, the first letter “L” and
*
λ is lightweight factor (0.75 to 1.0); ψt is reinforcement location factor (1.0 to
“D” denotes four-point loading and three-point loading
1.3); ψe is coating factor (1.0 to 1.5); ψs is bar diameter factor (0.8 to 1.0); ψg is bar
yield strength factor (1.0 to 1.3); cb is thickness of the bottom cover concrete; cso is tests, respectively; “D18” and “D25” denote the diameter of
thickness of the side cover concrete; csi is one-half of the center-to-center spacing of longitudinal bars; the following number 0, 2, or 4 denotes
bars; Atr is total area of confining reinforcement within spacing st across the potential
the clear spacing of spliced bars (that is, 0db for contact lap
splitting plane; ns is number of bars being developed or lap spliced along the splitting
plane; and st is center-to-center spacing of transverse bars. The confinement term splices, and 2db or 4db for noncontact lap splices); and the
(cf +Ktr)/db is limited to 2.5 considering pullout failure. When lap splices are used, last letter “s” denotes the bar splices confined by stirrups.
the required lap splice length ls is defined as 1.0ld or 1.3ld for the Class A or B splice
For specimens with stirrups, a D8 bar (diameter = 8 mm
[0.31 in.] and cross-sectional area = 50.3 mm2 [0.08 in.2])

Table 2—Design properties of test specimens
Cover and bar spacing, mm
ls/ls (ACI ls/ls (ACI
Specimens fc', MPa db, mm Stirrups cb cso csi cl (cf + Ktr)/db (cw + Katr)/db ls, mm 318-19) 408R-03)
L1-D18-0s 26.8 18 D8@200 33 33 156 0 2.50 2.66 300 0.48 0.73
L2-D18-0 25.0 18 — 33 33 156 0 2.33 2.33 300 0.43 0.62
L3-D18-2s 26.8 18 D8@200 33 33 120 2db = 36 2.50 2.66 300 0.48 0.73
L4-D18-2 25.3 18 — 33 33 120 2db = 36 2.33 2.33 300 0.43 0.63
L5-D18-4s 26.8 18 D8@200 33 33 84 4db = 72 2.50 2.66 300 0.48 0.73
L6-D18-4 24.4 18 — 33 33 84 4db = 72 2.33 2.33 300 0.42 0.62
L7-D25-0s 27.1 25 D8@200 33 33 142 0 2.22 2.12 300 0.26 0.32
L8-D25-0 25.1 25 — 33 33 142 0 1.82 1.82 300 0.20 0.27
L9-D25-2s 27.1 25 D8@200 33 33 92 2db = 50 2.22 2.12 300 0.26 0.32
L10-D25-2 25.1 25 — 33 33 92 2db = 50 1.82 1.82 300 0.20 0.27
L11-D25-4s 27.1 25 D8@200 33 33 42 4db = 100 2.22 2.12 300 0.26 0.32
L12-D25-4 25.1 25 — 33 33 42 4db = 100 1.82 1.82 300 0.20 0.27
D1-D18-0s 26.7 18 D8@200 33 33 156 0 2.50 2.66 300 0.47 0.73
D2-D18-0 28.9 18 — 33 33 156 0 2.33 2.33 300 0.46 0.66
D3-D18-2s 26.7 18 D8@200 33 33 120 2db = 36 2.50 2.66 300 0.47 0.73
D4-D18-2 28.9 18 — 33 33 120 2db = 36 2.33 2.33 300 0.46 0.66
D5-D18-4s 26.7 18 D8@200 33 33 84 4db = 72 2.50 2.66 300 0.47 0.73
D6-D18-4 28.9 18 — 33 33 84 4db = 72 2.33 2.33 300 0.46 0.66
D7-D25-0s 25.7 25 D8@200 33 33 142 0 2.22 2.12 300 0.25 0.32
D8-D25-0 26.6 25 — 33 33 142 0 1.82 1.82 300 0.21 0.28
D9-D25-2s 25.7 25 D8@200 33 33 92 2db = 50 2.22 2.12 300 0.25 0.32
D10-D25-2 31.6 25 — 33 33 92 2db = 50 1.82 1.82 300 0.23 0.29
D11-D25-4s 25.7 25 D8@200 33 33 42 4db = 100 2.22 2.12 300 0.25 0.32
D12-D25-4 29.8 25 — 33 33 42 4db = 100 1.82 1.82 300 0.22 0.29
Note: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi; cl is spacing of spliced bars.
was used for stirrups at a spacing of 200 mm (7.9 in.) in the (43.3 in.). In three-point loading tests, vertical loading was
lap splice region. applied to the midspan (that is, edge of the bar splices).
Linear variable differential transformers (LVDTs) were used
Materials and testing method to measure the midspan deflection and support deflection.
Table 3 shows the yield strength fy and tensile strength fu Four strain gauges were attached to the bottom flexural bars
of reinforcing bars. For D18 and D25 bars, the yield strength to measure the reinforcing bar strain at the edge of the lap
was 460 to 480 MPa (66.7 to 69.6 ksi). Table 4 shows the splice length.
concrete mixture proportions. Compression tests of cylin-
ders (Φ150 x 300 mm [Φ5.9 x 11.8 in.]) were performed on TEST RESULTS
the day of each test. The measured compressive strength of Four-point loading tests
the concrete cylinders was fc′ = 24.4 to 31.6 MPa (3.54 to Figure 5 shows the moment-deflection relationships of
4.58 ksi). test specimens under four-point loads. The predicted flex-
Figure 4 shows the test setup of simply supported beam ural strength was calculated from the section analysis using
specimens under four-point and three-point loads. A the reinforcing bar stress estimated from ACI 408R-03,11 the
force-controlled load was applied using a hydraulic jack modified Kent and Park model16,17 for confined and uncon-
with a capacity of 1000 kN (224.8 kip). Each loading step fined concrete, and the elasto-plastic model for reinforcing
was increased by 5% of the peak strength predicted by ACI bars. Using the stress-strain relationships of materials, the
408R-0311 until failure. The loading increment between each moment-curvature relationship was determined, and then
step was constant, which was monitored constantly during the beam moment corresponding to a given reinforcing bar
the test. The test was terminated when the load decreased stress was estimated. The nominal strength was calculated
to 85% of the actual peak strength or the specimen lost using the measured reinforcing bar yield strength.
its load-carrying capacity. In four-point loading tests, two In the specimens without stirrups, the peak strength
concentric loads were applied at a distance of 1100 mm decreased as the clear spacing of spliced bars increased.

Fig. 3—Configuration and reinforcement details of specimens. (Note: Units in mm.)
Compared to specimen L2-D18-0 using contact lap splices of spacing. On the other hand, specimen L9-D25-2s exhibited
D18 bars, the peak strength decreased by 10.1% and 19.9% almost the same strength to that of L7-D25-0s.
in specimens L4-D18-2 and L6-D18-4 using noncontact Figure 6 shows the crack pattern and failure mode of spec-
lap splices with 36 and 72 mm (1.4 and 2.8 in.) reinforcing imens at the end of the tests. All specimens failed before the
bar spacing, respectively. Compared to specimen L8-D25-0 spliced bars reached the yield strength. Stirrups improved
using contact lap splices of D25 bars, the peak strength crack resistance as well as bond resistance, and more cracks
decreased by 7.7% and 15.1% in specimens L10-D25-2 and occurred intensively in the lap splice region of the specimens
L12-D25-4 using noncontact lap splices with 50 and 100 mm without stirrups. However, more cracks occurred in speci-
(2.0 and 3.9 in.) reinforcing bar spacing, respectively. mens L7-D25-0s and L9-D25-2s with stirrups, compared
In specimens with stirrups, the opposite behavior occurred. with specimens L8-D25-0 and L10-D25-2 without stir-
Compared to specimen L1-D18-0s using contact lap splices rups. This is because the peak strength of L7-D25-0s and
of D18 bars, the peak strength increased by 12.6% and 15.0% L9-D25-2s is 51% and 64% greater than that of L8-D25-0
in specimens L3-D18-2s and L5-D18-4s using noncontact and L10-D25-2, respectively.
lap splices with 36 and 72 mm (1.4 and 2.8 in.) reinforcing Table 5 shows the summary of the test results. Reinforcing
bar spacing, respectively. However, compared to specimen bar stress fm was calculated from the section analysis at the
L7-D25-0s using contact lap splices of D25 bars, the peak peak strength. Average reinforcing bar stress fg was calcu-
strength decreased by 27.4% in specimen L11-D25-4s using lated from four strain gauges of each spliced bar. Strain
noncontact lap splices with 100 mm (3.9 in.) reinforcing bar values showed relatively large deviation in each specimen,
which caused a difference between the measured reinforcing

Table 3—Mechanical properties of reinforcing bars Table 4—Concrete mixture proportions
Reinforcing bars db, mm fy, MPa fu, MPa Unit weight, kgf/m3
Specified strength
D8 8 506 630 fc′, MPa w/c W C FA CA
D18 18 480 608 28.3 0.44 205 466 571 1158
D25 25 460 582 Note: W is water; C is cement; FA is fine aggregate; CA is coarse aggregate; w/c is
water-cement ratio; 1 MPa = 0.145 ksi; 1 kgf/m3 = 0.062 lb/ft3.
Note: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi.
Fig. 4—Test setup. (Note: Units in mm.)
Fig. 5—Moment-deflection and load-deflection relationships of four-point loading test specimens.

Fig. 6—Crack pattern and failure mode of four-point loading test specimens.
bar stress fm and fg. The ratio of fg to fm in four-point loading D11-D25-4 using noncontact lap splices with 50 and 100
tests was 0.76 to 1.24. The discrepancy of fg between speci- mm (2.0 and 3.9 in.) reinforcing bar spacing, respectively.
mens L3-D18-2s and L5-D18-4s was approximately 22.4% Figure 8 shows the crack pattern and failure mode of spec-
despite almost the same peak strength. Considering that the imens at the end of the tests. In general, specimens with stir-
strain results were possibly affected by strain gauge error, rups exhibited better crack resistance and bond resistance.
the reinforcing bar stress fm based on the peak strength was All specimens failed before the spliced bars reached the
used for the test results. yield strength, except specimen D5-D18-4s. In D5-D18-4s,
obvious diagonal cracks occurred, although the peak strength
Three-point loading tests was less than the nominal shear strength. However, diagonal
Figure 7 shows the moment-deflection and load- cracks did not occur in the opposite side.
deflection relationships of the test specimens under three- Table 5 shows the summary of the test results. In specimen
point loads. Unlike four-point loading test results affected D1-D18-0s, strain was not measured due to the malfunc-
by the use of stirrups, the relationship between the peak tion of the strain gauges. The ratio of fg to fm in three-point
strength and spacing of spliced bars was affected by the rein- loading tests was 0.70 to 1.26. The reinforcing bar stresses
forcing bar diameter. fm calculated from the section analysis was considered in
In the specimens without stirrups, the peak strength the test results, except specimen D5-D18-4s showing rein-
increased or decreased as the clear spacing of spliced bars forcing bar yielding. For specimen D5-D18-4s, the average
increased. Compared to specimen D2-D18-0 using contact reinforcing bar stress fg calculated from the two strain gauges
lap splices of D18 bars, the peak strength increased by at the maximum moment location was used.
12.7% and 6.2% in specimens D4-D18-2 and D6-D18-4
using noncontact lap splices with 36 and 72 mm (1.4 and DISCUSSION
2.8 in.) reinforcing bar spacing, respectively. On the other Effect of clear spacing of noncontact spliced bars
hand, compared to specimen D8-D25-0 using contact lap Figure 9 compares the reinforcing bar stress ratio of
splices of D25 bars, the peak strength decreased by 8.3% and noncontact (spaced) splices to contact splices in accordance
increased 0.3% in specimens D10-D25-2 and D12-D25-4 with the clear distance between spliced bars. Compared to
using noncontact lap splices with 50 and 100 mm (2.0 and contact splices with stirrups, the average bond strength of
3.9 in.) reinforcing bar spacing, respectively. noncontact splices increased by 2.9% (Fig. 9(a)). In spec-
The specimens with stirrups showed similar behavior imens without stirrups, the average strength decreased by
to the specimens without stirrups. Compared to spec- 5.4% when noncontact splices were used (Fig. 9(b)). Thus,
imen D1-D18-0s using contact lap splices of D18 bars, the sufficient transverse reinforcement needs to be provided for
peak strength increased by 9.7% and 38.2% in specimens the better bond performance of noncontact splices.
D3-D18-2s and D5-D18-4s using noncontact lap splices The test results also indicate that the use of noncontact
with 36 and 72 mm (1.4 and 2.8 in.) reinforcing bar spacing, lap splices may enhance the bond performance of rein-
respectively. However, compared to specimen D7-D25-0s forcing bars with small diameters, which was also reported
using contact lap splices of D25 bars, the peak strength in the study of Hamad and Mansour.6 Compared to spec-
decreased by 4.1% and 16.1% in specimens D9-D25-2 and imens using contact splices of D18 bars with stirrups, the

Table 5—Summary of test results
Test-to-prediction ratio of reinforcing bar stress
ACI
db, ls/ls (ACI ACI 408R- MC Hwang et Mu,
Specimens mm 408R-03) 318-1910 0311 Eurocode 212 201013 GB 5001014 al. model15 kN·m fg, MPa fm, MPa fg/fm
L1-D18-0s 18 0.73 1.34 0.85 1.33 1.20 1.91 0.85 59.3 332.5 306.1 1.09
L2-D18-0 18 0.62 1.57 0.98 1.46 1.34 2.09 1.05 62.2 339.6 321.7 1.06
L3-D18-2s 18 0.73 1.51 0.95 1.50 1.40 2.15 0.96 66.8 357.1 344.9 1.04
L4-D18-2 18 0.63 1.40 0.88 1.30 1.25 1.86 0.94 55.9 316.6 288.9 1.10
L5-D18-4s 18 0.73 1.54 0.97 1.58 1.50 2.19 0.98 68.2 437.0 352.2 1.24
L6-D18-4 18 0.62 1.27 0.79 1.23 1.20 1.69 0.85 49.8 194.8 257.6 0.76
L7-D25-0s 25 0.32 1.93 0.95 1.47 1.51 1.94 1.18 81.8 196.4 226.1 0.87
L8-D25-0 25 0.27 1.62 0.69 1.03 1.04 1.35 0.95 54.2 132.8 150.0 0.89
L9-D25-2s 25 0.32 1.93 0.95 1.47 1.61 1.94 1.18 81.8 190.3 226.1 0.84
L10-D25-2 25 0.27 1.49 0.63 0.95 1.02 1.24 0.88 50.0 142.2 138.3 1.03
L11-D25-4s 25 0.32 1.40 0.69 1.13 1.31 1.41 0.86 59.4 180.0 164.0 1.10
L12-D25-4 25 0.27 1.37 0.58 0.92 1.06 1.14 0.81 46.0 122.6 127.2 0.96
D1-D18-0s 18 0.73 1.38 0.87 1.37 1.23 1.96 0.87 71.5 None 369.3 None
D2-D18-0 18 0.66 1.45 0.94 1.32 1.24 1.90 0.97 73.0 351.1 376.3 0.93
D3-D18-2s 18 0.73 1.51 0.95 1.50 1.40 2.15 0.96 78.4 433.6 405.1 1.07
D4-D18-2 18 0.66 1.63 1.06 1.49 1.46 2.15 1.10 82.3 417.7 424.5 0.98
D5-D18-4s 18 0.73 1.84 1.16 1.88 1.79 2.61 1.16 98.8 492.1 None None
D6-D18-4 18 0.66 1.54 1.00 1.45 1.45 2.02 1.03 77.5 281.0 399.6 0.70
D7-D25-0s 25 0.32 2.04 1.00 1.57 1.59 2.06 1.26 98.9 230.2 274.1 0.84
D8-D25-0 25 0.28 1.75 0.76 1.10 1.12 1.45 1.03 71.1 178.5 196.5 0.91
D9-D25-2s 25 0.32 1.95 0.95 1.51 1.63 1.98 1.20 94.8 226.4 262.6 0.86
D10-D25-2 25 0.29 1.47 0.66 0.90 1.00 1.19 0.86 65.2 191.1 179.1 1.07
D11-D25-4s 25 0.32 1.71 0.84 1.39 1.60 1.73 1.05 83.0 287.8 229.8 1.25
D12-D25-4 25 0.29 1.64 0.73 1.07 1.27 1.34 0.97 70.9 246.7 195.2 1.26
Average — — 1.60 0.87 1.33 1.34 1.81 1.00 — — — —
COV — — 0.132 0.167 0.181 0.159 0.205 0.127 — — — —
Note: Mu is peak moment of test results; fg is average reinforcing bar stress calculated from strain measurement; fm is reinforcing bar stress calculated from section analysis; 1 mm =
0.039 in.;1 MPa = 0.145 ksi; and 1 kN·m = 0.73757 ft·kip.
developed bond stress of D18 bars confined by stirrups length. For this reason, the bond strength of D18 bars in
increased by 17.7% on average in noncontact lap splices. On noncontact lap splices increased. However, the effective lap
the other hand, compared to contact splices of D25 bars with splice length was significantly decreased in the noncontact
stirrups, the developed bond stress of D25 bars in noncon- lap splices of D25 bars, which decreased the bond strength.
tact splices with stirrups decreased by 11.9%. This result
may be attributed to significant local bond failure due to Effect of stirrups
the insufficient lap splice length of D25 bars. The lap splice In general, transverse reinforcement improves the bar
length for D25 bars in this study was only approximately bond strength in the lap slice length. As shown in the test
30% of the requirement of ACI 408R-03,11 while the lap results, compared to the specimens without stirrups, the peak
splice length for D18 bars was approximately 70% of the strength of the specimens with stirrups increased by 26.2% on
requirement. The clear spacing of spliced bars decreased the average, and the maximum increment was 63.4%. Figure 10
effective lap splice length in noncontact lap splices due to compares the ratio of the bar stress of specimens with stir-
the diagonal crack distribution (Fig. 1). On the other hand, rups to the bar stress of specimens without stirrups. For D18
the inner concrete between the spliced bars developed addi- bars, compared to specimens without stirrups, the average
tional confinement, which increased the bond stress. In bar stress decreased by 3.4% in contact splices, but increased
the noncontact lap splices of D18 bars, the bond strength by 7.4% and 30.0% in noncontact splices with the spacing of
increment due to the confinement was greater than the bond 2db and 4db, respectively. For D25 bars, compared to spec-
strength degradation due to the reduced effective lap splice imens without stirrups, the average bar stress increased by

Fig. 7—Moment-deflection and load-deflection relationships of three-point loading test specimens.
Fig. 8—Crack pattern and failure mode of three-point loading tests.

45.1%, 55.0%, and 23.3% in contact splices and noncon- the end of the lap splice of each reinforcing bar. Considering
tact splices with the spacing of 2db and 4db, respectively. In linear moment distribution, the moment ratio is Mu2/Mu = 0.7
beams with stirrups, the bar bond strength of the noncontact corresponding to the strain gauge location. The reinforcing
splices was improved. bar stresses fm and fm2 were calculated from section analysis
using the moment values Mu and Mu2.
Effect of moment distribution Figure 12 compares the bar stress ratio between fm2/fm and
In actual flexural members, nonuniformly distributed fg2/fg. The average ratio of fm2/fm was 0.699, which was close
moment and shear force are simultaneously applied, which to the moment ratio of 0.7. On the other hand, the average
differs from uniformly distributed moment condition in ratio of fg2/fg was 0.765, showing large variations from 0.641
four-point loading tests. Compared to four-point loading test to 0.936 due to nonuniform bond stress distribution. Partic-
results, three-point loading test results showed higher peak ularly, the ratio of fg2/fg in specimens using D25 bars was
strength and tensile strength of bar splices. Figure 11 shows larger than that of specimens using D18 bars showing an
the locations of strain gauges and moment distribution in the average value of 0.798.
three-point loading test specimens. The average bar stresses Figure 13 shows the ratio of the bar stress of three-point
fg and fg2 were determined from the strain measurement at loading specimens to the bar stress of four-point loading

Fig. 9—Reinforcing bar spacing and stress ratio relationships of test specimens.
Fig. 10—Comparison of reinforcing bar stress of lapped Fig. 12—Reinforcing bar stress ratio of two different loca-
bars with/without stirrups. tions in three-point loading specimens.
Fig. 13—Reinforcing bar stress ratio of specimens under

three-point/four-point loads.
increased, the bar stress ratio increments increased. For
contact splices, the average bar stress ratio increased by
22.5%. For noncontact splices with the spacing of 2db and
4db, the average bar stress ratio increased by 27.5% and
47.1%, respectively. In three-point loading tests, linear
Fig. 11—Detailed information of strain gauge location. moment distribution decreased the tension force of bar
specimens. The average bar stress ratio increased by 32.4% splices, which decreased the bond requirement. The effect of
in three-point loading specimens compared to four-point shear force on the bar bond strength was insignificant.
loading specimens. As the clear spacing of spliced bars

COMPARISON OF PREDICTIONS AND TEST considered in Eurocode 212 when noncontact splice with
RESULTS spacing more than 50 mm (1.97 in.) is used. The increment
In the three-point loading specimens, the linear distribu- of lap splice length requirement equals the part of spacing
tion of moment in the lap splice length decreased the bond of spliced bars above 50 mm (1.97 in.). For instance, in
requirement.9 However, current design codes and methods specimens with noncontact splices at a spacing of 72 mm
do not consider this situation when calculating the tensile (2.83 in.), an extra lap splice length of 22 mm (0.86 in.) was
strength of bar splices. Thus, for more accurate calculation, considered. In other prediction methods, ld = ls was used.
a modification for reinforcing bar stress or development The reinforcing bar stress factor αs was additionally consid-
length was proposed. ered for specimens under moment gradient in all existing
In a linearly distributed moment region, the bond demand methods. For direct comparison with the test results, safety
of bar splices can be defined as the average tensile stress fs0 factors were not used in the calculations of current design
between fsm and fsi (where fsm and fsi are reinforcing bar stress codes and methods, except ACI 318-19,10 which has an
corresponding to the maximum moment Mmax and minimum implied safety factor included in the design equation.
moment Mmin within the lap splice length, respectively) As seen in Table 5, ACI 318-1910 underestimated the
tensile strength of bar splices showing the average ratio =
f sm f si 1 M ls 2 a ls 1.60 and coefficient of variation (COV) = 0.132 in the stress
fs0 f sm 1 min f sm 1 f sm
2 2 M max 2 a 2a ratio of 1.27 to 2.04, and the prediction was conservative
by the implied safety factor. ACI 408R-0311 overestimated
(1) the tensile strength of the lap splices with an average ratio =
0.87 and COV = 0.167 in the stress ratio of 0.58 to 1.16.
where a is the shear span length; and ls is the lap splice Eurocode 212 and fib Model Code 201013 showed a similar
length. result with the average ratio of 1.33 and 1.34, respectively.
A reinforcing bar stress factor concerning moment gradient The former showed COV = 0.181 in the stress ratio of 0.90
can be defined as αs. When calculating the reinforcing bar to 1.88, and the latter showed COV = 0.159 in the stress ratio
strength of spliced bars under moment gradient with a given of 1.00 to 1.79. GB 50010-201014 showed the most conser-
development length, the modified reinforcing bar stress fs′ vative prediction with an average ratio = 1.81 and COV =
can be calculated as follows 0.205 in the stress ratio of 1.14 to 2.61. The Hwang et al.
model15 showed an average ratio = 1.00 and COV = 0.127 in
2a the stress ratio of 0.81 to 1.26.
f s s f s f s f y (2)
2 a ls Figure 14 compares the bar stress predictions of the
existing methods with the test results of this study and
where fs is the basic reinforcing bar stress calculated using existing studies. In Appendix B, Tables B1 and B2 list the
current methods; and fy is the reinforcing bar yield strength. principal test parameters and comparison between test results
For design codes or methods where reinforcing bar stress and predictions of 38 existing bar-splice test specimens,2,3,6,8
is linearly related to development length, the consideration respectively. The lap splice length ls ranged from 76 to
of moment gradient can be transferred to a reduction of 406 mm (3.0 to 16.0 in.), reinforcing bar diameter db ranged
required development length as follows from 12.0 to 20.0 mm (0.5 to 0.8 in.), concrete strength fc′
ranged from 19.4 to 63.2 MPa (2.8 to 9.2 ksi), yield strength
ld 2 a l s of the reinforcing bar fy ranged from 345 to 581 MPa (50.0
ld ld (3)
s 2a to 84.3 ksi), and spacing of spliced bars ranged from 1db to
where ld′ is the modified development length requirement; 10.7db. Note that stirrups were not used along the lap splice
and ld is the basic development length requirement. length in the existing noncontact lap splice test specimens.
In Appendix A, Fig. A1 and A2 show the test-prediction In general, ACI 408R-0311 provided the best prediction for
ratios of reinforcing bar stress for specimens in this study noncontact splices with an average ratio = 1.00 and COV =
with and without moment gradient factor αs, respectively, 0.206. ACI 318-1910 underestimated the reinforcing bar
according to the concrete compressive strength fc′. The tensile strength of the noncontact lap splices with an average
moment gradient factor αs improved the prediction of rein- ratio = 1.87 and COV = 0.315. Eurocode 2,12 fib Model
forcing bar stress for three-point loading specimens signifi- Code 2010,13 and GB 50010-201014 underestimated the
cantly. Thus, in the following comparison of prediction reinforcing bar tensile strength with an average ratio = 1.52,
results, the moment gradient factor αs was introduced for 1.58, and 2.05, and COV = 0.238, 0.270, and 0.267, respec-
prediction of three-point loading specimens. tively. The Hwang et al.15 model predicted the test results
Table 5 shows the ratio of the test results to the predic- relatively well, except four test results (an average ratio =
tions of the tensile stress of bar splices in test specimens. 1.17 and COV = 0.287). Note that these four specimens were
Reinforcing bar stress predictions were calculated according in the same test group of Hamad and Mansour,6 which used
to reinforcing bar development length. In ACI 318-19,10 the D20 bars with a concrete cover cb = cs = 20 mm (0.79 in.).
ratio between lap splice length ls and development length When considering the design of noncontact splices, current
ld is ld = ls/1.3 for the Class B lap splices. In Eurocode 2,12 design codes can be used to get a relatively reliable result.
addressing that all reinforcing bars are lap spliced, ld = ls/1.5 However, ACI 318-1910 and GB 50010-201014 underesti-
is used. Further, an extra lap splice length requirement is mated the test results significantly. ACI 408R-0311 provided

Fig. 14—Comparison of predictions and test results for all test specimens.
the most accurate and economic results by considering the contact splices, the average bar stress decreased by 2.1% and
noncontact splices as the calculation parameter csi deter- 21.8% in noncontact splices with the spacing of 2db and 4db,
mined by the spacing of the spliced bars indirectly. However, respectively.
ACI 408R-0311 overestimated the reinforcing bar stress of 3. As the clear spacing of spliced bars increased, diagonal
specimens with D25 bars (db = 25.0 mm [0.98 in.]) in this cracks between the spliced bars decreased the effective lap
study with an average ratio of 0.75. In actual applications, splice length in noncontact lap splices, but the confinement
a minimum lap splice length concerning the reinforcing effect of concrete increased the bond stress. Thus, when a
bar diameter should be applied to ensure the reliability of sufficient effective lap splice length and transverse reinforce-
noncontact lap splice design. To verify the applicability ment is used, the bond strength of noncontact lap splices is
of noncontact splices under various design conditions, greater than that of contact lap splices. However, when the
including concrete strength, reinforcing bar diameter, rein- spliced bars are too widely separated or transverse reinforce-
forcing bar yield strength, and concrete cover, further studies ment is not used, the bond strength of noncontact lap splices
are required. is less than that of contact lap splices.
4. Linear moment distribution (that is, moment ratio = 0.7
SUMMARY AND CONCLUSIONS in the lap splice length) increased the bar stress by an average
In the present study, lap splice tests were performed to of 32.4%, compared to uniformly distributed moment. Due
investigate the bond stress of noncontact spliced bars. The to linear moment distribution, the tension force of bar splices
test parameters were the bar diameter, spacing of spliced decreased, which decreased the bond requirement in lap
bars, stirrups, and moment distribution. The test results splices. Thus, when actual moment distribution is consid-
were compared with the predictions of current design codes ered, lap splice length can be decreased.
and existing methods. The primary results of this study are 5. The Hwang et al. model15 showed the best prediction of
summarized as follows. reinforcing bar stress for 24 specimens in this study with an
1. In specimens without stirrups, the bond stress of lapped average test-prediction ratio = 1.00 and coefficient of varia-
bars decreased as the spacing of spliced bars increased under tion (COV) = 0.127. ACI 318-1910 underestimated the rein-
pure moment. Compared to contact splices, the average bar forcing bar stress with an average ratio of 1.60 and COV =
stress decreased by 9.0% and 17.6% in noncontact splices 0.132. ACI 408R-0311 overestimated the reinforcing bar
with the spacing of 2db and 4db, respectively. However, for stress with an average ratio of 0.87 and COV = 0.167.
specimens under moment gradient, compared to contact 6. According to the test prediction of noncontact splice
splices, the average bar stress increased by 2.0% and 2.6% test results in this study and existing noncontact splice tests,
in noncontact splices with the spacing of 2db and 4db, ACI 408R-0311 showed the best prediction with an average
respectively. ratio = 1.00 and COV = 0.206. ACI 408R-0311 considered
2. In specimens with stirrups, the tensile strength of bar the effect of noncontact splices indirectly using the param-
splices was significantly affected by reinforcing bar diameter. eter csi, which is half of the bar clear spacing, and provided
For specimens using D18 reinforcing bars, the reinforcing preferable results. However, a minimum lap splice length
bar bond stress increased as the spacing of spliced bars concerning the reinforcing bar diameter should be applied
increased under both pure moment and moment gradient. to ensure the reliability of noncontact lap splice design for
Compared to contact splices, the average bar stress increased large-size reinforcing bars.
by 11.2% and 24.2% in noncontact splices with the spacing Note that this study focused on the bond strength of
of 2db and 4db, respectively. For specimens using D25 rein- noncontact lap splices under monotonic loading. In beam-
forcing bars, contrary results were obtained. Compared to column joints, column-footing connection, and/or beams

under repeated loading, the effect of cyclic loading on the ld = development length of reinforcing bar (mm, in.)
ld′ = modified development length of reinforcing bar (mm, in.)
bond strength is of importance. Thus, further study is still ls = lap splice length of reinforcing bar (mm, in.)
needed to investigate the bond behavior of noncontact lap Mmax = maximum flexural moment within lap splice length (kN·m,
splices under various design conditions. ft∙kip)
Mmin = minimum flexural moment within lap splice length (kN·m,
ft∙kip)
AUTHOR BIOS Mu = maximum flexural moment corresponding to peak strength
ACI member Hyeon-Jong Hwang is an Associate Professor in the School (kN·m, ft∙kip)
of Architecture at Konkuk University, Seoul, South Korea. He received his αs = reinforcing bar stress factor of considering moment distribution
BE, MS, and PhD in architectural engineering from Seoul National Univer-
sity, Seoul, South Korea. He is a member of ACI Subcommittee 318-L, Inter-
national Liaison, and Joint ACI-ASCE Committees 352, Joints and Connec- REFERENCES
tions in Monolithic Concrete Structures, and 408, Bond and Development 1. Chamberlin, S. J., “Spacing of Spliced Bars in Tension Pull-Out Spec-
of Steel Reinforcement. His research interests include inelastic analysis and imens,” ACI Journal Proceedings, V. 49, No. 12, Dec. 1952, pp. 261-274.
seismic design of reinforced concrete and composite structures. 2. Chamberlin, S. J., “Spacing of Spliced Bars in Beams,” ACI Journal
Proceedings, V. 54, No. 2, Feb. 1958, pp. 689-697.
Fan Yang is a Graduate Student in the College of Civil Engineering at 3. Chinn, J.; Ferguson, P. M.; and Thompson, J. N., “Lapped Splices in
Hunan University, Changsha, Hunan, China. He received his BE in civil Reinforced Concrete Beams,” ACI Journal Proceedings, V. 52, No. 10, Oct.
engineering from Hefei University of Technology, Hefei, Anhui, China, in 1955, pp. 201-213.
2013. His research interests include development of reinforcing bar stress 4. Goto, Y., and Otsuka, K., “Experimental Studies on Cracks Formed in
and bond performance of noncontact lap splices. Concrete around Deformed Tension Bars,” Technology Reports of Tohoku
University, V. 44, No. 1, June 1979, pp. 49-83.
Gao Ma is an Associate Professor in the College of Civil Engineering at 5. Sagan, V. E.; Gergely, P.; and White, R. N., “Behavior and Design of
Hunan University and Key Laboratory for Damage Diagnosis of Engi- Noncontact Lap Splices Subjected to Repeated Inelastic Tensile Loading,”
neering Structures of Hunan Province (Hunan University). He received ACI Structural Journal, V. 88, No. 4, July-Aug. 1991, pp. 420-431.
his BE from Wuhan University, Wuhan, Hubei, China, and his PhD from 6. Hamad, B. S., and Mansour, M. Y., “Bond Strength of Noncontact
Harbin Institute of Technology, Harbin, Heilongjiang, China. His research Tension Lap Splices,” ACI Structural Journal, V. 93, No. 3, May-June
interests include seismic analysis, fiber-reinforced polymer retrofit, and 1996, pp. 316-326.
nondestructive testing. 7. ACI Committee 318, “Building Code Requirements for Structural
Concrete (ACI 318-89) and Commentary (ACI 318R-89),” American
Concrete Institute, Farmington Hills, MI, 1989, 353 pp.
ACKNOWLEDGMENTS 8. Kilpatrick, A. E., and Gilbert, R. I., “A Preliminary Investigation of the
This research was financially supported by the National Research Foun- Strength and Ductility of Lapped Splices of Reinforcing Bars in Tension,”
dation of Korea (NRF) grant funded by the Korean government (MSIT) 22nd Australasian Conference on the Mechanics of Structures and Mate-
(No. 2020R1F1A1076322), the Korea Agency for Infrastructure Tech- rials (ACMSM 2012), Sydney, Australia, 2012, pp. 305-311.
nology Advancement (KAIA) funded by the Ministry of Land, Infrastruc- 9. Kim, C.-G.; Park, H.-G.; and Eom, T.-S., “Seismic Performance of
ture and Transport (No. 1615012117), and the National Natural Science Reinforced Concrete Columns with Lap Splices in Plastic Hinge Region,”
Foundation of China (Grant No. 51851110760 and 51878268). The authors ACI Structural Journal, V. 115, No. 1, Jan. 2018, pp. 235-245. doi:
are grateful to the authority for the support. 10.14359/51701109
10. ACI Committee 318, “Building Code Requirements for Structural
NOTATION Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
a = shear span length (mm, in.)
11. Joint ACI-ASCE Committee 408, “Bond and Development of
cb = thickness of bottom cover concrete (mm, in.)
Straight Reinforcing Bars in Tension (ACI 408R-03),” American Concrete
cl = clear distance between spliced bars (mm, in.)
Institute, Farmington Hills, MI, 2003, 49 pp.
csi = half of center-to-center bar spacing (mm, in.)
12. BS EN 1992-1-1:2004, “Eurocode 2: Design of Concrete Structures,”
cso = thickness of side cover concrete (mm, in.)
European Committee for Standardization, Brussels, Belgium, 2004, 225 pp.
db = reinforcing bar diameter (mm, in.)
13. Fédération internationale du béton, “fib Model Code for Concrete
fc′ = concrete compressive strength (MPa, psi)
Structures 2010,” fib, Lausanne, Switzerland, 2013, 420 pp.
fg = reinforcing bar stress calculated using strain results measured by
14. GB 50010-2010, “Code for Design of Concrete Structures,” Ministry
strain gauges (MPa, psi)
of Housing and Urban-Rural Development of the People’s Republic of
fm = reinforcing bar stress calculated from section analysis (MPa,
China, Beijing, China, 2015, 427 pp. (in Chinese)
psi)
15. Hwang, H.-J.; Park, H.-G.; and Yi, W.-J., “Nonuniform Bond
fp = predicted tensile stress of bar splice (MPa, psi)
Stress Distribution Model for Evaluation of Bar Development Length,”
fs = bond stress of bar splice (MPa, psi)
ACI Structural Journal, V. 114, No. 4, July-Aug. 2017, pp. 839-849. doi:
fs′ = modified bond stress of bar splice (MPa, psi)
10.14359/51689446
fs0 = average tensile stress of bar splice within lap splice length (MPa,
16. Kent, D. C., and Park, R., “Flexural Members with Confined
psi)
Concrete,” Journal of the Structural Division, ASCE, V. 97, No. 7, 1971,
fsi = minimum tensile stress of bar splice within lap splice length
pp. 1969-1990. doi: 10.1061/JSDEAG.0002957
(MPa, psi)
17. Park, R.; Priestley, M. J. N.; and Gill, W. D., “Ductility of Square-
fsm = maximum tensile stress of bar splice within lap splice length
Confined Concrete Columns,” Journal of the Structural Division, ASCE,
(MPa, psi)
V. 108, No. 4, 1982, pp. 929-950. doi: 10.1061/JSDEAG.0005933
fu = ultimate strength of reinforcing bar (MPa, psi)
fy = yield strength of reinforcing bar (MPa, psi)

APPENDIX A: COMPARISON BETWEEN TEST RESULTS AND PREDICTIONS WITH MOMENT
GRADIENT FACTOR
Fig. A1—Bar stress ratio of test specimens with moment gradient factor according to concrete strength.
Fig. A2—Bar stress ratio of test specimens without moment gradient factor according to concrete strength.

APPENDIX B: TEST PARAMETERS AND COMPARISON BETWEEN TEST RESULTS AND PREDICTIONS
Table B1—Test parameters of existing noncontact bar-splice test specimens

Specimens ls, mm db, mm fc′, MPa fy, MPa cl, mm (cf + Ktr)/db (cw + Katr)/db
1b 76 12.7 30.2 345 12.7 2.50 2.63
3b 152 12.7 30.7 345 12.7 2.50 2.63
3c 152 12.7 30.7 345 25.4 2.50 2.63
5b 305 12.7 32.0 345 12.7 2.50 2.63
5c 305 12.7 32.0 345 25.4 2.50 2.63
Chamberlin2
2b 76 12.7 30.5 345 12.7 2.50 2.88
4b 152 12.7 30.2 345 12.7 2.50 2.88
4c 152 12.7 30.2 345 25.4 2.50 2.88
6b 305 12.7 30.8 345 12.7 2.50 2.88
6c 305 12.7 30.8 345 25.4 2.50 2.88
D2 279 19.1 33.3 393 19.1 1.50 1.55
D3 279 19.1 30.0 393 25.4 1.83 2.21
Chinn et al. 3
D4 406 19.1 30.8 393 25.4 1.83 2.21
D5 279 19.1 28.8 393 25.4 1.83 2.21
D21 279 19.1 30.9 393 25.4 1.83 2.2
S2-14-300-30 300 14.0 22.4 469 30.0 1.93 1.93
S3-14-300-60 300 14.0 22.8 469 60.0 1.93 1.93
S4-14-300-90 300 14.0 21.4 469 90.0 1.93 1.93
S5-14-300-120 300 14.0 20.8 469 120.0 1.93 1.93
S6-14-300-150 300 14.0 22.1 469 150.0 1.43 1.89
S8-16-300-30 300 16.0 20.6 476 30.0 1.75 1.75
S9-16-300-60 300 16.0 22.1 476 60.0 1.75 1.75
Hamad and Mansour6
S10-16-300-90 300 16.0 23.2 476 90.0 1.75 1.75
S11-16-300-120 300 16.0 22.4 476 120.0 1.75 1.75
S12-16-300-150 300 16.0 23.4 476 150.0 0.94 1.40
S14-20-350-35 350 20.0 19.4 474 35.0 1.50 1.50
S15-20-350-70 350 20.0 21.4 474 70.0 1.50 1.50
S16-20-350-105 350 20.0 24.1 474 105.0 1.50 1.50
S17-20-350-140 350 20.0 21.9 474 140.0 1.00 1.35
MNC1 186 12.0 56.7 560 91.0 1.50 1.98
MNC2 133 12.0 55.3 560 91.0 1.50 1.98
MNC3 128 12.0 53.0 560 91.0 2.50 3.64
MNC4 87 12.0 54.1 560 91.0 2.50 3.64
Kilpatrick and Gilbert 8
HNC1 165 12.0 62.5 560 91.0 1.50 1.98
HNC2 115 12.0 63.2 560 91.0 1.50 1.98
HNC3 117 12.0 54.6 560 91.0 2.50 3.64
HNC4 80 12.0 55.5 560 91.0 2.50 3.64
SL-6 180 12.0 38.0 581 28.0 2.50 3.16
76 to 12.0 to
Total No. of specimens: 38 19.4 to 63.2 345 to 581 12.7 to 150 0.94 to 2.50 1.35 to 3.64
406 20.0
Note: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi.

Table B2—Comparison between existing noncontact bar-splice test results and predictions
ftest/fp
ACI ACI GB Hwang
Specimens ftest, MPa 318-1910 408R-0311 Eurocode 212 MC 201013 50010-201014 et al. model15
1b 130 1.47 0.62 1.50 1.37 2.10 0.93
3b 229 1.29 0.82 1.31 1.19 1.83 0.82
3c 232 1.31 0.84 1.33 1.21 1.85 0.83
5b 339 0.98 0.98 0.98 0.98 1.32 0.98
5c 331 0.96 0.96 0.96 0.96 1.29 0.96
Chamberlin2
2b 241 2.72 1.05 2.77 2.44 3.86 1.58
4b 303 1.72 1.00 1.76 1.55 2.44 1.01
4c 299 1.70 0.99 1.74 1.53 2.41 0.99
6b 350 1.01 1.01 1.01 1.01 1.39 1.01
6c 350 1.01 1.01 1.01 1.01 1.39 1.01
D2 207 1.53 0.86 1.09 1.30 1.30 1.04
D3 254 1.62 0.92 1.36 1.52 1.68 1.11
Chinn et al. 3
D4 323 1.40 0.95 1.16 1.31 1.45 0.98
D5 308 2.01 1.13 1.69 1.89 2.09 1.37
D21 306 1.93 1.11 1.60 1.81 1.98 1.32
S2-14-300-30 393 1.90 1.19 1.60 1.37 2.12 1.28
S3-14-300-60 406 1.94 1.22 1.66 1.46 2.17 1.31
S4-14-300-90 402 1.99 1.23 1.80 1.58 2.24 1.34
S5-14-300-120 371 1.86 1.14 1.77 1.61 2.11 1.25
S6-14-300-150 363 2.38 1.11 1.85 2.09 1.98 1.21
S8-16-300-30 338 2.15 1.20 1.71 1.51 2.21 1.44
S9-16-300-60 350 2.15 1.22 1.71 1.58 2.18 1.44
Hamad and Mansour6
S10-16-300-90 363 2.17 1.25 1.80 1.69 2.19 1.45
S11-16-300-120 331 2.02 1.15 1.75 1.72 2.04 1.35
S12-16-300-150 321 3.56 1.21 1.77 2.72 1.92 1.59
S14-20-350-35 363 2.97 1.47 2.13 2.04 2.64 2.18
S15-20-350-70 370 2.88 1.46 2.08 2.09 2.52 2.12
S16-20-350-105 374 2.74 1.44 2.02 2.18 2.36 2.02
S17-20-350-140 347 4.01 1.42 2.06 3.21 2.33 2.18
MNC1 383 2.07 0.96 1.39 1.58 1.84 1.05
MNC2 302 2.31 0.88 1.56 1.76 2.04 1.17
MNC3 354 1.72 0.88 1.42 1.49 2.52 0.80
MNC4 300 2.13 0.92 1.75 1.84 3.12 0.97
Kilpatrick and Gilbert 8
HNC1 356 2.06 0.92 1.38 1.57 1.87 1.05
HNC2 255 2.11 0.76 1.40 1.61 1.92 1.07
HNC3 352 1.85 0.92 1.52 1.60 2.71 0.85
HNC4 320 2.44 1.02 2.00 2.11 3.59 1.11
SL-6 475 1.88 1.07 1.74 1.67 2.74 1.04
Total 0.96 to 4.01 0.62 to 1.47 0.96 to 2.77 0.96 to 3.21 1.29 to 3.86 0.80 to 2.18
Total Average — 2.00 1.06 1.61 1.66 2.15 1.24
COV 0.326 0.183 0.228 0.277 0.261 0.295
Notes: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi.

Title No. 119-S26
Maximum Shear Strength of Reinforced Concrete Beams

by Shyh-Jiann Hwang, Yu-Hsuan Yang, and Yi-An Li
Current codes limit the maximum shear strength in reinforced beam may cause inaccuracies, because the maximum shear
concrete beams to prevent possible sudden shear failure due strength of the beam test specimen would be configured with
to over-reinforcement. The ACI 318-19 limit is criticized to be a larger amount of longitudinal reinforcement to avoid flex-
over-conservative, especially for high-strength concrete beams. ural failure and obtain the test data of the beam’s maximum
Based on the force-transfer mechanism of the beams, this paper
shear strength. ACI 318-191 requires that the tensile strain of
investigates this issue by deriving a model to determine the shear
the outermost longitudinal tension reinforcement shall not
strength of rectangular beams with a shear span-depth ratio
exceeding 2. According to the proposed model, it is found that there be less than 0.004 when the beam is at the nominal flex-
are two influential parameters of the maximum shear strength for ural strength. This limit of 0.004 expects the beam to have a
beams—that is, concrete strength and longitudinal tension rein- better deformability, so that the longitudinal reinforcement
forcement. Considering the softening phenomenon of reinforced of the beam is constrained to be under-reinforced. Therefore,
concrete, the use of √fcꞌ in the ACI 318-19 limiting equation seems beam test specimens with maximum shear strength usually
appropriate for high-strength concrete beams. However, the ACI do not meet the ACI 318-191 longitudinal reinforcement
318-19 limit on the maximum shear strength of beams seems requirements. It may not be appropriate to count different
conservative for beams with a larger amount of longitudinal beam experimental data for extrapolation applications.
tension reinforcement. Therefore, an analytical model is preferred, which is derived
Keywords: force-transfer mechanism; maximum shear strength; shear
according to the shear force-transfer mechanism and verified
failure mode; softened concrete strength. by the experimental data, for the maximum shear strength of
the beam.
INTRODUCTION Due to urbanization, there have been an increasing number
The designed maximum shear strength of reinforced of high-rise buildings, and the harsher environmental loads,
concrete beams should avoid brittle shear failure by concrete such as larger wind speeds or bigger earthquake loadings,
crushing under high shear force, which leads to the sudden have greatly increased the reinforcement used in high-
failure of beams before the shear reinforcement yields. rise buildings, so the beam size may be controlled by the
However, the relevant provisions of the current specifica- maximum shear strength regulations. If the maximum
tions are inconsistent, and the identification of relevant shear strength of the beam is overestimated, there would be
impact parameters is also different. U.S. Code ACI 318-191 safety concerns, but if the regulations are too conservative,
limits the maximum shear strength by √fcꞌ, Japan’s AIJ-992 designs would lose economic efficiency. In addition, high-
uses the maximum amount of shear reinforcement ρt,max to strength concrete is often used in high-rise buildings, but
limit the maximum shear strength, while AASHTO LRFD3 the maximum shear strength regulated by ACI 318-191 is
and CSA A23.3-144 both limit the maximum shear strength similar to ACI 318-63.7 The dated regulations are based on
with fcꞌ. observations of normal-strength concrete beam experiments,
The existing research results suggest different methods and whether they are applicable to existing high-strength
to limit the maximum shear strength of the beam. Lee and concrete beams is precarious. Therefore, it is necessary to
Hwang5 compared experimental observations with existing examine the maximum shear strength limit of the beam and
literature and recommended limiting the maximum shear seek a set of more precise limits.
reinforcement ρt,max to avoid brittle shear failure of beams In this paper, the analytical model of beam shear strength
by concrete crushing. Proestos et al.6 also compared exper- prediction is developed based on the shear force-transfer
imental observations with existing literature and recom- mechanism of the beams, suggested by MacGregor,8 and
mended using the provisions of AASHTO LRFD3 and CSA the application of the softened strut-and-tie model.9,10 It is
A23.3-144 to limit the maximum shear strength with fcꞌ. suggested that the model divide the shear failure of the beam
Both Lee and Hwang5 and Proestos et al.6 suggest that the into three types: the shear compression failure of concrete
maximum shear strength provision in ACI 318-191 is too crushing, the shear tension failure controlled by the trans-
conservative for high-strength concrete beams and should verse reinforcement yielding, and the shear balanced failure
be revised and improved. that occurs when shear compression and shear tension
Using experimental observations to formulate design
equations is a method commonly used in reinforced concrete MS No. S-2020-286.R1, doi: 10.14359/51734375, received August 2, 2021, and
technology. For example, ACI 318-191 and AIJ-992 use Institute. All rights reserved, including the making of copies unless permission is
empirical formulas. However, using experimental observa- obtained from the copyright proprietors. Pertinent discussion including author’s
tion directly to determine the maximum shear strength of the is received within four months of the paper’s print publication.

failures occur at the same time; the maximum shear strength If the beam’s Av is less than the minimum area of shear
of the beam is defined by the shear compression failure. This reinforcement, the size effect factor λs should be applied to
paper collects the existing experimental data to compare modify the shear strength of the beam as follows
with the analytical model and extracts the heavily reinforced
beam specimens to compare with different maximum shear 2
strength limitations, and then discusses the physical signif- s 1 (d in mm) (2)
d
icance of different design parameters. This paper seeks to 1
254
simplify the proposed model to allow easy practical applica-
tion for engineers. 2
s 1 (d in in.)
It is worth noting that the ACI 318-191 limit holds dual d
1
objectives. One is for preventing a compressive failure and 10
the other is for preventing excessive shear crack widths.
The ACI 318-191 upper limit on the shear strength of a
This paper deals with the maximum shear strength of
beam Vmax,ACI is determined as follows
beams to prevent possible sudden shear failure due to over-
reinforcement. However, the cracking behavior of the beams
under service loading is a complicated subject and it is not Vmax, ACI 0.66(w1/ 3 1)bw d f c (fcꞌ in MPa) (3)
within the scope of this paper.
RESEARCH SIGNIFICANCE Vmax, ACI 8(w1/ 3 1)bw d f c (fcꞌ in psi)

The maximum shear strength of beams is defined to
prevent possible sudden failure due to over-reinforcement. AASHTO LRFD3 and CSA A23.3-144 suggest a simple
This paper uses the softened strut-and-tie model to develop linear relationship between the maximum shear strength and
an analytical model for beam shear strength prediction, concrete cylinder strength.
which clearly defines the force-transfer mechanism and
corresponding value of the beam’s maximum shear strength. Vmax, AASHTO 0.25 bw d v f c 0.225 bw d f c (4)
A comparison of the existing experimental data shows that
the prediction results of the proposed model are reasonable. in which dv is the effective shear depth taken as the distance
This paper also discusses the maximum shear strength limita- between the resultants of tensile and compressive forces due
tions of different codes, so that the physical significance of to flexure. It is assumed as dv = 0.9d in this paper.
different design parameters can be clearly understood. Based on the experimental observation of 18 beams and
the analysis of the test results of 178 beams reported in the
EXISTING MODELS OF MAXIMUM SHEAR literature, Lee and Hwang5 suggested that the maximum
STRENGTH shear reinforcement ratio ρt,max should be limited to 0.2fcꞌ/fyt
To facilitate further discussion on the maximum shear because the shear failure mode changes from shear tension to
strength of beams, some existing models are introduced in shear compression failure when the value of ρtfyt/fcꞌ reaches
this section. approximately 0.2. If the value of ρt,max is substituted into
The shear strength of a beam Vn,ACI, according to ACI Eq. (1), the maximum shear strength suggested by Lee and
318-191 can be calculated by Hwang5 can be expressed as
Av f yt d Vmax, Lee (0.66w1/ 3 f c 0.2 f c)bw d (fcꞌ in MPa) (5)

Vn , ACI Vc Vs 0.66 s w1/ 3 f cbw d (fcꞌ in MPa)
s
(1)
Vmax, Lee (8w1/ 3 f c 0.2 f c)bw d (fcꞌ in psi)
Av f yt d
Vn , ACI Vc Vs 8 s w1/ 3 f cbw d (fcꞌ in psi) SHEAR FORCE-TRANSFER MECHANISM OF
s BEAMS
in which Vc is the shear strength provided by concrete; Vs is Reinforced concrete members change in their cross-
the shear strength provided by shear reinforcement; λs is the sectional dimensions, or where concentrated loads and reac-
size effect factor; fcꞌ is the compressive strength of concrete; tion forces act, which causes the internal force transmission
ρw is the ratio of the area of longitudinal tension reinforce- to be disturbed and the phenomenon of stress concentration
ment to bwd; bw is the web width of the beam section; d is the to appear. The range in which the internal force is disturbed
distance from the extreme compression fiber to the centroid is called the D-region,1 and its length is one or two times
of longitudinal tension reinforcement; Av is the area of trans- the member depth from the discontinuity of the cross-
verse reinforcement within spacing s; fyt is the yield strength sectional scale or the location of the concentrated load distur-
of transverse reinforcement; and s is the center-to-center bance. The part of the component in the middle of the two D-
spacing of transverse reinforcement. regions is the B-region,1 and its internal force transmission
has not been disturbed, so it is in a state of uniform stress
distribution.

Fig. 1—Shear force-transfer mechanism of beams.
Based on the force-transfer mechanism proposed by would not occur. However, the diagonal compression in the
MacGregor,8 Fig. 1 shows the shear flows of reinforced fan-shaped struts will cause shearing and tensile yielding
concrete beams under different loading conditions. There of the transverse steel reinforcement due to the insufficient
would be a D-region where the internal force is disturbed tensile strength of the combined effect of the transverse steel
at the loading end and reaction area of the beam. The shear reinforcement and concrete.
force-transfer mechanism within the D-region can be repre- If the transverse reinforcement of the beam is sufficient
sented by a fan-shaped compression field. At the critical and would not yield due to shear force, the diagonal compres-
section for flexure of the beam, the fan-shaped compression sion is transmitted between the cracks along the direction
struts will concentrate in the flexural compression zone of of the initial crack of the beam, which eventually causes
the beam, so it is easier to produce shear compression failure the concrete to be crushed, producing shear compression
of concrete crushing. The fan-shaped compression struts strength Vn,c. Because the beam has no axial force, the initial
are scattered in a more spread-out manner at the border shear cracking angle is approximately 45 degrees. Therefore,
with the B-region, so the phenomenon of concrete crushing when the beam is damaged by shear compression failure, the

tilt angle of the concrete diagonal strut to the horizontal axis 3.35
θ is 45 degrees. This phenomenon can be corroborated by 0.52 (fcꞌ in MPa)
f c
experimental observation. In the shear compression failure
test specimens of Lee and Hwang,5 the tilt angle of the main (8)
diagonal crack is mostly 45 degrees.
If the transverse reinforcement of the beam is insufficient
and starts to yield, but the applied shear force continues to 40
increase, the inclination angle of the main crack will grad- 0.52 (fcꞌ in psi)
f c
ually decrease. This is because if the diagonal compression
strut is inclined, the vertical component of the diagonal
compression is also reduced, and it is therefore easier to seek The effective area at the end of diagonal strut Astr can be
the balance of the internal tensile support force. However, calculated as
if the transverse reinforcement of the beam is indeed insuf-
ficient, the reduction in the tilt angle of the main shear
cracks of the beam is also limited. According to the sugges- Astr bw as (9)
tion from Thürlimann,11 the angle between the shear crack
and the horizontal axis should not be less than tan–1(1/2), where as is the effective depth of the diagonal strut.
because the associated crack strain would be so large that Because shear compression strength is directly propor-
it could lead to a shear failure. Therefore, this paper recom- tional to the effective area at the end of the diagonal strut, the
mends that the minimum angle of the diagonal concrete strut value of as should be carefully determined. As can be seen
shall not be less than 25 degrees. If θ = 25 degrees, and the from Fig. 1(a), there are two fan-shaped patterns of concrete
internal tensile support force of the transverse reinforcement cracking (compression fans) within the D-regions to indi-
of the beam and the concrete is still insufficient to balance cate the disturbed stress fields. The compression fan near the
the vertical component of the diagonal compression, the bearing plate contains a CCC node, resisting three compres-
crack width will continue to increase; then, the shear tension sive forces, which is most likely the location where concrete
strength Vn,t, which is mainly due to transverse steel yielding, crushing occurs. The compression fan near the reaction of
appears. a simply supported beam includes a CCT node, resisting
If the shear compression failure and the shear tensile two compressive forces and one tensile force, which is less
failure of the beam occur at the same time, that is, when likely to have concrete crushing potential. This is because
Vn,c = Vn,t, the inclination angle of the diagonal compression a zero moment at the support reaction and an extended end
strut of the beam should be between 25 and 45 degrees. anchorage of the beam longitudinal tension reinforcement
greatly reduce the stress concentration at the CCT node.13
PROPOSED MODEL FOR SHEAR STRENGTH The major contributing factor of concrete crushing within
PREDICTION the diagonal strut can be attributed to the relatively narrow
This paper proposes to calculate the shear strength for depth of the compression zone due to flexure near the end of
shear tension failure Vn,t using the shear strength Eq. (1) and strut, where the diagonal compressive flow is transformed
(2) of ACI 318-19.1 into concentrated stress and finally leads to the compressive
The shear strength for shear compression failure Vn,c is failure of the concrete. Therefore, the depth of the compres-
based on the softened strut-and-tie model9,10 for evaluation sion zone near the end of the strut is observed to be highly
correlated with the effective depth as.
Vn , c K f cAstr sin (6) The depth of the compression zone (kd) in the beams is
proposed to be calculated with the equation of a singly rein-
where K is the strut-and-tie index; ζ is the softening coeffi- forced beam as
cient of cracked reinforced concrete; Astr is the effective area
at the end of a diagonal strut; and θ is the inclination angle
between the primary diagonal strut and the horizontal axis.
kd w n
2

2w n w n d (10)
The strut-and-tie index K can be used to calculate the effi- where n is the ratio of the elastic modulus of reinforcement
ciency of the secondary strut transferred by transverse rein- to the elastic modulus of concrete. It is noted that compres-
forcements in the D-region of the beam according to10 sion reinforcement in a beam would reduce the depth of
the compression zone, leading to a reduced shear strength
K tan A cot A 1 0.14 B 1.64 (7a) according to the softened strut-and-tie model. This result
seems contrary to what is expected. For this reason, the
authors propose to use Eq. (10) regardless of whether the
t f yt t f yt beam has compression reinforcement.
A 12 1 ; B 30 1 (7b)
f c f c The approaches to determine as near the loading ends are
shown in Fig. 2. For a simply supported beam loaded with
where ρt is the area ratio of transverse reinforcement. two bearing plates (Fig. 2(a)), only half of the bearing plate’s
According to Hwang and Lee,9 the softening coefficient of length (ap/2) is included for making a contribution to as in
cracked reinforced concrete12 ζ can be simplified as addition to kd.13 Because the other half of the bearing plate’s

Fig. 2—Illustration of effective depth of concrete diagonal strut.
length belongs to the constant moment (zero shear) region Equation (8) reveals a special characteristic of high-
where only the horizontal concrete strut can be found, its strength concrete that is the more pronounced behavior
contribution is considered in kd already. For a beam loaded of reinforced concrete softening.12 The crack propagation
with a single bearing plate (Fig. 2(b)), only a quarter of the in high-strength concrete can cut through the aggregates,
bearing plate’s length (ap/4) is counted. It is assumed that which reduces the effect of aggregate interlocking and leads
there is a horizontal concrete strut underneath the bearing to a smaller value of the softening coefficient ζ. Therefore,
plate with a length of ap/2 to facilitate the force transfer for a beam with high-strength concrete (fcꞌ ≥ 42 MPa [6 ksi]),
between two diagonal concrete struts. As for a beam loaded the maximum shear strength can be expressed as
using end blocks (Fig. 2(c)), its end constraint of the concrete
crushing zone is similar to that of a corbel.14 Therefore, only Vmax, SST Vn , c ( 45) 2.70 bw as f c (fcꞌ in MPa)
kd is considered for as.
In the proposed model, the maximum shear strength of an (12)
over-reinforced beam Vmax,SST occurs at the crack inclination
angle of θ = 45 degrees. For θ = 45 degrees and sufficient
transverse reinforcement, Eq. (7) yields the strut-and-tie Vmax, SST Vn , c ( 45) 32.24 bw as f c (fcꞌ in psi)
index of K = 1.14. Therefore, for a beam with normal-
strength concrete (fcꞌ < 42 MPa [6 ksi]), the maximum shear Equation (12) is a function of the square root of the
strength can be expressed as concrete strength, which refers to the more pronounced soft-
ening phenomenon of the high-strength concrete.
Vmax, SST Vn , c ( 45) 0.42 bw as f c (11) Having determined the shear strengths Vn,t = Vn,ACI and
Vmax,SST associated with the two failure modes, the shear
Equation (11) is a function of the concrete strength and strength of the beam Vn,SST is obtained by the following
the longitudinal tension reinforcement ratio. The concrete equation
strength shows that the failure mechanism is crushing of the
concrete, and the longitudinal tension reinforcement ratio =Vn , SST min(
= Vn ,t , Vn , c ) min(Vn , ACI , Vmax , SST ) (13)
indicates the area of the concrete crushing zone.

Fig. 4—Shear strength estimations for beams with shear
span-depth ratios exceeding 2.
is 1.24 with a coefficient of variation of 0.20, and the average
strength ratio Vtest/Vn,ACI of ACI 318-191 is 1.29 with a coef-
ficient of variation of 0.20. Both methods yield reasonable
and comparable predictions because most of the beam spec-
imens exhibited shear tension failure. However, the beams
with shear compression failure (solid squares in Fig. 4) indi-
cate some discrepancy between the two methods. Further
comparisons are made for the experiments on heavily rein-
forced beams.
MODEL COMPARISON ON MAXIMUM SHEAR

STRENGTH
Test specimens with a calculated shear strength (Eq. (1))
Fig. 3—Flowchart of shear strength estimation of beams. larger than the maximum shear strength limited by
ACI 318-191 (Eq. (3)) were extracted from Table 1 and listed
As for a beam having a shear balanced failure, its shear in Table 2. Figure 5 shows the comparison of the test results
strength is Vn,t = Vn,ACI but its inclination angle θ of the from Table 2 with the predictions of the various models. The
primary diagonal strut lies between 25 and 45 degrees. The softened strut-and-tie model (Fig. 3) can reasonably repro-
length of the D-region can be determined by the angle θ duce the 58 test results with an average strength ratio Vtest/
(Fig. 1), which is useful information to estimate displace- Vn,SST of 1.08 and a coefficient of variation of 0.12 (Table 2).
ment.15-17 Figure 3 shows the iteration process of the shear For these heavily reinforced specimens, the proposed model
strength evaluation of beams, which demonstrates how to predicts different failure modes—that is, 37 shear compres-
calculate the shear strength and the associated inclination sion failures, 15 shear balanced failures, and six shear tension
angle θ of three different shear failure modes of beams. failures. The failure mode prediction can be roughly checked
by the review of the inclination angle θ of the primary diag-
EXPERIMENTAL VERIFICATION onal strut through the recorded crack patterns. Figure 6
The predictions of the proposed method are compared shows some specimen photos whose crack patterns seem to
with the 306 test data available in the literature. All of the agree well with the associated θ estimation according to the
beam specimens are screened by the following criteria: proposed model.
1. The test specimens should be non-prestressed beams Shown in Fig. 5, ACI 318-191 produces consistent but
with rectangular cross sections and normalweight concrete. conservative estimations with an average strength ratio
Beams with T-shaped cross sections are excluded. Vtest/Vmax,ACI of 1.37 and a coefficient of variation of 0.16
2. The shear span-depth ratio exceeds two. Deep beams (Table 2). The average strength ratio Vtest/Vmax,AASHTO for the
are excluded. AASHTO LRFD3 predictions is 0.91, and a coefficient of
3. The test specimens were reported to have failed in shear. variation of 0.21 (Table 2), with unconservative estimations.
Table 1 summarizes the references used for comparison. The Lee and Hwang5 method overestimates the test results
Figure 4 shows the comparison of the test results with the with an average strength ratio Vtest/Vmax,Lee&Hwang of 0.86 and
predictions of the proposed model and that of ACI 318-19.1 a coefficient of variation of 0.19 (Table 2).
The average strength ratio Vtest/Vn,SST of the proposed model

Table 1—Collected beam database
Double bearing plates Single bearing plate End blocks
Author No. Author No. Author No.

Mattock and Wang18 1 Mphonde28 99 to 118 Fukuhara and Kokusho40 217 to 244
Elzanaty et al.19 2 to 4 Mattock and Wang18 119 to 121 Kokusho et al.41 245 to 259
Johnson and Ramirez 20
5 to 10 Roller and Russell 29
122 to 128 Takagi et al. 42
260 to 278
Sarsam and Al-Musawi 21
11 to 24 Xie et al. 30
129 to 133 Matsuzaki et al. 43
279 to 287
Tan et al.22 25 to 28 Yoon et al.31 134 to 142 Kagami et al.44 288 to 294
Kong and Rangan 23
29 to 58 Kong and Rangan 23
143 to 148 Iwai et al.45
295 to 300
Ozcebe et al. 24
59 to 70 Collins and Kuchma 32
149 to 154 Nishiura et al. 46
301 to 306
Rahal and Al-Shaleh 25
71 to 79 Shin et al. 33
155 to 162 — —
Rahal 26
80 to 86 Frosch 34
163 to 164 — —
Tan 27
87 to 98 Angelakos et al. 35
165 to 169 — —
— — Tompos and Frosch 36
170 to 173 — —
— — Cladera and Marí37 174 to 185 — —
— — Lee and Kim38 186 to 197 — —
— — Lee and Hwang 5
198 to 214 — —
— — Mihaylov et al. 39
215 to 216 — —
Figure 7(a) shows the effect of the concrete strength fcꞌ amount of ρw. This suggests that the parameter ρw used in
on the estimations of maximum shear strength by various the ACI 318-191 limit (Eq. (3)) might need reconsideration.
models. On the whole, the predictions of the softened strut- Figure 8 presents further study of the effect of the concrete
and-tie model are consistent and reasonable for a broad strength fc' on the estimations of maximum shear strength.
range of concrete strengths. ACI 318-191 produces consis- The ratio of the longitudinal tension reinforcement ρw of
tent but conservative estimations. However, the AASHTO 3.4% is the average value of the beam specimens listed in
LRFD3 and Lee and Hwang5 methods overestimate the Table 2. As shown in Fig. 8, the softened strut-and-tie model
maximum shear strengths of the high-strength concrete with ρw = 3.4% follows the trends of AASHTO LRFD3 and
beams (Fig. 7(a)). This might be attributable to the use of the Lee and Hwang5 for normal-strength concrete beams but
concrete strength rather than the square root of the concrete bends down for the high-strength concrete beams. According
strength as the limiting parameter for the high-strength to Eq. (8), the proposed model suggests that the maximum
concrete beams (Eq. (4) and (5)). On the contrary, the soft- shear strength is a function of the concrete strength fcꞌ or
ened strut-and-tie model and ACI 318-191 adopt the square normal-strength concrete beams with fcꞌ < 42 MPa (6 ksi) and
root of the concrete strength as the primary variable and thus that the maximum shear strength is a function of the square
Eq. (12) and (3) can serve the limiting purpose well for the root of the concrete strength √fcꞌ for high-strength concrete
high-strength concrete beams. It is thus concluded that the beams with fcꞌ ≥ 42 MPa (6 ksi), as shown in Eq. (11) and
more pronounced softening phenomenon of high-strength (12), respectively.
concrete should be considered. If concrete crushing capacity To estimate the concrete crushing capacity for beam shear,
is evaluated, the softened concrete strength ζfcꞌ should be the area of the concrete crushing zone is also an important
used instead of the concrete strength fcꞌ alone. For high- factor, which is related to the amount of longitudinal tension
strength concrete, the parameter of ζfcꞌ is well presented by reinforcement of the beam. For the softened strut-and-tie
√fcꞌ . model, the calculated maximum shear strength decreases as
Figure 7(b) shows the effect of the longitudinal steel the area ratio of the longitudinal tension reinforcement ρw
ratio ρw on the estimations of maximum shear strength by decreases. A new curve of the proposed estimation with the
various models. The softened strut-and-tie model yields value of ρw = 1.7% is shown in Fig. 8. It can be seen from
consistent and reasonable estimations for a broad range of Fig. 8 that the maximum shear strengths according to ρw =
ρw. On the other hand, ACI 318-191 tends to underestimate 1.7% are lower than that of ρw = 3.4%.
the maximum shear strengths of the beams with a larger

Table 2—Details and estimations of 58 experiments on heavily reinforced beams
Test
Test parameters data Estimation
SST Vtest/Vmax,calc
Failure
ρ w, fcꞌ, Vtest, Vtest/ mode+ ACI Lee and
Test setup No. Specimen ID b × h, mm a/h % ρt, % MPa kN Vn,SST (θ) SST 318-191 AASHTO3 Hwang5
Tan27
89 A3 0.84 22 379 0.88 B (43.7°) — 1.09 0.92 0.83
90 A4 1.27 22 433 0.99 C (45°) 0.99 1.24 1.05 0.95
92 B2 0.84 28 445 1.01 T (25°) — 1.12 0.83 0.77
93 B3 1.27 28 533 0.97 C (45°) 0.97 1.34 0.99 0.92
200 × 500 2.5 3.85
94 B4 1.58 28 610 1.11 C (45°) 1.11 1.54 1.14 1.06
96 C2 0.84 32 547 1.22 T (25°) — 1.29 0.89 0.84
97 C3 1.69 32 729 1.18 C (45°) 1.18 1.72 1.18 1.11
98 C4 2.53 32 793 1.29 C (45°) 1.29 1.87 1.29 1.21
Lee and Hwang5
201 F20-4 3.83 0.80 27 667 0.95 B (25.1°) — 1.09 0.82 0.76
350 × 450 2.6
202 F20-5 4.79 1.03 27 836 0.96 C (45°) 0.96 1.34 1.03 0.94
204 F40-3 2.5 2.52 1.02 37 262 0.81 B (38.1°) — 0.99 0.61 0.60
205 F40-5 2.5 3.78 1.43 37 345 0.85 C (45°) 0.85 1.26 0.81 0.77
207 F60-2 2.5 2.55 1.06 63 355 1.02 T (25°) — 1.04 0.50 0.50
208 F60-3 2.5 3.82 1.41 63 437 0.97 B (25.8°) — 1.24 0.61 0.60
209 F60-4 200 × 300 2.3 5.54 1.95 63 491 0.93 C (45°) 0.93 1.46 0.75 0.73
211 F80-2 2.5 3.82 1.49 85 531 1.09 B (25.2°) — 1.30 0.55 0.56
212 F80-3 2.3 5.10 2.01 85 600 1.05 C (45°) 1.05 1.55 0.68 0.67
213 F80-4 2.3 5.54 2.53 85 633 1.08 C (45°) 1.08 1.63 0.72 0.71
214 F80-5 2.3 6.92 3.02 85 760 1.21 C (45°) 1.21 1.91 0.86 0.84
Fukuhara and Kokusho 40
232 (4)-7 5.10 0.27 20 275 1.04 B (44.6°) — 1.17 1.02 0.93
233 (4)-8 180 × 400 3.0 5.54 0.27 20 282 1.07 B (44.6°) — 1.21 1.06 0.96
234 (4)-9 6.92 0.40 20 314 1.17 C (45°) 1.17 1.34 1.17 1.06
Kokusho et al.41
245 B-210-6.0 0.31 20 241 0.87 C (45°) 0.87 1.00 0.86 0.78
246 B-210-7.4 0.48 20 279 1.01 C (45°) 1.01 1.16 1.00 0.91
247 B-210-9.2 0.74 20 322 1.16 C (45°) 1.16 1.34 1.15 1.05
248 B-210-11.0 1.06 20 356 1.28 C (45°) 1.28 1.48 1.27 1.16
251 B-360-6.0 0.31 38 366 1.09 T (25°) — 1.12 0.71 0.68
252 B-360-7.4 180 × 400 3.0 3.16 0.48 38 368 0.80 C (45°) 0.80 1.13 0.71 0.68
253 B-360-9.2 0.74 38 460 1.00 C (45°) 1.00 1.41 0.89 0.85
254 B-360-11.0 1.06 38 510 1.11 C (45°) 1.11 1.57 0.99 0.95
257 B-570-7.4 0.48 54 485 0.95 B (40.1°) — 1.24 0.65 0.64
258 B-570-9.2 0.74 54 549 1.01 C (45°) 1.01 1.41 0.74 0.73
259 B-570-11.0 1.06 54 593 1.09 C (45°) 1.09 1.52 0.80 0.79
Takagi et al.42
266 B-80-046 0.46 34 378 1.06 T (25°) — 1.12 0.74 0.71
200 × 400 4.0 3.09
267 B-80-058S 0.58 34 418 1.02 B (26.0°) — 1.24 0.82 0.78

Table 2 (cont.)—Details and estimations of 58 experiments on heavily reinforced beams
Test
Test parameters data Estimation
SST Vtest/Vmax,calc
Failure
ρw, fcꞌ, Vtest, Vtest/ mode+ ACI Lee and
Test setup No. Specimen ID b × h, mm a/h % ρt, % MPa kN Vn,SST (θ) SST 318-191 AASHTO3 Hwang5
268 B-80-059 0.59 34 462 1.05 B (42.0°) — 1.37 0.91 0.87
269 B-80-110S 1.10 34 528 1.15 C (45°) 1.15 1.56 1.03 0.99
270 B-80-121 4.0 1.21 34 507 1.10 C (45°) 1.10 1.50 0.99 0.95
273 B-120-059 200 × 400 3.09 0.59 35 475 1.01 C (45°) 1.01 1,38 0.90 0.87
274 B-120-121 1.21 35 541 1.15 C (45°) 1.15 1.57 1.03 0.98
277 B-1.5-058 0.58 35 439 1.07 B (25.1°) — 1.27 0.82 0.79
3.0
278 B-1.5-110 1.10 36 532 1.11 C (45°) 1.11 1.53 0.99 0.95
Matsuzaki et al. 43
281 210-0.59 0.59 23 330 1.01 C (45°) 1.01 1.19 0.95 0.88
282 210-0.89 0.89 23 378 1.16 C (45°) 1.16 1.36 1.09 1.01
283 210-1.18 1.19 23 424 1.30 C (45°) 1.30 1.53 1.22 1.13
284 210-1.18-U 200 × 400 3.0 2.88 1.19 23 412 1.26 C (45°) 1.26 1.48 1.19 1.10
285 210-1.18-135 1.19 23 437 1.34 C (45°) 1.34 1.58 1.26 1.17
286 360-0.89 0.89 37 462 0.95 C (45°) 0.95 1.31 0.83 0.80
287 570-0.89 0.89 66 652 1.20 B (25.6°) — 1.39 0.65 0.65
Iwai et al. 45
296 B-4 0.66 51 339 1.17 T (25°) — 1.41 0.76 0.75

297 B-5 1.69 51 479 1.42 C (45°) 1.42 1.99 1.08 1.06
150 × 300 3.0 3.06
299 B-7 0.84 74 435 1.25 B (38.0°) — 1.50 0.67 0.68
300 B-8 1.76 74 471 1.24 C (45°) 1.24 1.63 0.73 0.73
Nishiura et al. 46
S10-M-2.0-
301 0.40 20 271 0.96 B (42.2°) — 1.03 0.88 0.81
21-40-1
S10-M-2.0-
302 0.59 20 291 0.99 C (45°) 0.99 1.11 0.95 0.87
21-59-1
S10-M-2.0-
303 200 × 400 4.0 2.88 0.89 21 371 1.21 C (45°) 1.21 1.39 1.16 1.07
21-89-1
S10-M-2.0-
305 0.59 33 440 1.08 B (40.5°) — 1.33 0.89 0.85
36-59-1
S10-M-2.0-
306 0.89 29 456 1.14 C (45°) 1.14 1.45 1.03 0.98
36-89-1
Average (AVG) 3.41 — — — 1.08 — 1.10 1.37 0.91 0.86
Coefficient of variation (COV) 0.23 — — — 0.12 — 0.13 0.16 0.21 0.19
Note: C is compression failure; B is balanced failure; T is tension failure; 1 mm = 0.0394 in.; 1 MPa = 145 psi; 1 kN = 224.8 lb.
As shown in Fig. 8, it is of interest to note that ACI 318-191 high-rise buildings, it is suggested that a modification factor
yields conservative estimations of the maximum shear for the beams with a larger amount of longitudinal tension
strength for the beams with ρw = 3.4% by the SST estima- reinforcement can be included in the ACI 318-191 limit.
tion, but closer to that of the beams with ρw = 1.7%. The The maximum shear strength of T-shaped beams and deep
ACI 318-191 limit on the maximum shear strength of beams beams are not in the scope of this study. It is noted that the
is often criticized to be overly conservative. Based on this flange can enhance the shear capacity of T-shaped beams,47
study, the over-conservatism might be attributable to the and more studies are needed to address this issue. As for the
over-reinforced longitudinal tension reinforcement, espe- maximum shear strength of the deep beam, the inclination
cially for the laboratory beam tests for shear strength, not angle θ plays an important role,13 and this topic is studied in
due to the high-strength concrete. For the application of a companion paper.48

CONCLUSIONS occur simultaneously and its inclination angle θ lies between
A proposal for determining the shear strength of rectan- 25 and 45 degrees. The proposed model consistently repro-
gular beams with shear span-depth ratios exceeding 2 has duced 306 beam measured shear strengths with reasonable
been made. Three shear failure modes were classified and accuracy.
the associated strength estimations were provided in the 2. The maximum shear strength of beams can be defined
proposed model. The maximum shear strength is defined as the shear compression strength that originates from the
as the crushing of concrete struts prior to the yielding of crushing of concrete struts. The crushing strength should
shear reinforcement. The available test results in the liter- consider the softening phenomenon of cracked reinforced
ature were collected to compare the proposed model and concrete. This paper suggests that the maximum shear
other existing methods. Within the scope of this study, the strength is a function of the concrete strength fcꞌ for normal-
following conclusions can be made: strength concrete beams with fcꞌ < 42 MPa (6 ksi) and that
1. The proposed model suggests that the strength of the the maximum shear strength is a function of the square root
shear tension failure can be estimated by the ACI 318-191 of the concrete strength √fcꞌ for high-strength concrete beams
shear equation and its inclination angle θ of the primary diag- with fcꞌ ≥ 42 MPa (6 ksi).
onal strut is close to 25 degrees. The softened strut-and-tie 3. The longitudinal tension reinforcement of the beam
model is recommended to predict the strength of the shear is an important parameter to estimate the maximum shear
compression failure, and the inclination angle θ at failure is strength because it is related to the area of the concrete
suggested to be 45 degrees. Shear balanced failure means crushing zone. In the proposed model, the calculated
the shear tension failure and the shear compression failure maximum shear strength increases as the area ratio of the
longitudinal tension reinforcement ρw increases.
4. It is noted that the ACI 318-191 limit holds dual objec-
tives. One is for preventing a shear compressive failure and
the other is for preventing excessive shear crack widths. It is
found that the ACI 318-191 limit on shear strength to prevent
a shear compressive failure with a function of √fcꞌ is suitable
to regulate high-strength concrete deep beams, but it seems
conservative for beams with a larger amount of longitudinal
tension reinforcement. It is suggested that a modification
factor for the beams with a larger amount of longitudinal
tension reinforcement can be included in the ACI 318-191
limit, especially for the use of the high-rise buildings. As for
preventing excessive shear crack widths, the ACI 318-191
limit seems to serve well, but a format containing the dual
objectives needs further study.
5. AASHTO LRFD3 overestimates the maximum shear
strengths, especially for the beams with the high-strength
concrete. This is attributed to the use of the concrete strength
rather than the square root of the concrete strength as the
limiting parameter for the high-strength concrete beams.
Fig. 5—Shear strength estimations for experiments on
heavily reinforced beams.
Fig. 6—Comparison of failure mode estimations with cracking photos.

Taiwan University. His research interests include shear behavior of rein-
forced concrete members.
ACKNOWLEDGMENTS
The authors express their gratitude to the Ministry of Science and Tech-
nology of Taiwan and the National Center for Research on Earthquake
Engineering for the supporting funds provided throughout this study.
REFERENCES
Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
2. Architectural Institute of Japan (AIJ), “Standard for Structural Calcu-
lation of Reinforced Concrete Structures,” Architectural Institute of Japan,
Tokyo, Japan, 1999, 412 pp.
3. AASHTO, “LRFD Bridge Design Specifications and Commentary,”
seventh edition, American Association of State Highway Transportation
Officials, Washington, DC, 2014, 2160 pp.
4. CSA A23.3-14, “Design of Concrete Structures,” CSA Group, Toronto,
ON, Canada, 2014, 297 pp.
5. Lee, J.-Y., and Hwang, H.-B., “Maximum Shear Reinforcement
of Reinforced Concrete Beams,” ACI Structural Journal, V. 107, No. 5,
Sept.-Oct. 2010, pp. 580-588.
6. Proestos, G. T.; Bentz, E. C.; and Collins, M. P., “Maximum Shear
Capacity of Reinforced Concrete Members,” ACI Structural Journal,
V. 115, No. 5, Sept. 2018, pp. 1463-1473. doi: 10.14359/51702252
7. ACI Committee 318, “Building Code Requirements for Reinforced
Concrete (ACI 318-63),” American Concrete Institute, Farmington Hills,
MI, 1963, 144 pp.
8. MacGregor, J. G., Reinforced Concrete: Mechanics and Design, third
edition, Prentice Hall Inc., Upper Saddle River, NJ, 1997, 939 pp.
9. Hwang, S.-J., and Lee, H.-J., “Strength Prediction for Disconti-
nuity Regions by Softened Strut-and-Tie Model,” Journal of Structural
Engineering, ASCE, V. 128, No. 12, 2002, pp. 1519-1526. doi: 10.1061/
(ASCE)0733-9445(2002)128:12(1519)
10. Hwang, S.-J.; Tsai, R.-J.; Lam, W.-K.; and Moehle, J. P., “Simplifi-
cation of Softened Strut-and-Tie Model for Strength Prediction of Discon-
Fig. 7—Effect of fcꞌ and ρw on estimations of maximum shear tinuity Regions,” ACI Structural Journal, V. 114, No. 5, Sept.-Oct. 2017,
pp. 1239-1248. doi: 10.14359/51689787
strength. 11. Thürlimann, B., “Shear Strength of Reinforced and Prestressed
Concrete-CEB Approach,” Concrete Design: U.S. and European Prac-
tices, SP-59, American Concrete Institute, Farmington Hills, MI, 1979,
pp. 93-115.
12. Zhang, L.-X. B., and Hsu, T. T. C., “Behavior and Analysis
of 100 MPa Concrete Membrane Elements,” Journal of Structural
Engineering, ASCE, V. 124, No. 1, 1998, pp. 24-34. doi: 10.1061/
(ASCE)0733-9445(1998)124:1(24)
13. Lim, E., and Hwang, S.-J., “Modeling of the Strut-and-Tie Parame-
ters of Deep Beams for Shear Strength Prediction,” Engineering Structures,
V. 108, 2016, pp. 104-112. doi: 10.1016/j.engstruct.2015.11.024
14. Hwang, S.-J.; Lu, W.-Y.; and Lee, H.-J., “Shear Strength Prediction
for Reinforced Concrete Corbels,” ACI Structural Journal, V. 97, No. 4,
July-Aug. 2000, pp. 543-552.
15. Li, Y.-A., and Hwang, S.-J., “Prediction of Lateral Load Displacement
Curves for Reinforced Concrete Short Columns Failed in Shear,” Journal
of Structural Engineering, ASCE, V. 143, No. 2, 2017, p. 04016164. doi:
10.1061/(ASCE)ST.1943-541X.0001656
16. Li, Y.-A.; Weng, P.-W.; and Hwang, S.-J., “Seismic Performance
of Reinforced Concrete Intermediate Short Columns Failed in Shear,”
ACI Structural Journal, V. 116, No. 3, May 2019, pp. 195-206. doi:
10.14359/51713309
Fig. 8—Model comparison of maximum shear strength. 17. Shen, W.-C.; Hwang, S.-J.; Li, Y.-A.; Weng, P.-W.; and Moehle,
J. P., “Force-Displacement Model for Shear-Critical Reinforced Concrete
AUTHOR BIOS Columns,” ACI Structural Journal, V. 118, No. 1, Jan. 2021, pp. 241-249.
Shyh-Jiann Hwang, FACI, is a Professor of civil engineering at National 18. Mattock, A. H., and Wang, Z., “Shear Strength of Reinforced
Taiwan University, Taipei, Taiwan. He received his PhD from the University Concrete Members Subject to High Axial Compressive Stress,” ACI
of California, Berkeley, Berkeley, CA. He is a member of Joint ACI-ASCE Journal Proceedings, V. 81, No. 3, May-June 1984, pp. 287-298.
Committee 352, Joints and Connections in Monolithic Concrete Structures. 19. Elzanaty, A. H.; Nilson, A. H.; and Slate, F. O., “Shear Capacity of
His research interests include seismic behavior of reinforced concrete Reinforced Concrete Beams Using High-Strength Concrete,” ACI Journal
members and seismic retrofitting of reinforced concrete structures. Proceedings, V. 83, No. 2, Mar.-Apr. 1986, pp. 290-296.
20. Johnson, M. K., and Ramirez, J. A., “Minimum Shear Reinforcement
Yu-Hsuan Yang is a Structural Engineer at King-Le Chang & Associates, in Beams with Higher Strength Concrete,” ACI Structural Journal, V. 86,
Taipei, Taiwan. She received her MS in civil engineering from National No. 4, July-Aug. 1989, pp. 376-382.
Taiwan University. Her research interests include the shear behavior of 21. Sarsam, K. F., and Al-Musawi, J. M. S., “Shear Design of High- and
reinforced concrete beams. Normal Strength Concrete Beams with Web Reinforcement,” ACI Struc-
tural Journal, V. 89, No. 6, Nov.-Dec. 1992, pp. 658-664.
Yi-An Li is an Assistant Professor of civil engineering at National Chung
Hsing University, Taichung, Taiwan. He received his PhD from National

22. Tan, K.-H.; Teng, S.; Kong, F.-K.; and Lu, H.-Y., “Main Tension 36. Tompos, E. J., and Frosch, R. J., “Influence of Beam Size, Longitu-
Steel in High-Strength Concrete Deep and Short Beams,” ACI Structural dinal Reinforcement, and Stirrup Effectiveness on Concrete Shear Strength,”
Journal, V. 94, No. 6, Nov.-Dec. 1997, pp. 752-764. ACI Structural Journal, V. 99, No. 5, Sept.-Oct. 2002, pp. 559-567.
23. Kong, P. Y. L., and Rangan, B. V., “Shear Strength of High- 37. Cladera, A., and Marí, A. R., “Experimental Study on High-Strength
Performance Concrete Beams,” ACI Structural Journal, V. 95, No. 6, Concrete Beams Failing in Shear,” Engineering Structures, V. 27, No. 10,
Nov.-Dec. 1998, pp. 677-688. 2005, pp. 1519-1527. doi: 10.1016/j.engstruct.2005.04.010
24. Ozcebe, G.; Ersoy, U.; and Tankut, T., “Evaluation of Minimum 38. Lee, J.-Y., and Kim, U.-Y., “Effect of Longitudinal Tensile Rein-
Shear Reinforcement Requirements for Higher Strength Concrete,” ACI forcement Ratio and Shear Span-Depth Ratio on Minimum Shear Rein-
Structural Journal, V. 96, No. 3, May-June 1999, pp. 361-369. forcement in Beams,” ACI Structural Journal, V. 105, No. 2, Mar.-Apr.
25. Rahal, K. N., and Al-Shaleh, K. S., “Minimum Transverse Reinforce- 2008, pp. 134-144.
ment in 65 MPa Concrete Beams,” ACI Structural Journal, V. 101, No. 6, 39. Mihaylov, B. I.; Bentz, E. C.; and Collins, M. P., “Behavior of Large
Nov.-Dec. 2004, pp. 872-878. Deep Beams Subjected to Monotonic and Reversed Cyclic Shear,” ACI
26. Rahal, K. N., “Shear Behavior of Reinforced Concrete Beams with Structural Journal, V. 107, No. 6, Nov.-Dec. 2010, pp. 726-734.
Variable Thickness of Concrete Side Cover,” ACI Structural Journal, 40. Fukuhara, M., and Kokusho, S., “Effectiveness of High Tension Shear
V. 103, No. 2, Mar.-Apr. 2006, pp. 171-177. Reinforcement in RC Members: Beams Subjected to Bending Moment and
27. Tan, S., “Maximum Amount of Shear Reinforcement of Reinforced Shear Force,” Journal of Structural and Construction Engineering, Archi-
Concrete Beams,” master’s thesis, Department of Civil Engineering, tectural Institute of Japan, No. 320, 1982, pp. 12-20.
National Taiwan University, Taipei, Taiwan, 2010, 269 pp. 41. Kokusho, S.; Kobayashi, K.; Mitsugi, S.; and Kumagai, H., “Ulti-
28. Mphonde, A. G., “Shear Strength of High-Strength Concrete Beams,” mate Shear Strength of RC Beams with High Tension Shear Reinforcement
PhD thesis, University of Connecticut, Storrs, CT, 1984, 249 pp. and High Strength Concrete,” Journal of Structural and Construction Engi-
29. Roller, J. J., and Russell, H. G., “Shear Strength of High-Strength neering, Architectural Institute of Japan, No. 373, 1987, pp. 83-91.
Concrete Beams with Web Reinforcement,” ACI Structural Journal, V. 87, 42. Takagi, H.; Okude, H.; and Nitta, T., “Shear Strength of Beam
No. 2, Mar.-Apr. 1990, pp. 191-198. Depending the Strength of Web Reinforcements,” JCI Proceedings, V. 11,
30. Xie, Y.; Ahmad, S. H.; Yu, T.; Hino, S.; and Chung, W., “Shear No. 2, 1989, pp. 75-80.
Ductility of Reinforced Concrete Beams of Normal and High-Strength 43. Matsuzaki, Y.; Nakano, K.; Iso, M.; and Watanabe, H., “Experimental
Concrete,” ACI Structural Journal, V. 91, No. 2, Mar.-Apr. 1994, Study on Shear Characteristic of RC Beams with High Tension Shear Rein-
pp. 140-149. forcement,” JCI Proceedings, V. 12, No. 2, 1990, pp. 325-328.
31. Yoon, Y.-S.; Cook, W. D.; and Mitchell, D., “Minimum Shear Rein- 44. Kagami, Y.; Mazuhara, H.; Takagi, H.; and Karino, Y., “Splitting
forcement in Normal, Medium, and High-Strength Concrete Beams,” ACI Bond Failure of a Double Reinforced Concrete Beam, Part 1,” Summaries
Structural Journal, V. 93, No. 5, Sept.-Oct. 1996, pp. 576-584. of Technical Papers of Annual Meeting, Architectural Institute of Japan,
32. Collins, M. P., and Kuchma, D., “How Safe Are Our Large, Lightly 1991, pp. 353-354.
Reinforced Concrete Beams, Slabs, and Footings?” ACI Structural Journal, 45. Iwai, Y.; Enomoto, H.; Watanabe, F.; and Muguruma, H., “Shear
V. 96, No. 4, July-Aug. 1999, pp. 482-491. Strength of High Strength Concrete Beams, Part 1,” Summaries of Tech-
33. Shin, S.-W.; Lee, K.-S.; Moon, J.-I.; and Ghosh, S. K., “Shear nical Papers of Annual Meeting, Architectural Institute of Japan, 1991,
Strength of Reinforced High-Strength Concrete Beams with Shear Span-to- pp. 269-270.
Depth Ratios between 1.5 and 2.5,” ACI Structural Journal, V. 96, No. 4, 46. Nishiura, N.; Makitani, E.; and Shindou, K., “Shear Resistance of
July-Aug. 1999, pp. 549-556. Concrete Beams with High Strength Web Reinforcements,” JCI Proceed-
34. Frosch, R. J., “Behavior of Large-Scale Reinforced Concrete Beams ings, V. 15, No. 2, 1993, pp. 461-466.
with Minimum Shear Reinforcement,” ACI Structural Journal, V. 97, 47. Li, Y.-A.; Hsu, T. T. C.; and Hwang, S.-J., “Shear Strength of
No. 6, Nov.-Dec. 2000, pp. 814-820. Prestressed and Nonprestressed Concrete Beams,” Concrete International,
35. Angelakos, D.; Bentz, E. C.; and Collins, M. P., “Effect of Concrete V. 39, No. 9, Sept. 2017, pp. 53-57.
Strength and Minimum Stirrups on Shear Strength of Large Members,” ACI 48. Hwang, S.-J.; Yang, Y.-H.; and Li, Y.-A., “Maximum Shear Strength
Structural Journal, V. 98, No. 3, May-June 2001, pp. 290-300. of Reinforced Concrete Deep Beams,” ACI Structural Journal, V. 118,
No. 6, Nov. 2021, pp. 155-164.

Title No. 119-S27
Long-Term Behavior of Glass Fiber-Reinforced Polymer-

Reinforced Columns under Multi-Axial Loading
by Yail J. Kim
This paper presents the time-dependent behavior of non-slender of great concern. Epoxy-coated steel bars are frequently used
concrete columns reinforced with glass fiber-reinforced polymer to address durability requirements, whereas the functional
(GFRP) bars under concentric, uniaxial, and biaxial loadings. life of these bars is limited, and corrosion problems take
A service-level load is applied at 28 days after concrete casting place.4 As an alternative solution, steel reinforcement may
and sustained for up to 100 years. Conforming to force equilib-
be replaced by noncorrosive glass fiber-reinforced polymer
rium and displacement compatibility, a sectional analysis model
(GFRP) bars. GFRP composites have been broadly adopted
is developed in conjunction with a mathematical procedure, called
Bayesian inference, to address potential uncertainty. The formu- for the construction of flexural members5; on the contrary,
lated model is validated against published experimental data and their application to axial load-bearing members received
is employed to predict the columns’ response with variable eccen- restricted attention owing to the conservative nature of struc-
tricities, loading angles, concrete strengths, and section losses. tural design.6-8 Following early endeavors in the 2000s,9 the
When the short-term loading is shifted from concentric to uniaxial- research community recently commenced to reexamine the
eccentric, abrupt changes are noticed in the concrete and GFRP potential of GFRP-reinforced concrete columns, and active
strains. Biaxial loading brings about unfavorable distress investigations are underway on various subjects.10,11 Choo
compared with its uniaxial counterpart. Excessive biaxial loading et al.9 studied the behavior of concrete columns reinforced
angles between 30 and 40 degrees, measured from the horizontal with GFRP bars at variable slenderness ratios. Due to the
axis, swiftly raise the short-term concrete strains. Regarding the
absence of a yield plateau, balanced sections that are arche-
characteristics of the constituent materials, early-age responses
typal in steel-reinforced columns were not identified in the
are remarkable during the 2 months after loading. Under the
eccentric loadings, long-term concrete stresses are redistributed, GFRP-reinforced columns. A failure mode referred to as
and an irregular pattern is observed in the compression region for brittle tension was dominant in the GFRP columns when a
the first 10 years. Furthermore, creep and shrinkage relieve the reinforcement ratio was less than 1%. The need for under-
concrete stresses over time. The elastic and shrinkage components standing the long-term deformation of GFRP-reinforced
in the long-term concrete strains are appreciable with the uniaxial columns was raised. De Luca et al.6 tested full-scale concrete
and biaxial loadings, whereas the creep component accounts for columns reinforced with GFRP bars. Despite the GFRP’s
less than 1% of the total strains for 100 years. With an increase in linear-elastic attribute, the columns demonstrated gradual
the concrete strength, GFRP strains decrease; however, the rein- stiffness reductions and sufficient ductility in the post-peak
forcing bar strains rise as the concrete section spatially damages. behavior. There was an insignificant discrepancy between
When the column is biaxially loaded, the vulnerability of creep
the load-carrying capacities of the columns with GFRP
rupture is elevated in the GFRP reinforcement relative to the case
and steel bars. Afifi et al.12 evaluated the axial capacity of
of the uniaxial loading.
GFRP-reinforced circular columns against that of conven-
Keywords: fiber-reinforced polymer (FRP); long-term; modeling; sustained tional steel-reinforced columns. Within a reinforcement ratio
loading; time-dependent behavior. of 1.7 to 2.2%, the capacities of these columns differed by
7.0% on average. The contribution of GFRP to the axial
INTRODUCTION capacity was less than 10%, implying that the performance
Structural columns, indispensable to resist dead and live of reinforced concrete columns may not be reliant upon
loads from upper stories, transfer axial loads to the foun- reinforcing type. Moreover, the ductility and deformability
dation of a frame system. In practice, columns are eccen- of both column configurations were acceptable as long as
trically loaded and carry compression and bending distress. adequate transverse ties were provided. Hasan et al.13 carried
Sustained loading augments adverse stress in concrete out a numerical study on the moment-curvature relationship
members over time and can degrade serviceability and of GFRP-reinforced columns and reported that ductility of
load-carrying capacity. A number of factors are simultane- the columns was controlled by bending stiffness.
ously engaged in determining the state of existing concrete On most occasions, experimental testing and analytical
columns, including elastic, creep, and shrinkage compo- modeling in regard to GFRP-reinforced concrete columns
nents.1 Because of intrinsic complexities associated with the focused on short-term behavior under either uniaxial-
microstructural dislocation of a cement binder surrounding eccentric or concentric loading.7,10,14 One of the principal
aggregates,2 the time-dependent material characteristics of
concrete are difficult to quantify; as a result, corresponding MS No. S-2020-321, doi: 10.14359/51732649, received August 4, 2020, and
design expressions are largely empirical.3 reviewed under Institute publication policies. Copyright © 2022, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
When concrete structures are exposed to aggressive service obtained from the copyright proprietors. Pertinent discussion including author’s
environments, their longevity and maintenance expense are is received within four months of the paper’s print publication.

gaps between the state-of-the-art and actual practice is long- where θ and D are the parameter and data, respectively;
term performance with multi-axial eccentric loadings, which p(D/θ) is the likelihood function; and p(θ) is the prior proba-
is an important aspect for columns that permanently carry bility of the parameter. The posterior is interpreted as a prob-
assorted gravity loadings. This paper deals with the short- ability distribution of the parameter θ, given the data D. The
term and time-dependent responses of non-slender GFRP- prior information is updated in combination with the likeli-
reinforced concrete columns subjected to concentric, hood function. This simulation technique is useful to infer
uniaxial, and biaxial loadings. Theoretical models are formu- the outcomes of uncertain data, such as the time-dependent
lated and validated against published test data to predict the behavior of structural members. Because the denominator of
columns’ behavior. Through a parametric investigation, the the Bayesian formula is a normalization constant, Eq. (1) is
effects of material and geometric properties of the concrete abridged to determine the most probable value of the param-
section are accounted for, and the creep rupture of GFRP is eter in certain conditions
assessed from a design perspective. The scope of the current
study is on the use of longitudinal GFRP reinforcement p / D p D / p (2)
rather than transverse ties.
RESEARCH SIGNIFICANCE ANALYTICAL MODELING

In all circumstances, columns are subjected to axial The time-dependent properties of constituent materials
compression and bending moments at variable eccentrici- are delineated along with the behavior of GFRP-reinforced
ties, both of which can be sustained during a considerably concrete columns under short-term and sustained loadings.
long period. Accordingly, an adequate understanding of An analytical model is formulated based on the adjusted
the consequences of such multi-axial loadings is a prereq- effective modulus method16,17 alongside fundamental
uisite for the development of design guidelines concerning mechanics,18 as outlined in Fig. 1, to predict the short- and
GFRP-reinforced concrete columns. Aligning with contem- long-term responses of the columns when subjected to
porary efforts on expanding the application boundaries concentric, uniaxial, and biaxial loading conditions.
of GFRP bars, this research aims to elucidate the time-
dependent behavior of GFRP-reinforced columns under Materials
multi-axial eccentricities, which has not been studied thus Concrete—The strain of concrete under a constant load
far. Technical findings are expected to enrich a revised sustained for a certain period (εc(t)) is composed of
version of ACI 440.1R-155 that does not contain provisions
on vertical load-bearing members. c t e t cr t sh t (3)
BAYESIAN INFERENCE
Bayes’ theorem for predicting the posterior distribution of c 0
a parameter (p(θ/D)) is shown as follows15 e t (4)
Ee t
p D / p
p / D (1) c t c 0
p D / p d cr t (5)
Ee t
tdr
sh t shu (6)
f tdr
where εe(t), εcr(t), and εsh(t) are the elastic, creep, and
shrinkage strains, respectively; σc(0) and σc(t) are the
concrete stresses at the initial and long-term conditions,
respectively; tdr is the time from the end of initial curing; f
and α are the shrinkage constants (f = 35 and α = 1.0 for
7-day moisture-cured concrete3); and εshu is the ultimate
shrinkage strain (εshu = 780 × 10–6 for a typical service envi-
ronment3). It is assumed that the shrinkage is uniform across
the concrete section.19,20 The elastic moduli of concrete at
variable times may be obtained from3,21
Ec 0 57, 000 f c in psi (7a)
Ec 0 4730 f c in MPa (7b)
Fig. 1—Flowchart of modeling approach.

Fig. 2—Column section and notation: (a) coordinate system and strain diagrams; (b) loading schemes; and (c) benchmark
section. (Note: Units in mm; 1 mm = 0.0394 in.)
Ec 0 E f 0
Ee t (8) E fe t (15)
1 t t t0
k2
1 k1

f0
Ec 0
Ee t (9) Formulation
1 t t
Short-term behavior—The GFRP-reinforced column
where Ec(0) is the initial modulus of elasticity with concrete depicted in Fig. 2 is supposed to carry axial compression
strength fc′; and Ee(t) and Ee′(t) are the effective and the and bending moments consisting of Eq. (16) to (18)
age-adjusted effective moduli at time t, which are a function
of the creep (Eq. (10)) and aging (Eq. (11)) coefficients3,22 N dA N c N f (16)
A
t t0
0.6
t u (10) M x ydA M xc M xf (17)

10 t t0
0.6
A
t00.5 M y xdA M yc M yf (18)

t (11) A
1 t00.5
where N is the applied axial load; and Mx and My are the
where t0 is the time when the load is applied; and φu is the applied moments about the horizontal and vertical axes with
ultimate creep coefficient (φu = 2.35). eccentricities y and x, respectively; σ is the normal stress of
GFRP reinforcement—The effective elastic modulus of the section; Nc and Nf are the axial resistance by the concrete
GFRP (Efe(t)) may be expressed as23 and the GFRP bars, respectively; and Mc and Mf are the
moment resistance by the concrete and GFRP, respectively.
E f 0 For convenience, time indication is omitted in the deriva-
E fe t (12)
1 f t tion of short-term expressions (for example, N = N(0)).
Integrating the strain compatibility of the section (Eq. (19),
Fig. 2(a)) with the axial load and moment terms (Eq. (16) to
f t
f t 1 (13) (18)) at a point on the column section with the coordinate of
f 0 x and y (Fig. 2(b))
where Ef(0) and εf(0) are the initial elastic modulus and strain
of GFRP, respectively; φu(t) is the GFRP creep coefficient; 0 x y y x (19)
and εf(t) is the long-term strain that can be calculated by24
k2
N c Ec dA (20)
t Ac
f t 1 k1 cr f 0 (14)

f0 N fi
where tcr is the creep time (tcr = t – t0); τf0 is the reference time N f Afi E f (21)
i 1
(τf0= 1.0); and k1 and k2 are material constants (k1 = 0.48 and
k2 = 0.41 for GFRP bars under compression, respectively).
Combining Eq. (12) to (14) yields M xc yc dA yEc dA (22)
Ac Ac

M yc xc dA xEc dA (23) For the convenience of computation, the elements of the [M]
Ac Ac matrix in Eq. (38) are converted to those with transformed
properties
N fi
M xf y fi Afi E f (24) AcT QxT QyT
i 1
T QxT
M I xT

I xyT Ec (39)
N fi QyT I xyT I yT

M yf x fi Afi E f (25)
i 1 N fi
AcT Ac n f 1 Afi (40)
where Afi is the cross-sectional area of the i-th GFRP bar at i 1
the coordinate of xfi and yfi; and Nfi is the number of the bars.
Substituting Eq. (20) through (25) into Eq. (16) to (18) N fi
QxT Qx n f 1 Afi y fi (41)
N M 110 M 12 x M 13 y (26) i 1
N fi
M x M 210 M 22 x M 23 y (27) QyT Qy n f 1 Afi x fi (42)

i 1
N fi
M y M 310 M 32 x M 33 y (28) I xyT I xy n f 1 Afi x fi y fi (43)
i 1
N fi
N fi
M 11 Ec Ac Afi E f (29)
i 1
I xT I x n f 1 Afi y 2fi (44)
i 1
N fi
N fi
M 12 Ec Ac y Afi E f y fi (30)
i 1
I yT I y n f 1 Afi x 2fi (45)
i 1
where Q and I are the first and second moments of area,

N fi
M 13 Ec Ac x Afi E f x fi (31) respectively (QxT = QyT = IxyT = 0 when the origin of the
i 1 coordinate system equals the centroid of the section in a
symmetric column). Because the applied loading generates
N fi service-level responses, the linear constitutive relationship
M 21 Ec Ac y Afi E f y fi (32) of the concrete and the i-th GFRP bars is used to determine
i 1
their stresses in conjunction with Eq. (38).
Long-term behavior—To predict the behavior of the
N fi
column under sustained loading, the afore-derived formu-
M 22 Ec Ac y 2 Afi E f y 2fi (33)
i 1 lation is modified. Upon adding the time-dependent compo-
nents to Eq. (38), the long-term strain is given as
N fi
M 23 Ec Ac xy Afi E f x fi y fi (34) M 11t M 12t M 13t
1
N 1 Ac 0 0
i 1 t t
1 y x M 21 t
M 22 M 23 M x 0 J sh t I
xc x 0 J CR t
M 31t M 32t M 33t M 0 I
0
y yc y
N fi
M 31 Ec Ac x Afi E f x fi (35) (46)
i 1
N fi J sh t Ee t Ac sh t (47)
M 32 Ec Ac xy Afi E f x fi y fi (36)
i 1
t t 1
J CR t Ec 0
N fi
M 33 Ec Ac x 2 Afi E f x 2fi (37) (48)
i 1
1 t t
A summary of the aforementioned terms is given as a

matrix form Ac Ag Af (49)
1
M 11 M 12 M 13 N
bh3 N fi
1 y x M 21 M 22 M 23 M x (38) I xc Afi y 2fi (50)
12 i 1
M 31 M 32 M 33 M
y
hb3 N fi for brevity). The converged simulation results were then
I yc Afi x 2fi (51) incorporated to determine the elastic modulus of concrete
12 i 1
(Eq. (7), Fig. 3(b)) that was necessary for calculating the
The elements of the time-dependent matrix [Mt] are the same short- and long-term responses of the column. Likewise, the
as those enumerated in Eq. (29) to (37), except that Ec and Ef Bayesian inference procedure was performed with all other
are replaced by Ee′(t) and Efe(t), respectively. Similarly, the parameters to address potential uncertainties.
effective modular ratio (nf′ = Ef(t)/ Ee′(t)) is used in place of
the modular ratio (nf) for Eq. (40) to (45) alongside Eq. (39) Validation
with Ee′(t). It is worth noting that Eq. (49) to (51) are based Figure 4 shows the validation of the modeling approaches
on an untransformed section to determine the cross-sectional against experimental data. Due to the absence of long-term
area and the moment of inertia of the concrete (Ac, Ixc, and testing in the literature, short-term column responses were
Iyc). The stress of the concrete section at time t (σc(t)) is evaluated under concentric10,14 and uniaxial-eccentric7,25
calculated using loadings. The material and geometric properties of the spec-
imens are listed in Table 2. It should be noted that the cover
J CR t depth was assumed to be 20 mm (0.8 in.) when missed and
c t Ee sh c 0 (52) that the elastic modulus of concrete was calculated as per
Ec 0
ACI 318-19.21 Figures 4(a) and (b) are concerned with the
behavior of the concrete and GFRP subjected to concentric
Application of Bayesian theory compression. The predicted strains agreed with the measured
In accordance with Eq. (2), the posterior distributions of values up to 40% of the ultimate capacity, beyond which
the individual parameters constituting a GFRP-reinforced a bifurcation was noticed because of nonlinear damage in
concrete column are theoretically inferred using the statistical the columns. Given that the present modeling is coupled
properties enumerated in Table 1. The prior distribution of a with a service load level, which is related to the ordinary
parameter that represents existing knowledge was defined operation of column members, such discrepancy would not
by its own distribution type and the coefficient of variation cause a problem from a practical standpoint. The compres-
(Table 1). The likelihood distribution was constructed on the sive and tensile strains of the reinforcement under eccen-
basis of Monte Carlo simulations, which randomly generate tric loading are plotted in Fig. 4(c) and (d). Similar to the
parameter values. Thereafter, the most probable value concentric-loading case, good agreement was made between
acquired from the parameter’s posterior distribution was the measured and calculated strains within the service range.
sequentially updated in tandem with the analytical equations
detailed in the preceding sections. Figure 3(a) describes IMPLEMENTATION
randomly sampled data for concrete strength fc′ and corre- Employing the formulated models, an extensive analytical
sponding posterior distributions (selected graphs are shown study is conducted to explore the implications of consti-
tuting variables on the time-dependent behavior of GFRP-
Table 1—Statistical properties of major parameters reinforced concrete columns loaded in concentric, uniaxial,
and biaxial manners. The prediction range spans up to 100
Parameter Distribution COV Reference
years, and long-term loading is applied at 28 days after
Compressive strength of concrete casting. It is again noted that the applicability of the
Normal 0.125 Nowak and Collins28
concrete, fc′
models is limited to non-slender columns without cracking.
Elastic modulus of GFRP,
Normal 0.037 You et al.29
Ef Benchmark column
Geometry of concrete
Normal 0.03 Okeil et al.30
A square column designed by Zadeh and Nanni26 was
section taken as the default member for numerical investigations.
Geometry of GFRP Normal 0.015 Kulkarni31 The column section, consisting of 500 x 500 mm (20 x
Fig. 3—Bayesian inference: (a) posterior distributions for concrete strength; and (b) predicted elastic modulus of concrete.

Fig. 4—Validation of modeling approach (line is test; circle is prediction; units in mm; 1 mm = 0.0394 in.): (a) concentric
loading14; (b) concentric loading10; (c) eccentric loading25; and (d) eccentric loading.7
Table 2—Properties of GFRP-reinforced concrete columns used for validation
GFRP elastic modulus, Number of Total cross-sectional
Reference Width, b Height, h Concrete strength, fc′ Ef GFRP bars area of bars
Tobbi et al.14 350 mm (13.8 in.) 350 mm (13.8 in.) 35 MPa (5080 psi) 46,300 MPa (6320 ksi) 8 1016 mm2 (1.57 in.2)
Sun et al.25 180 mm (7.1 in.) 250 mm (9.8 in.) 33.51 MPa (4860 psi) 60,200 MPa (8730 ksi) 6 471 mm2 (0.73 in.2)
Guerin et al.7 405 mm (16.0 in) 405 mm (16.0 in.) 42.3 MPa (6140 psi) 51,300 MPa (7440 ksi) 6 1710 mm2 (2.65 in.2)
Tu et al.10 200 mm (7.9 in.) 200 mm (7.9 in.) 32.1 MPa (4660 psi) 44,250 MPa (6420 ksi) 4 456 mm2 (0.71 in.2)
20 in.), was intended to carry structural loads specified in to concentric and uniaxial-eccentric loadings, respectively.
ASCE/SEI 7-1027 with the following properties: compres- As stated earlier, the level of eccentricity was increased
sive strength, elastic modulus, and modulus of rupture from 0%h (concentric) to 20%h (eccentric), where h is the
for concrete were fc′ = 35 MPa (5080 psi), Ec = 28 GPa dimension of the column section (Fig. 5(a), inset), until
(4060 ksi), and fr = 3.7 MPa (535 psi), respectively, and the the concrete stress reached the preset limit of either 40%fc′
elastic modulus of GFRP was Ef = 46 GPa (6670 ksi). The or fr. The sign convention used herein is that positive and
ultimate strength and rupture strain of GFRP adopted herein negative strains represent tension and compression, respec-
were ffu = 1030 MPa (150 ksi) and εfu = 2.24%, respectively, tively. Under the concentric loading (0%h in Fig. 5(a)), the
which were within the bounds of generic GFRP properties concrete strain was –106 × 10–6 at all corners; however, a
stated in ACI 440.1R-15.5 The primary reinforcement was bifurcation was noticed with the increased eccentricity. The
eight No. 8 GFRP bars (cross-sectional area [Af] = 510 mm2 response slope of Corners B and C differed between 0%h
[0.79 in.2] each), as depicted in Fig. 2(c), and discrete No. 3 and 5%h when a transition occurred from compression to
ties (cross-sectional area [At] = 71 mm2 [0.11 in.2] each) were tension; afterward, their strain profiles were mirror imaged.
placed at a spacing of 200 mm (8 in.). For consistency with It should be noted that strain magnitudes at Corners B and C
the validation section, the service structural loading was set were the same as those at Corners D and A on account of the
to 750 kN (170 kip), which was 40% of the axial load at the symmetric behavior under the uniaxial loadings. The GFRP
ultimate limit state. A variety of loading schemes (Fig. 2(b)) strain was uniform at 0%h, as in the case of the concrete
were applied at incremental eccentricities within a concrete strain, and began to rotate about the geometric centroid
stress range varying from 0% to 40%fc′ in compression. (Fig. 5(b)). These observations align with the fact that plain
sections remain plain. The reinforcing bar in the tension zone
Short-term behavior (Bar No. 1 at abscissa = –187 mm [–7.4 in.]) experienced an
Multi-axial loadings—Figures 5(a) and (b) exhibit the abrupt increase in strain due to the presence of eccentricity
concrete and reinforcing bar strains of the column subjected (5%h). Concrete strains resulting from the biaxial loadings

Fig. 5—Short-term behavior: (a) concrete strain under uniaxial loading; (b) GFRP strain under uniaxial loading; (c) concrete
strain under biaxial loading at θ = 45 degrees; and (d) GFRP strain under biaxial loading at θ = 45 degrees.
are given in Fig. 5(c). Despite the analogous tendency, the section (Bar No. 3 and 8, respectively) were affected by the
maximum level of eccentricity was 10%h, which indicates vertical eccentricity; in contrast, the strain of Bar No. 5 at
the significance of biaxial loading in comparison with its the centroid was constant owing to the invariant eccentricity
uniaxial counterpart. Besides, the axis of sectional rotations about the vertical axis. Even when the distance between
in the square column shifted from the geometric centroid the loading point and reinforcing bar location dominated
to the location of Bar No. 1 (abscissa = 0 mm [0 in.] and the strain values (Fig. 6(d), inset), their interaction rate did
–187 mm [–7.4 in.], respectively), as shown in Fig. 5(d). not change.
Loading angle—The influence of variable eccentricities
associated with the biaxial loading is plotted in Fig. 6(a). To Long-term behavior
reposition a loading point from the abscissa, the horizontal Material-level response—The time-dependent charac-
eccentricity was fixed at the maximum value of 10%h, and teristics of the constituents are summarized in Fig. 7. The
the vertical eccentricity was increased, leading to an angular creep coefficient of concrete exponentially ascended and
change from 0 to 40 degrees (Fig. 6(a), inset). The strains at converged to φ(t) = 2.31 at 100 years (Fig. 7(a)), while the
Corners A and B decreased algebraically with the increased aging coefficient remained constant at χ(t) = 0.841 (Fig. 7(a),
angle, whereas the strains at Corners C and D showed an inset). The shrinkage of concrete demonstrated an asymp-
opposite trend because the plane of the sectional rotation totical strain pattern with time (Fig. 7(b)); specifically,
was in between Corners B (compression) and C (tension). the difference between 28 days and 1 year was a factor of
The dependency of the compressive concrete strains on the 1.43, which was higher than the difference of 0.09 between
biaxial eccentricity is graphed in Fig. 6(b). The fact that the 1 year and 100 years, representing sufficient hydration of the
gradually increasing strain ratio between 0 and 30 degrees cement binder.2 It is worth noting that the shrinkage strains
swiftly rose in the 30- to 40-degree domain corroborates are in compression. Shown in Fig. 7(c) are the effective and
the unfavorable distress of the excessive biaxial eccen- age-adjusted effective moduli of the concrete, both of which
tricity. Figure 6(c) displays the strains of GFRP along the revealed precipitous drops during the first 65 days, followed
top layer (Bar No. 1 to 3). Unlike the sectional rotations by stabilized values. Although the contribution of the aging
discussed previously, rigid translations were noticed toward coefficient was marginal (Eq. (11)), it ensures the gradual
the compression side as the vertical eccentricity was raised. application of sustained loadings.18 The effective modulus of
In other words, the mutual variation of biaxial eccentricities GFRP given in Fig. 7(d) manifests a downward propensity,
in the abscissa and ordinate was necessary for revolving the induced by the logarithmic growth of the creep coefficient
concrete section. The reinforcing bar strains at abscissa = (Fig. 7(d), inset).
187 mm (7.4 in.) are compared in Fig. 6(d). The strains Uniaxial loading—The concrete behavior in the first
above and below the geometric centroid of the column quadrant (Corner B), where consistent compression was

Fig. 6—Influence of loading angle with biaxial eccentricity: (a) concrete strain at each corner; (b) strain ratio in compression;
(c) strain of top-layer reinforcing bars; and (d) reinforcing bar strain at variable loading angle.
Fig. 7—Time-dependent material properties: (a) creep coefficient of concrete; (b) shrinkage strain in compression; (c) effective
modulus of concrete; and (d) effective modulus of GFRP.
applied, is provided in Fig. 8(a), contingent upon uniaxial were interacted in the long-term stress (Eq. (52)). Conse-
eccentricity varying from 0%h to 20%h. Contrary to the quently, concrete stresses were employed to better present
concrete strains under the short-term loading with the the ramifications of shrinkage and creep. As shown in the
linear stress-strain relationship, multiple strain components noticeable stress reduction from 0%h to 5%h, the eccentric

Fig. 8—Long-term concrete stress under uniaxial loading: (a) Corner B; (b) Corner C; (c) stress ratio at Corner B; and
(d) stress ratio at Corner C.
loading altered the uniform compression state across the crack the concrete. The radial representation of all GFRP
column section. With a further increase in the eccentricity, strains with respect to time and eccentricity is given in Fig.
the stresses steadily developed and exceeded the value of the 9(c) and (d), respectively. With the progression of time from
0%h case. These oscillating responses may cause cumulative 0 to 100 years, the almost uniform strain contour became
damage in the concrete; however, the occurrence probability distorted, especially at Bar No. 3, 5, and 8, which were
appears to be negligible because of the low stress levels, directly under the uniaxial compression (Fig. 9(c)). As far
which were less than 10.5% of the specified strength (fc′) as eccentricity is concerned, the strains of Bar No. 1, 4, and
at 100 years. The stresses in the third quadrant (Corner C, 6 were counterbalanced by the tension (Fig. 9(d)); therefore,
Fig. 8(b)) support the redistribution of the uniform compres- the likelihood of creep rupture was in part relieved (further
sion—that is, the absolute-magnitude change of 3.96 MPa discussions to follow).
(574 psi) from 0%h to 5%h at 100 years was 430% greater Biaxial loading—The ratios between the long- and short-
than the corresponding change from 5%h to 10%h. The term stresses at Corner B under selected biaxial loadings
time-dependent ratios of the short- and long-term stresses are shown in Fig. 10(a). Prior to the stabilized response, the
at Corner B are visible in Fig. 8(c). Except for the irregular previously discussed stress redistribution was reaffirmed
growth of 0%h and 5%h from 1 to 10 years that is ascribed to between 1 and 10 years. It was also obvious that the extent
the redistributed stresses, the ratios were basically preserved of the redistribution was affected by the biaxial eccentricity.
from 10 to 100 years, irrespective of eccentricity level. This The discrepancy between the stress ratios at Corners B and
fact implies that, when conducting structural design, a long- C dwindled with time (Fig. 10(b)), which means creep and
term period of 50 years is sufficient for GFRP-reinforced shrinkage alleviated the concrete stresses. The strain contour
columns subjected to uniaxial eccentricities. Regarding the of GFRP at 0%h was enlarged by the eccentric loading, partic-
stress ratio at Corner C on the other side of the loaded quad- ularly along the bottom layer (Bar No. 6 to 8, Fig. 10(c)). As
rant (Fig. 8(d)), no anomaly was observed, and the responses illustrated in Fig. 10(d), the biaxial loading altered the distri-
were consistent with both time and eccentricity. bution of the GFRP strains toward Bar No. 6, which was in
Figure 9(a) shows the strains of GFRP located in the first the third quadrant. The bars positioned along the horizontal
quadrant (Bar No. 3). Conforming to the stress redistribu- axis (No. 4 and 5) maintained the strain levels, regardless of
tion of concrete at Corner B, the transition from the concen- eccentricity type, because those were influenced only by the
tric to eccentric loadings was apparent in Bar No. 1, which sectional rotations about the vertical axis.
was subjected to local tension at 5%h (Fig. 9(a), inset). Comparison of loading configurations—The three
According to the normalized strains between the eccentric components of the time-dependent concrete strain (Eq. (3))
and concentric loadings (Fig. 9(b)), the maximum GFRP at Corner B are shown in Fig. 11(a) to (c), where their frac-
strain occurred at 0%h within a loading range that did not tions to the total strain in the primary compression region

Fig. 9—Long-term GFRP strain under uniaxial loading: (a) Bar No. 3; (b) comparison between concentric and eccentric
responses at 100 years; (c) comparison with time; and (d) comparison with eccentricity.
Fig. 10—Long-term behavior under biaxial loading: (a) ratio between long- and short-term stresses; (b) ratio between biaxial
and uniaxial stresses; (c) reinforcing bar strain with eccentricity; and (d) reinforcing bar strain with uniaxial and biaxial
eccentricities.

Fig. 11—Long-term strain components in concrete at Corner B: (a) elastic strain; (b) shrinkage strain; (c) creep strain; and
(d) average fraction to total strain.
were assessed. The elastic strain component under the of 100 years. The design provision of ACI 440.1R-155 is
biaxial loading accounted for 34.7% of the total strain at evaluated with an emphasis on the creep rupture of the rein-
100 years (Fig. 11(a)), which was 12.4% and 64.3% higher forcement. For the purpose of comparison between uniaxial
than the components of the concentric and uniaxial load- and biaxial loadings, a maximum eccentricity of 10% times
ings, respectively. This points out that when the sustained the column width is employed.
loading is removed, elastic recovery in the biaxially loaded
column would be greater than in the other columns. For the GFRP strain
shrinkage component, the uniaxial loading was prominent, Figure 12(a) exhibits the strains of GFRP with concrete
and the biaxial case was placed below the concentric case strength in compression (fc′). For clarity, two bars in the first
(Fig. 11(b)). Because the shrinkage strain is independent of and third quadrants were selected (Bar No. 3 and 6, respec-
external loading (Eq. (6)), these fractions are attributable tively). The compressive strains of Bar No. 3 adjacent to
to the development of the total strain depending upon the the loading point were mitigated by increasing the concrete
loading type. Notwithstanding the insignificant fractions strength. For instance, the strain of –1205 × 10–6 at fc′ =
of less than 1%, the creep strain component of the column 20 MPa (2900 psi) under the biaxial loading became –1049
under the biaxial loading almost doubled that of the column × 10–6 at fc′ = 50 MPa (7250 psi), which is attributed to the
under the uniaxial loading (Fig. 11(c)). The relatively large fact that the concrete section carried more stresses. The
differential between the initial and long-term stresses caused strains of Bar No. 6 on the other side of the loading point,
by the biaxial loading (Eq. (5)) is responsible for generating the third quadrant, revealed a different response owing to the
such creep responses. Figure 11(d) appraises the average distributed compression distress from the first quadrant. A
fractions of the individual components from 28 days to 100 graphical summary of these strains is provided in Fig. 12(b).
years. Overall, shrinkage was a salient factor in the strain It was evident that the biaxial loading was more influential
development, along with the loading configurations, while in the behavior of the reinforcement close to the compres-
creep was uninfluential. sive loading (Bar No. 3), relative to the uniaxial loading that
predominated the strains of Bar No. 6. The loss of the concrete
PARAMETRIC STUDY AND DESIGN ASSESSMENT section altered the GFRP strains as well (Fig. 12(c)). As the
This section examines the effects of concrete proper- magnitude of the loss increased, the concrete stress in the
ties on the long-term behavior of GFRP reinforcement in first quadrant was transferred to Bar No. 3 and, thereby, the
the benchmark column subjected to multi-axial loadings. compressive strain of the bar went up. The strain ratios given
The compressive strength of concrete spans from fc′ = 20 in Fig. 12(d) confirm the vulnerability of GFRP to the biaxial
to 50 MPa (2900 to 7250 psi), and a potential loss in the eccentricity in the loaded quadrant when the column experi-
column section varies up to 20% within the concrete cover, enced those section losses.
which can represent physical damage during a service period

Fig. 12—Parametric investigation under uniaxial and biaxial loadings at 100 years: (a) GFRP strain with concrete strength;
(b) GFRP strain ratio with concrete strength; (c) GFRP strain with concrete section loss; and (d) GFRP strain ratio with
concrete section loss.
Creep rupture of GFRP potential uncertainty. Upon experimental validation, a
The selected strains of GFRP at 100 years normalized parametric study was conducted with variable eccentrici-
by the creep-rupture limit specified in ACI 440.1R-155 are ties, loading angles, concrete strengths, and section losses.
shown in Fig. 13(a) and (b), which are related to the concrete The time-dependent performance of GFRP was appraised
strength and section loss variables, respectively. Under the against the creep-rupture provision of ACI 440.1R-15.5 The
uniaxial eccentricity of 10%h (Fig. 13(a)), the bar strains following conclusions are drawn.
on the compression side (first and fourth quadrants) were • On the short-term response of the column, the initial
notable as high as 20.7% of the limit. With an increase in the transition from concentric to uniaxial-eccentric loadings
concrete strength to fc′ = 50 MPa (7250 psi), the maximum (0%h to 5%h, where h is the dimension of the column
creep ratio decreased to 19.0%. The bars located on the section) caused abrupt changes in the concrete and
tension side (Bar No. 1, 4, and 6) revealed low vulnerability GFRP strains. Compared with the case of the uniaxial
to creep rupture. The loss of the concrete section under the loading, the maximum achievable eccentricities in the
biaxial loading resulted in erratic strain profiles (Fig. 13(b)), uncracked column decreased by 50% when biaxial
and the peak creep ratio was elevated to 25.1%, signifying loading was applied due to the augmented distress.
that physical damage in the column was crucial from a • The biaxial loading angles varying from 0 to 30 degrees
potential creep-rupture point of view. The load-dependent gradually elevated the short-term strains of the concrete,
creep strain ratios are visible in Fig. 13(c) and (d). The range whereas a rapid strain growth was noticed between 30
between the upper- and lower-bound ratios of the biaxially and 40 degrees. The mutual interaction of the biaxial
loaded columns was wider than the case of the uniaxially eccentricities in the horizontal and vertical axes was
loaded columns. For that reason, care should be exercised necessary to swivel the column section that comprised
when GFRP-reinforced columns are expected to carry the compression and tension domains.
sustained loadings at biaxial eccentricities. • The material characteristics of the column constitu-
ents were responsive to early ages, within approxi-
SUMMARY AND CONCLUSIONS mately 2 months from the loading, and then stabilized.
This paper has investigated the short- and long-term Although the long-term concrete stresses fluctuated,
behavior of glass fiber-reinforced polymer (GFRP)- dependent upon uniaxial eccentricity, the occurrence of
reinforced concrete columns under concentric, uniaxial, cumulative damage seemed negligible due to the insig-
and biaxial loadings. Based on force equilibrium and nificant stress levels that were below 10.5% of the spec-
displacement compatibility, an analytical model was formu- ified compressive strength (fc′ = 35 MPa [5080 psi]) at
lated in combination with Bayesian inference to address 100 years.

Fig. 13—Check of creep limit for GFRP bars at 100 years: (a) concrete strength under uniaxial loading; (b) concrete section
loss under biaxial loading; (c) assessment with concrete strength; and (d) assessment with concrete section loss.
• As time progressed, stress redistributions were noticed ience of Concrete Structures; and 440, Fiber-Reinforced Polymer Rein-
forcement, and Joint ACI-ASCE Committee 343, Concrete Bridge Design.
in the column concrete subjected to both uniaxial and He received the Chester Paul Siess Award for Excellence in Structural
biaxial loadings. During the first 10 years of loading, Research in 2019. His research interests include advanced composite
an irregular stress distribution was observed in the materials for rehabilitation, structural informatics, complex systems, and
science-based structural engineering, including statistical, interfacial, and
compression region where the eccentric loading was quantum physics.
sustained. The concrete stresses were relieved by creep
and shrinkage. Practically, 50 years would be a suffi- ACKNOWLEDGMENTS
ciently long period for the design of GFRP-reinforced The author would like to acknowledge support from the U.S. Depart-
ment of Transportation through the Mountain-Plains Consortium. Technical
concrete columns. contents presented in this paper are based on the opinion of the writer and
• The elastic component of the long-term concrete strains do not necessarily represent that of others.
was susceptible to the biaxial loading; by contrast, the
shrinkage component was more reactive in the uniaxi- REFERENCES
ally loaded column. The creep component was marginal 1. Mertol, H. C.; Rizkalla, S.; Zia, P.; and Mirmiran, A., “Creep and
Shrinkage Behavior of High-Strength Concrete and Minimum Rein-
in all loading configurations, less than 1% of the total forcement Ratio for Bridge Columns,” PCI Journal, V. 55, No. 3, 2010,
strains for 100 years. pp. 138-154. doi: 10.15554/pcij.06012010.138.154
• With the increased concrete strength from fc′ = 20 2. Neville, A. M., Properties of Concrete, fourth edition, John Wiley &
Sons, Inc., New York, NY, 1996.
to 50 MPa (2900 to 7250 psi), the strain level of the 3. ACI Committee 209, “Guide for Modeling and Calculating Shrinkage
reinforcement was reduced, and the influence was and Creep in Hardened Concrete (ACI 209.2R-08),” American Concrete
pronounced under the biaxial loading. The loss of the Institute, Farmington Hills, MI, 2008, 45 pp.
4. Manning, D. G., “Corrosion Performance of Epoxy-Coated Reinforcing
concrete section appreciably raised the reinforcing bar Steel: North American Experience,” Construction and Building Materials,
strains and altered their profiles. When subjected to V. 10, No. 5, 1996, pp. 349-365. doi: 10.1016/0950-0618(95)00028-3
the biaxial loading, the vulnerability of GFRP to creep 5. ACI Committee 440, “Guide for the Design and Construction of
Structural Concrete Reinforced with Fiber-Reinforced Polymer Bars (ACI
rupture was higher by 32.1% relative to that under the 440.1R-15),” American Concrete Institute, Farmington Hills, MI, 2015,
uniaxial loading. 88 pp.
6. De Luca, A.; Matta, F.; and Nanni, A., “Behavior of Full-Scale Glass
Fiber-Reinforced Polymer Reinforced Concrete Columns under Axial
AUTHOR BIOS Load,” ACI Structural Journal, V. 107, No. 5, Sept.-Oct. 2010, pp. 589-596.
Yail J. Kim, FACI, is President of the Bridge Engineering Institute, An
7. Guérin, M.; Mohamed, H. M.; Benmokrane, B.; Shield, C. K.; and
International Technical Society, and a Professor in the Department of
Nanni, A., “Effect of Glass Fiber-Reinforced Polymer Reinforcement
Civil Engineering at the University of Colorado Denver, Denver, CO. He
Ratio on Axial-Flexural Strength of Reinforced Concrete Columns,”
is Chair of ACI Subcommittee 440-I, FRP-Prestressed Concrete, and past
ACI Structural Journal, V. 115, No. 4, July 2018, pp. 1049-1061. doi:
Chair of ACI Committee 345, Bridge Construction and Preservation. He
10.14359/51701279
is a member of ACI Committees 342, Evaluation of Concrete Bridges and
8. AlAjarmeh, O. S.; Manalo, A. C.; Benmokrane, B.; Karunasena, W.;
Bridge Elements; 377, Performance-Based Structural Integrity & Resil-
Ferdous, W.; and Mendis, P., “A New Design-Oriented Model of Glass

Fiber-Reinforced Polymer-Reinforced Hollow Concrete Columns,” ACI 20. Kaklauskas, G., and Gribniak, V., “Eliminating Shrinkage Effect from
Structural Journal, V. 117, No. 2, Mar. 2020, pp. 141-156. Moment Curvature and Tension Stiffening Relationships of Reinforced
9. Choo, C. C.; Harik, I. E.; and Gesund, H., “Strength of Rectangular Concrete Members,” Journal of Structural Engineering, ASCE, V. 137,
Concrete Columns Reinforced with Fiber-Reinforced Polymer Bars,” ACI No. 12, 2011, pp. 1460-1469. doi: 10.1061/(ASCE)ST.1943-541X.0000395
Structural Journal, V. 103, No. 3, May-June 2006, pp. 452-459. 21. ACI Committee 318, “Building Code Requirements for Structural
10. Tu, J.; Gao, K.; He, L.; and Li, X., “Experimental Study on the Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
Axial Compression Performance of GFRP-Reinforced Concrete Square Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
Columns,” Advances in Structural Engineering, V. 22, No. 7, 2018, 22. CEB, “Structural Effects of Time-Dependent Behaviour of Concrete
pp. 1554-1565. doi: 10.1177/1369433218817988 (CEB Bulletin 215),” Comité Euro-International du Béton, Lausanne, Swit-
11. Zhou, J.; He, X.; and Shen, W., “Compression Behavior of zerland, 1993.
Seawater and Sea-Sand Concrete Reinforced with Fiber and Glass Fiber- 23. Tan, K. H., and Saha, M. K., “Long-Term Deflections of Rein-
Reinforced Polymer Bars,” ACI Structural Journal, V. 117, No. 4, July forced Concrete Beams Externally Bonded with FRP System,” Journal of
2020, pp. 103-114. Composites for Construction, ASCE, V. 10, No. 6, 2006, pp. 474-482. doi:
12. Afifi, M. Z.; Mohamed, H. M.; and Benmokrane, B., “Axial Capacity 10.1061/(ASCE)1090-0268(2006)10:6(474)
of Circular Concrete Columns Reinforced with GFRP Bars and Spirals,” 24. Tavares, C. M. L.; Ribeiro, M. C. S.; Ferreira, A. J. M.; and
Journal of Composites for Construction, ASCE, V. 18, No. 1, 2014, Guedes, R. M., “Creep Behaviour of FRP-Reinforced Polymer Concrete,”
pp. 363-366. doi: 10.1061/(ASCE)CC.1943-5614.0000438 Composite Structures, V. 57, No. 1-4, 2002, pp. 47-51. doi: 10.1016/
13. Hasan, H. A.; Karim, H.; Sheikh, M. N.; and Hadi, M. N. S., S0263-8223(02)00061-2
“Moment-Curvature Behavior of Glass Fiber-Reinforced Polymer 25. Sun, L.; Wei, M.; and Zhang, N., “Experimental Study on the
Bar-Reinforced Normal-Strength Concrete and High-Strength Concrete Behavior of GFRP Reinforced Concrete Columns under Eccentric Axial
Columns,” ACI Structural Journal, V. 116, No. 4, July 2019, pp. 65-75. doi: Load,” Construction and Building Materials, V. 152, 2017, pp. 214-225.
10.14359/51715573 doi: 10.1016/j.conbuildmat.2017.06.159
14. Tobbi, H.; Farghaly, A. S.; and Benmokrane, B., “Behavior of 26. Zadeh, H. J., and Nanni, A., “Design of RC Columns Using Glass FRP
Concentrically Loaded Fiber-Reinforced Polymer Reinforced Concrete Reinforcement,” Journal of Composites for Construction, ASCE, V. 17,
Columns with Varying Reinforcement Types And Ratios,” ACI Structural No. 3, 2013, pp. 294-304. doi: 10.1061/(ASCE)CC.1943-5614.0000354
Journal, V. 111, No. 2, Mar.-Apr. 2014, pp. 375-385. 27. ASCE/SEI 7-10, “Minimum Design Loads for Buildings and Other
15. Kruschke, J., Doing Bayesian Data Analysis, second edition, Else- Structures,” American Society of Civil Engineers, Reston, VA, 2010.
vier, Oxford, UK, 2015. 28. Nowak, A. S., and Collins, K. R., Reliability of Structures, second
16. Dilger, W., and Neville, A. M., “Method of Creep Analysis of Struc- edition, CRC Press, Boca Raton, FL, 2013.
tural Members,” Designing for Effects of Creep, Shrinkage and Tempera- 29. You, Y.-J.; Park, Y.-H.; Kim, H.-Y.; and Park, J.-S., “Hybrid Effect
ture in Concrete Structures, SP-27, American Concrete Institute, Farm- on Tensile Properties of FRP Rods with Various Material Compositions,”
ington Hills, MI, 1971, pp. 349-371. Composite Structures, V. 80, No. 1, 2007, pp. 117-122. doi: 10.1016/j.
17. Bazant, Z. P., “Prediction of Concrete Creep Effects Using Age- compstruct.2006.04.065
Adjusted Effective Modulus Method,” ACI Journal Proceedings, V. 69, No. 30. Okeil, A. M.; El-Tawil, S.; and Shahawy, M., “Flexural Reliability
4, Apr. 1972, pp. 212-217. of Reinforced Concrete Bridge Girders Strengthened with Carbon Fiber-
18. Dilger, W. H., “Creep Analysis of Prestressed Concrete Structures Reinforced Polymer Laminates,” Journal of Bridge Engineering, ASCE, V. 7,
Using Creep-Transformed Section Properties,” PCI Journal, V. 27, No. 1, No. 5, 2002, pp. 290-299. doi: 10.1061/(ASCE)1084-0702(2002)7:5(290)
1982, pp. 98-119. doi: 10.15554/pcij.01011982.98.119 31. Kulkarni, S. N., “Calibration of Flexural Design of Concrete
19. TRB, “Control of Cracking in Concrete: State of the Art,” Trans- Members Reinforced with FRP Bars,” MS thesis, Louisiana State Univer-
portation Research Circular No. E-C107, Transportation Research Board, sity, Baton Rouge, LA, 2006.
Washington, DC, 2006, 56 pp.

Title No. 119-S28
In-Plane Seismic Response of Autoclaved Aerated

Concrete Block Masonry-Infilled Reinforced Concrete
Frame Building
by Nikhil P. Zade, Avadhoot Bhosale, Pradip Sarkar, and Robin Davis
The use of lightweight autoclaved aerated concrete (AAC) block in factories, AAC blocks exhibit variability in strength prop-
masonry is gaining popularity in earthquake-resistant infilled rein- erties due to several manufacturing parameters, such as
forced concrete (RC) frame buildings due to its various benefits. temperature, humidity, curing conditions, constituent mate-
Therefore, appropriate knowledge of the strength properties of rials, proportions of the mixture, geometry, and supervision
AAC block masonry is necessary for a reasonable evaluation of the
(Bhosale et al. 2019). However, modeling the nonlinearity
seismic behavior of such buildings. In the present study, the uncer-
and uncertainty in the structural properties of AAC block
tainties related to the two most critical parameters that control the
resistance capacity of infilled masonry are investigated through masonry has not received any research attention, while these
laboratory experiments, and the best-fitted probability density two parameters constitute the primary input to assess the
functions are recommended. Furthermore, the in-plane seismic seismic safety of building structures. Similar studies with
performances of typical RC frame buildings infilled with AAC other building materials, such as reinforced concrete (RC),
block masonry are evaluated in a probabilistic framework consid- burnt clay, and fly ash brick masonry, are well reported by
ering the recommended probability density functions showing the many researchers (Sahoo et al. 2020; Sahu et al. 2019, 2020;
ineffectiveness of an assumed normal distribution for this purpose. Sherafati and Sohrabi 2016; Kilinc et al. 2012). Although
Although lightweight AAC block masonry slightly increases the the application of probabilistic methods in engineering
seismic risk of the building compared to traditional brick masonry design still is in the formative stages, efforts have been made
due to its lower strength properties, it can be safely used as an
(to varying degrees) to introduce such techniques into the
infill material in areas with high seismicity, as it achieves the code-
practice of earthquake-resistant design. Many international
prescribed reliability index.
codes and standards (ASCE/SEI 41-17 2017; ASCE/SEI
Keywords: autoclaved aerated concrete (AAC) block masonry; compres- 7-16 2016; IS 1893-1:2016 2016, and so on) now accept a
sive strength; fragility curve; infilled frame; seismic reliability; shear-bond probabilistic structural design that considers uncertainties in
strength. demand and capacity.
The present study aims to assess the in-plane seismic
INTRODUCTION performance of AAC block-infilled RC frame buildings in
Autoclave aerated concrete (AAC) block masonry is a probabilistic framework. To accomplish this, a series of
commonly used for the construction of load-bearing and laboratory experiments are conducted to develop the uncer-
framed structures due to its several advantages over tradi- tainty and nonlinearity models of AAC block masonry. The
tional brick masonry. Very low density, high fire resistance, suitable probability distribution model for the compressive
better finishing, and convenience of handling are some of and shear-bond strength of AAC block masonry is evaluated
the important benefits of AAC block masonry in building from the statistical analyses of the experimental results. The
construction (Raj 2020; Bhosale et al. 2019; Ravichandran available force-deformation backbone curve (Panagiotakos
2009). Increased use of AAC block can help to preserve the and Fardis 1996) is used to model the nonlinearity of AAC
natural agricultural clayey soil used for building materials. block masonry with experimentally obtained parameters.
Also, the energy consumption for burning clay bricks in An accepted probability-based seismic assessment method
developing countries can be decreased significantly by the (Vamvatsikos and Cornell 2002; Ellingwood 2001) is used
increased use of AAC blocks. The porous structure and low to study the effect of AAC block masonry on the seismic
thermal conductivity of AAC can offer better sound and safety of infilled RC frame buildings. The suitability of the
thermal insulation. A detailed description of the composi- assumed normal distribution of strength properties of AAC
tion and manufacturing process of AAC blocks can be found block masonry is also investigated. Further, the relative
elsewhere (RILEM Technical Committees 78-MCA and in-plane seismic performance of RC frame buildings infilled
51-ALC 1993).
It is the general perception of practicing engineers that
due to its reduced mass density, AAC block masonry can ACI Structural Journal, V. 119, No. 2, March 2022.
reduce the inertia force and thereby reduce the seismic risk MS No. S-2020-343.R3, doi: 10.14359/51734329, received May 22, 2021, and
of an infilled frame; however, this perception has not been Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including author’s
validated by any research findings. Although manufactured closure, if any, will be published ten months from this journal’s date if the discussion
is received within four months of the paper’s print publication.

with AAC block masonry is compared with that of clay and Table 1—Mean strength properties of constituent
fly-ash brick masonry. materials at 28 days
Constituent Compressive strength, Flexural strength,
REVIEW OF LITERATURE materials MPa MPa
The present study deals with the modeling of variability in
AAC block 3.63 (0.08) —
the strength properties of AAC block masonry and in-plane
seismic performance evaluation of AAC block masonry- Block adhesive 6.63 (0.19) 2.42 (0.29)
infilled RC frame buildings. A detailed literature review has Note: Strength values presented here are average test results of 20 specimens,
been carried out on the following three specific topics: 1) and figures in parenthesis represent associated coefficient of variation; 1000 psi =
6.895 MPa.
variability of strength properties of brick masonry; 2) AAC
block masonry; and 3) seismic performance evaluation of studies (Wijaya et al. 2020; Bhosale et al. 2018; Jeon et al.
infilled RC frames. 2015) have used this methodology for seismic risk analysis
The characteristic values of material strength, its disper- of infilled RC frame buildings using nonlinear dynamic
sion, and the probability distribution affect the design of any structural analyses. Most of the aforementioned studies
structure (Nowak and Collins 2000). Strength and load param- have considered equivalent diagonal struts to model infill
eters exhibit statistical regularity, which calls for appropriate masonry. Different backbone curves for the equivalent strut
probabilistic load and resistance models (Banerjee et al. model are also available in the published literature (Dolšek
2016). Experimental evaluation of the probability distribu- and Fajfar 2008; Žarnić and Gostič 1997; Panagiotakos and
tion of strength parameters for traditional masonry materials Fardis 1996; Saneinejad and Hobbs 1995) with their specific
(for example, fly ash and clay brick masonry) is reported in strengths and limitations. However, the nonlinear backbone
the literature. Several studies on the establishment of correct curve recommended by Panagiotakos and Fardis (1996) is
probability density functions for various strength properties widely used by the previous studies (Di Trapani et al. 2015;
of clay brick masonry are abundantly available in the liter- Jeon et al. 2015; Ricci et al. 2013; Celarec et al. 2012) to
ature (Sherafati and Sohrabi 2016; Montazerolghaem 2015; model infill for seismic risk assessment.
Sýkora and Holický 2010; Schueremans and Van Gemert
2006). Sahu et al. (2019, 2020) proposed best-fit statistical RESEARCH SIGNIFICANCE
distributions for compressive and shear-bond strength of Several studies have been conducted to evaluate the phys-
fly ash brick masonry. Despite the widespread use of AAC icomechanical properties of AAC block masonry and the
block masonry, no research study is available on the proba- behavior of AAC block masonry-infilled moment frames
bilistic model for its strength properties. against quasi-static lateral loading. However, this paper
A detailed literature review revealed that the majority presents a detailed analysis of such building frames against
of the past studies, including the codes and standards (Raj seismic loading for the first time, which can help select AAC
2020; Raj et al. 2020; Bhosale et al. 2019; ASTM C1693 as a suitable masonry material for the construction of infill
2011; IS 2185-3:2005 2005; Schober 2005; Narayanan and frames in seismically active areas. In addition, this document
Ramamurthy 2000; RILEM Technical Committees 78-MCA reports the appropriate probability distribution models for
and 51-ALC 1993) on AAC block masonry, are limited to compressive and shear-bond strength of unreinforced AAC
the evaluation of its several physical and mechanical proper- block masonry that will aid the reliability-based structural
ties. One important observation that can be drawn from the design of AAC block masonry-infilled buildings.
literature is that a significant variation exists for most of the
physical and mechanical properties of AAC block. Evalu- EXPERIMENTAL PROGRAM AND TEST RESULTS
ation of steel (Ravichandran 2009; Liu and Li 2004, 2005; The variability in the compressive and shear-bond
Memari et al. 1999) and RC (Milanesi et al. 2018; Celik strength of AAC block masonry is investigated by labora-
2016; Penna et al. 2008) moment frames infilled with AAC tory experiments. Commercially procured AAC blocks of
block masonry have been studied experimentally against 600 x 200 x 100 mm (± 5 mm) dimensions are cut into brick
static and quasi-static lateral loading by several researchers, sizes of 200 x 100 x 100 mm using a diamond blade cutter
and significant improvement in the strength and stiffness is to prepare the masonry prisms. Three brick units are bonded
observed compared to the bare frames. Milanesi et al. (2018) together using commercially available AAC block adhesive
suggested three different coupled arrangements of AAC conforming to ANSI A118.4 (2019) to form a prism. This
block masonry to improve the seismic performance of RC adhesive is a factory-prepared blend of portland cement,
infilled frames. However, comprehensive structural analyses graded aggregates, and polymers, designed to be used with
of AAC block masonry-infilled RC frames against seismic water to produce high-strength thixotropic mortar. This study
loading have not been adequately reported in the published uses a water-adhesive ratio of 0.3 (by weight) in accordance
literature. with ANSI 118.1 (2019) and the manufacturer’s recommen-
The probabilistic approach that considers uncertain- dation. The strength properties of the AAC block and the
ties in both load and resistance can result in a comprehen- block adhesive used in the present study are experimentally
sive structural evaluation. Cornell et al. (2002) developed evaluated as per the guidelines of relevant standards (ANSI
a probabilistic framework for the seismic assessment of A118.4 2019; ASTM C1693 2011) and presented in Table 1.
structures, which formed the basis for the design recom- Twenty AAC block masonry samples are prepared and
mendations of the SAC-FEMA project. Many published tested for each of the selected parameters (compressive

Fig. 1—Typical AAC brick units and masonry specimens.
Fig. 2—Experimental setup. (Note: 1 m = 3.28 ft.)

and shear-bond strength) following the published literature test without precompression is also carried out in the load-
(Chen et al. 2014; van Proosdij et al. 2016). The published controlled loading frame, as shown in Fig. 2(b). The failure
literature and standards have not specified any guidelines load divided by the shear area (two times the cross-sectional
for the experimental assessment for AAC masonry assem- area of the joint) gives the shear-bond strength of AAC block
blages. Therefore, the compressive and shear-bond strength masonry. Figure 3 shows the compressive and shear-bond
tests are carried out in the present study as per ASTM strength data obtained experimentally in the form of histo-
C1314 (2018) and BS EN 1052-3:2002 (2002), respectively, grams. The descriptive statistics of the strength data are
which are intended for clay brick masonry. Photographs of presented in Table 2.
typical AAC brick specimens and masonry assemblages are
presented in Fig. 1. DEVELOPMENT OF VARIABILITY MODELS
The compressive strength test is conducted in a load- This section focuses on representing the variability in
controlled compression testing frame under uniaxial the compressive and shear-bond strength of AAC block
compression, as shown in Fig. 2(a). The failure load divided masonry. Experimentally obtained compressive and shear-
by the cross-sectional area gives the compressive strength bond strength data of AAC block masonry specimens are
of the masonry (fm) prism. The shear-bond strength (τcr) transformed into probability density functions using the

shape and scale parameters. Normal, lognormal, gamma, the GOF tests lies in the method by which the test statis-
Weibull, Gumbel maximum, and Gumbel minimum are tics and critical values are calculated. The KS test is widely
the pre-decided two-point standard probability distribution used to analyze continuous distributions and is reported to
models selected in the present work, as these distributions be independent of sample size. However, KS tends to be
are reported (Sorrentino et al. 2016; Montazerolghaem 2015; more sensitive near the middle of the distribution than at the
Lawrence 1985) to describe the strength properties of brittle tails (Sahu et al. 2020). KSL is appropriate when population
materials better. Four nonparametric goodness-of-fit (GOF) parameters are not known. Both KSL and KS tests are iden-
tests, such as Kolmogorov-Smirnov (KS), Kolmogorov- tical in terms of the statistic value but different in terms of
Smirnov-Lilliefors (KSL), Anderson-Darling (AD), and the critical values (Lilliefors 1967). AD is a modification of
Chi-Square criterion (CS), are used in the present study. The the KS test and is more sensitive to the tail region (Stephens
philosophy of all the GOF tests is to measure the “statistic” 1974). The CS test is generally suitable when large random
value, which is defined as the difference between the actual variables are used.
and the chosen cumulative distribution function (CDF), and For each probability density function, the test statistic
compare this statistic value with a “critical” value. The crit- was determined and ranked in line with the methodology
ical value depends on the sample size and the significance followed in previous studies (Sahu et al. 2020; Chandrappa
level considered. It should be noted that a lower statistic and Biligiri 2017; Chen et al. 2014; Kilinc et al. 2012).
value indicates better performance by the chosen distribu- Commercial software EasyFit Professional (Version 5.6) is
tion in modeling the variability. The difference between used for all the statistical analyses in the present study. The
shape and scale parameters corresponding to the selected
Table 2—Descriptive statistics of compressive and statistical distribution functions obtained for the compres-
shear-bond strength of AAC block masonry sive and shear-bond strength of AAC block masonry speci-
Compressive Shear-bond mens are presented in Table 3.
Statistics strength strength The statistic values of the individual GOF tests and the
Sample size 20 20 total statistic for strength data are reported in Tables 4
and 5. The graphical representation of the chosen CDFs,
Range, MPa 2.54 0.28
along with the test data sets, is presented in Fig. 4. It can
Mean, MPa 2.24 0.23 be observed from Table 4 that for the compressive strength
Standard deviation, MPa 0.60 0.062 of AAC block masonry, KS/KSL has minimum statistics
Coefficient of variation 0.26 0.28 for lognormal distribution, AD has minimum statistics for
gamma distribution, and CS has minimum statistics for the
Standard error, MPa 0.13 0.014
Gumbel maximum distribution. It can be observed that not
Skewness 0.38 2.66 all GOF tests agree to choose one distribution to describe
Excess kurtosis 0.64 9.27 the variability of the strength properties. Therefore, the
Minimum, MPa 1.12 0.16 selection of the most appropriate distribution is made based
on the total statistic value, which is the square root of the
25th percentile, MPa 1.90 0.18
sum of the squares of the individual statistic of the selected
Median, MPa 2.23 0.21 GOF tests. The lowest total statistic corresponds to the
75th percentile, MPa 2.64 0.24 most appropriate probability density function. Accordingly,
Maximum, MPa 3.66 0.45 lognormal and Gumbel maximum are found to be the best-
fitted distribution for compressive and shear-bond strength
Note: 1000 psi = 6.895 MPa
Fig. 3—AAC block masonry strength data histogram. (Note: 1000 psi = 6.895 MPa.)

of AAC block masonry, respectively. It can be noted here SEISMIC PERFORMANCE EVALUATION
that there is a slight difference in the total statistic value of One of the primary objectives of this study is to inves-
the first-ranked distribution and other higher-ranked distri- tigate the performance of AAC block masonry relative to
butions. In addition, these statistic values may alter if the the traditional brick masonry for the seismic resistance of
data pool is changed. Therefore, any of the higher-ranked infilled RC frame buildings. A probability-based seismic
distributions with low total statistic values may be consid- evaluation method (Vamvatsikos and Cornell 2002; Elling-
ered to describe the variability in the strength properties of wood 2001) using nonlinear dynamic analysis is employed
AAC block masonry. for this purpose. In the present study, spectral acceleration
at the first mode of vibration (Sa,T1) and maximum inter-
story drift (ISD) are considered the ground-motion inten-
sity measure (IM) and demand parameter, respectively. The
Table 3—Distribution parameters for various seismic performance is evaluated through the probabilistic
probability distribution functions seismic demand model (PSDM), fragility, and reliability
curves.
Property Distribution Parameters
The fragility function represents the probability of exceed-
Normal µ = 2.237, σ = 0.591 ance of a selected demand parameter beyond the chosen
Lognormal µ = 0.771, σ = 0.268 structural limit state at a particular IM, which can be defined
Gamma α = 14.313, β = 0.156
by the following closed-form equation (Celik and Elling-
Compressive strength,
wood 2010)
Weibull α = 4.249, β = 2.376
MPa
Gumbel  
maximum
µ = 1.971, σ = 0.461
ln( Dˆ )  ln(Cˆ )

P (C  D IM )  
Gumbel  2  2  2  (1)
µ = 2.503, σ = 0.461  D IM c m

minimum
Normal µ = 0.222, σ = 0.062 where ϕ is the standardized Gaussian cumulative distribu-
Lognormal μ = -1.531, σ = 0.225
tion function; D and C are the demand and capacity at a
selected limit state, respectively; D^ and C^ are the median
Gamma α = 12.615, β = 0.018
values of D and C, respectively; and βD|IM, βc, and βm are
Shear-bond strength, Weibull α = 6.644, β = 0.225 the dispersions in demand at a given IM, capacity, and
MPa
Gumbel modeling, respectively. Nonlinear time-history analyses
μ = 0.194, σ = 0.049
maximum (NLTHA) are conducted to obtain the maximum ISD, which
Gumbel is the demand parameter D considered in the present study.
μ = 0.250, σ = 0.048
minimum This method uses two analytical approximations. The first is
the power-law relationship between the median drift demand
Note: µ is continuous location parameters; α is continuous shape parameters; and β
and σ are continuous scale parameters. 1000 psi = 6.895 MPa. and the intensity measure (Cornell et al. 2002)
Table 4—GOF test results for compressive strength of AAC block masonry
KS/KSL AD CS
Distribution Statistic Rank Statistic Rank Statistic Rank Total statistics Final rank
Normal 0.146 5 0.234 2 1.649 5 1.673 5
Lognormal 0.109 1 0.281 3 0.199 2 0.361 1
Gamma 0.111 2 0.228 1 0.274 3 0.373 2
Weibull 0.111 3 0.364 4 1.540 4 1.587 4
Gumbel maximum 0.125 4 0.414 5 0.174 1 0.467 3
Gumbel minimum 0.199 6 0.878 6 1.656 6 1.885 6
Table 5—GOF test results for shear-bond strength of AAC block masonry
KS/KSL AD CS
Distribution Statistic Rank Statistic Rank Statistic Rank Total statistics Final rank
Normal 0.205 5 1.426 4 0.019 2 1.441 4
Lognormal 0.138 2 0.622 1 0.527 5 0.827 2
Gamma 0.172 4 1.060 3 0.024 3 1.074 3
Weibull 0.134 1 2.192 5 0.459 4 2.243 5
Gumbel maximum 0.147 3 0.746 2 0.015 1 0.760 1
Gumbel minimum 0.266 6 3.619 6 2.930 6 4.664 6

Fig. 4—Cumulative distributions of AAC block masonry strength. (Note: 1000 psi = 6.895 MPa.)
and bay width are kept uniform as 3.2 and 5 m for both the
Dˆ  a ( IM )b (2) buildings. The buildings are assumed to be located in Bhuj,
Gujarat, a state in west-central India. Consequently, the
where a and b are the constant coefficients. The second building frames are designed as per relevant Indian standards
approximation is that the drift demand D is distributed (IS 1893-1:2016; IS 13920:2016) considering the seismic
lognormally (Shome 1999) with a standard deviation βD|IM load of this region, which corresponds to a peak ground
about the median. The three parameters a, b, and βD|IM are acceleration of 0.36g according to the Indian Standard (IS
obtained by regression analysis on the NLTHA results. The 1893-1:2016). The characteristic compressive strength of
power-law relation in Eq. (2) represents the PSDM for the concrete (fck) and characteristic yield strength of reinforce-
corresponding frame. The value of C^ for the RC moment- ment steel (fy) are considered 25 MPa (3626 psi) and 415
resisting infilled frame is considered 0.5% at the life-safety MPa (60 ksi), respectively. The details of the cross section
(LS) performance level (Wijaya et al. 2020; Morandi et al. and reinforcement of the frame elements of the two selected
2014; Hak et al. 2012). RC frames are presented in Tables 6 and 7. It should be noted
The seismic performances of the selected buildings are here that the design details of the frame remain uniform for
further examined by incorporating the seismic hazard of the all types of infill materials considered.
site for which the building has been designed. The seismic Selected buildings are modeled and analyzed using the
hazard function, H(IM), is the annual rate of exceeding a open-source software OpenSEES (Version 3.0.3) developed
given IM at a site due to all possible future earthquakes, by McKenna et al. (2018). A force-based nonlinear beam-
expressed through a complementary cumulative distribution column fiber element is used to model the beams and
function. The probability of failure due to a seismic event is columns, which considers the spread of plasticity along the
obtained by convolution of the seismic hazard curve with the element. The force-based approach is reported (Anvarsa-
corresponding fragility curve, as follows (Ellingwood 2001) marin et al. 2018; Terzic 2011) to be the economical and
accurate approach to investigate the seismic behavior of RC
dH IM i structures. Each beam and column element is discretized
PLS ( IM i ) P(C D IM i ) d IM (3)
d IM into five Gauss integration points. Figure 6 presents a sche-
matic diagram of the two-dimensional element and section
The reliability index, a measure of the safety margin corre-
discretization used in the present study. Detailed formula-
sponding to the probability of failure, can be found using the
tion of the fiber-based element can be available in Lee and
following equation
Mosalam (2004). The cross section of the RC members
consists of three types of materials: confined and unconfined
βLS(IMi) = –ϕ–1(PLS(IMi)) (4)
concrete, and reinforcing steel (Fig. 6). Cover concrete and
core concrete materials are considered separately, and the
STRUCTURAL MODELING
corresponding stress-strain relations are calculated based
Two typical infilled RC frame buildings (Fig. 5), one
on Mander et al. (1988). Concrete02 is a uniaxial material
six-story three-bay (6S3B) and one eight-story four-bay
model that includes tensile strength, and the linear tension-
(8S4B), are selected for the present study. Story height

Table 6—Design details of frame 6S3B
Member Floor number Width x depth, mm Longitudinal reinforcement Transverse reinforcement
1 450 x 450 8-28 φ (UD) 10 φ @ 200 mm c/c
Column 2 to 4 450 x 450 8-25 φ (UD) 10 φ @ 200 mm c/c
5 to 6 450 x 450 8-20 φ (UD) 10 φ @ 200 mm c/c
1 to 4 300 x 400 5-25 φ (T) + 3-25 φ (B) 10 φ @ 175 mm c/c
Beam 5 300 x 400 4-25 φ (T) + 3-25 φ (B) 10 φ @ 175 mm c/c
6 300 x 400 3-25 φ (T) + 2-25 φ (B) 10 φ @ 175 mm c/c
Note: UD is uniformly distributed; T is top reinforcement; and B is bottom reinforcement. 1 m = 3.28 ft.
Table 7—Design details of frame 8S4B

Member Floor number Width x depth, mm Longitudinal reinforcement Transverse reinforcement
1 450 x 450 16-28 φ (UD) 10 φ @ 200 mm c/c
Column 2 to 3 450 x 450 16-25 φ (UD) 10 φ @ 200 mm c/c
4 to 8 450 x 450 8-25 φ (UD) 10 φ @ 200 mm c/c
1 to 6 300 x 400 5-25 φ (T) + 3-25 φ (B) 10 φ @ 150 mm c/c
Beam 7 300 x 400 4-25 φ (T) + 3-25 φ (B) 10 φ @ 150 mm c/c
8 300 x 400 3-25 φ (T) + 2-25 φ (B) 10 φ @ 150 mm c/c
Note: UD is uniformly distributed; T is top reinforcement; and B is bottom reinforcement. 1 m = 3.28 ft.
E t sin 2
0.25
h h w (6)
4 Ec I c hw
where Ew, hw, and t are the modulus of elasticity, height, and
thickness of the infill panel, respectively; Ec and Ic are the
modulus of elasticity and the second moment of area of the
adjacent columns, respectively; and θ represents the infill
panel aspect ratio indicated in Fig. 7(a). The force-defor-
mation backbone curve for modeling the nonlinearity of the
equivalent strut adopted from published literature (Panag-
iotakos and Fardis 1996) is shown in Fig. 7(b). In the
absence of a specific model for AAC block masonry, the
aforementioned modeling approach is adopted in the present
study. A similar approach for modeling the equivalent strut
is also adopted by previous literature (Ravichandran 2009)
to analyze AAC block masonry-infilled frames.
The column end at the foundation is modeled by consid-
Fig. 5—Geometry of selected frames: (a) 6S3B; and (b)
ering fixity at the top of the foundation. All permanent mass
8S4B. (Note: 1 m = 3.28 ft.)
that moves with the structure is lumped at the appropriate
softening effect is used to model the concrete. Steel nodes as per Mazzoni et al. (2006). This includes the mass
reinforcing bars are modeled with isotropic strain- of beams, columns, slabs, infill walls, and the mass corre-
hardening properties as per Menegotto and Pinto (1973). sponding to the appropriate part (25%) of the live loads
Further details about the selected materials (that is, expected to be present in the structure during the ground
Concrete02 and Steel02) are available in the OpenSees shaking. The present study uses Rayleigh damping, which
material library (Mazzoni et al. 2006). Due to the complexity formulates the damping matrix as a linear combination of
of the micromodeling approach, the present study considers the mass matrix and stiffness matrix. As per the published
the macromodeling approach using double-bracing diag- literature and standards (Kaushik et al. 2006; EN 1998-
onal strut elements for modeling infill walls in line with the 1:2005 2005; FEMA-306 1999), out-of-plane collapses are
published literature (Wijaya et al. 2020; Bhosale et al. 2018; not common for infill walls of low slenderness value (ratio
Celarec et al. 2012). The equivalent width of the strut (w) is of the smaller of length or height to a thickness smaller than
formulated (IS 1893-1:2016 2016; Stafford Smith and Carter 15) and when walls are sufficiently confined in an RC frame,
1969) in terms of a nondimensional relative stiffness param- as is the case with most of the RC residential infilled frame
eter (λh) buildings. Therefore, the present study considers an in-plane
analysis of a representative two-dimensional middle frame
w = d(0.175λh–0.4) (5)

Fig. 6—Fiber-based modeling used in present study.
Fig. 7—Modeling of infill masonry: (a) schematic presentation of equivalent strut; and (b) backbone curve.
in the loading direction ignoring the torsional effects and Acc-475, is used to carry the nonlinear pseudodynamic
out-of-plane response. time-history analysis, and the roof displacement history
As discussed earlier, RC moment frames infilled with is recorded. Figure 8 shows the comparative plot between
AAC block masonry have been studied experimentally the roof displacement records of the experimental work
against static and quasi-static lateral loading by a few conducted on the ICONS infilled frame and the computa-
researchers (Milanesi et al. 2018; Celik 2016; Penna et al. tional results obtained from the present study. From this
2008). However, in the absence of the necessary information figure, it can be concluded that the modeling approach used
regarding the building models, nonlinear material character- in the present analysis will yield accurate behavior of the
istics, loading, and response history of those experiments, the infilled RC frame subjected to earthquake ground motion
modeling approach considered in this study is validated with with reasonable accuracy.
the results of a pseudodynamic experimental test conducted The uncertainty in the seismic loading is modeled consid-
at the European Laboratory for Structural Assessment for a ering a suite of earthquake ground motions. In the absence of
three-bay, four-story, clay brick-infilled full-scale RC frame sufficient ground-motion records in the Indian region where
(ICONS infilled frame). Details of the test specimens, mate- the buildings are assumed to be located, those recorded else-
rial properties, loading schemes, and response histories are where (California) with similar site conditions are consid-
available in Carvalho et al. (1999). A ground-motion record, ered for the present study, as recommended by Chopra

Fig. 8—Validation of modeling approach using experimental results. (Note: 1 m = 3.28 ft.)
(2007). Accordingly, 22 pairs of far-field ground-motion ASSESSMENT OF SEISMIC PERFORMANCE
records were obtained from the NGA-West2 Pacific Earth- This section investigates the in-plane seismic performance
quake Engineering Research Center (PEER) ground-mo- of the AAC block masonry-infilled RC building relative to
tion data set (FEMA P695 2009) and are used in the present the performance of the same building infilled with traditional
study. General characteristics of the selected ground motions masonry infill. As discussed earlier, an ND is often assumed
are as follows: magnitude > 6.5 on the Richter scale, epicen- in a probability-based seismic analysis in the absence of an
tral distance > 8 km, PGA > 0.2g, and peak ground velocity appropriate variability model. This section also examines
> 15 cm/s. Table 8 presents the details of the selected how such assumptions may affect the seismic evaluation
ground-motion records. of AAC-infilled RC frame buildings. Forty-four selected
The uncertainty in the structural capacity is modeled ground motions are linearly scaled from a PGA of 0.1g to
considering the variability of the strength properties of 1.0g. Each of the 44 building models generated through the
infill masonry only. The material properties of RC frames sampling of selected input variables (Table 9) is analyzed
are kept at their characteristic values. The description of the for one randomly selected ground motion with a randomly
variability in the strength properties of selected masonry selected PGA using NLTHA. The following sections present
materials is presented in Table 9. The proposed best-fitted the results of the probabilistic seismic analysis in terms of
distributions (BFs) are considered for modeling the vari- PSDMs, fragility, and reliability curves.
ability of AAC block masonry. It should be noted that in the
absence of an appropriate variability model for any material, PSDMs
a normal distribution (ND) is often assumed in probability- The demand parameter (ISD) obtained from the NLTHA
based seismic analysis. Therefore, an assumed ND for the and the corresponding intensity measure (Sa,T1) are plotted
strength parameters of AAC block masonry is additionally in Fig. 9 and 10 in a log-log scale. The power-law rela-
considered as a reference. The distribution parameters for tion between ISD and Sa,T1 (PSDM as shown in Eq. (2)) is
traditional brick masonry are taken from the published liter- developed through regression analysis and plotted in
ature, as shown in Table 9. The mass density and the elastic Fig. 9 and 10. Figure 9 presents the PSDM curves for the
modulus of all the materials considered in this study are AAC-infilled frame modeled with samples generated
presented in Table 10. It is apparent that the AAC block has considering the BF and assumed ND functions for the
a mass density of approximately one-third of that of tradi- selected strength properties of AAC block masonry. Figure 9
tional brick masonry, which leads to a significant reduction shows a higher ISD for an assumed ND when compared with
(approximately 30%) of the seismic weight of the buildings the proposed BF distribution. Therefore, the assumption of
infilled with AAC block masonry. the ND for describing the strength properties of AAC block
On the other hand, the average compressive strength (and masonry yields conservative results in this study. It can be
hence the average elastic modulus) of AAC block masonry concluded based on this result that the selection of proba-
is found to be approximately 50% lower than that of selected bility distribution may alter the outcome of seismic perfor-
traditional brick masonry, as shown in Table 9. This contrib- mance evaluation.
utes to the substantial reduction in building stiffness. The Further assessment of the seismic performances is
combined effect of low seismic mass and low stiffness did conducted considering only the BF of AAC block masonry.
not significantly influence the building period. In addition, Accordingly, PSDMs of the building for different selected
the elastic modulus, which is a dependent variable, assumes infill materials are presented in Fig. 10. This figure shows
the variability associated with the compressive strength of that, among all the selected infill materials, AAC block
the selected masonry materials. masonry leads to the highest drift demand for both the chosen
building frames. The regression coefficients (a and b) and
the dispersion in ISD (βD|IM) for each frame are presented
in Table 11. It is observed from this table that AAC infill

Table 8—Details of selected ground-motion records
PGA, g
Source (fault
Sl. Event Magnitude North East type) Recording station
1 Northridge 1994 6.7 0.42 0.52 Thrust Beverly Hills - Mulhol
2 Northridge 1994 6.7 0.41 0.48 Thrust Canyon Country-WLC
3 Düzce, Turkey 1999 7.1 0.73 0.82 Strike-slip Bolu
4 Hector Mine 1999 7.1 0.27 0.34 Strike-slip Hector
5 Imperial Valley 1979 6.5 0.24 0.35 Strike-slip Delta
6 Imperial Valley 1979 6.5 0.36 0.38 Strike-slip El Centro Array No. 11
7 Kobe, Japan 1995 6.9 0.51 0.5 Strike-slip Nishi-Akashi
8 Kobe, Japan 1995 6.9 0.24 0.21 Strike-slip Shin-Osaka
9 Kocaeli, Turkey 1999 7.5 0.31 0.36 Strike-slip Düzce
10 Kocaeli, Turkey 1999 7.5 0.22 0.15 Strike-slip Arçelik
11 Landers 1992 7.3 0.24 0.15 Strike-slip Yermo Fire Station
12 Landers 1992 7.3 0.28 0.42 Strike-slip Coolwater
13 Loma Prieta 1989 6.9 0.53 0.44 Strike-slip Capitola
14 Loma Prieta 1989 6.9 0.56 0.37 Strike-slip Gilroy Array No. 3
15 Manjil, Iran 1990 7.4 0.51 0.5 Strike-slip Abbar
16 Superstition Hills 1987 6.5 0.36 0.26 Strike-slip El Centro Imperial County
17 Superstition Hills 1987 6.5 0.45 0.3 Strike-slip Poe Road
18 Cape Mendocino 1992 7.0 0.39 0.55 Thrust Rio Dell Overpass
19 Chi-Chi, Taiwan 1999 7.6 0.35 0.44 Thrust CHY101
20 Chi-Chi, Taiwan 1999 7.6 0.47 0.51 Thrust TCU045
21 San Fernando 1971 6.6 0.21 0.17 Thrust LA - Hollywood Stor
22 Friuli, Italy 1976 6.5 0.35 0.31 Thrust Tolmezzo
Table 9—Probability distribution of selected infill masonry

Masonry Parameters Shape parameter, MPa Scale parameter, MPa Distribution Reference
0.771 0.27 Lognormal Present study
Compressive strength, fm
2.237 0.60 Normal Assumed
AAC block
0.194 0.05 Gumbel max Present study
Shear-bond strength, τcr
0.22 0.06 Normal Assumed
Compressive strength, fm 5.32 0.80 Normal Singhal and Rai
Clay brick
Shear-bond strength, τcr 0.43 0.10 Normal (2014)
Compressive strength, fm 3.90 0.58 Normal Basha and Kaushik

Fly-ash brick
Shear-bond strength, τcr 0.46 0.092 Normal (2015)
Note: 1000 psi = 6.895 MPa.
(for all selected distributions and frames) results in a higher traditional masonry due to its lower mass density. However,
value of the constant a, which corresponds to a higher value it is not really the true representation of their seismic
of ISD, leading to the higher vulnerability of the buildings. characteristics.
The values of βD|IM and R2 indicate that the dispersion in ISD
is more or less in a similar range for all considered frames, Fragility curves
and the power-law assumption of PSDM is quite acceptable. The fragility curves are developed for all the frames for
The building frames are additionally analyzed considering the selected performance limit state and are presented in Fig.
only the mass of the masonry, ignoring their strength and 12. The results presented in Fig. 12 support all the findings
stiffness contribution as usually done by practicing engi- of PSDMs shown in Fig. 10. The AAC-infilled RC frame is
neers. The resulting PSDMs of the two selected building found to be the most fragile among all the selected infilled
frames are presented in Fig. 11. Figure 11 shows the lowest frames. The order of the frames with different infill in
drift demand for the AAC in the open frame compared to terms of increasing probability of failure is fly ash < clay <

Table 10—Other physical and mechanical properties of selected infill masonry
Masonry Parameters Mean Reference
Density, kg/m 3
602.75 Bhosale et al. (2019)
AAC block
Modulus of elasticity, MPa 750fm Bhosale et al. (2019)
Density, kg/m 3
1773.50 Singhal and Rai (2014)
Clay brick
Modulus of elasticity, MPa 620fm Ellingwood and Tallin (1985)
Density, kg/m3 1660.00 Basha and Kaushik (2016)
Fly ash brick
Modulus of elasticity, MPa 600fm Basha and Kaushik (2015)
Density, kg/m 3
2500.00 IS 875-1:1987 (1987)
Reinforced concrete
Modulus of elasticity, MPa 5000 f ck IS 456:2000 (2000)
Note: 1000 psi = 6.895 MPa; 1 lb/ft3 = 16.02 kg/m3.
Fig. 9—PSDMs for selected distributions of AAC block masonry strength.
Fig. 10—PSDMs for selected infill masonry materials.

AAC_BF. Although there are many parameters governing the AAC-infilled open RC frame compared to the traditional
the results, the decrease in ISD due to a decrease in the mass masonry. However, this is not the appropriate representation
of the AAC block is nullified by the increase in ISD due to of the seismic performance of integral connected infilled RC
its low modulus of elasticity and strength compared to that frames due to incomplete modeling.
of the traditional brick masonry-infilled frame.
The fragility curves of the two selected building frames Reliability curves
are also developed considering only the mass of the masonry, The actual assessment of the seismic performance of
ignoring their strength and stiffness contribution for a typical the building must take into account the seismic hazard of
performance limit state (2% of ISD), and presented in the selected site. Accordingly, the reliability curves of
Fig. 13. This figure depicts the better seismic performance of the selected building models are developed considering

the Sa hazard curves of Bhuj, Gujarat (India) available in a quantitative estimate of the change in reliability for the
published literature (Nath and Thingbaijam 2012). The reli- selected infill materials, the reliability indexes are calculated
ability curves developed for the selected infilled frame are for the selected frames at the life-safety damage level with
presented in Fig. 14. Reliability curves endorse the conclu- four different mean annual probabilities of ground-motions
sions drawn from PSDM and fragility analyses. The order of occurrence and presented in Table 12. This table concludes
the building frames infilled with various masonry materials that AAC block masonry results in lower reliability in infilled
in increasing reliability is AAC_BF < clay < fly ash. From RC frame buildings than traditional brick masonry. The
these figures, it is also observed that the relative difference in reliability indexes of the AAC-infilled RC frame are found
the reliability decreases as the story height increases. To have to be approximately 5 to 10% and 2 to 5% lower than the
traditional infill masonry for 6S3B and 8S4B frames, respec-
Table 11—PSDMs of 6S3B and 8S4B frames tively. However, it was found that the reliability indexes of
the RC building infilled with AAC block masonry, calcu-
Frame Infill type ISD, % βD|IM R2
lated taking into account the hazard curve representative of
AAC_BF 0.917 (Sa)1.109 0.274 0.867 India’s highly active seismic zone, are within the allowable
AAC_ND 0.961 (Sa)1.098 0.248 0.877 limits of the design code. Therefore, AAC block masonry
6S3B
Clay 0.906 (Sa) 1.231
0.322 0.852 can be safely used as an infill material in high seismic areas.
Fly ash 0.908 (Sa) 1.277
0.241 0.919
SUMMARY AND CONCLUSIONS
AAC_BF 1.020 (Sa) 1.085
0.256 0.885 Autoclaved aerated concrete (AAC) block masonry is
AAC_ND 1.090 (Sa)1.058 0.326 0.806 gaining popularity in the construction of infilled reinforced
8S4B
Clay 0.951 (Sa) 1.121
0.275 0.875 concrete (RC) frame buildings because of its several advan-
tages. A number of studies on the characterization of different
Fly ash 0.945 (Sa) 1.090
0.301 0.843
Fig. 11—PSDMs of building being considered as open frames.
Fig. 12—Fragility curves for selected infill masonry materials.

Fig. 13—Fragility curves of building being considered as open frames.
Fig. 14—Reliability curves for selected infill masonry materials at LS (ISD = 0.5%).
Table 12—Reliability indexes at selected limit state
Mean annual probability of occurrence of ground motions
Frame Infill type 50% in 50 years 20% in 50 years 10% in 50 years 2% in 50 years
AAC_BF 4.750 3.912 3.493 3.118
6S3B Clay 4.994 4.053 3.569 3.128
Fly ash 5.395 4.275 3.685 3.161
AAC_BF 4.454 3.768 3.422 3.105
8S4B Clay 4.796 3.975 3.531 3.127
Fly ash 4.804 4.011 3.576 3.165
engineering properties of AAC block masonry are reported in best-fitted probability density function for the compressive
the published literature. However, no studies on modeling the and shear-bond strength of AAC block masonry; and 2) the
uncertainty in key strength properties of AAC block masonry effect of AAC block masonry on the in-plane seismic resis-
are available in the published literature. This information is tance of infilled RC frame buildings. The most appropriate
essential for the formulation of limit state design criteria and statistical distributions for the compressive and shear-bond
probability-based seismic analysis of AAC block masonry strength of AAC block masonry prisms are proposed based
structures. The lightweight nature of AAC block masonry on experimental data and goodness-of-fit (GOF) tests (KS/
is believed to improve the seismic resistance of infilled RC KSL, AD, and CS). Only two-parameter distributions suit-
frame buildings, although there are no reported studies to able for brittle material properties are considered in this
justify this general assumption. Therefore, the present study study.
focuses on two aspects: 1) experimental evaluation of the

Additionally, the performance of selected RC frame REFERENCES
buildings infilled with AAC block masonry relative to two ANSI A118.1, 2019, “American National Standard Specifications for
Dry-Set Cement Mortar,” American National Standards Institute, New
traditionally used masonry materials (clay and fly ash brick York, NY.
masonry) is evaluated through probabilistic seismic demand ANSI A118.4, 2019, “American National Standard Specifications for
models (PSDMs), fragility curves, and reliability curves Modified Dry-Set Cement Mortar,” American National Standards Institute,
New York, NY.
incorporating the proposed probability distributions. This Anvarsamarin, A.; Rofooei, F. R.; and Nekooei, M., 2018, “Soil-
study briefly compares the analysis results considering the Structure Interaction Effect on Fragility Curve of 3D Models of Concrete
proposed best-fitted distribution of the input strength prop- Moment-Resisting Buildings,” Shock and Vibration, V. 2018, Article
ID 7270137, 13 pp. doi: 10.1155/2018/7270137
erties of the AAC block masonry with the assumed normal ASCE/SEI 7-16, 2016, “Minimum Design Loads for Buildings and Other
distribution. The salient conclusions of the present study are Structures,” American Society of Civil Engineers, Reston, VA, 800 pp.
listed as follows: ASCE/SEI 41-17, 2017, “Seismic Evaluation and Retrofit of Existing
Buildings,” American Society of Civil Engineers, Reston, VA, 550 pp.
1. The variability in the compressive and shear-bond ASTM C1314-18, 2018, “Standard Test Method for Compressive
strength of AAC block masonry can be described using the Strength of Masonry Prism,” ASTM International, West Conshohocken,
lognormal and Gumbel maximum probability density func- PA, 9 pp.
ASTM C1693, 2011, “Standard Specification for Autoclaved Aerated
tions, respectively. This information can be used as an essen- Concrete (AAC),” ASTM International, West Conshohocken, PA, 7 pp.
tial input to the reliability-based analysis and design of AAC Banerjee, A. K.; Pramanik, D.; and Roy, R., 2016, “Seismic Structural
block masonry structures. Fragilities: Proposals for Improved Methodology per Spectral Matching
of Accelerogram,” Engineering Structures, V. 111, Mar., pp. 538-551. doi:
2. It was found that the assumption of the normal distribu- 10.1016/j.engstruct.2016.01.002
tion to describe the variability of the strength properties of Basha, S. H., and Kaushik, H. B., 2015, “Evaluation of Nonlinear Mate-
AAC block masonry gives conservative results. However, rial Properties of Fly Ash Brick Masonry under Compression and Shear,”
Journal of Materials in Civil Engineering, ASCE, V. 27, No. 8, Aug.,
choosing an appropriate distribution function is essential for p. 04014227. doi: 10.1061/(ASCE)MT.1943-5533.0001188
a more realistic assessment of seismic performances. Basha, S. H., and Kaushik, H. B., 2016, “Suitability of Fly Ash Brick
3. The mass density of AAC block masonry is found to Masonry as Infill in Reinforced Concrete Frames,” Materials and Struc-
tures, V. 49, No. 9, Sept., pp. 3831-3845. doi: 10.1617/s11527-015-0757-5
be approximately 35% of that of traditional clay and fly Bhosale, A.; Zade, N. P.; Davis, R.; and Sarkar, P., 2019, “Experimental
ash brick masonry. On the other hand, the compressive and Investigation of Autoclaved Aerated Concrete Masonry,” Journal of Mate-
shear-bond strength of AAC block masonry is in the range rials in Civil Engineering, ASCE, V. 31, No. 7, July, p. 04019109. doi:
10.1061/(ASCE)MT.1943-5533.0002762
of 40 to 60% of that of traditional brick masonry, which Bhosale, A. S.; Davis, R.; and Sarkar, P., 2018, “Seismic Safety of Verti-
ultimately reduces the modulus of elasticity of AAC block cally Irregular Buildings: Performance of Existing Indicators,” Journal of
masonry to 50 to 70%. Architectural Engineering, ASCE, V. 24, No. 3, Sept., p. 04018013. doi:
10.1061/(ASCE)AE.1943-5568.0000319
4. It has been found that the combined effect of lower BS EN 1052-3:2002, 2002, “Methods of Test for Masonry-Part 3: Deter-
seismic mass, lower strength, and lower rigidity of AAC mination of Initial Shear Strength,” British Standards Institution, London,
block masonry slightly reduces the seismic performance of UK, 18 pp.
Carvalho, E. C.; Coelho, E.; and Campos-Costa, A., 1999, “Preparation
infilled RC frame buildings compared to clay and fly ash of the Full-Scale Tests on Reinforced Concrete Frames-Characteristics of
brick masonry. The reduction of the reliability indexes of the the Test Specimens, Materials and Testing Conditions,” ICONS Report,
AAC block masonry-infilled RC frame is found to be within Innovative Seismic Design Concepts for New and Existing Structures,
European TMR Network, LNEC.
10% compared to traditional clay and fly ash brick masonry. Celarec, D.; Ricci, P.; and Dolšek, M., 2012, “The Sensitivity of Seismic
5. The reliability indexes of the RC frame buildings Response Parameters to the Uncertain Modelling Variables of Masonry-In-
infilled with AAC block masonry estimated considering a filled Reinforced Concrete Frames,” Engineering Structures, V. 35, Feb.,
pp. 165-177. doi: 10.1016/j.engstruct.2011.11.007
hazard curve representative of India’s highly active seismic Celik, O. C., 2016, “Effect of AAC Infill Walls on Structural System
zone is found to be within the acceptable limits of the design Dynamics of a Concrete Building,” Journal of Earthquake Engineering,
code. Therefore, AAC block masonry can be safely used as V. 20, No. 5, pp. 738-748. doi: 10.1080/13632469.2015.1104757
Celik, O. C., and Ellingwood, B. R., 2010, “Seismic Fragilities for
an infill material in high-seismic areas. Non-Ductile Reinforced Concrete Frames–Role of Aleatoric and Epis-
temic Uncertainties,” Structural Safety, V. 32, No. 1, Jan., pp. 1-12. doi:
AUTHOR BIOS 10.1016/j.strusafe.2009.04.003
Nikhil P. Zade is a PhD Student in the Department of Civil Engineering Chandrappa, A. K., and Biligiri, K. P., 2017, “Flexural-Fatigue Charac-
at National Institute of Technology, Rourkela, Rourkela, Odisha, India. teristics of Pervious Concrete: Statistical Distributions and Model Devel-
He received his BE in civil engineering from Sant Gadge Baba Amravati opment,” Construction and Building Materials, V. 153, Oct., pp. 1-15. doi:
(SGBA) University, Amravati, Maharashtra, India, and his MTech from 10.1016/j.conbuildmat.2017.07.081
National Institute of Technology, Rourkela. Chen, X.; Wu, S.; and Zhou, J., 2014, “Variability of Compressive
Strength of Concrete Cores,” Journal of Performance of Constructed
Avadhoot Bhosale is an Assistant Professor in the School of Civil Engi- Facilities, ASCE, V. 28, No. 4, Aug., p. 06014001. doi: 10.1061/(ASCE)
neering at MIT World Peace University Pune, Maharashtra, India. He CF.1943-5509.0000513
received his PhD from National Institute of Technology, Rourkela. Chopra, A. K., 2007, Dynamics of Structures: Theory and Applications
to Earthquake Engineering, Higher Education Press, Beijing, China.
Pradip Sarkar is a Professor in the Department of Civil Engineering Cornell, C. A.; Jalayer, F.; Hamburger, R. O.; and Foutch, D. A.,
at National Institute of Technology, Rourkela. He received his PhD from 2002, “Probabilistic Basis for 2000 SAC Federal Emergency Manage-
Indian Institute of Technology Madras, Chennai, Tamil Nadu, India. ment Agency Steel Moment Frame Guidelines,” Journal of Structural
Engineering, ASCE, V. 128, No. 4, Apr., pp. 526-533. doi: 10.1061/
Robin Davis is an Assistant Professor in the Department of Civil Engi- (ASCE)0733-9445(2002)128:4(526)
neering at National Institute of Technology Calicut, Kozhikode, Kerala, Di Trapani, F.; Macaluso, G.; Cavaleri, L.; and Papia, M., 2015,
India. He received his PhD from Indian Institute of Technology Madras. “Masonry Infills and RC Frames Interaction: Literature Overview and
State of the Art of Macromodeling Approach,” European Journal of
Environmental and Civil Engineering, V. 19, No. 9, pp. 1059-1095. doi:
10.1080/19648189.2014.996671

Dolšek, M., and Fajfar, P., 2008, “The Effect of Masonry Infills on the Simulation (OpenSees), Pacific Earthquake Engineering Research (PEER)
Seismic Response of a Four-Storey Reinforced Concrete Frame—A Deter- Center, Berkeley, CA.
ministic Assessment,” Engineering Structures, V. 30, No. 7, July, pp. 1991- Memari, A. M.; Aghakouchak, A. A.; Ghafory Ashtiany, M.; and Tiv, M.,
2001. doi: 10.1016/j.engstruct.2008.01.001 1999, “Full-Scale Dynamic Testing of a Steel Frame Building during
EasyFit version 5.6 [computer software], MathWave Technologies, Construction,” Engineering Structures, V. 21, No. 12, pp. 1115-1127. doi:
Dnipro, Ukraine. 10.1016/S0141-0296(98)00068-6
Ellingwood, B., and Tallin, A., 1985, “Limit States Criteria for Masonry Menegotto, M., and Pinto, P. E., 1973, “Method of Analysis for Cycli-
Construction,” Journal of Structural Engineering, ASCE, V. 111, No. 1, cally Loaded Reinforced Concrete Plane Frames Including Changes in
Jan., pp. 108-122. doi: 10.1061/(ASCE)0733-9445(1985)111:1(108) Geometry and Non-Elastic Behaviour of Elements under Combined Normal
Ellingwood, B. R., 2001, “Earthquake Risk Assessment of Building Force and Bending,” IABSE Reports of the Working Commissions, Interna-
Structures,” Reliability Engineering & System Safety, V. 74, No. 3, Dec., tional Association of Bridge and Structural Engineering, V. 13, pp. 15-22.
pp. 251-262. doi: 10.1016/S0951-8320(01)00105-3 doi: 10.5169/seals-13741
EN 1998-1:2005, 2005, “Eurocode 8: Design of Structures for Earth- Milanesi, R. R.; Morandi, P.; Penna, A.; and Magenes, G., 2018, “Seismic
quake Resistance-Part 1: General Rules, Seismic Actions and Rules for Performance of AAC Masonry Infill: From Traditional Systems to Innova-
Buildings,” European Committee for Standardization, Brussels, Belgium, tive Solutions,” ce/papers, V. 2, No. 4, 311-317.
229 pp. Montazerolghaem, M., 2015, “Analysis of Unreinforced Masonry Struc-
FEMA-306, 1999, “Evaluation of Earthquake Damaged Concrete and tures with Uncertain Data,” PhD thesis, Department of Civil and Architec-
Masonry Wall Buildings: Basic Procedures Manual,” Applied Technology ture, Technische Universität Berlin, Berlin, Germany.
Council (ATC-43) Project, The Partnership for Response and Recovery, Morandi, P.; Hak, S.; and Magenes, G., 2014, “In-Plane Experimental
Federal Emergency Management Agency, Washington, DC, 270 pp. Response of Strong Masonry Infills,” 9th International Masonry Confer-
FEMA P695, 2009, “Quantification of Building Seismic Performance ence, Guimarães, Portugal, International Masonry Society, 12 pp.
Factors,” Federal Emergency Management Agency, Washington, DC, 421 pp. Narayanan, N., and Ramamurthy, K., 2000, “Structure and Properties
Hak, S.; Morandi, P.; Magenes, G.; and Sullivan, T. J., 2012, “Damage of Aerated Concrete: A Review,” Cement and Concrete Composites, V. 22,
Control for Clay Masonry Infills in the Design of RC Frame Structures,” No. 5, Oct., pp. 321-329. doi: 10.1016/S0958-9465(00)00016-0
Journal of Earthquake Engineering, V. 16 No. sup1, pp. 1-35. Nath, S. K., and Thingbaijam, K. K. S., 2012, “Probabilistic Seismic
IS 456:2000, 2000, “Plain and Reinforced Concrete — Code of Prac- Hazard Assessment of India,” Seismological Research Letters, V. 83, No. 1,
tice,” Bureau of Indian Standards, New Delhi, India, 114 pp. pp. 135-149. doi: 10.1785/gssrl.83.1.135
IS 875-1:1987, 1987, “Code of Practice for Design Loads (Other than Nowak, A. S., and Collins, K. R., 2000, Reliability of Structures,
Earthquake) for Buildings and Structures-Part 1: Dead Loads — Unit McGraw-Hill, New York, 360 pp.
Weights of Building Materials and Stored Materials,” Bureau of Indian Panagiotakos, T. B., and Fardis, M. N., 1996, “Seismic Response of
Standards, New Delhi, India, 44 pp. Infilled RC Frames Structures,” 11th World Conference on Earthquake
IS 13920:2016, 2016, “Ductile Design and Detailing of Reinforced Engineering, Acapulco, Mexico.
Concrete Structures Subjected to Seismic Forces - Code of Practice (First Penna, A.; Magenes, G.; Calvi, G. M.; and Costa, A. A., 2008, “Seismic
Revision),” Bureau of Indian Standards, New Delhi, India. Performance of AAC Infill and Bearing Walls with Different Reinforce-
IS 1893-1:2016, 2016, “Criteria for Earthquake Resistant Design of ment Solutions,” 14th International Brick and Block Masonry Conference,
Structures-Part 1: General Provisions and Buildings,” Bureau of Indian Sydney, Australia, pp. 13-20.
Standards, New Delhi, India, 44 pp. Raj, A., 2020, “Strength Enhancement of Autoclaved Aerated Concrete
IS 2185-3:2005, 2005, “Concrete Masonry Units-Part 3: Autoclaved (AAC) Block and Its Masonry,” PhD thesis, Department of Mechanical
Cellular Aerated Concrete Blocks,” Bureau of Indian Standards, New Engineering, Indian Institute of Technology Guwahati, Guwahati, India.
Delhi, India. Raj, A.; Borsaikia, A. C.; and Dixit, U. S., 2020, “Evaluation of Mechan-
Jeon, J.-S.; Park, J.-H.; and DesRoches, R., 2015, “Seismic Fragility of ical Properties of Autoclaved Aerated Concrete (AAC) Block and its
Lightly Reinforced Concrete Frames with Masonry Infills,” Earthquake Masonry,” Journal of The Institution of Engineers (India): Series A, pp.
Engineering & Structural Dynamics, V. 44, No. 11, pp. 1783-1803. doi: 315-325.
10.1002/eqe.2555 Ravichandran, S. S., 2009, “Design Provisions for Autoclaved Aerated
Kaushik, H. B.; Rai, D. C.; and Jain, S. K., 2006, “Code Approaches to Concrete (AAC) Infilled Steel Moment Frames,” PhD thesis, The Univer-
Seismic Design of Masonry-Infilled Reinforced Concrete Frames: A State- sity of Texas at Austin, Austin, TX, 342 pp.
of-the-Art Review,” Earthquake Spectra, V. 22, No. 4, pp. 961-983. doi: Ricci, P.; De Risi, M. T.; Verderame, G. M.; and Manfredi, G., 2013,
10.1193/1.2360907 “Influence of Infill Distribution and Design Typology on Seismic Perfor-
Kilinc, K.; Celik, A. O.; Tuncan, M.; Tuncan, A.; Arslan, G.; and mance of Low- and Mid-Rise RC Buildings,” Bulletin of Earthquake Engi-
Arioz, O., 2012, “Statistical Distributions of In-Situ Microcore Concrete neering, V. 11, No. 5, pp. 1585-1616. doi: 10.1007/s10518-013-9453-4
Strength,” Construction and Building Materials, V. 26, No. 1, pp. 393-403. RILEM Technical Committees 78-MCA and 51-ALC, 1993, Autoclaved
doi: 10.1016/j.conbuildmat.2011.06.038 Aerated Concrete - Properties, Testing and Design, S. Aroni, G. J. de Groot,
Lawrence, S. J., 1985, “Random Variations in Brickwork Properties,” M. J. Robinson, G. Svanholm, and F. H. Wittman, eds., Taylor & Francis
7th International Brick Masonry Conference, Melbourne, Australia, pp. Group, London, UK, 428 pp.
537-547. Sahoo, K.; Dhir, P. K.; Teja, P. R. R.; Sarkar, P.; and Davis, R., 2020,
Lee, T.-H., and Mosalam, K. M., 2004, “Probabilistic Fiber Element “Variability of Silica Fume Concrete and Its Effect on Seismic Safety of
Modeling of Reinforced Concrete Structures,” Computers & Structures, Reinforced Concrete Buildings,” Journal of Materials in Civil Engi-
V. 82, No. 27, Oct., pp. 2285-2299. doi: 10.1016/j.compstruc.2004.05.013 neering, ASCE, V. 32, No. 4, Apr., p. 04020024. doi: 10.1061/(ASCE)
Lilliefors, H. W., 1967, “On the Kolmogorov-Smirnov Test for MT.1943-5533.0003072
Normality with Mean and Variance Unknown,” Journal of the Amer- Sahu, S.; Sarkar, P.; and Davis, R., 2019, “Quantification of Uncertainty
ican Statistical Association, V. 62, No. 318, June, pp. 399-402. doi: in Compressive Strength of Fly Ash Brick Masonry,” Journal of Building
10.1080/01621459.1967.10482916 Engineering, V. 26, Nov., p. 100843. doi: 10.1016/j.jobe.2019.100843
Liu, Y. S., and Li, G. Q., 2004, “Behavior of Steel Frames With and Sahu, S.; Sarkar, P.; and Davis, R., 2020, “Uncertainty in Bond Strength
Without AAC Infilled Walls Subjected to Static and Cyclic Horizontal of Unreinforced Fly-Ash Brick Masonry,” Journal of Materials in Civil
Loads,” 13th World Conference on Earthquake Engineering, Vancouver, Engineering, ASCE, V. 32, No. 3, Mar., p. 06020003. doi: 10.1061/(ASCE)
BC, Canada, Paper No. 1112, 11 pp. MT.1943-5533.0003095
Liu, Y.-S., and Li, G.-Q., 2005, “Experimental and Theoretical Research Saneinejad, A., and Hobbs, B., 1995, “Inelastic Design of Infilled
on Lateral Load Resistance of Steel Frames with Infilled Walls,” Journal of Frames,” Journal of Structural Engineering, ASCE, V. 121, No. 4, Apr.,
Building Structures, V. 26, No. 3, pp. 78-84. pp. 634-650. doi: 10.1061/(ASCE)0733-9445(1995)121:4(634)
Mander, J. B.; Priestley, M. J. N.; and Park, R., 1988, “Theoret- Schober, G., 2005, “The Most Important Aspects of Microstructure Influ-
ical Stress-Strain Model for Confined Concrete,” Journal of Structural encing Strength of AAC,” Autoclaved Aerated Concrete, M. C. Limbachiya
Engineering, ASCE, V. 114, No. 8, Sept., pp. 1804-1826. doi: 10.1061/ and J. J. Roberts, eds., Taylor & Francis Group, London, UK, pp. 145-153.
(ASCE)0733-9445(1988)114:8(1804) Schueremans, L., and Van Gemert, D., 2006, “Probability Density Func-
Mazzoni, S.; McKenna, F.; Scott, M. H.; and Fenves, G. L., 2006, tions for Masonry Material Parameters – A Way to Go?” Structural Analysis
“OpenSees Command Language Manual,” Open System for Earthquake of Historical Constructions, P. B. Lourenço, P. Roca, C. Modena, and S.
Engineering Simulation (OpenSees), Pacific Earthquake Engineering Agrawal, eds., New Delhi, India, pp. 921-928.
Research (PEER) Center, 465 pp. Sherafati, M. A., and Sohrabi, M. R., 2016, “Probabilistic Model for
McKenna, F.; McGann, C.; Arduino, P.; and Harmon, J. A., 2018, Bed Joint Shear-Sliding Strength of Clay-Brick Walls Based on Field
“OpenSEES Laboratory,” Open System for Earthquake Engineering Test Data,” ASCE-ASME Journal of Risk and Uncertainty in Engineering

Systems, Part A: Civil Engineering, V. 2, No. 4, Dec., p. 04016007. doi: Sýkora, M., and Holický, M., 2010, “Probabilistic Model for Masonry
10.1061/AJRUA6.0000879 Strength of Existing Structures,” Engineering Mechanics, V. 17, No. 1,
Shome, N., 1999, “Probabilistic Seismic Demand Analysis of Nonlinear pp. 61-70.
Structures,” PhD dissertation, Department of Civil and Environmental Terzic, V., 2011, “Force-Based Element vs. Displacement-based Element,”
Engineering, Stanford University, Stanford, CA, 320 pp. OpenSees, NEES, and NEEScomm, University of California, Berkeley,
Singhal, V., and Rai, D. C., 2014, “Suitability of Half-Scale Burnt Clay Berkeley, CA, 25 pp.
Bricks for Shake Table Tests on Masonry Walls,” Journal of Materials in Vamvatsikos, D., and Cornell, C. A., 2002, “Incremental Dynamic Analysis,”
Civil Engineering, ASCE, V. 26, No. 4, Apr., pp. 644-657. doi: 10.1061/ Earthquake Engineering & Structural Dynamics, V. 31, No. 3, pp. 491-514. doi:
(ASCE)MT.1943-5533.0000861 10.1002/eqe.141
Sorrentino, L.; Infantino, P.; and Liberatore, D., 2016, “Statistical Tests van Proosdij, A. S. J.; Sosef, M. S. M.; Wieringa, J. J.; and Raes, N.,
for the Goodness of Fit of Mortar Compressive Strength Distributions,” 2016, “Minimum Required Number of Specimen Records to Develop Accu-
16th International Brick and Block Masonry Conference (IBMAC), Padova, rate Species Distribution Models,” Ecography, V. 39, No. 6, pp. 542-552.
Italy, C. Modena, F. da Porto, and M. R. Valluzzi, eds., Taylor & Francis doi: 10.1111/ecog.01509
Group, London, UK, pp. 1921-1928. Wijaya, H.; Rajeev, P.; Gad, E.; and Amirsardari, A., 2020, “Effect of Infill-
Stafford Smith, B., and Carter, C., 1969, “A Method of Analysis for Wall Material Types and Modeling Techniques on the Seismic Response of
Infilled Frames,” Proceedings of the Institution of Civil Engineers, V. 44, Reinforced Concrete Buildings,” Natural Hazards Review, ASCE, V. 21,
No. 1, Sept., pp. 31-48. doi: 10.1680/iicep.1969.7290 No. 3, Aug., p. 04020031. doi: 10.1061/(ASCE)NH.1527-6996.0000395
Stephens, M. A., 1974, “EDF Statistics for Goodness of Fit and Some Žarnić, R., and Gostič, S., 1997, “Masonry Infilled Frames as an Effec-
Comparisons,” Journal of the American Statistical Association, V. 69, tive Structural Sub-Assemblage,” Seismic Design Methodologies for the
No. 347, Sept., pp. 730-737. doi: 10.1080/01621459.1974.10480196 Next Generation of Codes, P. Fajfar and H. Krawinkler, eds., Routledge,
London, UK, pp. 335-346.

Title No. 119-S29
Simplified Strut-and-Tie Model for Shear Strength

Prediction of Reinforced Concrete Low-Rise Walls
by Jimmy Chandra and Susanto Teng
In this study, an analytical model using the strut-and-tie concept shear stress distribution over an entire wall cross section
was developed to predict reinforced concrete (RC) low-rise wall was uniform which is only valid for RC low-rise walls with
shear strengths. In the model, the failure mode considered was boundary elements having Hw/Lw less than 1.0.10,11 More-
crushing of the diagonal compression strut. To accurately determine over, the calculation of RC low-rise wall shear strengths
the strut area, a formula for calculating depth of compression zone
using their models needs an iterative procedure to obtain a
at the bottom of wall was derived with the aid of nonlinear finite
solution that satisfies equilibrium and compatibility condi-
element analysis. A total of 100 RC low-rise wall specimens failing
in shear obtained from available literature were used to verify the tions as well as constitutive law of materials. Thus, it may
accuracy of wall strength predictions of the proposed strut-and- not be practical to be used by engineers to estimate the shear
tie model. Furthermore, strength predictions from building codes strength of RC low-rise walls.
and other analytical models were also included for comparison In this study, an analytical model for predicting RC
purposes. The analysis results show that the proposed strut-and-tie low-rise wall shear strengths was developed based on the
model is conservative and it has the lowest coefficient of varia- strut-and-tie concept. RC low-rise walls having Hw/Lw less
tion as compared to other methods in predicting the shear strength than 2.5 can be categorized as disturbed regions where a
of RC low-rise walls. In addition, the predictions of the proposed plane section does not remain plane. In this case, the strut-
model are quite consistent and less scattered for wide ranges of and-tie model is considered as a rational approach to predict
wall height-length ratios and concrete compressive strengths.
the strength of disturbed regions.14 Later on, experimental
Keywords: building code predictions; reinforced concrete (RC) wall shear wall strengths obtained from available literatures were used
strengths; strut-and-tie. to verify the accuracy of the proposed strut-and-tie model.
In addition, strength predictions from building codes4,5 and
INTRODUCTION other strut-and-tie models15,16 were included as well for
The use of reinforced concrete (RC) walls has become comparison purposes.
increasingly popular nowadays due to their superior perfor-
mance against lateral loads such as wind and earthquake RESEARCH SIGNIFICANCE
loads.1 In addition, not only for lateral loads, RC walls This study focused on the development of an analytical
can also be used to resist gravity loads as well. Thus, it is model based on the strut-and-tie concept to predict RC
important to be able to determine the strength of RC walls low-rise wall shear strengths. It is expected that the model
accurately to provide safe and economical design, as these could serve as a rational yet simple approach for predicting
are two major concerns for structural engineers. Previous the shear strength of RC low-rise walls. Furthermore, the
studies by the authors2,3 show that the flexural strength of study conducted here provides a new formula for calcu-
RC walls can be reasonably well predicted using flexural lating the depth of the compression zone at the bottom of
theory for members subjected to axial load and bending RC low-rise walls in which the assumption of plane section
moment. However, for the shear strength, empirical building remains plane (linear strain distribution) is not valid. The
code formulas4,5 underestimate RC wall shear strengths by formula was developed with the aid of nonlinear finite
a significant margin, especially for high-strength concrete element analysis (FEA) using ATENA software.17 This is
(HSC) walls, and the overall predictions are quite scattered. important to accurately predict the shear strength of RC
Therefore, there was a need to develop an analytical model low-rise walls.
based on rational theory to accurately predict the shear
strength of RC walls. BUILDING CODES AND OTHER ANALYTICAL
The rational theory for predicting RC members’ shear MODELS
strength was developed in early 1900s based on the truss ACI 318-194 and Eurocode 85 are two reference building
analogy.6,7 The theory was further developed to predict the codes that are adopted in many countries. As such, those two
shear strength of RC members more accurately.8,9 For RC
low-rise walls having a height-length ratio (Hw/Lw) less
than 2.5, many researches have been conducted to predict MS No. S-2020-477.R1, doi: 10.14359/51734330, received July 3, 2021, and
the shear strength.10-13 All those theories are able to predict reviewed under Institute publication policies. Copyright © 2022, American Concrete
the shear strength of RC low-rise walls with certain accu- obtained from the copyright proprietors. Pertinent discussion including author’s
racy. However, in their truss models, it was assumed that closure, if any, will be published ten months from this journal’s date if the discussion

Fig. 1—Strut-and-tie mechanisms proposed by Hwang and Lee.15
building codes and other strut-and-tie models proposed by Diagonal tension failure of web due to shear—If αs = MEd/
other researchers15,16 are reviewed briefly as follows. (VEdLw) ≥ 2.0, where MEd is the design bending moment at
the base of the wall and VEd is the design shear force, the
ACI 318-19 shear strength is given by VRd,s
According to ACI 318-19,4 the nominal shear strength Vn
of RC special structural walls can be calculated as follows Asw
VRd , s = zf ywd cot θ (2)
s
(
Vn = Acv α c λ f c′ + ρt f yt ) If αs = MEd/(VEd Lw) < 2.0, the shear strength is given
by VRd
ACI 318-19 also states that the value of Vn shall not exceed
0.83Acw√fcꞌ (in N).
VRd = VRd , c + 0.75ρh f yd , h bwo α s Lw (3)
Eurocode 8 (EN 1998-1:2004)
According to Eurocode 85 (EC8), the shear strength of RC
Hwang and Lee’s model
walls subjected to earthquake loadings can be taken as the
Hwang and Lee15 proposed a softened strut-and-tie model
lesser value of shear resistance from two failure modes: 1)
for calculating the shear strength of RC walls. The model
diagonal compression failure VRd,max; and 2) diagonal tension
has the term “softened” because it takes into account the
failures, either VRd,s or VRd.
softening behavior of cracked concrete. In the model, the
Diagonal compression failure of web due to shear—For
external forces were resisted by a combination of concrete
the case of diagonal compression failure, the shear strength
compression struts and steel tension ties as shown in Fig. 1.
is calculated as follows
There are three load paths—that is, vertical, horizontal, and
diagonal components which are calculated according to their
VRd , max = α cwbw zv1 f cd / ( cot θ + tan θ) (1) relative stiffness (Rv, Rh, and Rd), and these components are
combined together to become the diagonal compression
where the recommended value of αcw is as follows force acting on nodal zone Cd. The nominal capacity of the
nodal zone can be calculated using Eq. (4a). Then, the shear
1.0 for non-prestressed structures (1a) strength of RC wall according to this model can be taken as
the horizontal component of the diagonal compression force
(1.0 + σcp/fcd) for 0 < σcp ≤ 0.25fcd (1b) that is corresponding to the nominal capacity of the nodal
zone
1.25 for 0.25fcd < σcp ≤ 0.5fcd (1c)
Cd , n = K ζf c′Astr (4a)
2.5 (1.0 – σcp/fcd) for 0.5fcd < σcp < 1.0fcd (1d)
where K is strut-and-tie index, which is defined as follows
The recommended value for v1 is 0.6 [1.0 – fck/250] (fck
in MPa).
EC8 recommends that the values of cotθ and tanθ are
taken as 1.0.

Fh F
−D +
+ v
K= cos θ sin θ ≥ 1.00
Fh  sin 2 θ  Fv  cos 2 θ 
−D + 1 − + 1 −
cos θ  2  sin θ  2 
(4b)
and ζ is softening coefficient of cracked diagonal concrete

strut, which in this model, it is calculated as (3.35/√fcꞌ) ≤ 0.52.
Kassem’s model
Kassem16 proposed a strut-and-tie model and closed-form
design formula for predicting the shear strength of squat
walls. The model uses three shear-resisting mechanisms—
that is, diagonal, horizontal, and vertical mechanisms—
similar to Hwang-Lee’s model.15 In this model, a parametric
expression to calculate the shear strength of squat walls was
developed and calibrated using data of 645 walls obtained
from literature. The design formulas developed are as
follows (in SI units):
For walls with rectangular cross section
Fig. 2—Equilibrium of proposed strut-and-tie model.
Vn = 0.27 f c′  ψks sin ( 2θ) + 0.11ω h H ′ + 0.30ω v cot (θ) tw d w ≤ 0.83 f c′tw d w mechanism, a resultant force R is used to replace the axial
 dw  load P and tension force T in the equilibrium equation. The
(5a) resultant force R and lateral load V are equilibrated at point
A by diagonal compression force D and thus, it forms a strut-
For walls with flanged cross section and-tie model. The diagonal compression force D is equil-
ibrated at point B by compression force C and horizontal
Vn = 0.47 f c′  ψks sin ( 2θ) + 0.15ω h H ′ + 1.76ω v cot (θ) tw d w ≤ 1.25 reaction force that is equal to V. The governing failure mode
f c′tw d w
 dw  of the model is crushing of diagonal compression strut
(5b) which represents shear failure of the wall web. The internal
and external forces equilibrium of the model is described as
PROPOSED STRUT-AND-TIE MODEL follows
In this study, an analytical model for predicting RC
low-rise wall shear strengths was developed based on the R = C = Dsinθ (6)
strut-and-tie concept. The behavior of RC low-rise wall
having a height-length ratio (Hw/Lw) less than 2.5 is domi- V = Dcosθ (7)
nated by shear mode18,19 and it can be categorized as a
disturbed region where a plane section does not remain Determination of depth of compression zone at
plane and shear stress is not uniform within the wall panel. bottom of wall
Thus, the strut-and-tie model is considered a more appro- In this model, depth of compression zone at the bottom
priate approach to predict the strength as compared to the of wall c as displayed in Fig. 2 has to be determined first
sectional design model which includes concrete resis- before calculating the diagonal compression strut capacity.
tance to shear Vc due to tensile stresses in concrete.14,20 In Initially, the authors calculated the depth of compression
contrast to Hwang and Lee’s softened strut-and-tie model15 zone c based on flexural theory with the assumption of
that uses three compression struts, the model developed in linear strain distribution along the wall cross section. Never-
this study uses only one diagonal compression strut to be theless, this assumption led to inaccurate predictions of RC
simple. Furthermore, the contribution of web reinforcement wall shear strengths. This was because the assumption might
is accounted in terms of confinement effect to the diagonal not be valid for RC low-rise wall that can be categorized as
compression strut. disturbed region in which a plane section does not remain
plane. Thus, in this model, the value of c is calculated using
Equilibrium of proposed strut-and-tie model a formula that was derived using nonlinear FEA.
Initially, a typical RC low-rise wall with axial load P and First, some parameters that influence the depth of the
lateral load V as displayed in Fig. 2 has reaction forces at the compression zone were identified. Based on flexural theory,
bottom of the wall—that is, horizontal reaction force that these parameters are concrete strength fcꞌ, vertical reinforce-
is equal to V, vertical reaction force, and bending moment ment area in the edge column or boundary element Asb, and
that can be represented by a combination of tension force value of axial load P. Referring from the flexural theory for
T and compression force C. To simplify the load transfer a member subjected to axial load and bending moment, it is

Fig. 3—Comparison of envelope curves between experimental result and finite element analysis (FEA): (a) Specimen S-7 tested
by Gupta and Rangan12; and (b) Specimen J7 tested by Teng and Chandra.22 (Note: 1 kN = 0.22 kip.)
clear that the value of c decreases if the value of fcꞌ increases.
In contrast, the value of c increases if the value of Asb or P
increases. Moreover, the authors added shear span ratio or
wall height-length ratio (Hw/Lw) as additional parameter that
affects the value of c. This was because in similar cases of
disturbed region—that is, deep beams—it was shown that
the value of c increases if the shear span ratio decreases.21
Second, after identifying parameters influencing the value
of c and their qualitative relationships, the following step
was to determine quantitative relationships between these
parameters and the value of c. The main objective was to
express the value of c as a function of these parameters (fcꞌ,
Asb, P, and Hw/Lw). For this objective, nonlinear FEA using
ATENA software17 was used to determine multiplication
factors for each parameter. The software was used because
its superior capability to perform nonlinear analysis of RC Fig. 4—State of normal stresses in vertical axis at maximum
structures and it has advanced material models for concrete lateral load of typical wall specimen analyzed using ATENA
and steel reinforcement. In this study, RC wall specimen S-7 software.17
having Hw/Lw of 1.0 tested by Gupta and Rangan12 and RC An example of analysis results of a typical wall specimen is
wall specimen J7 having Hw/Lw of 2.0 tested by Teng and displayed in Fig. 4.
Chandra22 were used to validate the accuracy of the finite In total, 108 specimens were analyzed and the values of c
element model. The comparison of envelope curves between obtained at the peak loading condition of each specimen were
experimental result and FEA of those specimens is displayed measured. These values were then plotted against varying
in Fig. 3. It can be seen that the finite element model can parameters to obtain the quantitative relationships. These
predict well the force-drift relationship of those specimens. relationships can be seen in Fig. 5 to 7. The parameters fcꞌ
Subsequently, a parametric study using a typical wall and P are combined into one and normalized with wall web
specimen similar to the ones tested by Teng and Chandra22 area to become P/(fcꞌAw) because this is more frequently used
with varying parameters mentioned earlier was done to as a parameter. From the figures, it can be seen that the value
obtain the value of c at the peak loading condition of each of c increases linearly with increment of P/(fcꞌAw) and Asb/
specimen. For concrete strength, two values were used—that Aw. In contrast, the value of c decreases exponentially with
is, fcꞌ = 50 and 100 MPa (7.25 and 14.50 ksi). For vertical increment of Hw/Lw. These analysis results are consistent
reinforcement area in the edge column or boundary element, with qualitative relationships mentioned previously. Hence,
two values were used—that is, Asb = 1200 and 2400 mm2 the value of c can be expressed as follows
(1.86 and 3.72 in.2). For axial force, three values of axial
load ratio (ALR) were used—that is, 0.0, 0.1, and 0.2. For c4
Hw/Lw, three values were used—that is, 0.4, 1.0, and 2.0. In  P A H 
c = Lw  c1 + c2 + c3 sb   w  ≤ d w (8)
addition, the authors also attempted to vary the boundary  f c′Aw Aw   Lw 
element width bf—that is, 120, 250, and 500 mm (4.72, 9.84,
Equation (8) contains four constants that need to be deter-
and 19.69 in.)—to observe the relationship between c and bf.
mined. Constants c2, c3, and c4 can be derived from Fig.

Fig. 5—Values of c/Lw obtained from nonlinear FEA plotted against P/[fcꞌAw]: (a) cases for bf = 120 mm (4.72 in.); (b) cases
for bf = 250 mm (9.84 in.); and (c) cases for bf = 500 mm (19.69 in.).
Fig. 6—Values of c/Lw obtained from nonlinear FEA plotted against Asb/Aw: (a) cases for bf = 120 mm (4.72 in.); (b) cases for
bf = 250 mm (9.84 in.); and (c) cases for bf = 500 mm (19.69 in.).

Fig. 7—Values of c/Lw obtained from nonlinear FEA plotted against Hw/Lw: (a) cases for bf = 120 mm (4.72 in.); (b) cases for
bf = 250 mm (9.84 in.); and (c) cases for bf = 500 mm (19.69 in.).
Fig. 8—Relationships between c and varying parameters with average regression lines and their equations.
5 to 7 by plotting regression lines for each data series. From the figure, the values of c2, c3, and c4 were determined as
the equations of the regression lines, the constants were 0.5, 6.0, and –0.4, respectively. Subsequently, the value of c1
obtained and then the average constant value from all data was obtained by trial-and-error approach to achieve the most
series was calculated. The average regression lines as well suitable values of c that were in good agreement with the
as the average constant values are presented in Fig. 8. From values of c obtained from nonlinear FEA. Thus, the value of

c1 was found to be 0.35. Moreover, from the nonlinear FEA, Vn = Dncosθ (15)
it was noted that the value of c should not be taken greater
than effective depth of wall dw. In this model, dw is defined An example of RC wall shear strength calculation using the
as the distance from center to center of the edge columns or proposed strut-and-tie model can be seen in the Appendix.
boundary elements or it can be taken as 80% of wall length
0.8Lw in the case of an RC wall without edge columns or COMPARISON WITH EXPERIMENTAL RESULTS
boundary elements. To examine the accuracy of the proposed strut-and-tie
model, experimental wall strengths of 100 specimens
Capacity of diagonal compression strut collected from past experiments on RC low-rise walls
Capacity of the diagonal compression strut Dn is a product failing in shear12,19,22,25-37 were compared with calculated
of effective strut strength ζfcꞌ and the strut area Astr as shear strengths from the model. Subsequently, the predic-
described by tions from the proposed strut-and-tie model were also
compared with predictions from building codes4,5 and other
Dn = ζf c′Astr (9) strut-and-tie models.15,16 The analysis results are presented
in terms of ratio of the experimental shear strengths to calcu-
In this model, the value of effective strut strength is taken lated shear strengths Vexp/Vn. The ratio below 1.00 means that
as recommended by Eurocode 2.23 The code considers reduc- the prediction overestimates the shear strength whereas the
tion of concrete strut strength due to tensile stresses that ratio above 1.00 means that the prediction underestimates
cause cracks in the concrete strut. Moreover, in this model, the shear strength. These results are presented in Table 1.
increment of concrete strut strength because of confinement Moreover, the ratio was also plotted against Hw/Lw (refer to
effect from transverse reinforcement is also considered using Fig. 9) and fcꞌ (refer to Fig. 10) to observe the variation of
recommendation by FIP Commission 3.24 Thus, the soft- predictions as affected by those parameters.
ening coefficient for strut strength ζ in this model can be From the statistical parameters of Vexp/Vn as presented in
described as follows Table 1, it can be concluded that on average, all methods are
conservative in predicting the shear strength of RC low-rise
ζ = 0.6 (1 − fc′/250) × 0.80 (1 + 1.6αwωw) ≤ 0.85 (10) walls. Hwang and Lee’s model15 has the average value
Vexp/Vn of 1.29, which is the closest to 1.00 but overestimates
where αw and ωw are defined as the shear strength of many RC low-rise walls (approximately
s 22 out of 100 specimens), whereas the proposed strut-and-tie
α w = 1.6 ≤ 0.4 (11) model only overestimates seven out of 100 specimens. This
tw
means Hwang and Lee’s model15 needs a lower strength
reduction factor (below 0.68) as compared to the proposed
ρf y model (approximately 0.83) to ensure safe predictions of RC
ωw = 4 (12) low-rise wall shear strengths. The proposed model has the
f c′
average value Vexp/Vn of 1.35, which is the second closest
Because the definition of transverse reinforcement herein to 1.00 and it has the lowest coefficient of variation (COV)
that provides confinement effect to the concrete strut is the of 0.19 as compared to other methods. This is definitely
one that is perpendicular to the strut axis, it is needed to an advantage of the proposed model over other methods.
represent vertical and horizontal web reinforcement of the Furthermore, as can be seen in Table 1, Eurocode 85 is indeed
RC low-rise wall to be the transverse reinforcement of the the most conservative method with average value Vexp/Vn of
concrete strut as defined by FIP Commission 3.24 Therefore, 2.13 and the code underestimates the shear strength of all
in this model, the term ρfy is represented as 100 specimens collected in this study. Moreover, Kassem’s
model16 has the highest COV of 0.37 with the average value
ρf y = ρv f yv cos θ + ρh f yh sin θ (13) Vexp/Vn of 1.40 while ACI 3184 has similar average value
Vexp/Vn (1.41) and slightly lower COV (0.35).
where θ is defined as From Fig. 9 and 10, it can be seen that the predictions of
 H′  the proposed strut-and-tie model are quite consistent and less
θ = tan −1  (14) scattered for various ranges of Hw/Lw and fcꞌ, as compared
 Lw − r − 0.5c 
to the predictions by other methods. From Fig. 9, it can be
In this model, the value of θ is limited to 31 ≤ θ ≤ seen that ACI 318-194 is more conservative for walls with
59 degrees. lower Hw/Lw while it is the opposite for Hwang and Lee’s
The strut area Astr is defined as a product of strut depth model15 and Kassem’s model.16 From Fig. 10, except for
multiplied by strut width. Strut depth as is the perpendicular Kassem’s model,16 it can be seen that the predictions of most
projection of depth of compression zone at the bottom of methods are closer to 1.00 for walls with fcꞌ below 60 MPa
wall c to the strut axis as displayed in Fig. 2, while strut (8.70 ksi) and they are more conservative for walls with
width can simply be taken as the thickness of wall web tw. fcꞌ above 60 MPa (8.70 ksi). In addition, the predictions of
Finally, the nominal wall shear strength due to crushing of Eurocode 85 are quite scattered for various ranges of Hw/Lw
diagonal compression strut Vn is defined as and fcꞌ, and there is no clear trend that can be observed from
these figures.

Table 1—Ratio of experimental and calculated wall shear strengths
Vexp/Vn
No. Specimen ID fcꞌ, MPa P/(fcꞌAg) Hw/Lw ρvfyv, MPa ρhfyh, MPa ACI 318-19 4
EC85
Hwang-Lee15 Kassem16 Proposed model
Hirosawa25
1 72 17 0.12 0.94 2.07 1.09 1.33 1.71 1.13 1.79 1.42
2 73 21 0.10 0.94 2.07 1.09 1.28 1.68 1.00 1.61 1.28
3 74 21 0.10 0.94 2.07 2.40 0.82 1.45 1.01 1.60 1.17
4 75 14 0.15 0.94 2.07 2.40 0.97 2.09 1.39 2.07 1.47
5 76 15 0.14 0.94 2.07 4.47 0.92 1.94 1.30 1.88 1.16
6 77 18 0.11 0.94 2.07 4.47 0.91 1.78 1.23 1.84 1.18
7 79 14 0.15 0.94 2.07 2.57 0.71 1.52 1.01 1.50 1.09
8 82 21 0.10 1.88 1.63 2.40 0.72 1.22 0.95 1.66 1.20
9 83 18 0.11 1.88 1.63 2.40 0.70 1.26 1.02 1.73 1.25
Barda et al.26
10 B1-1 29 0.00 0.46 2.72 2.48 1.65 3.94 1.23 0.99 1.52
11 B2-1 16 0.00 0.46 2.76 2.50 1.51 3.45 1.72 1.06 1.39
12 B3-2 27 0.00 0.46 2.72 2.56 1.48 3.23 1.18 0.93 1.29
13 B6-4 21 0.00 0.46 1.24 2.48 1.25 2.72 1.39 1.22 1.33
14 B7-5 26 0.00 0.21 2.65 2.51 1.56 4.64 1.09 0.98 1.11
15 B8-5 23 0.00 0.96 2.64 2.48 1.24 2.24 1.82 1.02 1.57
Cardenas et al.19
16 SW-7 43 0.00 1.00 3.44 1.12 1.30 2.06 0.88 1.45 1.03
17 SW-8 42 0.00 1.00 13.45 1.26 1.36 2.02 0.97 1.28 0.96
Corley et al.27
18 B2 54 0.00 2.40 1.54 3.35 0.76 1.31 1.04 0.92 1.04
19 B5 45 0.00 2.40 1.46 3.16 0.91 1.56 1.27 1.14 1.30
20 B6 22 0.14 2.40 1.48 3.22 1.10 1.96 1.56 1.28 1.78
21 B7 49 0.08 2.40 1.42 3.08 1.18 2.05 1.11 1.07 1.40
22 B8 42 0.09 2.40 1.32 6.65 0.94 1.38 1.13 1.00 1.31
23 B9 44 0.09 2.40 1.34 2.91 1.25 2.17 1.12 1.12 1.49
24 B10 46 0.09 2.40 1.35 2.92 0.90 1.56 0.81 0.80 1.17
25 F1 38 0.00 2.40 1.58 3.73 0.90 1.45 1.41 1.20 1.51
26 F2 46 0.08 2.40 1.44 2.92 1.13 1.96 0.91 0.91 1.28
Maeda 28
27 MAE03 58 0.03 0.55 3.83 3.83 1.46 2.82 1.02 0.96 1.09
28 MAE07 58 0.03 0.55 6.42 6.42 1.52 2.38 1.10 1.10 1.11
Okamoto 29
29 W48M6 82 0.02 0.74 4.44 4.44 1.10 1.99 0.88 0.70 1.16
30 W48M4 82 0.02 0.74 4.12 4.12 1.12 1.97 0.86 0.68 1.14
31 W72M8 82 0.02 0.74 7.24 7.24 1.33 1.89 1.20 0.95 1.41
32 W72M6 82 0.02 0.74 6.65 6.65 1.30 1.93 1.17 0.93 1.38
33 W72M8 102 0.02 0.74 7.24 7.24 1.23 1.93 1.14 0.88 1.40
34 W96M8 102 0.02 0.74 9.41 9.41 1.44 2.04 1.33 1.03 1.49
Gupta and Rangan 12
35 S-1 79 0.00 1.00 5.45 2.89 1.11 1.58 0.99 0.61 1.03
36 S-2 65 0.07 1.00 5.45 2.89 1.96 2.24 1.32 1.06 1.55
37 S-3 69 0.13 1.00 5.45 2.89 2.28 2.28 1.23 1.21 1.53
38 S-4 75 0.00 1.00 8.00 2.89 1.58 2.16 1.43 0.82 1.32
39 S-5 73 0.06 1.00 8.00 2.89 2.10 2.43 1.42 1.10 1.49

Table 1 (cont.)—Ratio of experimental and calculated wall shear strengths
Vexp/Vn
No. Specimen ID fcꞌ, MPa P/(fcꞌAg) Hw/Lw ρvfyv, MPa ρhfyh, MPa ACI 318-19 4
EC85
Hwang-Lee15 Kassem16 Proposed model
40 S-6 71 0.13 1.00 8.00 2.89 2.59 2.60 1.40 1.37 1.62
41 S-7 71 0.06 1.00 5.45 5.45 1.52 2.05 1.41 1.12 1.56
Kabeyasawa and Hiraishi 30
42 W-08 103 0.09 1.18 5.75 5.75 1.48 1.93 1.35 1.10 1.89
43 W-12 138 0.09 1.18 5.75 5.75 1.46 1.95 1.21 1.02 1.99
44 No. 1 65 0.13 1.18 1.58 1.58 2.25 2.19 1.11 0.91 1.48
45 No. 2 71 0.12 1.18 2.75 2.75 1.90 1.93 1.18 0.99 1.55
46 No. 3 72 0.12 1.18 4.22 4.22 1.60 1.84 1.23 1.08 1.59
47 No. 4 103 0.14 1.18 4.22 4.22 1.84 1.88 1.22 1.11 1.70
48 No. 5 77 0.11 1.76 4.22 4.22 1.41 1.50 1.07 0.94 1.55
49 No. 6 74 0.12 1.18 9.31 9.31 1.45 1.86 1.26 1.09 1.34
50 No. 7 72 0.12 1.18 7.92 7.92 1.57 2.01 1.34 1.18 1.50
51 No. 8 76 0.11 1.18 11.52 11.52 1.66 2.13 1.45 1.25 1.45
Farvashany et al.31
52 HSCW1 104 0.04 1.25 6.74 2.51 2.20 2.36 1.56 1.00 1.62
53 HSCW2 93 0.09 1.25 6.74 2.51 2.60 2.48 1.60 1.18 1.78
54 HSCW3 86 0.09 1.25 4.01 2.51 1.96 1.85 1.19 0.91 1.38
55 HSCW4 91 0.22 1.25 4.01 2.51 2.68 1.99 1.13 1.23 1.56
56 HSCW5 84 0.09 1.25 6.74 4.01 1.93 2.07 1.42 1.18 1.66
57 HSCW6 90 0.05 1.25 6.74 4.01 1.77 1.94 1.49 1.06 1.63
58 HSCW7 102 0.08 1.25 4.01 4.01 1.85 1.94 1.39 1.08 1.67
Burgueno et al. 32
59 M05C 46 0.08 2.25 6.54 8.14 1.85 2.68 2.46 1.68 1.62
60 M05M 39 0.09 2.25 6.54 8.14 2.14 3.23 2.76 1.89 1.81
61 M10C 56 0.06 2.25 7.00 8.71 1.56 2.19 2.22 1.46 1.39
62 M10M 84 0.04 2.25 7.00 8.71 1.53 2.09 2.43 1.62 1.51
63 M15C 102 0.03 2.25 7.07 8.80 1.27 1.77 2.09 1.45 1.37
64 M15M 111 0.03 2.25 7.03 8.75 1.38 1.98 2.33 1.65 1.54
65 M20C 131 0.03 2.25 6.44 10.69 1.11 1.72 1.92 1.43 1.35
66 M20M 115 0.03 2.25 6.44 10.69 1.34 1.95 2.27 1.59 1.49
Cheng et al.33
67 M60 39 0.00 0.94 1.39 1.39 0.92 1.76 0.69 1.24 0.93
68 M115 38 0.00 0.94 1.21 2.41 0.68 1.14 0.68 1.23 0.83
69 H60 44 0.00 0.94 3.89 3.89 0.87 1.37 1.08 1.81 1.12
70 H115 44 0.00 0.94 3.30 3.30 0.88 1.39 0.99 1.70 1.13
71 H60X 42 0.00 0.94 3.89 3.89 0.88 1.41 1.10 1.85 1.14
Teng and Chandra 22
72 J1 103 0.05 1.00 1.71 1.71 2.85 3.25 1.62 1.29 1.65
73 J2 97 0.05 1.00 4.34 1.71 3.05 3.48 1.75 1.20 1.71
74 J3 111 0.05 1.00 1.71 4.34 2.09 2.36 1.71 1.51 1.87
75 J4 94 0.05 1.00 1.71 1.71 1.97 2.35 1.44 2.06 1.19
76 J5 103 0.05 2.00 1.71 1.71 1.73 4.36 1.07 0.88 1.07
77 J6 97 0.05 2.00 4.34 1.71 2.14 5.30 1.33 0.98 1.29
78 J7 111 0.05 2.00 1.71 4.34 1.46 2.58 1.23 1.24 1.52

Table 1 (cont.)—Ratio of experimental and calculated wall shear strengths
Vexp/Vn
No. Specimen ID fcꞌ, MPa P/(fcꞌAg) Hw/Lw ρvfyv, MPa ρhfyh, MPa ACI 318-194 EC85 Hwang-Lee15 Kassem16 Proposed model
Baek et al. 34
79 NS2 37 0.07 2.00 5.17 4.37 1.34 2.10 1.98 3.36 1.72
80 HS2 37 0.07 2.00 3.74 4.54 1.30 2.04 1.93 3.31 1.72
81 NS2L 37 0.07 2.00 3.10 2.16 1.40 2.57 1.31 2.46 1.43
82 HS2L 37 0.07 2.00 2.80 2.27 1.45 2.63 1.41 2.64 1.54
Baek et al. 35
83 NS1M 53 0.07 1.00 5.17 4.37 1.26 1.77 1.37 2.21 1.31
84 HS1M 53 0.07 1.00 4.67 4.54 1.17 1.64 1.28 2.06 1.22
85 NS0.5M 45 0.07 0.50 4.32 4.37 1.50 2.33 1.19 2.55 1.25
86 HS0.5M 37 0.07 0.50 3.87 4.54 1.54 2.40 1.29 2.74 1.29
Baek et al. 36
87 SW1 20 0.00 2.50 0.82 1.14 0.91 1.90 1.36 2.25 1.30
88 SW2 20 0.25 2.50 0.82 1.14 1.14 2.36 0.92 1.62 1.28
89 SW3 20 0.00 2.50 0.97 1.08 0.90 1.90 1.30 2.15 1.21
90 SW4 20 0.25 2.50 0.97 1.08 1.17 2.48 0.92 1.63 1.26
91 SW5 37 0.14 2.50 0.82 1.14 1.20 2.86 0.79 1.42 1.06
92 SW6 37 0.14 2.50 0.97 1.08 1.21 2.95 0.79 1.41 1.02
Hube et al. 37
93 WSL1 29 0.00 1.00 1.21 1.21 0.85 1.31 0.77 1.29 0.87
94 WSL3 29 0.00 1.00 1.56 1.56 1.01 1.50 1.02 1.70 1.13
95 WSL4 29 0.00 1.00 0.88 0.88 1.13 1.80 0.92 1.51 1.03
96 WSL5 29 0.00 1.00 0.89 0.89 1.00 1.58 0.82 1.34 0.91
97 WSL6 29 0.00 1.00 0.62 0.62 1.12 1.86 0.84 1.34 0.92
98 WSL7 29 0.00 1.00 1.51 1.51 0.89 1.34 0.87 1.46 1.00
99 WSL8 29 0.00 1.00 1.07 1.07 0.94 1.49 0.80 1.34 0.93
100 WSL9 29 0.00 1.00 1.12 1.12 1.07 1.67 0.92 1.54 1.07
Statistical parameters
Minimum value 0.68 1.14 0.68 0.61 0.83
Maximum value 3.05 5.30 2.76 3.36 1.99
Average value 1.41 2.13 1.29 1.40 1.35
Standard deviation 0.50 0.70 0.41 0.52 0.25
Coefficient of variation 0.35 0.33 0.32 0.37 0.19
CONCLUSIONS It is generally known that a strut-and-tie model serves as a

The authors have developed an analytical method based lower-bound theory.14
on the strut-and-tie concept to calculate the shear strength 2. As compared to building codes4,5 and other strut-and-tie
of reinforced concrete (RC) low-rise walls. The following models,15,16 the proposed strut-and-tie model has the lowest
conclusions can be made: coefficient of variation (0.19) in predicting the shear strength
1. The proposed strut-and-tie model was verified with a total of RC low-rise walls. This is clearly an advantage of the
of 100 RC low-rise walls (wall height-length ratio [Hw/Lw] proposed model over other methods. In addition, the predic-
less than 2.5) failing in shear that were selected from avail- tions of the proposed model are also quite consistent and less
able literature.12,19,22,25-37 The analysis results show that the scattered for wide ranges of Hw/Lw and concrete compressive
model is conservative in predicting the shear strength of strengths.
RC low-rise walls with an average value of the ratio of the
experimental shear strengths to calculated shear strengths AUTHOR BIOS
Vexp/Vn of 1.35. While Hwang and Lee’s model15 has the Jimmy Chandra is an Assistant Professor and currently Head of Struc-
tural Engineering Laboratory at Petra Christian University, Indonesia.
average value Vexp/Vn of 1.29, which is the closest to 1.00, He received his Doctor of Philosophy degree from Nanyang Technological
it overestimates the shear strength of 22 specimens whereas University, Singapore. His research interests include behavior and seismic
the proposed model only overestimates seven specimens. performance evaluation of reinforced concrete structures.

Fig. 9—Vexp/Vn plotted against Hw/Lw.
Fig. 10—Vexp/Vn plotted against fcꞌ. (Note: 1 MPa = 145.04 psi).

ACI member Susanto Teng is an Associate Professor at Nanyang Tech- through School of Civil and Environmental Engineering is also very much
nological University. He is a member of ACI Committee 435, Deflection appreciated.
of Concrete Building Structures; and Joint ACI-ASCE Committees 421,
Design of Reinforced Concrete Slabs, and 445, Shear and Torsion. His
research interests include behavior of structural concrete walls, shear NOTATION
strength of slabs, size effect in shear behavior of concrete members, compu- Acv = gross area of concrete section bounded by web thickness and
tational modeling of concrete structures, and durability of marine concrete length of section in direction of shear force considered
structures. Acw = area of concrete section of individual vertical wall segment
considered
Ag = wall gross cross-sectional area
ACKNOWLEDGMENTS Asb = total area of vertical reinforcement in one boundary element
This research is part of the Competitive Research Program “Underwater Astr = area of diagonal concrete strut
Infrastructure and Underwater City of the Future” funded by the National Asw = cross-sectional area of shear reinforcement
Research Foundation (NRF) of Singapore. The authors are grateful for Aw = wall web area
the funding. Support by Nanyang Technological University, Singapore, as = depth of diagonal concrete strut
bf = width of boundary element

bw = minimum width (thickness) of wall between tension and ρv = average vertical web reinforcement ratio
compression chords σcp = mean compressive stress, measured positive, in concrete due to
bwo = width of wall web design axial force
C = compression force in compression zone ωh = horizontal web reinforcement index which can be defined as [ρh
Cd = diagonal compression force acting on nodal zone fyh /fcꞌ]
Cd,n = nominal capacity of nodal zone ωv = vertical web reinforcement index which can be defined as [ρv fyv
c = depth of compression zone at bottom of wall /fcꞌ]
D = compression force in diagonal strut ωw = coefficient taking account of confinement effect of web rein-
Dn = nominal strength of diagonal concrete strut forcement to concrete strut strength, related to ratio of web
dw = effective depth of wall reinforcement
Fh = tension force in horizontal tie ψ = a non-dimensional function which can be defined as [0.95 −
Fv = tension force in vertical tie fcꞌ/250] (fcꞌ in MPa)
fcꞌ = concrete cylinder compressive strength ζ = softening coefficient of concrete in compression
fcd = design value of concrete compressive strength
fck = characteristic compressive cylinder strength of concrete at
28 days REFERENCES
fy = specified yield strength of reinforcement. 1. Fintel, M., “Shearwalls − An Answer for Seismic Resistance,”
fyb = yield strength of vertical reinforcement in boundary element Concrete International, V. 13, No. 7, July 1991, pp. 48-53.
fyd,h = design value of yield strength of horizontal web reinforcement 2. Chandra, J.; Liu, Y.; and Teng, S., “Analytical Study on High Strength
fyh = yield strength of horizontal shear reinforcement Concrete Shear Walls,” 36th Conference on Our World in Concrete and
fyt = yield strength of transverse reinforcement. Structures, Singapore, 2011, pp. 221-230.
fyv = yield strength of vertical shear reinforcement 3. Chandra, J., and Teng, S., “Shear Strength of Normal to High Strength
fywd = design yield strength of shear reinforcement Concrete Walls,” The 1st International Conference on Sustainable Civil
Hw = height of wall Engineering Structures and Construction Materials, Yogyakarta, Indonesia,
Hꞌ = distance measured from point of application of external shear 2012, pp. 138-145.
force to wall base 4. ACI Committee 318, “Building Code Requirements for Structural
K = strut-and-tie index Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
ks = ratio of depth of compression zone at wall base to effective Concrete Institute, Farmington Hills, MI, 2019, 623 pp.
depth of wall 5. Comite Europeen de Normalisation, “Eurocode 8: Design of Struc-
Lw = wall length tures for Earthquake Resistance Part 1: General Rules, Seismic Actions
MEd = design bending moment at base of wall and Rules for Buildings (EN 1998-1),” Comite Europeen de Normalisation
P = axial load applied at top of wall (CEN), Brussels, Belgium, 2004.
R = resultant force of external axial force and tension force in 6. Ritter, W., “Die Bauweise Hennebique,” Schweizerische Bauzeitung,
tension tie V. 33, No. 7, 1899, pp. 59-61.
Rd = wall shear ratio resisted by diagonal mechanism 7. Morsch, E., Concrete-Steel Construction, McGraw-Hill, New York,
Rh = wall shear ratio resisted by horizontal mechanism NY, 1909, 368 pp.
Rv = wall shear ratio resisted by vertical mechanism 8. Vecchio, F. J., and Collins, M. P., “Modified Compression-Field
r = distance measured from point of application of resultant force to Theory for Reinforced Concrete Elements Subjected to Shear,” ACI Journal
nearest wall edge Proceedings, V. 83, No. 2, Mar.-Apr. 1986, pp. 219-231.
s = spacing of shear (web) reinforcement 9. Hsu, T. T. C., “Softened Truss Model Theory for Shear and Torsion,”
T = tension force in tension tie ACI Structural Journal, V. 85, No. 6, Nov.-Dec. 1988, pp. 624-635.
tf = thickness of boundary element 10. Hsu, T. T. C., and Mo, Y. L., “Softening of Concrete in Low-Rise
tw = thickness of wall web Shear Walls,” ACI Journal Proceedings, V. 82, No. 6, Nov.-Dec. 1985, pp.
V = applied external shear force 883-889.
Vc = concrete contribution to overall shear strength 11. Mau, S. T., and Hsu, T. T. C., “Shear Design and Analysis of
VEd = design shear force Low-Rise Structural Walls,” JACI Journal Proceedings, V. 83, No. 2, 1986,
Vexp = experimental wall shear strength pp. 306-315.
Vn = nominal shear strength of RC wall 12. Gupta, A., and Rangan, B. V., “High-Strength Concrete (HSC) Struc-
VRd = shear resistance of a member with shear reinforcement tural Walls,” ACI Structural Journal, V. 95, No. 2, Mar.-Apr. 1998, pp.
VRd,c = design shear resistance of a member without shear reinforcement 194-204.
VRd,max = design value of maximum shear force which can be sustained by 13. Kassem, W., and Elsheikh, A., “Estimation of Shear Strength of
member Structural Shear Walls,” Journal of Structural Engineering, ASCE, V. 136,
VRd,s = design value of shear force which can be sustained by the No. 10, 2010, pp. 1215-1224. doi: 10.1061/(ASCE)ST.1943-541X.0000218
yielding shear reinforcement 14. Schlaich, J.; Schafer, K.; and Jennewein, M., “Toward A Consis-
z = inner lever arm, which is taken as 0.8 Lw (Lw is wall length) tent Design of Structural Concrete,” PCI Journal, V. 32, No. 3, 1987, pp.
αc = coefficient defining relative contribution of concrete strength to 74-150. doi: 10.15554/pcij.05011987.74.150
nominal wall shear strength which may be taken as 0.25 for Hw/ 15. Hwang, S. J., and Lee, H. J., “Strength Prediction for Disconti-
Lw ≤ 1.5, 0.17 for Hw/Lw ≥ 2.0, and varies linearly between 0.25 nuity Regions by Softened Strut-and-Tie Model,” Journal of Structural
and 0.17 for Hw/Lw between 1.5 and 2.0; where Hw/Lw is height- Engineering, ASCE, V. 128, No. 12, 2002, pp. 1519-1526. doi: 10.1061/
to-length ratio of wall (ASCE)0733-9445(2002)128:12(1519)
αcw = coefficient taking account of state of stress in compression chord 16. Kassem, W., “Shear Strength of Squat Walls: A Strut-and-Tie Model
αw = coefficient taking account of confinement effect of web rein- and Closed-Form Design Formula,” Engineering Structures, V. 84, 2015,
forcement to concrete strut strength, related to spacing of web pp. 430-438. doi: 10.1016/j.engstruct.2014.11.027
reinforcement 17. Cervenka, V., “ATENA: Software for Analysis of Concrete and Rein-
λ = modification factor reflecting reduced mechanical properties of forced Concrete Structures,” Cervenka Consulting Ltd., Prague, Czech
lightweight concrete, all relative to normal weight concrete of Republic, 2012.
same compressive strength 18. Cardenas, A. E., and Magura, D. D., “Strength of High-Rise Shear
θ = angle between concrete compression strut and wall axis perpen- Walls - Rectangular Cross Section,” Response of Multistory Concrete Struc-
dicular to shear force (Eurocode 8) tures to Lateral Forces, SP-36, M. Fintel and J. G. MacGregor, eds., Amer-
θ = angle of diagonal compression strut with respect to horizontal ican Concrete Institute, Farmington Hills, MI, 1972, pp. 119-150.
axis (Hwang-Lee’s model, Kassem’s model, and proposed 19. Cardenas, A. E.; Russell, H. G.; and Corley, W. G., “Strength of
model) Low-Rise Structural Walls,” Reinforced Concrete Structures Subjected
v1 = strength reduction factor for concrete cracked in shear to Wind and Earthquake Forces, SP-63, J. Schwaighofer, ed., American
ρ = reinforcement ratio Concrete Institute, Farmington Hills, MI, 1980, pp. 221-242.
ρb = ratio of vertical reinforcement in boundary element 20. Joint ASCE-ACI Committee 445, “Recent Approaches to
ρh = average horizontal web reinforcement ratio Shear Design of Structural Concrete,” Journal of Structural Engi-
ρt = ratio of area of distributed transverse (horizontal) shear neering, ASCE, V. 124, No. 12, 1998, pp. 1375-1417. doi: 10.1061/
reinforcement to gross concrete area perpendicular to that (ASCE)0733-9445(1998)124:12(1375)
reinforcement

21. Park, R., and Paulay, T., Reinforced Concrete Structures, John Wiley Thickness of wall web, tw = 100 mm
& Sons, Inc., New York, 1975, 769 pp.
22. Teng, S., and Chandra, J., “Cyclic Shear Behavior of High Strength
Width of boundary element, bf = 500 mm
Concrete Structural Walls,” ACI Structural Journal, V. 113, No. 6, Thickness of boundary element, tf = 120 mm
Nov.-Dec. 2016, pp. 1335-1345. doi: 10.14359/51689158 Ratio of vertical reinforcement in boundary element, ρb =
23. Comite Europeen de Normalisation, “Eurocode 2: Design of Concrete
Structures - Part 1-1: General Rules and Rules for Buildings (EN 1992-1-
0.0388
1),” Comite Europeen de Normalisation (CEN), Brussels, Belgium, 2004. Yield strength of vertical reinforcement in boundary element,
24. FIP Commission 3, “Practical Design of Structural Concrete,” FIB fyb = 630 MPa
- International Federation for Structural Concrete, London, UK, 1999,
113 pp.
Ratio of vertical shear (web) reinforcement in wall, ρv =
25. Hirosawa, M., “Past Experimental Results on Reinforced Concrete 0.0028
Shear Walls and Analysis on Them,” Building Research Institute, Ministry Yield strength of vertical shear reinforcement, fyv = 610 MPa
of Construction, Tsukuba, Japan, 1975, 279 pp.
26. Barda, F.; Hanson, J. M.; and Corley, W. G., “Shear Strength of
Ratio of horizontal shear (web) reinforcement in wall, ρh =
Low-Rise Walls with Boundary Elements,” Reinforced Concrete Structures 0.0028
in Seismic Zones, SP-53, N. M. Hawkins and D. Mitchell, eds., American Yield strength of horizontal shear reinforcement, fyh =
Concrete Institute, Farmington Hills, MI, 1977, pp. 149-202.
27. Corley, W. G.; Fiorato, A. E.; and Oesterle, R. G., “Structural Walls,”
610 MPa
Significant Developments in Engineering Practice and Research, SP-72, Experimental wall shear strength, Vexp = 595.76 kN
M. A. Sozen, ed., American Concrete Institute, Farmington Hills, MI, 1981,
pp. 77-132.
28. Maeda, Y., “Study on Load-Deflection Characteristics of Reinforced
Calculation of nominal shear strength (Vn)
Concrete Shear Walls of High Strength Concrete - Part 1 Lateral Loading according to proposed strut-and-tie model:
Test,” Research Institute Maeda Construction Corporation, Tokyo, Japan, 1. Calculate c using Eq. (8) and the corresponding Astr
1986, pp. 97-107. (in Japanese)
−0.4
29. Okamoto, S., “Study on Reactor Building Structure using Ultra-High  P A H 
Strength Materials: Part 1. Bending Shear Test of RC Shear Wall - Outline c = Lw  0.35 + 0.5 + 6 sb   w  ≤ dw
(in Japanese),” Summaries of technical papers of annual meeting, Architec-  f c′Aw Aw   Lw 
tural Institute of Japan, Tokyo, Japan, 1990, pp. 1469-1470. −0.4
30. Kabeyasawa, T., and Hiraishi, H., “Tests and Analyses of High-  1, 012, 000 2328   2000 
c = 1000  0.35 + 0.5 +6  
100, 000   1000 
Strength Reinforced Concrete Shear Walls in Japan,” High-Strength
Concrete in Seismic Regions, SP-176, C. W. French and M. E. Kreger, eds.,  103.3 × 100, 000
American Concrete Institute, Farmington Hills, MI, 1998, pp. 281-310.
31. Farvashany, F. E.; Foster, S. J.; and Rangan, B. V., “Strength and
c = 408.23 mm ≤ 880 mm (OK)
Deformation of High-Strength Concrete Shearwalls,” ACI Structural
Journal, V. 105, No. 1, Jan.-Feb. 2008, pp. 21-29. Calculating T assuming yielding of reinforcement:
32. Burgueno, R.; Liu, X.; and Hines, E. M., “Web Crushing Capacity of
High-Strength Concrete Structural Walls: Experimental Study,” ACI Struc-
T1 from vertical reinforcement in boundary element
tural Journal, V. 111, No. 1, Jan.-Feb. 2014, pp. 37-48.
33. Cheng, M. Y.; Hung, S. C.; Lequesne, R. D.; and Lepage, A., “Earth- T1 = ρb × bf × tf × fyb
quake-Resistant Squat Walls Reinforced with High-Strength Steel,” ACI
Structural Journal, V. 113, No. 5, Sept.-Oct. 2016, pp. 1065-1076. doi:
10.14359/51688825 T1 = 0.0388 × 500 × 120 × 630
34. Baek, J. W.; Park, H. G.; Shin, H. M.; and Yim, S. J., “Cyclic Loading
Test for Reinforced Concrete Walls (Aspect Ratio 2.0) with Grade 550 MPa
(80 ksi) Shear Reinforcing Bars,” ACI Structural Journal, V. 114, No. 3,
T1 = 1466.64 kN
May-June 2017, pp. 673-686. doi: 10.14359/51689437
35. Baek, J. W.; Park, H. G.; Lee, J. H.; and Bang, C. J., “Cyclic Loading T2 from vertical web reinforcement that is in tension
Test for Walls of Aspect Ratio 1.0 and 0.5 with Grade 550 MPa (80 ksi)
Shear Reinforcing Bars,” ACI Structural Journal, V. 114, No. 4, July-Aug.
2017, pp. 969-982. doi: 10.14359/51689680 T2 = ρv ×(Lw − c − tf) × tw × fyv
36. Baek, J. W.; Park, H. G.; Choi, K. K.; Seo, M. S.; and Chung, L.,
“Minimum Shear Reinforcement of Slender Walls with Grade 500 MPa
(72.5 ksi) Reinforcing Bars,” ACI Structural Journal, V. 115, No. 3, May
T2 = 0.0028 × (1000 – 408.23 – 120) × 100 × 610
2018, pp. 761-774. doi: 10.14359/51701281
37. Hube, M. A.; Maria, H. S.; Arroyo, O.; Vargas, A.; Almeida, J.; and T2 = 80.58 kN
Lopez, M., “Seismic Performance of Squat Thin Reinforced Concrete
Walls for Low-rise Constructions,” Earthquake Spectra, V. 36, No. 3, 2020,
pp. 1074-1095. doi: 10.1177/8755293020906841 Calculating r by taking wall edge in tension as reference
point
APPENDIX T1 × arm1 + T2 × arm 2 + P × 0.5 Lw
An example of RC wall shear strength calculation using r=
T1 + T2 + P
the authors’ proposed strut-and-tie model is given herein. A
specimen taken from Teng and Chandra22 is used—that is,
Specimen J5. The procedure is given as follows (in SI units): 1466.64 × 0.5 × 120 + 80.58 × 120 + 0.5 × (1000 − 408.23 − 120)  + 1012 × 500
r=
1466.64 + 80.58 + 1012
Specimen J5 data:
Concrete compressive strength, fcꞌ = 103.3 MPa r = 243.31 mm
Wall gross cross-sectional area, Ag = 196,000 mm2
Axial load applied at top of wall, P = 1012 kN (compression) Calculate θ using Eq. (14)
Wall height, Hw = 2000 mm  H′ 
Wall length, Lw = 1000 mm θ = tan −1 
 Lw − r − 0.5c 

 2200  ρfy = 2.34
θ = tan −1 
 1000 − 243.31 − 0.5 × 408.23  ρf y
ωw = 4
f c′
θ = 75.9 degrees
2.34
ωw = 4
Then, take θ = 59 degrees 103.3
Calculating Astr ωw = 0.09
Astr = as × tw
ζ = 0.6 (1 − fc'/250) × 0.80 (1 + 1.6αwωw) ≤ 0.85
Astr = c × sinθ × tw
ζ = 0.6 (1 − 103.3/250) × 0.80 (1 + 1.6 × 0.4 × 0.09)
Astr = 408.23 × sin59° × 100
ζ = 0.30
Astr = 34,992.14 mm2
3. Calculate Dn using Eq. (9)
2. Calculate ζ using Eq. (10):
Calculating αw using Eq. (11) Dn = ζfc'Astr
s
α w = 1.6 ≤ 0.4 Dn = 0.30 × 103.3 × 34,992.14
tw
Dn = 1077.18 kN
200
α w = 1.6
100 4. Calculate Vn using Eq. (15)
αw = 3.2 Vn = Dncosθ
Then, take αw = 0.4 Vn = 1077.18 × cos59°

Calculating ɷw using Eq. (12)
Vn = 554.79 kN
ρfy = ρvfyv cosθ + ρhfyh sinθ
Thus, Vexp/Vn = 595.76/554.79 = 1.07.
ρfy = 0.0028 × 610 × cos59° + 0.0028 × 610 × sin59°

Title No. 119-S30
Modeling of Alkali-Silica Reaction-Affected Shear-Critical

Reinforced Concrete Structures
by Anca C. Ferche and Frank J. Vecchio
Analytical procedures for enhanced nonlinear finite element aging and damaged concrete structures that employ rational
analysis of shear-critical reinforced concrete structures affected analytical procedures, with appropriate constitutive models
by alkali-silica reaction (ASR) are presented. A novel model that for degradation mechanisms and are capable of analyzing
addresses the directional variations in the mechanical properties of structures under general loading conditions, represent a
ASR-affected concrete is developed; in it, the residual mechanical
fundamental research topic for modern structural appraisal.
properties are evaluated based on the sustained long-term stress
The work presented in this paper is centered in the realm
condition and on the severity of the expansion. The proposed model
is implemented within a nonlinear finite element analysis program of macro-modeling, where the emphasis lies on the global
and validation analyses are carried out to examine the accuracy of behavior of a structure. Therefore, the ASR-induced defor-
the methodology proposed, as well as to identify mechanisms that mations, stresses, and deterioration of the mechanical
have a significant influence on the analysis of ASR-affected spec- properties need to be considered appropriately. There are
imens that are prone to brittle failure. It is found that more accu- numerous models in the literature; the majority of them were
rate predictions are obtained when considering directionality in the implemented within the framework of an FE method. The
mechanical properties using the model developed. The results also kinetics of the reaction is typically based on experimental
indicate that for ASR-affected structures in the field, material infor- studies, while the behavior of concrete is simulated as either
mation from either damaged or undamaged concrete can be used linear elastic or nonlinear. Some approaches presented in the
as valuable information for numerical analysis.
literature adopt an elasto-plastic behavior for concrete2 or a
Keywords: alkali-silica reaction (ASR); beams; direction-dependent visco-elasto-plastic damage model,3 while others are based
mechanical properties; finite element analysis; panels; shear walls. on a smeared fixed crack model.4,5
Esposito and Hendriks6 proposed a classification of the
INTRODUCTION ASR models available in the literature based on the level
Concrete is one of the most used materials in the world as at which the input and output parameters were defined. The
it continues to be the material of choice in the construction goal of this comprehensive review was to identify models
industry. However, concrete is vulnerable to several factors that could be used for structural analysis. A total of 40
that can cause premature deterioration. The most common modeling approaches were grouped based on concrete expan-
and severe forms of deterioration are usually caused by a sion, internal pressure, gel production, and ions diffusion-
combination of factors and are linked with the volume reaction. Apart from the models based on concrete expan-
expansion of concrete and reinforcing bars. Among these sion, the other three categories were found to not be directly
deterioration mechanisms are endogenous chemical reac- applicable to the evaluation of the structural response of
tions such as alkali-silica reactivity (ASR), which lead to ASR-affected structures. In addition, the models based on
expansion and cracking of concrete that in turn may result in concrete expansion, which have structural assessment as
cover spalling and corrosion of the embedded reinforcement. their primary goal, were found to require a large number
As such, structures in need of assessment and advanced of input parameters that in many cases were not available.
modeling are oftentimes existing buildings experiencing The study concluded that reliable computational modeling
different levels of distress. of the effects of ASR on the behavior of structures remains
Population growth together with economic development an unsolved issue.
will exert significant strain on resources. The OECD1 report The ASCET (Assessment of Structures Subjected to
on the projection of global material resources estimates that Concrete Pathologies) program was organized by the Cana-
material use will rise from 89 Gt in 2017 to 167 Gt in 2060. dian Nuclear Safety Commission (CNSC) and by the U.S.
This growth will be reflected in all major types of materials. Nuclear Regulatory Commission (U.S. NRC) to address
Greenhouse gas emissions are strongly linked to material aging management of nuclear concrete structures, taking
use policies; today, concrete manufacturing accounts for into account the effect of ASR on structural deterioration. As
9% of the total greenhouse gas emissions, and it is projected part of the ASCET benchmark exercise, researchers around
that in 2060, 12% of the emissions will be due to concrete
production. MS No. S-2021-012.R2, doi: 10.14359/51734331, received April 4, 2021, and
The current environmental and economic climate dictates reviewed under Institute publication policies. Copyright © 2022, American Concrete
a trend of prudent maintenance, assessment, and rehabilita- obtained from the copyright proprietors. Pertinent discussion including author’s
tion of existing structures. Finite element (FE) analyses of closure, if any, will be published ten months from this journal’s date if the discussion

Fig. 1—Modification factors.
the world submitted predictions for the behavior of five shear characterize the concrete material. When a more advanced
wall specimens tested at the University of Toronto.7 Three analysis is required, additional information is needed to
of these walls experienced various levels of ASR-induced describe the concrete behavior. Generally, the complete
deterioration, and two of them were cast with nonreactive uniaxial stress-strain behavior has to be defined analyti-
concrete as control specimens. The reports8,9 of the ASCET cally. Numerous constitutive models have been developed
workshops summarize the various numerical approaches to describe the concrete compression and tension response
employed by researchers to model the wall specimens. The based on the mechanical properties measured from tests:
modeling techniques varied in terms of the software used, compressive strength, tensile strength, and Young’s modulus
the model type constructed, the concrete and reinforce- of elasticity.10 These models, however, were developed for
ment models, the ASR expansion model, load application, sound, undamaged concrete. The cracking caused by chem-
boundary conditions, and analysis type. ical reactions such as ASR influences the concrete stress-
Some of the software programs employed were Abaqus, strain behavior. The degree to which the behavior of reac-
LS-DYNA, VecTor2, VecTor3, FINAS/STAR, MS ESSI tive concrete differs from nonreactive concrete was found
Simulator, or in-house developed software. The concrete to be influenced by crack plane orientation and severity of
models used were alternately based on fracture mechanics, expansion.11-13
damage plasticity, rigid-body-spring modeling, or smeared This paper proposes a novel model for the mechanical
crack modeling. ASR-induced expansion was simulated properties of ASR-affected concrete. Macroscopic modeling
either as equivalent thermal expansion or based on consti- at the material level of the anisotropic damage induced
tutive models for ASR expansion available in the literature. by ASR is focused on the concrete compressive strength,
Some researchers found that the boundary conditions of the tensile strength, and modulus of elasticity. For sound,
walls’ base blocks had a significant role on the computed undamaged concrete, these mechanical properties have an
response, whereas others found them to have a marginal isotropic character. Material tests on ASR-affected concrete,
effect. There was also variability in the representation of the however, have revealed an anisotropy of the mechanical
boundary conditions of the top beam, with some researchers properties dependent on the magnitude of the ASR damage
allowing rotation and others restraining it. and on the orientation of the ASR-induced crack planes.12,14
In general, the peak strength of the wall specimens was The proposed model addresses the directional variations in
captured relatively well by all researchers. On the other the mechanical properties of ASR-affected concrete through
hand, high variability in the calculated ductility, stiffness, reduction functions applied to the concrete compressive
and shape of the hysteresis loops of the reactive walls was strength, tensile strength, and modulus of elasticity (Fig. 1).
found across the predictions submitted. These analytical The model was implemented within the algorithms of the
results were not as closely matched to the experimental nonlinear FE analysis program VecTor2,10,15 and validation
values as were the ultimate strengths of the walls. studies were performed on ASR-affected specimens. The
One of the major conclusions resulting from this bench- reinforced concrete elements examined included panel spec-
mark exercise was that macro-modeling of ASR-affected imens,16 shear wall specimens tested by Habibi et al.,7 and
reinforced concrete structures is heavily reliant on multiple shear-critical beams tested by Deschenes et al.17
influencing behavioral mechanisms, apart from the ASR The validation studies performed on the behavior of
expansion model employed. ASR-affected specimens were focused on specimens
Proper modeling of material mechanical behavior is susceptible to brittle failure mechanisms. Sensitivity
essential for reliable structural assessment. For conven- analyses were also conducted to identify mechanisms that
tional linear elastic assessment procedures, the compressive have a significant influence on the computed responses.
strength, tensile strength, Young’s modulus of elasticity, Specimens tested in the literature that experienced ductile
and Poisson’s ratio are typically necessary and sufficient to failure were investigated elsewhere.17,18

RESEARCH SIGNIFICANCE direction are, in part, influenced by the strain and stress
In structures suffering from ASR, the level of expansion states along the other two orthogonal directions. Therefore,
and the direction of the ASR-induced crack planes are influ- distinction is made between stress-free, uniaxial, biaxial, and
enced by the reactivity of the aggregate and by the sustained triaxial stress states. The modification factors are applied
three-dimensional stress state. In the majority of structures to the concrete compressive strength, tensile strength, and
in the field, concrete is experiencing different levels of modulus of elasticity (Fig. 1) as determined from sound
long-term stress along different directions; as a result, ASR- concrete with no ASR-induced damage. For a field structure,
induced deterioration of concrete mechanical properties is these values can be determined from material tests on cores
direction-dependent. The model proposed herein addresses extracted from an undamaged part of the structure. When
the anisotropy of the ASR-affected concrete mechanical this is not feasible, an estimation can be made based on the
properties. The comprehensive analyses of ASR-affected specified strength at 28 days and the age of the structure.
shear-critical reinforced concrete elements provide a novel A database was compiled with material test results
understanding of the implications of the direction-dependent for ASR-affected concrete, centered on the experiments
concrete mechanical properties on the structural response. conducted by Ferche and Vecchio.14 The majority of the
tests reported in the literature were performed on plain
MODELING APPROACH concrete specimens, conditioned in an unrestrained state, that
A two-phased analysis procedure was previously developed randomly oriented cracks forming a map-cracking
developed for the assessment of ASR-affected structures pattern. Although this type of test cannot be used to inves-
using nonlinear FE analysis.18,19 The procedure was incor- tigate directional variations in the mechanical proper-
porated within the program VecTor2,10,15 which employs a ties, they serve an important role for evaluating the effect
smeared, rotating crack model for concrete behavior.20 The varying levels of ASR-induced damage have on the concrete
first phase of the procedure involves the ASR analysis, which mechanical properties. The compiled results, presented in
differentiates and evaluates two different mechanisms: the tabular form in Appendix A* (Tables A.1, A.2, A.3, and A.4),
deterioration of the concrete mechanical properties, and the are categorized based on the stress state during the condi-
ASR-induced strains and stresses developed under long- tioning period, the ASR-induced expansion, and the testing
term loading. Using the results obtained from the first phase, direction relative to the long-term stresses. All the experi-
an analysis is then performed to estimate the response of the mental studies compiled contained control specimens cast
structure to externally applied short-term loads. with nonreactive aggregate against which the behavior of the
The linear unrestrained ASR expansion, needed for the reactive concrete was compared.
ASR analysis, can either be input by the user or calculated The modification factors k f p .i, k ft .i, and k Ec .i, determined
with one of the two models implemented that include a from the compiled database, were defined for the concrete
kinetics component. The ASR-induced strains are evaluated compressive strength, tensile strength, and modulus of elas-
along the principal directions and are carried through the ticity as per Eq. (1), (2), and (3). These factors essentially
analysis as strain offsets. These expansions are influenced represent the ratio of the mechanical properties of the reac-
by the stress conditions and so, in this phase of the analysis, tive concrete to the mechanical properties measured on the
the loads applied to the structure should be sustained long- nonreactive concrete, normalized to the initial difference
term loads only. For the changes in the concrete mechan- measured at 28 days.
ical properties, two isotropic options were made available: For the cases when tension tests were not performed at
user-defined properties, or properties calculated based on the 28 days, the square root of the ratios of the compressive
recommendations made by the Institution of Structural Engi- strengths were used to normalize the tensile tests results
neers,11 depending on the free expansion and the undamaged (Eq. (2)).
concrete strength at 28 days. The mechanical properties
of concrete were, therefore, assumed to be uniform in all f p.Ri / f p. N
directions. k f p .i (1)
f c.Ri / f c. N
Two additional options regarding the mechanical prop-
erties of ASR-affected concrete were subsequently imple-
mented as part of this work: the anisotropic formulation of f t .Ri / f t . N f / f
k ft .i = (a) or k ft .i t .Ri t . N (b) (2)
the model described in what follows and its corresponding f t .Ri.28 / f sp. N .28 f c.Ri / f c. N
simplified isotropic version.
Ec.Ri / Ec. N
Model formulation k ft .i = (3)
Ec.Ri.28 / Ec. N .28
In the model proposed herein, anisotropy in the mechan-
ical properties of concrete is quantified as a function of the Note that N refers to nonreactive concrete; R to reactive
reactivity potential of the aggregate and the long-term stress concrete; and i represents the principal direction (i = 1, 2, 3,
state. Modification factors are determined along each prin- where f3 < f2 < f1).
cipal direction for the concrete compressive strength k f p .i,
tensile strength k ft .i , and modulus of elasticity k Ec .i (i = 1, *
The Appendix is available at www.concrete.org/publications in PDF format,
2, 3 represents the principal direction where f3 < f2 < f1). It 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
is worth emphasizing that the modification factors in one time of the request.

In the case of unrestrained long-term conditions, the modi- compressive stresses lower than 0.3 MPa (44 psi) are treated
fication factors were defined to represent the lower-bound essentially as a stress-free case.
values from all data points compiled. The proposed modifi- The elementary strain ei(fi) is calculated as
cation factors are defined in Eq. (4), (5), and (6) as functions
of the linear unrestrained ASR expansion, εASR. The long- ei(fi) = fi/Ec (7)
term stress state of uniaxial tension is treated similarly to the
unrestrained case in all directions. where fi is the long-term compressive stress, along principal
direction i (i = 1, 2, 3); and Ec is the concrete modulus of
1.00 205.5 ASR ASR 1.8 103 elasticity of sound concrete at the current age of the structure.
Note that for the elementary strain, ei(fi), the label e was
k f . free 0.63 1.75 ASR 1.8 103 ASR 13.2 103
chosen as opposed to the customary label for strain ε. This
p
0.75 10.9 ASR 13.2 103 ASR 27.0 103
choice was made to emphasize that in calculating the elemen-
0.45 ASR 27.0 103 tary strain, the modulus of elasticity corresponding to sound,

(4) unaffected concrete is used, assuming linear behavior.
The mechanical properties degradation function proposed
for the compressed directions, kj.i(fi), is given in Eq. (8)
1.00 344.4 ASR ASR 1.8 103
1 1
k E . free
3
0.42 21.9 ASR 1.8 10 ASR 13.2 10
3
k j .i f i k j ,i. free
3 3
0.09

0.09
0.19 4.35 ASR 13.2 10 ASR 27.0 10

c
1.05 ASR 1.05 ASSR
0.07
ASR 27.0 103 ei f i ei 0
(5) (8)
where i = 1, 2, 3 represents the principal directions, where f3

1.00 475.9 ASR ASR 0.6 103 < f2 < f1 (compression negative); j = fp, Ec, ft; fi is the long-
term compressive stress along the principal direction i; kj.free
0.77 86.8 ASR 0.6 103 ASR 2.5 103 is the modification factor evaluated assuming unrestrained
k ft . free
0.70 60.0 ASR 2.5 103 ASR 5.0 103 stress conditions, according to Eq. (4), (5), and (6); εASR is
0.40 ASR 5.0 103 the unrestrained linear ASR expansion; ei(fi) is the elemen-

tary strain in the principal direction i, corresponding to the
(6)
long-term compressive stress fi; and ei(0) is the elemen-
tary strain in the principal direction i, corresponding to a
Shown in Appendix A, Fig. A.1, are the proposed relation-
compressive stress of 0.3 MPa (44 psi).
ships plotted versus the individual data points. The lower
To illustrate the relationship between the modification
bounds proposed by the Institution of Structural Engineers11
factors kfp, kEc, kft, and the long-term compressive stress fi,
in 1992 are also plotted. In general, the modification factors
Fig. 2 provides plots of the compressive stress for two levels
decrease as the ASR expansion increases. However, for
of ASR expansion: 0.5 × 10–3 and 2.0 × 10–3, and a modulus
the same level of expansion, a rather significant scatter in
of elasticity of concrete equal to 25,000 MPa (3625 ksi).
the experimentally determined modification factors can be
Constant modification factors were calculated up to a
seen. For the unrestrained case, good agreement can be seen
compressive stress level of 0.3 MPa (44 psi) based on the
between the proposed relationships and the ISE reduction
relationships proposed in Eq. (4), (5), and (6), similar to a
functions, although they were derived from different spec-
stress-free condition. For levels of compressive stress larger
imen databases.
than 0.3 MPa (44 psi), the modification factors increased
In most instances, however, the concrete component in a
with the increase in stress, reflecting the beneficial effect of
structure will experience sustained loading. In the case of
compressive stresses to counteract the ASR-induced dete-
a long-term compressive stress state, a distinction is made
rioration. The most affected mechanical property was the
between the restrained and unrestrained directions. The modi-
tensile strength, followed by the modulus of elasticity, and
fication factors along the restrained directions are a function
lastly by the compressive strength for both levels of ASR
of the stress-free modification factors (k f p . free, k E . free, and
c expansion.
k ft . free) defined by Eq. (4), (5), and (6), and the ratio of the
In Appendix A, Fig. A.2, ratios of the ASR strain to the
linear unrestrained ASR expansion to the elementary strain
elementary strain, εASR/ei(fi), are plotted with respect to the
εASR/ei(fi). The elementary strain ei(fi) in the principal direc-
long-term compressive stress fi for various levels of ASR
tion i is calculated as per Eq. (7). The proposed relation-
expansion, assuming a modulus of elasticity of concrete
ships for the modification factors are applicable to compres-
equal to 25,000 MPa (3625 ksi). Due to the nature of Eq. (8),
sive stresses higher than 0.3 MPa (44 psi). This value was
an increase in the strain ratio, εASR/ei(fi) corresponds to a
chosen based on available experimental data on the devel-
decrease in the calculated modification factors, which trans-
opment of ASR expansion as a function of the compressive
lates to a more pronounced degradation of the mechanical
stress.21 Compressive stresses lower than 0.3 MPa (44 psi)
properties. For the same level of long-term stress fi, the strain
were found not to influence the level of expansion. As such,
ratio εASR/ei(fi) increases with the increase in ASR strain

experimental measured results for the specimens that were
part of the compiled database. The tables summarize: the
levels of unrestrained linear ASR expansion εASR; the long-
term conditioning stress state; the testing direction relative
to the long-term stresses; the measured mechanical proper-
ties (compressive strength fp, modulus of elasticity Ec, and
tensile strength ft); the experimentally determined modifica-
tion factors for the mechanical properties, kj.exp; the calcu-
lated modification factors, kj.calc; and the ratios of calculated
to experimental values for the modification factors. The index
j represents the mechanical property under consideration (j =
fp, Ec, ft). A wide variance of the experimental modification
factors is observed for similar expansion levels. As such, the
adopted model was chosen to be conservative, reflected in
the mean of the ratios of calculated to experimental values
for the modification factors.
For analytical procedures that cannot accommodate aniso-
tropic implementation for the mechanical properties of
concrete, an isotropic version is proposed, using the average
Fig. 2—Modification factors versus long-term compres- modification factors. Therefore, the isotropic model applies
sive stress for Ec = 25,000 MPa. (Note: Ec = 25,000 MPa the average modification factors calculated along each prin-
was chosen for illustrative purposes only. The model is not cipal direction to the corresponding mechanical properties,
limited to this value for Ec. 1 MPa = 145 psi.) as shown in Eq. (12), (13), and (14).
(Fig. A.2), or with the increase in the undamaged modulus
of elasticity.
f p.isotropic
k f p .1 k f p .2 k f p .3 f (12)
The decision behind establishing this relationship between 3
p
the modification factors, the ASR-induced strain, and the

undamaged modulus of elasticity of concrete was made based
on empirical observations. A higher ASR-induced expansion
Ec.isotropic
k Ec .1 k Ec .2 k Ec .3 E (13)
usually led to a more pronounced degradation of mechan- 3
c
ical properties. Due to the ability of a more porous concrete

matrix to accommodate the expansive ASR gel before suffi-
cient pressure develops to cause cracking, concrete with
f t .isotropic
k ft .1 k ft .2 k ft .3 f (14)
reduced porosity was found to be more severely affected by 3
t
ASR deleterious effects compared to concrete with higher

porosity levels.22-24 Porosity is strongly associated with the
modulus of elasticity of concrete; a reduced porosity gener- Finite element implementation
ally results in a higher modulus of elasticity when the same To implement the proposed model for the mechanical
components are used in the mixture design.25 As the concrete properties of ASR-affected concrete, an iterative solution
porosity is a property that in the majority of cases is not as algorithm is required due to the interdependency of the
readily available as the modulus of elasticity, it was decided ASR-induced strains, the stress state, and the mechanical
to use the modulus of elasticity as an indicator of the porosity properties. Such was done within the algorithms of nonlinear
level, thus influencing the calculated modification factors. FE analysis program VecTor2 in two forms: an anisotropic
The modification factors are applied to the unaffected implementation with directional-dependency of the mechan-
mechanical properties of concrete as such ical properties, and an isotropic implementation, indepen-
dent of the orientation of the principal stress field. The FE
f p.i k f p .i f p (9) implementation for a two-dimensional approach is summa-
rized in what follows.
The anisotropic model for the mechanical properties of
Ec.i k Ec .i Ec (10) concrete consists of evaluating modification factors for the
compressive strength, modulus of elasticity, and the tensile
strength along each principal direction. The modification
f t .i k ft .i f t (11) factors are calculated based on Eq. (4), (5), (6), and (8) in an
iterative process and are applied to the undamaged mechan-
where i = 1, 2, 3 represents the principal directions, where ical properties according to Eq. (9), (10), and (11).
f3 < f2 < f1. Upon the completion of the ASR analysis, in the second
Appendix A contains Tables A.1, A.2, A.3, and A.4 stage of the analysis, the modification factors are reevalu-
summarizing the calculated modification factors versus the ated at each step as the orientation of the principal stress
field changes from the initial orientation.

Table 1—Panel specimen properties and test results
εASR, × 10–3
ID ρx, % ρy, % ρz , % fcp, MPa Ec, MPa νcr, MPa γcr, × 10–3 νu, MPa γu, × 10–3 Reactive aggregate
AF1 3.31 0.42 — 57.2 33,700 2.19 0.24 6.75 9.59 0.12*
AF2 3.31 0.84 — 58.4 33,500 2.74 0.32 8.64† 6.54† Nonreactive
AF3 3.31 0.42 — 38.2 18,300 4.34 0.53 6.99 6.50 2.32
AF4 3.31 0.84 — 41.3 18,600 4.32 0.35 9.77 7.37 Jobe-Newman
AF5 3.31 0.42 — 52.5 21,000 3.96 0.33 6.99 6.67 1.23
AF6 3.31 0.84 — 52.1 20,100 4.32 0.49 9.63 7.39 Spratt
AF7 3.31 0.42 1.69 46.3 21,200 4.90 0.50 7.33 7.07 2.49
AF8 3.31 0.84 1.69 47.1 19,400 5.67 0.87 10.42 7.83 Jobe-Newman
AF9 3.31 0.20 — 46.9 18,900 3.90 0.25 4.79 3.81 2.36
AF10 3.31 1.66 — 50.9 21,200 5.21 0.48 10.79 ‡
5.98 Jobe-Newman
*
Expansion primarily attributed to swelling due to water absorption.
†
Edge failure.
‡
Tested under cyclic loading.
Note: 1 MPa = 145 psi.
As such, kASR.i, lASR.i, and mASR.i are defined as the direction VecTor2 dated November 2019, and were constructed with
cosines for the orientation of the principal stress field with VecTor2’s pre-processor software, FormWorks 4.3.15 The
respect to the global reference axes at the end of the ASR post-processor Augustus26 was used to process and visualize
analysis; i = 1, 2, 3 represent the principal directions where the results.
f3 < f2 < f1 (compression negative). Similarly, the direction
cosines ki, li, and mi are defined for the current orientation of PANEL ELEMENTS
the principal stress field with respect to the global reference A set of 10 reinforced concrete panel elements tested by
axes; i = 1, 2, 3 represent the principal directions where f3 < Ferche and Vecchio16 were investigated as part of the vali-
f2 < f1 (compression negative). dation studies. The panels, 890 mm (35 in.) square x 70 mm
For two-dimensional analyses, the modification factors (2.7 in.) thick, contained varying amounts of in-plane and
corresponding to the current first and second in-plane prin- out-of-plane reinforcement, and were cast with either nonre-
cipal directions, kj.1c and kj.2c (j = fp, Ec, ft), are evaluated as active, reactive fine (Jobe-Newman), or reactive coarse
(Spratt) aggregate. The specimens were conditioned under
k j .1 k1kASR.1 l1lASR.1 m1mASR.1
2 elevated humidity and temperature to accelerate the reaction
rate. At the end of the conditioning period, the panels were
k j .1c k j .2 k1kASR.2 l1lASRR.2 m1mASR.2 (15) tested under in-plane pure shear loading conditions. The
2

following properties and test results are shown in Table 1
k j .3 k1kASR.3 l1lASR.3 m1mASR.3
2
for the panel specimens: the reinforcement ratios ρx, ρy, and
ρz; the concrete compressive strength and modulus of elas-
k j .1 k2 kASR.1 l2 lASR.1 m2 mASR.1
2
ticity fcp and Ec; shear stress and strain at cracking νcr and
k j .2 c k j .2 k2 kASR.2 l2 lASRR.2 m2 mASR.2 (16)
2 γcr; ultimate stress and strain νu and γu; the ASR expansion

measured on accompanying prisms specimens εASR; as well
k j .3 k2 kASR.3 l2 lASR.3 m2 mASR.3
2
as the type of reactive aggregate used in the mixture.
Each panel was modeled as a single four-node plane
where kj.1, kj.2, and kj.3 are the modification factors calculated stress rectangular element. The analyses were performed in
along the principal directions at the end of the ASR analysis, force-controlled conditions. The shear stress was increased
considering long-term sustained loading conditions only, by 0.10 MPa (14.5 psi) at each analysis step for the mono-
according to the model proposed. tonically tested panels and by 0.20 MPa (29 psi) for the
For a two-dimensional analysis procedure, the modifica- cyclically loaded panel. The reinforcement was modeled
tion factors calculated in the out-of-plane direction remain as smeared reinforcement with the mechanical properties
constant throughout the analysis, and m1 = m2 = 0.0. measured from coupon tests, summarized in Table C.1
Given in Appendix B is the nonlinear analysis algo- of Appendix C. Also summarized in Appendix C are the
rithm employed in VecTor2. Highlighted with red are the concrete and reinforcement stresses at the end of the condi-
steps where ASR strains are calculated and included in the tioning period, before the beginning of the shear test (Table
concrete prestrains vector, {εco}, and where the mechanical C.2). The concrete compressive stresses in the x-direction
properties modification factors are evaluated. varied between 3.74 and 6.06 MPa (542 and 879 psi), while in
Validation studies were performed on ASR-affected speci- the y-direction they were in the 0.89 to 3.33 MPa (129 to 483
mens to obtain an indication of the accuracy of the proposed psi) interval, depending on the reactivity of the mixture and
model. All the analyses were performed with the version of

the reinforcement ratios provided. The x-direction reinforce-
ment had stresses ranging from 86 to 183 MPa (12.4 to 26.5
ksi), while the y-direction reinforcement reached stresses
between 201 and 444 MPa (29.2 and 64.4 ksi). Perfect bond
was assumed between concrete and reinforcement.
For the nonreactive panels, the concrete properties speci-
fied were the concrete compressive strength and the modulus
of elasticity, as measured from the cylinders at test day. For
the ASR-affected panels, three different cases were consid-
ered in terms of the concrete properties:
1. Using the concrete compressive strength as measured
from reactive cylinders at test day that were cast and condi-
tioned with the reactive panels, employing no degradation
functions for the mechanical properties of concrete—labeled
Cylinder.
from nonreactive cylinders at test day that were cast with
the control panels, together with the anisotropic model for
degradation of mechanical properties proposed—labeled
Anisotropic.
from nonreactive cylinders at test day that were cast with
the control panels, together with the isotropic model for
degradation of mechanical properties proposed—labeled
Isotropic.
Two ASR-induced strain calculation models, the Charl-
wood et al.21 model and the Saouma and Perotti27 model,
were investigated to quantify the influence of ASR strain
calculation on the analytical response. The Charlwood model
calculates ASR expansion as a function of the compressive
stresses and treats each principal direction independently.
The Saouma and Perotti model assumes ASR expansion is
volumetrically distributed and that expansions in the three
principal directions are interdependent. Thus, for each of
the three cases considered in terms of concrete properties,
two analyses were performed based on the ASR expansion
model: one using the Charlwood model and the other one
using the Saouma and Perotti model.
In addition, for each reactive panel, an analysis was
performed neglecting the effect of ASR-induced expan-
sion and using the concrete mechanical properties as deter-
mined from the reactive cylinders test. This analysis was Fig. 3—Numerical versus experimental response for nonre-
performed as it is similar to the approach taken by several active panel specimens. (Note: 1 MPa = 145 psi.)
researchers when estimating the strength of field ASR-
panels. Panel AF2 suffered an edge failure soon after the
affected structures.28-30
y-direction reinforcement started yielding. Numerically, the
Prior to modeling the ASR-affected specimens, the accu-
failure shear stress and strain predicted were close to the
racy of VecTor2 in predicting the response of the nonreac-
ultimate shear stress and strain measured experimentally,
tive control panels was investigated. The FE analysis results
illustrated in Fig. 3(b). It should be pointed out that the rein-
for the nonreactive panels AF1 and AF2 are shown in Fig. 3
forcement was cold-formed and did not have a well-defined
in comparison to the experimentally measured responses.
yield plateau, which contributed to the discrepancy in the
The numerical results were reasonably similar to the exper-
analysis results.
imental findings. Analytically, the failure mechanism was
initiated by the yielding of the reinforcement in the Y-
Panel AF3
direction for both nonreactive panels. This was consistent
The analysis of representative reactive Panel AF3 is
with the experimental observations.
discussed in what follows. Table C.3 presents the stress
Panel AF1 exhibited higher shear stiffness upon yielding
and strain state of the panel, upon the completion of the
of the y-direction reinforcement compared to the numer-
ASR conditioning period, at the end of each of the various
ical model, as shown in Fig. 3(a). A similar observation
ASR analyses undertaken. Thus, Table C.3 summarizes
was previously noted by Luo31 and Carnovale32 on similar

cracking strength prediction, the Charlwood model was
conservative regardless of the approach considered for the
concrete mechanical properties. The results obtained with
either anisotropic or cylindrical properties closely matched
the experimental values, whereas the isotropic model
resulted in a lower cracking strength. In terms of overall
predicted behavior, the anisotropic model matched more
closely the experimental behavior compared to the isotropic
model or the one using the cylinder properties. Similar to the
control panel, for AF1, upon yielding of the y-direction rein-
forcement, the calculated response was less stiff compared
to the experimental one.
The Saouma and Perotti model, used with either the aniso-
tropic model or with the model considering cylinder prop-
erties, overestimated the cracking strength of the panel, as
illustrated in Fig. 4(b). This was symptomatic of an overesti-
mation of the ASR-induced prestress. However, the Saouma
and Perotti model used with the isotropic model for degrada-
tion of concrete mechanical properties matched the experi-
mental cracking strength closely. This was the result of more
severe degradation of the mechanical properties predicted
by the isotropic model compared to the anisotropic one
which, in this case, counteracted better the initial high level
of prestress calculated.
The results obtained neglecting the effects of ASR expan-
sion and using the reactive cylinder properties resulted in
a significant underestimation of the cracking strength and
ultimate capacity. The ultimate shear strain, however, was
better predicted (Fig. 4). An additional analysis was carried
out for Panel AF3 with both ASR-induced expansion and
deterioration neglected. This analysis also underestimated
the response compared to the experimental one.
Given in Appendix C in Tables C.4 and C.5 are the modi-
fication factors calculated along the principal directions for
the initial and final load stages. The modification factors
change as the direction of the principal stresses changes, as
detailed previously. In-depth analyses of the panel speci-
mens can be found elsewhere.33
Discussion—panel elements
The results are summarized for the monotonically
Fig. 4—Numerical versus experimental response for Panel
tested reactive panels in Table 2. For each type of analysis
AF3. (Note: 1 MPa = 145 psi.)
performed, the mean and coefficient of variation are shown
the concrete stresses fcx and fcy, the total strains εx.total and for the cracking shear stress ratio νcr.Calc/νcr.Exp, the ultimate
εy.total, and the concrete strains εx.ASR and εy.ASR in the x- and shear stress ratio νu.Calc/νu.Exp, and the ultimate shear strain
y-directions. The reported measured strains and stresses ratio γu.Calc/γu.Exp.
were based on measurements of total surface strains using The anisotropic mechanical properties model employed
Zurich gauges. The data show that the ASR expansion with the Charlwood model for ASR expansion yielded
model influenced the calculated initial stress and strain state superior predictions compared to the rest of the analyses,
significantly, compared to the approach taken to model the followed closely by the model using the cylinder properties.
mechanical properties, which had a marginal effect. In this The analyses performed with the Saouma and Perotti model
case, the Saouma and Perotti (labeled S&P) predictions for overestimate the cracking strength, whereas the ultimate
the ASR-induced strains matched the experimental measure- shear strength matched well with the experimental capacity.
ments to a closer degree than did the Charlwood model. Reactive Panel AF10 was subjected to cyclic shear loading.
As expected, neglecting the ASR expansion resulted in no The target peak shear stress of each cycle was 10.8 MPa.
initial strains and stresses for the panel specimens. The panel failed after 15 cycles. Shown in Table C.6 are the
Shown in Fig. 4(a) are the results obtained for Panel AF3 results obtained for each different analysis performed. The
using the Charlwood model for expansion. In terms of the cracking strength, post-cracking stiffness, and the number
of cycles to failure were better captured using either the

Table 2—Summary of results: reactive panels*
νcr.Calc/νcr.Exp νu.Calc/νu.Exp γu.Calc/γu.Exp
ASR expansion Mechanical
model properties Mean COV, % Mean COV, % Mean COV, %
Anisotropic 0.86 7.82 0.89 4.35 0.87 12.89
Charlwood Isotropic 0.74 6.93 0.81 5.83 0.92 17.79
Cylinder 0.97 6.87 0.86 4.45 0.88 7.60
Anisotropic 1.19 9.76 0.93 4.77 0.70 17.94
Saouma and Perotti Isotropic 1.05 10.08 0.84 4.42 0.80 28.65
Cylinder 1.27 6.86 0.91 6.44 0.76 8.79
Neglected Cylinder 0.50 13.40 0.80 5.37 1.12 9.82
*
Includes Panels AF3, AF4, AF5, AF6, AF7, AF8, and AF9.
proposed anisotropic model or the reactive cylinder proper-

ties together with the Charlwood model for expansion.
The following trends can be identified following the
analytical investigation of the ASR-affected panels:
1. The Charlwood model for evaluating ASR-induced
strains produced results that better matched the experimental
behavior of the panels (cracking strength, post-cracking stiff-
ness, post-yielding stiffness, ultimate capacity) compared to
the Saouma and Perotti model.
2. The influence of the model for the concrete mechan-
ical properties on the evaluation of ASR-induced strains was
significantly smaller compared to the influence of the ASR
expansion model. This was found to be true regardless of the
level of ASR expansion or the reinforcement ratios.
3. The Saouma and Perotti model evaluated higher
ASR-induced strains compared to the Charlwood model.
For the biaxially reinforced panels, the strains calculated Fig. 5—Geometric details and reinforcement layout of shear
with the Saouma and Perotti model were closer to the exper- wall specimens. (Note: 1 mm = 0.04 in.)
imental measurements, while for the triaxially reinforced
panels AF7 and AF8, the Charlwood model resulted in better the top slab, was 750 mm (29.5 in.). The web was 100 mm
predictions. (4 in.) thick and 1300 mm (51.1 in.) wide. The boundary
4. The proposed anisotropic model for the mechanical elements had a thickness of 200 mm (7.8 in.) and a width of
properties of concrete resulted in the most accurate predic- 120 mm (4.7 in.). In the web region, the horizontal reinforce-
tions of the overall response of the panels. ment ratio was 0.80%, and the vertical reinforcement ratio
5. The isotropic model calculated more severe degradation was 0.77%. The flanges had 2.10% reinforcement ratio in
parameters for the concrete compressive strength, tensile the vertical direction, 0.67% in the horizontal direction, and
strength, and modulus of elasticity compared to the aniso- 0.44% in the out-of-plane direction.
tropic model in the first and second principal directions. This Two specimens were cast with nonreactive concrete—
resulted in more conservative predictions. REG A and REG B walls; three specimens—ASR A1, ASR
6. The isotropic model yielded more conservative results B1, and ASR B2 were cast with reactive Spratt aggregate.
in comparison to the model employing cylinder concrete The concrete compressive strength fcp and modulus of elas-
properties as well. ticity Ec, as determined from standard 100 mm (4 in.) Φ
7. For all panels, neglecting the ASR expansion resulted cylinders on the test day, are shown in Table 3. For the reac-
in significant underestimations of the cracking strength and tive specimens, the free expansions (εASR), determined from
ultimate shear stress. expansion prisms, were used as input for the ASR analyses.
Sensitivity studies were carried out to identify the mecha-
SHEAR WALLS nisms that play a significant role in the computed responses
To gauge the reliability of the model for the analysis of of these walls. The FE models had different levels of
elements with common structural applications, the shear complexity to address the parameters investigated. A predic-
walls tested by Habibi et al.7 were examined. All shear walls tive approach was adopted, underlying the steps commonly
were similar in terms of reinforcement details, geometric required for a blind prediction exercise or for structural
dimensions, and configuration, as shown in Fig. 5. The spec- appraisal. The sensitivity studies were classified into three
imens had a barbell-shaped cross section and were built inte- categories: modeling details, behavioral mechanisms,
grally with rigid top and bottom beams. The height of the and material response. In-depth analyses of the sensitivity
shear walls, from the top of the bottom slab to the soffit of studies are reported elsewhere.33

Table 3—Shear wall specimens: concrete properties and summary of results
δu.Calc, δu.Exp,
ID fcp, MPa Ec, MPa εASR, × 10–3 Pu.Calc, kN Pu.Exp, kN mm mm Pu.Calc/Pu.Exp δu.Calc/δu.Exp
REG A 79.0 47,150 0.33 *
1172 1180 7.00 6.10 0.99 1.15
REG B 80.1 46,650 0.33* 1178 1187 7.06 6.30 0.99 1.12
ASR A1 63.7 35,750 1.90 1180 1355 4.50 6.20 0.87 0.73
ASR B1 67.1 32,600 2.15 1205 1240 4.88 4.90 0.97 1.00
ASR B2 63.0 28,100 2.23 1187 1243 4.60 2.60 0.95 1.77
Mean — 0.96 1.15
COV, % — 4.72 29.8

*
Expansion primarily attributed to swelling due to water absorption.
Note: 1 kN = 0.225 kip; 1 mm = 0.04 in.
The following set of factors were found to not significantly

affect the calculated behavior: smeared versus discrete repre-
sentation of the reinforcement, three-dimensional effects,
bond strength, reinforcement buckling, cover spalling and
element erosion, ASR expansion models, nor hysteretic
behavior of the concrete. Nevertheless, these aspects may
have a crucial effect for specimens with different geometries
or failure modes.
Three factors however, were identified as having a notable
effect on the computed response: representation of boundary
conditions, strength enhancement due to confinement, and
concrete compression response. In this respect, the following
conclusions can be made:
1. The magnitude and direction of the ASR-induced
strains depend on internal and external restraints, as well as
on long-term loading conditions, all of which must be appro-
priately considered for the ASR analysis.
2. Confinement conditions exert a significant influence on
many aspects of behavior, such as cracking, ultimate load
capacity, and post-peak ductility. Modeling of confinement-
related mechanisms has to be made in a realistic fashion.
3. The uniaxial concrete compression model impacts
the computed ductility appreciably. It is not sufficient to
accurately capture the concrete compressive strength and
modulus of elasticity. The strain at peak stress and post-peak
stiffness need to be properly represented as well.
Following the sensitivity study, for the final iteration of
analyses, it was decided to adopt an FE model with a 20
mm (0.79 in.) mesh size, illustrated in Fig. 6(a). The wall-
floor connection (that is, the anchor bolt assembly) was
modeled by defining a layer of bearing material elements
beneath the bottom beam that provided stiffness in compres-
sion only, thus allowing for potential uplift. The anchor bolt
was represented by three truss bars with equivalent cross-
sectional area. In addition, as shown in Fig. 6(b), the Fig. 6—Finite element model for shear wall specimens.
anchoring steel plate provided at the top of the bottom beam
was also modeled using two layers of steel elements and one expansion was used together with the anisotropic model for
layer of bearing material to allow free expansion of the wall the mechanical properties of concrete. For the shear wall
specimen due to ASR and prevent additional confinement. specimens with experimentally measured unsymmetrical
The reinforcement was represented as smeared, and responses, the envelope containing the maximum measured
the Hoshikuma model was used for concrete in compres- load was selected.
sion; all other behavioral parameters were the default Comparisons between the envelope of the experimental
VecTor2 options, as given in Appendix D, Table D.2. For results and the monotonically computed responses are
the analyses of the reactive walls, the Charlwood model for shown in Fig. 7 for nonreactive wall REG A and for the

Fig. 8—Finite element model for beam specimens—deep
beam test.
measured values. The nonreactive walls developed a failure
mode governed by diagonal shear with sliding between the
wall and the bottom beam, and the ASR-affected walls failed
due to diagonal shear. The numerical analyses captured the
appropriate failure modes for all wall specimens.
SHEAR-CRITICAL BEAMS
The validation analyses examining the proposed model for
the mechanical properties of ASR-affected concrete included
large shear-critical beam specimens tested by Deschenes
et al.17 The reactive specimens were cast with Jobe-Newman
fine aggregate and two nonreactive specimens were used as
a basis of comparison for long-term structural performance.
The specimens were structurally identical, with a width of
533 mm (21 in.) and a height of 1067 mm (42 in.). Two inde-
pendent shear tests, a deep beam shear test and a sectional
shear test, were performed on each specimen; one at each
end. The shear span-depth ratio was 1.85 for the deep beam
tests and 3.0 for the sectional shear tests. The longitudinal
reinforcement ratio of 3.1% was chosen such that a shear
failure would be obtained. The minimum amount of trans-
Fig. 7—Numerical versus experimental responses for selected
verse reinforcement was provided to ensure that the spec-
shear walls. (Note: 1 kN = 0.225 kip; 1 mm = 0.04 in.)
imens represented the least conservative field scenario:
reactive ASR A1 wall. For both specimens, the initial stiff- 0.31% for the deep beam tests and 0.15% for the sectional
ness, peak load, ductility, and failure mode were captured shear tests. The specimens were conditioned outside, with a
well by the FE analysis. Shown in Table 3 is the summary sustained load applied to simulate long-term service loading.
of results for all wall specimens in terms of peak force The FE model constructed to represent these specimens,
Pu and ultimate displacement δu ratios for calculated-to- showing the support conditions and the load application
experimental values. Note that the calculated and experi- for the deep beam test scenario, is illustrated in Fig. 8. The
mental ultimate displacements were reported as the displace- longitudinal reinforcement was represented using discrete
ments recorded before a reduction higher than 15% in the truss bar elements, whereas the stirrups and the crack control
horizontal load was measured. The peak loads of all wall reinforcement were smeared within the concrete elements.
specimens were matched well by the analytical results, with The mesh size used was 50 x 50 mm (2 x 2 in.). A monotoni-
the general tendency for the predictions being on the conser- cally increasing displacement-controlled load was applied
vative side. The ultimate displacements were reasonably well in increments of 0.25 mm (9.8 × 10–3 in.) until failure. For
captured for the nonreactive walls and for the reactive wall the reactive specimens, the load applied during the condi-
ASR B2; however, for walls ASR A1 and ASR B2, greater tioning phase was introduced as nodal loading active during
dissimilarity was noted between the calculated and the the ASR analysis only. Additionally, the dead load of the
specimens was simulated as gravity loading during both

Table 4—Shear-critical beams: concrete properties and summary of results
Concrete Vu.Calc, kN Vu.Calc/Vu.Exp
εASR,
Specimen fcp, MPa × 10–3 Vu.Exp, kN Cylinder Anisotropic Cylinder Anisotropic
nR1 DB 50.3 — 2500 2105 0.84
R1 DB 31.7 0.90 2309 2202 2155 0.95 0.93
R2 DB 27.0 4.40 2440 2692 2590 1.10 1.06
nR1 SS 49.6 — 1230 1440 1.17
R1 SS 31.0 1.70 1496 1627 1530 1.09 1.02
R2 SS 29.0 6.30 1570 1409 1644 0.90 1.05
Note: 1 kN = 0.225 kip.
ASR analysis and shear testing. Perfect bond was assumed and R2. The simulated crack patterns match sufficiently well
between concrete and reinforcement. the observed crack patterns, capturing the reduced crack
The reinforcement properties specified in the analysis distribution observed for the reactive specimens.
were the yield strength, the modulus of elasticity, and the
ultimate strength, as summarized in Appendix E. The DISCUSSION AND FUTURE WORK
concrete mechanical property specified was the compres- The proposed model provided reasonably accurate results
sive strength. Two analyses were performed for the reactive over the range of conditions examined. However, there are
specimens depending on the model for the mechanical prop- aspects in need of further study.
erties of concrete. In one case, the compressive strength as The material-level program that served as a basis for the
determined at the test date, from standard 100 mm (4 in.) Φ development of the model investigated a limited set of stress
cylinder tests, was used in the analysis. The second analysis conditions and expansion levels. Additional tests on concrete
employed the proposed anisotropic model for the mechanical subjected to long-term multi-axial stresses and undergoing
properties of concrete in conjunction with the compressive ASR expansion will potentially lead to a more refined model
strength of sound concrete at test date. Table 4 summarizes for the mechanical properties of ASR-affected concrete.
the concrete compressive strengths used in the analyses, fcp, Additionally, mechanical tests on concrete specimens that
and the ASR expansion strains considered for the ASR analysis have experienced ASR under environmental field conditions
phase εASR. For the anisotropic analyses of the reactive speci- would be a valuable addition to the current database. Central
mens, the compressive strengths measured from the nonre- to this suggested investigation is the ability to estimate the
active specimens nR1 DB and nR1 SS were used. long-term stress condition.
Shown in Table 4 is a summary of the results presented in Data from the literature suggest that ASR-affected
terms of shear strength for each test. Note that the ultimate concrete exhibits a continuous increase in the Poisson’s ratio
shear strength was reported to be the shear force acting at the with the increase of normal stress. This can potentially have
midspan of the test region. Both analyses matched well the a significant influence on the level of confinement induced.
experimentally measured shear strengths. The methodology proposed in this study does not address
The differences between the results obtained using the this phenomenon. Future work on this topic could provide
anisotropic model versus the cylinder properties were negli- further valuable insight on the behavior of ASR-affected
gible for these specimens. One reason behind this was the structures.
reinforcement configuration of the beam specimens. The
transverse reinforcement provided confinement of the core CONCLUSIONS
of the beams, and therefore had a beneficial effect on the The proposed model for the mechanical properties of alkali-
overall strength of the specimens. Experimental observa- silica reaction (ASR)-affected concrete is shown to be a
tions noted that ASR-induced cracks did not penetrate within viable approach for modeling reactive shear-critical rein-
the structural core of the specimens. It is of interest for future forced concrete elements. There is unequivocal evidence
work to analyze the response of ASR-affected beams or slab that the concrete mechanical properties are differently
strips with no transverse reinforcement. affected by ASR-induced damage. Moreover, the deterio-
The reactive specimens were found to have a markedly ration of mechanical properties is direction-dependent. The
different progression of cracking compared to the nonreac- proposed model, compatible with smeared crack formula-
tive specimens. The reactive specimens developed diagonal tions, considers this anisotropy through empirically deter-
cracks at higher applied shear load and the density of cracks mined modification factors applied to the compressive
was considerably reduced in comparison to the nonreactive strength, modulus of elasticity, and tensile strength of unaf-
specimens. These test observations are in good agreement fected concrete. The following conclusions are derived from
with the observations on the behavior of the panel speci- this work:
mens tested as part of this work. Shown in Fig. 9 and 10 1. The proposed anisotropic model yields results that
are the experimental and analytical crack patterns at failure closely match experimental observations when used with
following the deep beam tests performed on specimens nR1 the Charlwood model for ASR expansion. The validation

Fig. 10—Reactive R2 DB crack diagrams at failure.
Fig. 9—Nonreactive nR1 DB crack diagrams at failure.
AUTHOR BIOS
investigation performed on reinforced concrete panels illus- Anca C. Ferche is an Assistant Professor in the Department of Civil,
Architectural and Environmental Engineering at The University of Texas
trated that the isotropic version of the model produced more at Austin, Austin, TX. She received her PhD from the University of Toronto,
conservative results compared to the anisotropic one. Toronto, ON, Canada, in 2020. Her research interests include performance
2. The modeling of ASR-affected shear walls produces assessment and analysis of reinforced concrete structures, concrete deteri-
oration mechanisms, and rehabilitation of structures.
results that are consistent with the findings obtained from
panel specimens. In the shear wall specimens examined, Frank J. Vecchio, FACI, is a Professor in the Department of Civil and
however, the influences were somewhat more muted due Mineral Engineering at the University of Toronto. He is a past member of
Joint ACI-ASCE Committees 441, Reinforced Concrete Columns, and 447,
to other prevailing mechanisms. The sensitivity study Finite Element Analysis of Reinforced Concrete Structures. He has received
performed for the shear walls revealed that apart from the ACI Structural Research Award (1998), Structural Engineering Award
the model for the mechanical properties of ASR-affected (1999), Wason Medal for Most Meritorious Paper (2011), Joe W. Kelly
Award (2016), and Arthur J. Boase Award (2020). His research interests
concrete, the following mechanisms play an important role include advanced constitutive modeling and analysis of reinforced concrete,
in the computed response: ASR expansion effects, modeling assessment and rehabilitation of structures, and response to extreme loads.
of confinement-related mechanisms, and representation of
the boundary conditions. ACKNOWLEDGMENTS
3. In the analyses of shear-critical beams containing at The authors would like to acknowledge NSERC for funding support
provided to this project. The authors would also like to acknowledge the
least the minimum shear reinforcement, a significant differ- Ontario Ministry of Transportation and The University of Texas at Austin
ence does not exist between the results obtained employing for providing the reactive aggregate.
the anisotropic model for the mechanical properties or the
isotropic model using the mechanical properties measured REFERENCES
on standard cylinder tests. For the beam specimens exam- 1. Organisation for Economic Co-operation and Development, Global
Material Resources Outlook to 2060: Economic Drivers and Environmental
ined, this was primarily due to the presence of transverse Consequences, OECD Publishing, Paris, France, 2019.
reinforcement that confined the structural core of the beams, 2. Ulm, F.-J.; Coussy, O.; Kefei, L.; and Larive, C., “Thermo-
and therefore reduced the ASR-induced deterioration. Chemo-Mechanics of ASR Expansion in Concrete Structures,” Journal of
Engineering Mechanics, V. 126, No. 3, 2000, pp. 233-242. doi: 10.1061/
4. Analyses of shear walls and shear-critical beams reveal (ASCE)0733-9399(2000)126:3(233)
no improvement in the accuracy of the computed results 3. Grimal, E.; Sellier, A.; Le Pape, Y.; and Bourdarot, E., “Creep,
when using the more computationally expensive Saouma Shrinkage, and Anisotropic Damage in Alkali-Aggregate Reaction Swelling
Mechanism – Part I: A Constitutive Model,” ACI Materials Journal, V. 105,
and Perotti model compared to the much simpler Charlwood No. 3, May-June 2008, pp. 227-235.
model. 4. Fairbairn, E. M. R.; Ribeiro, F. L. B.; Lopes, L. E.; Toledo-Filho, R. D.;
and Silvoso, M. M., “Modelling the Structural Behaviour of a Dam Affected

by Alkali–Silica Reaction,” Communications in Numerical Methods in 19. Ferche, A. C.; Panesar, D. K.; Sheikh, S. A.; and Vecchio, F. J.,
Engineering, V. 22, No. 1, 2006, pp. 1-12. doi: 10.1002/cnm.788 “Toward Macro-Modeling of Alkali-Silica Reaction-Affected Structures,”
5. Farage, M. C. R.; Alves, J. L. D.; and Fairbairn, E. M. R., “Macro- ACI Structural Journal, V. 114, No. 5, Sept.-Oct. 2017, pp. 1121-1129.
scopic Model of Concrete Subjected to Alkali–Aggregate Reaction,” 20. Vecchio, F. J., “Disturbed Stress Field Model for Reinforced Concrete:
Cement and Concrete Research, V. 34, No. 3, 2004, pp. 495-505. doi: Formulation,” Journal of Structural Engineering, ASCE, V. 126, No. 9,
10.1016/j.cemconres.2003.09.001 2000, pp. 1070-1077. doi: 10.1061/(ASCE)0733-9445(2000)126:9(1070)
6. Esposito, R., and Hendriks, M. A. N., “Literature Review of Model- 21. Charlwood, R. G.; Solymar, S. V.; and Curtis, D. D., “A Review of
ling Approaches for ASR in Concrete: A New Perspective,” European Alkali Aggregate Reactions in Hydroelectric Plants and Dams,” Proceed-
Journal of Environmental and Civil Engineering, V. 23, No. 11, 2019, ings of the International Conference of Alkali-Aggregate Reactions in
pp. 1311-1331. doi: 10.1080/19648189.2017.1347068 Hydroelectric Plants and Dams, Fredericton, NB, Canada, 1992.
7. Habibi, F.; Sheikh, S. A.; Vecchio, F. J.; and Panesar, D. K., “Effects 22. Hobbs, D. W., Alkali-Silica Reaction in Concrete, Thomas Telford
of Alkali-Silica Reaction on Concrete Squat Shear Walls,” ACI Structural Ltd., London, UK, 1988.
Journal, V. 115, No. 5, Sept. 2018, pp. 1329-1339. doi: 10.14359/51702238 23. Jensen, A. D.; Chatterji, S.; Christensen, P.; and Thaulow, N., “Studies
8. NEA/CSNI, “Report on the Phase 2 of the Assessment of Struc- of Alkali-Silica Reaction — Part II Effect of Air-Entrainment on Expan-
tures Subjected to Concrete Pathologies (ASCET),” NEA/CSNI/R(2018)4, sion,” Cement and Concrete Research, V. 14, No. 3, 1984, pp. 311-314. doi:
Nuclear Energy Agency, Boulogne-Billancourt, France, 2018. 10.1016/0008-8846(84)90046-2
9. NEA/CSNI, “Report on the Phase 3 of the Assessment of Structures 24. Collins, R. J., and Bareham, P. D., “Alkali-Silica Reac-
Subjected to Concrete Pathologies (ASCET),” NEA/CSNI/R(2019)11, tion: Suppression of Expansion Using Porous Aggregate,” Cement
Nuclear Energy Agency, Boulogne-Billancourt, France, 2019. and Concrete Research, V. 17, No. 1, 1987, pp. 89-96. doi:
10. Wong, P. S.; Vecchio, F. J.; and Trommels, H., “VecTor2 and Form- 10.1016/0008-8846(87)90063-9
Works User’s Manual,” Technical Report, Department of Civil Engineering, 25. Beaudoin, J. J.; Feldman, R. F.; and Tumidajski, P. J., “Pore Struc-
University of Toronto, Toronto, ON, Canada, 2013, 318 pp. ture of Hardened Portland Cement Pastes and Its Influence on Properties,”
11. Institution of Structural Engineers (ISE), “Structural Effects of Alkali- Advanced Cement Based Materials, V. 1, No. 5, 1994, pp. 224-236. doi:
Silica Reaction,” SETO, London, UK, 1992. 10.1016/1065-7355(94)90028-0
12. Barbosa, R. A.; Hansen, S. G.; Hansen, K. K.; Hoang, L. C.; and 26. Bentz, E., “Augustus: Post Processor for VecTor2 (Version 5.8.3),”
Grelk, B., “Influence of Alkali-Silica Reaction and Crack Orientation on 2013.
the Uniaxial Compressive Strength on Concrete Cores from Slab Bridges,” 27. Saouma, V., and Perotti, L., “Constitutive Model for Alkali-
Construction and Building Materials, V. 176, 2018, pp. 440-451. doi: Aggregate Reactions,” ACI Materials Journal, V. 103, No. 3, May-June
10.1016/j.conbuildmat.2018.03.096 2006, pp. 194-202.
13. Ahmed, T.; Burley, E.; Rigden, S.; and Abu-Tair, A. I., “The Effect 28. Barbosa, R. A.; Hansen, S. G.; Hoang, L. C.; and Hansen, K. K.,
of Alkali Reactivity on the Mechanical Properties of Concrete,” Construc- “Residual Shear Strength of a Severely ASR-Damaged Flat Slab
tion and Building Materials, V. 17, No. 2, 2003, pp. 123-144. doi: 10.1016/ Bridge,” Engineering Structures, V. 161, 2018, pp. 82-95. doi: 10.1016/j.
S0950-0618(02)00009-0 engstruct.2018.01.056
14. Ferche, A. C., and Vecchio, F. J., “Mechanical Properties of Alkali- 29. den Uijl, J. A., and Kaptijn, N., “Structural Consequences of ASR:
Silica Reaction-Affected Concrete,” ACI Materials Journal, V. 119, No. 1, an Example on Shear Capacity,” HERON, V. 47, No. 2, 2002, pp. 125-139.
Jan. 2022, pp. 251-262. doi: 10.14359/51734198 30. Schmidt, J. W.; Hansen, S. G.; Barbosa, R. A.; and Henriksen, A.,
15. VTAG, VecTor Analysis Group, Nonlinear finite element “Novel Shear Capacity Testing of ASR Damaged Full Scale Concrete
analysis software for reinforced concrete structures, 2019, Bridge,” Engineering Structures, V. 79, 2014, pp. 365-374. doi: 10.1016/j.
http://vectoranalysisgroup.com/. (last accessed Jan. 27, 2022) engstruct.2014.08.027
16. Ferche, A. C., and Vecchio, F. J., “Behavior of Alkali-Silica Reaction- 31. Luo, J. W., “Behaviour and Analysis of Steel Fibre-Reinforced
Affected Reinforced Concrete Elements Subjected to Shear,” ACI Struc- Concrete under Reversed Cyclic Loading,” MASc thesis, University of
tural Journal, V. 118, No. 4, July 2021, pp. 163-174. Toronto, Toronto, ON, Canada, 2014, 315 pp.
17. Deschenes, D. J.; Bayrak, O.; and Folliard, K. J., “ASR/DEF- 32. Carnovale, D. J., “Behaviour and Analysis of Steel and Macro-
Damaged Bent Caps: Shear Tests and Field Implications,” Technical Report Synthetic Fibre Reinforced Concrete Subjected to Reversed Cyclic Loading:
for the Texas Department of Transportation, 2009, 271 pp. A Pilot Investigation,” MASc thesis, University of Toronto, Toronto, ON,
18. Jurcut, A.-C., “Modelling of Alkali-Aggregate Reaction Effects in Canada, 2013, 323 pp.
Reinforced Concrete Structures,” MASc thesis, University of Toronto, 33. Ferche, A.-C., “Behaviour and Modelling of ASR-Affected
Toronto, ON, Canada, 2015, 136 pp. Shear-Critical Reinforced Concrete Structures,” PhD dissertation, Depart-
ment of Civil and Mineral Engineering, University of Toronto, Toronto,
ON, Canada, 395 pp.

Title No. 119-S31
Nonlinear Modeling Parameters of Reinforced Concrete

Coupling Beams
by Tae-Sung Eom, Seung-Jae Lee, Su-Min Kang, and Hong-Gun Park
The present study investigated nonlinear modeling parameters Ishikawa and Kimura (1996), and Shimazaki (2004) studied
for reinforced concrete coupling beams. Based on a review of the the cyclic behavior of short and long coupling beams. The
existing test results of 79 coupling beams, modeling parameters effects of concrete type (that is, normal-weight and light-
of varying aspect ratios and transverse reinforcement ratios were weight concrete), main reinforcement layouts (that is,
proposed. First, the effective stiffness and yield deformation were
conventional reinforcement, diagonal reinforcement, and
formulated as functions of aspect ratio. Second, the deformation
both), and transverse reinforcement details on the strength
capacity varying with reinforcement details was examined, and its
design equation was defined as a function of transverse reinforce- and ductility were investigated. Xiao et al. (1999) investi-
ment ratio. Third, the energy dissipation ratio to define hysteresis gated the behavior of high-strength concrete coupling beams
loop under cyclic loading was suggested as a function of aspect under cyclic loading. The tests showed that longitudinal
ratio and reinforcement details. For verification, the modeling reinforcements distributed across the beam section height
results of load-deformation relations were compared with the significantly improved the hysteretic energy dissipation and
existing test results. The predicted load-deformation relations ductility by reducing an accumulation of longitudinal elon-
showed good agreement with the test results. Finally, practical gation deformations. Accumulated longitudinal elongation
applications to computer modeling are discussed. and its effects on cyclic behavior were investigated by Lee
and Watanabe (2003) and Eom and Park (2013), respectively.
Keywords: coupled wall; coupling beam; effective stiffness; modeling
parameters; nonlinear modeling. Naish et al. (2013a, 2013b) studied the cyclic behavior
and modeling methods of conventionally reinforced and
INTRODUCTION diagonally reinforced coupling beams. Based on the results,
Reinforced concrete coupling beams used in tall buildings they proposed modeling parameters, including the effects of
develop large chord rotation under seismic loads owing to the aspect ratio l/h. Eom et al. (2009) studied a nonlinear
differences in vertical displacements Δ of the coupled walls truss model and hysteretic energy dissipation of diagonally
(refer to Fig. 1(a)) (Paulay 2002; Moehle et al. 2012). Such reinforced short coupling beams. Based on the results, they
coupling beams usually have low aspect ratios (length-to- proposed energy-based cyclic models for coupling beams,
height ratios, l/h) between 1 and 4, and their behavior can with emphasis on an accurate estimation of the hysteretic
be dominated either by shear or by flexure depending on l/h energy dissipation. Lim et al. (2016a, 2016b) studied seismic
(American Society of Civil Engineers 2017; Paulay 2002): behavior and strut-and-tie models of conventionally rein-
for coupling beams with l/h ≤ 2.0, behavior is controlled by forced and diagonally reinforced coupling beams with short
shear (ACI Committee 318 2019); for coupling beams with and intermediate aspect ratios (that is, 1 ≤ l/h ≤ 4). Park et al.
l/h ≥ 4.0, on the other hand, behavior is controlled by flexure; (2020) studied the seismic performance of coupling beams
and for coupling beams with 2 < l/h < 4, the behavior mode with different transverse reinforcement details. Ameen et al.
changes from “shear” to “flexure” as l/h increases. Further, (2020) studied the effects of high-strength steel bars on
the coupling beam behavior can be affected by reinforce- coupling beam behavior.
ment details, such as diagonal and transverse reinforcements In ASCE/SEI 41-17, modeling parameters for coupling
(Naish et al. 2013b; Galano and Vignoli 2000). Thus, when beams, such as effective stiffness (that is, flexural rigidity
determining modeling parameters for coupling beams, the 0.3EcIg and shear rigidity 0.4EcAw), deformation param-
effects of these design variables need to be addressed. eters (θd = 0.008 to 0.30 rad and θe = 0.014 to 0.05 rad),
There have been many studies on the behavior of coupling and residual strength ratio (c = 0.2 to 0.8), are specified
beams. Barney et al. (1978) investigated the effects of l/h, in Tables 10-5, 10-19, and 10-20 (refer to Fig. 1(b)). For
reinforcement details, and confined core size on the cyclic modeling in accordance with ASCE/SEI 41-17, its behavior
responses of coupling beams. The test results showed that mode (“controlled by flexure” or “controlled by shear”)
diagonal reinforcements significantly improved the perfor- needs to be defined; then, the modeling parameters θd, θe,
mance of coupling beams, although the level of improve- and c are determined, considering reinforcement details and
ments differed according to l/h; and larger concrete core shear stress level. Although ASCE/SEI 41 has updated these
size also improved ductility. Tegos and Penelis (1988) and ACI Structural Journal, V. 119, No. 3, March 2022.
Galano and Vignoli (2000) studied the behavior of short MS No. S-2021-040.R1, doi: 10.14359/51734139, received August 1, 2021, and
coupling beams with various reinforcement details under Institute. All rights reserved, including the making of copies unless permission is
monotonic and cyclic loadings. Kanakubo et al. (1996), closure, if any, will be published ten months from this journal’s date if the discussion

Fig. 1—Behavior of reinforced concrete coupling beams.
modeling parameters, some improvements are required RESEARCH SIGNIFICANCE
as follows. This study investigates modeling parameters for rein-
1. ASCE/SEI 41-17 does not specify any criteria for the forced concrete coupling beams for better accuracy in the
determination of the behavior mode of coupling beams. In nonlinear analysis of coupled wall structures. On the basis
some cases, it is unclear whether the behavior is controlled of existing test results, the effective stiffness varying with
by flexure or by shear. In fact, the behavior mode varies with aspect ratio and governing behavior mode are defined, and
aspect ratio and reinforcement details. the deformation capacity varying with reinforcement details
2. The effective stiffness of coupling beams varies is examined. In particular, a modeling parameter to define
according to l/h. Thus, 0.3EcIg and 0.4EcAw specified in hysteretic energy dissipation and its application to computer
ASCE/SEI 41-17 may not be adequate, particularly for short modeling are proposed. The accuracy of the proposed model
coupling beams. Naish et al. (2013b) and Son Vu et al. (2014) is verified through comparisons between modeling and
studied effective stiffness equations for coupling beams. On test results.
the basis of a review of the existing studies, Tall Buildings
Initiative (TBI) guidelines (Pacific Earthquake Engineering EFFECTIVE STIFFNESS AND YIELD DEFORMATION
Center 2017) define effective stiffness as 0.07(l/h)EcIg (≤ Existing test results
0.3EcIg). The force-deformation relation of a coupling beam can
3. The deformation parameters θd and θe in ASCE/ be represented with end moment M (or shear force V) and
SEI 41-17 are mainly functions of shear stress level (V/ chord rotation θ (= Δ/l). Figure 1(b) shows the idealized
[bh√fc′]) and transverse reinforcement (that is, “conforming” M-θ relation. The initial behavior before flexural yielding
or “non-conforming”). However, in addition to the shear is approximated as a straight line with effective stiffness Ke
stress level and transverse reinforcement, the aspect ratio connecting the origin and the pre-peak point of 0.6Mu (Mu
l/h can also affect θd and θe. Naish et al. (2013b) suggested is maximum load). The effective stiffness Ke is affected by
values of θd and θe varying with l/h for diagonally reinforced concrete cracking and reinforcement anchorage slip (Naish
coupling beams. et al. 2013b). Once Ke is determined, the yield rotation can
4. ASCE/SEI 41-17 and TBI guidelines do not specify be calculated as θy = Mu/Ke.
modeling criteria for hysteretic energy dissipation. The Table 1 shows the main test variables of 79 existing speci-
hysteretic energy dissipation of coupling beams varies mens collected to quantify the effective stiffness Ke and yield
significantly according to design variables such as reinforce- rotation θy (Barney et al. 1978; Kanakubo et al. 1996; Ishi-
ment layout and aspect ratio (Eom et al. 2009). kawa and Kimura 1996; Galano and Vignoli 2000; Shimazaki
In this study, on the basis of a review of existing test 2004; Lee and Watanabe 2003; Xiao et al. 1999; Naish et al.
results of 79 specimens, nonlinear modeling parameters 2013a; Lim et al. 2016a, 2016b; Park et al. 2020; Ameen et
for coupling beams were investigated as follows. First, the al. 2020). The ranges of test variables are as follows: 18.1
effective stiffness and yield deformation varying with l/h MPa (2.62 ksi) ≤ fc' ≤ 79.5 MPa (11.5 ksi); 0.16% ≤ ρl ≤
were theoretically formulated. Second, through a review of 4.21%; 0.52% ≤ ρd ≤ 4.24%; 1.00 ≤ l/h ≤ 4.94; conventional
the test results, the deformation capacity varying with rein- (longitudinal and transverse) and diagonal reinforcement
forcement details (reinforcement layout and transverse rein- layouts. ρl and ρd are longitudinal and diagonal reinforce-
forcement ratio) was investigated. Third, a modeling param- ment ratios, respectively. The values of ρl and ρd in Table 1
eter was proposed to define the hysteretic energy dissipation. were calculated by dividing the areas of total longitudinal
For verification, the modeling results of load-deformation bars or total diagonal bars by bwh (bw is web width).
relations were compared with the existing test results. For coupling beams with relatively short spans, the overall
behavior is affected by shear as well as flexure. Thus, the
stiffness and deformation measured from tests contain
contributions of shear and flexure. However, it is technically

Table 1—Summary of existing coupling beam specimens
Researchers and specimens fc', MPa ρl, % ρd, % l/h ρt, % Reinforcement layout* Energy dissipation classification†
NPM13 41.7 1.81 — 2.40 0.33 Conventional Low (pinched)
LPM 37.9 2.84 — 2.80 0.53 Conventional Low (pinched)
Kanakubo et al. NX7L 41.7 0.95 1.89 2.80 0.33 Diagonal (X) High
(1996) NHPL13 79.5 1.81 — 2.40 0.22 Conventional Low (pinched)
LPH 37.9 2.84 — 2.80 0.80 Conventional Low (pinched)
LX7L 37.9 0.95 1.89 2.80 0.33 Diagonal (X) High
S-No.1 54.0 0.99 1.98 2.50 0.21
S-No.2 51.0 0.99 1.98 2.50 0.32
S-No.3 54.0 0.99 1.98 2.50 0.21
S-No.4 51.0 0.99 1.98 2.50 0.32
Shimazaki (2004) Diagonal (X) High
S-No.5 51.0 0.99 1.98 2.50 0.64
S-No.6 64.0 0.99 1.98 2.50 0.64
S-No.7 48.0 0.99 2.85 2.50 0.64
S-No.8 32.0 0.99 1.98 2.50 0.64
IK-No.1 42.2 0.59 1.15 1.78 0.24
IK-No.2 45.9 1.17 1.72 1.78 0.24
IK-No.3 45.1 1.17 1.72 1.78 0.24
IK-No.4 48.2 1.17 1.72 1.78 0.24
Ishikawa and IK-No.5 25.4 1.17 1.72 1.78 0.47

Diagonal (X) High
Kimura (1996) IK-No.6 51.2 1.27 1.72 1.78 0.85
IK-No.7 51.7 1.27 1.72 1.78 0.24
IK-No.8 45.3 1.17 1.69 1.78 0.24
IK-No.9 45.8 1.17 1.69 1.78 0.24
IK-No.10 51.7 1.17 1.69 2.50 0.24
P01 48.9 1.14 — 1.50 0.84 Conventional NA (Monotonic)
P05 39.9 0.28 1.05 1.50 0.39 Diagonal (X) NA (Monotonic)
P10 46.8 0.28 1.05 1.50 0.31 Diagonal (X) NA (Monotonic)
Galano and P13 47.5 0.52 0.52 1.50 0.55 Diagonal (DX) NA (Monotonic)
Vignoli (2000) P02 44.5 1.14 — 1.50 0.84 Conventional Low (pinched)
P07 54.0 0.28 1.05 1.50 0.39 Diagonal (X) High
P12 41.6 0.28 1.05 1.50 0.31 Diagonal (X) High
P14 45.0 0.52 0.52 1.50 0.55 Diagonal (DX) High
C1 20.3 0.66 0.66 2.47 0.41 Diagonal (DX) Low (pinched)
C2 21.0 1.31 — 2.47 0.41 Conventional Low (pinched)
C3 20.5 0.66 1.31 2.47 0.41 Diagonal (DX) Low (pinched)
Barney et al. C4 24.1 0.66 1.31 2.47 0.41 Diagonal (DX) Low (pinched)
(1978) C5 21.7 1.31 — 2.47 0.41 Conventional Low (pinched)
C6 18.1 0.16 1.56 2.47 0.41 Diagonal (X) High
C7 25.6 1.31 — 4.94 0.41 Conventional Moderate
C8 23.9 0.16 1.56 4.94 0.41 Diagonal (X) High

Table 1 (cont.)—Summary of existing coupling beam specimens
BA2 29.3 0.27
BA3 29.3 0.20
BA4 29.3 0.80
BA5 29.3 0.53
Lee and
BA6 29.3 2.54 — 4.00 0.40 Conventional Moderate
Watanabe (2003)
BA7 43.4 0.20
BA10 43.4 0.60
BA11 43.4 0.40
BA12 43.4 0.30
HB4-6L-T100 2.11 — 4.00 0.70 Moderate
HB4-10L-T65 3.51 — 4.00 1.08 Moderate
HB4-12L-T65 4.21 — 4.00 1.08 Moderate
Xiao et al. (1999) 69.5 Conventional
HB3-6L-T100 2.11 — 3.00 0.70 Low (pinched)
HB3-10L-T50 3.51 — 3.00 1.40 Moderate
HB3-12L-T50 4.21 — 3.00 1.40 Moderate
CB24F 47.3 — 3.99 2.40 0.92 Diagonal (X) High
CB24D 47.3 — 3.99 2.40 1.10 Diagonal (X) High
CB24F-RC 50.4 — 3.99 2.40 0.92 Diagonal (X) High
Naish et al. CB24F-PT 50.0 — 3.99 2.40 0.92 Diagonal (X) High
(2013a) CB24F-1/2-PT 48.2 — 3.99 2.40 0.46 Diagonal (X) High
CB33F 47.3 — 3.33 3.33 0.92 Diagonal (X) High
CB33D 47.3 — 3.33 3.33 0.34 Diagonal (X) High
FB33 41.4 1.22 — 3.33 0.92 Conventional Moderate
CB10-1 34.5 0.57 3.24 1.00 1.52 Diagonal (X) High
Lim et al.
CB20-1 52.1 0.84 3.43 2.00 1.27 Diagonal (X) High
(2016b)
CB20-2 52.2 3.73 — 2.00 1.81 Conventional Low (pinched)
CB30-C 47.9 3.58 — 3.00 1.27 Conventional Low (pinched)
CB30-DA 39.7 0.34 4.24 3.00 0.63 Diagonal (X) High
Lim et al. CB30DB 38.4 0.48 4.24 3.00 1.27 Diagonal (X) High
(2016a)‡ CB30-H 58.0 2.21 2.12 3.00 1.27 Diagonal (X) Moderate
CB40-C 58.0 3.43 — 4.00 1.27 Conventional Moderate
CB40-H 59.0 2.82 0.77 4.00 1.27 Diagonal (X) Moderate
CCB 66.5 0.74 1.90 2.00 1.27 Moderate
Park et al.
CCBWC 66.5 0.74 1.90 2.00 0.63 Diagonal (X) Moderate
(2020)‡
SFRCWC 62.5 0.74 1.90 2.00 0.63 Moderate
CB1 41.0 — 4.00 1.89 0.73 High
Ameen et al. CB2 50.0 — 1.96 1.89 0.73 Moderate
Diagonal (X)
(2020) CB2D 43.0 — 1.96 1.89 0.73 Moderate
CB3D 43.0 — 2.94 1.89 0.73 Moderate
*
Reinforcement layouts “Conventional,” “Diagonal (X),” and “Diagonal (DX)” refer to Fig. 6.
†
Energy dissipations are qualitatively classified as “Low (pinched),” “Moderate,” and “High,” based on shapes of cyclic curves. Refer to Fig. 7(a).
‡
Specimens failed by shear before flexural yielding are excluded.
Note: 1 ksi = 6.90 MPa.
difficult to distinguish or measure the contributions of shear The values of αf and αs in Fig. 2 are calculated as follows.
and flexure accurately. For this reason, in current seismic αf in Fig. 2(a) is defined by dividing the measured Ke by the
evaluation and design guidelines, the overall behavior of a coupling beam flexural stiffness (6EcIg/l) without consider-
coupling beam is represented by using either a flexural model ation of shear stiffness; αf = Ke/[6EcIg/l]. On the other hand,
(refer to Fig. 2(a)) or a shear model (refer to Fig. 2(b)). In αs in Fig. 2(b) is defined by dividing the measured Ke by
each model, the overall stiffness and deformation represent the coupling beam shear stiffness (0.5GAwl) without consid-
the combined behavior of flexure and shear. ering flexural stiffness; αs = Ke/[0.5GAwl]. Ec is the elastic
Figure 2 shows the effective stiffness factors αf for flex- modulus of concrete, taken as 4700√fc′, and Ig is the second-
ural modeling and αs for shear modeling, varying with order moment of inertia of the uncracked concrete section.
aspect ratio l/h. Solid (+) and void (–) circles indicate the The shear modulus G is taken as 0.4Ec and Aw = bwh. The
test data under positive and negative loadings, respectively. values of αf in Fig. 2(a) show an increasing trend from 0 to

Fig. 2—Variations of effective stiffness factors αf and αs according to aspect ratio.
0.3 with increasing l/h, and the trend is limited to 0.3 as l/h
exceeds 4. The effective stiffness factor in ASCE/SEI 41-17,
αf = 0.3, does not catch such a trend. The effective stiffness
factor in TBI guidelines, αf = 0.07l/h (≤ 0.3, refer to the solid
line in Fig. 2(a)), agrees better with the test results, although
it slightly overestimates them. The values of αs in Fig. 2(b)
show a decreasing trend from 0.1 to 0.03 as l/h increases
from 1 to 5.
Figure 3 plots the measured yield rotations θy (= Mu/
Ke) of conventionally reinforced (crosses) and diagonally
reinforced (circles) coupling beams. The values of θy vary
between 0.0024 and 0.0175 rad. The correlation between θy
and l/h is not clear.
Effective stiffness
Figure 4(a) shows an idealized model for coupling beams.
Assuming a linear elastic behavior, the chord rotation of the Fig. 3—Yield rotation of coupling beams according to aspect
coupling beam can be simplified as the sum of flexural and ratio.
shear deformations as follows.
Ke f
f
Kf f K f (2)
M M M 1
f s or s K s
K e f K f s K s (1)

Ke s
where Ke is overall stiffness, including contributions of s
Ks K
flexure and shear; Kf is flexural stiffness (6EcIg/l, refer 1 s s
(3) f Kf
to Fig. 2(a)); and Ks is shear stiffness (0.5GAwl, refer to
Fig. 2(b)). In Eq. (1), βf and βs are factors accounting for By substituting βf = 0.3, βs = 0.04[l/h], and Kf/Ks = 2.5[h/l]2
a reduction in flexural and shear rigidities due to concrete into Eq. (2) and Eq. (3), αf and αs can be expressed as follows.
cracking, respectively. For a rectangular section of width bw
and height h, Ig and Aw are bwh3/12 and bwh, respectively,
and by taking G as 0.4Ec, Kf/Ks yields 2.5[h/l]2. Given that 0.3 0.3
f (4)
h h
3 3
flexural cracking occurs mostly at the beam ends regardless
1 18.8 1 20
of whether the beam span is slender or short, the reduction l l
factor for flexural rigidity can be approximate as βf = 0.3,
which is the same as that for ordinary reinforced concrete l l
0.04 0.04
beams. On the other hand, if l/h increases progressively, s h h
shear cracking reduces as the coupling beam behavior is 1 l 1 l
3 3
1 1 (5)
governed more by flexure than by shear. Thus, in this study, 18.8 h 20 h
the reduction factor for shear rigidity is approximated as
βs = 0.04[l/h], which is linearly proportional to l/h (refer to
As shown in Fig. 4(b) and (c), the predicted αf and αs
Fig. 2(b)).
plotted as the dashed lines trace the test results varying with
From Eq. (1), the effective stiffness factors αf and αs can
l/h. αf in Eq. (4) approaches 0.3 as l/h increases to 5; αs in
be defined as follows.

Fig. 4—Comparison of effective stiffness factors αf and αs with test values.
Eq. (5) reaches the peak at l/h = 2.2 and then decreases as
l/h increases.
Given that the αf in Eq. (4) and αs in Eq. (5) yield reason-
able estimates, the effective flexural and shear rigidities
of coupling beams for practical application to computer
modeling can be defined as follows.
EI e f Ec I g 0.3Ec I g (6)
l
GAe s GAw 0.04 GAw (7)
h
Yield deformation
The yield rotation θy of coupling beams can be calculated
using the effective stiffness Ke (= αfKf), as follows (refer to Fig. 5—Comparison of predicted yield rotations with test
Fig. 1(b) and Eq. (4)). values.
Mu Mu DEFORMATION CAPACITY
y In this section, the effects of aspect ratio (l/h), reinforce-
f Kf 0.3 6 Ec I g
ment layouts, transverse reinforcement ratio (ρt) on the
1 20 h / l l
3
deformation capacity (θd) in Fig. 1(b) were investigated
1 20 h / l M u l
(8)
3 as follows.
Figure 6 shows the values of θd varying with l/h and ρt.
1.8 Ec I g Figure 6(a) and (b) are the test results of conventionally rein-
where Mu is the measured maximum load of coupling beams forced and diagonally reinforced coupling beams, respec-
(that is, end moment). tively. θd is defined as the deformation at the post-peak point,
Figure 5 shows the values of θy,pred/θy,test varying with l/h, where the load is degraded to 80% of the maximum load
where θy,pred and θy,test are the calculated and measured yield (that is, 0.8Mu, refer to Fig. 1(b)). θd includes elastic defor-
rotations, respectively. The mean and covariance of θy,pred/ mation before yielding. In Fig. 6(a1) and (b1), correlations
θy,test are 0.97 and 0.26, respectively. The yield rotations of between θd and l/h are poor for both conventionally rein-
coupling beams are affected by reinforcement anchorage slip, forced and diagonally reinforced coupling beams. Particu-
flexural and shear cracking, and shear stress level, in addi- larly in Fig. 6(a1), θd tends to decrease as l/h increases. This
tion to l/h. Thus, Eq. (8) estimates the yield rotations with trend is contrary to expectations for the following reasons.
reasonable precision, although the scatter is relatively large. Generally, the flexural strength Mn of a coupling beam is
For design, the yield rotation of coupling beams can be proportional to ρbwh2. Thus, the shear force Vy (= 2Mn/l)
calculated by substituting the nominal flexural strength Mn corresponding to Mn increases with ρbwh2/l. This implies that
for Mu in Eq. (8). the shear stress v (= Vy/[bwh]) is inversely proportional to
l/h: v ∝ ρ/[l/h]. For this reason, the deformation parame-
ters d and e defined in ASCE/SEI 41-17 increase if the shear
stress level (that is, V/[bwh√fc′]) is low.

Fig. 6—Deformation capacity varying with aspect ratio and transverse reinforcement ratio.
Transverse reinforcements can affect the deformation d 10t 0.05 rad (9b)
capacity of coupling beams by improving shear capacity
and lateral confinement. Thus, the values of θd varying with In Fig. 6(b2), the test results denoted by the dashed line
transverse reinforcement ratio ρt are replotted in Fig. 6(a2) rectangles are coupling beams whose loading was stopped
and (b2). θd shows better correlations with ρt as follows. before failure (Shimazaki 2004) or transverse reinforcement
1. For conventionally reinforced coupling beams in details were inadequate (Park et al. 2020). Other than those
Fig. 6(a2), θd shows an increasing trend within approximately specimens, most of the test results are greater than the design
ρt ≤ 0.004. However, for ρt ≥ 0.004, θd is limited to approxi- values of θd defined by Eq. (9b) (that is, thick solid lines).
mately 0.04 to 0.05 rad. Conventionally reinforced coupling In Eq. (9a) and (9b), the upper limits of θd (0.035 rad for
beams mostly have large aspect ratios (that is, l/h3 2.5, refer conventionally reinforced coupling beams and 0.05 rad for
to Fig. 6(a1)), and their behavior is governed mainly by diagonally reinforced coupling beams) are set for conserva-
flexure. Thus, if ρt is greater than the value required for flex- tive design. The scatter in Fig. 6(a2) and (b2) implies that
ural yielding, the values of θd tend to remain constant. In this the deformation capacity of coupling beams can be affected
study, a design equation for θd is suggested as follows (refer by other design variables, such as transverse reinforcement
to thick solid lines in Fig. 6(a2)). details (“conforming” and “non-conforming” transverse
reinforcement in ASCE/SEI 41-7), the type of confining
d 10t 0.035 rad (9a) hoops, combined use of longitudinal and diagonal reinforce-
ments, and shear strength-to-demand ratio.
2. Diagonally reinforced coupling beams mostly have
short aspect ratios of l/h ≤ 2.5 (refer to Fig. 6(a)), and ENERGY DISSIPATION RATIO
their behavior is governed mainly by shear. Although not Cyclic force-deformation relations of coupling beams
clearly seen from the scattered data in Fig. 6(b2), transverse are significantly affected by design variables such as aspect
reinforcements can increase the deformation capacity by ratio, reinforcement layouts, and transverse reinforcements.
providing confinement to the concrete surrounding diag- Figure 7(a) shows three cyclic curves of coupling beams
onal reinforcements. Thus, the same design equation of θd classified qualitatively by the level of hysteretic energy
as that for conventionally reinforced coupling beams (that dissipation (ED): “high,” “moderate,” and “low (pinched)”
is, Eq. (9a)) is suggested for diagonally reinforced coupling (refer to notes in Table 1). Characteristics for cyclic force–
beams as follows (refer to thick solid lines in Fig. 6(b2)). deformation relations, such as unloading/reloading stiffness

Fig. 7—Energy-based cyclic model and energy dissipation ratio κ.
Fig. 8—Design variables to define energy dissipation ratio.

and pinching, are related to ED. As shown in Table 1, diago- of reinforced concrete members under cyclic loading is
nally reinforced coupling beams showing cyclic curves with contributed mainly by reinforcing steel bars, whereas the
limited or no pinching are classified as high or moderate contribution of concrete (that is, brittle material) to the energy
ED. On the other hand, conventionally reinforced coupling dissipation is almost negligible. Particularly for diagonally
beams showing cyclic curves with severe pinching are reinforced coupling beams where the aspect ratio is less than
mostly classified as low ED (refer to the pinching or shear 2.0, and the governing deformation mode is shear deforma-
slip in Fig. 7(a), Eom and Park 2013). tion, the diagonal bars crossing shear cracks dissipate signif-
Various hysteresis models, including the Clough and icant hysteretic energy by experiencing large plastic strain
modified Takeda models, have been proposed to describe reversals. Thus, the energy dissipation of coupling beams
the cyclic behavior of reinforced concrete members. They with diagonal reinforcement layout (X-type), EDX, can be
are available in computer programs for nonlinear analysis. calculated by summing energy dissipations of diagonal bars
Among the existing models, the energy-based hysteresis as follows (Fig. 8(a), Eom et al. 2009).
model adopted in Computers and Structures (2018) can be
conveniently used for coupling beams (refer to Fig. 7(b)). 2 yd
However, to use the energy-based hysteresis model, it is EDX 4 RB f yd Asd l sin P N (11)
cos sin
required to define ED quantitatively.
In this study, the energy dissipation ratio κ is used to define
ED (= κEep), where Eep is energy dissipation by the idealized where RB is the reduction factor that accounts for the Baus-
elastic-perfectly plastic behavior (refer to Fig. 7(b)). chinger effect, taken as 0.75 (Park and Eom 2006); fyd and
εyd are the yield stress and strain of diagonal bars; Asd is the
ED ED total cross-sectional area of diagonal bars in each direction;

Eep M P M N P N 2 y (10) and α is the inclination angle of diagonal bars relative to the
longitudinal axis.
The reduction factor RB is defined as the energy dissipa-
where MP and MN are the end moment strengths under position ratio by the actual cyclic behavior (including the Baus-
tive and negative loadings, respectively; θP and θN (≥ 0) are chinger effect) to the idealized one by the elastic-perfectly
the current maximum deformation histories in the positive plastic behavior.
and negative loadings, respectively; and θy is the yield rota- EDX in Eq. (11) is the hysteretic energy dissipation of the
tion (refer to Eq. (8)). ED and κ in Eq. (10) for coupling whole coupling beam. On the other hand, ED in Fig. 10(b)
beams can be estimated as follows. is the energy dissipation occurring at one end of the beam,
equivalent to 0.5EDX. Thus, by substituting ED = 0.5EDX into
Diagonally reinforced coupling beams Eq. (10), κ can be calculated as follows.
According to existing studies (Park and Eom 2006; Eom et
al. 2009; Eom and Park 2010a, 2013), the energy dissipation

Fig. 9—Modeling methods for coupling beams.
2 yd
P N
0.5 EDX 3 f yd Asd l sin cos sin

Eep 2MP M N P N 2 y
3 f yd Asd l sin

2 MP M N
(12)
According to Eom et al. (2009), the yield chord rotation

θy of diagonally reinforced coupling beams (X-type) can be
Fig. 10—Rigid plastic force-deformation relationship for
approximated as 0.5[εyd – εc]/[sin α∙cos α], where εc (< 0)
moment hinge model.
is the compressive strain of diagonal struts. By substituting
θy = 0.5[εyd – εc]/[sin α∙cos α] and eliminating the deforma- the hysteretic energy dissipation EDL is contributed mainly
tion terms, the simplified equation κ is derived from Eq. (12) by longitudinal bars at the plastic hinge region. According
with little loss of accuracy. to Eom and Park (2010a, 2013), EDL can be calculated as
fyd, Asd, l, and α are determined from material and follows (refer to Fig. 8(b)).
geometric properties. The flexural strengths of diagonally
reinforced coupling beam, MP and MN, should be calculated lp
on the basis of the equivalent stress block analysis at the EDL 2 RB f yl Asl hl P N 4 yl (13)
hl
critical section on the wall face (Fig. 8(a)); in this calcu-
lation, internal forces in the tensile zone are approximately where fyl and εyl are the yield stress and strain of longitudinal
taken as the tensile force of longitudinal bars (fylAsl) and the bars; Asl is the cross-sectional area of longitudinal bars at the
horizontal component of the tensile force of diagonal bars top or bottom of the beam section, whichever is smaller; hl is
(fyd Asd cos α). the distance between the centroids of top and bottom longi-
Note that κ in Eq. (12) is uniform, regardless of the values tudinal bars; lp is the plastic hinge length where longitudinal
of θP and –θN. If MP and MN (= Vnl/2) are calculated as bars undergo large plastic strain reversals, approximately
[2fyd Asd sin α]l/2 in accordance with ACI 318-19, Eq. (12) taken as 0.5h (Eom and Park 2010a).
yields a theoretical maximum value, κ = 0.75. However, If the cross-sectional areas of the top and bottom longi-
even for short coupling beams with diagonal reinforcement tudinal bars are different, the smaller of the two should be
only, κ decreases to approximately 0.5 as the concrete in the taken as Asl because only the reinforcing bars with the smaller
compression zone increases MP and MN by increasing the area contribute to the energy dissipation. If the longitudinal
moment arm (Fig. 8(a)). Thus, care should be taken when bars are distributed across the entire beam section, the sum
calculating MP and MN not to overestimate the hysteretic of cross-sectional areas of the outmost two layers of longi-
energy dissipation of diagonally reinforced coupling beams. tudinal bars is approximately taken as Asl, neglecting the
contribution of intermediate bars in the web.
Conventionally reinforced coupling beams EDL in Eq. (13) is the energy dissipation at one end of the
As shown in Table 1, conventionally reinforced coupling beam, which is the same as ED in Fig. 7(b). Thus, substi-
beams with l/h ≥ 3 can also dissipate considerable energy tuting ED = EDL into Eq. (10) yields κ as follows. Note, RB =
under inelastic cyclic loading. For instance, C7 (l/h = 4.93), 0.75.
tested by Barney et al. (1978); BA2 to BA7 and BA10 lp
to BA12 (l/h = 4.0), tested by Lee and Watanabe (2003); 2 RB f yl Asl hl P N 4 yl
E hl
and the coupling beams (l/h = 3.0 or 4.0), tested by Xiao DL
et al. (1999), except for HB3-6L-T100, showed hysteretic Eep M P M N P N 2 y
energy dissipations classified as “moderate.” For slender (14)
3 f yl Asl hl
coupling beams with conventional reinforcement layouts,
2 MP M N

According to Priestley (2000), the yield rotation θy in the end, a moment-plastic rotation (M-θp) relationship (that is,
plastic hinge region can be approximated as [1.7εyl/h]lp. By rigid plastic behavior) is used (refer to Fig. 10); and for the
substituting θy = [1.7εyl/h]lp and eliminating the deformation shear hinge element, a rigid plastic V-Δp relationship that is
terms, the simplified equation of κ is derived from Eq. (14) directly converted from the M-θp relationship is used (that
with little loss of accuracy. is, V = 2M/l and Δp = θpl). Basically, the two methods use
EDL in Eq. (13) and κ in Eq. (14) are applicable only when the compatible modeling parameters for the moment and
shear deformation is negligible. This may be reasonable for shear hinge elements. The moment hinges at beam ends in
slender members with roughly l/h ≥ 8. In fact, most rein- Fig. 9(a) are more natural for slender coupling beams with
forced concrete beams and coupling beams are subjected to conventional reinforcement layouts, the behavior of which
a significant decrease in their energy dissipation capacity is governed by flexural deformations. The shear hinge at the
owing to shear deformation and pinching (Fig. 7(a), Eom and midspan in Fig. 9(b) may be preferred for the modeling of
Park 2010b, 2013). Thus, the actual κ is less than Eq. (14). In short coupling beams with diagonal reinforcement layouts
this study, to apply κ in its current form to coupling beams controlled by shear deformations. Note that the modeling
with 2 ≤ l/h ≤ 8, where shear deformation and pinching are approaches in Fig. 9 are comparable to those in Naish et al.
not negligible, a modification factor λ (≤ 1.0) is multiplied (2013b), except that the slip/extension spring at each beam
as follows. end was not used.
To evaluate the validity of the proposed modeling param-
eters, the load-deformation relations of 17 existing coupling
3 f yl Asl hl
0.15 (15) beam specimens were predicted. Table 2 shows the test
2 MP M N variables and modeling parameters of the coupling beam
specimens. Mn is the flexural strengths of the beam critical
In Eq. (15), the lower limit of κ is taken as 0.15, based section, calculated from the equivalent stress block analysis
on existing test results of beam-column joints with severe in ACI 318-19, neglecting the contribution of steel reinforce-
pinching (Eom et al. 2015). ments in the compression zone (refer to Fig. 8(a)). Ke (= αf
Xiao et al. (1999) reported that the energy dissipation Kf) is the effective stiffnesses, and αf is the effective stiffness
capacity of conventionally reinforced coupling beams is factor in Eq. (4). θy and θd are the yield and ultimate rota-
significantly affected by aspect ratio l/h, the arrangement tions calculated by Eq. (8) and Eq. (9), respectively, and κ is
of longitudinal bars (that is, placing intermediate bars in the energy dissipation ratios, calculated either by Eq. (12) or
the web), and transverse reinforcement details. Values of λ Eq. (15). All modeling parameters in Table 2 were calculated
varying with these design variables can be estimated using on the basis of actual material strengths and reinforcement
the procedure developed by Eom and Park (2013). However, details reported in the literature. Then the load–deformation
such a sophisticated procedure may not be appropriate for relations of the coupling beams were predicted using the
practical application. In this study, an empirical approxima- calculated modeling parameters and Computers and Struc-
tion of λ is proposed as a function of l/h. tures (2018), as follows.
1. For the current version of Computers and Structures
l (16) (2018), the shear hinge model does not use κ for hysteresis
1.0
8h loops. Thus, the load-deformation relations of all specimens
were modeled with moment hinge elements in Fig. 9(a). As
For coupling beams satisfying the requirements of shown in Fig. 10, the plastic behavior of the moment hinge
“conforming transverse reinforcement” in ASCE/SEI 41-17 elements is defined by five-point backbone curves (YULRX
(spacing of closed stirrups ≤ d/3, and shear strength Vs curves) and parallelogram hysteresis loops.
of closed stirrups ≥ 3/4 of required coupling beam shear 2. For the backbone curves, the yield strength at point Y
strength), κ and λ in Eq. (15) and Eq. (16) yield reasonable and ultimate strength at points U and L were taken as MY =
estimates for design. For conventionally reinforced coupling Mn and MU = ML = 1.05Mn, respectively. The rotations at
beams classified as “non-conforming transverse reinforce- point Y and L were taken as θY = 0.00001 rad (that is, almost
ment” in ASCE/SEI 41-17, it is recommended to use the zero) and θL = [θd – θy], respectively, and the rotation at point
lower limit of κ = 0.15 for conservative design. U was taken as 0.95 θL. The strengths and rotations at points
R and X, which are beyond the scope of the study, were
APPLICATION TO COMPUTER MODELING not defined.
Figure 9 shows two modeling methods for coupling beams. 3. Two types of hysteresis loops were considered using
Both methods use an elastic beam element to account for the values of κ calculated by Eq. (12) and Eq. (15) (refer to
linear elastic behavior before flexural yielding; the effective the solid and dashed parallelograms in Fig. 10). The areas
flexural and shear rigidities of the beam element are taken of two hysteresis loops representing the energy dissipation
as 0.3EcIg in Eq. (6) and 0.04[l/h]GAw in Eq. (7), respec- capacity were the same, but the unloading/reloading stiff-
tively. After flexural yielding, plastic deformations (that is, nesses were different.
neglecting elastic deformations) are considered with different Figure 11 compares the predicted load–deformation rela-
hinge models; Fig. 9(a) uses a moment hinge element at each tions of conventionally reinforced coupling beams with test
beam end, whereas Fig. 9(b) uses a shear hinge element at results. The predicted envelope relations by the proposed
the beam midspan. For the moment hinge element at each modeling parameters (that is, Ke, θy, and θd) agree well

Table 2—Modeling parameters of coupling beam specimens
Reinforcement layout Specimens h, mm l, mm l/h Mn, kN-m Ke*, kN-m αf † θy, rad θd, rad κ
NHPL13 350 840 2.40 119 14,100 0.123 0.00571 0.022 0.194
LPH 350 980 2.80 160 20,900 0.157 0.01143 0.035 0.248
HB3-6L-T100 406 1218 3.00 147 36,100 0.172 0.00408 0.035 0.255
HB3-10L-T50 406 1218 3.00 218 36,100 0.172 0.00606 0.035 0.236
Conventionally
HB4-6L-T100 406 1624 4.00 147 35,900 0.229 0.00411 0.035 0.340
reinforced
HB4-10L-T65 406 1624 4.00 218 35,900 0.229 0.00609 0.035 0.319
BA4 300 1200 4.00 70.6 12,500 0.229 0.00564 0.035 0.364
BA5 300 1200 4.00 70.6 12,500 0.229 0.00564 0.035 0.364
BA10 300 1200 4.00 73.6 15,200 0.229 0.00483 0.035 0.349
P07 400 600 1.50 62.5 9480 0.043 0.00741 0.039 0.493
Diagonally IK-No.1 450 800 1.78 190 32,900 0.066 0.00578 0.024 0.493
IK-No.2 450 800 1.78 284 34,300 0.066 0.00829 0.024 0.443
285(+) 0.00813(+) 0.024(+)
IK-No.4‡ 450 800 1.78 35,100 0.066 0.428
302(–) 0.00860(–) 0.024(–)
S-No.1 400 1000 2.50 178 27,900 0.132 0.00637 0.021 0.444
Reinforced (X)
S-No.3 400 1000 2.50 177 27,100 0.132 0.00653 0.021 0.447
NX7L 350 980 2.80 164 18,200 0.157 0.00899 0.033 0.447
LX7L 350 980 2.80 162 14,100 0.157 0.01146 0.033 0.454
*
When calculating Ke, elastic modulus of concrete was determined as Ec = 4700√fc′, except NHPL13, LPH, NX7L, and LX7L, in which values of Ec were reported in the literature.
†
αf = Ke/Kf = Ke/[6EcIg/l] is effective stiffness factor multiplying to EcIg to represent whole effective stiffness of coupling beam in terms of flexural behavior only.
‡
IK-No. 4 was T-shaped beam. Thus, Mn and θy were different depending on loading directions.
Note: 1 ksi = 6.90 MPa; 1 k-ft. = 1.36 kN-m.
Fig. 11—Cyclic load-deformation relations of conventionally reinforced coupling beams: predictions versus tests. (Note:
1 in. = 25.4 mm; 1 k = 4.45 kN)
with the test results. Further, the accuracy of the predicted with the test results, except ultimate rotations θd. As shown
hysteresis loops defined by the values of κ was satisfactory, in Fig. 6, the design equation of θd is defined as the lower
although the unloading and reloading stiffnesses loops do bound for measured ultimate deformations.
not accurately capture the pinched cyclic curves. Figure 12 Consequently, for NX7L, LX7L, IK-No. 1, IK-No. 2, and
compares the predicted load-deformation relations of diag- IK-No. 4, the values of θd calculated by Eq. (9) are less than
onally reinforced coupling beams with test results. The the measured ones. Note that the κ values and consequent
predicted envelope relations and hysteresis loops agree well hysteresis loops are different depending on reinforcement

Fig. 12—Cyclic load-deformation relations of diagonally reinforced coupling beams: predictions versus tests. (Note: 1 in. =
25.4 mm; 1 k = 4.45 kN)
details. For conventionally reinforced coupling beams, the forcement details and l/h, can be used as a modeling param-
values of κ in Table 2 are relatively small, ranging from eter to consider ED. The range of κ is approximately 0.4 ≤ κ ≤
0.194 to 0.364, and thus the hysteresis loops in Fig. 11 are 0.5 for diagonally reinforced coupling beams and κ ≤ 0.35
relatively thin and narrow. On the other hand, for diagonally for conventionally reinforced coupling beams. For practical
reinforced coupling beams, the values of κ (= 0.428 to 0.493) application, κ can be used to define energy-based hysteresis
are relatively large and less variable, and the hysteresis loops models in computer programs for nonlinear modeling.
in Fig. 12 dissipate more energies.
AUTHOR BIOS
CONCLUSIONS AND RECOMMENDATIONS Tae-Sung Eom is an Associate Professor at Dankook University, Yongin,
Gyeonggi, South Korea. He received his BS, MS, and PhD in architec-
This study investigated nonlinear modeling parameters tural engineering from Seoul National University, Seoul, South Korea.
for coupling beams, considering the effects of aspect ratios His research interests include the experiment and analysis of reinforced
(l/h) and reinforcement details. The effective stiffness, yield concrete and composite structures, with an emphasis on performance-based
seismic design.
deformation, deformation capacity, and energy dissipation
ratio were defined on the basis of existing test results of 79 Seung-Jae Lee is a PhD Candidate at Dankook University. He received
coupling beam specimens reported in the literature. The his BS and MS in architectural engineering from Dankook University.
His research interests include seismic evaluation and design of reinforced
major findings are summarized as follows. concrete and masonry structures.
1. The effective stiffness Ke of coupling beams is affected
by l/h, because the governing deformation mode depends on Su-Min Kang is an Associate Professor at Soongsil University. He received
his BS and MS in architectural engineering from Seoul National Univer-
l/h. Such Ke can be defined using effective flexural and shear sity. His research interests include the experiment and design of reinforced
rigidities, EIe = 0.3EcIg and GAe =0.04[l/h]GAw, respectively. concrete structures.
The yield deformation can be estimated as θy = Mn/Ke (Mn =
Hong-Gun Park, FACI, is a Professor in the Department of Architecture
coupling beam flexural strength). and Architectural Engineering at Seoul National University. He received
2. Unlike expectations, the deformation capacity θd shows his BS and MS in architectural engineering from Seoul National University
stronger correlations with reinforcement layouts and trans- and his PhD in civil engineering from the University of Texas at Austin,
Austin, TX. His research interests include numerical analysis and seismic
verse reinforcement ratio, ρt, than l/h. Diagonally rein- design of reinforced concrete and composite structures.
forced coupling beams show an increasing trend of θd with
significant scatter as ρt increases, whereas conventionally ACKNOWLEDGMENTS
reinforced coupling beams show a trend of θd remaining This research was supported by the National Research Foundation of
Korea (NRF-2018R1A6A1A070258 and NRF-2019R1F1A1059248).
constant between 0.035 and 0.05 rad. A design equation for The authors would like to thank I. Kim and S. Cho for their comments
θd is conservatively defined as 10ρt, with the upper limits of and advice on the proposed models during the development of the KCI
0.035 rad for conventionally reinforced coupling beams and nonlinear modeling guideline for performance-based seismic design of RC
building structures.
0.05 rad for diagonally reinforced coupling beams.
3. The hysteretic energy dissipation ED of coupling beams
REFERENCES
is affected by reinforcement layouts and l/h. The energy ACI Committee 318, 2019, “Building Code Requirements for Structural
dissipation ratio κ (= ED/Eep), defined as functions of rein- Concrete (ACI 318-19) and Commentary,” American Concrete Institute,
Farmington Hills, MI, 624 pp.

Ameen, S.; Lequesne, R. D.; and Lepage, A., 2020, “Diagonally Rein- Ratio,” ACI Structural Journal, V. 113, No. 3, May-June, pp. 515-524. doi:
forced Concrete Coupling Beams with Grade 120 (830) High-Strength 10.14359/51688473
Steel Bars,” ACI Structural Journal, V. 117, No. 6, Nov., pp. 199-210. Lim, E.; Hwang, S.-J.; Wang, T.-W.; and Chang, Y.-H., 2016, “An Inves-
American Society of Civil Engineers, 2017, “Seismic Evaluation and tigation on the Seismic Behavior of Deep Reinforced Concrete Coupling
Retrofit of Existing Buildings (ASCE/SEI 41-17),” ASCE, Reston, VA, Beams,” ACI Structural Journal, V. 113, No. 2, Mar.-Apr., pp. 217-226.
p. 416. doi: 10.14359/51687939
Barney, G. B.; Shiu, K. N.; Rabbat, B. G.; and Fiorato, A. E., 1978, Moehle, J. P.; Ghodsi, T.; Hooper, J. D.; Fields, D. C.; and Gedhada, R.,
“Earthquake Resistant Structural Walls—Tests of Coupling Beams (NSF- 2012, “Seismic Design of Cast-in-Place Concrete Special Structural Walls
RA-760844),” Portland Cement Association, Skokie, IL, 151 p. and Coupling Beams: A Guide for Practicing Engineers (NIST GCR 11-917-
Computers and Structures, 2018, “Perform 3D: Nonlinear Analysis and 11REV-1: NEHRP Seismic Design Technical Brief No. 6),” National Insti-
Performance Assessment for 3D Structures User Guide (Version 7.0.0),” tute of Standards and Technology, Gaithersburg, MD, pp. 1-41.
CSI, Berkeley, CA. Naish, D.; Fry, A.; Klemencic, R.; and Wallace, J. W., 2013, “Reinforced
Eom, T.-S.; Hwang, H.-J.; and Park, H.-G., 2015, “Energy-Based Hyster- Concrete Coupling Beams—Part I: Testing,” ACI Structural Journal,
esis Model for RC Beam-Column Connections,” ACI Structural Journal, V. 110, No. 6, Nov.-Dec., pp. 1057-1066.
V. 112, No. 2, Mar.-Apr., pp. 157-166. Naish, D.; Fry, A.; Klemencic, R.; and Wallace, J. W., 2013, “Reinforced
Eom, T.-S., and Park, H.-G., 2010, “Evaluation of Energy Dissipa- Concrete Coupling Beams—Part II: Modeling,” ACI Structural Journal,
tion of Slender Reinforced Concrete Members and Its Applications,” V. 110, No. 6, Nov.-Dec., pp. 1067-1075.
Engineering Structures, V. 32, No. 9, pp. 2884-2893. doi: 10.1016/j. Pacific Earthquake Engineering Center, 2017, “Guidelines for Perfor-
engstruct.2010.05.007 mance-Based Seismic Design of Tall Buildings (TBI Guideline) (PEER
Eom, T.-S., and Park, H.-G., 2010, “Elongation of Reinforced Concrete Report 2017/06),” PEER, University of California, Berkeley, pp. 1-147.
Members Subjected to Cyclic Loading,” Journal of Structural Engi- Park, H.-G., and Eom, T.-S., 2006, “A Simplified Method for Esti-
neering, ASCE, V. 136, No. 9, pp. 1044-1054. doi: 10.1061/(ASCE) mating the Amount of Energy Dissipated by Flexure-Dominated Rein-
ST.1943-541X.0000201 forced Concrete Members for Moderate Cyclic Deformations,” Earthquake
Eom, T.-S., and Park, H.-G., 2013, “Evaluation of Shear Deformation and Spectra, V. 22, No. 2, pp. 459-490. doi: 10.1193/1.2197547
Energy Dissipation of Reinforced Concrete Members Subjected to Cyclic Park, W.-S.; Kang, T. H.-K.; Kim, S.; and Yun, H.-D., 2020, “Seismic
Loading,” ACI Structural Journal, V. 110, No. 5, Sept.-Oct., pp. 845-854. Performance of Moderately Short Concrete Coupling Beams with Various
Eom, T.-S.; Park, H.-G.; and Kang, S.-M., 2009, “Energy-Based Cyclic Reinforcements,” ACI Structural Journal, V. 117, No. 3, May, pp. 141-154.
Force-Displacement Relationship for Reinforced Concrete Short Coupling Paulay, T., 2002, “The Displacement Capacity of Reinforced Concrete
Beams,” Engineering Structures, V. 31, No. 9, pp. 2020-2031. doi: Coupled Walls,” Engineering Structures, V. 24, No. 9, pp. 1165-1175. doi:
10.1016/j.engstruct.2009.03.008 10.1016/S0141-0296(02)00050-0
Galano, L., and Vignoli, A., 2000, “Seismic Behavior of Short Coupling Priestley, M. J. N., 2000, “Performance-Based Seismic Design (Paper
Beams with Different Reinforcement Layouts,” ACI Structural Journal, No. 2831),” Proceedings, Twelfth World Conference on Earthquake Engi-
V. 97, No. 6, Nov.-Dec., pp. 876-885. neering, Auckland, New Zealand, pp. 1-22.
Ishikawa, Y., and Kimura, H., 1996, “Experimental Study on Seismic Shimazaki, K., 2004, “De-Bonded Diagonally Reinforced Beam for
Behavior of R/C Diagonally Reinforced Short Beams (Paper No. 1386),” Good Repairability (Paper No. 3173),” Proceedings, Thirteenth World
Proceedings, Eleventh World Conference on Earthquake Engineering, Conference on Earthquake Engineering, Vancouver, BC, pp. 1-14.
Acapulco, Mexico, pp. 1-8. Son Vu, N.; Li, B.; and Beyer, K., 2014, “Effective Stiffness of Rein-
Kanakubo, T.; Fujisawa, M.; Sako, N.; and Sonobe, Y., 1996, “Ductility forced Concrete Coupling Beams,” Engineering Structures, V. 76, Oct, pp.
of Short Span RC Beams (Paper No. 1369),” Proceedings, Eleventh World 371-382. doi: 10.1016/j.engstruct.2014.07.014
Conference on Earthquake Engineering, Acapulco, Mexico, pp. 1-8. Tegos, I. A., and Penelis, G. Gr., 1988, “Seismic Resistance of Short
Lee, J.-Y., and Watanabe, F., 2003, “Predicting the Longitudinal Columns and Coupling Beams Reinforced with Inclined Bars,” ACI Struc-
Axial Strain in the Plastic Hinge Regions of Reinforced Concrete Beams tural Journal, V. 85, No. 1, Jan.-Feb., pp. 82-88.
Subjected to Reversed Cyclic Loading,” Engineering Structures, V. 25, Xiao, Y.; Esmaeily-Ghasemabadi, A.; and Wu, H., 1999, “High-Strength
No. 7, pp. 927-939. doi: 10.1016/S0141-0296(03)00026-9 Concrete Short Beams Subjected to Cyclic Shear,” ACI Structural Journal,
Lim, E.; Hwang, S.-J.; Cheng, C.-H.; and Lin, P.-Y., 2016, “Cyclic Tests V. 96, No. 3, May-June, pp. 392-399.
of Reinforced Concrete Coupling Beam with Intermediate Span-Depth

ACIChapter!
NOTES:
JOIN AN
The American Concrete Institute has Chapters and Student Chapters located
throughout the world. Participation in a local chapter can be extremely
rewarding in terms of gaining greater technical knowledge and networking
with leaders in the concrete community.
Because chapters are distinct and independent legal entities, membership includes both ACI
members and non-ACI members and is made up of a diverse blend of architects, engineers,
consultants, contractors, educators, material suppliers, equipment suppliers, owners, and
students—basically anyone interested in concrete. Many active ACI members initially became
involved in ACI through their local chapter. In addition to technical programs and publications,
many chapters sponsor ACI Certification programs, ACI educational seminars, project award
recognition programs, and social events with the goal of advancing concrete knowledge.
Check out the Chapters Special Section from the November 2020 Concrete International:
www.concrete.org/publications/concreteinternational.aspx
Student Chapters
Join or form an ACI Student Chapter to maximize your influence, knowledge
sharing, and camaraderie! ACI has 240+ student chapters located throughout
the world, each providing opportunities for students to:
• Connect with their peers and participate in concrete-related activities such as: student
competitions, ACI Conventions, ACI Certification Programs, ACI Educational Seminars, local
chapter meetings, social events, and community service projects;
• Network with members of local chapters, many of whom have been in the industry for
decades and can help to develop professional relationships and offer career advice;
• Win recognition for their universities through the University Award; and
• Learn about the many scholarships and fellowships offered by the ACI Foundation and by
ACI’s local chapters.
   
www.concrete.org/chapters
Title No. 119-S32
Strength Reduction Factors for ACI 318 Strut-and-Tie

Method for Deep Beams
by Victor Aguilar, Robert W. Barnes, and Andrzej Nowak
The strut-and-tie approach has gained importance in reinforced

concrete design practice in the United States in the last two decades.
This method has proven suitable for designing shear-critical struc-
tural members where beam theory is not applicable. However, the
strength reduction factors specified for the ACI 318 strut-and-tie
method have not been calibrated based on the structural reliability
approach. Therefore, the reliability of members designed according
to these provisions is unknown. In this study, the reliability of deep
beams designed using the strut-and-tie method according to ACI
318 building code requirements for structural concrete was deter-
mined. Statistical parameters employed for loads, material uncer-
tainty, and fabrication uncertainty were based on published liter-
ature. The uncertainty in the analytical model was characterized
based on available test results. The findings indicate that current
design practice using the strut-and-tie method promotes the likeli-
hood of a nonductile failure mode relative to a ductile failure mode.
Inconsistencies in reliability with respect to concrete strength are
highlighted. The following reliability-based strength reduction
factors are suggested: ϕ = 0.65 for struts and nodal zones and ϕ =
0.90 for ties.
Keywords: D-region; design strength; ductility; nodal zone; reinforced

concrete; reliability; safety; shear; strut-and-tie model.
INTRODUCTION Fig. 1—Generic deep beam and strut-and-tie model. (Note:

Since 2002, the design of deep beams according to the 1 ft = 12 in.; 1 in. = 25.4 mm.)
ACI 318 Building Code requirements is based on either
nonlinear analysis or using the strut-and-tie method (STM).1 requires the designer to identify one possible load path for
The strut-and-tie approach has proven very useful for each load combination and assure that none of the compo-
shear-critical structures where beam theory does not apply— nents of the load path is overloaded.
for example, pile caps, corbels, beams with holes, anchorage The STM was formally published by Schlaich et al.,2
zones, and deep beams—which are the focus of this study. and it was introduced to U.S. practice in the first edition of
The classical beam theory does not apply to deep beams AASHTO LRFD Bridge Design Specifications3; later, it was
because the assumption of plane sections remain plane after included in Appendix A of ACI 318-021 for building design.
loading is not valid due to significant shear deformations. ACI 318-144 moved the STM to be part of the main body, and
Figure 1 shows the application of the strut-and-tie modeling ACI 318-195 updated several aspects of strut-and-tie design.
approach to a generic deep beam. The complex stress field This shows that strut-and-tie modeling approach has gained
of the concrete structure is represented as an equivalent attention and importance in practice. Theoretical back-
truss, with elements in compression (struts), elements in ground on strut-and-tie modeling is provided by Schlaich et
tension (ties), and the truss joints being called nodal zones al.2,6; ACI 445R7; and Collins and Mitchell,8 while several
(or nodes). The strength of struts and nodal zones control examples of application have been presented by Reineck9;
the dimensions of the concrete member, while the tie forces Reineck and Novak10; and Kuchma et al.11 Experimental
determine the amount of reinforcement required to resist the research on deep beams and comparison with strut-and-tie
external loads. The main assumptions in STM are: 1) equi- shear strength predictions can be found in the literature.12-25
librium must be satisfied; 2) the centroid of each component
and the lines of action of the applied loads must coincide at ACI Structural Journal, V. 119, No. 2, March 2022.
MS No. S-2021-044.R1, doi: 10.14359/51734332, received June 15, 2021, and
joints; and 3) failure occurs when a strut or node crushes or reviewed under Institute publication policies. Copyright © 2022, American Concrete
when a tie reaches its yielding capacity, so a mechanism is Institute. All rights reserved, including the making of copies unless permission is
formed (for statically determinate truss models). The STM closure, if any, will be published ten months from this journal’s date if the discussion

The main updates in ACI 318-195 STM provisions are: 1) combination, the design strength of each strut, tie, and nodal
the inclusion of a confinement factor that increases the effec- zone are to be equal or greater than the ultimate load as
tive compressive strength of struts and nodal zones when stated in Eq. (1)
subjected to triaxial compression; 2) the strut coefficient
was updated; 3) a size effect factor, consistent with the one Fus ≤ ϕFns; Fut ≤ ϕFnt; Fun ≤ ϕFnn (1)
specified for the new one-way shear model based on frac-
ture mechanics,26 was added to the maximum ultimate shear where Fus, Fut, and Fun are the factored force in a strut, in a
force limitation; and 4) modifications in the minimum distri- tie, and on a nodal face, respectively; Fns, Fnt, and Fnn are the
bution reinforcement were incorporated. nominal strength of a strut, a tie, and at a face of a nodal zone,
The load and resistance factors (resistance factors are also respectively; and ϕ is the resistance factor (strength reduc-
called strength reduction factors) for sectional strength in the tion factor). ACI 318-195 provisions specify ϕ = 0.75 for
ACI 318 building code requirements were calibrated based struts, ties, and nodes. The strength of each component in a
on a probabilistic approach.27-29 Nevertheless, the resistance strut-and-tie model are defined in the following subsections.
factors specified for the design using STM have not been
calibrated; hence, the safety level of designs based on STM Strut strength
is unknown. In general, as load factors are related to the The nominal compressive strength of a strut is defined
level of uncertainty in the load sources, the resistance factors differently for reinforced struts and unreinforced struts as
should reflect the difference in variation in the yield strength indicated in Eq. (2)
of the reinforcing bars, which defines the tie strength, and
the compressive strength of concrete, which controls the Fns = fceAcs for unreinforced struts
strut and nodal zone’s strength. Therefore, inconsistencies in Fns = fceAcs + As′fs′ for reinforced struts (2)
safety are suspected because all ACI 318 resistance factors
for strut-and-tie model components are ϕ = 0.75. where Acs is the cross-sectional area at the end of the strut
Experimental research has shown that the failure mode under consideration; As′ is the area of compression rein-
of deep beams can be ductile when the longitudinal reinforcement along the length of strut; fs′ is the stress in the
forcement and stirrups yield before the concrete crushes12; compression reinforcement of the strut at the nominal axial
however, this has occurred under laboratory conditions strength of the strut, which is permitted to take it equal to the
where material strength and dimensions are well controlled. yielding stress of the steel; and fce is the effective compres-
In-place conditions are uncertain: material properties and sive strength of concrete in a strut in accordance with Eq. (3)
dimensions are random variables; the concrete compres-
sive strength is significantly more variable than the yielding fce = 0.85βcβsfc′ (3)
stress of steel; furthermore, the dimensions of cast-in-place
concrete members are much more variable than the steel where βc is the strut and node confinement modification
bar dimensions, which are subject to strict fabrication toler- factor that accounts for the effect of concrete confinement
ances. Therefore, a high likelihood of a premature strut or when the resisting concrete area is larger than the loaded
nodal zone failure is also suspected. area, which definition is the same as the one used for bearing
An assessment of STM provisions in ACI 318-195 with strength (the effect of confinement in strut-and-tie models
regard to its reliability and likelihood of ductile failure mode has been studied by Tuchscherer et al.21 and Breen et al.30). βs
is described in this paper. The analysis includes: 1) deter- is called strut coefficient, and it is a factor used to account for
mination of the current safety margin in terms of reliability the effect of tensile stress presence and cracking reinforce-
index of individual components and the deep beam system; ment on the effective compressive strength of the concrete
2) selection of target reliability indexes; and 3) presentation in a strut; βs = 1.0 for strut with uniform cross-sectional area
of alternatives for strength reduction factors. along the length; βs = 0.75 for struts located in a region of
a member where the width of the compressed concrete at
RESEARCH SIGNIFICANCE mid-length if the strut can spread laterally if minimum strut
The important contribution of this study is the determina- reinforcement is provided; other cases are specified in the
tion of the current safety margin for deep beams designed ACI 318-195 code.
according to ACI 318-195 STM provisions, in terms of reli-
ability indexes, and the development of strength reduction Nodal zone strength
factors for strut, nodal zones, and ties with consideration of The nodal compressive strength of a nodal zone is taken
failure modes. as Eq. (4)
ACI 318-19 PROVISIONS FOR STRUT-AND-TIE Fnn = fceAnz (4)

METHOD
Chapter 23 of ACI 318-195 applies to the design of where Anz is the area of the nodal face under consideration
structural members, or region of members, where loads or which is perpendicular to the line of action of Fun; if Fun is
geometric discontinuities cause a nonlinear distribution of not perpendicular to the nodal face, then shear stresses at
longitudinal strains within the cross section, which there- the face need to be considered. Mohr’s circle can be used
fore includes deep beams. For each applicable factored load to find the principal compressive stress on the nodal face,

or equivalently, the concept of the extended nodal zone can g
also be applied; fce is the effective compressive strength of PF P g 0 (9)

concrete at a face of a nodal zone in accordance with Eq. (5) g
where μg is the mean of g; σg is the standard deviation of g;
fce = 0.85βcβnfc′ (5) and Φ is the standard normal cumulative distribution func-
tion. The reliability index β is defined as34,35
where βc is the strut and node confinement modification
factor previously described; and βn is the nodal zone coeffi- β = μg/σg (10)
cient, a factor used to account for the effect the anchorage of
ties on the effective compressive strength of a nodal zone. βn and the probability of failure is PF = Φ(−β).
= 1.0 if the nodal zone is bounded by struts, bearing areas, or For each limit state, the acceptable safety margin, in
both; βn = 0.80 if the nodal zone is anchoring one tie; and βn terms of a target reliability index (βT) must be selected
= 0.60 if the nodal zone is anchoring two or more ties. first to calibrate strength reduction factors. The selection
of βT depends on the consequences of failure and the cost.
Tie strength The relevant questions include: What happens when the
The tie strength can be calculated as follows when nonpre- limit state is reached? Does the structure have an ability to
stressed reinforcement is used continue resisting the loads? Or failure propagates to other
components and causes the overall collapse, is failure in a
Fnt = Atsfy (6) ductile mode or is it brittle? Therefore, the target reliability
index for the limit state of flexural capacity that occurs in a
where Ats is the area of reinforcement in a tie; and fy is the ductile manner can be lower than for the limit state of shear
yield stress of the reinforcement. capacity as failure occurs in a brittle manner. In a framed
building, target reliability index for columns must be larger
RELIABILITY ANALYSIS than for beams because the consequences of overloading are
The objective of reliability analysis is to incorporate the different. The other factor that is used in determining βT is
uncertainties involved in load and resistance in the design the cost. Safety can be considered as a commodity that can
process to determine a probabilistic safety margin. There be bought. If it is cheap, a high βT is justified; however, when
are many sources of uncertainty inherent to the design of a unit of safety is expensive, a lower βT is acceptable.
concrete structures. Therefore, random variables are inev- In previous ACI 318 calibration,28 the target reliability
itable in structural design; absolute safety or zero proba- index for new design of reinforced concrete beams is βT
bility of failure cannot be achieved. Consequently, struc- = 3.5. This is equivalent to the probability of failure PF =
tural members must be designed to perform their functions 0.233 × 10–3. However, this does not mean that beams fail
with a finite probability of failure.31 The structural reli- at a rate of 1 in 20,000, but that the ultimate limit state of
ability approach—a probabilistic approach—has been flexural capacity is reached. The beam does not fail because
extensively used to calibrate design procedures in modern the system ductility and redundancy allow for redistribution
structural design codes to achieve a pre-selected safety of part of the load. In this study, the target reliability index
margin.28,32,33 This section describes the fundamentals of for shear capacity of newly designed reinforced concrete
reliability analysis, and it provides the statistical models members is assumed ideally as βT = 4.0. This is equivalent
used for the analysis and calibration performed. to the probability of failure PF = 0.317 × 10–4. The reliability
index of deep beams design according with ACI 318-19
Fundamentals of reliability analysis current specifications is determined and used as a minimum
Consider a design formula in the load and resistance factor acceptable target reliability index, as well.
design (LRFD) format (Eq. (7))
Load models
1.2Dn + 1.6Ln ≤ ϕRn (7) The statistical models for load effects were taken from
previous reliability-based calibrations.32,36 Table 1 provides
where Dn, Ln, and Rn, are the nominal values of the dead load the statistical parameters for dead and live load in this
effect, live load effect, and resistance, respectively; and ϕ is analysis in terms of the bias factor (λ, ratio of mean-to-nom-
the strength reduction factor. The corresponding limit state inal), coefficient of variation (COV, ratio of standard devia-
function is tion to mean), and distribution type.
g = R – (D + L) (8) Table 1—Statistical parameters of the load
components
where D, L, and R are random variables representing dead Distri-
load effect, live load effect, and resistance, respectively. A Item λ COV bution Source
positive value of g represents satisfactory performance, and Dead load from cast- Ellingwood et al.36
1.05 0.10 Normal
a negative value of g corresponds to failure. If g is normally in-place component Nowak32
distributed, the probability of failure is 50 years maximum Extreme Nowak and Szerszen36
1.00 0.18
building live load Type I Nowak and Collins31

Resistance model the uncertainty in the forces in each truss component are also
The load-carrying capacity of a structural element is a considered. The statistical parameters used for each random
random variable due to various categories of uncertainties— variable are based on previously published research29,36 and
for example, material properties, fabrication tolerances, and are summarized in Table 2.
analysis simplifications. Each category is represented by a
statistical distribution, and therefore the resistance can be Uncertainty in analysis model
expressed as Eq. (11) The professional factor is defined as the ratio of test results
to predictions based on the analytical method employed.
R = RnMFP (11) Park and Kuchma38 calculated strengths of deep beams based
on ACI 318-0539 using the strut-and-tie specifications and
where Rn is the nominal value of resistance; M is the material compared the results with the measured capacity of 214 deep
factor, which is a distribution that represents the uncertainty beam tests from eight different sources.13-20 The essence of
in material properties; F is the fabrication factor, which is a the ACI 318 STM provisions has not changed since it was
distribution that represents the uncertainty in as-built dimen- first included in ACI 318-021; thus, this information can be
sions; and P is the analysis or professional factor, which used to determine a probabilistic model that captures the
accounts for uncertainties in the analytical model itself. uncertainty in this analysis model (ACI 318 STM). Park
The resistance of the components in strut-and-tie models and Kuchma38 provide ratios between the measured shear
is a function of relevant material properties (for example, at failure to predicted shear strength (Vtest/Vpred). The cumu-
concrete compressive strength or reinforcement yield lative distribution function (CDF) of Vtest/Vpred is plotted for
strength) and member geometry (strut dimensions or amount each source of experimental data on the normal probability
of reinforcing steel). For this analysis, it is assumed that the scale in Fig. 2(a). Failure of a structural member is a conse-
strength of concrete in the actual structural member is less quence of some combination of overload and understrength;
than the specified compressive strength by 10% on average.37 therefore, the probability of failure is determined by the
In the system reliability analysis, the model uncertainty, and upper tail of the load effect distribution and the lower tail of
Table 2—Statistical parameters of random variables involved in resistance model

Material factor λ COV Distribution Source
3000 1.31 0.170
4000 1.24 0.150
5000 1.19 0.135
6000 1.15 0.125
fc′, psi Normal Nowak et al.29
7000 1.13 0.115
8000 1.11 0.110
10,000 1.09 0.110
12,000 1.08 0.110
No. 3 1.18 0.040
No. 4 1.13 0.030
No. 5 1.12 0.020
No. 6 1.12 0.020
No. 7 1.14 0.030
fy = 60,000 psi Lognormal Nowak et al.29
No. 8 1.13 0.025
No. 9 1.14 0.020
No. 10 1.13 0.020
No. 11 1.13 0.020
No. 14 1.14 0.020
Fabrication factor λ COV
Width of cast-in-place beam 1.01 0.040 Ellingwood et al.36
Normal
Area of reinforcement 1.00 0.015 Nowak et al.29
Professional factor λ COV

Normal-strength concrete 1.76 0.142
High-strength concrete 1.31 0.145 Normal This study
Component forces 1.00 0.200

Fig. 2—Vtest/Vpred Cumulative distribution functions on
Fig. 3—Vtest/Vpred based on fc′ presented in: (a) scatter plot;
normal probability scale: (a) based on source; (b) dataset
and (b) box plot.
for fc′ ≤ 6000 psi; and (c) dataset for fc′ > 6000 psi.
Because the strut-and-tie model is selected by the designer
the resistance (strength) distribution. For this reason, accu-
as a simplified load path to represent a more complex stress
rately modeling the lower tail of each source of uncertainty
field, the forces in the struts, nodal zones, and ties may vary
in the resistance statistical model is essential. Appropriate
somewhat. Considering the geometry and loading of the
distributions were fitted to the lower tail of each dataset for
example provided by Uribe and Alcocer40 (shown in Fig. 1),
Vtest/Vpred with good agreement. Thus, the bias factor ranges
the vertical location of the top strut of the STM was varied
from λ = 1.24 – 2.60 and the coefficient of variation ranges
by ±4 in. (100 mm), and the vertical location of the tie was
from COV = 0.10 – 0.31. Researchers24,38 have recognized
varied ±2 in. (50 mm). Thus, the forces calculated in the
that the conservatism of these predictions decreases as the
components varied over a range of 0.8P. The uncertainty in
concrete compressive strength increases. Figure 3 shows
the load path definition is assumed as normally distributed.
Vtest/Vpred ratios as a function of the concrete compressive
Assuming that any meaningful variation occurs within ±3σ
strength in a scatter plot and a box plot. In the box plot, the
of the mean, 0.8P was assigned to the difference between μ –
central mark inside the box indicates the median, and the
3σ and μ + 3σ. Thus, the coefficient of variation is estimated
bottom and top edges of the box indicate the 25th and 75th
as COV = 0.133. This variation depends on the dimension
percentiles, respectively. The whiskers indicate the extreme
of the STM; therefore, COV = 0.20 was conservatively
data. Consequently, two categories of professional factor are
assumed for further analysis.
defined: normal-strength concrete fc′ ≤ 6000 psi, and high-
strength concrete fc′ > 6000 psi. The resulting CDFs and
System reliability index
the fitted normal distributions are shown in Fig. 2(b); these
Strut-and-tie models frequently are statically determinate
distributions were used to model professional factors (refer
trusses. Therefore, one of these systems can be classified as
to Table 2) in the reliability analysis. Although according
a series system, because if one of the truss components fails,
to Fig. 2 and Fig. 3, the predictions by the ACI 318 STM
the truss will not be able to carry the load and the structure
approach are shown conservative in almost all cases, consid-
will fail. For a series system, the system reliability index
eration of sources of uncertainty is needed to study the safety
is bounded by two extremes; the lower bound is the prob-
of deep beam designed using strut-and-tie provisions.
ability of failure when all elements are fully correlated, and

to 0.7 vary as shown in Fig. 4. The reliability index for a tie
is much larger than for struts and nodal zones, which indi-
cates that a tie failure is currently much less likely than a
nonductile strut or nodal zone failure.
Because the reliability indexes obtained for different sizes
of bars are very consistent, a single distribution for all bar
sizes could reasonably be selected for further simulations—
that is, λ = 1.13 and COV = 0.03, as in Nowak et al.29 For
Dn/(Dn + Ln) ≥ 0.6, the tie reliability index is not reported
because it would require a considerable number of simula-
tions to obtain accurate results; the tie reliability index for
Dn/(Dn + Ln) = 0.7 is expected to be close to 6. The reliability
indexes of struts and nodal zones, which range between 2.9
and 3.4, vary with concrete strength. The least β values are
obtained for the extreme concrete strengths: fc′ = 3000 psi
and fc′ = 12,000 psi. These two cases were adopted for further
analyses as references for the normal-strength concrete and
high-strength concrete, respectively.
For the load ratio of 0.5, the reliability index for individual
components of deep beams was found to be β = 3.0 for struts
and nodal zones for normal-strength concrete, and β = 3.3
for high-strength concrete. The tie reliability index was
found to be β = 5.0. Therefore, when using current ACI 318
resistance factors (ϕ = 0.75), a failure that starts by concrete
crushing or splitting is far more likely than a failure initiated
by yielding of the tie reinforcement.
System reliability of generic strut-and-tie deep

beam design
A typical deep beam supporting two concentrated loads
and a corresponding strut-and-tie model are shown in Fig. 1.
Fig. 4—Reliability of individual components: (a) ties; (b) Because the load path is a statically determinate truss, the
struts and nodal zones made of normal-strength concrete; and failure of one individual component theoretically causes
(c) struts and nodal zones made of high-strength concrete. failure of the system; therefore, Eq. (12) is applicable.
the upper bound is the probability of failure when all compo- Table 3 and Table 4 summarize the results of 100 million
nents are statistically independent, as shown in Eq. (12)31 simulations at a load ratio of 0.5 for several sets of resistance
factors for normal-strength concrete and for high-strength
n concrete, respectively. The presented results include the
max Pi PF 1 1 Pi (12) component reliability index (βi), the system reliability index
i 1
(βsys), and the percentage of cases where the tie resistance
where Pi is the probability of failure of the individual controlled the failure mode—that is, the probability of Rt
component i; and n is the total number of components in the being less than Rs and Rnz. The first section of each table
system. In this study, it is assumed that concrete components corresponds to the current design practice with all strength
(strut and nodes) are fully correlated but these components reduction factors equal ϕ = 0.75. Thus, the reliability index
are uncorrelated with the steel reinforcement (ties). of the system (which includes the uncertainty in the load
path definition and the analysis model) was found to be β
CURRENT SAFETY MARGIN OF DEEP BEAMS = 4.0 and β = 3.2 for normal-strength concrete and high-
DESIGNED WITH STM strength concrete, respectively, when using the current
Reliability index of individual components strength reduction factors. Subsequent sections of Table 3
Monte Carlo simulations are performed to determine the and Table 4 show the effects on component and system reli-
probability of failure of each component of the strut-and-tie ability when implementing different sets of strength reduc-
model. Then, the reliability index can be approximated using tion factors.
Eq. (9) and Eq. (10). Assuming there is no uncertainty in In reference to Table 3, for normal-strength concrete,
the βs, βn, and βc factors, the uncertainties associated with among the simulated deep beams for which there was a
struts and nodal zones are equal (related to material and failure, none of the cases was controlled by tie failure. Simi-
dimensions), so the reliability of these components is effec- larly, as shown in Table 4, for high-strength concrete, within
tively assumed to be equal. When using the current resis- the simulated scenarios that represented a failure, only 0.3%
tance factors (ϕ = 0.75), the reliability indexes of individual of the cases were controlled by tie failure.
components for load ratios, Dn/(Dn + Ln), ranging from 0.3

Table 3—Component and system reliability of Table 4—Component and system reliability of
deep beams for normal-strength concrete deep beams for high-strength concrete
Component ϕ βi P(Rt < Rs,nz) βsys Component ϕ βi P(Rt < Rs,nz) βsys
Strut 0.75 3.0 Strut 0.75 3.3
Node 0.75 3.0 0 4.0 Node 0.75 3.3 0.3% 3.2
Tie 0.75 5.0 Tie 0.75 5.0
Strut 0.75 3.0 Strut 0.75 3.3

Node 0.75 3.0 1% 4.0 Node 0.75 3.3 70% 2.9
Tie 0.90 4.0 Tie 1.00 3.4
Strut 0.75 3.0 Strut 0.65 3.9

Node 0.75 3.0 6% 4.0 Node 0.65 3.9 78% 3.4
Tie 1.00 3.4 Tie 0.90 4.0
Strut 0.65 3.4 Strut 0.60 4.3

Node 0.65 3.4 3% 4.2 Node 0.60 4.3 58% 3.8
Tie 0.90 4.0 Tie 0.80 4.7
Strut 0.65 3.4
Node 0.65 3.4 18% 4.2 Strut 0.50 5.1
Tie 1.00 3.4 Node 0.50 5.1 93% 4.1
Tie 0.75 5.0
Target reliability index and performance
As in previous calibrations, past satisfactory practice factors larger than or equal to the reliability index that corresponds
are taken into consideration for the selection of the target β. to current design practice. Also, the factors should be
In this study, special consideration is also given to reduction selected so that the component reliability index of a tie is
of the probability of a nonductile failure. Failure in struts or approximately equal to or less than the component reliability
nodal zones due to high compressive stress is an undesirable index of struts and nodes. Several combinations of strength
nonductile mode of failure. Conversely, if the reinforcement reduction factors were analyzed, some of which are reported
is adequately developed and the ties yield well before the in Table 3 and Table 4.
concrete fails, the failure mode can be ductile. Hence, within Based on the simulation results, for normal-strength
the elements of a strut-and-tie model, the reliability index concrete, the combination of ϕ = 0.65 for strut and nodal
of the tie should be the least of all the components, so that zones and ϕ = 1.0 for ties produces a higher system reliability
reinforcement is likely to yield before concrete is crushed or index than the reliability corresponding to current practice.
splits. This is a crucial assumption in the development of the Also, the probability of a ductile failure mode is increased
strut-and-tie method based on plasticity models. from zero to 18%. For high-strength concrete, the combina-
In a previous calibration of ACI 318 building code, the tion of ϕ = 0.65 for strut and nodal zones and ϕ = 0.9 for ties
target reliability index of βT = 3.5 was considered for the provides a more consistent design with an increased system
ultimate limit state of flexural capacity for slender beams.28 reliability index (from β = 3.2 to β = 3.4, but still lower than
A target reliability index for the ultimate limit state of shear the ideal targeted value βT = 4.0), and the probability of a
strength has not yet been agreed upon the engineering ductile failure mode greatly increases from 0.3 to 78%.
community. Nevertheless, studies have suggested a target For simplicity in design and consistency with other
reliability of βT = 3.5 or larger for this case considering the compression- and tension-controlled failure modes in ACI
consequences of a nonductile failure.28 In this study, the 318, a combination of ϕ = 0.65 for strut and nodal zones and
system reliability index obtained using the ACI 318 STM ϕ = 0.9 for ties could be adopted for all concrete strengths.
current design practices was found to be β = 4.0 for normal- Figure 5 shows the system reliability index for current strength
strength concrete and β = 3.2 for high-strength concrete. reduction factors and for this recommended combination of
These values were then selected as reference values for the strength reduction factors, for load ratios from 0.3 to 0.7, for
calibration of component strength reduction factors. normal-strength concrete and high-strength concrete.
Suggested strength reduction factors for STM DISCUSSION
deep beam design The reliability indexes of deep beams for normal-strength
The STM strength reduction factors should provide a concrete and high-strength concrete are significantly
system reliability index that is ideally βT = 4.0, or at least

future research and calibrations. It is anticipated that this
refinement might affect the relative probability of ductile
versus nonductile failure modes. This is a topic of poten-
tial improvement that extends to all structural actions and
elements, not just the strut-and-tie method.
The results of this study indicate that the reliability of
deep beams built with normal-strength concrete designed in
accordance with ACI 318-195 STM provisions (βsys = 4.0)
exceeds the reliability targeted in the previous calibration of
the ACI 318 building code requirements for flexural ultimate
limit state (βT = 3.5) and meets the ideal target reliability for
shear ultimate limit state defined in this study (βT = 4.0).
Although the overall probability of failure is low, an unde-
sirable nonductile failure is much more likely than a ductile
failure. For high-strength concrete, the system reliability of
deep beams (βsys = 3.2) is less than the reliability targeted in
the previous calibration of the ACI 318 for flexural ultimate
limit state (βT = 3.5) and less than the reliability targeted
for ultimate limit state of shear resistance. In this case, the
undesirable nonductile failure is also far more likely than a
ductile failure.
Fig. 5—System reliability index for: (a) current practice all Consider Table 3 and Table 4. Several combinations of
ϕ = 0.75; and (b) ϕ = 0.65 for struts and nodal zones and ϕ strength reduction factors were considered targeting a system
= 0.9 for ties. reliability index at least as great as the reliability associated
different due to different levels of conservatism within the with current practice, while assuring that the component
analysis model itself. Figure 3 shows that ACI 318-19 STM reliability index of ties is less than or equal to the component
predictions are overly conservative for normal-strength reliability index of struts and nodes.
concretes, and conservative for high-strength concretes. For normal-strength concrete, the combination of strength
Improvement of STM specifications to accomplish a uniform reduction factor ϕ = 0.65 for strut and nodal zones and ϕ =
level of conservatism for a wide range of concrete compres- 1.0 for ties gives a system reliability index of β = 4.2, which
sive strengths would reduce this disparity in the reliability slightly exceeds the target and current-practice system reli-
level. One possibility is to modify the definition of the ability index (β = 4.0). The combination of ϕ = 0.65 for strut
concrete effective compressive strength (fce). Researchers and nodal zones and ϕ = 0.9 for ties also gives a system
have suggested new equations for estimating fce of struts and reliability index of β = 4.2. For high-strength concrete, the
nodal zones including the influence of the concrete compres- combination of ϕ = 0.65 for strut and nodal zones and ϕ =
sive strength, the span-depth ratio, and the tensile strain 0.9 results in a system reliability index of β = 3.4, which
perpendicular to the strut.25,41-44 To avoid overly complex is slightly greater than the reliability index of the current
design provisions, the strength reduction factors should not practice (β = 3.2) and approximately matches the ACI
be different for normal- and high-strength concrete. When a 318 past target for flexural ultimate limit state (βT = 3.5);
consensus can be reached on strut and nodal zone effective however, it is still less than the defined target value (βT =
strength relationships that result in more uniform reliability 4.0). For high-strength concrete deep beams to achieve the
across all practical ranges of concrete strength, an updated ideal target system reliability index (βΤ = 4.0, which is also
reliability-based calibration might be warranted. the current-practice reliability index for normal-strength
Consistent with previous code calibration studies, this concrete) a combination of ϕ = 0.50 for strut and nodal zones
study included only Grade 60 reinforcement. High-strength and ϕ = 0.75 for ties would be required.
steel is increasingly used, and ACI 318-19 allows the use For the current STM provisions, the simplest and most
of Grade 80 and Grade 100 reinforcement for several appli- effective practical path forward would be to adopt the
cations. Hence, a statistical analysis of actual steel yield following combination of strength reduction factors: ϕ = 0.65
strength data for these grades would be valuable for future for struts and nodal zones, and ϕ = 0.90 for ties. Adopting
reliability assessment for a range of structural members, these factors would only slightly increase the system reli-
including deep beams and other shear-critical members. ability for all ranges of concrete strengths but greatly
In this research, the in-place concrete strength assumption increase the relative probability of a ductile failure mode
was adopted for consistency with previous reliability-based for high-strength concrete members. This recommendation
calibrations of ACI 318 strength reduction factors—that is, is based solely on tests and analyses performed on deep
in-place concrete strength is assumed less than the spec- beams; however, it should represent at least an incremental
ified concrete compressive strength by 10% on average. improvement in reliability for other D-regions designed
However, a more sophisticated statistical approach and using STM. As test data from other shear-critical members
more recent in-place strength data should be considered for are collected and analyzed, more specific reliability analyses
will become feasible.

The main practical effect of adopting the strength reduc- on calibration of AASHTO LRFD and ACI 318. He is a member of Joint
ACI-ASCE Committee 343, Concrete Bridge Design, and ACI Committee
tion factors suggested in this paper would be a reduction 348, Structural Reliability and Safety.
in the amount of tie reinforcement required. For members
for which the capacity of the D-region is controlled by the ACKNOWLEDGMENTS
strut or nodal zone strength, the proposed changes would The authors thank the manuscript reviewers for their insightful comments,
which resulted in several improvements to this paper.
require some combination of the following potential reme-
dies: a larger cross section, larger bearing/contact surfaces,
minimum distributed reinforcement, or a greater specified NOTATION
Acs = cross-sectional area of strut perpendicular to axis of strut
concrete compressive strength. Anz = area of face of nodal zone or section through nodal zone
As′ = area of compression reinforcement
Ats = area of nonprestressed reinforcement in tie
CONCLUSIONS COV = coefficient of variation defined as standard deviation-to-mean
The reliability of deep beams designed in accordance ratio
with ACI 318-195 strut-and-tie method was studied. This D = statistical distribution of dead load effect
Dn = nominal dead load effect
reliability analysis was performed using simulations based F = fabrication factor. Statistical distribution that represents vari-
on existing statistical models of load components, material ability in resistance of structural component due to uncertainty
uncertainty, and fabrication geometry uncertainty. in member dimensions
Fnn = nominal strength at face of nodal zone
The research performed in this study supports the Fns = nominal strength of strut
following conclusions: Fnt = nominal strength of tie
1. The current practice of using the same strength reduc- Fun = factored (ultimate) force on face of node
Fus = factored (ultimate) compressive force in strut
tion factor for all STM components (struts, nodal zones, and Fut = factored (ultimate) tensile force in tie
ties) promotes the likelihood of a nonductile failure mode fc′ = specified compressive strength of concrete
relative to a ductile failure mode. fce = effective compressive strength of concrete in strut or nodal zone
fs′ = compressive stress in reinforcement under factored loads
2. The reliability of high-strength concrete deep beams fy = specified yield strength for nonprestressed reinforcement
designed in accordance with ACI 318-19 STM is not consis- g = limit state function
tent with the reliability targeted in the previous calibration of L = statistical distribution of live load effect
Ln = nominal live load effect
ACI 318 requirements for sectional strength. M = material factor. Statistical distribution that represents variability
3. Updating the ACI 318 STM strength reduction factors in resistance of structural component due to uncertainty in mate-
to ϕ = 0.65 for struts and nodal zones and ϕ = 0.90 for ties rial properties
P = professional or analysis factor. Statistical distribution that
would: (a) slightly increase the system reliability across all represents uncertainty in resistance of structural member due to
ranges of concrete strength; and (b) increase the relative analysis simplification or assumptions
probability of a ductile failure mode—especially for higher PF = probability of exceeding limit state or probability of failure
Pi = probability of failure of individual component in system
concrete strengths. R = statistical distribution of strength of structural component
4. A disparity in the reliability of deep beams constructed Rn = nominal strength of structural component
with different concrete strengths due to the nonuniform Rnz = statistical distribution of strength of nodal zone
Rs = statistical distribution of strength of strut
level of conservatism inherent to the analysis method was Rt = statistical distribution of strength of tie
observed. Improvement of the ACI 318 effective compres- Vpred = shear strength predicted from strut-and-tie method
sive strength for struts and nodal zones could decrease the Vtest = measure shear strength from test
β = reliability index
system reliability disparity between deep beams constructed βc = confinement modification factor
with high-strength concrete and those constructed with βi = reliability index of individual component in system
normal-strength concrete. βn = nodal zone coefficient
βs = strut coefficient
βsys = reliability index of system
AUTHOR BIOS Φ(x) = standard cumulative normal distribution
Victor Aguilar is Assistant Professor at Facultad de Ingeniería y ϕ = resistance factor
Tecnología, Universidad San Sebastián, Concepción, Chile. He received λ = bias factor defined as mean-to-nominal ratio
his BCE from Universidad Austral de Chile and his MS and PhD from μg = mean of limit state distribution
Auburn University, Auburn, AL. His research interests include the design of σg = standard deviation of limit state distribution
reinforced concrete structures, strut-and-tie models, and the application of
statistics and probability in civil engineering.
REFERENCES
Robert W. Barnes, FACI, is the Brasfield and Gorrie Associate Professor 1. ACI Committee 318, “Building Code Requirements for Structural
of Civil and Environmental Engineering at Auburn University. He received Concrete (ACI 318-02) and Commentary (ACI 318R-02),” American
his BCE from the Georgia Institute of Technology, Atlanta, GA, and his MSE Concrete Institute, Farmington Hills, MI, 2002, 445 pp.
and PhD from the University of Texas at Austin, Austin, TX. He is a member 2. Schlaich, J.; Schafer, K.; and Jennewein, M., “Toward a Consistent
of Joint ACI-PCI Committee 319, Precast Structural Concrete Code; Joint Design of Structural Concrete,” PCI Journal, V. 32, No. 3, 1987, pp. 74-150.
ACI-ASCE Committee 445, Shear and Torsion; and ACI Subcommittee doi: 10.15554/pcij.05011987.74.150
318-P, Precast and Prestressed Concrete. He received the ACI Structural 3. AASHTO, “LRFD Bridge Design Specifications,” first edition, Amer-
Research Award in 2005. His research interests include structural concrete ican Association of State Highway and Transportation Officials, Wash-
behavior and design. ington, DC, 1994.
Andrzej Nowak, FACI, is Professor and Department Chair of Civil Engi- Concrete (ACI 318-14) and Commentary (ACI 318R-14),” American
neering at Auburn University. He received his MS and PhD from Poli- Concrete Institute, Farmington Hills, MI, 2014, 503 pp.
technika Warszawska, Warszawa, Poland. His major contribution was the 5. ACI Committee 318, “Building Code Requirements for Structural
development of a reliability-based calibration procedure for calculation Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
of load and resistance factors. The procedure was successfully applied Concrete Institute, Farmington Hills, MI, 2019, 624 pp.

6. Schlaich, J., and Schafer, K., “Design and Detailing of Structural for Reinforced Concrete,” ACI Structural Journal, V. 116, No. 4, July-Aug.
Concrete Using Strut-and-Tie Models,” Structural Engineering, V. 69, 2019, doi: 10.14359/51716739
No. 6, 1991, pp. 113-125. 27. Nowak, A. S., and Szerszen, M. M., “Calibration of Design Code for
7. Joint ACI-ASCE Committee 445, “Recent Approaches to Shear Buildings (ACI 318) Part I: Statistical Models for Resistance,” ACI Struc-
Design of Structural Concrete (ACI 445R-99),” American Concrete Insti- tural Journal, V. 100, No. 3, May-June 2003, pp. 377-382.
tute, Farmington Hills, MI, 1999, 55 pp. 28. Szerszen, M. M., and Nowak, A. S. “Calibration of Design Code for
8. Collins, M. P., and Mitchell, D., Prestressed Concrete Structures, Buildings (ACI 318) Part 2: Reliability Analysis and Resistance Factors,”
Prentice Hall Englewood Cliffs, NJ, 1991. ACI Structural Journal, V. 100, No. 3, May-June 2003, pp. 383-391.
9. Reineck, K.-H., ed., Examples for the Design of Structural Concrete 29. Nowak, A. S.; Rakoczy, A. M.; and Szeliga, E. K., “Revised Statis-
with Strut-and-Tie Models, SP-208, American Concrete Institute, Farm- tical Resistance Models for R/C Structural Components,” Andy Scanlon
ington Hills, MI, 2002, 250 pp. Symposium on Serviceability and Safety of Concrete Structures: From
10. Reineck, K.-H., and Novak, L. C., eds., Further Examples for the Research to Practice, P. H. Bischoff and E. Musselman, eds., American
Design of Structural Concrete with Strut-and-Tie Models, SP-273, Amer- Concrete Institute, Farmington Hills, MI, 2012, pp. 1-16.
ican Concrete Institute, Farmington Hills, MI, 2010, 288 pp. 30. Breen, J. E., “Anchorage Zone Reinforcement for Post-Tensioned
11. Kuchma, D. A.; Yindeesuk, S.; and Nagle, T., “Experimental Vali- Concrete Girders,” V. 356, Transportation Research Board, Washington,
dation of Strut-and-Tie Method for Complex Regions,” ACI Structural DC, 1994.
Journal, V. 105, No. 5, Sept.-Oct. 2008, pp. 578-589. 31. Nowak, A. S., and Collins, K. R., Reliability of Structures, second
12. Rogowsky, D. M., and MacGregor, J. G., “Design of Reinforced edition, CRC Press, Boca Raton, FL, 2013.
Concrete Deep Beams,” Concrete International, V. 8, No. 8, Aug. 1986, 32. Nowak, A. S., “NCHRP Report 368: Calibration of LRFD Bridge
pp. 49-58. Design Code,” Transportation Research Board, National Research Council,
13. Smith, K. N., and Vantsiotis, A. S., “Shear Strength of Deep Beams,” Washington, DC, 1999.
ACI Journal Proceedings, V. 79, No. 3, Mar. 1982, pp. 201-213. 33. Nowak, A. S., and Lind, N. C., “Practical Code Calibration
14. Kong, F.-K.; Robins, P. J.; and Cole, D. F., “Web Reinforcement Procedures,” Canadian Journal of Civil Engineering, V. 6, No. 1, 1979,
Effects on Deep Beams,” ACI Journal Proceedings, V. 67, No. 12, Dec. pp. 112-119. doi: 10.1139/l79-012
1970, pp. 1010-1017. 34. Cornell, C. A., “Bounds on the Reliability of Structural Systems,”
15. Clark, A. P., “Diagonal Tension in Reinforced Concrete Beams,” ACI Journal of the Structural Division, V. 93, No. 1, 1967, pp. 171-200. doi:
Journal Proceedings, V. 48, No. 10, Oct. 1951, pp. 145-156. 10.1061/JSDEAG.0001577
16. Oh, J. K., and Shin, S. W., “Shear Strength of Reinforced High- 35. Cornell, C. A., “A Probability-Based Structural Code,” ACI Journal
Strength Concrete Deep Beams,” ACI Structural Journal, V. 98, No. 2, Proceedings, V. 66, No. 12, Dec. 1969, pp. 974-985.
Mar.-Apr. 2001, pp. 164-173. 36. Ellingwood, B.; Galambos, T. V.; MacGregor, J. G. et al., “Develop-
17. Aguilar, G.; Matamoros, A. B.; and Parra-Montesinos, G., “Experi- ment of a Probability Based Load Criterion for American National Standard
mental Evaluation of Design Procedures for Shear Strength of Deep Rein- A 58,” NBS Special Report 577, U.S. Department of Commerce, National
forced Concrete Beams,” ACI Structural Journal, V. 99, No. 4, July-Aug. Bureau of Standards, 1980.
2002, pp. 539-548. 37. Bartlett, F. M., and MacGregor, J. G. “Variation of In-Place Concrete
18. Quintero-Febres, C. G.; Parra-Montesinos, G.; and Wight, J. K., Strength in Structures,” ACI Materials Journal, V. 96, No. 2, Mar.-Apr.
“Strength of Struts in Deep Concrete Members Designed Using Strut- 1999, pp. 261-270.
and-Tie Method,” ACI Structural Journal, V. 103, No. 4, July-Aug. 2006, 38. Park, J., and Kuchma, D. A., “Strut-and-Tie Model Analysis for
p. 577. Strength Prediction of Deep Beams,” ACI Structural Journal, V. 104, No. 6,
19. Tan, K.-H.; Kong, F.-K.; and Teng, S., “High-Strength Concrete Nov.-Dec. 2007, pp. 657-666.
Deep Beams with Effective Span and Shear Span Variations,” ACI Struc- 39. ACI Committee 318, “Building Code Requirements for Structural
tural Journal, V. 92, No. 4, July-Aug. 1995, pp. 395-405. Concrete (ACI 318-05) and Commentary (ACI 318R-05),” American
20. Anderson, N. S., and Ramirez, J. A., “Detailing of Stirrup Reinforce- Concrete Institute, Farmington Hills, MI, 2005, 473 pp.
ment,” ACI Structural Journal, V. 86, No. 5, Sept.-Oct. 1989, pp. 507-515. 40. Uribe, C. M., and Alcocer, S. M., “Deep Beam Design in Accordance
21. Tuchscherer, R.; Birrcher, D.; and Huizinga, M., “Confinement with ACI 318-2002,” Examples for the Design of Structural Concrete with
of Deep Beam Nodal Regions,” ACI Structural Journal, V. 107, No. 6, Strut-and-Tie Models, K.-H. Reineck, ed., American Concrete Institute,
Nov.-Dec. 2010, p. 709 Farmington Hills, MI, 2002, pp. 65-80.
22. Hwang, S.-J.; Lu, W.-Y.; and Lee, H.-J., “Shear Strength Prediction 41. Shuraim, A. B., and El-Sayed, A. K., “Experimental Verification
for Deep Beams,” ACI Structural Journal, V. 97, No. 3, May-June 2000, of Strut and Tie Model for HSC Deep Beams without Shear Reinforce-
pp. 367-376. ment,” Engineering Structures, V. 117, 2016, pp. 71-85. doi: 10.1016/j.
23. Garay-Moran, J. D., and Lubell, A. S., “Behavior of Deep Beams engstruct.2016.03.002
Containing High-Strength Longitudinal Reinforcement,” ACI Structural 42. Su, R. K. L., and Looi, D. T. W., “Revisiting Unreinforced Strut Effi-
Journal, V. 113, No. 1, Jan.-Feb. 2016, doi: 10.14359/51687910 ciency Factor,” ACI Structural Journal, V. 113, No. 2, Mar.-Apr. 2016, doi:
24. Ismail, K. S.; Guadagnini, M.; and Pilakoutas, K., “Shear Behavior 10.14359/51688062
of Reinforced Concrete Deep Beams,” ACI Structural Journal, V. 114, No. 43. Mohamed, K.; Farghaly, A. S.; and Benmokrane, B., “Strut Efficien-
1, Jan.-Feb. 2016, doi: 10.14359/51689151 cy-Based Design for Concrete Deep Beams Reinforced with Fiber-Rein-
25. Chen, H.; Yi, W.-J.; and Hwang, H.-J., “Cracking Strut-and-Tie forced Polymer Bars,” ACI Structural Journal, V. 113, No. 4, July-Aug.
Model for Shear Strength Evaluation of Reinforced Concrete Deep 2016, doi: 10.14359/51688476
Beams,” Engineering Structures, V. 163, 2018, pp. 396-408. doi: 10.1016/j. 44. Ismail, K. S.; Guadagnini, M.; and Pilakoutas, K., “Strut-and-Tie
engstruct.2018.02.077 Modeling of Reinforced Concrete Deep Beams,” Journal of Structural
26. Kuchma, D. A.; Wei, S.; Sanders, D. H.; Belarbi, A.; and Novak, L. Engineering, ASCE, V. 144, No. 2, 2018, p. 04017216 doi: 10.1061/
C., “Development of the One-Way Shear Design Provisions of ACI 318-19 (ASCE)ST.1943-541X.0001974

Title No. 119-S33
Reliability Assessment and Strength Reduction Factor

Calibration for Screw Anchors Concrete Breakout
by Ahmad N. Tarawneh, Eman F. Saleh, and Sereen A. Majdalaweyh
The screw anchor design model was recently adopted in ACI 318-19,

and similar strength reduction factors for post-installed anchors
were applied to screw anchors despite its reliable performance,
which has been proven in experimental studies. Accordingly,
this study presents a comprehensive reliability assessment of the
safety level for the design of screw anchors governed by concrete
breakout. A worldwide database that includes single anchors
embedded in cracked and uncracked concrete, anchors with edge
effects, and a group of anchors is used in the reliability analysis.
Anchor design inherently has different uncertainties related to
multiple variables like concrete strength and variability in load.
These variables have been modeled using multiple distributions.
Furthermore, in this paper, anchors are grouped into three classes
based on the installation effort sensitivity classes 1 to 3, where
this sensitivity is described in a beta distribution. Based on a 3.5
selected reliability index threshold, the ACI 318-19 strength reduc-
tion factors result in conservative reliability indexes that exceed
the selected threshold. Thus, revised strength reduction factors (ϕ)
are proposed. For anchors embedded in uncracked concrete, the
proposed ϕ factors are 0.80 for anchor classes 1 and 2 and 0.75
for anchor class 3. For anchors embedded in cracked concrete, ϕ
factors are 0.75, 0.7, and 0.65 for classes 1, 2, and 3, respectively.
The study also presents a sensitivity analysis to quantify the effect
of each model error parameter on the results. Fig. 1—Screw anchor.
Keywords: first-order reliability method (FORM); Monte Carlo simulation between these factors is required.3 Screw anchors have
(MCS); reliability analysis; screw anchors; strength reduction factors. different failure modes based on the undercut degree and the
embedment depth. Anchors with high embedment depth and
INTRODUCTION small undercut degree exhibit a pullout failure mode, while
Post-installed anchorage systems are generally classi- anchors with a shallower embedment and high undercut
fied into two main categories: bonded anchors, which use experience concrete breakout failure. Between these two
a bonding agent with concrete to develop tensile resistance; extreme failure modes, a mixed pullout/breakout failure
and mechanical anchors, which use friction and mechanical mode is exhibited.3
interlocking with concrete to develop tensile resistance.1 The design of screw anchors was recently adopted in
Mechanical screw anchors can be described as steel anchors ACI 318-19.4 Strength reduction factors associated with
with threads that cut into concrete during installation to post-installed anchors in ACI 318-19 are based on the prob-
provide mechanical interlock (Fig. 1). Screw anchors are abilistic study by Farrow and Klingner in 1995.5 However,
considered a relatively new anchorage system that is gaining their study does not consider the reduction of the anchor
greater attention in the industry for their reliable perfor- capacity to the 5% fractile required by the ACI Standard.
mance and ease of installation.2,3 In contrast to adhesive Also, this study implemented several assumptions for
anchors, which require detailed hole-cleaning steps, screw simplification that are considered a deficiency, such as
anchors can be installed—and removed—easily in a drilled considering live load only, using simple distribution, and a
hole with a hand or impact wrench. limited number of simulations. Moreover, the database used
Similar to other types of anchors, the tensile capacity in Farrow and Klingner includes only anchors embedded
of screw anchors is a function of the embedment depth, in uncracked concrete and does not include screw anchors.
concrete strength, and the undercut degree. The undercut Accordingly, a reliability analysis is needed to assess the
degree is defined by the difference between the outer diam- ACI Structural Journal, V. 119, No. 2, March 2022.
eter of the screw anchor and the hole diameter and represents MS No. S-2021-054.R2, doi: 10.14359/51734190, received September 6, 2021, and
the part of the threads that interlock with the concrete2,3 Institute. All rights reserved, including the making of copies unless permission is
(Fig. 1). A higher undercut degree results in higher capacity obtained from the copyright proprietors. Pertinent discussion including author’s
but also increases the installation difficulty; thus, a balance is received within four months of the paper’s print publication.

safety level of the Code design procedure and to calibrate the ϕ factors for anchor design under concrete breakout failure
strength reduction factors for screw anchors. mode. In anchorage design, using the aforementioned
parameters will maintain a uniform level of safety across
RESEARCH SIGNIFICANCE different conditions that might affect the anchorage perfor-
The availability of new experimental data for the tensile mance. For example, the supplementary reinforcement
strength of screw anchors, as well as the incorporation of allows more deformation capacity and thus a higher strength
the new screw anchors resistance model into ACI 318-19, reduction factor.
requires an assessment of the safety level and the verifica- The strength reduction factor of 0.65 for post-installed
tion of or updating the resistance factors, ϕ, covering the full anchors under tensile loads adopted in ACI 318 is based
range of practical design parameters (for example, concrete on the Monte Carlo reliability analysis done by Farrow
strength). This study provides a comprehensive reliability and Klingner in 1995. Farrow and Klingner indicated that
evaluation and strength reduction factors derivation for a strength reduction factor of 0.65 provides an acceptable
screw anchors’ breakout capacity. The analysis includes a safety limit, probability of failure Pf, and reliability index β.
single anchor embedded in uncracked and cracked concrete, The probability of failure is the conditional probability when
anchors with edge effects, and a group of anchors. the demand (applied loading) exceeds the anchor capacity or
resistance (R). The reliability index represents the number
ACI 318-19 ANCHOR STRENGTH REDUCTION of standard deviations between the zero and the mean in
FACTORS the (resistance-load) distribution. As mentioned earlier, the
Anchorage design according to ACI 318-19 is based on reliability analysis by Farrow and Klingner did not consider
the load and resistance factor design (LRFD) safety concept. the 5% fractile reduction required by the current Standard.
Accordingly, at the ultimate limit state for all relevant load Therefore, maintaining the same 0.65 resistance factor with
combinations, the design strength, obtained by multiplying the 5% fractile will increase the safety level significantly.
the nominal strength Rn by the strength reduction factor ϕ, Farrow and Klingner reported that a current design of
should exceed the required strength expressed in terms of concrete structures accepts reliability index values of 3.0
loads Qi multiplied by load factors γ, where i is the load to 3.5.5 In addition, ASCE/SEI 7-166 reported a targeted
type. Hence, the basic requirement that should be satisfied probability of failure and reliability index of 5.0 × 10−5 and
is expressed in Eq. (1) 3.5, respectively, for sudden or widespread progression of
damage failure. Accordingly, a reliability index of 3.5 will
ϕRn ≥ ΣγiQi (1) be used as a targeted threshold in this study.
In ultimate limit state assessment, the design strength DESIGN OF ANCHORS FOR TENSILE
of the anchorage is based on the nominal resistance to all CONCRETE BREAKOUT
possible failure modes, and the lowest resistance failure The concrete cone breakout capacity is calculated using
mode controls the design. Failure modes include steel anchor the concrete capacity design model (CCD) proposed by
failure, concrete breakout, side blowout, pullout or pull- Fuchs et al.7 The CCD model was adopted by the European
through failure, concrete pryout, and concrete splitting, all Organisation for Technical Assessment (EOTA) in 1997 and
under applicable shear and/or tension loading. ACI 318-19 by ACI 318 in 2002.3 The model is based on the concrete
provides ϕ factors for anchorage design depending on the failure of a 35-degree cone originating at the end of the
following parameters: 1) failure mode; 2) type of loading anchor, as shown in Fig. 2, where hef is the effective embed-
(that is, tension or shear loading); 3) type of anchor installa- ment depth. The CCD basic mean breakout strength of single
tion (that is, cast-in or post-installed anchors); 4) sensitivity post-installed anchors loaded in tension is given by Eq. (2),
of the anchor to installation effort; and 5) the presence or where kc = 35 (13.5 for SI units) for uncracked concrete and
absence of supplementary reinforcement. Table 1 shows the kc = 24.5 (9.45 for SI units) for cracked concrete; fc′ is the
Table 1—ACI 318-19 anchor strength reduction factor governed by concrete breakout
Strength reduction factor ϕ
Type of anchor Anchor category from ACI 355.2 or Tension breakout, bond, or side-face Shear (concrete
Supplementary reinforcement installation ACI 355.4M blowout breakout)
Cast-in anchors Not applicable 0.75
Supplementary reinforcement 1 0.75

0.75
present Post-installed anchors 2 0.65
3 0.55
Cast-in anchors Not applicable 0.70
Supplementary reinforcement 1 0.65

0.70
not present Post-installed anchors 2 0.55
3 0.45

Fig. 2—Concrete capacity design model (CCD).
cylinder concrete compressive strength measured, in psi
(MPa for SI units); and hef is the effective embedment depth,
in inches (mm for SI units). Equation (2) results from multi-
plying the surface area of a concrete breakout cone with the
tensile capacity of concrete and accounting for the size effect
(Fig. 2). For design, the kc factor should be evaluated for
experimental data at 5% fractile as per ACI 318 and ACI
355 (characteristic value). This reduction indicates that 95%
of the predicted capacities will exceed the tested capacities
with a confidence of 90%, assuming a normal distribution.
The capacity of anchors evaluated based on the CCD
method (Eq. (2)) should be modified to include the edge
effects, spacing, the geometric arrangement of the anchor
group, the eccentricity of the resultant tensile force of
fasteners that exhibit tension, the concrete condition at
service (that is, cracked or uncracked), and the possibility Fig. 3—Effective depth of screw anchors proposed by
of concrete splitting on the capacity of the anchors (Eq. (3)). Kuenzlen.8
The edge and/or group effect is included by applying the
modification factors ANC/ANCO and ψed,N, where ANC is the per ICC-ES AC 193. The results from Olsen confirmed the
projected area of the anchor arrangement; and ANCO is the applicability of the reduced effective embedment depth
total projected area of a single anchor far from the edge. equation (Eq. (4)).
Additional modification factors include ψc,N, ψcp,N, and ψec,N,
which represent an increasing factor for uncracked concrete, WORLDWIDE SCREW ANCHORS DATABASE
breakout splitting factor, and breakout eccentricity effect, The worldwide database reported in Olsen et al.3 includes a
respectively. Readers can refer to ACI 318-19 for calculating total of 853 screw anchor tests and 17 unique thread profiles.
the modification factors. Of the 853 tests, 462 single anchors were tested in uncracked
concrete, 248 single anchors were tested in cracked concrete,
N b kc f chef1.5 (2) and 143 were groups of two or four anchors. Table 2 summa-
rizes the data set and the range of tested parameters.
The predicted values of the breakout failures in the
Ncb = (ANC/ANCO)ψed,Nψc,Nψcp,Nψec,NNb (3) database have been evaluated for each category using the
CCD design model, as discussed in the previous section.
In 2004, Kuenzlen8 presented a design procedure for Anchors with embedment depths of less than 40 mm have
screw anchors, which follows the CCD but with the reduced high variability owing to the surface effect of the concrete,
effective embedment depth given by Eq. (4). and these anchors are therefore excluded from testing and
qualification for metal anchors in concrete (ETAG 001
hef = 0.85(hnom – 0.5ht – hs) (4) and AC193). Accordingly, they were excluded from the
worldwide database.
where hnom is the nominal length of the anchor; ht is the Ratios of the tested capacity to predicted capacity using
thread spacing; and hs is the tip length after the last thread CCD for each category (single anchor in uncracked and
(Fig. 3). The effective embedment depth for screw anchors is cracked concrete, anchor with edge effects, and a group of
reduced because the tip of the screw has no threads and does anchors) are shown in Fig. 4. The figure also indicates the
not engage with the concrete. Equation (4) was validated 5% fractile reduction limit required by ACI 318-19. The 5%
by Kuenzlen using data from 500 tests of screw anchors. fractile reduction implies that only 5% of actual (experi-
Olsen et al.3 expanded the data from Kuenzlen with an addi- mental) capacities will fall below the predicted capacities
tional 353 test data conducted by independent laboratories with a confidence of 90%. Equation (4) is used to calculate
the value of kc that achieves this reduction, where Fm is the
Table 2—Summary of worldwide screw anchor database
Nominal embedment range, Drill cutting diameter range, Concrete compressive strength, MPa
Test Number of tests mm (in.) mm (in.) (psi)
Single anchor, uncracked 407 41.3 to 127 (1.63 to 5) 6.5 to 18.4 (0.256 to 0.725) 13 to 68 (1904 to 9911)
Single anchor, cracked 166 28.2 to 111.13 (1.63 to 4.38) 8.2 to 18.3 (0.323 to 0.721) 20.4 to 49.7 (2959 to 7208)
Single anchor, edge effect 99 50 to 111.13 (1.96 to 4.38) 8.2 to 18.3 (0.323 to 0.721) 17.5 to 22.3 (2538 to 3236)
Anchor group 106 50 to 70 (1.97 to 2.76) 10.2 (0.4) 22.3 to 28.9 (3234 to 4188)
Fig. 4—Comparison of data set for various anchor types under tension loading: (a) single screw anchor in uncracked concrete;
(b) single screw anchor in cracked concrete; (c) screw anchors with edge effects; and (d) group of anchors.
mean value (mean capacity); ν is the coefficient of variation maxi[P(Fi)] ≤ P(F) ≤ 1 – Πi[1 – P(Fi)] (5)
(COV); and the K value is a factor for one-sided tolerance
limits for normal distributions,9 corresponding to a 5% prob- where P(Fi) is the probability of failure of mode i; and P(F)
ability of non-exceedance with a confidence of 90%. is the probability of failure of the actual series system.
In design, if the anchor was detailed according to Code
F5% = Fm(1 – Kν) (4) provisions (for example, ACI 318) to preclude side blowout,
splitting failure, and steel failure (for example, limiting
edge distance, member thickness, anchor steel diameter, and
SYSTEM RELIABILITY UNDER MULTIPLE so on) and to ensure that cone breakout will be the most
FAILURE MODES probable mode of failure,11 the probability of failure of the
As discussed earlier, the anchor capacity in concrete is a series system can be defined to be approximately equal to the
function of its failure modes. In the anchor design proce- concrete cone failure (that is, P(F) ≈ P(Fconcrete cone)).
dure, the capacity for each failure mode is calculated, and
the mode with the lowest capacity governs the design. Thus, Load model
all these failure modes perform as a series system in which Load combinations (Eq. (6)) were considered in the load
the probability of failure of a single anchor can be defined by model in the reliability analysis. It should be noted that
the failure of one or more of these modes10 (Fig. 5). ACI 318-19 includes various load cases that may apply to
The probability of failure of the aforementioned series anchor design; however, to reduce the complexity of the anal-
system can be evaluated using probability bounds (that is, ysis, only the load cases in Eq. (6) are considered. According
lower and upper bound of the probability of failure), on the to Nowak and Collins,12 the dead load is normally distrib-
assumption that the failure modes in the series are some- uted with a bias factor and a COV of 1.05 and 0.1, respec-
where between fully dependent and completely independent tively. The live load follows the extreme value distribution
(Eq. (5)).10 (Type I) with a bias and COV of 1.00 and 0.18, respectively.

with the one derived using standard tests. Several models
have been suggested to define the bias and COV of the
concrete strength (JCSS,14 Wood,15 CEB,16 and Nowak and
Szerszen17). However, these models are only valid for new
concrete construction where the strength of standard cubes
and cylinders can be obtained. Concrete in existing struc-
Fig. 5—Single anchor failure as series system. tures, which is mostly the case in post-installed anchors,
requires consideration of the variability of the in-place
1.4 PD concrete strength test procedure (for example, the rebound
max Rn (6)
1.2 PD 1.6 PL hammer test, pulse-velocity test, core drilling, and so on),
uncertainty in the relationship that translates the test values
Resistance model into the equivalent cylinder or cube strength, the number
Selecting the resistance model for screw anchors capacity of tests, and the variability of concrete in the structure
is crucial to the reliability analysis. As discussed earlier, the (ACI 228.1R-03).
multiplicative form of resistance that relates tested strength Nowak and Szerszen17 proposed representing the bias in
R and nominal strength Rn is based on the ACI 318-19 design in-place concrete compressive strength with a normal distri-
equation of screw anchors. This multiplicative form can be bution with a bias factor (λfc′) evaluated as in Eq. (9) and a
expressed in a simple form, as shown in Eq. (7)11 COV of 0.1. To represent the variability in in-place concrete,
the in-place COV will be increased by 0.05.
R = P . M . F . Q . Rn (7)
λfc′ = –2.47 × 10–5(fc′)3 + 3.17 × 10–3(fc′)2 – 1.35 × 10–1fc′
where P, M, and F are random variables with specific distri- + 3.0649 ≥ 1.15 (fc′ in MPa) (9)
butions; P is the professional factor that considers the differ-
ences between the tested strength and nominal strength due Based on ACI 228.1R-03, the test anchors are tested in
to the inability of the nominal resistance model to capture member cast within two nominal compressive strength
the true behavior; M is the material factor that accounts for ranges: 1) low-strength concrete ranges between 17 and
the variability in material properties; F is the fabrication 28 MPa; and 2) high-strength concrete ranges between 46
factor that includes section properties (for example, dimen- and 60 MPa.
sion) variability; and Q is a random variable representing the Accordingly, any strength reduction factor derived herein
sensitivity of anchors to bad workmanship. is only valid within these ranges of compressive strength (for
To use this model (Eq. (7)), the distribution of each random example, anchors tested in ultra-high-strength concrete may
variable (P, M, and F) should be defined. For simplicity, the not exhibit the same behavior and may require different ϕ).
bias (that is, R/Rn and COV of the system) can be estimated The influence of site imperfections, quality of workman-
from Eq. (8a) and (8b) respectively ship, and sensitivity of the anchor to installation effort can
also be added using the following equation (Eq. (10))
(R/Rn) = λR = P . M . F . Q (8a)
Q = 1 – PW(1 – S) (10)
VR2/ Rn V R 2 VP2 VM2 VF2 VQ2 (8b) where Q is a quality control modification factor; PW is the
probability of having bad workmanship; and S is the sensi-
where R, P, M, F, and Q are the mean values of resistance, tivity of the anchor to installation effort.
professional factor, material factor, fabrication factor, and According to CEB,16 anchors can be grouped into three
sensitivity factor, respectively. classes based on their sensitivity to installation effort. Each
To obtain the uncertainty in resistance due to material and sensitivity class can be described by a beta distribution
fabrication errors, Monte Carlo simulations (MCSs) were with four parameters, the first two describing the shape of
performed on the nominal resistance model (that is, concrete the distribution and the last two representing the lower and
cone breakout model) given the statistical properties (bias, upper bounds as follows
COV, and the distribution type) of the design parameters. • Class I: System with low sensitivity to installation
Table 3 summarizes the statistical properties of the design effort: S~Beta (10; 4; 0.9; 1)
parameters. According to Bergmeister,13 the variability • Class II: System with normal sensitivity to installation
in the embedment depth is considered very small; thus, effort: S~Beta (3; 3; 0.75; 1)
embedment depth can be considered a deterministic value. • Class III: System with high but acceptable sensitivity to
The same observation can be made regarding the spacing installation effort: S~Beta (5; 11; 0.65; 1)
between the anchors, and, therefore, spacing is also consid- The professional factor can be obtained from a compar-
ered a deterministic value.13 Assessing the uncertainty in ison of the experimental strength and the predicted strength
concrete compressive strength, fc′ is more difficult and Rexp/Rn (MacGregor et al. 1983). According to MacGregor
requires the consideration of variability in concrete strength et al. (1983), the variability of the professional factor can be
within the structure, the age of the concrete at first loading, obtained by subtracting the uncertainty due to impression in
and variation in in-place compressive strength compared the test method Vtest and the variability of errors introduced

Table 3—Statistical properties of design parameters in concrete
Random variable Distribution Mean COV Reference
hef Deterministic — — Bergmeister13
Anchors spacing Deterministic — — Bergmeister13
fc′ Normal Eq. (8) 0.15 Nowak and Szerszen17 + ACI 228.1R-03
PW Beta Beta (3; 5; 0; 0.6) Hokkanen and Scheiwiller10
Class I Beta (10; 4; 0.9; 1)
S Beta Class II Beta (3; 3; 0.75; 1) CEB16
Class III Beta (5; 11; 0.65; 1)
Class I 1.58 0.158
λR
Normal Class II 1.54 0.16 Monte Carlo simulation (106 samples)
(uncracked)
Class III 1.50 0.162
Class I 1.98 0.194
λR
Normal Class II 1.94 0.198 Monte Carlo simulation (106 samples)
(cracked)
Class III 1.89 0.197
λD Normal 1.05 0.1 Szerszen and Nowak21
λL Extreme Type I 1.0 0.18 Szerszen and Nowak21
Note: PW is the probability of having bad workmanship; S is sensitivity of anchor to installation effort; λ is model error bias factor.
due to differences between the strength measured in actual combinations introduced in the design codes (for example,
and control conditions Vcond from the variation of the exper- ACI 318-19).
imental-to-predicted strength VRexp/Rn (refer to Eq. (11)). Vtest
and Vcond are hard to calibrate, and it is assumed herein that RELIABILITY ANALYSIS OF SINGLE ANCHOR
Vtest and Vcond are equal to 0.02 and 0.04, respectively. BREAKOUT IN CRACKED AND UNCRACKED
CONCRETE
One of the objectives of reliability analysis is to calibrate
V Vtest Vcond (11)
2
VP
2 2
Rexp / Rn the strength reduction factor ϕ to obtain certain target reli-
ability over the range of practical design and loading condi-
Based on the observations from the worldwide anchor tions. Generally, the target reliability level is set based on the
database, the VRexp/Rn was approximately 0.15 and 0.3 for consequence of failure, cost of construction, and experience
uncracked and uncracked concrete, respectively. with similar existing construction.12 As discussed earlier, the
targeted reliability index for screw anchors in this study is set
Limit state function at 3.5.5,6 The first-order reliability method (FORM)18 is used
The limit state function for a single screw anchor can be in this reliability analysis, following the steps outlined here:
written in the following form • Define the distribution of λR, λD, and λL.
• Transform and standardize the variables R, D, and L to
G = R – D – L = λRRn – λDDn – λLLn the independent standardized normal variables r, d, and
l.
• Transform G(R, D, L) to G(r, d, l).
R L • For various values of Ln/Dn (that is, calibration points)
Dn R n D L n
Dn Dn and selected ϕ, compute the reliability index, β.
• Select ϕ that minimize the objective function (β – βT)2,
where βT is the targeted reliability index.
1.2 Dn 1.6 Ln L Figures 6 and 7 show the results of the reliability analysis
Dn R D L n (12)
Dn Dn for different strength reduction factor values and different
λR, λD, and λL represent the bias of the resistance, dead load, sensitivity classes for single anchors embedded in uncracked
and live load, respectively. The bias terms are random vari- and cracked concrete, respectively. As shown in the figure,
ables with their distribution. The distribution of λR is derived the optimal ϕ for anchor design is highly dependent on the
using the Monte Carlo technique presented in the previous sensitivity classes. Additionally, it is shown that current
section, and λD and λL follow the normal and extreme value strength reduction factors adopted by ACI 318-19 lead to a
distribution Type I, respectively (refer to Table 3). Rn, Dn, conservative result, with reliability exceeding the selected
and Ln represent the nominal resistance, dead load, and live threshold. Proposed strength reduction factors ϕ for screw
load, respectively. The nominal values are assumed to be anchors governed by breakout failure were chosen to achieve
deterministic and based on the resistance model and load the selected reliability index, as shown in Table 4, and have
been rounded to the nearest 0.05.

Fig. 6—Reliability index β versus load ratio for single anchors in uncracked concrete with different ϕ; (a) for class I system:
system with low sensitivity to installation effort; (b) for class II system: system with normal sensitivity to installation effort; and
(c) for class III system: system with high but acceptable sensitivity to installation effort.
Reliability analysis for anchors with edge effects 2. The nominal mean of resistance (Rn) distribution
and group of anchors in uncracked concrete is computed using the LRFD equation (Eq. (6)). The
To verify the adequacy of the proposed strength reduc- nominal resistance mean equals the factored load (using the
tion factors, reliability analysis for screw anchors with edge controlling load combination) divided by the strength reduc-
effects and a group of screw anchors was conducted using tion factor. Both the ACI and the proposed strength reduc-
ACI 318-19 and the proposed ϕ factors. For the reliability tion factor of 0.65 and 0.8, respectively, were used.
analysis of group anchors, a Monte Carlo-based reliability The values of the experimental-to-predicted ratios were
analysis is conducted assuming class 1 anchors because it is used to choose the distribution type and distribution param-
more common in anchorage. The reliability analysis proce- eters. The experimental-to-predicted ratios were fitted to
dure employed in Farrow et al.20 was modified to overcome either generalized extreme value distribution or normal
the limitations of their study by considering more realistic distribution based on normality tests and probability plots.
distribution types, adding combinations of live and dead Measurements of live loads in building codes are gener-
loads to cover a wide range of live-to-dead-load ratios (L/D), ally higher than what the building experiences. For example,
and using up to 1 × 109 simulations to guarantee convergence. building codes specify a live load at the 95th percentile for
The modified reliability procedure is illustrated in the offices.5 This means the design load is higher than or equal
following steps: to 95% of the observed live loads. This value corresponds
1. Start by assuming a dead load distribution with a mean to 1.325 for live load extreme Type I distribution. Figure 8
of one. Szerszen and Nowak21 reported that dead load presents a simple illustrative diagram of the adopted proce-
exhibits a normal distribution with a bias and COV of 1.05 dure assuming an L/D of three.
and 0.1, respectively. The nominal live load mean is deter- 3. With three distributions determined, the MCS is
mined based on the selected L/D. Szerszen and Nowak21 performed to generate up to 1 × 109 simulations, to reach the
reported that live load exhibits extreme Type I distribution optimal number of simulations. Then, the limit state function
with a bias and COV of 1.00 and 0.18, respectively. The in Eq. (13) is used to find the probability of failure at each
actual mean equals the nominal mean multiplied by the bias. simulation. Where R is the ultimate capacity for the column,
Extreme Type I distribution is also known as Gumbel PD and PL are the dead and live load, respectively.
distribution and is used to model the maximum (or the
minimum) of many samples of various distributions. G = R – PD – PL (13)

A negative value for the G function represents a failure state (D + L). Similar to previous reliability results, the reliability
but a safe state otherwise. The probability of failure is calcu- index for anchors decreases with the increase in load ratios,
lated by dividing the frequency for which the G function with the lowest β value corresponding to the D/(D + L) ratio
presents a failure state by the total number of simulations. of 0.88, at which the controlling load combination switches
The reliability analysis is conducted for a range of L/D (Eq. (6)). This observation shows that including only live
from 0 to 4. The L/D is usually expressed by the term D/ load in the analysis may lead to unconservative results.
(D + L). The adopted range of the L/D corresponds to a D/ Figure 10 shows the ACI strength reduction factor results
(D + L) ratio from 0.2 to 1. in the reliability index exceeding the selected threshold
4. A convergence study has been conducted for each simu- (β = 3.5). On the other hand, the revised strength reduc-
lation to ensure convergence. Figure 9 presents a sample of tion factors allow additional strength of the anchors corre-
the convergence study that shows the relationship between sponding to 15.3% for anchors in uncracked concrete and
the number of simulations versus Pf. 23.1% for anchors in uncracked concrete while maintaining
uniform and acceptable safety levels.
Reliability analyses results Regarding the small drop in the reliability index below the
Figure 10 shows the reliability index for screw anchors acceptable limit at some of the proposed reduction factors, it
with edge effects and a group of anchors as a function of D/ is important to point out that from the outset of establishing
the very notion of “targeted reliability,” the prime concern
was to maintain a consistent reliability level within the
Table 4—Proposed strength reduction factors ϕ
broad domain of load levels without excessive conservatism.
for screw anchors governed by concrete tension
Thus, this small drop was deemed negligible because it was
breakout
considered small (a 2% drop beyond the acceptable level)
Anchor class Uncracked concrete Cracked concrete and noticed only in a very narrow range of loads.
1 0.8 0.75
2 0.8 0.70
SENSITIVITY ANALYSIS
A larger data set allows more accurate calibration of the
3 0.75 0.65
model error. Herein, a sensitivity analysis is performed to
Fig. 7—Reliability index β versus load ratio for single anchors in cracked concrete with different ϕ: (a) for class I system:
system with low sensitivity to installation effort; (b) for class II system: system with normal sensitivity to installation effort; and
(c) for class III system: system with high but acceptable sensitivity to installation effort.

Fig. 8—Illustrative diagram for determining loads and
resistance distributions.
Fig. 10—Reliability index for anchors with edge effects

and group of anchors using ACI and proposed strength
reduction factors.
and uncracked concrete, anchors with edge effects, and a
group of anchors. The reliability analysis is conducted using
the first-order reliability method (FORM) for single anchors
Fig. 9—Convergence plot of Monte Carlo results as function embedded in cracked and uncracked concrete for different
of number of trials. anchor classes (which represent anchor sensitivity to instal-
lation). In addition, to verify the adequacy of the proposed
identify the influence of the change in model error param-
resistance factors, reliability analysis for anchors with edge
eters on the reliability study results. Based on the sensi-
effects and group of anchors has also been conducted using
tivity study (refer to Fig. 11), several conclusions can be
Monte Carlo simulation (MCS).
drawn. Regarding the effect of the mean of the model error,
The results showed that ACI 318 strength reduction
the influence seems to follow a parabolic trend (refer to
factors for the concrete breakout, which are based on an
Fig. 10(a)). The relationship between the reliability index
anchors database that does not include screw anchors, result
and mean of resistance bias may help decide on a different
in a conservative reliability index exceeding 3.5. Based on
fractile strength model (for example, 10% fractile strength
the reliability results, revised strength reduction factors for
model) that ensures a certain bias between the test results
breakout failure are proposed to provide a uniform reliability
and predicted values as a measure of safety. Concerning
index of 3.5 as follows: for anchors embedded in uncracked
bias variability (that is, COV of λR), higher variability has
concrete, the proposed ϕ factors are 0.80 for anchor classes
a negative influence on the reliability of the anchor design
1 and 2, and 0.75 for anchor class 3. For anchors embedded
(refer to Fig. 11(b)), whereas this rate of reduction seems to
in cracked concrete, the ϕ factors are 0.75, 0.7, and 0.65 for
decrease at larger COV of λR. Furthermore, the decrease in
classes 1, 2, and 3, respectively.
reliability with an increase in the COV of resistance shows
signs of independence from the load ratio at a COV greater
AUTHOR BIOS
than 0.1. The COV of λR has a greater influence on the reli- ACI member Ahmad N. Tarawneh is an Assistant Professor in the Depart-
ability index than the mean of λR, whereas a small change ment of Civil Engineering, Faculty of Engineering at The Hashemite Univer-
may cause a significant reduction in reliability. Concerning sity, Zarqa, Jordan, and Head of the Building Environment and Structural
Systems Center. He received his BS and MS from The University of Jordan,
the effect of live load variability, as expected, the effect is Amman, Jordan, and his PhD from Clemson University, Clemson, SC. He
larger at a higher live load ratio and an increasing rate with is a member of ACI Committees 355, Anchorage to Concrete, and 440,
the COV of live load (refer to Fig. 11(c)). This implies that Fiber-Reinforced Polymer Reinforcement. His research interests include
anchorage to concrete and the use of fiber-reinforced polymers (FRPs) in
an accurate assessment of live load variability is necessary reinforced concrete structural elements.
at high live load ratio conditions to certify the designed
target reliability. Eman F. Saleh is an Assistant Professor in the Department of Civil Engi-
neering, Faculty of Engineering at The Hashemite University. She received
her BS and MS from The University of Jordan, and her PhD from Illinois
CONCLUSIONS Institute of Technology, Chicago, IL. Her research interests include reli-
The screw anchor design breakout model was recently ability analysis, risk assessment, and nondestructive assessment of concrete
damage.
adopted in ACI 318-19. Accordingly, this study performs a
comprehensive reliability analysis to assess the safety level Sereen A. Majdalaweyh is a PhD Candidate in the Department of Civil
and the adequacy of the ACI 318-19 strength reduction Engineering at Clemson University. She received her BS from Mu’tah
University, Mu’tah, Karak, Jordan, and her MS from The University of
factors (ϕ). The study uses a worldwide database of exper- Jordan. Her research interests include seismic loss and fragility assessment
imental data on screw anchors, including those in cracked

Fig. 11—Class I system reliability index β of single anchor in cracked concrete for different load ratio with ϕ = 0.75: (a) sensi-
tivity analysis of mean of resistance λR; (b) sensitivity analysis of COV of resistance λR; and (c) sensitivity analysis of COV of
live load.
of nonstructural components and contents, and optimal risk mitigation of 9. Owen, D. B., “Factors for One-Sided Tolerance Limits and for Vari-
these components. ables Sampling Plans (SCR-607),” Sandia Corporation Monograph, Wash-
ington, DC, Mar. 1963, 414 pp.
ACKNOWLEDGMENTS 10. Hokkanen, P. F., and Scheiwiller, A. P., “Probabilistic Models for
The authors would like to acknowledge and thank S. Sawyer at DEWALT the Resistance of Concrete Anchors,” V. 242, Institute of Structural Engi-
for her support in providing the screw anchors database. Also, the authors neering, Swiss Federal Institute of Technology Zürich (ETH Zürich),
acknowledge the Palmetto Cluster (Supercomputer) team at Clemson Zürich, Switzerland, May 1999, 58 pp.
University for their support. 11. Melchers, R. E., and Beck, A. T., Structural Reliability Analysis and
Prediction, John Wiley & Sons, Inc., Hoboken, NJ, 2018, 528 pp.
12. Nowak, A. S., and Collins, K. R., Reliability of Structures, CRC
REFERENCES Press, London, UK, 2012, 407 pp.
1. Cook, R. A.; Burtz, J. L.; and Ansley, M. H., “Design Guidelines and 13. Bergmeister, K., “Bemessung von Dübelverbindungen im Stahl-
Specifications for Engineered Grouts Used in Anchorage and Pile Splice betonbau auf der Grundlage von Teilsicherheitsbeiwerten (Bericht Nr.
Applications (FDOT Research Report BC354 RPWO #48),” University of 7/6-90/2),” Institute of Construction Materials, University of Stuttgart,
Florida, Gainesville, FL, Aug. 2003, 129 pp. Stuttgart, Germany, 1990.
2. Tarawneh, A. N.; Ross, B. E.; and Cousins, T. E., “Tensile Behavior 14. Vrouwenvelder, T., “Special Issue Devoted to the Work of the Joint
and Design of Screw Anchors in Thin Concrete Members,” ACI Structural Committee on Structural Safety,” Structural Safety, V. 19, No. 3, 1997.
Journal, V. 117, No. 1, Jan. 2020, pp. 91-102. doi: 10.14359/51718011 15. Wood, S. L., “Evaluation of the Long-Term Properties of Concrete,”
3. Olsen, J.; Pregartner, T.; and Lamanna, A. J., “Basis for Design of ACI Materials Journal, V. 88, No. 6, Nov.-Dec. 1991, pp. 630-643.
Screw Anchors in Concrete,” ACI Structural Journal, V. 109, No. 4, 16. Comité Euro-International du Béton, Design of Fastenings in
July-Aug. 2012, pp. 559-568. Concrete: Design Guide, CEB Bulletin No. 233, Thomas Telford Ltd.,
4. ACI Committee 318, “Building Code Requirements for Structural London, UK, 1997, 83 pp.
Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American 17. Nowak, A. S., and Szerszen, M. M., “Calibration of Design Code
Concrete Institute, Farmington Hills, MI, 2019, 624 pp. for Buildings (ACI 318): Part 1—Statistical Models for Resistance,” ACI
5. Farrow, C. B., and Klingner, R. E., “Tensile Capacity of Anchors with Structural Journal, V. 100, No. 3, May-June 2003, pp. 377-382.
Partial or Overlapping Failure Surfaces: Evaluation of Existing Formulas 18. Hasofer, A. M., and Lind, N. C., “Exact and Invariant Second-
on an LRFD Basis,” ACI Structural Journal, V. 92, No. 6, Nov.-Dec. 1995, Moment Code Format,” Journal of the Engineering Mechanics Divi-
pp. 698-708. sion, ASCE, V. 100, No. 1, Feb. 1974, pp. 111-121. doi: 10.1061/
6. ASCE/SEI 7-16, “Minimum Design Loads and Associated Criteria JMCEA3.0001848
for Buildings and Other Structures,” American Society of Civil Engineers, 19. MacGregor, J. G.; Mirza, S. A.; and Ellingwood, B., “Statistical
Reston, VA, 2017, 800 pp. Analysis of Resistance of Reinforced and Prestressed Concrete Members,”
7. Fuchs, W.; Eligehausen, R.; and Breen, J. E., “Concrete Capacity ACI Journal Proceedings, V. 80, No. 3, May-June, 1983, pp. 167-176.
Design (CCD) Approach for Fastening to Concrete,” ACI Structural 20. Farrow, C. B., Frigui, I.; and Klingner, R. E., “Tensile Capacity of
Journal, V. 92, No. 1, Jan.-Feb. 1995, pp. 73-94. Single Anchors in Concrete: Evaluation of Existing Formulas on an LRFD
8. Kuenzlen, J. H., “Load-Bearing Behavior of Screw Anchors under Basis,” ACI Structural Journal, V. 93, No. 1, Jan.-Feb. 1996, pp. 128-137.
Axial Tension,” doctoral thesis, the University of Stuttgart, Stuttgart, 21. Szerszen, M. M., and Nowak, A. S., “Calibration of Design Code for
Germany, 2004. (in German) Buildings (ACI 318): Part 2—Reliability Analysis and Resistance Factors,”
ACI Structural Journal, V. 100, No. 3, May-June 2003, pp. 383-391.

Title No. 119-S34
Fatigue Performance of Recycled Aggregate Concrete

Beams with Corroded Steel Reinforcement
by Guoqing Dong, Jin Wu, and Xing Zhao
This experimental research investigates the effects of corrosion on concrete-steel interface.16 For reinforced concrete beams
the fatigue performance of reinforced recycled aggregate concrete with low corrosion degree, the bond strength between rein-
(RAC) beams as well as the mechanical properties of corroded forcement and concrete increased due to the aggregation of
reinforcement in beams after fatigue loading. In addition, the corrosion products.17 Even if the mass loss is approximately
decreasing trend of fatigue life with rising corrosion degree and
5%, the flexure load-carrying capacity of the corroded rein-
the loss of fatigue life induced by steel corrosion for reinforced
forced concrete beam would increase. Despite the fact that
concrete beams are exhibited and analyzed, respectively. It was
found that greater fatigue stress amplitude and increased corrosion the generated corrosion products leads to an increase in the
degree tended to enlarge the maximum width of cracks and midspan bond between reinforcement and concrete, these increased
deflection, as well as aggravating the damage degree of reinforced weak bond forces cannot compensate for the loss of bond
RAC beams at failure. After corrosion fatigue, characteristics of strength due to steel corrosion. This is the main reason for
load-displacement curves for reinforcements changed significantly: the reduction in the strength of reinforced concrete beams
the yield strength and ultimate strength decreased, features of yield with low corrosion degree.18 From another perspective, the
plateau changed. The more the mass loss percentage of steel bars, loss of cross section of reinforcement not only results in the
the less the fatigue life of reinforced RAC beams. With the increase reduction of flexure load-carrying capacity and ductility
of corrosion degree, the reinforced RAC beams subject to large of reinforced concrete beams but also greatly changes the
fatigue stress amplitude shows a steeper decline in fatigue life than
failure behavior of the reinforcement itself. The literature
reinforced RAC beams subject to relatively small fatigue stress
shows that brittle failure of steel bar would occur when the
amplitude. Compared with reinforced natural aggregate concrete
(NAC), reinforced RAC suffers larger fatigue life loss and exhibits mass loss exceeded 12%.19 However, Yu et al.20 proposed
a steeper decline in fatigue life with rising corrosion degree. that, although corrosion causes the reinforcement to be more
brittle when stretched, the ductility of reinforced concrete
Keywords: fatigue life; fatigue stress amplitude; mass loss percentage; beams depends on the initial ductility of steel bar.
recycled aggregate concrete. Over the past ten decades, there is much research involving
the fatigue behavior of RAC, taking into account of the substi-
INTRODUCTION tution level of RCA. Regarding the condition of compres-
In the coastal areas, the corrosive environment with high sive fatigue loading, as the content of RCA increased, both
chloride ions content and the extensive use of deicing chlo- the fatigue life and residual strength decreased.21 Further-
ride salt in winter have accelerated the corrosion of rein- more, considering the influence of water-binder ratio (w/b)
forcements in concrete structures1,2 under the erosion of on strength development, Thomas et al.22,23 also found that
chloride ions. In addition to the aforementioned chemical RAC has a higher strength loss than NAC in the case of the
effect, fatigue load generated during vehicle driving also same water-cement ratio (w/c). In a word, the incorporation
endangers the service life of concrete structures. At present, of RCA would greatly weaken the ability of concrete to resist
the research scope of recycled aggregate concrete (RAC) cyclic deformation, and this was closely related to the low
is more and more extensive. If RAC is applied to coastal strength capacity of the aggregate. When the replacement
bridges or other structures in marine environment,3 it is ratio of RCA reached 100%, no obvious difference existed in
bound to be affected by the combination of fatigue load and fatigue behavior between NAC and RAC.24 However, from
reinforcement corrosion.4 the perspective of the Weibull distribution model established
Many scholars were devoted to the study of natural aggre- on the basis of fatigue life data,25 the variability of fatigue
gate concrete (NAC), taking into account the influence of life distribution of RAC with 100% substitution ratio was
pure fatigue load or pure steel corrosion, and a large number much greater than that of NAC. Additionally, combining
of research results were demonstrated regarding residual this variability with the fatigue life S-N curve and taking
flexure load-carrying capacity,5,6 ductile behavior,7,8 and the survival probability into account,26 the fatigue equations
proposed practical finite models.9-11 The existing literature of NAC and RAC corresponding to definite probability of
indicates that corrosion of reinforcements would change failure were established,27,28 which is of great significance
the failure mode of reinforced concrete beams. Under high for the fatigue life assessment of concrete structures.
degree of corrosion condition, brittle failure and bond split-
ting of reinforced concrete beams occurred.12,13 Researchers MS No. S-2021-057.R2, doi: 10.14359/51734333, received August 18, 2021, and
reached two consensus on the analysis of these phenomena. reviewed under Institute publication policies. Copyright © 2022, American Concrete
One is the loss of the cross section of the reinforcing bar14,15 obtained from the copyright proprietors. Pertinent discussion including author’s
and the other is the breakdown of bond strength at the closure, if any, will be published ten months from this journal’s date if the discussion

In addition, some research on the fatigue performance of fatigue behavior as well as the prediction and evaluation of
corroded reinforced concrete beams has been conducted. Sun fatigue life for in-service reinforced RAC bridge structures
et al.29 found that the flexural stiffness under the coupling of in the marine environment.
corrosion and repeated loading experienced two stages: the
decline of the flexural stiffness was slow at first and then RESEARCH SIGNIFICANCE
entered a stable stage. For the steel reinforcement after the In general, reinforced NAC structures exhibit degraded
combined action of corrosion and repeated loading, it can be mechanical performance under the combined action of
seen from stress-strain curves that its mechanical behavior corrosion and fatigue. With the aim of sustainable concrete
has changed significantly. Under the simultaneous action of development, if reinforced RAC is applied to coastal bridge
corrosion and sustained load, Hariche et al.30 found that the engineering, it will inevitably suffer the dual adverse effect
deflection of reinforced concrete beams increased continu- of corrosion and fatigue; this moment, whether reinforced
ously with the progressive mass loss of steel bars, especially RAC can fulfill strength and serviceability similarly to rein-
increasing rapidly and accompanied by a large number of forced NAC, it is not clearly and rarely involved in existing
cracks in the early stage of corroding steel. Similar conclu- literature. This paper mainly investigates several service
sions were also obtained by Yang et al.31 In general, the indexes of reinforced NAC and RAC beams, focusing on
fatigue performance of corroded reinforced concrete struc- analyzing the influence of steel corrosion and fatigue stress
tures largely depends on the fatigue performance of corroded magnitude on these service indexes. Through the compara-
reinforcements. The obvious rust pits appear on the rein- tive analysis of the theoretical results of the reinforced NAC
forcement surface of reinforced concrete structures in the beam and RAC beam, the obtained conclusion could lay a
corrosive ambient. Under fatigue load, the fatigue cracks solid foundation for future research, as well as effectively
initiate at the rust pits subject to large stress amplitude, then guiding the safe operation of in-service reinforced RAC
propagate unsteadily until reinforcement break. The earlier structures in the marine environment.
the fatigue crack initiation and the deeper the crack depth,
the shorter the fatigue failure process of concrete structures. EXPERIMENTAL PROGRAM
The effect of this action on RAC structures is more obvious Details and materials of specimens
than that on NAC structures. It is because that the external Sixteen reinforced concrete beams were fabricated and
load and the temperature change result in a large amount of divided into three groups. The first Group A included two
microcracks inside mortar attached to the surface of original beams: one uncorroded NAC beam was used as a control
aggregate. In addition, the load applied during the process of specimen, and the other RAC beam made with 100% content
crushing causes great damage to the interior of RCA. There- of RCA was used for static test to determine the flexure
fore, weak areas with the characteristics of high porosity load-carrying capacity. The second Group B and third Group
and low strength are formed in reinforced RAC structures, C, each group involves seven RAC beams with 100% RCA,
which provides more channels for chloride ions to invade among which, one beam was exposed to fatigue load without
the surface of reinforcements and further leads to severe corrosion, the other six beams were subjected to fatigue load
corrosion of reinforcements. The review of the existing after accelerated corrosion. The fatigue stress amplitude of
research shows that more information was available on the Group B beams was larger than that of Group C beams. All
performance of reinforced concrete beams subject to the the beams had the same dimensions and were designed with
coupled action of corrosion and sustained load, such as the a width of 200 mm (7.9 in.), a height of 300 mm (11.8 in.),
development of crack, changes in the stiffness of test beams and a total length of 2500 mm (98.5 in.). Two deformed
and the mechanical properties of steel bars, and the flexure steel bars (HRB400 hot-rolled crescent) with a diameter of
load-carrying capacity. 12 mm (0.5 in.) served as the longitudinal tensile reinforce-
The existing related literatures mostly focused on behavior ment. Two deformed steel bars (HRB400 hot-rolled cres-
of reinforced NAC beams subject to combined action of cent) with a diameter of 10 mm (0.4 in.) were used as the
corrosion and fatigue as well as reinforced RAC beams top hanger bars, and HRB400 hot-rolled crescent steel bars
subject to pure corrosion or pure fatigue. However, rela- with a diameter of 8 mm (0.3 in.) arranged every 150 mm
tively little research has dealt with the fatigue performance (5.9 in.) throughout the length of the beam were used as stir-
of corroded reinforced RAC beams. Furthermore, for the rups to prevent shear failure. Anti-rust paint was applied at
research on mechanical properties of reinforcements in RAC the contact position between the tensile reinforcements and
after corrosion fatigue, little research has been reported so stirrups to prevent stirrups from being corroded. Figure 1
far. In this paper, further efforts were made through accel- depicts the specific dimensions and the reinforcing details.
erated corrosion and fatigue loading test to research fatigue Mechanical properties of the main reinforcements obtained
behavior of corroded reinforced RAC beams. After failure in test are listed in Table 1.
of the beams, the steel bars were removed from the beams The employed cement was portland cement produced
to investigate the change in mechanical properties of the by a China cement factory, and its basic properties tested
corroded steel bars. The degradation law and loss of the according to GB 175-200732 are listed in Table 2. The RCA
fatigue life of reinforced RAC beams induced by steel corro- was originated from the crushed concrete from demolished
sion were presented and quantitatively analyzed, respec- buildings in Nanjing, and the relevant apparent density,
tively. Our findings are expected to provide the theoretical crushing index, and measured water absorption rate were
supports and practical technique methods for the research on 2356 kg/m3 (148 lb/ft3), 16.1%, and 4.7%, respectively.

Fig. 1—Details of specimens. (Note: Dimensions in mm; 1 mm = 0.0394 in.)
Table 1—Mechanical properties of reinforcement
Yield Ultimate
Diameter, strength, strength, Elongation,
Type mm (in.) MPa (ksi) MPa (ksi) %
Longitudinal tensile
12 (0.5) 547.0 (79.3) 644.5 (93.5) 22.8
reinforcement
Top hanger bars 10 (0.4) 553.2 (80.2) 632.5 (91.7) 23.1
Stirrup 8 (0.3) 559.8 (81.2) 621.1 (90) 23.5
Table 2—Basic properties of P.O 42.5 portland

Fig. 2—Schematic diagram of accelerated corrosion. cement
Item P.O 42.5
NCA was made from crushing basalt. The natural river sand
with fineness modulus of 2.78 was selected as fine aggregate Type Ordinary portland cement
(FA). The basic properties of RCA, NCA, and FA are listed Specific surface area, m /kg (ft /lb) 2 2
3720 (18,069)
in Table 3. The mixture proportion of concrete designed in Initial setting time, min 198
this study is listed in Table 4. Six 150 x 150 x 150 mm (5.9 x
Final setting time, min 248
5.9 x 5.9 in.) cubes were prepared when each test beam was
cast to determine the 28-day compressive strength of RAC. 3-day bending strength, MPa (ksi) 5.9 (0.9)
The measured average compressive strength of RAC was 3-day compressive strength, MPa (ksi) 31.5 (4.6)
29.5 MPa (4.28 ksi). 28-day bending strength, MPa (ksi) 8.8 (1.3)
28-day compressive strength, MPa (ksi) 52.8 (7.7)
Accelerated corrosion of steel
An accelerated corrosion technique was used to induce
Table 3—Basic properties of RCA, NCA, and FA
corrosion of longitudinal tensile reinforcements in beams
after curing for 28 days. The specific principle could be Type
further explained looking at Fig. 2. The bottom surfaces of Property RCA NCA FA
beams were covered with sponge, and 5% sodium chloride
Apparent density, kg/m (lb/ft ) 2356 (148) 2570 (161)
3 3
2629 (165)
solution was sprayed to retain the surface’s moisture. The
tensile steel bars were then connected with the anode of the Bulk density, kg/m (lb/ft )
3 3
1286 (80.8) 1330 (83.5) 1666 (104.7)
constant voltage and current power supply by wire, and the Crushing index, % 16.1 9.96 —
cathode was connected with the stainless-steel sheet placed Mud content, % 3.03 0.35 1.7
in the bottom of the beams. The imposed current density is
Water absorption, % 4.7 0.76 —
approximately 0.8 mA/cm2. Figure 3 shows the test sites of
accelerated corrosion of beams.
Table 4—Mixture proportion of concrete
According to Faraday’s law, the expected mass loss
percentage of steel (ρ) can be obtained by controlling Type
corrosion duration (t) and current intensity (i). The rela- Property RAC NAC
tionship between power-on time (t) and expected mass loss Replacement ratio of RCA, % 100 0
percentage stated by Faraday’s law could be expressed by
Cement, kg/m3 (lb/ft3) 412 (25.9) 412 (25.9)
Eq. (1)
FA, kg/m (lb/ft )
3 3
584 (36.6) 584 (36.6)
Nem RCA, kg/m (lb/ft )
* 3 3
1030 (64.6) 0 (0)
t (1)
3600 M i r NCA, kg/m3 (lb/ft3) 0 (0) 1030 (64.6)
where N is the Avogadro constant (6.02 × 1023 mol–1); e is Water, kg/m3 (lb/ft3) 185 (11.6) 185 (11.6)
the quantity of electric charge (1.6 × 10–19 C); m is the linear *
Saturated surface-dry moisture state.
mass density of steel (the measured linear mass density of
12 mm [0.5 in.] HRB400 was 0.838 g/mm [0.047 lb/in.]); M After the fatigue failure of each beam, two longitudinal
is the molar amount of electrolytic steel (56 g/mol [0.1 lb/ tensile reinforcements were removed and then cut at each
mol]); and r is the radius of steel (6 mm [0.2 in.]). section of 300 mm (11.8 in.), as shown in Fig. 4. Three

Fig. 3—Schematic diagram of accelerated corrosion of beams.
Fig. 4—Schematic diagram of sampling corroded reinforcement. (Note: Dimensions in mm; 1 mm = 0.0394 in.)
Table 5—Mass loss percentage and fatigue life of tested beams
Expected mass loss Measured mass loss
Beam No. Stress range ∆σ, MPa (ksi) Corrosion duration, h percentage, % percentage, % Fatigue life N, 104 cycles
NAC 312 (45.2) 0 0 0.00 45.1
B0 312 (45.2) 0 0 0.00 43.1
B1 312 (45.2) 48 2 2.10 41.7
B2 312 (45.2) 96 4 4.10 35
B3 312 (45.2) 144 6 5.88 34.3
B4 312 (45.2) 192 8 10.58 6.7
B5 312 (45.2) 240 10 18.02 2.8
B6 312 (45.2) 288 12 23.20 0.3
C0 240 (34.8) 0 0 0.00 163.1
C1 240 (34.8) 48 2 4.48 115.6
C2 240 (34.8) 96 4 7.26 78.2
C3 240 (34.8) 144 6 9.30 60.1
C4 240 (34.8) 192 8 11.91 58.6
C5 240 (34.8) 240 10 15.40 32.6
C6 240 (34.8) 288 12 22.30 5.8
Note: B and C represent Group B test beams and Group C test beams, respectively. 0-6 indicates that the beams are ordered according to corrosion duration in each Group.
sections of steel from one longitudinal tensile reinforcement where m1 and m2 denote the mass of corroded bars after
were subjected to static tensile test to study its mechanical acid solution cleaning and the mass of uncorroded bars,
properties under the coupling of corrosion and fatigue load. respectively.
Six sections of steel from another were cleaned with 12%
hydrochloric acid solution to remove attaching rust products, Fatigue test
then the measured mass loss percentage of each segmented After the accelerated corrosion, the fatigue test was carried
reinforcement was calculated according to the method out by MTS hydraulic servo loading system, as shown in
proposed in GB/T 50082,33 which could be expressed by Fig. 5. Auxiliary apparatus at both ends of each beam were
Eq. (2). The average measured value of six steel samples used to prevent shifting of the beams on the supports during
was regarded as the measured mass loss percentage (ƞ) of cycling. Three displacement gauges, two of which were
the corroded beam. Table 5 lists the mass loss percentage placed on the supports to measure their settlement, and the
and corresponding corrosion duration. other was used to measure the midspan deflection of the
beam. Repeated loading with constant amplitude acted on
m2 m1 the midspan of beam. The fatigue load was performed using
100% (2)
m2 a sinusoidal wave with a frequency of 4 Hz.

The calculated ultimate flexure load-carrying capacity of
the beam under single-point midspan loading was 53.54 kN
(12 kip). For reinforced Groups A and B beams, the upper
limit fatigue load and the lower limit fatigue load were 38.5
and 9.2 kN (8.7 and 2.1 kip), respectively, and the corre-
sponding calculated maximum and minimum fatigue stress
in tensile steel were 371.2 and 59.4 MPa (53.8 and 8.6 ksi),
respectively. The stress range of tensile bars was 312 MPa
(45.2 ksi). With respect to reinforced Group C beams, the
applied maximum and minimum fatigue load were 30.4 and
8.0 kN (6.8 and 1.8 kip), respectively. The corresponding
upper limit fatigue stress for tensile reinforcements was
285.7 MPa (41.4 ksi), the lower limit fatigue stress was
45.7 MPa (6.7 ksi), and the stress range was 240 MPa Fig. 5—Fatigue test setup.
(34.8 ksi).
Before the start of fatigue test, static load starting from
0 kN (0 kip) to the upper limit fatigue load was pre-loaded,
ensuring good contact between the beam and the test setup
in normal working state. Then, unloaded to 0 kN (0 kip)
and cyclic load was applied. When the repeated cycles
reached to 1 × 104, 5 × 104, 10 × 104, 20 × 104, 50 × 104,
100 × 104, 150 × 104, and 200 × 104, respectively, fatigue
test suspended. Then the monotonic loading static test was
conducted from 0 kN (0 kip) to the maximum load of the
upper fatigue load under a step loading process.12,29,34,41 At
each step of static loading, maximum width of cracks was
measured by a concrete crack width observation instrument
and data measured in displacement gauges were transferred
to the computer and saved. The reinforcement fractured,
concrete in compression zone crushed, the maximum crack
width at the tensile longitudinal steel bar reached 1.5 mm
(0.1 in.) and the midspan deflection exceeded 1/50 of the
beam span; if one of the aforementioned phenomena
occurred, it indicates the fatigue failure of the beam. If there
is no fatigue failure after 200 × 104 cycle, the static load was
applied to failure.
EXPERIMENTAL RESULTS AND DISCUSSIONS Fig. 6—Corroded reinforcing bars.

Mass loss percentage and mechanical properties along the circumference of the cross section of the steel bar
of steel bar after corrosion fatigue exhibited a large difference, and the steel bar adjacent to the
As illustrated in Table 5, the mass loss percentage of steel concrete protective layer was seriously corroded.
bar increases with the corrosion duration. With the increase After corrosion fatigue failure of the test beams, in accor-
of corrosion duration, the passive film shaped on the surface dance with the position identified in Fig. 4, the removed
of the steel bar was seriously damaged. Under the combined tensile reinforcements were cut into segments 1, 2, and
action of electrolysis and oxidation, a large amount of gener- 3, which were used for static tensile test. Figure 7 shows
ated corrosion products accumulated on the surface of the the load-displacement curves for the segmented reinforce-
steel bar, which greatly decreased the mass of the steel bar. It ments sampled from the Group B beams after corrosion
is noteworthy that the mass loss percentage of the steel bars fatigue. Under the mode of single point midspan loading,
embedded in two reinforced RAC beams is indeed different, the reinforcements at different positions along the length
when these two beams suffer the same corrosion duration. of the beam underwent various fatigue stress amplitude;
This is because the internal defects inside the reinforced therefore, the segmented reinforcements within the same
RAC beams with 100% replacement rate are significantly beam would inevitably endure various degree of fatigue
different, which generates a great impact on the distribution damage. As shown in Fig. 7(a), for uncorroded reinforce-
of concrete protective layer as resistance and further induces ments embedded in RAC beam B0, under the same number
the uneven corrosion of steel bar. As shown in Fig. 6, from of fatigue cycles, steel bar segment 1 bore the larger fatigue
the appearance of the corroded steel bar samples, uneven stress amplitude compared with the segment 2 and segment
corrosion as well as apparent local corrosion and pit corro- 3, it can be concluded that the increased amplitude of fatigue
sion along the length of the longitudinal tensile reinforce- stress decreased the yield plateau, and the final displacement
ment could be observed. In addition, the corrosion degree representing the ultimate strain was reduced. Regarding the

Fig. 7—Load-displacement curves for segmented reinforcements embedded in reinforced RAC beams. (Note: 1 mm = 0.0394 in.;
1 kN = 224.82 lbf.)
corroded reinforced RAC beams, similar conclusion could RAC beams, both the yield strength and ultimate strength
be obtained from the load-displacement curves depicted in decreased nonlinearly with the mass loss percentage. The
each figure. increased mass loss percentage brought about the inho-
Although each corroded reinforced RAC beam in Group mogeneous corrosion of steel bars, the presence of a large
B suffers the same fatigue stress amplitude, the number of number of rust pits will minimize the cross-sectional area of
fatigue cycles for each beam after corrosion fatigue was reinforcements and further induced stress concentration. In
different due to the various mass loss percentage of rein- addition, the occurrence and accumulation of fatigue crack
forcement. Figures 7(b) through (g) show that, under the will aggravate the fatigue damage within reinforcements.
combined action of corrosion and fatigue load, increased For the steel bars at different segment positions in the same
mass loss percentage can significantly change the feature for reinforced RAC beam, it can be observed that the increased
the yield plateau of steel bar. The yield plateau can be clearly mass loss percentage contributed to the large difference in
observed from the load-displacement curves of RAC beams both yield strength and ultimate strength—that is to say, the
with mass loss percentage of 0.00% and 2.10%. As the mass larger the mass loss percentage, the greater the yield strength
loss percentage increased, the yield plateau began to tilt, and ultimate strength loss of the steel bars that underwent the
shorten, and then disappear. The load-displacement curves largest amplitude of fatigue stress in the same beam. There-
of reinforcement segment 1, 2, and 3 exhibits the same fore, the mass loss percentage and the amplitude of fatigue
mechanical behavior. The aforementioned finding confirms stress were responsible for the reduction in the mechanical
that the coupled action of higher corrosion level and fatigue properties of the steel bars after corrosion fatigue.
stress amplitude significantly changes the mechanical prop-
erties of steel bars in RAC beams. Cracks due to corrosion
Figure 8 shows the yield strength and ultimate strength Figure 9 shows the corrosion-induced cracks at the bottom
of segmented reinforcement after corrosion fatigue. For the surface of reinforced RAC beams after accelerated corro-
steel bars at the same segment position in different reinforced sion. Both reinforced Groups B and C beams show similar

Fig. 8—Strength degradation of segmented reinforcements with various mass loss percentage after fatigue load. (Note: 1 MPa
= 145 psi.)
Fig. 9—Diagram of corrosion-induced cracks at bottom surface of reinforced RAC beams.

crack distribution with the increase of mass loss percentage.
Regarding the reinforced RAC beams B1, B2, and C1 with
small mass loss percentage, merely corrosion-induced cracks
along the longitudinal direction of the tensile steel bars
appeared at the bottom surface and there was no connection
between them. As the mass loss percentage increased, a small
amount of transverse corrosion-induced cracks emerged.
The larger the mass loss percentage, the more uneven the
distribution of transverse corrosion-induced cracks. For
reinforced RAC beams B6 and C6 subject to severe corro-
sion, the occurrence of transverse corrosion-induced cracks
intersected with longitudinal corrosion-induced cracks can
be observed at the bottom surface, especially noticeable
with a large number of corrosion-induced cracks on the side
surface of the beams. With the continual accumulation of
corrosion products, the bond strength formed at the inter-
face between reinforcement and concrete will be greatly Fig. 10—Mass loss percentage versus maximum width of
reduced, and microcracks in ITZ around reinforcements corrosion-induced crack. (Note: 1 mm = 0.0394 in.)
were extremely easy to broaden under the action of the
expansion force. The maximum width of corrosion-induced Maximum width of cracks due to fatigue load
crack versus the mass loss percentage of steel bar is shown During the process of applying the fatigue load, load-in-
in Fig. 10. It can be seen that there is a good linear correla- duced cracks started from the bottom side of the beams
tion between the maximum width of the corrosion-induced and propagated along the stirrup in the vertical direction.
crack and the mass loss percentage. The more the mass loss Figure 1 shows the maximum width of these transverse
percentage of steel reinforcement, the greater the maximum load-induced cracks when the number of fatigue load cycles
width of corrosion-induced crack of the beam, which further reached the expected values. Maximum width of these
demonstrates that the width of corrosion-induced crack is transverse cracks appeared on the surface of all test beams
greatly affected by the corrosion damage. presents an increasing trend with the continual increase of
fatigue cycles. It can be found by comparing Fig. 11(a) with
(b) that, despite the fact that the reinforced NAC and RAC

Fig. 11—Maximum crack width after expected fatigue cycles. (Note: 1 mm = 0.0394 in.; 1 kN = 224.82 lbf.)

beam were susceptible to the same level of fatigue stress fatigue stress amplitude increases the crack width of rein-
and static load, two main differences are demonstrated: after forced RAC after expected cycles, which can be attributed to
each number of expected fatigue cycles, the reinforced RAC the deteriorated bond strength and greater strain in concrete
beam shows larger maximum crack width than reinforced and reinforcement.
NAC beam, as well as the steeper slope of maximum crack
width increasing with the static load level. These differences Midspan deflection
are attributed to the lower tensile splitting strength of RAC Figure 12 shows the measured midspan deflection of the
and the poor properties of RCA.35 Generally, the interface beams when the number of fatigue load cycles reached the
between RCA and the old mortar as well as the interface expected values. Under the initial loading, due to the concrete
between RCA and the new mortar are loose and porous, in the tension zone at the bottom of the beams cracked
which becomes the weakest zone inside RAC beam. Under laterally and withdrew from work, the midspan deflection
monotonic load or fatigue load, microcracks in ITZ will increased rapidly and the flexural stiffness was obviously
widen and intersect with the cracks within crushed RCA with weakened. It can be seen by contrasting Fig. 12(a) and (b)
high crushing value due to its high porosity. As expected, the that, after the same expected fatigue cycles, the midspan
width of cracks on the surface of RAC beam grows largely. deflection of reinforced RAC beam were larger than that of
By comparing the maximum crack width of Beam B1 the NAC beam. Due to the lower elastic modulus of recy-
with that of Beam B0, as well as the maximum crack width cled concrete compared to that of normal concrete, a larger
of Beam C1 with that of Beam C0, it is observed that the deformation of recycled concrete under fatigue load results
maximum crack width of less-corroded reinforced RAC in larger deflection.
beam is smaller than that of uncorroded reinforced RAC Comparing the reinforced RAC beams B0 with B1, or the
beam. As is well known, the cracking behavior of concrete, beams C0 with C1 subject to low fatigue stress amplitude, it
namely the crack width, is closely related to the strain of can be observed that for the RAC beam with a little increase
concrete around the tensile reinforcement. When the mass in mass loss percentage of embedded steel bars, the midspan
loss percentage is small, corrosion products attached deflection reduced and the flexural stiffness increased.
to the surface of reinforcement can fill the voids inside Owing to the voids of RAC around the reinforcement filled
surrounding concrete, making the cement matrix around the with a certain amount of corrosion products, the existence of
RCA more dense. Moreover, the volume expansion force rust layer bonds the steel bars with the surrounding concrete
generated by the accumulation of a small amount of corro- tightly together to resist deformation under cyclic fatigue
sion products will not lead to cracking or even spalling of the load. Besides, it should be noted that in Fig. 12(f), during the
surrounding concrete; it will exert radial expansion pressure range from 10,000 to 50,000 cycles, abnormal reduction in
on the steel-concrete interface, which can increase the bond the midspan deflection of the RAC beam B2 appeared. This
strength between reinforcement and concrete to a certain may be attributed to the fact that the behavior of corrosion
extent, further restraining the deformation of concrete under products creeping and migrating to cracking location under
fatigue load. cyclic loading could induce the closure of cracks. There-
With respect to the maximum crack width of corroded fore, the mass loss percentage of steel bars within a certain
reinforced RAC beams, it can be stated that the increased range is beneficial to decrease the flexural stiffness loss for
mass loss percentage induces obvious change, those are the RAC beams under fatigue load. To obtain the aforemen-
larger maximum crack width and its greater growth trend. tioned specific range of mass loss percentage and the change
For example, compared to the maximum crack width of of corresponding midspan deflection, it is inevitable to carry
Beam B2 after 10,000 cycles, increases of 13%, 47%, and out fatigue tests on a large number of corroded test beams.
67% have been noticed for Beams B3, B4, and B5, respec- With respect to the reinforced RAC beams B4, B5, and
tively. Similarly, compared to the maximum crack width C6 subject to severe corrosion, their midspan deflection
of Beam C2 after 200,000 cycles, increases of 30%, 34%, under the initial loading was relatively large and increased
and 117% have been noticed for Beams C3, C4, and C5, by a large margin when the fatigue cycles was increased to
respectively. This clearly indicates that the increase of 50,000. This is because a large amount of corrosion prod-
reinforcement corrosion has a marked effect on reinforced ucts were generated, and the cracking of RAC was induced
RAC beam’s cracking behavior. The reason for this is that by its volume expansion. Additionally, although a thick rust
under the condition of high mass loss percentage, a large layer exists between the steel bars and the RAC due to the
amount of corrosion products accumulated on the steel-con- severe corrosion, the bond strength and the friction between
crete interface makes the reinforcement lose the restriction the steel bars and the RAC were greatly reduced as a result
on the concrete, and the elastic modulus of RAC is inher- of the rust layer stratified under cyclic load.
ently low; these directly lead to the large strain of RAC on
the bottom of the beams under fatigue load. Furthermore, Fatigue failure mode
as aforementioned, transverse corrosion-induced cracks The fatigue failure modes of reinforced Groups B and C
appeared on the bottom surface of the severe corroded RAC RAC beams with various mass loss percentage were shown
beams; these cracks intersect with the flexural cracks, which in Fig. 13 and Fig. 14, respectively. All reinforced RAC
generates influence on the crack development. For Beams beams exhibited typical flexural failure of normal section; it
B4 and C4, B6 and C6 with similar mass loss percentage, was marked by the fracture of the steel bars near the loading
it can be also found through comparing that the increase of point and the crushing of concrete in the compression zone

Fig. 12—Midspan deflection of beams. (Note: 1 mm = 0.0394 in.; 1 kN = 224.82 lbf.)

Fig. 13—Fatigue failure modes of reinforced Group B RAC beams.
Fig. 14—Fatigue failure modes of reinforced Group C RAC beams.

once the cycles of fatigue load reached the fatigue life. The in the form of blocks, as shown in Fig. 13(g) and Fig. 14(g).
sound of steel bars fracturing with a snap and the squeaking The increased degree of corrosion induced the accumulation
sound of concrete being crushed were clearly heard, which of a substantial amount of corrosion products at the inter-
occurred without any warning. Meanwhile, a main crack in face between reinforcements and concrete. On the one hand,
the midspan expanded rapidly and ran through the entire the volume expansion of corrosion products caused the
section of the beam. surrounding concrete to crack. On the other hand, the bond
The reinforcement embedded in uncorroded and less strength between reinforcements and concrete was greatly
corroded RAC beams could cooperate with surrounding reduced, which further destroys the mechanism of the joint
concrete well to resist fatigue load, noting that a relatively combination of steel bar and concrete to resist fatigue load.
complete crack was clearly observed from the bottom surface Additionally, when the mass loss percentage was similar, it
of the reinforced RAC beams B0, B1, C0 and C1 after can be observed by contrasting Beams B3 and C2, B4, and
corrosion fatigue. As the mass loss percentage increased, C3 that, reinforced Group B RAC beams subject to higher
the failure mode was marked by the appearance of a large amplitude of fatigue stress were more severely damaged
number of cracks with small width near the main crack and compared with the reinforced Group C RAC beams. This
mutual connection. With the further increase in mass loss finding clearly demonstrates that increased fatigue stress
percentage, the failure mode was also accompanied by the amplitude will cause more serious damage to reinforced
phenomenon that the concrete in the tension zone peeled off RAC beams.

Fig. 15—Fatigue life versus mass loss percentage. (Note: 1 kN = 224.82 lbf.)
Fatigue life degenerated, coupling effects of three factors contributes
Fatigue life influenced by corrosion mass loss—As to the reduction in fatigue life. As the mass loss percentage
expected, reinforced NAC beam presents larger fatigue life increased further, depth of rust pits rises slowly and the bond
compared to the Beam B0 subject to the same fatigue stress strength became stable, whereas the cross-sectional area of
amplitude. With respect to the comparison of fatigue life of reinforcement was weakened significantly, which leads to a
reinforced RAC beams with various mass loss percentage in considerable reduction in fatigue life.
Group B or Group C, the greater the mass loss percentage With reference to the comparison of fatigue life between
of steel bars, the lesser the fatigue life. For the severely reinforced Groups B and C RAC beams, it can be noticed
corroded reinforced beams B5 and B6, the fatigue lives are from Fig. 15 that corroded reinforced RAC beams exposed
28,000 and 3000 cycles, which are only 6.5% and 0.7% of to larger fatigue stress amplitude has shorter fatigue life.
that of the uncorroded Beam B0. Similarly, when the mass This is because that with the increase of fatigue cycles, the
loss percentage increased from 0 to 15.4% and 22.30%, the stress of tensile reinforcement in beams subject to larger
fatigue life of the beams in Group C decreased by 80% and fatigue stress amplitude increased rapidly. Note that with
96%, respectively. This means that corrosion significantly the increase of mass loss percentage, the fatigue life of
reduces the fatigue life of reinforced RAC beams. corroded Group B beams drops steeper compared with that
Previous research suggests that under the same level of of corroded Group C beams, indicating that corrosion shows
fatigue load, the fatigue life of corroded reinforced NAC a greater influence on the fatigue life of RAC beams subject
beam exhibits obvious sensitivity to the corrosion degree to high fatigue stress amplitude.
of reinforcement.12,36,37 Figure 15 shows the relationship
between the fatigue cycles at failure and the mass loss Comparison between NAC and RAC
percentage for reinforced Groups B and C RAC beams. Available results of numerous research on the fatigue
It can be observed that the change in fatigue life of rein- behavior of corroded reinforced NAC beams indicated that
forced RAC beams follows the same trend as that of rein- an increase in mass loss percentage decrease the fatigue
forced NAC beams. Under dual adverse effects of corrosion life. The fatigue life of reinforced RAC beams in this study
and cyclic load, the occurrence of fatigue cracks results also exhibits a comparable tendency with the mass loss
in decrease of ultimate strain, yield strength, and ultimate percentage. From the perspective of resource conservation,
strength of reinforcing bars, as well as the loss of ductility. it is very meaningful to examine whether the fatigue life of
The higher the corrosion level, the greater the yield strength reinforced RAC structures is identical or similar to that of
and ultimate strength loss of steel bars that underwent the reinforced NAC structures, when reinforced RAC struc-
largest amplitude of fatigue stress in the midspan. Further- tures works in corrosion environment instead of reinforced
more, the subsequent three critical factors, namely the cross- NAC structures.
sectional area of reinforcement, the number and depth of rust With the purpose of quantitatively investigating the fatigue
pit together with the bond strength between the reinforce- life loss induced by steel corrosion, the defined coefficient of
ment and surrounding concrete, were responsible for the fatigue life reduction (βf) can be expressed as
reduction in fatigue life of reinforced concrete beams. When
the mass loss percentage was small, despite the fact that the βf = Ne/Nc (3)
limited amount of corrosion products gives rise to the bond
strength between reinforcement and surrounding concrete,38 where Ne is the experimental fatigue life of concrete beams
along with an insignificant reduction in cross-sectional area subject to the combined action of corrosion and fatigue; and
of reinforcement, the existence of rust pits and the induced Nc is the calculated fatigue life of concrete beams exposed
stress concentration under fatigue load are the main reasons to pure fatigue.
for the decrease of fatigue life.39 With the incessant increase Fatigue test results of concrete beams were generally put
of mass loss percentage, cross-sectional area of reinforce- forward in the form of the relationship between the fatigue
ment was far more reduced, the number and the depth of life and the employed fatigue stress amplitude (∆σ). Based
rust pits were all increased as well as the bond strength on the current test results,12,40-44 the experimental fatigue

life of corroded NAC beams was already known, as well
as the S-N curve and fitting formula of NAC beams after
pure fatigue depicted in Fig. 16 were obtained. Submit-
ting the fatigue stress amplitude applied on corroded NAC
beams into the fitting formula, the calculated fatigue life was
acquired. Regarding the fatigue life reduction coefficient of
RAC beams, it was acquired by the ratio of the experimental
fatigue life of corroded beams to that of uncorroded beam
in each Group RAC beams. Figure 17 shows the fatigue life
reduction coefficient of reinforced NAC and RAC beams
versus the mass loss percentage.
A distinct difference was clearly observed between the
impact of corrosion on fatigue life of reinforced NAC and
the impact of corrosion on fatigue life of reinforced RAC. As
shown in Fig. 17, with the increase of mass loss percentage,
reinforced RAC presents the fatigue life reduction coef-
ficient smaller and decremental slope larger than those of Fig. 16—S-N relationship of uncorroded NAC beams.
reinforced NAC. These findings indicate that compared
with reinforced NAC, reinforced RAC at the same corrosive
environment suffers larger fatigue life loss and its fatigue life
decreases more rapidly with mass loss percentage.
It is widely known that, because of the poor properties
of RCA, high porosity and large water absorption capacity
result in RCA more susceptible to erosion induced by chlo-
ride ions; this drawback is particularly obvious in rein-
forced RAC made with 100% content of RCA. Compared
with reinforced NAC, chloride solution is more effortlessly
close to the surface of the steel bars embedded in reinforced
RAC, inducing the large-scale corrosion of reinforce-
ment. The measured mass loss percentage is essentially the
average mass loss of steel bars, which cannot truly reflect the
following aspects, including the cross section loss of rein-
forcement, the deviatoric stress induced by uneven corrosion
in the same section, and the depth of rust pits. Even though
Fig. 17—Fatigue life reduction coefficient versus mass loss
reinforced RAC and NAC have the same measured mass
percentage.
loss percentage, in fact, the local corrosion degree of rein-
forcement in reinforced RAC is undoubtedly more serious 2. When the mass loss percentage is small, merely corro-
than that of reinforced NAC. Besides, the ITZ between RCA sion-induced cracks along the longitudinal direction of the
and new mortar and origin mortar is relatively loose and tensile steel bars appeared. With the increase of mass loss
porous, a large amount of existing microcracks will widen percentage, unevenly distributed transverse corrosion-in-
and interconnect under the action of fatigue load. Moreover, duced cracks emerged and intersected with longitudinal
the bond strength between reinforcement and surrounding corrosion-induced cracks.
concrete is greatly deteriorated due to the low tensile split- 3. Under fatigue load, reinforced RAC beam showed
ting strength and compressive strength of recycled concrete. larger maximum crack width and midspan deflection than
These adverse factors will lead to the premature shedding of reinforced NAC beam. Compared with the uncorroded rein-
reinforced RAC cover compared to reinforced NAC. forced RAC beam, the maximum crack width of less-cor-
roded reinforced RAC beam was smaller, and the midspan
CONCLUSIONS deflection reduced and the flexural stiffness increased. As the
Based on the experimental results, the following findings mass loss percentage increased, the maximum crack width
and conclusions can be drawn: and midspan deflection became larger and their growth trend
1. For uncorroded or corroded reinforcements subject to was greater.
cycling loading, the increased amplitude of fatigue stress 4. All reinforced RAC beams exhibited typical flexural
would decrease the yield plateau and ultimate strain. Corro- failure of normal section, which was marked by the fracture
sion effect has substantial influence on mechanical properties of the steel bars near the loading point and the crushing of
of reinforcements subject to fatigue loading. After corrosion concrete in the compression zone. From the appearance of
fatigue, characteristics of load-displacement curves for steel the damaged RAC beams, it can be seen that a relatively
bars changed significantly: the yield strength and ultimate complete crack in the midspan of less corroded beam, a
strength decreased, features of yield plateau changed. large amount of cracks intersected in the midspan of medium

corroded beam and concrete spalling in the midspan of Cohesive Surface Bonding Approach,” Computers and Concrete, South
Korea, V. 22, No. 2, Aug. 2018, pp. 167-182.
severely corroded beam. 11. Du, Y. G.; Chan, A. H. C.; Clark, L. A.; Wang, X. T.; Gurkalo, F.;
The more the mass loss percentage of steel bar, the less the and Bartos, S., “Finite Element Analysis of Cracking and Delamination of
fatigue life of reinforced RAC beams. With the increase of Concrete Beam Due to Steel Corrosion,” Engineering Structures, England,
V. 56, Nov, 2013, pp. 8-21. doi: 10.1016/j.engstruct.2013.04.005
corrosion degree, the reinforced RAC beams subject to large 12. Yi, W. J.; Kunnath, S. K.; Sun, X. D.; Shi, C. J.; and Tang, F. J.,
fatigue stress amplitude shows a steeper decline in fatigue “Fatigue Behavior of Reinforced Concrete Beams with Corroded Steel
life than reinforced RAC beams subject to relatively small Reinforcement,” ACI Structural Journal, V. 107, No. 5, Sept.-Oct. 2010,
pp. 526-533.
fatigue stress amplitude. Compared with reinforced NAC, 13. Yoon, S.; Wang, K. J.; and Weiss, J.; andShah, S. P., “Interaction
reinforced RAC suffers larger fatigue life loss and exhibits between Loading, Corrosion, and Serviceability of Reinforced Concrete,”
a steeper decline in fatigue life with rising corrosion degree. ACI Structural Journal, V. 97, No. 6, Nov.-Dec. 2000, pp. 637-644.
14. Oyado, M.; Kanakubo, T.; Sato, T.; and Yamamoto, Y., “Bending
Performance of Reinforced Concrete Member Deteriorated by Corrosion,”
AUTHOR BIOS Structure and Infrastructure Engineering, England, V. 7, No. 1-2, 2011,
Guoqing Dong is a PhD Student in the Department of Civil Engineering pp. 121-130. doi: 10.1080/15732471003588510
at Nanjing University of Aeronautics and Astronautics, Nanjing, China. 15. Castel, A.; Francois, R.; and Arliguie, G., “Mechanical Behaviour
He received his BS from Nanjing Tech University, Nanjing, China, and his of Corroded Reinforced Concrete Beams-Part 1: Experimental Study of
MS from Nanjing University of Aeronautics and Astronautics. His research Corroded Beams,” Materials and Structures, Netherlands, V. 33, No. 233,
interests include durability of concrete. 2000, pp. 539-544. doi: 10.1007/BF02480533
16. Mangat, P. S., and Elgarf, M. S., “Flexural Strength of Concrete
Jin Wu is a Professor in the Department of Civil and Airport Engineering Beams with Corroding Reinforcement,” ACI Structural Journal, V. 96,
at Nanjing University of Aeronautics and Astronautics. He received his No. 1, Jan.-Feb. 1999, pp. 149-158.
PhD from Hohai University, Nanjing, China. His research interests include 17. Fang, C. Q.; Yang, S.; and Zhang, Z., “Bending Characteristics of
intelligent monitoring on the durability of engineering structure, recycled Corroded Reinforced Concrete Beam under Repeated Loading,” Structural
aggregate concrete, and recycled concrete structures. Engineering and Mechanics, South Korea, V. 47, No. 6, 2013, pp. 773-790.
doi: 10.12989/sem.2013.47.6.773
Xing Zhao is a Lecturer in the Department of Civil and Airport Engineering 18. Jnaid, F., and Aboutaha, R. S., “Residual Flexural Strength of
at Nanjing University of Aeronautics and Astronautics. She received her Corroded Reinforced Concrete Beams,” Engineering Structures, England,
PhD from Southeast University, Nanjing, China. Her research interests V. 119, July 2016, pp. 198-216. doi: 10.1016/j.engstruct.2016.04.018
include fatigue properties of fiber-reinforced composites under multifield 19. Almusallam, A. A., “Effect of Degree of Corrosion on the Properties
coupling and finite element methodology. of Reinforcing Steel Bars,” Construction and Building Materials, England,
V. 15, No. 8, 2001, pp. 361-368. doi: 10.1016/S0950-0618(01)00009-5
ACKNOWLEDGMENTS 20. Yu, L. W.; Francois, R.; Dang, V. H.; and Hostis, L. V.; and ,
The work was carried out with the financial support from Transporta- “Gagne, R., “Structural Performance of RC Beams Damaged by Natural
tion Science and Technology Program of Jiangsu Provincial Department of Corrosion under Sustained Loading in a Chloride Environment,” Engi-
Water Resources (2016Y22). neering Structures, England, V. 96, Aug. 2015, pp. 30-40. doi: 10.1016/j.
engstruct.2015.04.001
21. Peng, Q. M.; Wang, L.; and Lu, Q., “Influence of Recycled Aggregate
REFERENCES Replacement Percentage on Fatigue Performance of Recycled Aggregate
1. Thomas, M., “Chloride Thresholds in Marine Concrete,” Cement Concrete,” Construction and Building Materials, England, V. 169, Apr.
and Concrete Research, England, V. 26, No. 4, 1996, pp. 513-519. doi: 2018, pp. 347-353. doi: 10.1016/j.conbuildmat.2018.02.196
10.1016/0008-8846(96)00035-X 22. Thomas, C.; Setien, J.; Polanco, J. A.; Lombillo, I.; and Cimentada,
2. Angst, U.; Elsener, B.; Larsen, C. K.; and Vennesland, Ø., “Critical A., “Fatigue Limit of Recycled Aggregate Concrete,” Construction and
Chloride Content in Reinforced Concrete-A Review,” Cement and Concrete Building Materials, England, V. 52, Feb. 2014, pp. 146-154. doi: 10.1016/j.
Research, England, V. 39, No. 12, 2009, pp. 1122-1138. doi: 10.1016/j. conbuildmat.2013.11.032
cemconres.2009.08.006 23. Thomas, C.; Sosa, I.; Setien, J.; Polanco, J. A.; and Cimentada, A.
3. Liu, K. H.; Yan, J. C.; Alam, M. S.; and Zou, C. Y., “Seismic Fragility I., “Evaluation of the Fatigue Behavior of Recycled Aggregate Concrete,”
Analysis of Deteriorating Recycled Aggregate Concrete Bridge Columns Journal of Cleaner Production, V. 65, Feb. 2014, pp. 397-405. doi:
Subjected to Freeze-Thaw Cycles,” Engineering Structures, England, 10.1016/j.jclepro.2013.09.036
V. 187, No. 12, 2019, pp. 1-15. doi: 10.1016/j.engstruct.2019.01.134 24. Xiao, J. Z.; Li, H.; and Yang, Z. J., “Fatigue Behavior of Recycled
4. Aldaca, I. C. U.; Briz, E.; Biezma, M. V.; and Puente, I., “A Plain Linear Aggregate Concrete under Compression and Bending Cyclic Loading,”
Rule for Fatigue Analysis under Natural Loading Considering the Coupled Construction and Building Materials, England, V. 38, Jan. 2013,
Fatigue and Corrosion Effect,” International Journal of Fatigue, England, pp. 681-688. doi: 10.1016/j.conbuildmat.2012.09.024
V. 122, No. 12, 2019, pp. 141-151. doi: 10.1016/j.ijfatigue.2019.01.008 25. Arora, S., and Singh, S. P., “Analysis of Flexural Fatigue Failure
5. Torres-Acosta, A. A.; Navarro-Gutierrez, S.; and Teran-Guillen, J., of Concrete Made with 100% Coarse Recycled Concrete Aggregates,”
“Residual Flexure Capacity of Corroded Reinforced Concrete Beams,” Construction and Building Materials, England, V. 102, Jan. 2016,
Engineering Structures, England, V. 29, No. 6, 2007, pp. 1145-1152. doi: pp. 782-791. doi: 10.1016/j.conbuildmat.2015.10.098
10.1016/j.engstruct.2006.07.018 26. Murali, G.; Indhumathi, T.; Karthikeyan, K.; and Ramkumar, V. R.,
6. Chen, H. P., “Residual Flexural Capacity and Performance Assess- “Analysis of Flexural Fatigue Failure of Concrete Made with 100% Coarse
ment of Corroded Reinforced Concrete Beams,” Journal of Structural Recycled and Natural Aggregates,” Computers and Concrete, South Korea,
Engineering, ASCE, V. 144, No. 12, 2018, p. 04018213 doi: 10.1061/ V. 21, No. 3, Mar. 2018, pp. 291-298.
(ASCE)ST.1943-541X.0002144 27. Arora, S., and Singh, S. P., “Fatigue Strength and Failure Probability
7. Du, Y. G.; Clark, L. A.; and Chan, A. H. C., “Impact of Reinforcement of Concrete Made with RCA,” Magazine of Concrete Research, England,
Corrosion on Ductile Behavior of Reinforced Concrete Beams,” ACI Struc- V. 69, No. 2, 2017, pp. 55-67. doi: 10.1680/jmacr.15.00353
tural Journal, V. 104, No. 3, May-June 2007, pp. 285-293. 28. Arora, S., and Singh, S. P., “Probability of Flexural Fatigue Failure
8. Dang, V. H., and Francois, R., “Prediction of Ductility Factor of of Concrete Made with Recycled Concrete Aggregates,” IOP Conference
Corroded Reinforced Concrete Beams Exposed to Long Term Aging in Series: Materials Science and Engineering, England, V. 431, Aug. 2018,
Chloride Environment,” Cement and Concrete Composites, England, V. 53, pp. 102004.
Oct. 2014, pp. 136-147. doi: 10.1016/j.cemconcomp.2014.06.002 29. Sun, J. Z.; Huang, Q.; and Ren, Y., “Performance Deterioration of
9. Mohammed, A.; Almansour, H.; and Martin-Perez, B., “Simplified Corroded RC Beams and Reinforcing Bars under Repeated Loading,”
Finite Element Model for Evaluation of Ultimate Capacity of Corro- Construction and Building Materials, England, V. 96, Oct. 2015,
sion-Damaged Reinforced Concrete Beam-Columns,” Internal Journal pp. 404-415. doi: 10.1016/j.conbuildmat.2015.08.066
of Advanced Structural Engineering, Germany, V. 10, No. 4, 2018, 30. Hariche, L.; Ballim, Y.; Bouhicha, M.; and Kenai, S., “Effects of
pp. 381-400. doi: 10.1007/s40091-018-0204-2 Reinforcement Configuration and Sustained Load on the Behavior of Rein-
10. Al-Osta, M. A.; Al-Sakkaf, H. A.; Sharif, A. M.; Ahmad, S.; and forced Concrete Beams Affected by Reinforcing Steel Corrosion,” Cement
Baluch, M. H., “Finite Element Modeling of Corroded RC Beams Using & Concrete Composites, England, V. 34, No. 10, 2012, pp. 1202-1209. doi:
10.1016/j.cemconcomp.2012.07.010

31. Yang, S.; Fang, C. Q.; Yuan, Z. J.; and Yi, M. Y., “Effects of Rein- 39. Stewart, M. G., “Mechanical Behavior of Pitting Corrosion of Flex-
forcement Corrosion and Repeated Loads on Performance of Reinforced ural and Shear Reinforcement and Its Effect on Structural Reliability of
Concrete Beam,” Advances in Structural Engineering, V. 18, No. 8, 2015, Corroding RC Beams,” Structural Safety, Netherlands, V. 31, No. 1, 2009,
pp. 1257-1271. doi: 10.1260/1369-4332.18.8.1257 pp. 19-30. doi: 10.1016/j.strusafe.2007.12.001
32. GB 175-2007, “Common Portland Cement,” China, 2007. 40. Wang, H. C.; Gong, J. X.; and Qu, X. H., “An Experimental Study
33. GB/T 50082, “Standard for Test Methods Long-Term Performance on the Fatigue Behavior of Corroded Reinforced Concrete Beams,” China
and Durability of Ordinary Concrete,” China, 2009. Civil Engineering Journal, China, V. 38, No. 11, Oct. 2005, pp. 32-37.
34. GB/T 50152-92, “Standard Methods for Testing of Concrete Struc- 41. Ma, Y. F.; Wang, G. D.; Su, X. C.; Wang, L.; and Zhang, J. R.,
tures,” China, 1992. “Experimental and Modelling of the Flexural Performance Degradation
35. Seara-Paz, S.; Gonzalez-Fonteboa, B.; Martinez-Abella, F.; and Eiras- of Corroded RC Beams under Fatigue Load,” Construction and Building
Lopez, J., “Flexural Performance of Reinforced Concrete Beams Made with Materials, England, V. 191, Dec. 2018, pp. 994-1003. doi: 10.1016/j.
Recycled Concrete Coarse Aggregate,” Engineering Structures, England, conbuildmat.2018.10.031
V. 156, Feb. 2018, pp. 32-45. doi: 10.1016/j.engstruct.2017.11.015 42. Oyado, M.; Hasegawa, M.; and Sato, T., “Characteristics of Fatigue
36. Song, L., Fan, Z.W. and Hou, J. “Experimental and Analytical Inves- and Evaluation of RC Beam Damaged by Accelerated Corrosion,” Quar-
tigation of the Fatigue Flexural Behavior of Corroded Reinforced Concrete terly Report of RTRI, Japan, V. 44, No. 2, 2003, pp. 72-77. doi: 10.2219/
Beams,” International Journal of Concrete Structures and Materials, South rtriqr.44.72
Korea, V. 13, No. 1, Apr. 2019, pp. 1-14. 43. Masoud, S.; Soudki, K.; and Topper, T., “Postrepair Fatigue Perfor-
37. Zhang, W. P.; Ye, Z. W.; Gu, X. L.; Liu, S. G.; and Li, S. B., “Assess- mance of FRP-Repaired Corroded RC Beams: Experimental and Analytic
ment of Fatigue Life for Corroded Reinforced Concrete Beams under Investigation,” Journal of Composites for Construction, ASCE, V. 9, No. 5,
Uniaxial Bending,” Journal of Structural Engineering, ASCE, V. 143, 2005, pp. 441-449. doi: 10.1061/(ASCE)1090-0268(2005)9:5(441)
No. 7, 2017, p. 04017048 doi: 10.1061/(ASCE)ST.1943-541X.0001778 44. Wu, J.; Wang, C. X.; Xu, J.; Xiao, Z.; and Wu, F. H., “Study on
38. Fang, C.; Yang, S.; and Zhang, Z., “Bending Characteristics of Flexural Behavior of Corroded Reinforced Concrete Beams under Fatigue
Corroded Reinforced Concrete Beam under Repeated Loading,” Structural Loads,” China Civil Engineering Journal, China, V. 45, No. 10, Oct. 2012,
Engineering and Mechanics, South Korea, V. 47, No. 6, 2013, pp. 773-790. pp. 118-124.
doi: 10.12989/sem.2013.47.6.773

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.
Title No. 119-S35
Local Bond-Slip Behavior of Reinforcing Bars in High-

Performance Steel Fiber-Reinforced Concrete Beams
by Rita-Elizabeth Saikali, S. J. Pantazopoulou, and D. Palermo
This paper presents the results of a comprehensive experimental a nonzero normal compressive stress, σlat, in the cementi-
study on the impact of the tensile properties of high-performance tious component of the composite is required for equilib-
steel fiber-reinforced concrete (HPFRC) on the characteristic local rium (Fig. 1(b)). Stress σlat acts as internal confinement in
bond-slip relationship of steel reinforcement. Due to the tensile the composite (Chasioti and Vecchio 2017). From previous
toughness of the material, hoop stresses develop in the cover when
experimental studies, the internal confining pressure
the bar ribs displace relative to the surrounding concrete. This hoop
generated by the fibers was found to be of the same order
action enhances the overall bond-slip performance. To quantify
this effect, 13 beam specimens comprising HPFRC and containing of magnitude as the post-cracking tensile strength of the
a short embedment length of a 16 mm (0.63 in.) diameter, deformed HPFRC (Georgiou and Pantazopoulou 2018). The presence
reinforcing bar were investigated to quantify experimentally the of an internal confinement in the bulk cementitious mate-
average local bond strength and bond stress distribution over the rial offers several avenues to be investigated regarding the
anchorage. Two study variables were considered: a) the concrete structural response of HPFRC materials when implemented
cover (1Db and 2Db) as the exclusive confining factor; and b) the in construction, such as detailing for flexure, shear, and rein-
tensile strength and toughness of the mixtures used as the concrete forcement anchorage. Of those, the latter is the most funda-
matrix. The average bond strength was found to be proportional mental, as it pertains to the essence of the composite action
to the equivalent flexural tensile strength of concrete, reaching of reinforced concrete. Confining pressure imparted by the
a peak value up to 20 MPa (3 ksi) and a substantial bar-slip of
fibers in HPFRC is considered to contribute to the develop-
19 mm (0.75 in.) at the end of testing. The smaller concrete cover
ment capacity of anchorages through the same mechanism
provided sufficient confinement to develop bar yielding over an
anchorage length of 5Db without additional transverse reinforce- as stirrups in conventional concrete (which was quantified
ment. Increasing the cover resulted in a tougher pullout-splitting through the Ktr coefficient in ACI 408R-03).
failure mode with reduced deterioration over the cover. Anchorage and development capacity depend on the
reinforcement-to-concrete bond strength, which in conven-
Keywords: anchorage; bond-slip law; development; high-performance tional concrete is controlled by several design variables,
steel fiber-reinforced concrete (HPFRC); pullout; strain-hardening concrete. including bar cover, bar diameter, concrete tensile strength,
and the presence of confinement. These aspects are reflected
INTRODUCTION in the local bond-slip law of reinforcement embedded in
The toughness of high-performance steel fiber-reinforced conventional concrete (ACI 408R-03, fib Model Code 2010).
concrete (HPFRC) is manifested by significant flexural However, the local bond-slip behavior of reinforcement
strength that is sustained after cracking, up to deformation embedded in HPFRC has not been thoroughly quantified in
levels that can be more than a hundred times greater than the terms of the salient variables. Given that bond controls the
cracking strain of plain concrete (Habel et al. 2008; Schmidt composite action of reinforced concrete (Chao et al. 2009),
et al. 2004; Doiron 2017; Sritharan et al. 2018). These prop- complementing the experimental database to support the
erties are mainly due to the essential role of the steel fibers, establishment of local bond-slip relationships, which could
which act as randomly distributed reinforcement in the be used in the design and assessment of HPFRC structures,
densely packed cementitious mortar, where coarse aggre- is an important milestone for its practical implementation.
gates have been eliminated to maximize the impact of fibers In the present study, beam-type specimens are used to
in bridging flaws and cracks. Fibers engaged in tension characterize the local bond stress-slip relationships repre-
cause—by virtue of equilibrium—passive compressive sentative of HPFRC matrixes. Because the test bar is
stress in concrete (Manita and Pantazopoulou 2002; Susetyo developed in the presence of a flexural tension strain gradient,
et al. 2013; Archontas and Pantazopoulou 2015). this test setup is considered to be particularly relevant to the
The passive stress is referred to as internal confinement state of bond that develops in the tension zones of actual
and can be derived from the equilibrium of an infinitesimal flexural members. The experimental program investigated
element of HPFRC under shear, as shown in Fig. 1(a). The the effect of two parameters that affect the mechanics of
HPFRC material is a composite comprising: a) fibers that are
stressed in tension when bridging cracks; and b) the cemen- ACI Structural Journal, V. 119, No. 2, March 2022.
MS No. S-2021-058.R2, doi: 10.14359/51734334, received August 22, 2021, and
titious gel-sand mortar (that is, no coarse aggregates). After reviewed under Institute publication policies. Copyright © 2022, American Concrete
cracking, at an arbitrary section oriented in the y-direction Institute. All rights reserved, including the making of copies unless permission is
through the element, tensile fiber stresses ff are revealed; closure, if any, will be published ten months from this journal’s date if the discussion

Fig. 1—Rise of passive confinement in UHPFRC and
HPFRC materials: (a) element under shear; and (b) internal
confining stress σlat to equilibrate internal fiber tension at
arbitrary section.
bond: cover thickness and the tensile strength of the HPFRC
matrix. Three different design mixtures were considered to
explore the contribution of individual tensile properties in
determining the bond strength. For each mixture, the effect
of concrete cover was studied using two different values of
clear bar-cover thickness, C (that is, 1Db or 2Db, where Db is
the bar diameter).
To extract an average local constitutive law for bond for this
class of material, very short anchorage lengths were consid-
ered in the investigation (5Db). This choice was meant to
induce nearly uniform bond stress near failure, such that the
measured average bond strength, fb,ave (obtained by dividing
the developed bar force with the lateral contact area), could
satisfactorily approximate the true local bond strength, fb,max.
The flexural tensile strength, ft′, of the three materials used
as cementitious matrixes was obtained through inverse Fig. 2—(a) Definition of terms; (b) cross section of pulled bar
analysis of prism flexural tests (ASTM C1856/C1856M-17 and cover showing radial compression and hoop tension;
2017). The measured fb,ave values were quantified in terms (c) longitudinal concrete stresses exerted by bearing of ribs
of basic material characteristics of the HPFRC matrix (that of pulled bar as it displaces through cover; and (d) bond
is, its uniaxial compressive and tensile cracking strengths). stresses along anchorage length for input bar strain equal
to 0.002.
RESEARCH SIGNIFICANCE
fb,ave = Fb/πDbLb. Given that bond stress is variable along
A prerequisite for the introduction of high-strength and
the anchorage length, the average value, fb,ave, approaches
ultra-high-strength fiber-reinforced concrete used widely
the true strength, fb,max, only when the distribution of bond
in engineering practice is the comprehensive understanding
stresses at peak developed load is nearly uniform over the
and characterization of the bond interaction with embedded
available bar length. To verify that the anchorage length used
reinforcement and the evaluation of the development
in the present experimental program yields nearly uniform
capacity of bar anchorages. The paper presents experimental
bond stress distributions at peak bar force, the actual solu-
results from beam tests conducted to investigate the effect
tion of the differential equation for the bond is considered in
of the cover and the tensile strength and strain toughness of
the subsequent section.
the concrete matrix, providing insights regarding the mech-
Bond stresses, fb(x), develop on the lateral surface of an
anistic behavior of bars embedded in high-strength HPFRC
embedded bar to resist the translation of the bar relative to
materials and adding to the database of tests that are needed
the surrounding concrete matrix. (The independent variable
to support development of relevant design provisions.
x denotes distance along the anchorage from the loaded end,
whereas x = Lb – x is the distance of the same point from
BOND PARAMETERS
the free end of the anchorage (refer to Fig. 2(a)). The rela-
Length of anchorage
tive translation, known as bar slip, s(x), is necessitated by
The length of the anchorage, Lb, was an important design
the accumulation of length change difference between the
selection in the experimental program. The objective of the
two contacting materials. The incremental value of slip is
tests was to measure the bond strength, fb,max, obtained from
calculated from the difference between bar and matrix longi-
the maximum bar developed force, Fb, divided by the lateral
tudinal strains, εs(x) and εct(x), respectively, over an infini-
contact area, which is equivalent to the average stress: fb,max ≈
tesimal length dx.

(1) stresses for the same problem is shown in red in Fig. 2(d)
ds ( s ( x) ct ( x)) dx for Lb = 5Db, and in dashed burgundy for Lb = 10Db. (The
Due to bond action, axial stresses, fs, from the bar are grad- bar strain studied could not be developed for fb,max = 20 MPa
ually transferred to the concrete over the bonded length, Lb, [2.9 ksi] and smaller values of Lb.) Note that bond stresses
so that at the free end (end of the bonded length), the entire are practically uniform in the case of Lb = 5Db, but attenuate
force of the bar has been transferred to the concrete whereas nonlinearly for Lb = 10Db. It is concluded that fb,ave is a real-
the bar stress has attenuated to zero. Assuming a linear rela- istic approximation of fb,max in the case of Lb = 5Db, but not
tionship between bond stress, fb(x) and slip, s(x), (that is, fb = in the case of Lb = 10Db.
k ∙ s), and considering the reinforcing bar before yielding, Therefore, if the distinction between average and
Eq. (1) leads to (Eleftheriou et al. 2014; Yankelevsky and maximum bond stress is not considered, unrealistic conclu-
Jabareen 2005) sions may be drawn from long anchorage tests, as the average
bond obtained from the peak developed force divided by
d 2 s 4( 1) A E the nominal lateral contact area typically degrades with
f b ( x), b s (2) increasing value of Lb. To approximate a realistic estimate
dx 2 Db Es Ac Ec
of local bond strength, it is therefore common practice to
In Eq. (2), Ab is the bar area, and Ac is the area of the use small lengths of embedment, such that the true distribu-
concrete ring around the bar that is effectively engaged in tion of local bond stresses approximates the assumption of
tension (Fig. 2(b)). This is taken here as Ac = π ∙ C ∙ (C + Db); a uniform distribution underlying the definition of average
that is, doubling the value of the cover C effectively multi- bond (fib Bulletin 72 2014; ACI 408R-03; Yuan and Graybeal
plies the value of constant ψ in Eq. (2) by a factor of 3 for the 2014). In light of these findings and considering the prac-
same material moduli. The solution of Eq. (2) is as follows tical fabrication requirements of the beam specimens, in the
(Eleftheriou et al. 2014) present study, the length of 5Db was selected to assess bond
strength as an important advancement in understanding the
ε s ( x) = ε o ⋅
(e − Ωx
− e ( b)
Ω x−2L
) ; s ( x ) = D E Ω ⋅ ε ⋅ (e
b s
− Ωx
+ e ( b)
Ω x−2L
) ;Ω = 4k ( ψ + 1) mechanics of bond in novel cementitious matrixes such as
(1 − e −2 ΩLb
) 4k
o
(1 − e −2 ΩLb
) Db Es HPFRC. The same anchorage length is also used throughout
ε ct ( x ) = − ψε o ⋅ eΩx − e − Ωx
the literature as a benchmark for bond strength experiments
; fb ( x) = k ⋅ s ( x)
eΩLb − e −ΩLb (fib Bulletin 72 2014; ACI 408R-03; Malvar 1992).
(3)
Concrete cover and test setup
Parameter k is the stiffness of the average bond-slip law For ribbed bars, the main mechanism of force transfer
(fb = k ∙ s). The anchorage length, Lb, is taken equal to q ∙ Db; between the cementitious matrix and the steel bar is contin-
taking the distance x from the loaded end also as a multiple gent on the bearing of the ribs (fib Bulletin 72 2014; Phan
of the bar diameter (x = r ∙ Db), this may be rearranged as 2012), where the longitudinal component of the bearing
follows forces is bonded. The transverse component is resolved
into radial compressive and hoop tensile stresses in the
f b ( x) = f b (r ⋅ Db ) = ε o ⋅
Db Es Ω e
⋅
(
− ΩrDb
+ eΩDb (r − 2 q) ) cover (Tepfers 1979; Tastani and Pantazopoulou 2013).
The two components are depicted in Fig. 2(b) and (c). In
(4)
4 (
1 − e −2 ΩqDb ) plain concrete cover, radial splitting occurs when the hoop
Figure 2(d) plots the variation of fb(x) for different values strength of the cover is exceeded, whereas confinement
of q (r is the independent variable–coordinate along the enhances bond strength and bar development capacity to
anchorage length, r ≤ q) for a bar with Db = 16 mm (0.63 in.). the extent that cover splitting may be suppressed and bar
Bond distributions correspond to elastic bar behavior (for pullout failure may prevail (Malvar 1992; Pantazopoulou
bar strain at the onset of yielding, that is, εo = 0.002), and and Papoulia 2001; ACI 408R-03; Kabir and Islam 2004;
elastic bond-slip law for C = Db; lines correspond to different Dominguez 2005; Tastani and Pantazopoulou 2010; fib
available anchorage lengths (q = 3 or Lb = 3Db in light blue; Bulletin 72 2014).
q = 5 or Lb = 5Db in black; and q = 10 or Lb = 10Db in gray). In the case of fiber-reinforced cementitious composite, the
Note that the difference between the bond values at the steel fibers enhance bond of embedded bars without external
beginning and the end of the anchorage length indicates how confinement (Harajli et al. 2002; ACI Committee 408 2022).
intensely nonuniform is the bond stress distribution (and In HPFRC, the dense concrete matrix, which in fresh condi-
therefore fb,ave in these cases is significantly less than fb,max, tions is also self-consolidating, generally has an improved
which is sought through the tests). It is noted that this differ- quality interface with the reinforcement; whereas, due to its
ence diminishes for smaller q (or Lb) values tending towards strain-hardening behavior after cracking, it arrests the prop-
a more uniform distribution. This finding underscores the agation of splitting failure in the cover, thereby enhancing
importance of using small anchorage lengths for the fb,ave bond strength and toughness to bar slip, altering the mode
to be a realistic approximation of fb,max. For an estimated of failure from splitting to pullout (Spasojević 2008; Wille
maximum bond strength equal to 20 MPa (2.9 ksi) (which is et al. 2012; Bandelt and Billington 2014; Georgiou et al.
the estimated required fb,ave at the onset of bar yielding), at a 2017). Using the thick-ring analogy, it was previously illus-
limiting slip of 1 mm (0.039 in.), followed by a plastic bond- trated that the internal confining effect of HPFRC effectively
slip behavior (refer to Fig. 2(d)), the distribution of bond increases the fracture energy of the cover, thus enabling a

much higher strain development capacity for the anchorage
(ACI Committee 408 2022; Eleftheriou et al. 2017)
Cover is an essential variable in determining bond
strength. One of the first standard pullout tests conducted
on UHPFRC illustrated that increasing the concrete cover
from 20 to 45 mm (0.79 to 1.77 in.) mildly improved the
local peak bond stress from 58 to 65 MPa (8.41 to 9.43 ksi),
respectively while preserving a similar pre-peak, bond-
stress versus bar-slip relationship, as in the case of common
concrete. However, a thicker cover led to a more gradual
decay in the post-peak response (Holschemacher et al. 2005).
Using a pullout splice test setup on identical reinforcing bars Fig. 3—General description of specimens under study.
embedded in two different cementitious strain hardening
materials (with compressive strengths fc′ = 134 and 93 MPa transverse cracking on bond strength. A total of 13 specimens
[19.44 and 13.49 ksi], respectively), it was demonstrated were designed based on the specimen bond setup proposed
that increasing the clear cover from 2Db to 3.5Db increased by Tastani et al. (2016) to directly measure the developed
the developed bar stress from 600 to 800 MPa (87 to 116 ksi) average bond strength based on the global response of the
for the stronger matrix, but only to 650 MPa (94.27 ksi) for beam while following a simple testing arrangement. The
the weaker matrix (Yuan and Graybeal 2014). Based on total length of the beams was set to 900 mm (35.43 in.) with
modified pullout specimens with UHPC technology (fc′ = a square cross section of 152 mm (6 in.) (Fig. 3). The beams
144 MPa [20.89 ksi]) and a bar embedment length of 6Db, were simply supported and subjected to two equal loads
it was determined that increasing the cementitious cover (Ptot/2) placed symmetrically at the midspan at a constant
from 1.75Db to 2Db and 3.5Db caused an increase in the distance of 200 mm (7.87 in.), thus creating a central region
measured bond stress from 11 to 12 MPa (1.6 to 1.74 ksi) of constant moment. The shear span was equal to 300 mm
and 16 MPa (2.32 ksi), respectively (Ronanki et al. 2016). (11.81 in.), that is, twice the beam height, to discourage a
The conflicting evidence regarding the effect of cover in shear-dominant behavior. On account of the high post-
UHPFRC and HPFRC materials suggests that competing cracking tensile strength of HPFRC, no stirrups were neces-
mechanisms of local failure (for example, crushing of the sary to resist the shear forces in the specimen shear spans.
cementitious material in front of the ribs followed by bar In a broader context, reducing the use of stirrups is a moti-
pullout as opposed to cover splitting) may govern in the vating objective of the ongoing development of HPFRC
weaker matrix, thereby obscuring the influence of cover in technology, where dowel action, supported by the stirrups
UHPFRC materials. To examine these effects, two values in conventional concrete, is now provided by the distributed
for the cover thickness were considered in the experimental fiber reinforcement and their crack bridging effect.
program presented herein (that is, C = Db and 2Db). The focus of this study was to isolate and quantify the
Based on experimental evidence, it is also apparent that the contribution of the HPFRC cover to bond strength. For this
setup and specimen form used in bond studies play a critical reason, no other sources of confinement were placed in the
role in the magnitude of local bond strength obtained through test region (such as stirrups or direct loads bearing on the
testing and influences the conservatism and the relevance of cover region) (Orangun et al. 1977; Hamad and Rteil 2006;
the results with the actual state of bond developing along Harajli 2010). All the specimens had a 25 mm (0.98 in.) wide
bars functioning as reinforcement in structural members (fib notch at midspan; the notch depth was 40 mm (1.57 in.) for
Bulletin 72 2014; ACI 408R-03). In the present study, the specimens with C = 1Db (Case C1), and 52 mm (2.05 in.) for
specimen form was designed to model the circumstances of specimens with C = 2Db (Case C2). The beams were rein-
a tension bar being developed in the tensile zone of flexural forced with one primary 15M bar (Db = 16 mm [0.63 in.],
members. It is noted here that no standardized procedure fy = 425 MPa [61.64 ksi]) running along the centerline of
has been established as of yet for bond testing of HPFRC; the specimen in the tension zone. The notch was intended to
collective evaluation of results from previous experimental model the function of a crack in actual structural members,
studies show that the effect of specimen shape and test and it enabled the calculation of the developed bar force
setup is as dominant in this case as in conventional concrete from the measured flexural strength at the midspan. On one
(Tsiotsias and Pantazopoulou 2021). To eliminate any effects side of the notch (referred to as the support length), the bar
of interaction with shear, the test region (that is, the bonded was fully anchored continuously to the end of the beam. On
length of the test bar) was placed within the central segment the other side of the notch (which represents the test half of
of constant moment. This detail is a key difference between the span), the bar was anchored over a short length equal to
the RILEM beam bond test and the present setup. 5Db measured from the face of the notch. To achieve nearly
uniform distribution of bond stress, the 5Db segment occurred
Experimental program within the constant moment region of the test setup in all
General description—Beam anchorage tests were consid- cases. Over the remaining length of the span, the bar was
ered in this study because they simulate the actual state of unbonded with the aid of a polyvinyl chloride (PVC) sleeve
stress occurring in the concrete cover in the tension zone tube acting as a bond breaker. The sleeve and bar protruded
of flexural beam members while considering the effect of beyond the end-face of the beam to facilitate measurement

of the free-end slip externally through a specially designed Table 1—Details of bond test specimens and code
instrumentation jig. The beam half-span that coincided with identification
the test anchorage was reinforced with two auxiliary 10M Fibers (lf, mm; Code
bars placed symmetrically in the tension zone of the cross Lb, df, mm; Vf,%; No. of identifica-
section to resist the applied flexural moment in the segment Series mm C, mm lf/df) beams tion
beyond the end of the anchorage length of the test bar. The 80 C1 16 2 K1
(13, 0.2, 2, 65)
auxiliary bars had a side cover Cs = 16 mm (0.63 in.) and a K (Com K)
Straight
80 C2 32 2 K2
clear transverse distance of 54 mm (2.13 in.) from the main
15M test bar to minimize the influence on the measured 80 C1 16 (13, 0.2, 2.5, 65) 3 I1
I (In-house)
bond strength. 80 C2 32 Straight 3 I2
The developed bar force was determined at the notch 80 C1 16 (20, 0.2, 1, 100) 2 F1
from the statics of the notched cross section and global and (25, 0.3, 1,
F (Com F)
equilibrium. As previously noted, the bonded length in 80 C2 32 83.3) 1 F2
the test span was made short by design of the experiment, Straight
such that the assumption of an approximately uniform bond Note: 1 mm = 0.0394 in.
stress distribution at the peak force development (that is,
prior to anchorage failure) could be substantiated, as was illustrates the plan view of the specimens. All bars were
demonstrated in Fig. 2(c). The average bond stress, fb,ave, supported by the notch material and the end wooden block,
was calculated from the ratio of the developed bar force, Fb, eliminating the need for chairs through the cover over the
divided by the contact bar area (fs is the bar stress at the test span. Differentiating between the sets of each series is
loaded end). directly related to the height of the notch: 40 mm (1.57 in.)
for C = 1Db and 52 mm (2.05 in.) for C = 2Db as depicted
Fb A f f in Fig. 4(d) and (e), respectively. Further details of the spec-
f b , ave b s s MPa where Lb 5 Db (5)
Db Lb 5Db2 20 imen geometry are depicted in Fig. 4(f) through (i).
Test specimens and parameters of study—The experi- Test procedure and instrumentation—A two-point loading
mental program studied the effect of two parameters on local jig was designed and attached to a static hydraulic universal
bond strength: cover thickness and tensile characteristics of test system to realize the four-point bending test used in this
the cementitious fiber-reinforced matrix. To evaluate the study. The load was imposed through a displacement-con-
effect of the tensile matrix properties on the average local trolled rate of 0.005 mm/min (0.0002 in./min). Rollers were
bond strength of anchored reinforcing steel, three series of fastened to the jig through bead-ring bearings to enable the
beams comprising different concrete mixture designs were movement of both in-plane and out-of-plane directions,
fabricated. The test series are identified as K, F and I for eliminating any type of friction between the loading system
commercial K, for commercial F, and for In-house HPFRC and the beam. During testing, several parameters were
mixtures, respectively. (The commercial mixtures were measured at a frequency of 1 Hz, including the applied load,
provided in pre-blended sacks and mixed on-site under mid-deflection, and slip of the reinforcing bar. Testing was
the supervision of expert personnel of the providers.) Each terminated once the residual post-peak capacity was reduced
series comprised two sets of specimens concerning different to 20 to 30% of the peak load.
values for the concrete clear cover, expressed in multiples of The net deflection of the beam was calculated at the
the bar diameter: 1Db (Case C1) and 2Db (Case C2). There- midspan, excluding any external effects caused by either
fore, the average bond stress-slip relationship was compared the settling of the specimen or the movement of the end
between the three series and the different specimen groups supports. To achieve this objective, a linear potentiometer
to study the importance of the individual HPFRC properties (LP) was attached at the midspan of a stiff aluminum bar that
and the cover, respectively. In addition, the global behavior was simply supported on a steel yoke, which in turn, was
was evaluated in terms of mode of failure and crack pattern. secured at the midheight of the beam over the end supports,
Table 1 presents a summary of the specimens tested, where as shown in Fig. 5(a). During testing, the linear potentiom-
the last column lists the specimen identification referred to eter was bearing on a steel angle plate that was hot-glued
herein. to the top compression fiber of the beam. Simultaneously,
Figures 4(a) and (b) present the side view of the spec- the midspan deflection of the beam was monitored through
imen details corresponding to a clear cementitious cover C, digital image correlation (DIC) analyses using the MATLAB
of 1Db = 16 mm (0.63 in.) (Case C1) and 2Db = 32 mm module GeoPIV-RG (White et al. 2003; Dutton 2012).
(1.26 in.) (Case C2), respectively. The beam formwork Figure 5(b) presents the front of the test setup highlighting
was 16 mm (0.63 in.) longer than the length of the beam to the camera, the lighting, the timer, and the speckled middle
enable an extension of the primary reinforcing bar for slip region of the beam used for monitoring. The two deflection
measurement (a thick wooden block filled this space during measurement methodologies were used to compare and vali-
casting, with perforations to support the bar ends). Thus, the date the response of the specimens. Slip of the test bar was
main reinforcing bar was approximately 916 mm (36 in.) measured using a second linear potentiometer clamped to an
in length, extending over the entire formwork and passing instrumentation frame, which in turn was attached to the end
through a foam board used to create the midspan notch and of the beam, and traveled following the end rotation of the
the wooden block at the end of the test span. Figure 4(c) beam, as illustrated in Fig. 5(c).

Fig. 4—Dimensions and reinforcement detailing: (a) side view for C = 1Db (Case C1); (b) side view for C = 2Db (C2); (c)
section D; (d) section A; (e) section E; (f) section B; (g) section F; (h) section C; (i) section G; and (j) and specimen I1B ready
to be tested. (Note: All measurements in mm; 1 mm = 0.0394 in.)
Fig. 5—(a) Frame holding linear potentiometer for midspan deflection; (b) camera and lighting setup for midspan deflection
with DIC; and (c) frame holding linear potentiometer for bar slip measurement.
Material characteristics—Of the three HPFRC matrixes The dry F-mixture comprised portland cement, fine sand,
studied in this experiment program, two were proprietary and silica fume at a ratio of 1:1:0.25 per weight (with 910
pre-blended products (Series K and F) and one was devel- kg/m3 [56.81 lb/ft3] of cement in the concrete). The w/c was
oped in-house (Series I) following a previous mixture design 0.22, and a high-range water-reducing admixture was added
by Shao (2016). The three mixtures comprise fine aggregates at a ratio of 4.2% per weight of cement. Two types of steel
solely. fibers were used in Series F: 1% of 20 mm (0.79 in.) long
The dry K-mixture comprises portland cement, quartz straight steel fibers with a diameter of 0.2 mm (0.008 in.)
sand, and silica fume at a ratio of 1:0.6:0.2 per weight (with (fiber aspect ratio lf/df = 100) and 1% of 25 mm (0.98 in.)
1000 kg/m3 [62.43 lb/ft3] of cement in the concrete). Liquid long hooked-end steel fibers with a diameter of 0.3 mm
admixtures were added at a ratio of 3.5% of the cement (0.01 in.) (fiber aspect ratio lf/df = 83.3).
weight. The water-cement ratio (w/c) was 0.25; however, it The I-mixture was prepared using GUL cement (containing
was added 50% in liquid form and 50% in solid (ice) form. 6 to 15% limestone powder), silica sand (maximum grain size
Short, 0.2 mm (0.008 in.) diameter, 13 mm (0.51 in.) long of 0.530 mm [0.02 in.]), slag, and silica fume at a weight ratio
steel fibers (fiber aspect ratio lf/df = 65) were added at a volu- of 1:0.92:0.5:0.167, containing 724.13 kg/m3 (45.21 lb/ft3)
metric ratio of 2%. of cement in the concrete. The w/c was 0.33, whereas a poly-
carboxylate high-range water-reducing admixture with 40%

Fig. 6—(a) Cross section of a 75 x 75 mm (2.95 x 2.95 in.) prism cross section indicating the density of fibers (I-mixture, 2.5%
volume ratio, 13 mm [0.51 in.] fibers); (b) flow test diameter of 210 mm (8.27 in.) for Series K; (c) for Series I; and (d) for
Series F.
Table 2—Compressive and tensile properties of HPFRC mixtures

Series Sets fc′, MPa ft,max, MPa Δmax, mm fr, MPa fcr; fFu, MPa εcr; εtu Δεt = εtu – εcr
30.08 O : 10.62; 13.5
*
0.00037; 0.00474 0.00437
K K1 and K2 122.6 0.64 1.3 ÷ 4
22.98 R*: 10.42; 12.03 0.00043; 0.00528 0.00485
I1 153.5 28.8 0.56 O: 9.06; 11.63 0.00014; 0.00681 0.00667
I 1.63 ÷ 5
I2 125.0 26.4 0.40 O: 9.21; 10.46 0.00032; 0.0047 0.00438
35.28 O: 11; 16.06 0.00061; 0.01202 0.0114
F F1 and F2 128.4 1.24 3.5 ÷ 8
39.57 R: 11.91; 17.76 0.00046; 0.01635 0.01589
*
O is one directional, layered casting; R is random casting.
Note: 1 mm = 0.0394 in.; 1 MPa = 0.145 ksi.
solid content was added at a low ratio of 1.55% of the weight mm2 (43.75/in.2) long fibers for K, I, and F mixtures, respec-
of cement to reduce the risk of segregation. Short, 0.2 mm tively. These values are good approximations of the average
(0.008 in.) diameter, 13 mm (0.51 in.) long steel fibers (fiber numbers of fibers per unit area counted through sections
aspect ratio lf/df = 65) were used for Series I with a volu- cut from the prism specimens after testing, as illustrated in
metric fraction of 2.5%. Fig. 6(a) and discussed by Saikali (2019).
Fibers used in all cases were brass coated. The reinforcing The tensile strength is a function of bond strength along
bars used in the specimens had yield and nominal tensile the fiber and the length of available anchorage, whereas
strengths of 425 and 584 MPa (61.64 and 84.70 ksi), respec- for straight fibers, this cannot exceed half the fiber length,
tively, measured from uniaxial tension tests. lf/2. Considering that fiber slip rather than fiber rupture is
To interpret the differences in the experimental responses the preferred mechanism in HPFRC to create the apparent
of the different matrixes, the number of fibers embedded strain hardening in the response, it may be concluded that
in 1 m3 of material was estimated from the total volume of the maximum attainable post-cracking tensile strength in the
fibers used, divided by the volume of a single fiber (πdf2 ∙ HPFRC composite may be estimated from the product
lf/4): K-mixture had 2% × 109mm3/(0.22 × 0.785 × 13) = 49 ×
106 fibers/m3 (0.785 = π/4); I-mixture had 61.2 × 106 fibers/ fr = Vf ∙ (lf/df) ∙ τb (6)
m3; and F-mixture had 15.9 × 106 of 20 mm straight fibers,
and 5.66 × 106 of 25 mm hooked fibers/m3. Therefore, the where τb is the interfacial bond strength that develops between
F-mixture had the smallest total number of fibers per unit fibers and the matrix (τb is in the range of 1 to 3 MPa [0.15
volume, where approximately 25% of those fibers had mark- to 0.45 ksi] or more for smooth fibers and 3 to 6 MPa [0.45
edly greater anchorage strength owing to their hooked ends. to 0.9 ksi] or more for hooked fibers [Pantazopoulou et al.
However, it is noted that a significant amount of slip (and a 2019]). Using these values, the estimated passive confining
commensurate crack width) would be required to mobilize stress fr is 1.3 ÷ 4 MPa (0.18 ÷ 0.55 ksi), 1.6 ÷ 5 MPa (0.23 ÷
the anchorage strength. Assuming only 50% of the fibers to 0.7 ksi), and 3.5 ÷ 8 MPa (0.51 ÷ 1.15 ksi) for the three
be effective at any given cross section, the number of fibers, mixtures (K, I, and F), respectively (Table 2). The different
nf, crossing a unit area through HPFRC is estimated from numbers of fibers actively engaged in tension through a
the cross-sectional area of a single fiber having a diameter critical failure plane, considering their different lengths of
(df) and the volumetric ratio of the fibers, Vf, (Georgiou and anchorage and anchoring conditions, suggest that I-mix-
Pantazopoulou 2016) as nf = 0.64Vf/(df)2. For the HPFRC ture is expected to have higher tensile strength compared to
materials considered in the present study, the estimated Κ-mixture on account of the larger fiber content; whereas
number of fibers crossing a unit area is 0.32/mm2 (200/in.2), F-mixture is expected to be more compliant, owing to the
0.4/mm2 (250/in.2), and 0.16/mm2 (100/in.2) short and 0.07/ smaller number of fibers overall, but more resilient and

strong on account of the higher anchorage capacity for part
of the fibers—the hooked fibers engage mechanical inter-
locking with the matrix, whereas the strength of the smooth
fibers depends on friction.
Specimens that belonged to the same series were cast
together, with the exception of the I-mixture that was batched
for each set of beams (I1 and I2) separately. Figures 6(b)
through (d) present the 2-minute flow table diameter (equal
to 210 mm [8.27 in.]) measured based on ASTM C1856
(2017) and ASTM C230 (2014) for Series I and Series
K-F, respectively. ASTM C230 (2014) measures first the
“static” flow based on the average of two diameters recorded
2 minutes after lifting the mold and secondly the “dynamic”
flow accounting for 20 drops. However, ASTM C1856 Fig. 7—Definition of cracking and ultimate tensile strength
(2017) measures only the “static” flow. Specimen casting and strain in ascending and hardening range; crack mouth
was performed in layers following a direction parallel to opening displacement (w0) in post-peak range.
the longitudinal axis of the beam. Thereafter, the specimens prism specimens using inverse analysis on the average
were covered by a plastic sheet for 2 days to prevent loss response curves of the specimens, as specified in Annex
of water required for cement’s hydration. Subsequently, the 8 of the 2019 edition of CSA-S6 Code. The results of the
specimens were demolded from the formwork and covered procedure are provided in Table 2, including fcr and fFu as
with wet burlap sheets and a vapor barrier and stored at the strengths at visible cracking and the end of the hardening
ambient temperature for curing until the day of testing. range of the tensile stress-strain law, respectively, and εcr
From each batch of concrete, triplet sets of cylinders and εtu as the corresponding axial tensile strains; whereas the
(75 mm [2.95 in.] diameter by 150 mm [5.9 in.] length) and strain toughness of the material is defined by the length of
flexural prisms (75 x 75 x 280 mm [2.95 x 2.95 x 11 in.]) the hardening plateau, ∆εt.
were prepared and cast to be used for HPFRC characteri-
zation according to ASTM C1856 (2017). Three cylinders Discussion of test results
and prisms for each series were tested for material char- At each applied load stage in the beam tests, the average
acterization. Each series included three prism specimens bond stress, fb,ave, was calculated directly from the developed
obtained using one-directional, layered casting and three bar uniaxial tensile stress, fs, using Eq. (5). The steel stress
specimens randomly cast (identified accordingly in Table 2). was calculated based on sectional analysis of the beam at the
All prisms were tested under third-point loading placed on notched cross section in the central constant moment region
the side corresponding to 90 degrees from the casting surface while considering the reduced concrete section (due to the
to ensure that a smooth surface would be in contact with notch) and a tensile stress-strain law obtained from the mate-
the loading and support hardware. Following the same prin- rial characterization procedure described in the preceding
ciple as previously described for the beam specimens, a section from third-point flexural prism tests.
yoke was pinned at the midheight sections of the prisms at Table 3 presents the test results at first cracking and at peak
the roller supports to hold in place an extruded bar which, load for all the beams (Fig. 8). The reported values include
in this manner, was oriented along the chord of the spec- the applied load at first cracking F1,stcr; the measured midspan
imen (refer to Fig. 3). A linear displacement transducer was st
deflection at first cracking ∆1,cr ; the maximum sustained load
attached to the chord, measuring the net midspan deflection Fpeak, and the corresponding values of midspan deflection,
of the prisms without influence from support displacement Δpeak; the maximum sustained steel stress at the peak load fs;
and rotation. the corresponding maximum average bond strength fb,max at
Table 2 lists the compressive strength and the tensile flex- maximum stress (from Eq. (5)); and the associated bar slip-
ural strength obtained at the time of testing of prism samples page. To fully define the post-peak branch, the table includes
(all specimens were tested approximately 120 days after the midspan deflection and bar slip corresponding to a post-
casting). All series developed compressive strengths in the peak residual load equal to 35 kN (7.88 kip) and the failure
range of 120 to 130 MPa (17.4 to 18.85 ksi) with the excep- mode. At this load level, it is assumed that uniform frictional
tion of I1, which reached 153.5 MPa (22.26 ksi). The equiv- bond develops along the bonded length, with corresponding
alent average flexural tensile strength from the prism tests, values of fb,ave being 11 and 12 MPa (1.6 and 1.8 ksi) for
calculated based on engineering beam theory (fmax = 3Ptot/b ꞏ C = 1Db and 2Db, respectively. The derived relationships
h) as well as the corresponding midspan displacement, ∆max, between average bond stress and free-end slip of the rein-
was substantially higher in the case of F-mixture compared forcing bars were considered, along with information about
to K-mixture and I-mixture. Owing to the different batches failure mode and crack development to illustrate the effect of
produced for Series I, sets I1 and I2 developed two different concrete cover thickness.
tensile flexural strengths. Milestone values associated with Effect of material type—Figure 8 depicts the load versus
the characteristic bilinear tensile stress-strain and the post- midspan deflection of the beams. The vertical midspan
peak softening tensile-stress crack mouth opening displace- displacements obtained from the DIC verified the recorded
ment w0 (defined in Fig. 7) were obtained from the flexural values from the LP. Due to the higher resolution (in time)

F1,stcr , st
∆1,cr , Fpeak, Δpeak, fs, fbave at ult., Slip, Sliponset of yielding, Slip35kN, Failure
Series Beam kN mm kN mm MPa MPa mm mm mm mode
K1A 53.5 0.6 75.1 1.62 284.1 14.21 0.56 2.14 05.57 P-S
K1B 48.8 0.5 69.7 1.54 252.2 12.61 0.52 1.45 04.58 P-S
K
K2A 42.3 0.6 56.4 1.50 187.9 9.39 0.42 1.10 04.98 P-S
K2B 52.1 0.8 68.7 1.86 261.22 13.06 0.62 2.33 04.50 P-S
I1A 54.4 0.6 72.6 2.00 262.0 13.10 0.81 2.85 06.12 P-S
I1B 68.6 0.9 91.5 2.17 383.06 19.15 0.89 4.20 07.60 M
I1C 58.0 0.7 79.5 1.78 310.43 15.52 0.58 2.66 05.48 M
I
I2A 57.7 0.6 80.3 2.04 370.69 18.53 0.71 3.98 07.67 P-S
I2B 60.0 0.1 83.3 2.61 384.22 19.21 0.73 4.74 06.63 P-S
I2C 60.2 0.7 86.0 2.20 397.78 19.89 0.44 3.67 05.93 P-S
F1A 66.4 0.8 92.1 2.96 381.01 19.05 1.39 5.91 11.84 P-S
F F1B 75.7 1.0 101.0 3.24 425 21.25 1.58 1.58 12.39 P-S/M
F2 58.5 1.0 83.7 3.13 357.09 17.85 1.04 6.56 08.79 P-S
Note: P-S is pullout-splitting; M is multiple splitting cracks; and P-S/M is m-type split with pullout; 1 mm = 0.0394 in.; 1 MPa = 0.145 ksi; 1 kN = 0.2248 kip.
Fig. 8—Load-versus-midspan deflection measured from linear potentiometer and DIC for: (a) and (b) Series K; (c) and (d)
Series I; and (e) and (f) Series F.
of the instrumentation readings in comparison to the DIC region between the point at first crack and the peak load in the
intervals, some local disturbances in the resistance curves response curves of the beams became wider and higher with
occurring upon the opening of cracks could only be captured increasing strain toughness in the tensile stress-strain curve of
by the instrumentation. the material, Δεt (Table 2). On average, Series I attained a 12%
In all cases, the onset of visible cracking occurred in the higher peak load than Series K, whereas the corresponding
ascending branch of the response curve, at approximately difference in the deformation at peak load was 26% for the
70% of the peak load. The effect of the material type is illus- two series. Similarly, the corresponding values for Series F
trated by the comparison of the resistance curves of otherwise were 33% and 96% higher than the respective values of Series
identical specimens but comprising different cementitious K. The same trends were also observed for the beams with
matrixes, as presented in Fig. 8(a), (c), and (e) for Series a concrete cover of 2Db, as shown in Fig. 8(b), (d), and (f).
K, I, and F with a concrete cover of 1Db, respectively. The Due to the high tensile strength and strain toughness of the

Fig. 9—Front face crack pattern after failure: (a) comparison between K1 and K2; (b) comparison between I1 and I2; and (c)
comparison between F1 and F2.
F-mixture (hybrid mixture of 25 mm [0.98 in.] hooked and 20 the concrete’s tensile flexural strength and cover thickness
mm [0.79 in.] straight fibers), the corresponding beam series on the local bond stress-slip relationship. As in conventional
experienced higher deformation capacity and strength and a concrete, the initial ascending portion of the bond-slip rela-
milder post-peak descending branch. tionship is rather stiff, relying on adhesion and partly on
Effect of cover thickness—Increasing the minimum friction which benefits from the intrinsic confinement mobi-
concrete cover by 100% while keeping the cementitious lized by the fibers that cross the potential splitting plane.
material and the cross-sectional dimensions of the beam The gradual softening of the ascending branch of the bond-
consistent led to an apparent reduction of the load response slip response corresponds to the onset of concrete cracking
curve of the specimens on account of the smaller effec- transverse to the bar as it pulls out at the notch. The two sets
tive depth of the specimens with a thicker cover. This was of Series K and I with varying concrete covers presented a
also reflected in the initial stiffness of the response curves. similar pre-peak response, whereas the F matrix, although it
Specimens K1 sustained an average peak load of 72.4 kN attained a higher bond strength, deviated from linearity at an
(16.28 kip) with a midspan deflection of 1.6 mm (0.06 in.) earlier amount of slip, and exhibited overall, a more compliant
(Fig. 8(a)), whereas specimens K2 sustained an average fiber contribution. This difference in the response results from
value of 62.6 kN (14.07 kip) for the same mid-deflection, the number of fibers crossing a unit area of the F matrix, which
Fig. 8(b). A similar downward shift was observed for Series was approximately 42 and 35% of the number of the respec-
F when increasing the concrete clear cover from 1Db to 2Db, tive value of the K and the I mixtures. This indicates that the
as depicted in Fig. 8(e) and (f). Conversely, the average load larger number of fibers was a critical variable in the greater
versus midspan deflection response for Series I was approxi- effective confinement in the matrix prior to the onset of crack
mately the same for both concrete cover values, as shown in propagation. The F-mixture, however, demonstrated superior
Fig. 8(c) and 8(d). In fact, the average peak load was higher toughness owing to the mechanical anchorage provided by
for I2 with a value of 83.2 kN (18.70 kip) at a mid-deflection the hooks in approximately one-quarter of its fibers, although
of 2.3 mm (0.09 in.) in comparison to I1 with an average engagement of this action required some degree of slippage
maximum load of 81.2 kN (18.25 kip) and a mid-deflection and fiber pullout prior to its activation.
of 2 mm (0.08 in.). Given that the resistance of the specimens was controlled
All six sets of beams presented the same crack pattern on the by bond failure, the bond stress-slip responses fully inherit
side face of the beam, as shown in Fig. 9. The first crack occurred the characteristics of the load-midspan displacement curves,
in the constant moment region, emanating from one of the right- as is evident by comparing Fig. 8 and 10. Before any type
angle corners of the notch. (This was expected as the notch of shear failure in the beams, the observed governing bond
represents a local discontinuity.) Further increase of the applied failure mode confirms the appropriateness of the specimen
load led to the propagation of these cracks upwards in a diagonal setup for bond testing.
or vertical direction towards the loading roller nearest and above Figure 11(a) summarizes the coordinates of the mile-
the test bar. The cracks subsequently rotated almost parallel to the stone points of the bond-slip relationships shown in Fig. 10.
longitudinal axis in the compression zone of the beam section. For a short anchorage length of 5Db to develop yielding of
The increase of the concrete cover did not affect the crack a 425 MPa (61.64 ksi), 15M reinforcing bar, the required
development of the front face, as illustrated by the comparison bond stress is 20 MPa (2.9 ksi). This bond stress was attained
of the two sets of the same series in Fig. 9(a), (b), and (c). The either at or after the peak load, indicating the initiation of bar
height of the notch is a differentiating factor between the two yielding. Some scatter was observed as to the slip magnitude
sets of each series, as discussed previously. However, it was when the bar yielded, attributed to the curvature of the beam,
evident that the crack-mouth opening displacement was greater which was maximum in the notch and was concentrated in the
in Series F specimens, consistent with the definition of terminal constant moment region. (Curvature causes a variable distri-
pullout failure, which is understood to occur when the crack bution of normal bar strains in the reinforcement cross section
width exceeds half the typical fiber length (in the F series 1% of and reverses the direction of bond stresses at the top and
the fibers were 25 mm [0.98 in.] long and hooked). bottom of the bar [Tsiotsias and Pantazopoulou 2021b]). The
Effect of concrete cover and material type on the average filled circles and triangles of a particular color denote average
bond strength and slip—Figures 10(a), (b), and(c) present bond stress and the corresponding free-end slip values at the
results for the average measured bond stress-slip response onset of bar yielding for specimens with 1Db and 2Db clear
of the different specimens of K, I, and F series, respectively. cover, respectively, with blue, orange, and green identifying
The tests enable comparisons regarding the contribution of K, I, and F mixtures, respectively. Empty symbols with the

Fig. 10—Experimental average bond stress-slip versus concrete cover for (a) Series K; (b) Series I; and (c) Series F.
Fig. 11—Summary of experimental values: (a) peak average bond stress and corresponding free-end slip; (b) free end slip at a
residual load of 35 kN (7.88 kip); (c) bond strength normalized with fcr; and (d) bond strength normalized with √fc′.
same color code are used in Fig. 11(b) to define the average concrete, the bearing action it exerts on the bar due to flex-
bond stress and slip magnitude corresponding to a residual ural curvature is sustained, while in conventional concrete,
load of 35 kN (7.88 kip) in the beam resistance curve. Note cover splitting and yield penetration have a rapid deprecating
that although bar yielding was sustained at or beyond the effect on bond strength, which breaks down immediately in
peak load carried by several beams, the behavior is charac- the absence of stirrups.
teristically different from that of plain concrete, where yield Because bond strength, fb,ave, is obtained from the
penetration in the absence of stirrups is accompanied by developed bar force as per Eq. (5), the effect of post-cracking
rapid debonding and loss of anchorage development capacity. strain toughness of the material—which was noted on the
Because of the tensile strain-hardening toughness of the cover resistance curves of the specimens—is also reflected in the

Fig. 12—Specimens cracked bottom face with C = 1Db: (a) Series K; (b) Series I; (c) Series F and with C = 2Db; (d) Series K;
(e) Series I; and (f) Series F.
values of the average fb,max per group of specimens with 1Db 2010 for normal concrete (Darwin et al. 1996); alternatively,
cover thickness. The average bond strength at peak load was bond strength may be estimated from fb,max = 1.6fcr (Fig. 11(c)).
13.4, 15.9, and 20.15 MPa (1.94, 2.3, and 2.9 ksi) for K, I, The influence of both studied parameters, concrete
and F series, respectively (Fig. 11(a)). The corresponding strength, and concrete cover was also manifested in the
slip values at attainment of fb,max were 0.54, 0.76, and 1.5 mm failure modes experienced by the specimens. In all cases, a
(0.02, 0.03, and 0.06 in.). The same trend was observed for the narrow longitudinal splitting crack developed gradually on
beams with a concrete cover of 2Db as illustrated in Fig. 11(a) the bottom face of the specimens, initiating from the loaded
for the three series, indicating a mild increment of bond end of the bar and extending throughout the end of the
strength with the increase of the concrete’s tensile strength. It embedment length of 5Db (Fig. 12). Failure in the anchorage
is noteworthy that the values of bond at peak load are ordered of the test bar was marked by pullout-splitting, in some cases
in the same manner as the values of the internal confinement, combined with a multi-crack split emanating from the end
fr (Table 2). The influence of the material type is also reflected of the anchorage (a similar crack pattern was observed by
in the slip values associated with 35 kN (7.9 kip) residual load. Yuan and Graybeal [2014]). For the specimens with C =
For a concrete cover of 1Db, Series F underwent the largest 2Db, the mode was predominantly a pullout type of failure,
slip of 15.52 mm (0.61 in.) compared to 8.52 mm (0.34 in.) with a fine splitting crack developing after the peak load;
for Series I and 7.26 mm (0.29 in.) for Series K (Fig. 11(b)). transverse cracking was observed at the end of the bonded
Similarly, for a concrete cover of 2Db, the 15M reinforcing bar length, extending orthogonally to the bar, accompanied with
pulled out 7.07, 9.39, and 17.25 mm (0.28, 0.37, and 0.68 in.) much less cracking over the anchorage as compared to the
for Series K, I, and F, respectively, for an applied residual load specimens with the smaller cover (Fig. 12(d), (e), and(f)).
of 35 kN (7.88 kip) in the descending branch. At advanced levels of pullout slip (refer to Fig. 12(b)), a
The effect of concrete cover, C, on bond strength is wide longitudinal crack also developed in the case of the F
illustrated in Fig. 11(c) that plots the ratio of the normal- specimen. Evidence of pullout was also visible on the bar
ized maximum bond strength fb,max/fcr with respect to C/ ribs, which carried a wedge of crushed concrete (examined
Db (values of the cracking stresses, fcr, are obtained through after the end of the tests). Transverse cracks (perpendic-
inverse analysis from the third point bending tests as per the ular to the bar) developed along the length of embedment
CSA-S6 Annex 8, listed in Table 2). Figure 11(d) plots the in all cases due to the pullout of the reinforcing bar. The
ratios fb,max/√fc′ (where fc′ is the compressive strength obtained number of cracks was reduced with increased cover thick-
from cylinder tests, also listed in Table 2) with respect to C/ ness supporting the evidence of altering the failure mode to
Db. For Series I, both graphs indicate a direct correspondence clear pullout in cases with a thicker cover.
between these two parameters, where doubling the concrete With the fracture energy mobilized by the HPFRC mate-
cover increases the normalized bond stress values by 30% in rials, it was possible—with a short embedment length of
both plots. However, the sensitivity of the normalized fb,max 5Db—to yield the 15M (Db = 16 mm [0.63 in.]) reinforcing
/√fc′ to the cover thickness was not evident in the case of the bar for all the specimens past the peak load. In that range of
more resilient mixture of Series F. It is expected that beyond large deflection and slip, as the tensile strain in the concrete
a limit, the level of confinement affected by the HPFRC mate- tension zone increased beyond the ultimate value (εtu listed in
rial on the bar may be adequately high to alter the mode of Table 2), the contribution of concrete in tension diminished,
failure towards local crushing of the cementitious material in with more load being transferred to the bar, which entered the
front of the ribs with subsequent pullout. This alteration of post-yielding range. With the exception of specimen F1B,
the mode of failure obscures the cover thickness effect. The which yielded at the peak load (Table 3), all other specimens
experimental bond strength values are approximated conser- experienced bar yielding at advanced levels of displacement
vatively by the relationship, fb,max = 1.4 ∙ √fc′; note that this when cracking in the tension zone had eliminated the concrete
does not follow the upper limit of 2.5√fc set by Model Code contribution to sectional equilibrium. For these cases, bar

yielding is identified in Fig. 8 by vertical lines corresponding fiber-reinforced concrete (HPFRC) and to characterize the
to the intersection of the response curve with horizontal lines local bond-slip relationship with specific reference to the
drawn at load values corresponding to bar yielding without role of critical design variables on this property. The high
contribution of concrete in the tension zone (60 kN [13.5 kip] mechanical properties of HPFRC lead to a very ductile bond-
for C = 1Db and 53 kN [12 kip] for C = 2Db). Steel stresses slip response that fully inherited the characteristics of the
and developed deformations at the peak load are presented in load-displacement response of the beam specimens. Average
Table 3. Notably, in the specimens of Series F (C = 1Db or values for the peak bond strength with the bar before yielding
2Db) and Series I (C = 2Db), where higher bond stresses were ranged between 9.3 and 21.25 MPa (1.35 and 3 ksi), with
sustained, the reinforcing bar exceeded its nominal ultimate the corresponding free-end slip ranging between 0.4 and
tensile strength and fractured. Over the yielded part of the bar, 1.6 mm (0.02 and 0.06 in.), respectively. Beyond yielding,
bond stress diminishes to zero as long as the stress in the bar the average bond stress reached the conservative value of
is constant and equal to fy (Tastani and Pantazopoulou 2013). 29.5 MPa (4.25 ksi), whereas the true local bond strength
Thus, the effective length of anchorage after bar yielding was likely higher because the actual bonded length was less
is reduced from the nominal value that corresponds to 5Db than that provided on account of yield penetration. Both the
anchorage due to yield-strain penetration over the available bond strength and corresponding slip values at the peak and
length. Thus, for the response beyond yielding (after peak residual point increased with the magnitude of tensile strain
load), the actual local bond stresses are even higher than the toughness of the material.
estimated average values obtained from Eq. (5) and Table 3, Three different fine-aggregate HPFRC mixtures were used,
which represent a lower bound of the true fb,max. having similar fiber content but different fiber lengths; for
Relevance to tests by other investigators—effect of test the same content, mixtures with longer fibers demonstrated
setup and curvature—Normalized bond strength values earlier compliance (departure from the linear ascending
(Fig. 11(d)) are compared with values obtained from other branch in the response curves of the specimens) but a far
UHPFRC and HPFRC bond tests reported in the literature greater toughness, a characteristic that was inherited in the
for similar bonded lengths (Alkaysi and El-Tawil 2016; resulting bond-slip laws obtained through the experiments.
Asano and Kanakubo 2012; Cai et al. 2020; Fehling et al. Only short anchorages were tested in the program such that
2012; Hota and Naaman 1997; Jansson et al. 2012; Jung- the peak average bond strength would approximate closely
wirth and Muttoni 2004; Kanakubo and Hosoya 2015; the true local bond strength; this was illustrated through
Lagier et al. 2012, 2015; Marchand et al. 2016; Pishro and the solution of the field equation of bond for that stage.
Feng 2017; Saleem et al. 2013; Siong et al. 2016; Wong- Increasing the concrete cover from 1Db to 2Db had practi-
Chang et al. 2017; Tsiotsias and Pantazopoulou 2021a; Zhou cally no apparent effect on the average bond strength for the
and Qiao 2018; Zhou et al. 2019). These tests had been K and F series. This was due to the alteration of the mode
conducted using various test setups (standard pullout, modi- of failure from mixed pullout-splitting in the case of C =
fied pullout [Yuan and Graybeal 2014], direct tension pullout 1Db, to pullout with crushing of the cementitious material in
[DTP], tension lap splice, beam end, and beam lap splices). front of the ribs in the case of C = 2Db. The change in failure
Beam-type tests are very few (Bandelt and Billington 2014; mode was caused by the internal confining pressure effected
Harajli et al. 2002), as compared to the vast majority of other by HPFRC fibers in the beam specimens, which was more
HPFRC bond tests; all beam tests were limited to a 2Db effective in the case of thicker cover, particularly in more
concrete cover, whereas standard pullout specimens ranged resilient matrixes with longer fibers, thereby obscuring the
from 2 to 16Db cover values. Within these small concrete effect of the cover. By comparing the splitting cracks for the
covers, the anchored beam tests presented herein reached three design mixtures, it is apparent that specimens with a
more conservative bond strength values to those of the stan- cover thickness of 1Db failed by pullout-splitting mode after
dard and modified pullout tests, with fb,max being in the range having developed a higher number of microcracks along the
of 1 to 1.5√fc′. Both beam and standard pullout tests illus- main splitting crack than those with a thicker concrete cover
trated insensitivity to the increase in concrete cover C/Db, (2Db). The thicker cover precluded splitting on account of
whereas this was not the case for the direct tension pullout the significant reserves of tensile resistance by the cover.
and lap splice tests that presented lower fb,max values—less Based on the available data, a conservative approximation
than 1√fc′ for a normalized C/Db = 1 and 1√fc′ to 1.5√fc′ for the bond strength obtained from the tests was fb,max = 1.4√fc′
when increasing the cover to 2Db (Tsiotsias and Pantazo- or 1.6fcr, where fcr is the tensile strength of the material at the
poulou 2021a). Evidently, the amount of internal confine- onset of cracking, obtained from inverse analysis of the third
ment mobilized in direct tension pullout and lap splice tests point flexural tests; slip values associated with peak response
is much lower than in the beam and standard pullout tests as were in the range of 0.5 mm (0.02 in.), whereas the specimens
would be necessitated to cause crushing and compaction of sustained a residual strength of approximately one-fifth of the
the cementitious material in front of the ribs, thereby altering peak value for up to slip values that exceeded 12 mm (0.47 in.).
the mode of failure to pullout. Extending the development length requirements so as to include
HPFRC requires the collective review of the available experi-
CONCLUSIONS mental evidence from several different investigations covering
A detailed experimental study was conducted using the entire range of parametric sensitivities of the problem.
beam specimens to determine the mechanical behavior of In comparison with the current development length expres-
bond of reinforcement anchored in high-performance steel sions for normal concrete, the following trends are observed:

1) anchorages in HPFRC behave as confined, that is, corre- Bandelt, M. J., and Billington, S. L., 2014, “Bond Behavior of Steel
Reinforcement in High-Performance Fiber-Reinforced Cementitious
sponding to a value of the upper limit of the (cmin + Ktr)/Db coef- Composite Flexural Members,” Materials and Structures, doi: 10.1617/
ficient (≤ 2.5, ACI 408). In fact, the intensity of confinement in s11527-014-0475-4
this type of concrete is so significant that break-out failure has Cai, J.; Pan, J.; Tan, J.; and Li, X., 2020, “Bond Behaviours of Deformed
Steel Rebars in Engineered Cementitious Composites (ECC) And
been reported (Yuan and Graybeal 2014); 2) because of the high Concrete,” Construction and Building Materials, V. 252, p. 119082. doi:
confining pressure, bond strength supports the development of 10.1016/j.conbuildmat.2020.119082
bar yielding for very short anchorage lengths in the range of Chao, S. H.; Naaman, A. E.; and Parra-Montesinos, G. J., 2009, “Bond
Behavior of Reinforcing Bars in Tensile Strain-Hardening Fiber-Reinforced
5Db; 3) specimen form plays a significant role in the results; Cement Composites,” ACI Structural Journal, V. 106, No. 6, Nov.-Dec.,
and 4) while splitting failure is not mitigated, the response of pp. 897-906.
the anchorage past the cover splitting is robust, and exhibits a Chasioti, S., and Vecchio, F., 2017, “Effect of Fiber Hybridization on
Basic Mechanical Properties of Concrete,” ACI Materials Journal, V. 114,
slow post-peak degradation. No. 3, May-June, pp. 375-384. doi: 10.14359/51689479
The results illustrate that the embedment length of 5Db, CSA-S6, 2019, “Canadian Highway Bridge Design Code,” Canadian
which was considered adequately short to approximate Standards Association, Toronto, ON, Canada, 1177 pp.
Darwin, D.; Zuo, J.; Tholen, M. L.; and Idun, E. K., 1996, “Development
the true local bond strength with the average bond stress Length Criteria for Conventional and High Relative Rib Area Reinforcing
measurement, may be too conservative for high strength Bars,” ACI Structural Journal, V. 93, No. 3, May-June, pp. 347-359.
HPFRC matrixes. This length which is an established norm Doiron, G., 2017, “UHPC Pier Repair / Retrofit Examples of Completed
Projects in North America. AFGC-ACI-Fib-RILEM,” Int. Symposium on
for conventional concrete, was sufficient in most cases to Ultra-High Performance Fibre-Reinforced Concrete, Montpellier, France,
fully develop bar yielding. V. 1, pp. 983-992.
Dominguez, R. N., 2005, “Etude de La Liaison Acier-Béton : De La
Modélisation Du Phénomène à La Formulation d’un Élément Fini Enrichi
AUTHOR BIOS béton Armé,” Mémoire de Thèse, Ecole Normale Supérieure de Cachan,
Rita-Elizabeth Saikali is an Engineer-in-Training at Yukon’s Transportation
Cachan, France.
Engineering Branch (Bridges, Highways and Public Works), Whitehorse, YT,
Dutton, M., 2012, “Digital Image Correlation For Evaluating Structural
Canada. She received her MASc in structural engineering from York University,
Engineering Materials,” master’s thesis, Queen’s University, Kingston, ON,
Toronto, ON, Canada, in 2019 and her 5-year diploma in civil and environ-
Canada.
mental engineering specializing in public works and transportation from Saint
Eleftheriou, T. E.; Tastani, S. P.; and Pantazopoulou, S. J., 2017,
Joseph University, Beirut, Lebanon, in 2016. Her research interests include
“Development of Reinforcing Bars in SRCC Matrix: Modeling and
bond and development of steel reinforcement with high-performance concrete.
Interpretation,” Journal of Structural Engineering, ASCE, V. 143, No. 9,
p. 04017108. doi: 10.1061/(ASCE)ST.1943-541X.0001845
S. J. Pantazopoulou, FACI, is a Professor of civil engineering at York
Fehling, E.; Lorenz, P.; and Leutbecher, T., 2012, “Experimental Investi-
University. She received her MSc and PhD from the University of Cali-
gations on Anchorage of Rebars in UHPC,” Proceedings of Hipermat 2012
fornia, Berkeley, Berkeley, CA, and her 5-year university diploma in civil
3rd International Symposium on UHPC and Nanotechnology for High Perfor-
engineering from the National Technical University of Athens (NTUA),
mance Construction Materials, M. Schmidt, ed., Kassel, Germany, pp. 533-540.
Athens, Greece. She is a member of ACI Committees 341, Earthquake-Re-
fib, 2010, “Model Code 2010,” Fédération Internationale du Béton,
sistant Concrete Bridges, and 374, Performance-Based Seismic Design
Lausanne, Switzerland.
of Concrete Buildings; and Joint ACI-ASCE Committees 408, Bond and
fib Bulletin 72, 2014, “Bond and Anchorage of Reinforcement – Back-
Development of Steel Reinforcement, and 445, Shear and Torsion. Her
ground to the fib Model Code 2010,” Fédération Internationale du Béton,
research interests include earthquake engineering and seismic assessment,
Lausanne, Switzerland, 170 pp.
and design and retrofit of structures with novel engineering materials,
Georgiou, A., and Pantazopoulou, S. J., 2016, “Effect of Fiber Length
including engineered cementitious composites (ECCs) and ultra-high-per-
and Surface Characteristics on the Mechanical Properties of Cementitious
formance fiber-reinforced concrete (UHPFRC).
Composites,” Construction and Building Materials, Elsevier, V. 125, pp.
1216-1228. doi: 10.1016/j.conbuildmat.2016.09.009
D. Palermo is a Professor of structural engineering at York University
Georgiou, A.; Tastani, S.; and Pantazopoulou, S. J., 2017, “Testing
and an Associate Editor of the Canadian Journal of Civil Engineering. He
Procedure for Determining the Bond-Slip Law of Steel Bars in Strain-Hard-
received his BS, MASc, and PhD from the University of Toronto, Toronto,
ening Cementitious Composites,” Proc., Strain-Hardening Cement-Based
ON, Canada. His research interests include seismic repair and retrofitting
Composites, Dresden, Germany.
of concrete structures, nonlinear finite element modeling, seismic appli-
Georgiou, A. V., and Pantazopoulou, S. J., 2018, “Experimental Inves-
cations of shape memory alloys, UHPFRC and ECCs, tall masonry wall
tigation of the Confining Effect of Fibers in SHFRCC,” Composite Struc-
systems, and the response of structures to tsunami loading.
tures, V. 202, pp. 29-37. doi: 10.1016/j.compstruct.2017.09.110
Habel, K.; Charron, J.-P.; Braike, S.; Hooton, D.; Gauvreau, P.; and
ACKNOWLEDGMENTS Massicotte, B., 2008, “Ultra-High Performance Fibre Reinforced Concrete
The research was conducted in the High-Bay Laboratory of York Univer- Mix Design in Central Canada,” Canadian Journal of Civil Engineering,
sity. Lafarge, SIKA, FACCA-Dura, Stoney Creek Norchem, BASF, and V. 35, No. 2, pp. 217-224. doi: 10.1139/L07-114
Holcim generously donated prepackaged UHPC mixtures, silica fume, Hamad, B. S., and Rteil, A. A., 2006, “Comparison of Roles of FRP
high-range water-reducing admixtures, and slag. Sheets, Stirrups, and Steel Fibers in Confining Bond Critical Regions,”
Journal of Composites for Construction, ASCE, V. 10, No. 4, pp. 330-336.
doi: 10.1061/(ASCE)1090-0268(2006)10:4(330)
REFERENCES Harajli, M.; Hamad, B.; and Karam, K., 2002, “Bond-Slip Response of
Alkaysi, M., and El-Tawil, S., 2016, “Bond between Ultra-High Perfor- Reinforcing Bars Embedded in Plain and Fiber Concrete,” Journal of Mate-
mance Concrete and Steel Bars,” 1st International Interactive Symposium rials in Civil Engineering, ASCE, V. 14, No. 6, pp. 503-511. doi: 10.1061/
on UHPC, Des Moines, IA. (ASCE)0899-1561(2002)14:6(503)
Archontas, N. D., and Pantazopoulou, S. J., 2015, “Microstructural Harajli, M. H., 2010, “Bond Behavior in Steel Fiber-Reinforced Concrete
Behavior and Mechanics of Nano-Modified Cementitious Materials,” Tech- Zones Under Static and Cyclic Loading: Experimental Evaluations and
nopress, V. 3, No. 1, pp. 15-37. doi: 10.12989/.2015.3.014 Analytical Modeling,” Journal of Materials in Civil Engineering, ASCE,
Asano, K., and Kanakubo, T., 2012, “Study on Size Effect in Bond V. 22, No. 7, pp. 674-686. doi: 10.1061/(ASCE)MT.1943-5533.0000067
Splitting Behavior of ECC,” High Performance Fiber Reinforced Cement Holschemacher, K.; Weisse, D.; and Klotz, S., 2005, “Bond of Reinforce-
Composites, V. 6, pp. 137-144. doi: 10.1007/978-94-007-2436-5_17 ment in Ultra-High Strength Concrete,” Proceedings of the 7th International
ASTM C1856/C1856M-17, 2017, “Standard Practice for Fabricating Symposium on the Utilization of High-Strength/High-Performance Concrete,
and Testing Specimens of Ultra-High-Performance Concrete,” ASTM SP-228, American Concrete Institute, Farmington Hills, MI, pp. 513-528.
International, West Conshohocken, PA. Hota, S., and Naaman, A. E., 1997, “Bond Stress-Slip Response of
ASTM C230/C230M-14, 2014, “Standard Specification for Flow Table Reinforcing Bars Embedded in FRC Matrices under Monotonic and Cyclic
for Use in Tests of Hydraulic Cement,” ASTM International, West Consho- Loading,” ACI Structural Journal, V. 94, No. 5, Sept.-Oct., pp. 525-537.
hocken, PA.

Joint ACI Committee-ASCE 408, 2003, “Bond and Development of Shao, B., 2016. “Mix Development of PLC-Based Ultra-High Perfor-
Straight Reinforcing Bars in Tension (ACI 408R-03),” American Concrete mance Fibre Reinforced Concrete and Characterization of Key Mechanical
Institute, Farmington Hills, MI, 49 pp. Properties and Time-Dependent Behaviour,” MASc thesis, University of
Joint ACI-ASCE Committee 408, 2022, “Bond and Development Toronto, Toronto, ON, Canada, 130 pp.
of Straight Reinforcing Bars in Tension,” under development by ACI Siong, W. L.; Shao-Bo, K.; Kang, H. T.; and En-Hua, Y., 2016,
Committee 408, Farmington Hills, MI. (in progress) “Experimental and Analytical Investigation on Bond-Slip Behaviour of
Jansson, A.; Lofgren, I.; Lundgren, K.; and Gylltoft, K., 2012, “Bond Deformed Bars Embedded in Engineered Cementitious Composites,”
of Reinforcement in Self-Compacting Steel-Fibre-Reinforced Concrete,” Construction and Building Materials, V. 127, pp. 494-503. doi: 10.1016/j.
Magazine of Concrete Research, V. 64, No. 7, pp. 617-630. doi: 10.1680/ conbuildmat.2016.10.036
macr.11.00091 Spasojević, A., 2008. “Structural Implications of Ultra-High Perfor-
Jungwirth, J., and Muttoni, A., 2004, “Structural Behavior of Tension mance Fibre-Reinforced Concrete in Bridge Design,” PhD thesis, Ecole
Members in Ultra High Performance Concrete,” Proceedings of Interna- Polytechnique Fédérale de Lausanne, Lausanne, Switzerland.
tional Symposium on Ultra High Performance Concrete, Kassel, Germany, Sritharan, S.; Doiron, G.; Bierwagen, D.; Keierleber, B.; and Abu-Ha-
pp. 553-544. wash, A., 2018, “First Application of UHPC Bridge Deck Overlay
Kabir, R., and Islam, M., 2014, “Bond Stress Behavior between Concrete in North America,” Transportation Research Record: Journal of the
and Steel Rebar : Critical Investigation of Pull-Out Test via Finite Element Transportation Research Board, V. 2672, No. 26, pp. 40-47. doi:
Modeling,” International Journal of Civil and Structural Engineering., 10.1177/0361198118755665
V. 5, No. 1, pp. 80-90. Susetyo, J.; Gauvreau, P.; and Vecchio, F. J., 2013, “Steel Fiber-Rein-
Kanakubo, T., and Hosoya, H., 2015, “Bond-Splitting Strength of Rein- forced Concrete Panels in Shear: Analysis and Modeling,” ACI Structural
forced Strain-Hardening Cement Composite Elements with Small Bar Journal, V. 110, No. 2, Mar.-Apr., pp. 285-295.
Spacing,” ACI Structural Journal, V. 112, No. 2, Mar.-Apr., pp. 189-198. Tastani, S. P.; Konsta-Gdoutos, M. S.; Pantazopoulou, S. J.; and
doi: 10.14359/51687228 Balopoulos, V., 2016, “Effect of Carbon Nanotubes and Polypropylene
Lagier, F.; Massicotte, B.; and Charron, J. P., 2012, “Bond Splitting Fibers on Bond of Reinforcing Bars in Strain Resilient Cementitious
Strength of Lap Splice Embedded in Ultra High Performance Fibre Rein- Composites,” Frontiers of Structural and Civil Engineering, V. 10, No. 2,
forced Concrete Under Direct Tension,” Bond in Concrete 2012, J. W. pp. 214-223. doi: 10.1007/s11709-016-0332-3
Cairns, G. Metelli, and G. A. Plizzari, eds., pp. 351-358. Tastani, S. P., and Pantazopoulou, S. J., 2010, “Direct Tension
Lagier, F.; Massicotte, B.; and Charron, J. P., 2015, “Bond Strength of Pullout Bond Test: Experimental Results,” Journal of Structural Engi-
Tension Lap Splice Specimens in UHPFRC,” Construction and Building neering, ASCE, V. 136, No. 6, pp. 731-743. doi: 10.1061/(ASCE)
Materials, V. 93, pp. 84-94. doi: 10.1016/j.conbuildmat.2015.05.009 ST.1943-541X.0000159
Malvar, J., 1992, “Bond of Reinforcement under Controlled Confine- Tastani, S. P., and Pantazopoulou, S. J., 2013, “Reinforcement and
ment,” ACI Materials Journal, V. 89, No. 6, Nov.-Dec., pp. 593-601. Concrete Bond: State Determination Along the Development Length,”
Manita, P., and Pantazopoulou, S. J., 2002, “Constitutive Model for the Journal of Structural Engineering, ASCE, V. 139, No. 9, pp. 1567-1581.
Stress-Strain Response of Fibre-Reinforced Concrete in Compression,” doi: 10.1061/(ASCE)ST.1943-541X.0000725
Proceedings, Intnl. Seminar: Composite Materials in Concrete Construc- Tepfers, R., 1979, “Cracking of Concrete Cover along Anchored
tion R. K. Dhir, K. A. Paine, and M. D. Newlands, eds., Thomas Telford, Deformed Reinforcing Bars,” Magazine of Concrete Research, V. 31,
London, UK, pp. 44-48. No. 106, pp. 3-12. doi: 10.1680/macr.1979.31.106.3
Marchand, P.; Baby, F.; Khadour, A.; Battesti, T.; Rivillon, P.; Quiertant, Tsiotsias, K., and Pantazopoulou, S. J., 2021b, “Analytical Investigation
M.; Nguyen, H. H.; Genereux, G.; Deveaud, J. P.; Simon, A.; and Toutle- on the Effect of Test Setup on Experimental Bond Strength,” CivilEng, V. 2,
monde, F., 2016, “Bond Behavior Of Reinforcing Bars In UHPFRC: Experi- No. 1, pp. 14-34. doi: 10.3390/civileng2010002
mental Investigation,” Materials and Structures, V. 45, No. 5, pp. 1359-5997. Tsiotsias T., and Pantazopoulou, S. J., 2021a, “Bond Behavior of
Orangun, C. O.; Jirsa, I. O.; and Breen, J. E., 1977, “A Reevaluation of Ultra-High-Performance Fiber Reinforced Concrete (UHPFRC) Under
Test Data on Development Length and Splices,” ACI Journal Proceedings, Direct Tension Pullout,” Engineering Structures, V. 243, p. 112701, doi:
V. 74, No. 3, Mar., pp. 114-122. 10.1016/j.engstruct.2021.112701
Pantazopoulou, S. J.; Palermo, D.; Yang, Y.; Eshghi, N.; Saikali, R.; White, D. J.; Take, W. A.; and Bolton, M. D., 2003, “Soil Deforma-
and Chasioti, S., 2019, “Development of Specification for Determining tion Measurement Using Particle Image Velocimetry (PIV) and Photo-
the Tensile Behaviour of HPFRC Materials for Structural Applications in grammetry,” Geotechnique, V. 53, No. 7, pp. 619-631. doi: 10.1680/
Highway Bridges,” Final Report, HIIFP 2017 - Open 15 - Agreement No. geot.2003.53.7.619
9017-R-0032. Wille, K., Naaman, A. E., El-Tawil, S., and Parra-Montesinos, G. J,
Pantazopoulou, S. J., and Papoulia, K. D., 2001, “Modeling Cover- 2012, “Ultra-High Performance Concrete and Fiber Reinforced Concrete:
Cracking Due to Reinforcement Corrosion in RC Structures,” Journal of Achieving Strength and Ductility without Heat Curing,” Materials and
Engineering Mechanics, ASCE, V. 127, No. 4, pp. 342-351. doi: 10.1061/ Structures/Materiaux et Constructions, V. 45, No. 3, pp. 309-324.
(ASCE)0733-9399(2001)127:4(342) Wong-Chang, C.; Seok-Joon, J.; and Hyun-Do, Y., 2017, “Bond and
Phan, T. S., 2012, “Modélisation Numérique de l’interface Acier-Béton : Cracking Behavior of Lap-Spliced Reinforcing Bars Embedded in Hybrid
Application Au Comportement Des Structures En Béton Renforcées Par Fiber Reinforced Strain-Hardening Cementitious Composite (SHCC),”
Des Aciers Plats Crantés,” PhD thesis, Université Paris-Est, Champs-sur- Composites. Part B, Engineering, V. 108, pp. 35-44. doi: 10.1016/j.
Marne, France. compositesb.2016.09.086
Pishro, A. A., and Feng, X., 2017, “Experimental Study on Bond Stress Yankelevsky, D. Z., and Jabareen, M., 2005, “Analysis of 1-D Tension
between Ultra High Performance Concrete and Steel Reinforcement,” Civil Stiffening with Discrete Cracks,” Serviceability of Concrete: A Symposium
Engineering Journal, V. 3, No. 12, pp. 12-35. Honoring Dr. Edward G. Nawy, SP-225, American Concrete Institute,
Ronanki, V. S.; Valentim, D. B.; and Aaleti, S., 2016, “Development Farmington HIlls, MI, pp. 55-82.
Length of Reinforcing Bars in UHPC: An Experimental and Analytical Yoo, D. Y.; Kwon, K. Y.; Park, J. J.; and Yoon, Y. S., 2015, “Local Bond-
Investigation,” First International Interactive Symposium on UHPC, Des Slip Response of GFRP Rebar In UHPFRC,” Composite Structures, V. 120,
Moines, IA, V. 4, pp. 1-9. pp. 53-64. doi: 10.1016/j.compstruct.2014.09.055
Saikali, R. E., 2019. “Bond Behaviour of Steel Reinforcing Bars Yuan, J., and Graybeal, B. A., 2014, “Bond Behavior of Reinforcing
Embedded in Ultra-High-Performance Steel Fiber Reinforced Concrete,” Steel in Ultra-High Performance Concrete,” ACI Structural Journal, V. 112,
MASc thesis, York University, Toronto, ON, Canada, 292 pp. No. 6, Nov.-Dec., pp. 851-860.
Saleem, M. A.; Mirmiran, A.; Xia, J.; and Mackie, K., 2013, Zhou, Y.; Fu, H.; Li, P.; Zhao, D.; Sui, L.; and Li, L., 2019, “Bond Behavior
“Development Length of High-Strength Steel Rebar in Ultrahigh Perfor- Between Steel Bar and Engineered Cementitious Composite (ECC) Consid-
mance Concrete,” Journal of Materials in Civil Engineering, ASCE, V. 25, ering Lateral FRP Confinement: Test And Modeling,” Composite Struc-
No. 8, pp. 991-998. doi: 10.1061/(ASCE)MT.1943-5533.0000571 tures, V. 226, p. 111206. doi: 10.1016/j.compstruct.2019.111206
Schmidt, M.; Fehling, E.; and Geinsenhansluke, C., 2004. “Ultra High Zhou, Z., and Qiao, P., 2018, “Bond Behavior of Epoxy-Coated Rebar in
Performance Concrete (UHPC),” Proceedings of the International Sympo- Ultra-High-Performance Concrete,” Construction and Building Materials,
sium on Ultra High Performance Concrete, Kassel University Press, V. 182, pp. 406-417. doi: 10.1016/j.conbuildmat.2018.06.113
Kassel, Germany, Sept. 13-15.

ARE YOU A RESEARCHER?
SIGN UP FOR TODAY!
ORCID provides a persistent digital identifer that distinguishes you
from every other researcher and, through integration in key research
workflows such as manuscript and grant submission, supports
automated linkages between you and your professional activities,
ensuring that your work is recognized.
Individuals may use ORCID services freely and it’s as easy as 1-2-3:
1 REGISTER
2 ADD YOUR INFO
3 USE YOUR ORCID ID
For more information and to register, visit:

WWW.ORCID.ORG
Title No. 119-S36
Uncertainty Modeling and Analysis of Punching Shear

Resistance of Flat Slabs
by Ming-Yue Jiang, Qing-Xuan Shi, and Wang-Hu Zhao
An approach is provided to modify model uncertainty related to and revised according to the actual conditions of each
punching shear resistance of flat slabs. Model uncertainty of six country. Among the mechanical models, the critical shear
shear strength models in different design codes are investigated crack theory (CSCT) developed by Muttoni9 has shown
systematically. Based on 452 punching tests of concentrically wide consistency and generality, serving as a basis for the
loaded slabs, the computational accuracy of six punching shear
punching provisions of fib Model Code 2010 (MC2010).10
strength models is compared. Characteristic values of model
Based on the previously developed beam shear model, Ghali
uncertainty with different specific assurance rates are determined
based on the probabilistic characteristics of computational model and Gayed11 advocated the universal design as a supple-
uncertainties. The comparison shows that the accuracy of the draft ment to ACI 318. Schmidt et al.12 clarified the contribution
of the next generation of the Eurocode 2 model is superior to the of concrete or steel in RC slabs and compared it to design
Chinese, American, and Canadian code models, which usually provisions based on Eurocode 2 (EC2),13 MC2010, and
underestimate the punching shear resistance of flat slabs. Further- the draft (of April 2018) of the next generation Eurocode 2
more, the model uncertainty of the Chinese, American, and Cana- (prEC2).14 To account for the impacts of warping, Rashwan
dian code models follow a normal distribution, whereas the model et al.15,16 introduced a new factor to the ACI equation for
uncertainty of the other three models obey lognormal distribution. checking punching shear. It has been demonstrated that
punching tests should be performed with warping effects
Keywords: design code; model uncertainty; punching shear strength model;
reinforced concrete slab-column connections; statistical characteristics. in mind to prevent underestimating the maximum punching
shear strength demand.
INTRODUCTION This paper covers the data from 452 tests of punching
Because of the advantages of simple structures and low shear resistance of RC slab-column connections. The
cost, reinforced concrete (RC) floors without beams have design provisions according to concrete codes including
been widely used in commercial and residential build- ACI 318, EC2, MC2010, GB50010-2010,17 CSA A23.3-
ings. However, the slab-column connections are prone to 14,18 and prEC2 are summarized. The model uncertainty
punching shear failure, which has no obvious signs before statistics were derived by comparing the mean predicted
occurring and can easily lead to the collapse of the whole punching shear resistance to the measured punching shear
structure. Therefore, the calculation of the punching shear resistance by the Kolmogorov-Smirnov (K-S) test. More-
resistance of flat slabs is considered one of the important over, the punching shear strength of a slab-column connec-
links in the design of RC beamless floors. tion depends on many parameters, such as slab thickness,
In the past few decades, researchers have proposed concrete strength, and shear reinforcement. Einpaul et al.19
various theoretical strength models to provide a theoretical advised that slenderness has an important influence on the
background for the punching shear strengths of slab-column punching strength of slabs with shear reinforcement, despite
connections. Yankelevsky and Leibowitz1 and Nielsen2 the fact that it is neglected in many codes of practice. There-
developed strength models based on the theory of plas- fore, it is necessary to compare the deviation degree between
ticity. Park et al.3 developed a strain-based strength model the punching bearing resistance of RC slat and column joints
to forecast the punching shear strength of interior slab- in various national codes to lay a foundation for accurately
column connections. Alexander and Simmonds4 proposed a predicting the punching shear resistance.
strut-and-tie model for slab-column connections. Johansen,5
Bažant and Cao,6 and Kueres et al.7 developed design equa- RESEARCH SIGNIFICANCE
tions based on fracture mechanics. Hegger et al.8 performed At present, most of the calculation formulas for punching
experimental studies on concrete footings and found that bearing resistance stipulated in concrete codes (such as
the failure shear crack angle of footings is steeper than that ACI 318-14,20 CSA A23.3, and EC2) are developed based
of slender slabs, and proposed a new shear strength model on existing test results. When dealing with practical prob-
addressing the test results. lems, uncertainties in model predictions are unavoidable,
Because the problem of punching is so intricate, no widely
accepted design model is available for design practice. The MS No. S-2021-063.R1, doi: 10.14359/51734376, received June 28, 2021, and
semi-theoretical and semi-empirical algorithms are adopted reviewed under Institute publication policies. Copyright © 2022, American Concrete
in the design codes of concrete structures in various coun- obtained from the copyright proprietors. Pertinent discussion including author’s
tries, and then the punching calculation formula is adjusted is received within four months of the paper’s print publication.

Table 1—Summary of punching shear strength models
Design value of punching shear capacity without
Concrete strength Effective depth Kρ Size effect shear reinforcement
Model Vc d — ζ VR
  4
0.083  2 +  f c′b0 d
  β
 α d 
ACI 318-14 0.33√fc′ d — — Vc , ACI = min 0.083  s + 2 f c′b0 d
  b 0 

 0 . 33 3 f ′b
c 0 d

Current Eurocode 2 0.18(fck)1/3 d β p = 3 100ρ kζ = 1 + √(200/d) ≤ 2.0 Vc,MC = CRd,ckζ(100ρfck)1/3 ∙ u1 ∙ d ≥ vmin ∙ u1 ∙ d
1000 Vc,FIP = kψ(√fck/γc)b0dv

Model Code 2010 √fck dv — βd = 4
d kψ = 1/(1.5 + 0.6 ∙ ψ ∙ d ∙ kdg) ≤ 0.6
GB50010-2010 ft d — βh Vc,GB ≤ 0.7βhftηumh0
2
1 0.19c f c b0 d
c

d
CSA A23.3-14 0.38√fc′ d — — Vc ,CSA min s 0.19 c f c b0 d
b0

0.38c f c b0 d

Draft of next Vc,NEC2 = τn,NEC2 ∙ b0 ∙ dv

generation of √fck dv (100ρ)1/3 (400/h0)1/4 = (0.6/γc) ∙ kpb ∙ (100 ∙ ρ1 ∙ fck ∙ (ddg/dv))1/3 ∙ b0 ∙ dv
Eurocode 2 ≤ (0.6/γc) ∙ √fck ∙ b0 ∙ dv
Note: “—” indicates items not measured. Units are in SI.
which leads to a certain degree of uncertainty in the input EC2 are much larger than ACI 318, MC2010, GB50010,
parameters and the calculated output response. This paper CSA A23.3, and prEC2. More detailed background infor-
presents the statistical characterization of model uncertainty mation, provisions for detailing, and derivations of the
in six concrete code predictive models for punching shear. design concepts are available in the sources provided in the
The results could serve as a basis for the development of references.
punching shear design in the future.
SOURCE OF DATA
PUNCHING SHEAR STRENGTH MODELS To analyze the uncertainty of the calculation model of
The punching shear design provisions for flat slabs the aforementioned RC slab punching resistance model,
according to ACI 318-14, EC2, MC2010, GB50010, CSA 452 groups (listed in Table A1 in the Appendix) of RC
A23.3, and prEC2 are briefly summarized, focusing on the slab punching shear resistance test data were collected
determination of concrete and steel contributions (Table 1). from the database22 and Yi et al.23 It is impossible to calcu-
For the calculation of rotations around the supported area, late the result of the model specification, so some data has
MC2010 proposes a different level of approximations been omitted. All of the specimens were RC slab-column
(LoA).2 For assessing the physical parameters of the design connections without shear reinforcement that were subjected
equations, LoA I offers uncomplicated and safe assump- to concentric punching shear loads. The essential design
tions21 that will suffice for most preliminary design purposes; factors influencing the punching shear strength of flat slabs
hence, LoA I is applied for the evaluations presented in this are discussed in terms of their frequency of occurrence. The
paper. range of major parameters for these specimens including the
The primary parameters affecting the punching shear compressive strength of the concrete (fc), the tension rein-
strength of the slab-column connection are the compres- forcement ratio (ρ), the edge length of the column section
sive strength of the concrete (fc), the size effect, the edge (c), and the effective depth of the slab (d) are shown in
length of the column section (c), and the effective depth of Fig. 2. The data identified herein represent ranges for each
the slab (d). EC2, MC2010, and prEC2 address the effect of parameter separately considered—that is, without referring
the tension reinforcement ratio (ρ) in their evaluation of the to associated values of different parameters among various
punching shear strength. The current design codes also differ specimens within the database. The distribution of the main
in the definition of the location of the critical section of the characteristics of the specimens covers the effective depth
slab-column connection. Figure 1 shows the definition of the range of 29 to 300 mm, the edge length of the column section
perimeter of the critical section when the column is circular, ranges from 50 to 901 mm, and the flexural steel ratios range
square, or rectangular. The critical sections specified in between 0.1 and 7.3%. The typical ranges of normal- and

Fig. 1—Critical shear perimeter in codes of practice.
Fig. 2—Distribution of influencing parameter values of specimens (452 tests). (Note: 1 mm = 39.4 × 10–3 in.; 1 MPa = 145 psi;
and 1 kN = 1000 N = 224.8 lb.)
high-strength concrete are 11 to 127 MPa, respectively. It slab thickness. To uniformly describe the uncertainty that
should be noted that the actual measured parameters, rather exists in the bias between the calculated value and the exper-
than notional values, are used in the computations. The best imental test value, the computation of the model uncertainty
estimate model predictions are achieved when the design is defined as
model is used to forecast punching shear resistance while
ignoring all conservative bias or safety factors added for θ = Vt/Vcal (1)
design objectives.
where Vt is the actual measured value of the punching shear
DEFINITION OF MODEL UNCERTAINTY resistance; and Vcal is the calculated value of the punching
Due to the different basic assumptions adopted by the shear resistance.
punching shear strength models and the significant differ-
ences in the influencing factors considered, there is a large ASSESSMENT OF MODEL UNCERTAINTIES IN
deviation between the calculated results of each model and PUNCHING SHEAR RESISTANCE PREDICTION
the experimental test values. For example, Fig. 3 shows the Visual presentation of model uncertainty
influence of various factors on the punching shear resistance realization using histograms
of flat slabs. The calculation results of different models also The comparison and analysis of each model’s calcu-
have different responses at different intervals according to lated and tested resistance values are shown in Fig 4. The
the concrete strength. When the concrete strength is 20 ≤ 45-degree line indicates that the calculated value of the
fc′ < 40 MPa, the average value of punching shear resis- model is equal to the experimental test value, and the +40%
tance obtained according to the EC2 model is 1.05, and the and –40% lines indicate that the ratio of the model calculated
standard deviation is 0.05, indicating that the EC2 model value to the experimental test value is 1.4 and 0.6, respec-
predicts the punching shear resistance of flat slabs in this tively. The calculated value of the EC2 model deviates little
range for calculating concrete strength. GB50010, ACI 318, from the experimental value, and the scattered points are
and CSA A23.3 neglect the influence of the flexural rein- distributed near the 45-degree line. Some scattered points in
forcement ratio, implying that the design value of punching the GB50010 model are distributed outside the –40-degree
shear resistance of RC slabs is the same regardless of the line, which proves that the calculated value of the model is
flexural reinforcement ratio, which is contrary to reality. The lower than the actual punching shear resistance.
GB50010 model has a considerable complication when it To a large extent, both the design equations specified in
comes to effective depth: the measured value and the calcu- the design code and some of the product’s particular perfor-
lated value exhibit a substantial variance, regardless of the mance attributes used in the design are based on experimental

Fig. 3—Influence of various factors on model uncertainty.
campaigns. Characterization of the model uncertainty only be used to test the hypothesis of the normal distribution
random variable was achieved by comparing the observed and does not apply to other distribution types.
punching shear strength acquired from physical testing of The K-S test can effectively determine the probability and
typical punching shear resistance predictions from ACI 318, statistical information of the uncertainty of each calculation
EC2, MC2010, GB5001015, CSA A23.3, and prEC2. The mode, including the distribution type, mean, and standard
database of 452 experiments described in the Appendix was deviation. When the significance level (α = 0.05) and the
used to predict resistance for each analogous test. The model specimen capacity is 452, the critical value of the K-S test is
uncertainty observations are condensed in the histogram for 0.059.25 When the maximum absolute value of the difference
simple interpretation. The histograms of the model uncer- between the theoretical distribution function of the calculated
tainty of each punching shear resistance are shown in Fig. 5. model uncertainty coefficient and the cumulative frequency
The histogram widths show the degree of scattering in the function (usually called the D value) is less than the critical
model uncertainty. The fitting-probability density curves of value of the K-S test, the distribution is not rejected.
the normal distribution and the lognormal distribution are
concluded (Fig. 5). Statistics analysis of model uncertainty
Table 2 and Fig. 6 show the statistical characteristics of
Kolmogorov-Smirnov test for distribution model uncertainty related to six punching shear strength
discrimination models (ACI 318, EC2, MC2010, GB50010, CSA A23.3,
Based on the measured data of the 452 groups of RC slabs and prEC2) using the experimental database provided earlier.
subjected to punching shear capacity in the Appendix, 452 The basic parameters of these two distributions and the K-S
samples of the uncertainty of the calculation mode of each test D value are shown in Table 3. N(μ, σ) and LN(λ, ξ) repre-
punching shear resistance model can be calculated, and sent normal distribution and lognormal distribution respec-
the K-S test24 can be used to determine the punching shear tively. μ and σ are the parameters of the normal distribution,
capacity. The K-S test is a test method that compares the and λ and ζ are the parameters of the lognormal distribution.
frequency distribution f(x) with the theoretical distribution Taking the distribution type with the smallest D value as the
g(x). It can be tested when the data distribution is uncer- optimal distribution of each model value and checking the
tain. The use of the data is relatively complete and does not D value according to the K-S test in Table 3, it can be deter-
depend on the location of the data mean. Furthermore, the mined that the k values of the ACI, GB, and CSA models
K-S test is largely unrestricted by the distribution type and obey the normal distribution. On the contrary, the k values
has a wide range of applications. However, the t-test can of the other models obey the lognormal distribution. The

Fig. 4—Comparison of experimental value Vt and predicted punching shear strength Vcal. (Note: 1 kN = 1000 N = 224.8 lb.)
fitting-probability density curves of normal and lognormal situation, and the safety margin of the GB, ACI, and CSA
distribution in each graph are summarized in Fig. 7. The models is extremely high.
fitting curves of the prEC2 model under the two distributions
are significantly higher and more compact. The EC2 model Model uncertainty partial factor related to
is also better able to predict the punching shear resistance of punching shear model
the slab-column connections. Epistemic uncertainty can be improved by obtaining
According to the analysis results of the 452 sets of test new data or refining the model.26 The following causes of
data, the percentages of the ACI, EC2, MC2010, GB, CSA, uncertainty can be identified in the context of the situation
and prEC2 models with calculation uncertainty θ > 1 of the mentioned previously: the uncertainty of material property
total number of specimens are 76.5%, 55.3%, 54.6%, 85.8%, constants and load values, uncertain modeling mistakes
62.2%, and 17.7%, respectively. It shows that the calcula- caused by the physical sub-models used to represent the
tion results of the EC2 model are most similar to the actual derived variables being chosen incorrectly, uncertainty
errors caused by measurement observations, and so on. The

Fig. 5—Histograms of model uncertainty and probability density function.
uncertainty associated with the models’ validity and correct- in the structure have been found.27 To account for the uncer-
ness should be taken into account during evaluation, either tainty associated with resistance models, the partial factor
by using suitable partial factors in deterministic verifications γRd might be used
or by using additional random variables in reliability studies
to indicate model uncertainties. The updated design resis- γRd = 1/[μθ(1 – αR ∙ β ∙ Ωθ)]
tance R is assumed to be obtained by the following general (for a normal random variable) (3)
model
R = f(X, a)/(γM ∙ γRd) (2) γRd = 1/[λθexp(–αR ∙ β ∙ υθ)]

(for a lognormal random variable) (4)
where f(X, a) is the characteristic resistance for punching
shear failure; γM is the partial safety factor for material and
geometrical uncertainties; and γRd is the partial safety factor u
(5)
related to model uncertainty.
1 2
In actual engineering, it is necessary to modify the calcu-
lation model to reduce the deviation between the calculation
model and the actual situation. Updated design values for ( )
υθ = ln 1 + Ωθ 2 (6)
each random variable is a more realistic method. A proba-
bilistic analysis may be used to compute the updated failure where αR is the first-order reliability method (FORM) sensi-
probability once updated distributions for the basic variables tivity factor (αR = 0.32)28; β is the target reliability index;

Table 2—Statistical characteristics of model uncertainty random variables
Model uncertainty symbol Mean Standard deviation Coefficient of variation
Model θ μθ σθ Ωθ
ACI θACI 1.283 0.408 0.318
EC2 θEC2 1.065 0.277 0.260
MC θMC 1.063 0.375 0.353
GB θGB 1.485 0.474 0.319
CSA θCSA 1.123 0.358 0.319
prEC2 θprEC2 0.863 0.191 0.222
Table 3—Basic parameters of each distribution and D value of K-S test

N(μ, σ) LN(λ, ξ) D
Model μ σ λ ξ N(μ, σ) LN(λ, ξ) Distribution
ACI 1.283 0.408 0.087 0.135 0.058 0.064 Normal
EC2 1.065 0.277 0.015 0.102 0.081 0.030 Lognormal
MC 1.063 0.375 0.000 0.151 0.088 0.055 Lognormal
GB 1.485 0.474 0.151 0.135 0.055 0.065 Normal
CSA 1.123 0.358 0.030 0.135 0.057 0.065 Normal
PrEC2 0.863 0.191 –0.074 0.090 0.077 0.033 Lognormal
Eurocode 2 (EC2) and the draft of the next generation of

the Eurocode 2 (prEC2) are regarded as the best predictors
of punching shear strength among all the models evaluated
in this research. Other models usually underestimate the
punching shear resistance of the slab-column connections.
2. EC2, Model Code 2010, and prEC2 punching shear
model uncertainty variables obey lognormal distribution,
while the other models’ uncertainty variables obey normal
distribution.
3. The suggested project is particularly universal because
it is independent of the random variable’s probabilistic
model. It is also appropriate for calibrating design formu-
lations using model uncertainty partial factors related to
punching shear models.
Fig. 6—Mean and coefficient of variation of model AUTHOR BIOS

Ming-Yue Jiang is a Doctoral Candidate in the Department of Civil Engi-
uncertainty. neering at Xi’an University of Architecture and Technology, Xi’an, Shaanxi,
China, where she received her BS and MS in 2017 and 2020, respec-
and μθ and Ωθ are the mean and coefficient of variation of the tively. Her research interests include punching shear mechanics of rein-
model uncertainty random variable, respectively. forced concrete slabs and nonlinear analysis of steel-concrete composite
Figure 8 demonstrates the relationship between model structures.
uncertainty factor γRd and the target dependability β, using Qing-Xuan Shi is a Professor in the Department of Civil Engineering at
EC2 as an example. The goal reliabilities in the figure encom- Xi’an University of Architecture and Technology, where he received his BS,
pass the target reliabilities for the ultimate limit states for MS, and PhD in 1984, 1987, and 2002, respectively. His research interests
include steel-concrete composite structures, design and analysis theory,
structures throughout a 50-year design working life period. and seismic behavior of concrete structures.
CONCLUSIONS Wang-Hu Zhao is a Master’s Degree Candidate in the Department of

Civil Engineering at Xi’an University of Architecture and Technology.
The model uncertainty related to the six design code func- He received his BS from Hebei University of Science and Technology,
tions for predicting punching shear failure are examined Shijiazhuang, Hebei, China, in 2020. His research interests include steel-
using a database of flat slabs with punching shear failure. concrete composite structures.
Based on the analysis results, the following conclusions can
ACKNOWLEDGMENTS
be drawn: The work presented in this paper was supported by the National
1. A good predictive model is a model with a mean close Natural Science Foundation of China (No. 51878540), which is gratefully
to 1 (low bias) and low standard deviation. As a result, acknowledged.

Fig. 7—Comparison of probability density curves.
3. Park, H.-G.; Choi, K.-K.; and Chung, L., “Strain-Based Strength Model
for Direct Punching Shear of Interior Slab–Column Connections,” Engi-
neering Structures, V. 33, No. 3, Mar. 2011, pp. 1062-1073. doi: 10.1016/j.
4. Alexander, S. D. B., and Simmonds, S. H., “Ultimate Strength of Slab-
Column Connections,” ACI Structural Journal, V. 84, No. 3, May-June
1987, pp. 255-261.
5. Johansen, K. W., Yield-Line Theory, Cement and Concrete Associa-
tion, London, UK, 1962, 181 pp.
6. Bažant, Z. P., and Cao, Z., “Size Effect in Punching Shear Failure of
Slabs,” ACI Structural Journal, V. 84, No. 1, Jan.-Feb. 1987, pp. 44-53.
7. Kueres, R.; Ricker, M.; Classen, M.; and Hegger, J., “Fracture
Kinematics of Reinforced Concrete Slabs Failing in Punching,” Engi-
neering Structures, V. 171, Sept. 2018, pp. 269-279. doi: 10.1016/j.
8. Hegger, J.; Ricker, M.; Ulke, B.; and Ziegler, M., “Investigations on the
Punching Behavior of Reinforced Concrete Footings,” Engineering Structures,
V. 29, No. 9, Sept. 2007, pp. 2233-2241. doi: 10.1016/j.engstruct.2006.11.012
9. 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.
10. Fédération internationale du béton, “fib Model Code for Concrete
Structures 2010 (MC2010),” fib, Lausanne, Switzerland, 2013, 434 pp.
Fig. 8—Variation of the partial factor γRd with β (α = 0.32). 11. Ghali, G., and Gayed, R. B., “Universal Design for Punching Resis-
tance of Concrete Slabs,” ACI Structural Journal, V. 116, No. 1, Jan. 2019,
NOTATION pp. 207-212. doi: 10.14359/51710866
b0 = control perimeter in distance 0.5d from face of loaded area 12. Schmidt, P.; Kueres, D.; and Hegger, J., “Contribution of Concrete
CRd,c = constant factor and Shear Reinforcement to the Punching Shear Resistance of Flat Slabs,”
ddg = size parameter describing failure zone roughness Engineering Structures, V. 203, Jan. 2020, Article No. 109872, 11 pp. doi:
dv = shear resisting effective depth 10.1016/j.engstruct.2019.109872
fc = compressive strength of concrete 13. EN 1992-1-1:2004/AC:2010/A1:2014, “Eurocode 2: Design of Concrete
ft = tensile strength of concrete Structures – Part 1-1: General Rules and Rules for Buildings,” European
kdg = coefficient for aggregate size Committee for Standardization, Brussels, Belgium, 2014.
kmax = factor limiting maximum punching shear resistance 14. PT1prEN 1992-1-1:2018-04, “Eurocode 2: Design of Concrete
kpb = punching shear gradient enhancement coefficient Structures - Part 1-1: General Rules for Buildings, Bridges and Civil Engi-
kψ = factor accounting for deformation of slab neering Structures - Third and Final Draft by the Project Team SC2.T1,”
kζ = factor accounting for influence of size effects European Committee for Standardization, Brussels, Belgium, 2018.
u0 = perimeter of loaded area 15. Rashwan, M. R.; Rashed, Y. F.; Mehanny, S. S. F.; and Mohareb, R. W.,
u1 = control perimeter at 2.0d from face of loaded area “Novel Warping-Included Punching Parameters for Interior Rectangular
α = significance level Columns in Flat Slabs,” Engineering Analysis with Boundary Elements,
αs = 4 for interior columns, 3 for edge columns, and 2 for corner V. 112, Mar. 2020, pp. 1-12. doi: 10.1016/j.enganabound.2019.11.009
columns 16. Rashwan, M. R.; Rashed, Y. F.; and Mehanny, S. S. F., “Theoret-
βc = ratio of long side to short side of column, concentrated load, or ical Envelop for the Punching Shear Formula,” Journal of Engineering and
reaction area Applied Sciences, V. 66, No. 1, Feb. 2019, pp. 91-107.
βh = factor accounting for influence of size effects 17. GB50010-2010, “Code for Seismic Design of Buildings,” China
γc = partial safety factor for concrete Architecture & Building Press, Beijing, China, 2010.
φc = resistance factor for concrete 18. CSA A23.3-14, “Design of Concrete Structures,” CSA Group,
μ m = control perimeter in distance 0.5d from face of loaded area Toronto, ON, Canada, 2014, 297 pp.
μ1 = control perimeter at 2.0d from face of loaded area 19. Einpaul, J.; Bujnak, J.; Fernández Ruiz, M.; and Muttoni, A., “Study
ψ = rotation of slab around loaded area on Influence of Column Size and Slab Slenderness on Punching Strength,”
ACI Structural Journal, V. 113, No. 1, Jan.-Feb. 2016, pp. 135-145. doi:
10.14359/51687945
REFERENCES 20. ACI Committee 318, “Building Code Requirements for Structural
1. Yankelevsky, D. Z., and Leibowitz, O., “Punching Shear in Concrete
Concrete (ACI 318-14) and Commentary (ACI 318R-14),” American Concrete
Slabs,” International Journal of Mechanical Sciences, V. 41, No. 1, Jan.
Institute, Farmington Hills, MI, 2014, 520 pp.
1999, pp. 1-15. doi: 10.1016/S0020-7403(97)00086-6
21. Muttoni, A., and Fernández Ruiz, M., “The Levels-of-Approximation
2. Nielsen, M. P., Limit Analysis and Concrete Plasticity, second edition,
Approach in MC 2010: Application to Punching Shear Provisions,” Struc-
CRC Press, New York, 1999, pp. 306-328.
tural Concrete, V. 13, No. 1, 2012, pp. 32-41. doi: 10.1002/suco.201100032

22. Ospina, C. E.; Birkle, G.; Widianto, W.; Wang, Y.; Fernando, S. R.; 25. Modarres, M.; Kaminskiy, M. P.; and Krivtsov, V., Reliability Engi-
Fernando, S.; Catlin, A. C.; and Pujol, S., “NEES: ACI 445 Punching Shear neering and Risk Analysis: A Practical Guide, third edition, Taylor &
Collected Databank,” American Concrete Institute, Farmington Hills, MI, Francis Group, Boca Raton, FL, 2017, 522 pp.
2015. 26. EN 1990:2002, “Eurocode – Basis of Structural Design,” European
23. Yi, W. J.; Hong, F.; and Peng, J., “Experimental Study on Shear Resis- Committee for Standardization, Brussels, Belgium, 2002, 119 pp.
tance of Reinforced Concrete Slab-Column Joints,” Building Structures, 27. Der Kiureghian, A., and Ditlevsen, O., “Aleatory or Epistemic? Does
V. 46, No. 15, 2016, pp. 11-18+46. it Matter?” Structural Safety, V. 31, No. 2, Mar. 2009, pp. 105-112. doi:
24. Massey, F. J., Jr., “The Kolmogorov-Smirnov Test for Goodness of 10.1016/j.strusafe.2008.06.020
Fit,” Journal of the American Statistical Association, V. 46, No. 253, Mar. 28. Olalusi, O. B., and Kolawole, J. T., “Probabilistic Safety Analysis
1951, pp. 68-78. doi: 10.1080/01621459.1951.10500769 of Prestressed Concrete Beams Failing Due to Web-Crushing,” Structures,
V. 15, Aug. 2018, pp. 370-377. doi: 10.1016/j.istruc.2018.08.004
APPENDIX
Table A1—Source of punching shear data
h0, mm c, mm ρ, % fc′, MPa Vt, kN
Lower Upper Lower Upper Lower Upper Lower Upper Lower Upper
Author Year n limit limit limit limit limit limit limit limit limit limit
Yi et al.23 2016 9 150 150 250 250 0.860 1.730 23.46 55.39 443 860
Parra 2011 1 165 165 152 152 0.752 0.752 45.00 45.00 600 600
Widianto et al. 2009 2 126 126 406 406 0.493 1.046 28.10 31.40 311 401
Ozden et al. 2006 6 100 100 200 200 0.727 2.264 20.00 75.00 188 489
Guandalini 2005 9 96 210 130 260 0.219 1.496 27.70 40.50 118 1024
Birkle 2004 3 124 260 250 350 1.100 1.539 31.40 36.20 483 1046
Timm 2003 3 172 246 175 250 1.186 1.242 32.80 40.70 668 1356
Oliveira et al. 2003 15 106 109 120 120 0.962 0.990 54.00 67.00 240 446
Ospina et al. 2003 1 120 120 250 250 0.838 0.838 36.80 36.8 365 365
Ospina 2001 1 109 109 400 400 0.922 0.922 29.80 29.80 541 541
Li 2000 4 100 300 200 200 0.758 0.971 39.40 39.40 330 1381
Matthys and Taerwe 2000 17 86 126 80 230 0.188 3.764 26.30 96.70 142 347
McHarg et al. 2000 1 125 125 225 225 0.974 0.974 30.00 30.00 306 306
Ozawa et al. 2000 4 60 75 100 100 1.100 1.950 29.10 39.00 120 180
Ghannoum 1998 3 127 127 225 225 0.958 0.958 37.20 67.10 301 443
Sistonen et al. 1997 10 170 177 201 901 0.434 1.117 24.10 32.70 478 1111
Hallgren 1996 5 194 201 250 250 0.625 1.257 85.70 112.90 889 1041
Ramdane 1996 15 98 102 150 150 0.581 1.272 33.60 127.00 169 405
Banthia et al. 1995 1 55 55 100 100 0.350 0.350 41.00 41.00 60 60
Urban 1994 13 90 104 80 320 0.757 1.503 18.30 36.80 176 360
Theodorakopoulos and
1993 4 100 100 100 200 0.353 0.530 43.10 45.80 137 191
Swamy
Tomaszewicz 1993 8 88 200 100 150 1.745 2.618 70.20 119.00 330 1450
Alexander and Simmonds 1992 6 107 134 200 200 0.499 0.625 26.00 35.60 257 343
Marzouk and Hussein 1991 17 70 125 150 300 0.493 2.377 30.00 80.00 178 645
Gardner 1990 30 33 123 51 203 0.448 7.305 13.20 52.10 24 357
Lovrovich and McLean 1990 5 83 83 102 102 1.752 1.752 40.00 40.00 129 479
Lunt 1998 1 112 112 250 250 0.105 0.105 29.30 29.30 207 207
Rankin and Long 1987 27 35 54 100 100 0.423 1.995 34.80 47.10 29 126
Regan 1986 41 64 200 54 250 0.748 1.625 11.90 53.30 105 1099
Schaefers 1984 1 113 113 210 210 0.834 0.834 23.10 23.10 280 280
Swamy and Ali 1982 1 100 100 150 150 0.557 0.557 46.60 46.60 198 198
Pralong et al. 1979 1 171 171 300 300 1.176 1.176 27.10 27.10 626 626
Ladner et al. 1977 4 80 80 100 320 1.795 1.795 37.50 41.90 183 324
Marti et al. 1977 1 143 143 300 300 1.483 1.483 43.20 43.20 628 628
Criswell 1974 8 121 127 254 508 0.750 1.560 26.60 35.70 273 580

Table A1 (cont.)—Source of punching shear data
h0, mm c, mm ρ, % fc′, MPa Vt, kN
Lower Upper Lower Upper Lower Upper Lower Upper Lower Upper
Author Year n limit limit limit limit limit limit limit limit limit limit
Long and Masterson 1974 2 30 32 76 76 0.980 1.770 27.70 29.50 31 34
Ladner et al. 1973 1 109 109 226 226 1.201 1.201 39.60 39.60 362 362
Nylander and Sundquist 1972 4 96 201 120 240 0.711 1.175 26.80 32.00 183 661
Vanderbilt 1972 15 38 38 76 305 1.309 2.618 19.90 27.60 39 114
Hawkins et al. 1971 3 117 121 114 152 0.752 0.860 25.90 29.50 314 320
Nightingale 1970 11 38 38 114 254 0.620 1.484 30.20 46.30 56 83
Corley and Hawkins 1968 2 111 111 203 254 1.008 1.512 18.70 20.40 266 334
Mowrer and Vanderbilt 1967 2 51 51 152 152 1.670 2.200 12.40 15.20 113 123
Base 1966 20 102 124 203 203 1.000 1.900 14.00 41.40 242 531
Manterola 1966 12 107 107 100 450 0.367 1.057 24.20 39.70 165 397
Yitzhaki 1966 15 78 83 119 333 0.526 2.020 12.20 27.00 98 307
Taylor and Hayes 1965 8 61 64 51 152 0.873 1.851 23.00 32.40 64 136
Moe 1961 7 114 114 152 457 1.050 1.519 20.80 27.60 311 433
Kinnunen and Nylander 1960 12 117 128 150 300 0.529 2.290 30.70 39.30 255 540
Base 1959 9 60 64 100 100 0.690 2.920 32.80 37.30 82 118
Rosenthal 1959 4 60 80 100 229 0.690 2.920 17.90 37.30 82 245
Elstner and Hognestad 1956 25 114 121 254 356 0.496 3.806 12.80 50.50 178 578
Forssell and Holmberg 1946 7 101 111 140 140 0.637 0.700 15.40 15.40 172 198
Richart and Kluge 1939 14 130 130 51 356 0.724 0.724 26.20 26.20 151 249
Graf 1938 1 271 271 300 300 1.037 1.037 16.40 16.40 1165 1165
Note: 1 mm = 39.4 × 10–3 in.; 1 MPa = 145 psi; and 1 kN = 1000 N = 224.8 lb.

Title No. 119-S37
Vehicle Collision with Reinforced Concrete Columns

Wrapped with Fiber-Reinforced Polymer Composites
by Tao Liu, Lin Chen, Jinjun Xu, Cristoforo Demartino, and Thomas H.-K. Kang
The dominant failure mode for existing reinforced concrete (RC)

columns exposed to vehicle collisions is generally shear. Fiber-
reinforced polymer (FRP) composites have been widely used to
retrofit RC structures. This study numerically investigates FRP’s
effectiveness in reducing potential collision damage and preventing
shear failure in full-scale RC pier columns retrofitted with FRP
wraps under vehicle collision. Developed finite element (FE)
models of plain and FRP-wrapped RC columns were well vali-
dated by comparing numerical results against laboratory impact
test results. The validated FE modeling method was extended to
build full-scale unstrengthened and FRP-wrapped RC pier column
models. A parametric study was performed examining the effect of
vehicle velocity and layer number and types of FRP wraps, as well
as fiber orientation, on the structural responses. Numerical results
showed that an increase in the number of layers of FRP wraps
could significantly reduce column damage and displacements but
has less influence on the dissipation of impact force, shear force,
and moment of the columns. Two dimensionless damage indexes
were proposed for the assessment of shear failure.
Keywords: dynamic response; fiber-reinforced polymer (FRP);

finite element (FE) method; reinforced concrete (RC) pier column;
vehicle collision.
INTRODUCTION
Vehicle collisions with reinforced concrete (RC) columns
supporting overcrossings, undercrossings, separations,
interchanges, and viaducts have recently drawn much atten-
tion due to their non-negligible occurrence probability
and large economic and life losses. Vehicular impact can
induce large damage or even structural failure of the pier
column, resulting in potential superstructure collapse due to Fig. 1—Images of RC columns subjected to vehicular colli-
loss of the vertical member. For instance, in April 2009, a sion: (a) after impact; and (b) after repair.
multi-column bent supporting structure of the overpass at
the intersection between the G4 Expressway and the S322 Do et al. 2018; Chen et al. 2020), which can potentially
Highway in Hunan Province in China was hit by a cement induce the sudden loss of vertical bearing capacity and the
tanker truck (Fig. 1(a)). The collision led to the separation of failure of the superstructure. Hence, there is a growing need
the impacted column from the foundation and bent cap, the to implement effective retrofit strategies to improve the
subsequent failure of the adjacent columns, and, finally, the impact-shear resistance of RC pier columns against vehicle
local collapse of the superstructures itself; it also caused two collision. Degradation of long-term performances of RC
deaths and closure of the road for more than 60 days with columns may increase the demand for such retrofit.
very large direct and indirect costs (Chen and Xiao 2012). On the other hand, fiber-reinforced polymer (FRP)
The repaired structure is shown in Fig. 1(b). Similar traffic composites have been extensively used for retrofitting RC
collisions were largely documented in the literature (Buth et structural members due to their excellent performances, such
al. 2010; Chen and Xiao 2012; Chen et al. 2021). Therefore, as lightweight, high-strength, corrosion resistance, and easy
attention should be paid to the development of the design installment. Several studies found that FRP strengthening
and retrofit of RC bridge columns subjected to vehicular ACI Structural Journal, V. 119, No. 2, March 2022.
impact loads accounting for dynamic effects to reduce this MS No. S-2021-068.R1, doi: 10.14359/51734335, received August 14, 2021, and
large risk. Institute. All rights reserved, including the making of copies unless permission is
RC pier columns subjected to vehicular impacts were obtained from the copyright proprietors. Pertinent discussion including author’s
demonstrated to be prone to shear failure (Buth et al. 2011; is received within four months of the paper’s print publication.

Fig. 2—Details of numerical modeling: (a) impact test setup; and (b) exemplified FE model of FRP-strengthened column.
(Note: Units in mm: 1 mm = 0.039 in.)
techniques can improve shear capacity and impact resis- RC columns with and without carbon FRP (CFRP) wrap-
tance of RC beams, turning brittle shear failure into a flex- ping under horizontal impact reported by Xu et al. (2020).
ural one (Ibrahim Ary and Kang 2012; Kang and Ibrahim Figure 2(a) shows part of the test setup. The tests refer to
Ary 2012; Pham and Hao 2016; Liu and Xiao 2017; Huo et four columns with a circular cross section 330 mm (13.0 in.)
al. 2018; Liu et al. 2020). There are only a few studies inves- in diameter and a height of 1700 mm (66.9 in.) fixed at the
tigating the strengthening effect of applying FRP compos- base and free at the top—that is, cantilever boundary condi-
ites to improve the resistance of RC columns against impact tion. The average unconfined cylinder compressive strength
loading (Sha and Hao 2015; Fan et al. 2018, 2020; Xu et al. for all the columns was 30 MPa (4.4 ksi). The transverse stir-
2020; Al-Bukhaiti et al. 2021; Liu et al. 2021; Zhou et al. rups for all four columns had the same diameter of 6.5 mm
2021). These studies revealed that the FRP composites can and yielding strength fy = 427 MPa (61.9 ksi), while two
control shear failure, significantly reducing deformation and different spacings of 100 mm (3.9 in.) (C2H0 and C2H2)
damage of RC columns. and 330 mm (13.0 in.) (C3H0 and C3H4) were used along
The current paper aims to investigate the effectiveness the entire height. Sixteen longitudinal steel bars were set
of wrapping FRP composites around a full-scale RC pier uniformly around the circumference of all the columns, and
column, which collapsed in 2009, to improve their resis- the diameter was 8 mm (0.3 in.) (fy = 436 MPa [63.2 ksi])
tance against vehicular collision. Nonlinear finite element for C2H0 and C2H2, and 12 mm (0.5 in.) (fy = 471 MPa
(FE) program LS-DYNA (2010) was used to simulate the [68.3 ksi]) for C3H0 and C3H4. Compared with the
FRP composites-wrapped RC pier columns against vehicle unstrengthened columns (C2H0 and C3H0), the strength-
collisions. First, the proposed model was validated against ened columns (C2H2 and C3H4) have CFRP wraps up to a
laboratory impact tests of RC columns with and without FRP height of 900 mm (35.4 in.) above the base with two (C2H2)
wrapping (Xu et al. 2020). After validation, the FE modeling or four (C3H4) layers. The CFRP wraps consisted of the
method was used to simulate an actual RC pier substruc- epoxy resin adhesive and unidirectional carbon fiber sheets
ture (three columns and bent cap) under vehicular impact. with a nominal thickness of 0.167 mm (0.007 in.), a tensile
Finally, an evaluation of the effectiveness of wrapping FRP strength of 3494 MPa (506.6 ksi), and a tensile modulus of
composites around columns to improve impact resistance 240 GPa (34,800 ksi). The epoxy resin adhesive achieved
is provided, where two dimensionless damage indexes a static tensile strength of 40 MPa (5.8 ksi), compressive
are proposed. strength of 70 MPa (10 ksi), and shear strength of 20 MPa
(2.9 ksi). Its tensile modulus of elasticity and tensile elonga-
RESEARCH SIGNIFICANCE tion were 4 GPa (580 ksi) and 1.3%, respectively. The CFRP
The dynamic performance of FRP-retrofitted RC columns wraps were placed with the fiber direction parallel to the
subjected to vehicle collision is still unclear. The main transverse direction of the column. The test truck equipped
purpose of this study is to quantitatively assess the effec- with an instrumental rigid hammer at the front impacted four
tiveness of wrapping FRP composites around RC columns columns (C2H0, C2H2, C3H0, and C3H4) with a velocity
to improve the impact resistance against vehicular collision. and mass of 4.5 m/s (177 in./s) and 1582 kg (3480 lb),
The following aspects are investigated or proposed: 1) the respectively. The impact height was at approximately 0.4 m
effect of FRP-wrapping variables on the dynamic response/ (15.7 in.) from the base to the center of the frontal hammer.
mode of RC columns; 2) the effect of FRP properties on Other details of the laboratory tests can be found in the liter-
the damage and deformation of the retrofitted columns; ature (Demartino et al. 2017; Xu et al. 2020).
and 3) damage indexes to assess the shear damage of RC Three-dimensional FE models were developed to simu-
columns under vehicle collision. late the experimental tests, and one example of strengthened
columns is shown in Fig. 2(b). The RC column, column base,
FINITE ELEMENT MODEL VALIDATION and anchoring steel bolts were modeled based on their real
The validation of the explicit FE modeling was performed geometry, but the test truck including the frontal hammer
against four tests (C2H0, C2H2, C3H0, and C3H4) of was simplified to a solid block with an identical total mass

Fig. 3—Comparison between simulation results and test results: (a) energy status in FE results; (b) impact force-time histories;
(c) displacement-time histories at 0.4 m (15.7 in.) impact height; and (d) horizontal displacement distribution along height of
column. (Note: 1 kJ = 0.08 ton; 1 kN = 0.225 kip; 1 mm = 0.039 in.; and 1 m = 3.28 ft.)
and the same frontal surface of the hammer. CFRP wraps experimental results, despite the loss of test data of the three
were modeled by creating a thin tube surrounding the periph- columns (C2H0, C2H2, and C3H0). As to the horizontal
eral surface of the column. More details of FE modeling are displacement distribution along the column height presented
presented in Appendix A*. Figure 3(a) presents the energy in Fig. 3(d), the FE model can well capture the displace-
status in FE results. The hourglass energy is approximately ment distribution of the columns before the failure, though
10% of the internal energy. The simulated impact force- some discrepancies exist thereafter. Figure 4(a) compares
time histories are compared with the experimental results, the experimental and simulated concrete damage evolution
as shown in Fig. 3(b). Despite some differences in the process of column C3H0, which indicates that the simulated
first peak values (approximately 10%), it is clear that the damage is quite consistent with the associated test results.
impact force-time histories predicted by the FE simulations Concrete damage states of all four columns at 50 millisec-
follow the patterns recorded in the experimental tests. As onds are compared in Fig. 4(b), which shows that shear
shown in Fig. 3(c), the simulated displacement-time histo- failure and flexural-dominant damage of columns are well
ries at a height of 0.4 m (15.7 in.) also agree well with the captured by the FE models.
Based on these comparisons, the developed FE model
*
The Appendix is available at www.concrete.org/publications in PDF format, is considered capable of predicting the impact response of
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
unstrengthened and FRP-wrapped RC columns under hori-
time of the request. zontal impact loading.

Fig. 4—Comparison of concrete damage between FE simulations and impact tests: (a) evolution process of column C3H0
during impact; and (b) damage states of columns C2H0, C2H2, C3H0, and C3H4 at 50 milliseconds.
RC PIER COLUMNS UNDER VEHICLE 1700 x 1100 mm (66.9 x 43.3 in.), as depicted in Fig. 5. The
COLLISION—SIMULATION height of the 1000 mm (39.4 in.) diameter columns above
The previously validated modeling method was extended ground was 5.4 m (212.6 in.). The steel reinforcement for
to simulate a full-scale RC pier substructure (three columns each column comprised 16 vertical bars having a diameter
and bent cap), which collapsed in 2009, subjected to a vehi- of 20 mm (0.8 in.) set uniformly along the circumference,
cle-column collision. which were confined by 8 mm (0.3 in.) diameter trans-
verse stirrups with a spacing of 200 mm (7.9 in.) along the
RC pier model column height. The gravity load associated with the super-
The full-scale RC pier model (three columns and bent cap) structure was simplified as a uniform load acting vertically
was developed according to the prototype bridge pier at the downward applied to the top of the bent cap. A prespecified
intersection of Highway S322 and the G4 Expressway in axial compression ratio of 0.1 (that is, the ratio of the axial
Hunan Province in China, with which a truck collided, as force to the product of the cross-section area of the columns
shown in Fig. 1(a) (Chen and Xiao 2012). Some simplifica- and concrete compressive strength) was used for the pier
tions and assumptions were made during the modeling. In columns to determine the gravity load.
particular, the bent cap is not connected with the superstruc- The modeling was similar to that introduced in the former
ture and is simply receiving the vertical load, and the foun- section. It should be noted that the impacted column was
dation is modeled as fully constrained. Due to the absence meshed at an average element size of approximately 25 mm
of the material properties (that is, the strengths of concrete (1.0 in.), but an average 50 mm (2.0 in.) element size was
and reinforcement) at the time of the accident, their nominal used for the other two columns, the bent cap, and the foun-
values that appeared in the blueprints of the bridge were dation to reduce computational cost. Both the bent cap and
used. The unconfined compressive strength of the concrete the foundation were assumed to have elastic properties of
was assumed to be 30 MPa (4.4 ksi). The yield strength concrete. The simplified three-dimensional FE model of the
of the vertical steel and hoops was assumed to be 335 and RC bridge pier is presented in Fig. 5.
235 MPa (48.6 and 34.1 ksi), respectively.
Similar to the RC pier model developed by Chen et al. Vehicle model
(2020), three circular RC columns were spaced at 5711 mm Due to the lack of detailed information about the truck
(224.8 in.) on the center with a bent cap cross-section size of during the collision accident reported by Chen and Xiao

Fig. 5—FE models of RC bridge pier and truck vehicle. (Note: Units in mm: 1 mm = 0.039 in.)
(2012), a generally accepted medium-duty truck model was the friction coefficient of the contact surface between the
employed in this study, as shown in Fig. 5. This truck model, vehicle and the column was taken as 0.3. Gravity loads from
developed by the U.S. National Crash Analysis Center the bridge superstructure were still acting during the second
(NCAC), was initially designed to computationally evaluate phase, but the mass damping was eliminated. The total simu-
roadside safety hardware. Miele et al. (2005) and Mohan lation duration of the two phases is 0.5 seconds.
et al. (2007) optimized and validated this model against two
impact tests with concrete barriers. Further, it was also used Simulation results
by El-Tawil et al. (2005) and other researchers to simulate To induce failure of the impacted column, the vehicle
the vehicle-pier collision scenarios. model with sufficient initial kinetic energy (velocity and
The truck engine, including clutch and gearbox, and truck total mass) is needed. After several trials, the case for the
cargo were both modeled using solid elements with elastic vehicle model traveling at a velocity of 100 km/h (62.1 mph)
properties. The engine weighs 840 kg (1852 lb), while the with a gross mass of 8 ton (17,640 lb) was selected.
density of the cargo is adjustable to achieve various truck The numerically simulated damage of the pier model at the
gross masses. The remaining mass of the truck, excluding time point of 0.5 seconds after the collision was compared
the engine and cargo, is approximately 4430 kg (9768 lb). with the real crash photo, as shown in Fig. 6. It can be
The cargo is assumed fully bonded to the bottom of the truck observed from Fig. 6 that the failure pattern between them is
container, which was made of 1.7 mm (0.07 in.) thick steel quite similar: the fracture of the impacted column 1 (C1) and
plates modeled with shell elements and was connected to the the local collapse of the upper bent cap, along with damage
chassis through Z-shaped steel beams, as shown in Fig. 5. at the top of the adjacent column 2 (C2). The similarity in
The entire truck model contained approximately 35,000 terms of the failure pattern cannot be considered a validation
elements and its further details can be found in the liter- of the proposed FE model but can support the good direction
ature (Miele et al. 2005; Mohan et al. 2007). It should be of the prediction.
noted that the nominal impact height of the truck model is The impact process between the truck and the RC pier at
approximately 1050 mm (41.3 in.) and 1350 mm (53.1 in.) critical time points is depicted in Fig. 7. At 0.123 seconds,
off the ground for engine and cargo impacts, respectively. corresponding to the moment of peak impact force induced
For simplicity, an average of 1200 mm (47.2 in.) is herein by engine impact, concrete damage is concentrated at the
adopted as the single nominal impact height off the ground local impacted zone. Here, shear cracks develop from the
(Chen et al. 2016). impact zone at an inclination angle of 45 degrees to the
bottom, while flexural cracks appear at the rear of the impact
Simulation process surface. At 0.15 seconds, more cracks develop from the
The simulation process was divided into two phases; bottom to the top of the impacted column. Subsequently,
Phase I was gravity application and stabilization. For the concrete damage is distributed along the entire height of the
first 0.05 seconds, gravity loads from the superstructure were impacted column, and shear failure occurs near the bottom
applied to the top of the bent cap at a steady rate. Gravity at 0.2 seconds. Thereafter, the shear failure of column C1
loads remained constant for subsequent sequences, but the worsens and the adjacent columns C2 and C3 begin to expe-
column would vibrate due to the initial quick loading. Thus, rience damage due to compression at the top, especially
mass damping was included to have the column stabilize column C2.
quicker. Phase II was truck impact. After gravity applica-
tion and stabilization (approximately 0.1 seconds later), the PARAMETRIC ANALYSES OF FRP-WRAPPED RC
truck hits the column. Automatic surface-to-surface contact PIER COLUMNS UNDER VEHICLE COLLISIONS
type was used to simulate the collision between the vehicle Parametric analyses of the RC pier column model used
and the column. In accordance with El-Tawil et al. (2005), in the previous section were conducted with and without

Fig. 6—Damage comparison of RC pier between simulation result and post-accident result.
Fig. 7—Impact process of truck collisions with unstrengthened and CFRP-wrapped RC pier columns: (a) case C0-M8-V100;
and (b) case C4-M8-V100.

FRP wrapping under different impact scenarios. Only the of 1.2 m (47.2 in.) are compared in Fig. 8. From Fig. 8(a),
column (C1) that was directly impacted was strengthened it is observed that the duration period of impact force lasts
for the FRP-wrapped scenarios, while the other two adjacent approximately 0.2 seconds after the truck collides with
columns (C2 and C3) were not wrapped. The arrangement of the RC column. Three peaks with different amplitudes are
the FRP-wrapping scheme was set along the entire height of primarily depicted in all the curves. The first peak is induced
column C1, as shown in Fig. 5. by the impact of the bumper at the front of the truck, while
the second, along with greater amplitude, is caused by
Parametric details the impact of the engine. The third peak is due to cargo
The main parameters considered include vehicle velocity impact. As depicted in Fig. 8, the maximum impact force
and layer number, type, and arrangement orientation of and displacement increase drastically with the increase of
FRP wraps. For analysis control, the geometry and mate- the truck impact velocity. When velocity is kept constant, an
rial properties of the RC pier column model were kept increase in cargo mass makes little difference in the displace-
unchanged. The axial compression ratio was set at 0.1 and ment but results in an increase of the third peak impact. For
kept constant. The total mass M of the vehicle model with the case of C0-M8-V100, the impact force has no visible
a truck gross mass of 8 ton (17,640 lb) was used for all the third peak due to shear failure of the RC column and insuffi-
strengthened columns, while only two cases of unstrength- cient residual resistance against the cargo impact.
ened columns considered the increase of total mass to 12 The concrete damage states of unstrengthened RC pier
and 16 ton (26,460 and 35,280 lb) achieved by increasing columns at 0.3 seconds are shown in Fig. 9. For cases
the cargo mass. As a key parameter considered herein, the with the same total mass of 8 ton (17,640 lb), an increase
initial vehicular velocity V varied from 80 to 140 km/h (49.7 in vehicle velocity led to a change in the type of damage
to 87.0 mph). Hence, a range from 1975 to 6049 kJ (163.9 to to the impacted column. For velocities lower than 80 km/h
502.0 ton) was reached for the initial vehicle kinetic energy (49.7 mph), local cracking at the collision zone and flex-
Ek (Ek = 0.5MV2), and from 178 to 356 kN·s (40 to 80 kip·s) ural cracks at the opposite zone were noted. At a higher
for the initial vehicle momentum Mk (Mk = MV). velocity of 100 km/h (62.1 mph) up to 120 km/h (74.5 mph),
Details for the parametric analysis are summarized in shear-dominant failure occurred with complete fracture
Table 1. There are six series for all impact cases. The former along with a local collapse of the superstructure. In addi-
four series were defined in terms of the layer number of tion, adjacent columns were subject to severe compressive
unidirectional CFRP sheets, that is, series C0 without CFRP, crushing damage near the top of the columns after the failure
series C2 with two layers (0.334 mm [0.013 in.] thickness) of the impacted column, because the vertical loads originally
of CFRP, series C4 with four layers (0.668 mm [0.026 in.] born by the impacted column were transferred to adjacent
thickness), and series C8 with eight layers (1.336 mm columns, resulting in the crushing damage due to insuf-
[0.052 in.] thickness). One additional C2O series with two ficiency of axial load-bearing capacity. When the impact
layers of CFRP sheets was focused on the layer arrange- velocity was kept at the lower value of 80 km/h (49.7 mph),
ment orientation. The orientation is in a counterclockwise an increase in the total truck mass from 8 to 16 ton (17,640
direction from the Z-axis of column C1, shown in Fig. 5; for to 35,280 lb) caused a little worse local damage adjacent to
example, a value of 0 degrees means an arrangement along the collision zone of column C1.
the axial direction of the column for longitudinal strength- Relationships between the maximum displacement at the
ening, 90 degrees for hoop arrangement around the column height of 1.2 m (47.2 in.) for column C1 and the dynamic
for transverse strengthening, and 45 or 135 degrees indicates response values are depicted in Fig. 10. Here, displacements
diagonal arrangement. The last G-series was designed to appear more sensitive to variation in the vehicular impact
investigate the effect of different FRP types, such as glass velocity, and the growth relationship appears to change almost
fiber-reinforced polymers (GFRP). The properties of CFRP exponentially. If the vehicular velocity remains constant and
and GFRP used in the FE model are listed in Table 2. sufficiently low, an increase in the vehicular kinetic energy
The impact scenarios are numbered as follows for ease of and momentum achieved by the cargo mass may not lead to
reference. For example, the label of C2-M8-V100 represents worsening damage of the RC column. The maximum impact
an RC pier column with two layers of CFRP wraps impacted force presents a strong relationship with the damage state
by the truck model with a total mass of 8 ton (17,640 lb) and the displacement of the impacted column.
(M8) at a speed of 100 km/h (62.1 mph) (V100). To distin- From these analyses, it is found that for a certain type of
guish the case considering the variation of fiber orientation truck with an almost similar engine mass, its velocity plays
of CFRP wraps, a number is incorporated into the labels of a more critical role in the collision with the pier column than
the C2O series; for example, C2O1 means the arrangement the total mass. Although the cargo impact can be capable of
of two CFRP layers in a 0/0 degrees orientation, and C2O2 causing more severe damage to columns in some collision
for a 0/90 degrees orientation, as well as C2O3 for a 135/45 scenarios (Chen et al. 2020), its effect generally depends on
degrees orientation, where the former orientation represents the cargo mass and stiffness and its connection to the truck
the internal layer, and the rear is the outside layer. chassis. The latter parameters can be quite different for
trucks with various functions. These factors make it complex
Analysis results and difficult to study the effect of cargo impact. Besides, the
Effect of vehicular velocity and mass—The dynamic case C0-M16-V80 with a large cargo mass showed a similar
time histories of impact force and displacement at a height

Table 1—Parametric details and results
FRP wraps
Ek, Mk, β, D m, Vm, Failure
Series Cases kJ kN·s n degrees mm Pm, kN kN Vf, kN Vn, kN Ek/VnD Dm/L Pm/Vn Vm/Vn mode
C0-M8-V80 1975 178 — — 4.4 7285 5639 — 2320 0.85 0.004 3.14 2.43 NF
C0-M8-V90 2500 200 — — 5.5 9396 6516 — 2320 1.08 0.005 4.05 2.81 NF
C0-M8-V100 3086 222 — — 309.1 *
10,787 6876 — 2320 1.33 0.258 4.65 2.96 SF
C0
C0-M8-V110 3736 244 — — 671.0 *
12,767 7267 — 2320 1.61 0.559 5.50 3.13 SF
C0-M12-V80 2963 267 — — 5.3 7287 5603 — 2320 1.28 0.004 3.14 2.42 NF
C0-M16-V80 3951 356 — — 5.2 7437 5729 — 2320 1.70 0.004 3.21 2.47 NF
C2-M8-V80 1975 178 2 90 4.3 7001 5655 504 2824 0.70 0.004 2.48 2.00 NF
C2-M8-V90 2500 200 2 90 6.4 9778 6557 504 2824 0.89 0.005 3.46 2.32 NF
C2 C2-M8-V100 3086 222 2 90 48.9 11,815 7075 504 2824 1.09 0.041 4.18 2.51 NF
C2-M8-V110 3736 244 2 90 123.8* 12,675 7441 504 2824 1.32 0.103 4.49 2.63 SF
C2-M8-V120 4444 267 2 90 301.7 *
12,828 7618 504 2824 1.57 0.251 4.54 2.70 SF
C4-M8-V90 2500 200 4 90 6.1 9668 6468 1007 3327 0.75 0.005 2.91 1.94 NF
C4-M8-V100 3086 222 4 90 13.7 10,691 7020 1007 3327 0.93 0.011 3.21 2.11 NF
C4 C4-M8-V110 3736 244 4 90 15.1 12,648 7455 1007 3327 1.12 0.013 3.80 2.24 NF
C4-M8-V120 4444 267 4 90 233.2 *
12,892 7704 1007 3327 1.34 0.194 3.87 2.32 SF
C4-M8-V130 5216 289 4 90 336.0 *
13,438 7724 1007 3327 1.57 0.280 4.04 2.32 SF
C8-M8-V100 3086 222 8 90 9 11,569 7138 2015 4335 0.71 0.008 2.67 1.65 NF
C8-M8-V110 3736 244 8 90 11.9 13,040 7716 2015 4335 0.86 0.010 3.01 1.78 NF
C8 C8-M8-V120 4444 267 8 90 18.0 12,864 7862 2015 4335 1.03 0.015 2.97 1.81 NF
C8-M8-V130 5216 289 8 90 50.3 13,895 7843 2015 4335 1.20 0.042 3.21 1.81 NF
C8-M8-V140 6049 311 8 90 333.9 *
15,536 8045 2015 4335 1.40 0.278 3.58 1.86 SF
C2O1-M8-V100 3086 222 2 0/0 146.5* 11,831 7109 0 2320 1.33 0.122 5.10 3.06 SF
C2O2-M8-V100 3086 222 2 0/90 31.9 11,422 7059 252 2572 1.20 0.027 4.44 2.74 NF
C2O
C2O2-M8-V110 3736 244 2 0/90 68.3 12,311 7297 252 2572 1.45 0.057 4.79 2.84 SF
C2O3-M8-V110 3736 244 2 135/45 250.0 *
12,867 7393 356 2676 1.40 0.208 4.81 2.76 SF
G2-M8-V80 1975 178 2 90 4.3 7356 5658 168 2488 0.79 0.004 2.96 2.27 NF
G2 G2-M8-V100 3086 222 2 90 157.7 10,973 6988 168 2488 1.24 0.131 4.41 2.81 SF
G2-M8-V120 4444 267 2 90 450 12,683 7773 168 2488 1.79 0.375 5.10 3.12 SF
G6-M8-V80 1975 178 6 90 4.2 7284 5696 504 2824 0.70 0.004 2.58 2.02 NF
G6 G6-M8-V100 3086 222 6 90 19.1 11,499 7108 504 2824 1.09 0.016 4.07 2.52 NF
G6-M8-V120 4444 267 6 90 209 13,647 7840 504 2824 1.57 0.174 4.83 2.78 SF
Note: Ek is initial vehicle kinetic energy; Mk is initial vehicle momentum; n and β are layer number and arrangement orientation of FRP wraps, respectively; Dm is peak displacement
of impacted column at height of 1.2 m, which would be displacement (*) at time point of 0.3 seconds if impacted column failed severely; Pm is largest peak impact force; Vm is peak
shear force at bottom surface of impacted column; Vf is contribution of static shear strength from FRP; Vn is static shear strength of RC columns with or without FRP retrofit as per the
literature (Priestley et al. 1994, 1996); L is impact height: L = 1200 mm; and D is diameter of circular column (D = 1000 mm). In “Failure mode” column, “NF” represents no shear
failure; and “SF” represents shear failure. 1 kJ = 0.083 ton; 1 kN·s = 0.225 kip·s; 1 GPa = 145 ksi; 1 MPa = 0.145 ksi; 1 mm = 0.039 in; 1 kN = 0.225 kip.
second peak impact force induced by cargo impact as the including layer number, type, and fiber orientation of FRP
first peak induced by engine impact, as shown in Fig. 8(a). wraps, were considered to investigate their effectiveness
Because the focus of this study is to investigate the in reducing damage during vehicular impact. Final failure
effectiveness of FRP strengthening to improve the impact modes of the directly impacted column C1 for all the cases
resistance of RC pier columns under vehicle collisions, the are given in Table 1. Note that column C1 may be considered
variations of engine mass, cargo mass, and cargo stiffness to fail in shear if it suffers visible lateral displacement and
are not considered in the following parametric analyses on clear diagonal shear crack.
FRP-strengthened RC columns. As presented in Fig. 7, the damage evolution of RC pier
Effect of FRP properties—Continuous wrapping using columns in C4-M8-V100 was compared with the un-retro-
unidirectional fiber sheets was adopted for strengthening fitted RC column C0-M8-V100. Before 0.15 seconds, when
the RC column, C1, along its entire height. Three variables, the engine impact was dominant, both cases presented

similar damage states; however, the states started to be quite upper part. In the meanwhile, only minor concrete damage
different after 0.15 seconds. The FRP wrapping with four appeared at the top and bottom of the two adjacent columns
layers could fully sustain the shear damage at the bottom to C2 and C3.
keep the integrity of column C1 and reduce its damage at the The damage states of FRP-wrapped RC pier columns
at 0.3 seconds in typical cases are presented in Fig. 11.
Table 2—FRP parameters Compared with C0-M8-V100, two layers of transverse
Materials Input parameters Parameter values CFRP wraps in C2-M8-V100 can control shear damage at
Single-layer thickness 0.167 mm
the bottom of column C1, but has a fracture of CFRP at the
bottom and little effect on limiting flexural cracks at the top
Mass density 1777 kg/m3
and in the middle. Comparatively, two longitudinal layers
Unidirectional Longitudinal modulus 240 GPa of CFRPs, as in case C2O1-M8-V100, cannot prevent shear
CFRP sheet Longitudinal tensile strength 3494 MPa failure near the bottom but can reduce flexure cracks in
Elongation 1.7%
the middle. For cases where the number of layers of CFRP
wraps was increased beyond two, no evident shear damage
Maximum strain for fiber tension 0.0145
occurred near the bottom when impacted at a velocity of 100
Single-layer thickness 0.167 mm km/h (62.1 mph).
Mass density 1800 kg/m3 When the initial velocity was increased to 110 km/h
Unidirectional
Longitudinal modulus 80 GPa
(68.3 mph), two layers of transverse CFRP wraps could not
GFRP sheet prevent shear failure in the case of C2-M8-V110, but the
Longitudinal tensile strength 1600 MPa
combination of one longitudinal layer and one transverse
Maximum strain for fiber tension 0.02 wrap of CFRP in case C2O2-M8-V110 noticeably reduced
Note: 1 mm = 0.039 in.; 1 kg/m3 = 0.0624 lb/ft3; 1 MPa = 0.145 ksi; and 1 GPa = shear damage at the bottom of column C1. Conversely, case
145 ksi.
Fig. 8—Dynamic time histories of column C1: (a) impact force; and (b) displacement at 1.2 m (47.2 in.) height. (Note: 1 kN =
0.225 kip; and 1 mm = 0.039 in.)
Fig. 9—Concrete damage of unstrengthened RC pier columns at 0.3 seconds: (a) C0-M8-V80; (b) C0-M8-V100; (c) C0-M8-
V120; and (d) C0-M16-V80.

Fig. 10—Relationships between Dm versus Ek, Mk, and Pm. (Note: 1 mm = 0.039 in; 1 kJ = 0.083 ton; 1 kN·s = 0.225 kip·s;
and 1 kN= 0.225 kip.)
Fig. 11—Damage states of directly impacted column C1 at 0.3 seconds in typical cases.
C2O3-M8-V110 with two layers in 135/45 degrees orienta- actual shear strength than two-layer CFRP wraps, despite the
tion showed relatively increased shear damage. Also, adding same confinement stiffness.
two more layers, as in the case of C4-M8-V110, could Figure 12 shows the effects of FRP wrapping on the impact
control shear failure making the response to flexure mode. force, shear force, and moment at the bottom of column C1.
With a further increase in velocity to 120 km/h (74.5 mph), Impact characteristic tendencies and close peak values of
column C1 in case C4-M8-V120 suffered shear failure at impact force, shear force, and moment at the bottom of the
an elevation lower than that in case C2-M8-V110, while RC column during bumper and engine impacts remained
cases C8-M8-V120 wrapped with four more layers of FRP independent, that is, similar to that of the non-retrofitted
seemed to respond in a local or flexure mode. At an impact RC column no matter the number of FRP wraps. However,
velocity of 140 km/h (86.9 mph), the column bottom in case visible differences during cargo impact were noted. The
of C8-M8-V140 appeared to fail in a flexure-shear mode. differences may be ascribed to the enhancement of impact
Under the same impact characteristics, two layers of resistance by the number of FRP wraps, whereby the column
GFRP wraps in case G2-M8-V100 led to more serious shear can completely sustain the former bumper and engine
damage than that in case C2-M8-V100, which is attributed impacts and have residual resistance against cargo impact.
to lower contribution of shear strength and weaker confine- Retrofitting with CFRP or GFRP led to almost the same
ment stiffness from GFRP wraps when using the same dynamic response in regard to impact force, shear force, and
amount as CFRP. Conversely, case G6-M8-V100 presented moment for cases C2- M8-V100 and G2-M8-V100.
slighter local damage and less fractured FRP wraps than case Time histories of displacement at the height of 1.2 m
C2-M8-V100, because six-layer GFRP wraps had stronger (47.2 in.) for column C1 from 0.1 to 0.35 seconds were
chosen and plotted in Fig. 13. With an increase in the

Fig. 12—Effects of FRP parameters on dynamic responses of column C1: (a) impact force; (b) shear force at bottom; and
(c) moment at bottom. (Note: 1 kN= 0.225 kip; and 1 kN·m = 0.738 ft·kip.)
Fig. 13—Effects on displacement at 1.2 m (47.2 in.) height of column C1: (a) velocity, layers, and types of FRP; and (b) velocity
and fiber orientation. (Note: 1 mm = 0.039 in.)
number of FRP layers, the horizontal displacement for the than GFRP wraps when using the same amounts, due to
RC column decreases significantly compared to its non- larger confinement stiffness and shear strength. However, the
retrofitted counterpart under the same loading conditions. A column in case G6-M8-V100 deflected less than that in case
comparison between cases C2-M8-V100 and G2-M8-V100 C2-M8-V100, as listed in Table 1, because the actual shear
indicates that CFRP wraps can reduce displacement better

Fig. 14—Relationships between: (a) Ek versus Dm; (b) Ek versus Pm; and (c) Ek versus Vm. (Note: 1 mm = 0.039 in.; 1 kJ =
0.083 ton; and 1 kN = 0.225 kip.)
Dm grows almost exponentially with an increase in vehic-
ular kinetic energy. A significant reduction in Dm is observed
for FRP-wrapped columns compared to its counterpart. FRP
wraps greatly improve column stiffness, and thus the ability
to resist higher kinetic energy impact and displacement.
An almost linear growth relationship of Pm versus Ek
is shown in Fig. 14 with slight differences between each
series—less than 10%. However, Vm is shown to grow
slowly after kinetic energy reaches a certain value, with little
difference within each series—less than 6%.
It was found that under the same impact conditions, the
amount and type of FRP wraps have little effect on Pm and
Vm. The assumed reason is that during the transient period
when stress is transmitted from the impact zone to the
bottom support, crack and deformation of the column are
not instantly developed, whereby the outside FRP wraps are
Fig. 15—Relationship between Pm and Vm for NF cases. not engaged. However, subsequently, the effect of the FRP
(Note: 1 kN = 0.225 kip.) wraps tends to be more noticeable along with column defor-
strength contributed from six layers of GFRP is stronger mation development. The peaks are occurring at the very
than two-layer CFRP wraps despite the same confinement. beginning time of the impact.
Regarding flexural strengthening, a comparison of case For all cases without shear failure (NF cases), the rela-
C2O1-M8-V100 to C2-M8-V100 indicates that vertical tionship between Pm and Vm is plotted in Fig. 15. Only NF
layers of FRP wraps induce larger displacement than that cases are considered because if the dynamic shear resistance
of transverse layers (refer to Fig. 13(b)). However, in a is exceeded, Vm (the peak force at the base) is limited to the
combination arrangement, as in the case of C2O2-M8-V100, shear capacity. A linear regression curve was used to fit the
where one internal vertical layer and one outside transverse relationship between Pm and Vm. The equation (R2 = 0.983)
layer is used, a decrease in displacement is noted, which is also given in Fig. 15. A good linear relationship between
becomes more visible as the velocity increases to 110 km/h Pm and Vm is observed with almost overlapped values for the
(68.3 mph), indicating better effectiveness in the use of a unstrengthened and strengthened columns. The reason is that
vertical and transverse wrap combination. This is attributed the behavior is almost elastic (prior to the shear damage) and
to the transverse FRP wraps improving the anchorage effect the difference in terms of stiffness between the unstrength-
of vertical wraps to enhance the bending stiffness while ened and strengthened columns is negligible.
confining the horizontal deformation of the column concrete.
Except for case C2O3-M8-V110, the diagonal arrange- Damage assessment of RC pier columns
ment of the FRP wraps showed larger displacement than its To propose a reasonable performance-based design method
counterparts. The larger displacement is attributed to insuffi- of unstrengthened and FRP-wrapped RC pier columns
ciency in improving bending stiffness and confinement. against vehicle collision, the determination of an appropriate
The relationships between Ek versus Dm, Pm, and Vm are damage index is one of the critical objectives. Because there
illustrated in Fig. 14, where Dm is the peak displacement of were no evident flexure-dominant failures for both the no
column C1 at a height of 1.2 m (47.2 in.) or the displacement failure (NF) and shear failure (SF) in these simulation cases,
at 0.3 seconds if the impacted column failed severely; Pm is this study only focuses on the shear failure. Two ratios, Dm/L
the largest peak impact force; and Vm is the peak shear force (where L is the impact height, L = 1200 mm [47.2 in.]) and
at the bottom surface of the impacted column. Ek/(VnD), are introduced to identify the occurrence of shear
failure. Vn is the nominal static shear strength of RC columns

The region for Dm/L between 0.04 and 0.1 is a transition
between no shear failure and shear failure conditions, and
the damage state (that is, NF or SF) of the columns in this
region could not be clearly determined due to the limited
simulations in this study. Accordingly, Dm/L can be under-
stood as a damage index for the identification of the shear
failure, that is, Dm/L < 0.04 for NF and Dm/L > 0.1 for SF.
It is interesting to find that the ratio Ek/(VnD) for the simu-
lated cases has a good correlation with Dm/L, observed from
Fig. 16. For the SF cases (with Dm/L > 0.1), the ratios of
Ek/(VnD) are all larger than 1.2, while they are all less than
1.0 for the NF cases (with Dm/L < 0.04). Hence, it can be
demonstrated that Ek/(VnD) is also suitable to be used as a
damage index to assess the shear failure condition for the
RC columns with or without FRP wraps subjected to vehicle
collision. Furthermore, considering all the parameters Ek, Vn,
Fig. 16—Relationship between Ek/(VnD) and Dm/L. and D could be more easily determined than Dm and L in a
vehicle-column impact scenario, the damage index Ek/(VnD)
with or without FRP wrapping retrofit as per the literature
could be included in the design guide more easily than Dm/L.
(Priestley et al. 1994, 1996). The procedure for the calcula-
It should be noted that the preliminary limits of the damage
tion of Vn is summarized in Appendix B. D is the diameter
indexes (that is, Ek/(VnD) < 1.0 or Dm/L < 0.04 for NF and
of the circular columns (D = 1000 mm [39.4 in.]) in this
Ek/(VnD) > 1.2 or Dm/L > 0.1 for SF) proposed herein are
study and can be the dimension of other shaped columns.
based on a limited number of simulations, and the analysis
The values of Dm/L and Ek/(VnD) for the unstrengthened and
of more cases is fundamental for the improvement of the
FRP-wrapped columns are summarized in Table 1.
prediction performances.
The ratio of Dm/L is quite similar to the interstory drift
angle widely employed in seismic engineering (Kang and
CONCLUSIONS
Wallace 2008). The ratio of Ek/(VnD) can be considered a
In this study, finite element (FE) models of prototype
kind of demand-capacity ratio for the RC columns subjected
reinforced concrete (RC) pier columns collided by trucks
to impact loading. The input energy Ek of the impactor (for
were built and validated by laboratory experimental results.
example, a vehicle) is commonly accepted to reflect the
Retrofitting of the RC column using fiber-reinforced polymer
impact demand; and the nominal static shear strength Vn
(FRP) wrapping was incorporated in the FE model to inves-
represents the capacity of the RC columns that have been
tigate its effectiveness in improving the impact resistance
demonstrated to be prone to shear failure under vehicle
against vehicle collision. The primary variables consid-
impacts. In addition, the diameter D is also included to
ered were the number of FRP wrap layers and the vehic-
construct a dimensionless ratio. Previously, Auyeung et al.
ular velocity. The dynamic responses of the FRP-retrofitted
(2019) proposed a similar damage index of Ek/(φVnD) to
RC pier column were analyzed. The main conclusions are
assess the damage states of the RC pier columns under
as follows:
vehicle collisions, where φVn is the dynamic shear capacity;
1. The vehicle-column collision is characterized by three
and D is the diameter of the RC pier columns. Because the
main sequential impacts: bumper, engine, and cargo impacts.
calculation process for the dynamic shear capacity is more
Shear failure of the RC pier column is initially caused by
complicated, this study just adopts the nominal static shear
engine impact rather than the cargo impact. During bumper
capacity of the RC columns for simplicity.
and engine impacts, RC pier columns retrofitted with different
The relationship curves between Ek/(VnD) versus Dm/L
FRP-wrap layers exhibited little difference in impact force,
are plotted in Fig. 16 for the six-series cases, neglecting
less than 10%, along with shear force and moment near the
C0-M12-V80 and C0-M16-V80 in the C0 series, because
bottom of the column, less than 6%. However, differences
the variation of the gross truck mass is not considered
become rather apparent during cargo impact, especially for
herein. The variation in Ek is due to the variation of the
the cases subjected to higher vehicular velocity.
vehicle velocity.
2. FRP wrapping can greatly improve the capacity to
As shown in Fig. 16, the ratios of Dm/L for the limited
control shear damage within RC columns under vehicle colli-
simulation cases are mostly scattered in the interval less
sion, especially for carbon FRP (CFRP) wrapping, which
than 0.04 or larger than 0.1, and almost none in the region
was relatively better than glass FRP (GFRP). Increasing the
between 0.04 and 0.1. For the cases with Dm/L larger than
FRP layer number can lead to a possible change in the failure
0.1, the impacted columns, whether strengthened with FRP
pattern from shear to flexure-shear mode. The two-layer
or not, exhibit clear diagonal shear crack at the bottom, indi-
arrangement of FRP wraps with one internal layer set in the
cating the reaching of the shear capacity. These cases are
vertical direction and the other external layer set in the trans-
categorized as SF. The columns in the cases having Dm/L
verse direction (0/90 degrees) showed better displacement
less than 0.04 just show minor damage along the column
reduction than that with two layers only set in the transverse
height and no shear failure, thus belonging to the NF case.
direction of the columns. In contrast, FRP wrapping with two

layers set in a 135/45 degrees orientation showed relatively Auyeung, S.; Alipour, A.; and Saini, D., 2019, “Performance-Based
Design of Bridge Piers under Vehicle Collision,” Engineering Structures,
more serious shear damage compared to its counterparts. V. 191, July, pp. 752-765. doi: 10.1016/j.engstruct.2019.03.005
3. The ratios Dm/L or Ek/(VnD) can be used as a damage Buth, C. E.; Brackin, M. S.; Williams, W. F.; and Fry, G. T., 2011, “Colli-
index to evaluate the shear damage state (that is, no shear sion Loads on Bridge Piers: Phase 2. Report of Guidelines for Designing
Bridge Piers and Abutments for Vehicle Collisions,” Report No. FHWA/
failure versus shear failure) of an RC pier column with or TX-11/9-4973-2, Texas Transportation Institute, The Texas A&M Univer-
without FRP strengthening when subjected to vehicle colli- sity System, College Station, TX, 102 pp.
sion. Considering the simplicity to determine its value, the Buth, C. E.; Williams, W. F.; Brackin, M. S.; Lord, D.; Geedipally, S. R.;
and Abu-Odeh, A. Y., 2010, “Analysis of Large Truck Collisions with
damage index Ek/(VnD) could be more easily employed in Bridge Piers: Phase 1. Report of Guidelines for Designing Bridge Piers and
the design guide than Dm/L. Abutments for Vehicle Collisions,” Report No. FHWA/TX-10/9-4973-1,
Further work is required to consider the effects of more Texas Transportation Institute, The Texas A&M University System, College
Station, TX, 186 pp.
parameters, such as vehicle type and mass, and column Chen, L., and Xiao, Y., 2012, “Review of Studies on Vehicle Anti-
properties (for example, diameter and height, steel reinforce- Collision on Bridge Piers,” Journal of Highway and Transportation
ment, axial compression ratio, and so on), to better quantify Research and Development, V. 29, No. 8, pp. 140-148.
Chen, L.; Wu, H.; and Liu, T., 2020, “Shear Performance Evaluation
the contribution of FRP wraps to the impact performances of of Reinforced Concrete Piers Subjected to Vehicle Collision,” Journal
RC pier columns. of Structural Engineering, ASCE, V. 146, No. 4, Apr., p. 04020026. doi:
10.1061/(ASCE)ST.1943-541X.0002571
Chen, L.; Wu, H.; and Liu, T., 2021, “Vehicle Collision with Bridge
AUTHOR BIOS Piers: A State-of-the-Art Review,” Advances in Structural Engineering,
Tao Liu is a Lecturer in the School of Civil Engineering at Hunan Univer-
V. 24, No. 2, Jan., pp. 385-400. doi: 10.1177/1369433220953510
sity of Science and Technology, Xiangtan, Hunan, China. He was a Visiting
Chen, L.; Xiao, Y.; and El-Tawil, S., 2016, “Impact Tests of Model RC
Research Assistant at Seoul National University, Seoul, Korea. He received
Columns by an Equivalent Truck Frame,” Journal of Structural Engi-
his BS, MS, and PhD in civil engineering from Hunan University, Changsha,
neering, ASCE, V. 142, No. 5, May, p. 04016002. doi: 10.1061/(ASCE)
Hunan, China. His research interests include the behavior and retrofit of
ST.1943-541X.0001449
reinforced concrete structures with fiber-reinforced polymer composites
Demartino, C.; Wu, J. G.; and Xiao, Y., 2017, “Response of Shear-
under impact loading.
Deficient Reinforced Circular RC Columns under Lateral Impact Loading,”
International Journal of Impact Engineering, V. 109, Nov., pp. 196-213.
Lin Chen is an Associate Professor in the School of Civil Engineering at
doi: 10.1016/j.ijimpeng.2017.06.011
Hunan University of Science and Technology. He received his BS and PhD
Do, T. V.; Pham, T. M.; and Hao, H., 2018, “Dynamic Responses and
in civil engineering from Hunan University. His research interests include
Failure Modes of Bridge Columns under Vehicle Collision,” Engineering
the dynamic behavior of concrete structures under impact loading and
Structures, V. 156, Feb., pp. 243-259. doi: 10.1016/j.engstruct.2017.11.053
vehicle collision on bridge piers.
El-Tawil, S.; Severino, E.; and Fonseca, P., 2005, “Vehicle Collision with
Bridge Piers,” Journal of Bridge Engineering, ASCE, V. 10, No. 3, May,
Jinjun Xu is an Assistant Professor in the School of Civil Engineering at
pp. 345-353. doi: 10.1061/(ASCE)1084-0702(2005)10:3(345)
Nanjing Tech University, Nanjing, Jiangsu, China. He received his PhD
Fan, W.; Shen, D.; Zhang, Z.; Huang, X.; and Shao, X., 2020, “A Novel
in civil engineering from Guangxi University, Nanning, Guangxi, China.
UHPFRC-Based Protective Structure for Bridge Columns against Vehicle
His research interests include the performance of sustainable construction
Collisions: Experiment, Simulation, and Optimization,” Engineering Struc-
materials and composite structures.
tures, V. 207, Mar., p. 110247. doi: 10.1016/j.engstruct.2020.110247
Fan, W.; Xu, X.; Zhang, Z.; and Shao, X., 2018, “Performance and Sensi-
Cristoforo Demartino is an Assistant Professor at Zhejiang University-
tivity Analysis of UHPFRC-Strengthened Bridge Columns Subjected to
University of Illinois at Urbana-Champaign Institute, Haining, China,
Vehicle Collisions,” Engineering Structures, V. 173, Oct., pp. 251-268. doi:
and was a Postdoctoral Researcher in the School of Civil Engineering at
10.1016/j.engstruct.2018.06.113
Nanjing Tech University. He received his PhD from the University of Naples
Huo, J.; Li, Z.; Zhao, L.; Liu, J.; and Xiao, Y., 2018, “Dynamic Behavior
Federico II, Naples, Italy, and his MS degrees from the Sapienza University
of Carbon Fiber-Reinforced Polymer-Strengthened Reinforced Concrete
of Rome, Rome, Italy, and Mediterranea University of Reggio Calabria,
Beams without Stirrups under Impact Loading,” ACI Structural Journal,
Reggio Calabria, Italy, where he also received his BS. His research interests
V. 115, No. 3, May, pp. 775-787. doi: 10.14359/51701283
include the dynamic behavior of structures under natural and man-made
Ibrahim Ary, M., and Kang, T. H.-K., 2012, “Shear-Strengthening of
loading, such as earthquakes, wind, and impact.
Reinforced & Prestressed Concrete Beams Using FRP: Part I – Review
of Previous Research,” International Journal of Concrete Structures and
Thomas H.-K. Kang, FACI, is a Professor of structural engineering and
Materials, V. 6, No. 1, Mar., pp. 41-47.
Director of the Engineering Innovation Center at Seoul National Univer-
Kang, T. H.-K., and Ibrahim Ary, M., 2012, “Shear-Strengthening of
sity. He is a member of ACI Subcommittee 318-T, Post-Tensioned Concrete;
Reinforced & Prestressed Concrete Beams Using FRP: Part II – Experi-
Joint ACI-PTI Committee 320, Post-Tensioned Structural Concrete Code;
mental Investigation,” International Journal of Concrete Structures and
Joint ACI-ASCE Committees 352, Joints and Connections in Monolithic
Materials, V. 6, No. 1, Mar., pp. 49-57. doi: 10.1007/s40069-012-0005-0
Concrete Structures, and 423, Prestressed Concrete; and Joint ACI-ASME
Kang, T. H.-K., and Wallace, J. W., 2008, “Seismic Performance of Rein-
Committee 359, Concrete Containments for Nuclear Reactors. He received
forced Concrete Slab-Column Connections with Thin Plate Stirrups,” ACI
the ACI Wason Medal for Most Meritorious Paper in 2009. His research
Structural Journal, V. 105, No. 5, Sept.-Oct., pp. 617-625.
interests include the design and behavior of reinforced, prestressed, and
Liu, T., and Xiao, Y., 2017, “Impact Behavior of CFRP-Strip-Wrapped
post-tensioned concrete structures.
RC Beams without Stirrups,” Journal of Composites for Construc-
tion, ASCE, V. 21, No. 5, Oct., p. 04017035. doi: 10.1061/(ASCE)
ACKNOWLEDGMENTS CC.1943-5614.0000815
The research was supported by the National Natural Science Foundation Liu, T.; Kang, T. H.-K.; Nghiem, A.; and Xiao, Y., 2020, “Impact Testing
of China (51808214), the visiting scholarship of the Institute of Construc- of Reinforced Concrete Beams Shear-Strengthened with Fiber-Reinforced
tion and Environmental Engineering at Seoul National University provided Polymer Wraps,” ACI Structural Journal, V. 117, No. 3, May, pp. 297-310.
to the first author, and the Korea Foundation of Nuclear Safety (2003007- Liu, Y.; Dong, A.; Zhao, S.; Zeng, Y.; and Wang, Z., 2021, “The Effect of
0120-CG100). Also, the authors appreciated the support from Hunan Provin- CFRP-Shear Strengthening on Existing Circular RC Columns under Impact
cial Natural Science Foundation of China (2020JJ5177 and 2018JJ3186). Loads,” Construction and Building Materials, V. 302, Oct., p. 124185. doi:
10.1016/j.conbuildmat.2021.124185
LS-DYNA, 2010, “LS-DYNA Keyword User’s Manual Volume I: Version
REFERENCES 971,” Livermore Software Technology Corporation, Livermore, CA.
Al-Bukhaiti, K.; Yanhui, L.; Shichun, Z.; Abas, H.; and Aoran, D., 2021,
Miele, C. R.; Plaxico, C. A.; Kennedy, J. C.; Simunovic, S.; and Zisi, N.,
“Dynamic Equilibrium of CFRP-RC Square Elements under Unequal
2005, “Heavy Vehicle Infrastructure Asset Interaction and Collision,”
Lateral Impact,” Materials (Basel), V. 14, No. 13, p. 3591. doi: 10.3390/
National Transportation Research Center, Knoxville, TN, Sept., 50 pp.
ma14133591
Mohan, P.; Marzougui, D.; and Kan, C. D., 2007, “Validation of a Single
Unit Truck Model for Roadside Hardware Impact,” International Journal of

Vehicle Systems Modelling and Testing, V. 2, No. 1, pp. 1-15. doi: 10.1504/ Sha, Y., and Hao, H., 2015, “Laboratory Tests and Numerical Simula-
IJVSMT.2007.011423 tions of CFRP Strengthened RC Pier Subjected to Barge Impact Load,”
Pham, T. M., and Hao, H., 2016, “Impact Behavior of FRP-Strengthened International Journal of Structural Stability and Dynamics, V. 15, No. 2,
RC Beams without Stirrups,” Journal of Composites for Construction, ASCE, 3 pp. doi: 10.1142/S0219455414500370
V. 20, No. 4, Aug., p. 04016011. doi: 10.1061/(ASCE)CC.1943-5614.0000671 Xu, J.; Demartino, C.; Shan, B.; Heo, Y. A.; and Xiao, Y., 2020, “Experi-
Priestley, M. J. N.; Seible, F.; and Calvi, G. M., 1996. Seismic Design mental Investigation on Performance of Cantilever CFRP-Wrapped Circular
and Retrofit of Bridges, John Wiley & Sons, Inc., Saddle River, NJ, 704 pp. RC Columns under Lateral Low-Velocity Impact,” Composite Structures,
Priestley, M. J. N.; Verma, R.; and Xiao, Y., 1994, “Seismic Shear V. 242, June, doi: 10.1016/j.compstruct.2020.112143
Strength of Reinforced Concrete Columns,” Journal of Structural Engi- Zhou, S. C.; Demartino, C.; Xu, J. J.; and Xiao, Y., 2021, “Effective-
neering, ASCE, V. 120, No. 8, Aug., pp. 2310-2329. doi: 10.1061/ ness of CFRP Seismic-Retrofit of Circular RC Bridge Piers under Vehic-
(ASCE)0733-9445(1994)120:8(2310) ular Lateral Impact Loading,” Engineering Structures, V. 243, Sept. doi:
10.1016/j.engstruct.2021.112602

ACI Faculty Why Join the
Network Faculty Network?
The Faculty Network is a support group for educators
interested in ACI and the concrete industry. Members
receive notifications about classroom resources, fellowships
and scholarships, funding for research, online learning,
and calls for papers and presenters.
Free 1-Year Educator Membership

ACI offers complimentary membership to teaching professionals who have not been an ACI member within
the past 5 years.
Free Desk Copies

Faculty Network members can request a complimentary print or PDF copy of:
• ACI CODE-318 Building Code Requirements for Structural Concrete and Commentary (includes a free subscription to
ACI 318 PLUS)
• ACI CODE-530 Building Code Requirements and Specification for Masonry Structures and Companion Commentaries
• ACI CODE-562 Assessment, Repair, and Rehabilitation of Existing Concrete Structures—Code and Commentary
• MNL-3(20) Guide to the Code for Assessment, Repair, and Rehabilitation of Existing Concrete Structures
• MNL-5(19) Contractor’s Guide to Quality Concrete Construction, 4th Edition
• MNL-17(21) ACI Reinforced Concrete Design Handbook (available digitally with ACI 318 PLUS subscription)
• MNL-66(20) ACI Detailing Manual (available digitally with ACI 318 PLUS subscription)
Networking
ACI hosts a Faculty Network Reception twice a year during the ACI Concrete Conventions, giving an
opportunity to exchange ideas and network.
Faculty Network members receive a complimentary annual subscription that
provides users with convenient digital interactive access to ACI CODE-318-19,
the ACI Detailing Manual, and the numerous design examples in the ACI Reinforced Concrete Design Handbook.
The platform allows professors to create custom user notes that can be distributed to the students to view
alongside the Code. Student members are provided 1-year access to ACI 318 PLUS when they purchase their
printed copy of ACI 318-19 at the student price of $99 (plus shipping).
Professors’
Workshop
Materials | Pavements | Structures Building the Future
Sponsor:
The Professors’ Workshop is designed The Concrete Research Council (CRC)

to provide instructors in civil engineering, seeks concrete research projects that
architecture, architectural engineering, further the knowledge and sustainability
materials science, and construction of concrete materials, construction,
management programs the tools and structures in coordination with
to engage students in the latest ACI Committees. Annual Request for
developments in concrete design, Proposals (RFP) are received annually
construction, and materials. between August 1 and December 1.
ADVANCE YOUR CLASSROOM WITH ACI

JOIN OR RENEW TODAY! concrete.org/educatorsandresearchers
Title No. 119-S38
Effect of Salt-Frost Cycles and Impact Loads on Residual

Axial Capacity of Reinforced Concrete Column
by Guoqing Dong, Jin Wu, and Xing Zhao
This paper presents the results of experimental and analytical and mortar paste, and cumulative cracks appear when the
studies of the performance of reinforced concrete (RC) columns stress exceeded the tensile strength of the materials, further
subjected to the salt-frost cycles (SFC), the impact, and their deteriorating the mechanical properties of concrete.5 Test
combined action under axial compression loading, respectively. results showed that the compressive strength and the elastic
Three groups of RC columns, with each RC column equipped with
modulus of the concrete decreased with the number of freez-
six cubic specimens, were constructed. Rapid SFC tests aiming at
ing-and-thawing cycles.6-8 The axial capacity of the RC
exploring the damage appearance of one group of RC columns and
the relative dynamic modulus of elasticity (RDME), the mass loss columns decreased from 41.56 to 33.75 kN (9.3 to 7.6 kip)
(Δm), as well as the cubic compressive strength loss (Δfcu) of the with the number of cycles increasing from 0 to 300.9 The
specimens after SFC were carried out. The drop-hammer impact existence of cracks was conducive to the penetration of chlo-
loading tests were performed to investigate the damage appearance ride ions into the concrete,10,11 resulting in a great corrosion
and the dynamic responses of the other group of axially loaded RC speed of reinforcement,11,12 further decreasing the ultimate
columns. The failure modes, load-midspan vertical displacement, compressive capacity of RC columns.13,14
and axial capacity of the remaining group of RC columns subjected Over the past decades, extensive research has been devoted
to the combined action of SFC and impact under axial compres- to the behavior of RC columns subjected to the static load
sion loading were also investigated. Based on the analytical study, or the impact load, including load-deflection and cracking
the models to calculate the residual axial capacity of RC columns
response.15 Liu et al.16 reported that the applied axial load
subjected to the SFC, the impact, and their combined action are
could improve the impact resistance of RC columns when
proposed, respectively.
the impact energy was small, whereas the presence of axial
Keywords: failure pattern; impact; reinforced concrete column; residual load accompanied with the high-energy impact could lead to
axial capacity; salt-frost cycles. the complete collapse of column with a low reinforcement
ratio. Cai et al.17 conducted horizontal truck-impact tests on
INTRODUCTION axially loaded RC columns to investigate the effect of slen-
As one of the basic supporting components, reinforced derness ratio, impact weight, and velocity on the dynamic
concrete (RC) columns in the coastal regions are subjected response. Wan et al.18 investigated the damage process of
to corrosion induced by salt solution and freezing-and- RC piers against ship collision with scaled model tests of
thawing cycles, which can lead to a dramatic decrease in ship-pier collisions and finite element simulations.
structural durability and bearing capacity of structures. Addi- The dynamic response of RC columns subjected to impact
tionally, RC columns in bridge structures and wharves can is significant to the safety design of RC columns, and the
be impacted by vessels, ending up with great loss of axial estimation of the residual axial capacity of RC columns after
capacity or overall failure. Thus, it is significant to inves- impact is considerable for the assessment and maintenance
tigate the residual axial capacity of the RC columns after of structures. Some research was conducted for the inves-
salt-frost cycles (SFC) and impact. tigation of the effect of blast on the residual axial capacity
Resistance to freezing-and-thawing cycles is one of the of concrete-filled double-skin tubes,19 RC columns,20-22 and
basic indexes of concrete’s durability. Much research on the concrete-filled fiber-reinforced polymer tubes. However, the
influence of freezing-and-thawing cycles on the properties effect of blast and impact on the residual axial capacity of
of concrete have been conducted. Two main problems— RC columns are not totally identical. Thus, further inves-
namely, surface scaling and internal cracking—were found tigation is necessary for better understanding the effect of
when the concrete was subjected to SFC.1 Liu et al.2 found impact on the residual axial capacity of RC columns.
that surface scaling of the concrete exposed to SFC was much Despite some existing literature on salt-frost resistance of
more severe than that exposed to pure-water freezing-and- concrete and dynamic response of RC columns subjected to
thawing cycles. The worst frost damage in concrete occurred impact, the combined effect of the two factors on the residual
when the concrete underwent freezing-and-thawing cycles axial capacity of RC columns is not clear. In this paper,
in chloride solutions with a concentration of approximately the damage performance and the residual axial capacity of
3% by weight.3 Pigeon et al.1 and Ge et al.4 found that one RC columns subjected to the combined action of SFC and
feasible way to improve concrete’s resistance to freezing-
and-thawing cycles was to ensure adequate entrained air MS No. S-2021-073.R2, doi: 10.14359/51734336, received September 26, 2021, and
voids in hardened concrete. reviewed under Institute publication policies. Copyright © 2022, American Concrete
Volume expansion due to icing of water during the obtained from the copyright proprietors. Pertinent discussion including author’s
cooling period can lead to tensile stress within aggregate is received within four months of the paper’s print publication.

impact are investigated to provide some information for the areas, as well as a theoretical basis for the calculation of the
estimation of residual axial capacity of existing RC columns axial capacity after external impact.
in the cold coastal regions.
EXPERIMENTAL PROGRAM
RESEARCH SIGNIFICANCE Materials and specimen details
This study vividly presents the failure modes and mechan- The cement employed was portland cement Type P.II42.5R
ical behavior of RC columns subjected to the combined and its basic properties tested according to GB 175-200723
action of SFC and impact under axial compression load. The are listed in Table 1. Natural river sand with a nominal size
uniaxial compression test result shows that the axial capacity of 0 to 5 mm (0 to 0.2 in.) was used as fine aggregate (FA).
loss of the RC columns under combined action of SFC and The basic properties of FA tested according to JGJ 52-200624
impact was not equal to the sum of the axial capacity loss are given in Table 2 and its grading distribution is illustrated
induced by SFC and the axial capacity loss induced by in Fig. 1. Natural coarse aggregate (NCA) was made from
impact. In view of this, models for calculating the residual crushing basalt and its basic properties tested according to
axial capacity of the RC columns subjected to the SFC, the JGJ 52-200624 are listed in Table 2. The grading distribu-
impact, and their combined action are proposed, respec- tion of NCA is depicted in Fig. 1. The optimum dosage of
tively. Findings in this paper could provide a basis for future concrete air-entraining agent, which was used to improve
research on the damage mechanism of RC columns in coastal the salt-frost resistance of concrete, was obtained through
tests and was set to 0.01% of the mass of the cement.
Table 1—Basic properties of P.II42.5R portland Mixture proportions referred to the method proposed in JGJ
cement 55-201125 are given in Table 3. The air content of each batch
Item P.II42.5R
of fresh concrete was tested according to GB/T 5008026 and
the average value was 5.5%. HRB400 hot-rolled ribbed bar
Specific surface area, m /kg (ft /lb)
2 2
3720 (18,069)
with a diameter of 10 mm (0.4 in.) was used for longitudinal
Initial setting time, min 198 reinforcement. Tie reinforcement was HPB300 hot-rolled
Final setting time, min 248 plain bar having a diameter of 6 mm (0.2 in.). Standard
Ignition loss, % 2.76%
tension tests were conducted on reinforcements according to
GB/T 228.1,27 and the mechanical properties were presented
Stability Qualified
in Table 4.
3-day bending strength, MPa (ksi) 5.9 (0.9) In total, 15 RC columns of the same size, with a cross
3-day compressive strength, MPa (ksi) 31.5 (4.6) section of 120 x 120 mm (4.7 x 4.7 in.) and a length of
28-day bending strength, MPa (ksi) 8.8 (1.3)
1200 mm (47.3 in.), were fabricated. Figure 2 shows the
schematic reinforcement details and dimensions. These
28-day compressive strength, MPa (ksi) 52.8 (7.7)
columns have identical configuration of reinforcement, thick-
ness of concrete cover of 15 mm (0.6 in.), and design rules
Table 2—Basic properties of FA and NCA confirming to GB 50010.28 The longitudinal reinforcement
Aggregate type FA NCA
Specific density, kg/m (lb/ft )
3 3
2629 (165) 2570 (161.3)
Bulk density, kg/m (lb/ft )
3 3
1666 (104.5) 1330 (83.5)
Crushing value, % — 9.96
Sediment percentage, % 1.7 0.35
Water absorption, % — 0.76
Fineness modulus 2.6 —
Table 3—Mixture proportion

Ingredients Dosage, kg/m3 (lb/ft3)
Water 185 (11.6)
Cement 393.6 (24.7)
NCA 1083.8 (68)
FA 636.2 (39.9)
Air-entraining agent 0.03936 (0.002)
Fig. 1—Grading distribution of NCA and FA.
Table 4—Mechanical properties of reinforcement
Type Code Diameter, mm (in.) Yield strength, MPa (ksi) Ultimate strength, MPa (ksi) Elongation, % Modulus of elasticity, MPa (ksi)
HRB400 C 10 (0.4) 437.65 (63.5) 594.22 (86.2) 23.2 2.0 × 105 (2.9 × 104)
HPB300 B 6 (0.2) 393.53 (57.1) 611.86 (88.7) 25.3 2.0 × 105 (2.9 × 104)

Fig. 2—Reinforcement details and dimensions of RC columns.
Table 5—Grouping of RC columns
Specimens Height of impact
group Name Salt frost cycles, N hammer, mm (in.)
SFC0-IL0 0 0 (0)
SFC25-IL0 25 0 (0)
SFC SFC50-IL0 50 0 (0)
SFC75-IL0 75 0 (0)
Fig. 3—Nomenclature rule of RC columns.
SFC125-IL0 125 0 (0)
SFC0-IL100 0 100 (0.4)
SFC0-IL150 0 150 (5.9)
Impacts
SFC0-IL200 0 200 (7.9)
SFC0-IL250 0 250 (9.9)
SFC0-IL200 0 200 (7.9)
SFC25-IL200 25 200 (7.9)
Combination SFC50-IL200 50 200 (7.9)
of SFC and
impacts SFC75-IL200 75 200 (7.9)
SFC100-IL200 100 200 (7.9)
Fig. 4—Test setup for measurement of RDME.
SFC125-IL200 125 200 (7.9)
consisted of four HRB400 bars, resulting in a longitudinal
reinforcement ratio of 2.18%. The transverse reinforcement Rapid salt-frost cycle test
was provided by HPB300 hot-rolled plain stirrups with Rapid SFC test was carried out according to the method
spacing equal to 60 mm (2.4 in.), resulting in a transverse proposed in GB/T 50082.29 First, the RC columns and cubic
reinforcement ratio of 0.79%. specimens for SFC test were immersed in water for 4 days,
Two parameters, including the number of SFC and the and then were taken out of water and toweled surface-dry.
height of the impact hammer, were considered for the nomen- Subsequently, the initial mass and dynamic elastic modulus
clature of RC columns, as depicted in Fig. 3. The grouping of the columns and the cubic specimens were measured.
of specimens is exemplified in Table 5. Two series—namely, Then, the columns and the cubic specimens were put into
SFCX-IL0 and SFC0-ILY—were designed to investigate the a freezing-and-thawing test tank. Sodium chloride solution
effect of single damage factor—that is, SFC or impact— on with a concentration of 3.5% by weight was poured into the
the residual axial capacity of the RC columns. The series tank until the liquid level of approximately 5 mm (0.2 in.)
of SFCX-IL0 consists of five RC columns with X varying above the top surface of the specimens. Each SFC includes
among 0, 25, 50, 75, and 125. The series of SFC0-ILY two steps (that is, freezing and thawing) with a total dura-
consists of four RC columns with Y varying between 100 and tion of 2 to 4 hours. During each cycle, the highest tempera-
250. Moreover, to investigate the combined effect of SFC ture and lowest temperature were controlled within –18 ±
and impact on the residual axial capacity of the RC columns, 2°C (–0.4 ± 3.6°F) and 5 ± 2°C (41 ± 3.6°F), respectively.
the series SFCX-IL200 consisted of six specimens with X After every 25 SFC, the RC columns and cubic specimens
varying among 0, 25, 50, 75, 100, and 125 were designed. were taken out from tank to have the mass and the relative
Six 150 x 150 x 150 mm (5.9 x 5.9 x 5.9 in.) cubic specimens dynamic modulus of elasticity (RDME) measured.
were cast with each column to test their cubic compressive
strength. All cast specimens were cured for 24 hours at room Relative dynamic modulus of elasticity test
temperature of approximately 20°C (68°F), then demolded. Figure 4 shows the test setup for the measurement of
Subsequently, all specimens were water-cured for 7 days and RDME. A nonmetal ultrasonic detection analyzer was used
then naturally cured. to measure the velocity of ultrasonic wave passing through
the cross section of RC columns.30 The RDME of the RC
columns subjected to various numbers of SFC was deter-
mined as Eq. (1) and (2)

Fig. 5—Schematic setup for exerting axial compression force on RC columns.
hinge. The two ends of each plain bar were threaded to
accommodate nuts. Ensuring that uniform axial force was
exerted on an RC column by fastening nuts, the tensile strain
of the respective plain bar should be consistent. Hence, strain
gauges attached to the midspan of plain bars were connected
to the static strain acquisition instrument and the axial force
was controlled by real-time acquisition of the strain value.
Moreover, each plain bar must be guaranteed to be in the
unyielding state; therefore, the conservative tensile stain
value was set at 100 με.
The setup for the lateral impact test on RC columns shown
in Fig. 6 consists of mainly two parts—that is, a supporting
frame and impact loading frame. The steel supporting frame
consists of two built-up steel short columns, two sets of
cylinder bearing, and eight threaded steel bar with 16 nuts
and four beam channels. Four holes were drilled on the top
Fig. 6—Setup for lateral impact loading test on RC columns. steel plate of each built-up short column and two holes were
drilled on the plate of each beam channel. A load sensor was
1 v 1 2v installed in each cylinder bearing to measure support reac-
Ed V 2 (1)
1 v tion. The impact loading frame shown in Fig. 6 consists of
three parts—that is, slide rail, I-beam, and hammerhead with
a semispherical end. The hammerhead was welded on the
EdN VN2
RDME
= = (2) center of the midspan of the I-beam’s bottom flange plate.
Ed 0 V02 The drop hammer was hung by a computer-controlled hook
where Ed is the dynamic elastic modulus; ρ and v are the and could slide up and down along the vertical slide rail.
density and Poisson’s ratio of concrete, respectively, both of One laser displacement sensor capable of measuring
which are considered as unchanged before and after SFC; V displacement as large as 50 mm (2.0 in.) was arranged at
is the velocity of ultrasonic pulse passing through concrete the bottom of the midspan of the RC columns to measure
(m/s); Ed0 and EdN are the dynamic elastic modulus of the the midspan deflection. One dynamic acquisition instrument
RC columns before SFC and after N SFC (GPa), respec- was used to collect support reaction at two supports and
tively; and V0 and VN are the velocity of ultrasonic pulse midspan deflection. Before the impact test, the drop hammer
passing through the RC columns before SFC and after N was held at the planned heights and the initial data collected
SFC, respectively. by dynamic acquisition instrument were cleared to zero.
Impact energy (Eimp) was used for the measurement of
Impact test the intensity of impact exerted on the RC columns and was
To simulate the axial compression state of RC columns calculated according to Eq. (3)
under impact, the axial compression force was exerted on
each RC column through the specially designed setup shown Eimp = mhgh (3)
in Fig. 5. The setup consists of four 25 mm (1.0 in.) in diam-
eter and 1500 mm (59.1 in.) in-length plain bars with the where mh is the mass of drop hammer in kg; g is acceleration
elastic modulus of 210 GPa (3.0 × 104 ksi), eight nuts, two of gravity; and h is the vertical distance between the bottom
steel plates with four preset holes, and two sets of steel ball of drop hammer and the top of the RC columns—that is,
impact height.

Fig. 7—Strengthening at end of RC columns.
Uniaxial compression test
As shown in Fig. 7, ensuring that the axial force was
uniformly exerted on the RC columns, the end sections of
each specimen were ground flat. Besides, two ends were
strengthened by carbon-fiber sheets in case of premature
failure at the end of the RC columns. The uniaxial compres-
sion tests for RC columns were carried out on the test setup
shown in Fig. 8. One rigid cylindrical support was installed
between the top end of each RC column and load plate to
eliminate end fixity moments. Four displacement sensors
with a 200 mm (7.9 in.) range were used to measure the
midspan longitudinal displacement of the RC columns.
Figure 9 shows the midspan measuring arrangement. S1 Fig. 8—Test setup for measurement of residual axial capacity
and S3 were the surface where the RC column was in direct of RC columns.
contact with drop hammer and the surface where the impact
tension zone was located when RC column was subjected to
impact, respectively. In addition, a load sensor was used to
measure the axial load exerted on the RC columns. The axial
load and midspan longitudinal displacement were simulta-
neously collected by the static acquisition instrument.
EXPERIMENTAL RESULTS AND DISCUSSIONS

Salt-frost damage of RC columns
Appearance of RC columns after SFC—The appearance
of the RC columns after various numbers of SFC is shown in
Fig. 10. The appearance of the RC columns became rougher
with the increase of SFC. It was observed that the appear- Fig. 9—Measuring arrangement of impacted column under
ance of the column after 25 SFC was nearly intact and the axial compression load.
surface was relatively flat; merely the surface cement mortar
interfacial transition zone (ITZ) and mortar under the tensile
dropped off slightly. As the number of SFC increased to
stress caused by freezing expansion of salt solution.31 As the
50, a large area of surface cement mortar exfoliated. After
number of SFC increased, external solution continuously
more than 75 SFC, surface spalling became more severe
infiltrated and microcracks gradually bordered and further
and coarse aggregate was exposed obviously. Moreover,
interconnected, then a large area of mortar began to exfoliate.
different areas in the same surface of the specimen presented
Rust staining due to reinforcement corrosion was observed
various degrees of salt-frost damage.
on the surface of each RC column after SFC. The hardened
After exposure to 25 to 50 SFC, the increasing porosity
cement paste on the surface of concrete could not only dissolve
could provide a large number of channels for the external
chloride, but also acted as a molecular filter in salt solution to
solution to enter into the concrete.30 In addition, accumulated
prevent chloride ions from entering the internal void space.
salt-frost damage in concrete induced destructive porous
Therefore, the water in salt solution penetrated into the void
structure, resulting in the enhancement of capillary absorption,
space, while a large amount of chloride ions accumulated
further increasing the amount of water absorbed by capillary
within a certain range from the surface and penetrated the
action. In the process of SFC, microcracks appeared in the
concrete cover, thus inducing the corrosion of reinforcement.

Fig. 10—Appearance of RC columns after SFC.
m0 mN
m 100% (4)
m0
where m0 and mN are the mass of the specimens before SFC
and after N SFC, respectively
f cu f cu , N
f cu 100% (5)
f cu
where fcu and fcu,N are the cubic compressive strength of the
specimens before SFC and after N SFC, respectively (%).
The RDME, Δm, and Δfcu of the specimens after SFC are
shown in Fig. 11. The RDME of the specimens presented
a downward trend after SFC and was characterized by
two phases. In the first phase—namely the process of SFC
Fig. 11—RDME, Δm, and Δfcu versus number of SFC. increasing from 0 to 100—the RDME of the specimens
decreased by 4.08%, 2.42%, 5.57%, and 4.7%, respectively.
RDME, mass loss, and cubic compressive strength loss— Microcracks occurred and developed within the mortar and
The RDME, mass loss (Δm), and cubic compressive strength interfacial transition zone (ITZ) of concrete due to repeated
loss (Δfcu) of the specimens are calculated according to SFC,31 indicating cumulative salt-frost damage in the spec-
Eq. (2), Eq. (4), and Eq. (5), respectively imens. When SFC increased from 100 to 125, namely the
second phase, the RDME of the specimens decreased by

Fig. 12—Appearance of RC columns after impact.
17.62%, showing a significant decline. Microcracks within extended to the right side. Under the same impact energy, the
the mortar and ITZ bordered continuously and intercon- coarse aggregate on the surface of columns SFC25-IL200,
nected, which led to a decrease in the velocity ultrasonic SFC50-IL200, and SFC75-IL200 has almost no exposure,
wave passing through the specimens. whereas the surface mortar being crushed near the impact
The mass of the specimens subjected to less than 50 SFC point became more severe with the increase of SFC. Within
were observed to increase slightly. The slight increase in 25 to 75 SFC, obvious impact damage could be observed
mass was due to the penetration of sodium chloride solution after the exerted impact load; this fully illustrates that the
into the specimens, the mass of which was larger than the application of SFC has a certain effect on impact damage.
mass loss due to surface spalling. With the number of SFC For columns SFC100-IL200 and SFC125-IL200 severely
increased further, the mass of the specimens declined and damaged by SFC, the mortar near the impact point was
decreased to 3.16% when the number of SFC was 125. The not completely crushed and the impact damage was rela-
reason is that the surface spalling became more severe—the tively reduced compared to columns SFC50-IL200 and
mass of which was larger than the mass gained from the SFC75-IL200. This is because that there was less mortar
penetration of solution. on the surface of the columns due to the severe salt-frost
When the number of SFC increased to 25, the cubic damage. Most of the impact load was borne by the exposed
compressive strength of the specimens increased slightly. coarse aggregate, while the mortar on the surface of the
The mass loss of –0.32% indicates that the penetration of columns only bore a small part.
sodium chloride solution kept the specimens in a saturated Dynamic response—Table 6 lists the midspan deflection
state, which promoted the internal hydration reaction of (δimp), residual deflection (δr), support reaction force (Nr),
concrete and further increased the strength of the specimen. maximum crack width (wm), and others including impact
With an increase in the number of SFC, the cubic compres- height (h), impact energy (Eimp), and cubic compressive
sive strength of the specimens decreased. This is because strength of the specimens after SFC (fcu,N). For the series of
that the massive spalling of the surface concrete decreased SFC0-ILY RC columns, the δimp, δr, Nr, and wm increased
the stress area of the specimens under axial load. In addition, with the impact height, indicating increasing damage due
the development and interconnection of microcracks within to impact. In terms of the series SFCX-IL200 withstanding
mortar and ITZ resulted in a decrease of the internal bond the same impact energy, the value Nr increased continually
performance. with the increasing number of SFC, whereas the value δimp,
δr, and wm, which characterized the impact damage degree,
Impact damage of RC columns experienced two stages of change. In the first stage, namely
Appearance of RC columns after impact—Figure 12 the number of SFC ranged from 0 to 75, these three variables
shows the appearance of the RC columns after impact. The continued to increase. In the second stage, a downward trend
dents and transverse cracks were observed clearly on the top was obviously shown as the number of SFC increased from
surface which drop hammer directly fell to and the bottom 75 to 125.
surface, respectively. The concrete in impact compression Serious permanent damage (that is, plastic deflection and
area presented various degrees of damage. In Fig. 12(a), microscopic cracks) was found in RC columns after impact.
visible oblique cracks caused by heavy hammer impact In this paper, the δr of the RC columns after impact was
appeared on the top surface of column SFC0-IL200 and regarded as the index to characterize impact damage degree,

Table 6—Dynamic response for RC columns
Specimen h, mm (in.) Eimp, kJ fcu,N, MPa (ksi) δimp, mm (in.) δr, mm (in.) Nr, kN (kip) wm, mm (in.)
SFC0-IL100 100 (3.9) 0.1862 25.17 (3.6) 3.803 (0.1) 0.405 (0.02) 69.53 (15.6) 0.34 (0.01)
SFC0-IL150 150 (5.9) 0.2793 26.86 (3.9) 5.314 (3.9) 1.517 (0.06) 73.96 (16.6) 0.50 (0.02)
SFC0-IL200 200 (7.9) 0.3724 32.87 (4.8) 6.219 (0.2) 2.335 (0.09) 77.98 (17.5) 0.70 (0.03)
SFC0-IL250 250 (9.9) 0.4655 30.41 (4.4) 7.461 (0.3) 2.847 (0.1) 105.34 (23.7) 0.85 (0.03)
SFC0-IL200 200 (7.9) 0.3724 28.01 (4.1) 5.670 (0.2) 2.072 (0.08) 72.68 (16.3) 0.70 (0.03)
SFC25-IL200 200 (7.9) 0.3724 26.65 (3.9) 5.713 (0.2) 1.960 (0.08) 87.27 (19.6) 0.92 (0.04)
SFC50-IL200 200 (7.9) 0.3724 24.34 (7.9) 6.330 (0.2) 2.500 (0.1) 94.62 (21.3) 1.04 (0.04)
SFC75-IL200 200 (7.9) 0.3724 19.92 (3.5) 6.534 (0.3) 2.620 (0.1) 110.27 (24.8) 1.08 (0.04)
SFC100-IL200 200 (7.9) 0.3724 19.75 (2.9) 6.321 (0.2) 2.220 (0.09) 112.13 (25.2) 0.96 (0.04)
SFC125-IL200 200 (7.9) 0.3724 20.03 (2.9) 5.880 (0.2) 2.130 (0.08) 112.50 (25.3) 0.88 (0.03)
Behavior of RC columns under axial compression

load
Failure modes of series SFCX-IL200—The failure modes
of the RC columns for axial loading are shown in Fig. 14. It
was observed from the surface S2 of each specimen that the
failure modes of RC columns subjected to less than 100 SFC
are all bending failure under axial compression load, whereas
the crush locations of concrete were different when failure
occurred. The position where the concrete of columns SFC0-
IL200 and SFC50-IL200 was crushed was at the midspan.
With regard to the position of crushed concrete for columns
SFC25-IL200 and SFC75-IL200, it was slightly above the
midspan. In addition to the concrete at the midspan being
crushed, the upper end of SFC100-IL200 was compressed
and accompanied by the appearance of vertical cracks.
Different from the previous five columns, the failure mode
Fig. 13—δr versus Eimp/Pu. of SFC125-IL200 was marked by the concrete at the upper
end being crushed. This clearly indicates that the damage
which could be influenced by a lot of factors, including due to midspan impact was responsible for the failure of
strength of concrete, reinforcement ratio, size of spec- the RC columns subjected to a small number of SFC. Due
imen, and impact energy. When being laterally impacted to the increase of SFC, salt-frost damage accumulated in
at the midspan by drop hammer, the RC columns could be columns and concrete loosened—for example, the concrete
regarded as RC beams and impact damage occurred resulting at the end of column SFC125-IL200 was severely damaged
from the deficiency in flexural load-carrying capacity of the and became the weakest part, which was responsible for its
specimens. One empirical equation (Eq. (6)) for calculating failure.
the δr of RC beams after impact was proposed by Kishi and Load-midspan vertical displacement curve—The midspan
Mikami.32 vertical displacement curves for axial loading are shown
in Fig. 15. At the initial stage of exerting axial load, the
Eimp vertical displacements of each column increased approx-
r 1 (6)
Pu imately linearly. The significant difference was observed
that, with regard to the curves of columns SFC0-IL200
where α1 is the fitting parameter; and Pu is the static flexural
and SFC25-IL200 subjected to a small number of SFC, the
load-carrying capacity of RC columns.
vertical displacement on impact tensile surface S3 increased
The Pu of the RC columns was calculated according to
faster than that on surface S1, S2, and S4, while the area
GB 5001028 and the measured cubic compressive strength of
where the vertical displacement of columns SFC100-IL200
the specimens after 25 to 125 SFC was used for the calcu-
and SFC125-IL200 subjected to a large number of SFC
lation of Pu. The δr in dependence on the Eimp/Pu of test RC
increased greatly was the impact compression surface S1.
columns is shown in Fig. 13. It was observed that the δr
With the increase of axial load, the vertical displacement
of the RC columns tended to increase with the increase of
on surface S1 of each column showed a significant decline,
Eimp/Pu.
and the vertical displacement on surface S3 increased
rapidly. It was because, under the combination of SFC
and impact, tensile surface S3 showed a larger number
of cracks and wider cracks than compression surface S1.

Fig. 14—Failure modes of RC columns.
Under the axial load, the closure degree of cracks on surface on surface S1 decreased, indicating that the concrete in the
S3 was much greater than that on surface S1. When axial midspan was not completely crushed.
load reached approximately 50% of the ultimate load, the
vertical displacement on surface S1, S2, S3, and S4 of each Axial capacity of RC columns after SFC and impact
column showed a similar increasing trend, indicating that The axial capacity of the normal RC columns (Nc)
the concrete in the midspan of the column was in the state of proposed in GB 5001028 was calculated by the following
uniform compression. Eq. (7) and (8)
It can be seen from the curves S1 depicted in Fig. 15(a)
through (c) that the rate of concrete in impact compres- N c f ck A f yk As (7)
sion zone withdrawing from work due to being crushed
was accelerated with the increase of SFC, whereas the
phenomenon of concrete withdrawing from work early no f ck 0.88 c1 c 2 f cu (8)
longer occurred, as illustrated in Fig. 15(d) and (e). Once
the concrete in the midspan was crushed, the compressive where ϕ is the stability coefficient of the RC column, which
stress of the steel bar would increase until it yielded. When is related to the slenderness ratio of the RC column; A is the
the bearing capacity reached the limit, for columns SFC0- concrete area in the section of RC column; As is the longi-
IL200, SFC25-IL200, SFC50-IL200, SFC75-IL200, and tudinal bars’ area in the section of the RC column; fyk is the
SFC100-IL200, the concrete on surface S1 was crushed yield strength of longitudinal bars; fck is the axial compres-
and the concrete on S3 surface cracked due to tension. For sion strength of concrete; αc1 = 0.76 is the ratio of axial
column SFC125-IL200, it can be clearly seen from the compressive strength to cubic compressive strength; and
curve depicted in Fig. 15(f) that the vertical displacement αc2 = 1.0 is the brittleness reduction factor.

Fig. 15—Axial load versus midspan longitudinal displacement.
Table 7—Experimental data and DFsalt of RC columns subjected to SFC
Specimen fck, MPa (ksi) Nu, kN (kip) Nc, kN (kip) Nc', kN (kip) DFsalt, % DFsalt′, % ∆DFsalt, %
SFC0-IL0 27.12 (3.9) 652.00 (146.6) 517.47 (116.3) 652.00 (146.6) 0.00 0.14 0.14
SFC25-IL0 28.00 (4.1) 640.20 (143.9) 529.91 (119.1) 667.69 (150.1) 4.12 3.77 0.35
SFC50-IL0 29.69 (4.3) 628.00 (141.2) 553.76 (124.5) 697.74 (156.9) 10.00 10.83 0.83
SFC75-IL0 30.31 (4.4) 600.00 (134.9) 562.53 (126.5) 708.79 (159.4) 15.35 14.17 1.18
SFC125-IL0 27.89 (4.0) 562.40 (126.4) 528.31 (118.8) 665.67 (149.7) 15.51 16.07 0.56
Note: DFsalt′ is calculated according to Eq. (11); ∆DFsalt = |DFsalt – DF′salt|.
Before SFC, the impact or their combined action, the capacity (517.47 kN [116.3 kip]) was defined as a correc-
axial capacity of the intact RC columns could be obtained tion factor ϕ1 in this paper, which was applied to modify the
by Eq. (7) and (8). In the uniaxial compression test, the calculated axial capacity of RC columns. The corrected axial
measured bearing capacity of the RC columns after SFC, capacity of intact RC columns (Nc') was calculated by the
the impact or their combined action was defined as the following Eq. (9)
residual axial capacity (Nu). Undoubtedly, there is a certain
gap between the calculated axial capacity and the measured N c 1 N c (9)
axial capacity for intact RC columns. For instance, regarding
column SFC0-IL0 without salt-frost and impact damage, where ϕ1 = 1.26 is the correction factor; and Nc' is the correc-
the calculated axial capacity Nc = 517.47 kN (116.3 kip) tion value of the calculated axial capacity.
is smaller than the measured axial capacity, which is equal Three damage factors DFsalt, DFimp, and DFi+s were defined
to the residual axial capacity Nu = 652.00 kN (146.6 kip), to investigate the influence of the SFC, the impact, and their
as shown in Table 7. Therefore, the ratio of the measured combined action on the Nu of the RC columns, respectively,
axial capacity (652.00 kN [146.6 kip]) to the calculated axial which were all calculated according to Eq. (10)

N c N u chloride ions in salt solution into concrete, aggravating the
DF 100% (10) corrosion of steel bars, further decreasing the bond strength
N c
between the concrete and the steel bars. This is the reason for
Table 7 shows the experimental data and the damage factor the loss of axial capacity of the RC columns after exposure
DFsalt of the RC columns. With the increase of SFC, the to SFC.
DFsalt became larger, indicating that the axial capacity of the The ratio of δr to the clear height (H) of the columns,
RC columns suffered a huge loss due to the cumulative salt- proposed by Li et al.,20 was defined as the damage factor
frost damage. Figure 16 depicts the relationship between α. Table 8 lists the test results of the series RC columns
the number of SFC (N) and the damage factor DFsalt. The SFX0-ILY subjected to impact, as well as the results of
correlation coefficients (R2) of the fitting curve is higher another four RC columns investigated by Li et al.20 It can be
than 0.9, demonstrating that the Eq. (11) could accurately observed that both the residual deflection δr and the damage
quantify the axial capacity loss of the RC columns caused factor DFimp increased with the falling height of the hammer.
by SFC. Figure 17 depicts the relationship between the DFimp and
the factor α. The good correlation indicates that Eq. (12)
0.16694 could accurately quantify the axial capacity loss of the RC
DFsalt 0.16832 (11)
N
2.682 columns caused by impact.
1
40.31
r
In the process of SFC, the salt solution penetrated into DFimp 0.72131 1 e 137.673 (12)
H
the mortar layer and ITZ through capillary holes inside the
mortar. When the expansion force generated by the icing of The increase in damage factor α results in an increase in
the salt solution exceeded the tensile strength of concrete, the coefficient DFimp. For the RC columns with small δr, the
a large number of microcracks appeared in the mortar and ultimate failure mode was axial compression failure. For the
ITZ. As the number of SFC increased, the cumulative salt- RC columns with large δr, additional bending moment gener-
frost damage in concrete was intensified. The spalling of the ated near the impact point due to eccentric compression, and
surface mortar as well as the enlargement and interconnec- lateral displacement increased with the axial compression
tion of microcracks are more conducive to the penetration of load. Finally, the concrete in the compression zone was
Fig. 16—DFsalt versus N. Fig. 17—DFimp versus α.

Table 8—Experimental results of RC columns subjected to impact
Specimen Nu, kN (kip) Nc′, kN (kip) δr, mm (in.) H, mm (in.) α, 10–3 DFimp, % DFimp′, % ∆DFimp, %
SFC0-IL100 594.00 (133.5) 617.33 (138.8) 0.405 (0.02) 1200 (47.3) 0.338 3.78 3.28 0.5
SFC0-IL150 573.50 (128.9) 647.38 (145.5) 1.517 (0.06) 1200 (47.3) 1.264 11.41 11.52 0.11
SFC0-IL200 651.00 (146.4) 768.50 (172.8) 2.335 (0.09) 1200 (47.3) 1.727 15.29 15.26 0.03
SFC0-IL250 505.50 (113.6) 710.50 (159.7) 2.847 (0.11) 1200 (47.3) 2.373 28.85 20.1 8.75
S1 982.00 (220.8) 2030.00 (456.4) 30.4 (1.20) 2400 (94.6) 12.667 51.63 59.52 7.89
S2 771.00 (173.3) 2030.00 (456.4) 41.23 (1.62) 2400 (94.6) 17.179 62.02 65.35 3.33
S3 535.00 (120.3) 2030.00 (456.4) 55.03 (2.17) 2400 (94.6) 22.929 73.65 69.06 4.59
S4 526.00 (118.3) 2030.00 (456.4) 61.17 (2.41) 2400 (94.6) 25.488 74.09 69.97 4.12
Note: DFimp′ is calculated according to Eq. (12); ∆DFimp = |DFimp – DFimp′|.

Table 9—Experimental results of RC columns subjected to SFC and impact
Specimen Nu, kN (kip) Nc′, kN (kip) DFi+s, % DFsalt,% DFimp, % ∆DFi+s, %
SFC0-IL200 548.00 (123.2) 681.18 (153.1) 19.55 0 15.26 4.26
SFC25-IL200 551.30 (123.9) 664.80 (149.5) 17.07 3.77 14.53 1.23
SFC50-IL200 505.30 (113.6) 656.20 (147.5) 23.00 10.83 17.99 5.82
SFC75-IL200 458.70 (103.1) 630.11 (141.7) 27.20 14.17 18.73 5.7
SFC100-IL200 439.30 (98.7) 647.34 (145.5) 32.14 15.49 16.22 0.43
SFC125-IL200 466.70 (104.9) 692.46 (155.7) 32.60 16.07 15.64 0.89
Note: ∆DFi+s =|DFi+s – (DFsalt + DFimp)|.
Table 10—Calculation results of series SFCX-IL200 RC columns

Specimen DFi+s, % DFsalt, % DFimp, % a b DFi+s′, % error, %
SFC0-IL200 19.55 0 15.26 0 1.0643 16.24 16.93
SFC25-IL200 17.07 3.77 14.53 0.3529 0.9870 15.67 8.19
SFC50-IL200 23.00 10.83 17.99 0.6417 0.9098 23.32 1.37
SFC75-IL200 27.20 14.17 18.73 0.9104 0.8325 28.49 4.75
SFC100-IL200 32.14 15.49 16.22 1.1668 0.7553 30.32 5.65
SFC125-IL200 32.60 16.07 15.64 1.4144 0.6780 33.33 2.25
crushed, showing typical compression and bending failure.

It can be stated that the δr induced by lateral impact has an
0.02197 N 0.86258 0.16832
obvious effect on the failure modes of the RC columns under 0.16694
DFi s
axial compression load. N
2.682

1 (15)
For each RC column in the series of SFCX-IL200 40.31
subjected to the combined action of SFC and impact, the
DFsalt, DFimp, and DFi+s calculated according to Eq. (10) -0.00309N 1.06426 0.72131 1-e137.673
through (12) are listed in Table 9. It is noted that the DFi+s
DFi s DFi s
is not a simple superposition of DFsalt and DFimp. In view of error 100% (16)
this, Eq. (13) was proposed to further figure out the rule of DFi s
combined effect of the SFC and the impact on the residual As demonstrated in Table 10, a increased, whereas b
axial capacity of RC columns decreased with the number of SFC. It can be stated that the
salt-frost damage in the SFCX-IL200 RC columns is the
DFi s a DFsalt b DFimp (13) main reason for the axial capacity loss. The increasing DFi+s
illustrates that the combined action of the SFC and the impact
where a and b are the salt-frost coupling coefficient and could aggravate the axial capacity loss for the RC columns.
impact coupling coefficient, respectively. The axial capacity loss for the RC columns increased with
From aforementioned Fig. 14, it can be observed that six the number of SFC. It is closely related to three aspects—
RC columns subjected to the various number of SFC and the that is, surface spalling, microcracks, and reinforcement
same impact energy shows different failure modes. There- corrosion. First, severe spalling of surface mortar reduced
fore, there must be a relationship between the coupling coef- the effective cross-sectional area of the columns under axial
ficients and the number of SFC. The relationship between compression load. Second, development and interconnec-
the number of SFC and coupling coefficients was fitted by tion of microcracks decreased the strength of the mortar and
Eq. (14) concrete in the ITZ. Third, reinforcement corrosion due to
the infiltration of chloride ions degraded the bond strength
a 0.02197 N 0.86258 between concrete and steel bars. In addition, the RC columns
(14)
b 0.00309 N 1.06426 with residual deflection induced by impact were in eccentric
compression state when axial compression load applied,
Rearranging Eq. (11) through (14), the combined damage which was responsible for the partial axial capacity loss.
factor DFi+s in dependence on the number of SFC and In the current code GB 50010,28 aforementioned Eq. (7)
damage factor α was expressed by Eq. (15). Hence, the was used to calculate the compressive bearing capacity for
calculated combined damage factor DFi+s′ could be obtained normal section of the reinforced concrete. Based on the test
by substituting the number of SFC and damage factor α into results and analysis shown in this paper, Eq. (17) through
Eq. (15). In addition, the error between DFi+s and DFi+s′ of (19) can be used to calculate the residual axial capacity of
the series of SFCX-IL200 RC columns was calculated by the RC columns subjected to the SFC, the impact and their
Eq. (16), as listed in Table 10. combined action, respectively.

the vertical displacement on impact compression surface

N salt 1 DFsalt f ck A f yk As first shortened and then elongated.
5. The axial capacity loss for the RC columns subjected
to the combined action of SFC and impact was not a simple
(17)
DF 0.16832 0.16694 superposition of the axial capacity loss for the RC column
salt subjected to the SFC and the impact. The formulas for calcu-
N
2.682
1 lating the residual axial capacity of RC columns subjected

40.31 to the SFC, the impact and their combined action are estab-
lished, respectively.
N imp 1 DFimp f ck A f yk As
(18) AUTHOR BIOS
DF 0.72131 1 e 137.673 r
Guoqing Dong is a PhD Student in the Department of Civil and Airport
imp H Engineering at Nanjing University of Aeronautics and Astronautics,
Nanjing, China, where he received his MS. His research interests include
durability of concrete.

Jin Wu is a Professor in Department of Civil and Airport engineering at
N (1 DF )( f A f A ) Nanjing University of Aeronautics and Astronautics. He received his PhD
is is ck yk s from Hohai University, Nanjing, China. His research interests include rein-
forced concrete structures.

Xing Zhao is a Lecturer in the Department of Civil and Airport Engineering

(19) at Nanjing University of Aeronautics and Astronautics. She received her
DFi s 0.02197 N 0.72131 1 e137.673 PhD from Southeast University, Nanjing, China. Her research interests
0.86258
include fiber-reinforced concrete structures.

ACKNOWLEDGMENTS
0.00309 N 1.06426 0.16832 0.16694 The study was carried out with the financial support from National Key
R&D program for 13th Five-year Plan of China (No. 2018YFD1101001).
N
2.682
1
40.31 REFERENCES
1. Pigeon, M.; Marchand, J.; and Pleau, R., “Frost Resistant Concrete,”
CONCLUSIONS Construction and Building Materials, V. 10, No. 5, 1996, pp. 339-348. doi:
Based on the experimental results, the following conclu- 10.1016/0950-0618(95)00067-4
sions can be drawn. 2. Liu, Z. C., and Hansen, W., “Pore Damage in Cementitious Binders
Caused by Deicer Salt Frost Exposure,” Construction and Building Mate-
1. The appearance of the reinforced concrete (RC) rials, V. 98, Nov. 2015, pp. 204-216. doi: 10.1016/j.conbuildmat.2015.06.066
columns suffered increasing degree of salt-frost damage 3. Marchand, J.; Pigeon, M.; Bager, D.; and Talbot, C., “Influence
with the increase of salt-frost cycles (SFC). With the of Chloride Solution Concentration on Deicer Salt Scaling Deteriora-
tion of Concrete,” ACI Materials Journal, V. 96, No. 4, July-Aug. 1999,
increase of SFC, the relative dynamic modulus of elasticity pp. 429-435.
(RDME) of the RC columns decreased at an increasing rate, 4. Ge, Y.; Yang, W. C.; Yuan, J.; Zhang, B. S.; and Xiong, A. L., “Dete-
whereas the mass of the RC columns first increased and then rioration of Concrete Freezing-Thawing in Different Salts Solution,” Key
Engineering Materials, Proceedings, V. 405-406, 2009, pp. 315-321. doi:
continuously decreased. The cubic compressive strength of 10.4028/www.scientific.net/KEM.405-406.315
the specimens first increased and then decreased after SFC. 5. Powers, T. C., “A Working Hypothesis for Further Studies of Frost
2. For the series of SFCX-IL200 RC columns subjected to Resistance of Concrete,” ACI Journal Proceedings, V. 16, No. 4, Apr. 1945,
pp. 245-272.
impact, within 0 to 75 SFC, the surface mortar being crushed 6. Muttaqin, H.; Hidetoshi, O.; Yasuhiko, S.; and Tamon, U., “Stress-
near the impact point became more severe with the increase Strain Model of Concrete Damaged by Freezing and Thawing Cycles,”
of SFC, whereas the impact damage to the RC columns was Journal of Advanced Concrete Technology, V. 2, No. 1, 2004, pp. 89-99.
doi: 10.3151/jact.2.89
relatively reduced after 75 SFC. 7. Shang, H. S., and Song, Y. P., “Experimental Study of Strength and
3. The axial compression failure modes of the RC columns Deformation of Plain Concrete under Biaxial Compression after Freezing
subjected to the combination of SFC and impact were found and Thawing Cycles,” Cement and Concrete Research, V. 36, No. 10, 2006,
pp. 1857-1864. doi: 10.1016/j.cemconres.2006.05.018
varied with the number of SFC. The failure modes of RC 8. Duan, A.; Jin, W.; and Qian, J., “Effect of Freeze–Thaw Cycles on
columns subjected to less than 100 SFC were all bending the Stress–Strain Curves of Unconfined and Confined Concrete,” Mate-
failure under axial compression load, whereas the failure rials and Structures, V. 44, No. 7, 2011, pp. 1309-1324. doi: 10.1617/
s11527-010-9702-9
mode of the RC column subjected to 125 SFC was marked 9. Xu, S.; Li, A.; Ji, Z.; and Wang, Y., “Seismic Performance of Rein-
by the concrete at the upper end being crushed. forced Concrete Columns after Freeze–Thaw Cycles,” Construction
4. The midspan vertical displacement on each surface of and Building Materials, V. 102, Jan. 2016, pp. 861-871. doi: 10.1016/j.
conbuildmat.2015.10.168
the RC columns increased linearly under initial axial loading. 10. Kuosa, H.; Ferreira, R. M.; Holt, E.; Leivo, M.; and Vesikari, E.,
With the increase of axial load, the vertical displacement on “Effect of Coupled Deterioration by Freeze–Thaw, Carbonation and Chlo-
impact compression surface decreased, whereas on impact, rides on Concrete Service Life,” Cement and Concrete Composites, V. 47,
Mar. 2014, pp. 32-40. doi: 10.1016/j.cemconcomp.2013.10.008
tensile surface increased. When axial load reached approx- 11. Zhang, P.; Cong, Y.; Vogel, M.; Liu, Z.; and Harald, S., “Müller,
imately 50% of the ultimate load, the vertical displacement Z. Y.; and Zhao, T. J., “Steel Reinforcement Corrosion in Concrete under
on each surface showed a similar increasing trend. When Combined Actions: The Role of Freeze-Thaw Cycles, Chloride Ingress, and
Surface Impregnation,” Construction and Building Materials, V. 148, Sept.
bearing capacity reached the limit, with the increase of SFC, 2017, pp. 113-121. doi: 10.1016/j.conbuildmat.2017.05.078

12. Shi, X.; Xie, N.; Fortune, K.; and Gong, J., “Durability of Steel Rein- 21. Roller, C.; Mayrhofer, C.; Riedel, W.; and Thoma, K., “Residual
forced Concrete in Chloride Environments: An Overview,” Construction Load Capacity of Exposed and Hardened Concrete Columns under Explo-
and Building Materials, V. 30, May 2012, pp. 125-138. doi: 10.1016/j. sion Loads,” Engineering Structures, V. 5, Oct. 2013, pp. 66-72. doi:
conbuildmat.2011.12.038 10.1016/j.engstruct.2011.12.004
13. Xia, J.; Jin, W. L.; and Li, L. Y., “Performance of Corroded Rein- 22. Echevarria, A.; Zaghi, A. E.; Chiarito, V.; Christenson, R.; and
forced Concrete Columns under the Action of Eccentric Loads,” Journal Woodson, S., “Experimental Comparison of the Performance and Residual
of Materials in Civil Engineering, ASCE, V. 28, No. 1, 2016, p. 04015087. Capacity of CFFT and RC Bridge Columns Subjected to Blasts,” Journal of
doi: 10.1061/(ASCE)MT.1943-5533.0001352 Bridge Engineering, ASCE, V. 21, No. 1, 2016, p. 04015026. doi: 10.1061/
14. Wu, X.; Chen, L.; Li, H.; and Xu, J., “Experimental Study of the (ASCE)BE.1943-5592.0000762
Mechanical Properties of Reinforced Concrete Compression Members 23. GB 175-2007, Common Portland Cement, Standards Press of China,
under the Combined Action of Sustained Load and Corrosion,” Construc- Beijing, China, 2007.
tion and Building Materials, V. 202, Mar. 2019, pp. 11-22. doi: 10.1016/j. 24. JGJ 52, “Standard for Technical Requirements and Test Method of
conbuildmat.2018.12.156 Sand and Crushed Stone (or Gravel) for Ordinary Concrete,” Ministry of
15. Remennikov, A., and Kaewunruen, S., “Impact Resistance of Rein- Construction of the People’s Republic of China, Beijing, China, 2006.
forced Concrete Columns: Experimental Studies and Design Consid- 25. JGJ 55, “Specification for Mix Proportion Design of Ordinary
erations,” Faculty of Engineering-papers., V. 202, No. 5, Jan. 2006, Concrete,” Ministry of Housing and Urban-Rural Construction, Beijing,
pp. 817-824. China, 2011.
16. Liu, B.; Fan, W.; Guo, W.; Chen, B. S.; and Liu, R., “Experimental 26. GB/T 50080, “Standard for Test Method of Performance on Ordinary
Investigation and Improved FE Modeling of Axially-Loaded Circular RC Fresh Concrete,” Standards Press of China, Beijing, China, 2002.
Columns under Lateral Impact Loading,” Engineering Structures, England, 27. GB/T 228.1, “Metallic Materials – Tensile Testing – Part 1: Method
V. 152, Dec. 2017, pp. 619-642. doi: 10.1016/j.engstruct.2017.09.009 of Test at Room Temperature,” Standards Press of China, Beijing, China,
17. Cai, J.; Ye, J. B.; Chen, Q. J.; Liu, X. P.; and Wang, Y. Q., 2010.
“Dynamic Behaviour of Axially-Loaded RC Columns under Horizontal 28. GB 50010, “Code for Design of Concrete Structures,” Standards
Impact Loading,” Engineering Structures, England, V. 168, Aug. 2018, Press of China, Beijing, China, 2010.
pp. 684-697. doi: 10.1016/j.engstruct.2018.04.095 29. GB/T 50082, “Standard for Test Methods Long-Term Performance
18. Wan, Y. L.; Zhu, L.; Fang, H.; Liu, W. Q.; and Mao, Y. F., “Experi- and Durability of Ordinary Concrete,” Standards Press of China, Beijing,
mental Testing and Numerical Simulations of Ship Impact on Axially Loaded China, 2009.
Reinforced Concrete Piers,” International Journal of Impact Engineering, 30. Zhang, P.; Wittmann, F. H.; Vogel, M.; Muller, H. S.; and Zhao, T.
V. 125, Mar. 2019, pp. 246-262. doi: 10.1016/j.ijimpeng.2018.11.016 J., “Influence of Freeze-Thaw Cycles on Capillary Absorption and Chloride
19. Zhang, F. R.; Wu, C. Q.; Li, Z. X.; and Zhao, X. L., “Residual Axial Penetration Into Concrete,” Cement and Concrete Research, V. 100, Oct.
Capacity of CFDST Columns Infilled with UHPFRC after Close-Range 2017, pp. 60-67. doi: 10.1016/j.cemconres.2017.05.018
Blast Loading,” Thin Wall Structures, V. 96, Nov. 2015, pp. 314-327. doi: 31. Hu, J., and Wu, J., “Mechanical Properties and Uni-Axial Compres-
10.1016/j.tws.2015.08.020 sion Stress-Strain Relation of Recycled Coarse Aggregate Concrete
20. Li, B.; Nair, A.; and Kai, Q., “Residual Axial Capacity of Reinforced Subjected to Salt-Frost Cycles,” Construction and Building Materials,
Concrete Columns with Simulated Blast Damage,” Journal of Performance V. 197, Feb. 2019, pp. 652-666. doi: 10.1016/j.conbuildmat.2018.11.213
of Constructed. Facilities, Proceedings, V. 26, No. 3, May-June 2012, 32. Kishi, N., and Mikami, H., “Empirical Formulas for Designing Rein-
pp. 287-299. forced Concrete Beams under Impact Loading,” ACI Structural Journal,
V. 109, No. 4, July-Aug. 2012, pp. 509-519.

Title No. 119-S39
Nonlinear Modeling of Reinforced Concrete Columns

under Cyclic Loading
by Tae-Sung Eom, Seung-Jae Lee, Chul-Goo Kim, and Hong-Gun Park
Nonlinear modeling parameters for reinforced concrete (RC)
columns under cyclic loading were investigated with an emphasis
on failure mode and hysteretic energy dissipation. By reviewing
existing shear and axial strength models, a failure mode-based
modeling method to define the force-deformation relations was
developed. The effective stiffness, yield deformation, and ulti-
mate deformation to define the envelope relation were proposed
as functions of axial compression ratio, shear span ratio, and the
shear strengths of concrete and transverse reinforcement. Further-
more, the energy dissipation ratio (κ), defined as the ratio of the
hysteretic energies dissipated by the actual behavior and idealized
elastic-perfectly plastic behavior, was formulated, and the cyclic
relation including hysteresis loops and unloading/reloading stiff-
ness was defined using κ. For verification, the modeling results of
load-deformation relations were compared with existing test results
of columns under cyclic loading. Based on the investigation results,
recommendations for the practical application of the proposed
method were discussed.
Fig. 1—Failure modes of RC columns according to shear
strength degradation.
Keywords: column; cyclic loading; deformation capacity; hysteresis loop;
nonlinear modeling; reinforced concrete. large enough over the flexural shear demand (that is, Vy/Vn ≤
0.6), flexure failure occurs at relatively large drift ratios due
INTRODUCTION to the crushing and spalling of confined concrete (Lim et al.
Reinforced concrete (RC) columns under inelastic cyclic 2017; Kim et al. 2019a).
loading show various failure modes depending on load 4. Axial load failure after severe shear damage: If columns
conditions (that is, axial and shear loads, shear span ratio, subjected to high compressive load have inadequate trans-
and so on) and reinforcement details (that is, transverse verse reinforcement and they are vulnerable to shear
reinforcement, longitudinal splice, and so on). According damage, an axial load failure at the loss of gravity load-car-
to ASCE/SEI 41 (ASCE 2013, 2017; ASCE standard for rying capacity can occur (Elwood and Moehle 2005).
seismic evaluation and retrofit of existing buildings) four These four column failure modes have been addressed
failure modes can occur in RC columns as follows (refer to in the 2013 and 2017 versions of ASCE/SEI 41. ASCE/
Fig. 1). SEI 41-13 classifies column failure modes into three cate-
1. Brittle shear failure before flexural yielding: Brittle gories, according to the shear strength ratio and transverse
shear failure is expected to occur if the shear capacity (that reinforcement details, and it specifies an equation for post-
is, nominal shear strength Vn) is less than the shear demand Vy yield shear strength developed by Sezen and Moehle (2004)
at the development of the flexural capacity (that is, nominal to check shear failure after flexural yielding. On the other
flexural strength Mn), which is not acceptable for design. hand, ASCE/SEI 41-17 does not address the failure mode
2. Post-yield shear failure: This failure mode occurs classification as the modeling parameters of RC columns
mostly in lightly reinforced columns or columns with inad- are updated to statistical equations based on the column test
equate transverse reinforcement. Although the nominal database by Ghannoum and Matamoros (2014). Instead,
shear strength (Vn) is greater than the shear demand (Vy), ASCE/SEI 41-17 defines the modeling parameters as func-
a shear failure can occur after flexural yielding in columns tions of three governing factors (that is, axial compression
with 0.6 ≤ Vy/Vn ≤ 1.0 as the shear capacity degrades with ratio, transverse reinforcement ratio, and shear demand-to-
increasing inelastic deformation (that is, VnS in Fig. 1), due capacity ratio Vy/Vn) in Tables 10.8 and 10.9. When consid-
to concrete spalling, bar buckling, and other damage (ASCE ering uncertainty in the calculation of modeling parameters,
41-17). To address this failure mode and consequent defor- such statistical approach is reasonable and yields better
mation capacity, various post-yield shear strength degrada- ACI Structural Journal, V. 119, No. 2, March 2022.
tion models have been proposed in existing studies (Priestley MS No. S-2021-098.R1, doi: 10.14359/51734141, received August 2, 2021, and
et al. 1994; Sezen and Moehle 2004; Kim et al. 2019a). Institute. All rights reserved, including the making of copies unless permission is
3. Flexural failure: If columns have sufficient and obtained from the copyright proprietors. Pertinent discussion including author’s
adequate transverse reinforcement and the shear capacity is is received within four months of the paper’s print publication.

prediction results of load-deformation relations. Neverthe- FAILURE MODE AND DEFORMATION CAPACITY
less, a failure mode-based approach could be more attractive Review of existing shear strength degradation
because it can give insight into the behavior of columns. models
Another demanding issue for the nonlinear modeling of The concept of shear strength degradation with ductility
RC columns is hysteretic energy dissipation. The hysteretic in RC columns under cyclic loading (that is, seismic shear
energy dissipation is difficult to address as it is affected by strength VnS; refer to Fig. 1) has been discussed in the liter-
various design variables (that is, load conditions, reinforce- ature (ATC 1981; Priestley et al. 1994; Sezen and Moehle
ment details, deformation histories, and so on). Current 2004; Kim et al. 2019a). Priestley et al. (1994) proposed
guidelines including ASCE/SEI 41 do not specify criteria a post-yield shear strength degradation model for rectan-
or recommendations to quantitatively assess the hysteretic gular and circular columns. In their model, the column
energy dissipation for nonlinear modeling. On the other shear strength Vn consists of three independent components:
hand, in existing studies (Park and Eom 2006; Eom and Park concrete component Vc, axial load component Vp, and truss
2010a, 2013; Eom et al. 2009, 2015), the energy dissipation component Vs (that is, transverse reinforcement component)
of various RC members was estimated, and the results were (that is, Vn = Vc + Vp + Vs). Among the three components,
used to define the hysteresis loop and unloading/reloading only the concrete component Vc (= k√fc'Aw) is defined to
stiffness in cyclic load-deformation relations. Thus, for degrade in a linear fashion with increasing member displace-
better modeling, a parameter that accounts for the hysteretic ment ductility μ: k decreases linearly from 0.29 to 0.1 (3.5 to
energy dissipation needs to be investigated. 1.2 in psi) as μ increases from 2 to 4. Although this approach
In the present study, nonlinear modeling parameters for yields good predictions, distinction between Vc and Vp is not
RC columns under cyclic loading were investigated. Based clear particularly after diagonal shear cracking.
on existing test results, the effective stiffness and yield defor- Sezen and Moehle (2004) proposed a shear strength
mation varying with axial compression ratio and shear span degradation model (that is, VnS = knl[Vc + αcVs]), which was
ratio were proposed. Further, by employing the concepts of adopted in ASCE/SEI 41 (2013, 2017). knl is 1.0 for μ ≤ 2,
post-yield shear strength degradation (Kim et al. 2019a) and 0.7 for μ ≥ 6, and varies linearly for 2 < μ < 6; and αc is 1.0
axial load failure at the loss of gravity load-carrying capacity for s/d ≤ 0.75, 0.0 for s/d ≥ 1.0, and varies linearly for 0.75 ≤
(Elwood and Moehle 2005), deformation parameters to s/d ≤ 1.0. knl and αc are empirical parameters that account for
define the force-deformation relation were proposed as func- degrading shear strength due to inelastic deformation and
tions of displacement ductility and reinforcement details. nonconforming transverse reinforcement spacing, respec-
Finally, to define hysteresis loops and unloading/reloading tively. The concrete component Vc is defined by taking into
stiffness under cyclic loading, the energy dissipation ratio consideration the effects of shear span (= M/V, the ratio of
(κ), defined as the ratio of the hysteretic energies dissipated moment to shear under considered load combination) and
by the actual behavior and idealized elastic-perfectly plastic axial load. Unlike the Priestley model, the degradation factor
behavior, was formulated, and an energy-based model was knl is applied to both Vc and Vs, whereas the lower limit of knl
proposed using κ. The validity of the proposed method is taken as a higher value of 0.7. This model results in satis-
was verified by comparisons between the predicted and factory predictions of column shear strengths, as commented
measured load-deformation relations of existing columns in in ASCE/SEI 41-17. Furthermore, it can capture the deforma-
the literature. tion capacity with reasonable precision, as reported in Kim
et al. (2019a), although the scatter between the predictions
RESEARCH SIGNIFICANCE and tests is significant (that is, the coefficient of variance for
Although nonlinear modeling parameters for RC columns the ratio of measured-to-calculated maximum deformations
in ASCE/SEI 41-17 were updated based on statistics of for 44 columns is 0.56).
test results, some issues still need further improvements Kim et al. (2019a) investigated the deformation capacity
as follows. First, because modeling parameters defining of RC columns under cyclic loading, using the concept of
force-deformation relation are mainly determined from seismic shear strength degradation. For convenience in
empirical equations, it is difficult to identify failure modes. practical application, the column shear strength degrading
Second, the effective stiffness cannot account for the effects depending on μ is defined using code-specified design
of shear span and is not accurate enough to define yield strengths, as follows (refer to Fig. 1)
deformation. Third, there is no parameter to define the energy
dissipation capacity and hysteretic model of columns under VnS = ηVc + Vs (1)
cyclic loading. In the present study, a modeling approach
to define the force-deformation relation of columns based
on failure modes was proposed. The energy dissipation N
ratio was formulated as functions of section geometry and Vc 0.17 1
14 A f c Aw (2)
reinforcement details, and an energy-based model to define g
hysteresis loops including unloading/reloading stiffness
was proposed. d
Vs = f yt Av (3)
s

(5 ) / 3 for seismic hoops transverse reinforcements on the deformation capacity, more
and 0 1 (4) comprehensive terms, adequate reinforcement and inade-
(4 ) / 2 for non-seismic hoops quate reinforcement, are used in Eq. (5).
where N is axial compressive load; fc′ is concrete compres- For practical application, the minimum and maximum
sive strength; Aw is effective concrete area for shear, taken values of [VM – Vs]/Vc in Eq. (5) should be limited as 0 and 1
approximately as 0.8Ag; Ag is gross area of the column for the following reasons.
section; fyt is steel yield stress of transverse reinforcements; 1. For safe design, the nominal shear capacity Vn (= Vc +
Av is cross-sectional area of transverse reinforcements; Vs) should not be less than the shear demand VM at flexural
d is effective depth of the column section; s is vertical yielding. If VM > Vn (that is, [VM – Vs]/Vc > 1.0) and thus
spacing of transverse reinforcements; and μ is member brittle shear failure is expected to occur, nonlinear modeling
displacement ductility. is not needed.
The term seismic hoops indicates the hoops with 2. If VM < Vs and [VM – Vs]/Vc < 0, the column is expected
135-degree end hooks or continuous wound hoops, whereas to show flexural failure as the lower limit of the degraded VnS
non-seismic hoops indicates hoops with 90-degree end is greater than VM (refer to Fig. 1). Based on existing column
hooks or lap-spliced hoops. test results showing flexural failure, the allowable maximum
Unlike the previous shear strength models by Priestley value of μus is taken as 5 for adequate reinforcements and 4
et al. (1994) and Sezen and Moehle (2004) focusing on an for inadequate reinforcements (Kim et al. 2019a).
accurate prediction of degraded column shear strengths, Once μus is calculated, the ultimate drift ratio at the onset
the model in Eq. (1) to (4) was proposed as a simple and of post-yield shear failure or flexural failure, δus, can be
straightforward method focusing on the prediction of defor- determined as follows
mation capacity. Thus, although the accuracy in predicting
degraded shear strengths may not be satisfactory in some δus = μusδy (6)
cases, the applicability and accuracy of predicting deforma-
tion capacity has been improved. The comparison results
between the predictions and test results of 86 column speci- δy = αVM/Ke (7)
mens are available in Kim et al. (2019a).
where δy is yield drift ratio; Ke is effective member stiffness;
Post-yield shear failure mode (δδus) α is ratio of the yield strength to the ultimate strength; and VM
Based on the shear strength degradation model proposed is shear force at the development of the flexural strength Mn.
by Kim et al. (2019a), the inelastic deformation at the onset Equations (6) and (7) show that to predict δus with accu-
of post-yield shear failure, δus, can be defined as follows. racy, the values of Ke and δy need to be accurately estimated
As shown in Fig. 1, the member displacement ductility at first. Thus, in this study, the values of Ke and δy were inves-
failure, μus (≥ 2.0) can be calculated by letting VnS = VM in tigated as follows, based on the results of 40 columns tested
Eq. (1) and substituting 4 for η in Eq. (1). Rewriting these by the authors (refer to Table 1) (Eom et al. 2014; Lim et al.
equations with respect to μ yields the following relations 2017; Kim et al. 2018, 2019a,b, 2020).
VM Vs Effective stiffness and yield deformation

5 3 5 for adequate reinforcement Figure 2 shows a backbone curve (that is, drawn from test
Vc results) and the effective stiffness (Ke,test) defined from the
us
VM Vs backbone curve. The measured Ke,test is taken as the slope
4 2 V 4 for inadequate reinforcement of a secant line connecting the origin and pre-peak point of
c
0.6Vu, where Vu is the maximum load (ASCE/SEI 41-17).
(5) For cantilever column specimens, the effective flexural
rigidity of the section (EIe,test) can be expressed as EIe,test =
where 0 ≤ [VM – Vs]/Vc ≤ 1.0; VM is shear force at the devel- Ke,test ls2/3, where ls = shear span length. Figure 3(a) shows
opment of the nominal flexural strength (Mn); and Vc and Vs the values of EIe,test/EcIg (that is, calculated from the back-
are nominal shear strengths of concrete and transverse rein- bone curves for positive and negative directions) varying
forcements, calculated by Eq. (2) and (3), respectively. with axial load ratio n (= N/[Agfc′]). Ec is the concrete elastic
Note that the classification of reinforcement details in modulus, taken as 4700√fc' (in MPa), and Ig is the moment
Eq. (5) is different from that in Eq. (4). Originally, Kim et of inertia of the gross section. For comparison, the effective
al. (2019a) suggested the classification of seismic hoops flexural rigidity specified in ASCE/SEI 41-17, EIeASCE/EcIg =
and non-seismic hoops to account for the effects of poor [n + 0.2], (0.3 ≤ EIeASCE/EcIg ≤ 0.7), is also plotted in the
lateral confinements on the deformation capacity, based on figure. Overall, the ASCE/SEI 41-17 equation captures the
the reported test results. However, other test results in the increasing trend of EIe,test/EcIg according to n well, although
literature (Lim et al. 2017; Kim et al. 2018, 2019b) have the scatter is significant. Figure 3(b) shows the variation of
showed that the deformation capacity can also be degraded EIe,test/EcIg with shear span ratio ls/h. ls and h are the shear
by inadequate longitudinal reinforcement details, such as lap span (that is, distance between the lateral loading point and
splice within the plastic hinge region and the use of bundled the location of the critical section) and column depth in the
bars. Thus, to consider the effects of both longitudinal and direction of lateral loading, respectively. Except for the

Table 1—Properties of existing column specimens
Specimens fc', MPa fy, MPa ρl, % ρt, % ls/h n, N/[Agfc'] Longitudinal reinforcement Hoop details
CH *
0.81 — 135-degree hook
Eom et al. SRO1* 0.41 — Continuous
34.7 635 2.53 3.00 0.17
(2014) SRO2* 0.41 — Continuous
SR* 0.72 3-bundled Continuous
C1 26.9 578 0.38 0.24 2-bundled 90-degree hook
C2 26.9 578 0.38 0.13 2-bundled Continuous
Lim et al. C3 26.9 578 0.38 0.24 2-bundled Continuous
2.53 3.00
(2017) C4 26.9 578 0.38 0.24 — Continuous
C1R 25.2 588 0.38 0.20 2-bundled 90-degree hook
C3R 25.2 588 0.43 0.20 2-bundled Continuous
SL00S1B 32.0 571 2.53 0.38 3.00 0.17 —
SL30S1B 32.0 571 2.53 0.38 3.00 0.17 Splice (OB)†
SL40S1B 32.0 571 2.53 0.38 3.00 0.17 Splice (OB)†
Kim et al. SL30S2B 25.0 550 2.53 0.38 6.00 0.18 Splice (OB)†
90-degree hook
(2018) SL40S2B 25.0 550 2.53 0.38 6.00 0.18 Splice (OB)†
SL40S2B-1 37.0 521 3.21 0.38 6.00 0.15 Splice (OB)†
SL50S2B 37.0 590 2.53 0.38 6.00 0.15 Splice (OB)†
SL50S2B-1 27.0 588 2.53 0.19 6.00 0.19 Splice (OB)†
SAd2 571 2.53 0.39 3.00 0.17 — 135-degree hook
SBd2 571 2.53 0.39 3.00 0.17 — 90-degree hook
SBd4 571 2.53 0.77 3.00 0.17 — 90-degree hook
SCd2 571 2.53 0.39 3.00 0.17 — Spliced
SDd2 571 2.53 0.39 3.00 0.10 3-bundled 90-degree hook
Kim et al.
RAd2 32.0 566 2.42 0.32 4.80 0.10 — 135-degree hook
(2019a)
RFd2 566 2.42 0.32 4.80 0.17 — 90-degree hook
RFd3 566 2.42 0.48 4.80 0.17 — 90-degree hook
RGd2 566 2.42 0.32 4.80 0.17 — Spliced
RHd2 566 2.42 0.32 4.80 0.17 — Spliced with hooks
RHd3 566 2.42 0.48 4.80 0.17 — Spliced with hooks
SL30S2T 27.0 0.19 Splice (OT)†
SL40S2T 27.0 0.19 Splice (OT)†
Kim et al.
SL50S2T 37.0 559 2.53 0.38 6.00 0.15 Splice (OT)† 90-degree hook
(2019b)
SL40S2S 37.0 0.15 Splice (S)†
SL50S2S 37.0 0.15 Splice (S)†
S-135-C 0.75 — 135-degree hook
S-135-V 0.38 — 135-degree hook
Kim et al. S-90V-C 0.75 — 90-degree hook + V tie
23.0 551 2.29 4.00 0.35
(2020) S-U-V 0.38 — Spliced with hooks
I-135-C 0.56 — 135-degree hook
I-135-V 0.28 — 135-degree hook
*
P-Δ effects were included in test setup.
†
OB is offset-bent bottom bars; OT is offset-bent top bars; S is side-by-side lap-spliced without offset bend.
l
EI e 0.2 n s Ec I g and 0.3Ec I g EI e 0.7 Ec I g
4.5h
(8)
where n and ls/h are the axial compression ratio and shear
span ratio, respectively.
Figure 4(a) shows the variation of EIe,test/EIe,calc with n.
EIe,calc is the effective stiffness value calculated by Eq. (8)
and EIe,test is the measured value obtained from the backbone
curves. For comparison, the values of EIe,test/EIeASCE are also
plotted in Fig. 4(b). When comparing the values of EIe,test/
Fig. 2—Definition of effective stiffness and yield drift ratio. EIe,calc (that is, proposed method) and EIe,test/EIeASCE (that is,
ASCE 41-17), the mean values of the proposed method are
specimens with a high axial compression ratio of n = 0.35,
comparable to those of ASCE 41-17 (that is, 1.01 ~ 1.04),
the values of EIe,test/EcIg increase almost in a linear fashion
whereas the coefficients of variance (= 0.12 and 0.13) of the
with ls/h. Based on the trends in Fig. 3, the effective flexural
proposed method are significantly less than those (= 0.22
rigidity was approximated as follows
and 0.29) of ASCE 41-17.

Fig. 3—Variations of effective stiffness with axial compression ratio and shear span ratio (test results).
Fig. 4—Ratios of measured-to-calculated effective stiffnesses.
Fig. 5—Comparison of measured and calculated yield drift ratios.

Figure 5(a) shows the yield drift ratios δy,test calculated Ke = 3EIe/ls2). In calculations of δy,calc, to exclude uncertainty
from the backbone curves for positive and negative direc- in strength estimates, the measured strengths Vu (= Mu/ls)
tions (refer to Fig. 2). The values of δy,test show a decreasing were used for VM and α was taken as 1.0 (that is, δy,calc = Vu/
trend with increasing n, although the correlation is rather Ke). Overall, the calculated values of δy,calc agree with the
weak. Figure 5(b) shows the variation of the values of δy,test/ measured values of δy,test: the means of δy,test/δy,calc are 0.98
δy,calc with n. The values of δy,calc were calculated by Eq. (7) and 0.99, and the coefficients of variance are 0.14 and 0.15.
using the effective flexural rigidity EIe in Eq. (8) (that is,

Fig. 6—Envelope force-deformation relations and modeling parameters.
This result shows that the proposed effective flexural rigidity relation is idealized with four points A, B, C, and E. Points
EIe in Eq. (8) is simple but satisfactory. B and C represent the yield and ultimate points, respectively,
More elaborate investigations on the effective stiffness of and E represents the axial load failure point where the column
RC columns were reported in Elwood and Eberhard (2009). completely loses its lateral and axial load-carrying capacities.
The effective flexural rigidity for rectangular columns Note that ASCE/SEI 41 defines the generalized force-deforma-
proposed by Elwood and Eberhard is defined as follows tion relation with five points, A to E, as plotted in Fig. 6(b). Thus,
the forces and deformations at the yield, ultimate, and axial load
0.45 2.5n failure points in Fig. 6(a) and (b) use the same symbols for clear
EI e Ec I g and 0.2Ec I g EI e 1.0 Ec I g
1 110 db / h h / ls comparison. For ASCE/SEI 41-17, deformation parameters a
and b to define points C and E are determined from empirical
(9) equations based on statistics, regardless of failure modes. For
the proposed method, on the other hand, deformations δC and
where db/h is the ratio of longitudinal bar diameter-to-column δE at points C and E are calculated according to expected failure
depth, approximately taken as 1/18 for building columns modes determined by design quantities such as axial load (N)
(Elwood and Eberhard 2009). For the same 40 column spec- and flexural and shear capacities (VM, Vc, and Vs).
imens listed in Table 1, the values of EIe by Eq. (9) were In Fig. 6(a), the force and deformation at the yield point
compared to those by Eq. (8): for EIe,test/EIe,calc, Eq. (9) yields B are defined as αVM and δB = αδy, respectively. VM is the
the mean values of 1.07 (positive direction) and 1.06 (negative column shear force corresponding to the flexural strength
direction), and the coefficients of variance of 0.11 (positive Mn, and α can be approximately taken as a value between 0.9
direction) and 0.16 (negative direction). Thus, both Eq. (8) and and 1.0. δy is the yield drift ratio calculated by using Eq. (7)
(9) can be used to define the effective stiffness of columns. and (8), as follows.
Axial load failure mode (δδua) VM M n / ls M n ls

Another column failure mode that needs to be considered y (11)
Ke 3EI e / ls2 3EI e
for nonlinear modeling is the axial load failure of shear-
damaged columns under high compressive load. Elwood and The force and deformation at point C are defined as VM (=
Moehle (2005) proposed a model to predict the deformation Mn/ls) and δC = δus. δus (= μusδy) is the ultimate drift ratio at
capacity of such columns, as follows the onset of post-yield shear failure or flexural failure, calcu-
lated by Eq. (5) and (6). The force and deformation at point
0.48 E are defined as 0 and δE = δua. δua is the lateral drift ratio
ua (10) at which the column loses its axial load-carrying capacity,
4.6 N / Vs
calculated by Eq. (10). Because δus and δua are calculated
where δua is lateral drift ratio at which the columns lose their independently, the calculated value of δus can be less than the
axial load-carrying capacity; N is axial compressive load; calculated value of δua, which is not acceptable for practical
and Vs is shear capacity of transverse reinforcement calcu- application. Thus, δC at point C is limited to be not greater
lated by Eq. (3). than 80% of δua: δC = δus ≤ 0.8δua (refer to Fig. 6(a)).
Equation (10) is based on the shear friction capacity on a Note that a moment hinge model using M-θ relation should
shear failure plane at an inclination angle of 65 degrees relative be used for the nonlinear modeling of columns, instead of
to the longitudinal axis. The axial capacity model in Eq. (10) the V-δ relations in Fig. 6. The method to convert the V-δ
captures the column deformation capacity δu varying with N relation to the M-θ relation is discussed in “Recommenda-
and Vs well (Elwood and Moehle 2005; Kim et al. 2020). tions for practical application.”
MODELING OF ENVELOPE RELATIONS Comparisons between modeling and test results

Modeling parameters for envelope relations Figure 7 compares the proposed force-deformation rela-
Figure 6(a) shows the envelope force-deformation relation tions (that is, V-δ relations) in Fig. 6(a) with the test results
for RC columns proposed in this study. The force-deformation
Fig. 7—Comparison of predicted envelope relations with test results. (Note: 1 k = 4.45 kN.)

in the literature (Eom et al. 2014; Lim et al. 2017; Kim et lap splice lengths and non-seismic hoop details (that is,
al. 2018, 2019a,b, 2020). Table 1 summarizes the properties SL30S1B, SL30S2B, SL30S2T, SCd2, RGd2, and RHd2).
of existing column specimens used for the verification. All
columns are rectangular or square sections, and the test vari- ENERGY DISSIPATION RATIO AND
ables cover the following ranges: 23.0 MPa ≤ fc′ ≤ 37.0 MPa; CYCLIC RELATION
521 MPa ≤ fy ≤ 635 MPa; 0.1 ≤ n(= N/[fc′Ag]) ≤ 0.35; 3 ≤ ls/h Figure 8(a) shows a typical load-deformation relation
≤ 6; seismic hoops (with 135-degree hooks for anchorage, of RC columns under cyclic loading. The characteristics
hooked U-bars, or continuously wound) and non-seismic of the cyclic curve, such as unloading/reloading stiffness,
hoops (that is, lap-spliced hoops); and longitudinal rein- are affected by the energy dissipation ED. ED herein is the
forcement details (that is, bundled and lap-spliced). quantity defined as the loop area of a cyclic curve repeated
For the verification of the proposed model, the forces and at current deformations in the positive and negative direc-
deformations at points B (α = 0.95), C, and E were defined tions. According to Park and Eom (2006) and Eom and Park
using the calculated strengths VM (= Mn/ls), Mn, Vc, and Vs (2010a), the quantity of ED of RC columns is affected by
based on measured material strengths. When calculating reinforcement details (that is, quantity and arrangement),
the values of μus (≥ 2.0) using Eq. (5), columns with one whereas not significantly affected by axial compressive load.
or more of the following reinforcement details were clas- Thus, as the axial compression load increases, ED does not
sified as “inadequate reinforcement” (refer to Table 1): 1) change much.
non-seismic hoops with 90-degree hook, and lap-spliced Figure 8(b) shows the energy-based cyclic model for RC
hoops (except for “spliced with hooks”); 2) lap splice of members proposed by Eom and Park (2010a). The ener-
longitudinal reinforcement in the plastic hinge region; and gy-based cyclic model defines the hysteresis loop as a paral-
3) bundled bars of two or more. All columns were tested lelogram in which the area is the same as the energy dissipa-
using loading setups where the second-order effect due to tion ED. The unloading stiffness is taken as the same value as
axial load was negligible, except for CH, SRO1, SRO2, and the effective stiffness Ke, whereas the reloading stiffness is
SR. For these four specimens, the second-order effect in the automatically determined by setting the loop area to be the
measured load-deformation relations (that is, test results) same as ED. Thus, to model the cyclic relation of columns
was removed by subtracting Nδ from measured lateral loads with accuracy, first, the energy dissipation ED varying with
(P), where P is measured lateral load, N is axial compressive design variables needs to be accurately estimated. In this
load, and δ is lateral drift ratio (that is, [P – Nδ]) (Kim et study, the energy dissipation ratio κ is used as a modeling
al. 2019a). parameter to define the hysteretic energy dissipation ED and
Figure 7 compares the envelope force-deformation rela- cyclic relations, as follows.
tions predicted by the proposed method (circles and thick
solid lines), ASCE/SEI 41-13 (triangles and thin solid Energy dissipation ratio
lines), and ASCE/SEI 41-17 (squares and dashed lines). The The ideal energy dissipation ratio is defined as κo = ED/
thick dashed lines indicate the seismic shear strengths VnS Eep, where Eep is the energy dissipation by the idealized elas-
degrading with member deformation ductility, calculated tic-perfectly plastic behavior (refer to Fig. 8(b))
by Eq. (1) to (4). Overall, the proposed method and ASCE/
SEI 41-17 capture the envelopes of cyclic responses (that is, ED ED
o (12)
backbone curves) well. Interestingly, the proposed method Eep M P M N P N 2 y
and ASCE/SEI 41-17 yield similar values of δC and δE (that
is, drift ratios at the ultimate and axial load failure points, where MP and MN are column moments at θP and θN, respec-
respectively) although their approaches to define δC and δE tively; θP and θN (≥ 0) are cyclic maximum deformations
are different. ASCE/SEI 41-13 shows good or conservative in the positive and negative directions, respectively; θy is
agreements with the test results, except for some specimens yield rotation, taken as 2.12εyllp/h for rectangular columns
with inadequate reinforcement details such as insufficient and 2.35εyllp/D for circular columns (Priestley 2000); εyl
is yield strain of longitudinal reinforcements; h and D are
Fig. 8—Energy-based hysteresis model and energy dissipation ratio κ.

Fig. 9—Parameters to define energy dissipation ratio κ.
rectangular and circular column depths in the loading direc- longitudinal bars, taken as ρl∙πD2/4, where ρl is longitudinal
tion, respectively; and lp is plastic hinge length, taken as 0.5h reinforcement ratio and Dl is diameter of the circumference
for rectangular columns and 0.5D for circular columns (Eom of longitudinal bars.
and Park 2010a). Substituting ED in Eq. (13a) and (13b) into Eq. (12) yields
Although modeling parameters for circular columns are κo as follows (RB = 0.75 and θy = 2.12εyllp/h for rectangular
not within the scope of this study, this section includes columns and 2.35εyllp/D for circular columns)
the energy dissipation ratio of both rectangular and
circular columns. ED 2 RB f yl Asl hl P N 4 yl l p / hl
o
The energy dissipation ED in Eq. (12) can be calculated as
Eep M P M N P N 2 y
follows. The energy dissipation of RC members is mainly (14a)
contributed by longitudinal reinforcing steel bars (ductile 3 f yl Asl hl
material), whereas the contribution of concrete (brittle mate- for rectangular columns
2 MP M N
rial) is almost negligible. Furthermore, given that plastic
deformations occur mostly in the plastic hinge region (= ED 4 RB f yl Ast Dl / 2 P N 2 yl l p / Dl
lp), ED can be calculated on the basis of deformations in o
Eep MP M N P N 2 y
the plastic hinge region (refer to Fig. 8 and Fig. 9, Eom and
Park 2010a) 3 f yl Ast Ds
for circular columns
2 M P M N
lp (14b)
ED 2 RB f yl Asl hl P N 4 yl
hl
for rectangular columns (13a) where fyl, Asl, hl, Ast, and Dl are the quantities determined
from material and geometric properties (refer to Fig. 9(a)).
The column moments in the positive and negative directions,
MP and MN, respectively, are calculated based on P-M inter-
D 2 Dl
ED 4 RB f yl l P N yl l p action relations. The values of MP and MN can be different as
4 2 column axial load varies with the direction of loading.

D l p The ideal energy dissipation ratio κo in Eq. (14) is appli-
4 RB f yl Ast l P N 2 yl cable to slender columns where shear deformation is negli-
2 Dl
gible. In general, when the shear span ratio ls/h or ls/D is
for circular columns (13b) less than 3.0, the energy dissipation can decrease due to
shear deformation (Eom and Park 2013). Further, inade-
where RB is reduction factor that accounts for the Baus- quate details of longitudinal and transverse reinforcements,
chinger effects, taken as 0.75; and fyl and εyl are yield stress such as bundled bars, insufficient development length,
and strain of longitudinal bars. For rectangular columns, Asl non-seismic hoops, and lap splice in the plastic hinge also
is sum of longitudinal bar areas at one end and hl is distance decrease energy dissipation by increasing bond slip (Kim et
between the centroids of Asl at both ends. If the arrangement al. 2018, 2019b). However, it is difficult to quantitatively
of longitudinal bars is unsymmetrical, only the smaller Asl address the effects of shear span ratio and inadequate details.
experiences full plastic strains during cyclic loading; thus, Thus, in the present study, the empirical modification factors
the smaller one is taken as Asl (refer to Asl = Asl2 in Fig. 9(a)). λs and λr were introduced to address the degraded energy
When longitudinal bars are distributed across the column dissipation
depth, the contribution of intermediate bars in the web to
ED is limited; thus, Asl should be determined by summing 3 f yl Asl hl
areas of the longitudinal bars in the outmost layer (refer to s r 0.15 for rectangular columns
2 MP M N
Fig. 9(a)) For circular columns, Ast is total sum of areas of
(15a)

3 f yl Ast Ds For rectangular columns in Table 1, the values of κo
s r 0.15 for circular columns and κ calculated by Eq. (14a) and (15a), respectively, are
2 M P M N
presented in Table 2. The ideal energy dissipation ratios κo
(15b) are relatively large (0.371 ≤ κo ≤ 0.650). On the other hand,
the actual energy dissipation ratios (κ) multiplied by the
where, λs (≤ 1.0) and λr (≤ 1.0) are the reduction factors that modification factors λs and λr are significantly reduced to
account for the effects of shear span ratio and inadequate 0.15 ≤ κ ≤ 0.465. This indicates that “inadequate details”
details (refer to Fig. 9(b)), respectively. Based on the existing caused deficient energy dissipation. For example, for CH,
test results, the lower limit of κ for severely pinched cyclic SRO1, SRO2, S-135-C, S-135-V, S-90V-C, and S-U-V with
curves was proposed as 0.15 in Eq. (15a) and (15b) (Eom “adequate” details (that is, λr = 1.0), the measured cyclic
et al. 2015). Note that κ is different from the ideal energy curves show relatively good energy dissipation (refer to
dissipation κo in that the reduction in energy dissipation due Fig. 7); thus, the calculated values of κ are relatively large
to pinching and bond slip is accounted for by λs and λr. (0.371 ≤ κ ≤ 0.465). On the other hand, in other specimens
The factor λs that accounts for a reduction in energy dissi- with “inadequate” details, the measured cyclic curves show
pation due to shear deformation can be calculated according limited or poor energy dissipation; thus, in such specimens
to the procedure developed by Eom and Park (2010b, 2013). in Fig. 7, the values of κ are reduced to the minimum value
In the present study, for practical application, an approxima- of 0.15, due to the modification factors λr (= 0.13 ~ 0.89).
tion of λs is proposed as follows.
Modeling of cyclic relations
ls l Figure 10(a) shows the modeling method for cantilever
s or s 1.0 (16) columns. A cantilever column is idealized with the elastic
3h 3D
column element with a moment hinge at the bottom. For the
The factor λr for reinforcement details is related to bond elastic column element that accounts for elastic deforma-
slip. It is difficult to accurately assess or quantify λr as a tion, the effective flexural rigidity EIe defined by Eq. (8) is
function of design variables. In this study, λr was empirically assigned along with the elastic shear rigidity GAe (≈ 0.4EcAg,
defined by design variables related to bond slip, as follows ASCE/SEI 41-17). Thus, the moment hinge element depicts
only plastic rotation. Figure 10(b) shows the load-deforma-
r ld lb ls ts ta 1.0 (17) tion relation for the moment hinge element, implemented by
CSI Perform 3D (2018). The envelope relation in Fig. 6(a)
proposed in this study can be modeled with the YULRX
ld , pro ls , pro relation in CSI Perform 3D (refer to Fig. 10(b)): points B,
ld or 1.0 (18a) C, and E in Fig. 6(a) are mapped to points Y, L, and R in
ld , req ls , req
Fig. 10(b), respectively (MY = 0.95ML [α = 0.95], ML= VMls,
and MR = 0; θY = 0, θL = δC – δB, and θR = δE – δB). The cyclic
1
lb 1.0 (18b) relation is plotted as a solid parallelogram and the loop area
nbb representing the energy dissipation per load cycle (or ED) is
determined by the energy dissipation ratio κ calculated by
1.0 without lap splice Eq. (15a) and (15b).
ls (18c) The proposed energy dissipation ratio κ and cyclic rela-
0.5 with lap splice in plastic hing
ge
tion were applied to existing column specimens in the liter-
ature (Eom et al. 2014; Kim et al. 2019b, 2020; Xiao and
sSMF Martirossyan 1998; Pujol 2002; Muguruma et al. 1989;
ts 1.0 (18d)
s Ng et al. 1978; Kunnath et al. 1997). Figure 11 compares
the load-deformation relations predicted by the proposed
1.0 for seismic hoops method with the test results. The predicted envelope and
ta (18e) cyclic relations were obtained by CSI Perform 3D, based
0.5 for non-seismic hoops on calculated strengths and modeling parameters (that is,
where ld,pro and ld,req are required and provided development effective stiffness EIe, deformations δus and δua, and energy
lengths of longitudinal reinforcements, respectively; ls,pro dissipation capacity κ). The modeling results of three rect-
and ls,req are required and provided splice lengths of longitu- angular column specimens with ls/h ≥ 3.0 are plotted in
dinal reinforcements, respectively; nbb is number of longitu- Fig. 11(a); four rectangular column specimens with ls/h <
dinal bundled bars in a bundle; sSMF is maximum transverse 3.0 in Fig. 11(b); and three circular column specimens (ls/h ≥
reinforcement spacing required for special moment frame 3.0) in Fig. 11(c). For all column specimens, the predicted
columns (that is, ACI 318-19 2019, Section 18.7.5.3); s is envelope and cyclic responses agree with the test results.
provided transverse reinforcement spacing; and “seismic Particularly, the hysteresis loops defined by the energy dissi-
hoops” indicate the hoops with anchorage details that are pation ratios κ (= 0.15 ~ 0.596) capture the overall cyclic
adequate to secure good confinement without loosening, responses with reasonable precision.
such as 135-degree-hooked hoops, continuously wound
hoops, and spliced hoops with hooks.

Table 2—Energy dissipation ratios of existing test specimens
Modification factors*
Specimens Mn, kN-m fyl, MPa Asl, mm 2
hl, mm Ideal κo αld αlb αls αts αta λr Actual κ
CH 800 2027 375 0.453 1.00 1.00 0.453
Eom et al. SRO1 800 2027 385 0.465 1.00 1.00 0.465
635 1.00 1.00 1.00 1.00
(2014) SRO2 800 2027 382 0.461 1.00 1.00 0.461
SR 891 3040 368 0.598 0.33 0.33 0.199
C1† 390 578 2027 251 0.566 0.88† 0.50 0.61 0.50 0.13 0.150
C2† 361 578 2027 251 0.611 0.88† 0.50 0.61 1.00 0.27 0.163
Lim et al. C3† 390 578 2027 251 0.566 0.88† 0.50 0.61 1.00 0.27 0.151
1.00
(2017) C4† 372 578 1520 269 0.477 0.88† 1.00 0.61 1.00 0.53 0.255
C1R 346 588 2027 251 0.650 1.00 0.50 0.61 0.50 0.15 0.150
C3R 385 588 2027 258 0.598 1.00 0.50 1.00 1.00 0.50 0.299
SL00S1B 374 571 1520 219 0.382 1.00 1.00 0.61 0.30 0.150
SL30S1B 340 571 1520 219 0.420 0.63 0.50 0.61 0.10 0.150
SL40S1B 340 571 1520 219 0.420 0.83 0.50 0.61 0.13 0.150
Kim et al. SL30S2B 294 550 1520 219 0.468 0.57 0.50 0.61 0.09 0.150
1.00 0.50
(2018) SL40S2B 294 550 1520 219 0.468 0.76 0.50 0.61 0.12 0.150
SL40S2B-1 367 521 1520 219 0.402 1.00 0.50 0.61 0.15 0.150
SL50S2B 393 590 1927 209 0.400 1.00 0.50 0.61 0.15 0.150
SL50S2B-1 312 588 1520 219 0.471 0.93 0.50 0.30 0.07 0.150
SAd2 374 571 1520 269 0.468 1.00 0.61 1.00 0.61 0.284
SBd2 374 571 1520 269 0.468 1.00 0.61 0.50 0.30 0.150
SBd4 374 571 1520 269 0.468 1.00 1.00 0.50 0.50 0.234
SCd2 374 571 1520 269 0.468 1.00 0.61 0.50 0.30 0.150
SDd2 386 571 2027 269 0.605 0.50 0.61 0.50 0.15 0.150
Kim et al.
RAd2 239 566 1936 149 0.512 1.00 1.00 1.00 0.60 1.00 0.60 0.305
(2019a)
RFd2 262 566 1936 149 0.467 1.00 0.60 0.50 0.30 0.150
RFd3 262 566 1936 149 0.467 1.00 0.89 0.50 0.45 0.209
RGd2 262 566 1936 149 0.467 1.00 0.60 0.50 0.30 0.150
RHd2 262 566 1936 149 0.467 1.00 0.60 1.00 0.60 0.278
RHd3 262 566 1936 149 0.467 1.00 0.89 1.00 0.89 0.417
SL30S2T 358 0.505 0.56 0.09 0.150
SL40S2T 358 0.505 0.74 0.12 0.150
Kim et al.
SL50S2T 398 559 1520 269 0.455 1.00 1.00 0.50 0.61 0.50 0.15 0.150
(2019b)
SL40S2S 398 0.455 0.87 0.13 0.150
SL50S2S 398 0.455 1.00 0.15 0.150
S-135-C 578 0.371 1.00 1.00 0.371
S-135-V 578 0.371 1.00 1.00 0.371
Kim et al. S-90V-C‡ 578 0.371 1.00 1.00 1.00 0.371
551 1548 322 1.00 1.00 1.00
(2020) S-U-V 571 0.375 1.00 ‡
1.00 0.375
I-135-C 535 0.401 0.75 0.75 0.300
I-135-V 535 0.401 0.75 0.75 0.300
*
For all specimens, ls/h ≥ 3.0 and thus λs = 1.0.
†
For C1, C2, C3, and C4, αta = 0.88 since anchorage length at base is 88% of required development length.
‡
For S-90V-C, αta is taken as 1.0 as V-ties prevent the 90-degree-hooked hoops from loosening.
Fig. 10—Moment hinge model for cantilever columns.

Fig. 11—Comparison of predicted cyclic relations with test results.
relation (that is, M-θp relation without elastic deformation)
are calculated as follows (refer to Fig. 6(a)).
a C B (19)
b E B (20)
A flowchart to define the envelope force-deformation rela-

tion according to the proposed method is shown in Fig. 13.
The energy dissipation capacity κ for cyclic modeling is
defined by Eq. (15). As shown in Fig. 10(b), the proposed
modeling method for moment hinge elements can be imple-
mented by CSI Perform 3D.
The modeling parameters proposed in this study can be
applied to rectangular or square columns with the following
Fig. 12—Modeling method for columns in buildings.
ranges of test variables: 23.0 MPa ≤ fc′ ≤ 37.0 MPa; 521 MPa
RECOMMENDATIONS FOR PRACTICAL ≤ fy ≤ 635 MPa; 0.1 ≤ n(= N/[fc′Ag]) ≤ 0.35; and 3 ≤ ls/h ≤ 6.
APPLICATION If the proposed method is applied to columns with test vari-
Figure 12 shows the modeling method for columns in build- ables beyond the ones verified in this study, verification or
ings. Columns are idealized with an elastic column element calibration with test results may be required.
with moment hinge elements at both ends. For the elastic
column element, the effective flexural rigidity EIe defined by CONCLUSIONS
Eq. (8) is used. In the calculation of EIe, the column shear In this study, nonlinear modeling parameters for reinforced
span is approximately taken as half of the net story height concrete (RC) columns under cyclic loading were investi-
(that is, ls = 0.5hn). For the moment hinge elements, the gated. The envelope and cyclic force-deformation relations
modeling parameters a and b to define the force-deformation of columns were developed considering failure modes and

Seung-Jae Lee is a PhD Candidate at Dankook University, where he
received his BS and MS in architectural engineering. His research interests
include seismic evaluation and design of reinforced concrete and masonry
structures.
Chul-Goo Kim is an Assistant Professor at Ewha Womans University,

Seoul, South Korea. He received his BS, MS, and PhD in architectural
engineering from Seoul National University. His research interests include
shear behavior and earthquake design of reinforced concrete structures.
Hong-Gun Park, FACI, is a Professor in the Department of Architecture

& Architectural Engineering at Seoul National University. He received his
BS and MS in architectural engineering from Seoul National University and
his PhD in civil engineering from the University of Texas at Austin, Austin,
TX. His research interests include numerical analysis and seismic design of
reinforced concrete and composite structures.
ACKNOWLEDGMENTS
This work was supported by the National Research Foundation of Korea
(NRF-2018R1A6A1A07025819) and the Ministry of Land, Infrastructure,
and Transport of Korea (20CTAP-C157615-01). The authors would like to
thank I. Kim and S. Cho for their comments and advice on the proposed
models during the development of the KCI nonlinear modeling guideline
for performance-based seismic design of RC building structures.
REFERENCES
ACI Committee 318, 2019, “Building Code Requirements for Struc-
Fig. 13—Flowchart to define envelope force-deformation tural Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
Concrete Institute, Farmington Hills, MI, 624 pp.
relation. ASCE/SEI 41-13, 2013, “Seismic Evaluation and Retrofit of Existing
Buildings,” American Society of Civil Engineers, ASCE, Reston, VA,
hysteretic energy dissipation varying with design variables. 416 pp.
The characteristics of the proposed method are summarized ASCE/SEI 41-17, 2017, “Seismic Evaluation and Retrofit of Existing
as follows. Buildings,” American Society of Civil Engineers, ASCE, Reston, VA,
416 pp.
1. The effective stiffness of columns is affected by axial ATC-6, 1981, “Seismic Design Guidelines for Highway Bridges,” ATC,
load ratio (n) and shear span ratio (ls/h). The flexural rigidity Berkley, CA.
EIe, defined as a function of n and ls/h, showed good agree- Computers and Structures, 2018, “Perform 3D, Nonlinear Analysis and
Performance Assessment for 3D Structures, User Guide (Version 7),” CSI,
ments with existing test results. Further the yield deforma- Berkeley, CA.
tions calculated using EIe also showed good agreements with Elwood, K. J., and Eberhard, M. O., 2009, “Effective Stiffness of
existing test results. Reinforced Concrete Columns,” ACI Structural Journal, V. 106, No. 4,
July-Aug., pp. 476-484.
2. The deformation parameters of the envelope relation Elwood, K. J., and Moehle, J. P., 2005, “Axial Capacity Model for
were defined as the deformations at the development of Shear-Damaged Columns,” ACI Structural Journal, V. 102, No. 4,
post-yield shear failure and axial load failure, respectively. July-Aug., pp. 578-587.
Eom, T.-S.; Hwang, H.-J.; and Park, H.-G., 2015, “Energy-Based Hyster-
The modeling results of the proposed method showed good esis Model for RC Beam-Column Connections,” ACI Structural Journal,
agreements with the measured backbone curves. Further, the V. 112, No. 2, Mar.-Apr., pp. 157-166.
predicted envelope relations of the proposed method agreed Eom, T.-S.; Kang, S.-M.; Park, H.-G.; Choi, T.-W.; and Jin, J.-M., 2014,
“Cyclic Loading Test for Reinforced Concrete Columns with Continuous
well with those of the ASCE/SEI 41-17 model using statis- Rectangular and Polygonal Hoops,” Engineering Structures, V. 67, May,
tical equations. pp. 39-49. doi: 10.1016/j.engstruct.2014.02.023
3. To define the cyclic relation including hysteresis loop Eom, T.-S., and Park, H.-G., 2010a, “Evaluation of Energy Dissipa-
tion of Slender Reinforced Concrete Members and Its Applications,”
and unloading/reloading stiffness, an energy-based cyclic Engineering Structures, V. 32, No. 9, pp. 2884-2893. doi: 10.1016/j.
model using the energy dissipation ratio (κ) was proposed. engstruct.2010.05.007
Energy dissipation ratios for rectangular and circular Eom, T.-S., and Park, H.-G., 2010b, “Elongation of Reinforced
Concrete Members Subjected to Cyclic Loading,” Journal of Structural
columns were formulated. Deficient energy dissipation in Engineering, ASCE, V. 136, No. 9, pp. 1044-1054. doi: 10.1061/(ASCE)
columns with inadequate reinforcements was also addressed ST.1943-541X.0000201
using empirical reduction factors. The predicted cyclic Eom, T.-S., and Park, H.-G., 2013, “Evaluation of Shear Deformation and
Energy Dissipation of Reinforced Concrete Members Subjected to Cyclic
responses agreed with the measured ones. Loading,” ACI Structural Journal, V. 110, No. 5, Sept.-Oct. pp. 845-884.
In this study, the modeling parameters such as effective Eom, T.-S.; Park, H.-G.; and Kang, S.-M., 2009, “Energy-Based Cyclic
stiffness, yield deformation, ultimate deformation, and Force-Displacement Relationship for Reinforced Concrete Short Coupling
Beams,” Engineering Structures, V. 31, No. 9, pp. 2020-2031. doi:
failure deformation were proposed based on the test results 10.1016/j.engstruct.2009.03.008
of rectangular columns. Thus, for the nonlinear modeling of Ghannoum, W. M., and Matamoros, A. B., 2014, “Nonlinear Modeling
circular columns, further study is required. Parameters and Acceptance Criteria for Concrete Columns,” Seismic
Assessment of Existing Reinforced Concrete Buildings, SP-297, K. J.
Elwood, J. Dragovich, and I. Kim, eds., American Concrete Institute, Farm-
AUTHOR BIOS ington Hills, MI, pp. 1-24.
Tae-Sung Eom is a Professor at Dankook University, Yongin, Gyeonggi, Kim, C.-G.; Eom, T.-S.; and Park, H.-G., 2020, “Cyclic Load Test of
South Korea. He received his BS, MS, and PhD in architectural engineering Reinforced Concrete Columns with V-Shaped Ties,” ACI Structural
from Seoul National University, Seoul, South Korea. His research interests Journal, V. 117, No. 3, May, pp. 91-101.
include experiment and analysis of reinforced concrete and composite Kim, C.-G.; Park, H.-G.; and Eom, T.-S., 2018, “Seismic Performance
structures with an emphasis on performance-based seismic design. of Reinforced Concrete Columns with Lap Splices in Plastic Hinge

Region,” ACI Structural Journal, V. 115, No. 1, Jan., pp. 235-245. doi: Park, H.-G., and Eom, T.-S., 2006, “A Simplified Method for Esti-
10.14359/51701109 mating the Amount of Energy Dissipated by Flexure-Dominated Rein-
Kim, C.-G.; Park, H.-G.; and Eom, T.-S., 2019a, “Cyclic Load Test and forced Concrete Members for Moderate Cyclic Deformations,” Earthquake
Shear Strength Degradation Model for Columns with Limited Ductility Tie Spectra, V. 22, No. 2, pp. 459-490. doi: 10.1193/1.2197547
Details,” Journal of Structural Engineering, ASCE, V. 145, No. 2, pp. 1-17. Priestley, M. J. N.; Verma, R.; and Xiao, Y., 1994, “Seismic Shear
doi: 10.1061/(ASCE)ST.1943-541X.0002254 Strength of Reinforced Concrete Columns,” Journal of Structural
Kim, C.-G.; Park, H.-G.; and Eom, T.-S., 2019b, “Effects of Type of Engineering, ASCE, V. 120, No. 8, pp. 2310-2329. doi: 10.1061/
Bar Lap Splice on Reinforced Concrete Columns Subjected to Cyclic (ASCE)0733-9445(1994)120:8(2310)
Loading,” ACI Structural Journal, V. 116, No. 2, Mar., pp. 183-194. doi: Priestley, M. J. N., 2000, “Performance Based Seismic Design,” Proceed-
10.14359/51711142 ings of the 12th World Conference on Earthquake Engineering, Paper No.
Kunnath, S. K.; El-Bahy, A.; Taylor, A.; and Stone, W., 1997, “Cumu- 2831.
lative Seismic Damage of Reinforced Concrete Bridge Piers (Technical Pujol, S., 2002, “Drift Capacity of Reinforced Concrete Columns
Report NCEER-97-0006),” National Center for Earthquake Engineering Subjected to Displacement Reversals,” PhD thesis, Purdue University, West
Research, Buffalo, NY. Lafayette, IN.
Lim, J.-J.; Park, H.-G.; and Eom, T.-S., 2017, “Cyclic Load Tests of Sezen, H., and Moehle, J. P., 2004, “Shear Strength Model for
Reinforced Concrete Columns with High-Strength Bundled Bars,” ACI Lightly Reinforced Concrete Columns,” Journal of Structural Engi-
Structural Journal, V. 114, No. 1, Jan.-Feb., pp. 197-207. neering, ASCE, V. 130, No. 11, pp. 1692-1703. doi: 10.1061/
Muguruma, H.; Watanabe, F.; and Komuro, T., 1989, “Applicability of (ASCE)0733-9445(2004)130:11(1692)
High Strength Concrete to Reinforced Concrete Ductile Column,” Transac- Xiao, Y., and Martirossyan, A., 1998, “Seismic Performance of High-
tions of the Japan Concrete Institute, V. 11, pp. 309-316. Strength Concrete Columns,” Journal of Structural Engineering, ASCE,
Ng, K. H.; Priestley, M. J. N.; and Park, R., 1978, “Seismic Behaviour of V. 124, No. 3, pp. 241-251. doi: 10.1061/(ASCE)0733-9445(1998)124:3(241)
Circular Reinforced Concrete Bridge Piers (Report 78-14),” Department of
Civil Engineering, University of Canterbury, Christchurch, New Zealand.

Title No. 119-S40
Shear Strength of Flanged Squat Walls with 690 MPa

Reinforcing Bars
by Ju-Hyung Kim and Hong-Gun Park
Six squat walls were tested under cyclic loading to investigate the ACI 318 nominal shear strength (the mean over-strength
effects of flanges and high-strength reinforcing bars on the shear factor of wall specimens with barbells or flanges1,3-6,8,13,15,36-39
strength. Test parameters were the grade of reinforcing bar, wall was 1.95). In particular, for squat walls, vertical reinforcing
geometry, and reinforcement ratio. The strength and deformation bars contributed to shear strength through the strut-tie mech-
capacity of the specimens with high-strength reinforcing bar were
anism. Based on existing test results, ASCE/SEI 43-0540 and
comparable to that of normal-strength reinforcing bar specimen
Gulec and Whittaker24 proposed shear strength equations
with identical design strength. The shear strength of flanged walls
was two times the nominal shear strength of ACI 318-19, regardless that included the effect of vertical reinforcing bars for squat
of the web reinforcement ratios. The test results showed that the walls, which equations are currently used for the seismic
vertical reinforcing bar and concrete in the flanges contributed to evaluation of shear walls in NPPs. In the equations, vertical
the shear strength. Shear strength was calculated using a multiple reinforcing bars in web and flanges significantly increase
shear panel model, which included the effect of flanges. The results wall shear strength. However, the number of existing tests
showed that the strength contribution of the flanges accounted for for squat walls with boundary elements is limited, partic-
40% of the overall shear strength. This result indicates that for ularly for the condition of NPP walls with relatively thick
the economical design of squat walls, the effect of flanges can be flanges and high reinforcement ratios. Thus, the applica-
considered. bility of the design equations to such conditions of NPP
Keywords: cyclic loading tests; flanged walls; high-strength reinforcing
walls should be verified. Further, several studies regarding
bar; shear strength; squat walls. squat walls18,21,41,42 reported that, unlike walls with moderate
aspect ratio,25 the web crushing strength (that is, maximum
INTRODUCTION shear strength of 0.67√fc'Acv [8√fc'Acv]) of ACI 318 signifi-
Existing experimental studies for squat walls (height- cantly underestimated the peak shear strength of squat walls.
to-length ratio ≤ 1.0)1-21 have focused on the evaluation of The present study performed cyclic shear tests to investi-
shear strength according to failure modes.1,20,22-29 While gate the effect of flanges on the shear strength of squat walls
the majority of existing studies have focused on isolated with high-strength reinforcing bars. Several specimens were
rectangular walls,2,16-20,22,23,26,27,30-33 the reality is that many designed with high reinforcement ratios that exceeded the
buildings have walls with flanges. In particular, safety-re- current maximum limitation, to verify the effect of flanges
lated auxiliary buildings in nuclear power plants (NPPs) on the walls with a high reinforcement ratio. On the other
consist of grid-type squat walls. In this case, the shear hand, several specimens were designed with minimum hori-
strength contribution of flanges may be significant. If the zontal or vertical reinforcements, to investigate the effect of
shear strength contribution of flanges is considered for the flanges on the walls with low reinforcement ratios.
design of walls, the constructability and economy of walls
can be improved, particularly for NPPs, using the high rein- RESEARCH SIGNIFICANCE
forcement ratio of large-diameter reinforcing bars. Further, Current shear design methods of walls are based on
in the recent version of ACI 318 (ACI 318-1934), the use of existing test results for rectangular walls. However, when
a high-strength reinforcing bar (Grade 690 MPa [100 ksi]) the effect of flanges and high-strength reinforcing bars are
was permitted for the shear design of walls. Thus, if in addi- considered for walls with flanges, the area of shear rein-
tion to the effect of flanges, the effect of a high-strength reinforcement can be reduced, which improves economy and
forcing bar is considered, the economy and constructability constructability, particularly for NPP walls with high rein-
can be further improved. For this reason, existing studies forcement ratios of large diameter reinforcing bars. The
have conducted tests on the use of high-strength reinforcing present study provides experimental evidence of the effect
bars in walls.16,18,21,35 The results demonstrated that the of flanges and high-strength reinforcing bars on the shear
tested shear strength of walls with high-strength reinforcing strength of squat walls.
bars was equivalent to that of normal-strength reinforcing
bar walls provided that the design shear strength was iden-
tical (that is, the same Vs). MS No. S-2021-101.R1, doi: 10.14359/51734142, received August 2, 2021, and
Regarding the effect of flanges, Kim and Park21 reported reviewed under Institute publication policies. Copyright © 2022, American Concrete
that the shear strength of walls with boundary elements obtained from the copyright proprietors. Pertinent discussion including author’s
(barbells or flanges) was significantly greater than the is received within four months of the paper’s print publication.

point to the wall bottom) was 1850 mm (72.8 in.). For the
five flanged wall specimens, the flange width was 800 mm
(31.5 in.). The concrete strengths of the specimens at the
time of the test were 46.8 and 54.8 MPa (6.79 and 7.95 ksi)
(refer to Table 1). Grade 420 MPa (60 ksi) reinforcing bars
were used for Specimen N1, while Grade 690 MPa (100 ksi)
reinforcing bars were used for the other specimens (H1 to
H5). Table 1 and Fig. 2 list the properties and details of the
specimens.
The design shear strengths Vn of the specimens were
calculated according to the ACI 318-19 equations. To avoid
flexural failure, all specimens were designed with large
flexural strength: the shear force corresponding to moment
capacity Vf was at least two times greater than the minimum
of Vn and the web crushing strength Vn,WC. Consequently, the
Fig. 1—Stress-strain relationship of normal-strength and vertical reinforcement ratios of flanges were significantly
high-strength reinforcing bars. greater than the web reinforcement ratios.
TEST PLAN Specimens N1 and H1 were designed with high horizontal
Test parameters and vertical reinforcement ratios (1.5ρmax = 0.98% for Grade
Six reinforced concrete walls with an aspect ratio (wall 420 MPa [60 ksi], and 0.63% for Grade 690 MPa [100 ksi],
height-to-length ratio) of 1.0 were tested. The specimens respectively) to verify the web crushing strength (that is,
consisted of five flanged walls and one rectangular wall. maximum shear strength) of ACI 318-19, which has been
To evaluate the contribution of flanges to shear strength, all reported to be significantly conservative for squat flanged
specimens were designed to fail in shear. Test parameters walls.21 Further, the two specimens were designed to have
were the grade of reinforcing bar (Grade 420 MPa [60 ksi] identical strength of shear reinforcement (that is, the same
or Grade 690 MPa [100 ksi]), wall geometry (flanged or ρhfyh), to directly compare the overall behavior between the
rectangular cross section), and reinforcement ratio of the normal-strength reinforcing bar specimen and high-strength
web (high or low). In the present study, the high reinforce- reinforcing bar specimen. The effective vertical reinforce-
ment ratio indicates 1.5ρmax (1.5 times the maximum rein- ment ratio of flanges ρvfyv in the two specimens was also the
forcement ratio18 of ACI 318-19 = 1.5(vmax − vc) = 1.5(0.67 same.
− 0.25)√fc'/fy [1.5(8 − 3)√fc'/fy] = 1.5∙0.42∙√50/690 = 0.64%, Specimens H2 and H3 were designed to have a lower
where vmax = 0.67√fc'/fy [8√fc'/fy](web crushing strength), vc vertical reinforcement ratio or a horizontal reinforcement
= 0.25√fc'/fy [3√fc'/fy](concrete shear strength) for walls with ratio that is close to the minimum reinforcement ratio of ACI
height-to-length ratio ≤ 1.5). The low reinforcement ratio 318-19 (0.25%). The other design parameters were the same
indicates the reinforcement ratio of 0.32%, which is slightly as those of H1. In the case of H2, the web vertical reinforce-
greater than the minimum reinforcement ratio (0.25%) of ment ratio (ρv = 0.32%) was only half of the web horizontal
ACI 318-19. reinforcement ratio (ρh = 0.63%). On the other hand, in
H3, the web horizontal reinforcement ratio was lower: ρh =
High-strength reinforcing bars 0.32% and ρv = 0.63%. In ACI 318-19, for special structural
Among the six specimens, five specimens were designed walls, the web vertical reinforcement ratio should be greater
with Grade 690 MPa (100 ksi) reinforcing bars. Figure 1 than the web horizontal reinforcement ratio. However, for
shows the tension test results of two types of Grade 420 MPa flanged walls, the vertical reinforcing bars in flanges are
(60 ksi) reinforcing bars and four types of Grade 690 MPa expected to contribute to the shear force, replacing the role
(100 ksi) reinforcing bars, which were used for test speci- of the vertical reinforcement in the web. By comparing the
mens. Numbers in parentheses denote the mean values of the two specimens (H2 and H3) with the control Specimen H1,
yield strength and tensile strength of each reinforcing bar. the contribution of the vertical and horizontal web reinforce-
High-strength (Grade 690 MPa [100 ksi]) reinforcing bars ment on shear strength can be evaluated for flanged walls.
showed relatively short or no yield plateaus, compared to In the case of Specimen H4, a low ratio was used for both
normal-strength (Grade 420 MPa [60 ksi]) reinforcing bars. horizontal and vertical reinforcement (ρh = ρv = 0.32%).
The yield strain of high-strength reinforcing bars (average Rectangular wall H5 was designed to have the same amount
yield strain of 5490 με, 0.2% offset) was greater than the of horizontal and vertical reinforcement as those of flanged
typical normal-strength concrete compressive strain. wall H4 (that is, web reinforcement area of H5 = overall
reinforcement area of H4 including web and flanges), to
Test specimens compare the shear strengths of the walls with different geom-
The dimensions of the six specimens were the same (the etries. A previous study21 reported that the shear strength of a
height, length, and thickness of the walls were 1600 mm rectangular wall with concentrated vertical reinforcing bars
[63 in.], 1600 mm [63 in.], and 200 mm [7.9 in.], respec- at the boundary region was comparable to that of a flanged
tively). The shear span length (distance from the loading wall with the same area of vertical reinforcing bars. In Spec-
imen H5 of the present study, vertical reinforcements were

Table 1—Test parameters
Specimens N1 H1 H2 H3 H4 H5
Design failure mode Shear Shear Shear Shear Shear Shear
Wall geometry Flanged Flanged Flanged Flanged Flanged Rect.
Total wall length lw, mm 1600 1600 1600 1600 1600 1600
Wall height hw, mm 1600 1600 1600 1600 1600 1600
hw/lw 1.0 1.0 1.0 1.0 1.0 1.0
fc', MPa 54.8 46.8 54.8 54.8 46.8 46.8
Thickness tw, mm 200 200 200 200 200 200
Horizontal reinforcing bar dia. D16 D13 D13 D13 D13 D13
Yield strength fyh, MPa 464 720 720 720 720 720
Web Reinforcement ratio ρh, % 0.98 0.63 0.63 0.32 0.32 0.32
Vertical reinforcing bar dia. D16 D13 D13 D13 D13 D25
Yield strength fyv, MPa 464 720 720 720 720 751
Reinforcement ratio ρv, % 0.98 0.63 0.32 0.63 0.32 2.53
Thickness hbe, mm 200 200 200 200 200 200
Width tbe, mm 800 800 800 800 800 200
Horizontal reinforcing bar dia. D16 D13 D13 D13 D13 D13
Yield strength fyh, MPa 464 720 720 720 720 720
Flange
Reinforcement ratio ρh, % 0.98 0.63 0.63 0.32 0.32 0.32
Vertical reinforcing bar dia. D29 D22/D25 D22/D25 D22/D25 D22/D25 D29
Yield strength fyv, MPa 468 734/751 734/751 734/751 734/751 754
Reinforcement ratio ρv, % 4.82 3.20 3.20 3.20 3.20 6.42
Vf, kN 3437 3489 3269 3489 3269 2513
Design Vn, kN 2046 1996 2041 1212 1167 1167
Vn,WC, kN 1574 1454 1574 1574 1454 1454
Failure mode WC WC WC+DT DT DT DT
Test results Vpeak, kN 2867 2797 2865 2336 2306 1820
Drift at Vpeak, % 1.18 1.16 1.33 0.92 0.99 0.88
Over-strength factor, Vtest/min(Vn,Vn,WC) 1.82 1.92 1.82 1.93 1.98 1.56
Note: DT is diagonal tension failure; WC is web crushing failure; 1 kN = 0.225 kip; 1 MPa = 0.145 ksi; 1 mm = 0.0394 in.
uniformly distributed along the wall length, to evaluate the slip (10). The global lateral displacement was measured by
effect of uniformly distributed vertical reinforcing bars. LVDT 11 located at the center of the top slab. Strains were
also measured by 30 foil type linear strain gauges attached
Test procedure and instrumentation to reinforcing bars in the web and flanges (Appendix A).*
Figure 3 shows that a reversed cyclic lateral load was
applied to the top slab under displacement control. The TEST RESULTS
loading protocol was planned considering the maximum drift Failure modes
ratio of the shear failure specimen.43 Reversed cyclic load- Figure 4 shows the failure mode and crack pattern of the
ings were repeated three times for a given drift ratio. Axial specimens. All specimens failed in shear, as intended in
load was not applied to the specimens because in low-rise the design. In the web, diagonal cracks were propagated.
buildings, the axial force demand is low. Figure 3 shows Due to the diagonal crack mode of the web, shear defor-
that 11 linear variable displacement transducers (LVDTs) mations were observed at the upper part of the tension
were installed to measure the deformation. Two LVDTs 1 flange and lower part of the compression flange. However,
and 2 were diagonally placed in the web to measure shear depending on the web reinforcement ratio, two types of
deformation. Four vertical LVDTs 4 to 7 were placed in the failure modes occurred. For Specimens N1, H1, and H2
flanges to measure flexure deformation. Horizontal LVDT 3
was used to measure the relative shear sliding between the
*
wall and base. Three LVDTs were placed at the wall base
to measure the possible foundation rocking (8 and 9) and from ACI headquarters for a fee equal to the cost of reproduction plus handling at the
time of the request.

Fig. 2—Dimensions and reinforcement details of wall specimens. (Note: 1 mm = 0.0394 in.)
with a high horizontal reinforcement ratio (ρh = 0.98% or Lateral load-displacement relationships
0.63%), closely spaced diagonal cracking and concrete Figure 5 presents the lateral load-displacement relation-
spalling occurred in the web. For those specimens, minor ships of the test specimens. The lateral displacement and
shear cracks were observed in the flanges at peak strength. corresponding drift ratio indicate the net wall deformation,
The test result indicates that the specimens failed in web excluding the horizontal slip and rocking of wall foundation.
crushing. On the other hand, for Specimens H3 and H4 with In the present study, the greater of the positive and negative
a low horizontal reinforcement ratio (ρh = 0.32%), fewer strengths was defined as the peak shear strength. As shear
but wider diagonal tension cracks occurred in the web. At failure occurred, all specimens showed pinched hysteresis
peak strength, diagonal cracking penetrated into the whole loops. The peak strengths of all the specimens were greater
cross section from tension flange to compression flange: the than the nominal shear strength of ACI 318-19 (Vn in Fig. 5).
specimen failed in diagonal tension cracking. Compared to For flanged walls, the mean over-strength factor Vpeak/Vn was
the tension flange, the compression flange showed severe 1.89, while for the rectangular wall specimen H5, the over-
crack damage and steep crack angle, due to the effect of flex- strength factor was 1.56.
ural compression. In the case of the rectangular wall (H5),
after diagonal tension cracking, shear-compression failure Effects of test parameters
occurred in the compression zone. Failure modes of normal- Figure 6 compares the envelope curves of specimens
strength reinforcing bar specimen (N1) and high-strength and presents the effects of high-strength reinforcing bar
reinforcing bar specimen (H1) were almost identical, though (Fig. 6(a)), web horizontal reinforcement ratio (Fig. 6(b)),
due to the lower horizontal and vertical reinforcement ratios, and flange (Fig. 6(c)).
H1 showed a relatively fewer number of diagonal cracks.

Fig. 3—Test setup and load protocol. (Note: 1 mm = 0.0394 in.)
In Fig. 6(a), N1 and H1 with identical nominal shear Figure 6(c) shows that the shear strength of the flanged
strength show identical peak shear strength and deformation wall (H4) was 40% greater than that of the rectangular wall
capacity. This result indicates that 690 MPa reinforcing bars (H5), with the same area of horizontal and vertical reinforce-
can be used for the shear design of squat walls. However, ment (that is, the design shear strength of the two specimens
after shear cracking, H1 shows lower stiffness than N1, due was the same). This result indicates that when the area of
to the smaller reinforcement ratio. vertical reinforcing bars is concentrated in the flanges, the
Figure 6(b) presents the influence of the web horizontal effect of the flanges can increase the shear strength, as well
reinforcement ratio. Specimens with a high horizontal as the flexure strength. Thus, to accurately evaluate the shear
reinforcement ratio (X-symbol, H1 and H2) show greater strength of flanged walls, the contribution of the flange
peak shear strength and deformation capacity, compared should be considered.
to the specimens with a low horizontal reinforcement ratio
(O-symbol, H3 and H4). The difference of peak strength Contributions of displacement components
between the high and low reinforcement ratios is approxi- Figure 7 presents the displacement components of the test
mately 500 kN (112.4 kip), which corresponds to the differ- specimens. Using the instrumentation in Fig. 3, the contri-
ence of shear strength provided by the horizontal reinforce- butions of shear, flexural, and sliding deformations were
ment (535 kN = [(0.63 − 0.32) × 1200 × 200 × 719] (% × measured.44 As shown in the figure, at peak strength, the
mm × mm × MPa). On the other hand, the shear strengths of shear deformation accounted for approximately 60% of the
H1 and H3 were close to those of H2 and H4, respectively. overall lateral displacement for all specimens, including the
This result indicates that the vertical reinforcement ratio in rectangular wall. On the other hand, shear sliding at the wall-
the web did not significantly affect the shear strength of the base interface was restrained by a large amount of vertical
squat walls with an aspect ratio of 1.0. reinforcement in the web and flanges which was used to
avoid early flexural yielding before shear failure. However,

Fig. 4—Crack patterns and failure modes at peak strength.
Fig. 5—Lateral load-displacement relationships of test specimens: (a) N1; (b) H1; (c) H2; (d) H3; (e) H4; and (f) H5. (Note:
1 kN = 0.225 kip; 1 mm = 0.0394 in.)

Fig. 6—Comparison of envelope curves: (a) effect of high-strength reinforcing bar; (b) effect of web horizontal reinforcement
ratio; and (c) effect of flange. (Note:1 kN = 0.225 kip.)
Fig. 7—Displacement components to lateral deformation: (a) N1; (b) H1; (c) H2; (d) H3; (e) H4; and (f) H5.
Fig. 7(c) to (e) shows that specimens with a low vertical after diagonal cracking without flexural yielding. Thus, the
reinforcement ratio showed a relatively greater contribution shear deformation accounted for approximately 60% of the
of shear sliding at the early stage of loading. overall lateral displacement for all specimens, including
the rectangular wall (Fig. 7). For this reason, the crack
Strains of horizontal and vertical reinforcement in width in the web and tension flange was the greatest, which
web cause large strains in the web. This result indicates that the
Strain gauges were placed at three locations of each rein- vertical reinforcing bars resisted shear, as well as flexural
forcing bar to measure the strain distributions. Figure 8 moment. Also, in the case of the horizontal reinforcing bars,
shows the locations of the strain gauges. Each mark in the the average strain was greatest at the mid height of walls
figure indicates the average of the strains measured at the (Fig. 8(b)), though local peak strains occurred along the
same height at peak strength (Appendix A provides the strain diagonal cracks. For specimens with a high reinforcement
distributions of all strain gauges at each loading step). For ratio (H1 and H2), the horizontal reinforcing bars remained
all specimens, the strains of vertical reinforcing bars in the elastic at peak strength. This result agreed with the existing
web and tension flange near the mid height of the wall were test results of squat walls by Luna and Whittaker20 and Kim
greater than those at the bottom of the wall (Fig. 8(a) and and Park.21 On the other hand, for specimens with a low
Appendix A). This is because the specimens failed in shear

Table 2—Average crack angle and crack width at peak strength
Specimen Peak strength, kN Drift ratio at peak, % Crack angle, deg Average diagonal crack width, mm
N1 2867 1.18 41 1.00
H1 2797 1.16 40 1.11
H2 2865 1.33 42 1.48
H3 2336 0.92 38 1.03
H4 2306 0.99 40 1.31
H5 1820 0.88 37 0.95
Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.
Fig. 8—Average shear strain at peak strength: (a) vertical reinforcing bars; and (b) horizontal reinforcing bars.
reinforcement ratio (H3 and H4), several horizontal rein- wall. For this reason, the crack angle was not significantly
forcing bars near diagonal cracks exceeded the yield strain. different, regardless of the reinforcement ratio.
To compare the strain distributions between specimens,
average crack angle (with respect to the vertical axis, Fig. 4) CONTRIBUTION OF FLANGES TO SHEAR
and crack width were evaluated. The average crack angle in STRENGTH
the web was estimated from the major diagonal cracks prop- Shear strength predictions of existing design
agated over more than half the height of the wall, following methods
Luna and Whittaker.20 The average crack width was esti- The shear strengths of the test specimens were compared
mated based on the diagonal deformation (measured using with the predictions of ACI 318-19,34 ASCE/SEI 43-05,40
LVDTs 1 and 2) and the number of major cracks propagated and Gulec and Whittaker,24 which are used in the evaluation
over more than half the height of the wall: the diagonal of the shear wall strength in NPPs (ACI 318 [or ACI 349] for
deformation divided by the number of major cracks. Table 2 squat rectangular walls and ASCE 43-05 or Gulec and Whit-
summarizes the crack angle and average crack width at the taker equations for squat flanged walls). Figure 9 plots the
peak strength of each specimen. over-strength factors Vtest/Vn of the test specimens according
In three specimens in which the horizontal reinforce- to the effective horizontal reinforcement ratio normalized by
ment ratio was the same as the vertical reinforcement ratio the concrete strength (ρhfyh/√fc'; normal-strength reinforcing
(N1: 0.98%, H1: 0.63%, and H4: 0.32%), the average crack bar Specimen N1 was excluded from the figure because the
angles were 40 degrees. Specimen H3, in which the hori- difference between N1 and H1 was negligible). As reported
zontal reinforcement ratio (0.32%) was lower than the in the previous study,21 the design equations generally under-
vertical reinforcement ratio (0.63%), showed greater strains estimated the shear strength of flanged walls (N1 to H4).
of horizontal reinforcing bars and smaller crack angle (38 The mean over-strength factors of the flanged wall speci-
degrees). Rectangular Specimen H5 showed the lowest mens were 1.89 (ACI 318-19), 1.94 (ASCE 43-05), and 1.70
vertical strains in the web, due to the greatest vertical rein- (Gulec and Whittaker), respectively. For rectangular wall
forcement ratio (2.53%), resulting in the lowest crack angle Specimen H5, the over-strength factors were 1.56, 0.85,
of 37 degrees. Due to the small aspect ratio (=1.0) of the and 1.00. In particular, the equation of ACI 318-19 signifi-
squat wall specimens, the major shear crack was propagated cantly underestimated the tested shear strength, owing to the
diagonally connecting the top and bottom corners of the conservative web crushing criteria (that is, the permissible

maximum shear strength). This result agreed with the result strain measurement and crack orientation was conducted. In
of the existing study21 that reported that the mean over- general, the major role of boundary elements is to provide
strength factors of the existing 92 barbell or flanged wall chords for inclined compression strut. However, in the case
specimens were 1.95, 1.59, and 1.67. Although the current of thick flanges (particularly in the case of NPP walls) or
design equations of ASCE and Gulec and Whittaker directly boundary columns, the boundary elements can directly
consider the contribution of vertical reinforcing bars, the provide shear resistance as a small shear panel. Therefore,
predictions (triangle and X-symbols in Fig. 9) showed large in the present study, the boundary elements were considered
variations, according to the vertical reinforcement ratio in as shear panels subjected to combined axial force and shear
the web. For rectangular specimen H5, ASCE and Gulec (Fig. 10(b)).
and Whittaker yielded unsafe prediction, considering the For numerical analysis, a simplified model in Fig. 10(b)
high vertical reinforcement ratio (2.53%) of the web. On the was considered, based on the observation that the shear
other hand, ACI 318 showed an over-strength factor of 1.56, deformations in the flanges were concentrated on the upper
neglecting the contribution of the vertical reinforcing bars. part of the tension flange and the lower part of the compres-
sion flange, due to the diagonal crack patterns of the web
Evaluation of shear strength contributions of web (Fig. 10(a)). The total shear strength was defined as the sum
and flanges of the shear contributions of the web, tension flange, and
The test results showed only the overall behavior of the compression flange (Eq. (1))
walls. Thus, to investigate the shear strength contributions
of the web and flanges, numerical analysis based on the Vpred. = Vw + Vcf + Vtf = τw(lw − 2hbe)tw + τcfhbetbe + τtfhbetbe
(1)
where Vw = τw(lw − 2hbe)tw shear strength of the web; Vcf

= τcfhbetbe shear strength of the compression flange; Vtf =
τtfhbetbe shear strength of the tension flange; lw is total wall
length, 1600 mm; tw is web thickness, 200 mm; hbe is flange
thickness, 200 mm; and tbe is flange width, 800 mm.
The shear stress of each element (τw, τcf, and τtf) was calcu-
lated based on the test results. In the case of the strains of
horizontal reinforcing bars, large variations occurred in the
strain measurement due to local strain peaks near major
cracks. On the other hand, in the case of vertical rein-
forcing bars in web, the variation of strains was relatively
small. Further, average shear strains can be calculated from
diagonal LVDTs 1 and 2, and crack angles can be directly
measured from damage observation. Thus, in the numerical
analysis model, the average shear strain γ, average crack
angle θ, and average strain of vertical reinforcing bars εv
Fig. 9—Comparison with existing design equations.
Fig. 10—Equivalent shear panel model to calculate shear strengths of web and flanges: (a) crack patterns of web and flanges;
and (b) proposed model.

Table 3—Average strain values of Specimen H1 at peak strength
Input (measured data) Estimation
Structural element γ θ, deg εv ε2 εh ε1
Web 0.0078 40 0.0024 0.0008 0.0038 0.0071
Tension flange 0.0078 35 0.0032 −0.0004 0.0044 0.0088
Compression flange 0.0078 29 −0.0005 0.0027 0.0060 0.0066
Fig. 11—Estimation of shear strength contribution of web and flanges: (a) prediction based on measured strains; and (b)
prediction based on MCFT. (Note: 1 kN = 0.225 kip.)
at peak strength were used for input values. The horizontal where f1 = 0.33√fc'/(1 + √500ε1).
strain was estimated from the three input values γ, θ, and εv. Consequently, the shear strengths of the web and flanges
Using Mohr’s circle of strains in the cracked concrete of specimen H1 were calculated as follows: τw = 6.34 MPa,
panel,41,45 the average shear strains of the horizontal rein- τtf = 2.69 MPa, and τcf = 3.52 MPa. Due to the flexural
forcing bars εh and the principal strains ε1 and ε2 were calcu- compression, the shear stress in the compression flange τcf
lated as follows was greater than that of the tension flange τtf. The effect of
compressive stress was considered through the steeper crack
ε2 = γtanθ/2 − εv (2) angle. Considering the dimensions of the web and flanges,
the shear resistances provided by each element were calcu-
lated: Vw = (6.34 MPa × 1200 mm × 200 mm) = 1522 kN, Vtf
εh = {εv + (1 − tan2θ) ε2}/tan2θ (3) = (2.69 MPa × 200 mm × 800 mm) = 430 kN, and Vcf = (3.52
MPa × 200 mm × 800 mm) = 563 kN. The overall predicted
shear strength was Vpred. = 1522 + 430 + 563 = 2515 kN,
ε1 = εv + εh + ε2 (4) while Vtest = 2797 kN (Vtest/Vpred. = 1.11). Following the same
procedure, the shear strength of the web and flanges of all
For example, Table 3 presents the input (measured) and the flanged wall specimens were calculated.
output (estimated) strain values of Specimen H1. It was Figure 11(a) shows the test results (black circle), numer-
assumed that the three shear panels (web, tension flange, and ical analysis (white circle), and the contributions of the
compression flange) are subjected to the same shear strain web (cross mark) and the flanges (X-mark) of the numer-
(γ = 0.0078). The estimated strain of the web horizontal rein- ical analysis based on the measured strains. The percent-
forcing bar (3800 με) yielded a reasonable estimate, when ages in the figure indicate the shear strength contributions
compared to the measured strain distribution of horizontal of the web and flanges. The numbers in parenthesis present
reinforcing bars in Fig. 8(b). the strength ratios of Vtest to Vpred. In general, the numer-
Using the force equilibrium of each panel with the average ical analysis results agreed with the measured peak shear
strain values, the shear stress of cracked concrete for the strength (mean over-strength factor of 1.11 for the five spec-
given strain state was calculated as the sum of the tensile imens). In the numerical analysis, the flanges significantly
strengths of the horizontal reinforcement and concrete increased the shear strength of walls, accounting for approx-
(Eq. (5)). For the tensile stress of cracked concrete f1, an imately 40% of the overall shear strength. The contribution
empirical tension stiffening model41 was used was greater for the specimens with a low horizontal rein-
forcement ratio in the web (H3 and H4). However, consid-
v = (f1 + ρhfsh)cotθ = f1 + ρsEsεhcotθ (5) ering that the sectional area of flanges accounts for 57% of

the gross-sectional area, the ratio of shear contribution is less bar Specimen N1 (Grade 420 MPa [60 ksi]). However, due
than the ratio of sectional area. The lower strength contribu- to the relatively low reinforcement ratios and wider crack
tion is attributed to the lower stiffness of the flanges, due to width of H1, the stiffness was lower than that of N1.
the out-of-plane deformation. 2. The flanged wall Specimens H1 and H2 with high
Alternatively, the shear capacity provided by the three horizontal reinforcement ratios showed greater peak shear
shear panels was estimated following a numerical strain- strength and deformation capacity, compared to the flanged
based approach (Modified Compression Field Theory),45,46 wall Specimens H3 and H4 with low horizontal reinforce-
without using the measured strains (that is, all stresses and ment ratios.
strains were estimated from the force equilibrium, compat- 3. The shear strength of the flanged wall Specimen H4
ibility condition, and material model). In the present study, was 40% greater than that of the rectangular wall Specimen
the effects of flexural compression and tension were addi- H5 with the same area of reinforcement. The result indicates
tionally considered for multiple panels of the web and that the effect of flanges can increase the shear strength, as
flanges. The shear strength of each panel was designed to well as the flexure strength.
be varied according to the flexural compression and tension 4. The current design equations significantly underes-
generated by the applied load. Thus, iterative calculations timate the peak shear strength of squat flanged walls. The
were required to calculate the peak strength satisfying the mean over-strength factors (Vtest/Vpred.) of flanged wall spec-
force equilibrium to the applied load of the cracked panels imens were 1.89 for ACI 318-19, 1.94 for ASCE 43-05,
(Eq. (1)). Each panel was designed to lose its load-carrying and 1.70 for Gulec and Whittaker. The excessive over-
capacity when compressive failure occurs, while in the strength factor of ACI 318 was due to the underestimated
case of reinforcing bar yielding, ductile behavior (main- web crushing strength and lack of consideration of the shear
taining the capacity) was assumed. For diagonal tension contribution of flanges. The shear strength equations of
failure specimens, in general, the shear panel of compres- ASCE 43-05 and Gulec and Whittaker also underestimated
sion flange failed in shear–compression due to the higher the tested peak shear strength, though the contribution of
demand of flexural compression. For web crushing failure flanges was considered.
specimens, relatively weak web shear panels failed in diag- 5. The shear strengths of the tested specimens were
onal compression while both flanges did not reach the peak predicted using a multiple shear panel model based on strain
strength. Appendix B presents the numerical procedure to measurement or modified compression field theory. The two
calculate the shear strength of a flanged wall. prediction results showed that flanges of the test specimens
Figure 11(b) shows the results of the numerical predic- contributed to at least 40% of the overall shear strength.
tion. Due to the simplified compatibility condition of the Due to the flexural compression, the shear strength contri-
procedure (the same shear deformation between the panels, bution of the compression flange was greater than that of the
while ignoring the restraint provided by the rigid top slab), tension flange.
the contribution of the web was generally underestimated,
compared to the evaluation based on the strain measurement AUTHOR BIOS
(Fig. 11(a)). The numerical procedure yielded safe predic- Ju-Hyung Kim is a PhD Candidate at Seoul National University, Seoul,
South Korea. He received his BE and MS from the Department of Architec-
tions of the shear strength, with the mean over-strength ture and Architectural Engineering at Seoul National University.
factor of 1.10.
It should be noted that, in the present specimens, the Hong-Gun Park, FACI, is a Professor in the Department of Architecture
and Architectural Engineering at Seoul National University. He received
vertical reinforcement ratio in the flanges was intentionally his BE and MS in architectural engineering from Seoul National University
increased to avoid early flexural cracking. Thus, the shear and his PhD in civil engineering from the University of Texas at Austin,
contribution of the flanges in the actual walls of existing Austin, TX. His research interests include inelastic analysis and the seismic
design of reinforced concrete structures.
buildings may be lower than the contribution of the present
test specimens. ACKNOWLEDGMENTS
This work was supported by the Korea Institute of Energy Technology
CONCLUSIONS Evaluation and Planning (KETEP) and the Ministry of Trade, Industry &
Energy (MOTIE) of the Republic of Korea (No. 20201510100020).
In the present study, the shear strength of squat (hw/lw =
1.0) flanged walls with high-strength reinforcing bars was
REFERENCES
studied. Six walls were tested to investigate the effects of 1. Barda, F.; Hanson, J. M.; and Corley, W. G., “Shear Strength of
high-strength reinforcing bar, flange, and reinforcement Low-Rise Walls with Boundary Elements,” Reinforced Concrete Structures
ratio. The test results showed that high-strength rein- in Seismic Zones, SP-53, N. M. Hawkins and D. Mitchell, eds., American
Concrete Institute, Farmington Hills, MI, 1977, pp. 149-202.
forcing bars (Grade 690 MPa [100 ksi]) are applicable to 2. Cardenas, A. E.; Russell, H. G.; and Corley, W. G., “Strength of
shear reinforcement, and flanges significantly increased the Low-Rise Structural Walls,” Reinforced Concrete Structures Subjected to
shear strength. The use of high-strength reinforcing bars Wind and Earthquake Forces, SP-63, American Concrete Institute, Farm-
ington Hills, MI, 1980, pp. 221-242.
is expected to resolve the problems in constructability and 3. Paulay, T.; Priestley, M. J. N.; and Synge, A. J., “Ductility in Earth-
economy induced by densely reinforced structures in NPPs. quake Resisting Squat Shearwalls,” ACI Journal Proceedings, V. 79, No. 4,
The principal findings are summarized as follows: July-Aug. 1982, pp. 257-269.
4. Fukuzawa, R.; Chiba, O.; Hatori, T.; Yagishita, K.; Watabe, M.; and
1. The shear strength and deformation capacity of high- Yamanouchi, H., “Study on Load-Deflection Characteristics of Heavily
strength reinforcing bar (Grade 690 MPa [100 ksi]) Specimen Reinforced Concrete Shear Walls,” Proceedings, Ninth World Conference
H1 were comparable to those of normal-strength reinforcing on Earthquake Engineering, Tokyo, Japan, 1988, pp. 543-548.

5. Saito, H.; Kikuchi, R.; Kanechika, M.; and Okamoto, K., “Experi- 26. Moehle, J., Seismic Design of Reinforced Concrete Buildings,
mental Study on the Effect of Concrete Strength on Shear Wall Behavior,” McGraw-Hill Education, New York, 2015, 760 pp.
Proceedings, Tenth International Conference on Structural Mechanics in 27. Kassem, W., “Shear Strength of Squat Walls: A Strut-and-Tie Model
Reactor Technology, Anaheim, CA, 1989, pp. 232-277. and Closed-Form Design Formula,” Engineering Structures, V. 84, Feb.
6. Sugano, S., “An Experimental Study on Strength and Ductility of 2015, pp. 430-438. doi: 10.1016/j.engstruct.2014.11.027
High Strength Concrete Shear Walls (in Japanese),” Meetings, Summaries 28. Ma, J., and Li, B., “Experimental and Analytical Studies on H-Shaped
of Technical Papers of Annual Meeting, 1991, pp. 437-438. Reinforced Concrete Squat Walls,” ACI Structural Journal, V. 115, No. 2,
7. Committee on the Safety of Nuclear Installations OECD Nuclear Mar. 2018, pp. 425-438. doi: 10.14359/51701144
Energy Agency, “Seismic Shear Wall ISP NUPEC’s Seismic Ultimate 29. Moretti, M. L.; Kono, S.; and Obara, T., “On the Shear Strength of
Dynamic Response Test Comparison Report (OCDE/GD(96)188),” Organ- Reinforced Concrete Walls,” ACI Structural Journal, V. 117, No. 4, July
isation for Economic Co-Operation and Development, Paris, France, 1996, 2020, pp. 293-304.
412 pp. 30. Luna, B. N.; Rivera, J. P.; and Whittaker, A. S., “Seismic
8. Kabeyasawa, T., and Hiraishi, H., “Tests and Analyses of High- Behavior of Low-Aspect-Ratio Reinforced Concrete Shear Walls,” ACI
Strength Reinforced Concrete Shear Walls in Japan,” High-Strength Structural Journal, V. 112, No. 5, Sept.-Oct. 2015, pp. 593-604. doi:
Concrete in Seismic Regions, SP-176, American Concrete Institute, Farm- 10.14359/51687709
ington Hills, MI, 1998, pp. 281-310. 31. Yeh, R.-L.; Tseng, C.-C.; and Hwang, S.-J., “Shear Strength of
9. Salonikios, T. N.; Kappos, A.; Tegos, I.; and Georgios, P., “Cyclic Reinforced Concrete Vertical Wall Segments under Seismic Loading,”
Load Behavior of Low-Slenderness Reinforced Concrete Walls: Design ACI Structural Journal, V. 115, No. 5, Sept. 2018, pp. 1485-1494. doi:
Basis and Test Results,” ACI Structural Journal, V. 96, No. 4, July-Aug. 10.14359/51702377
1999, pp. 649-660. 32. Mihaylov, B. I.; Hannewald, P.; and Beyer, K., “Three-Parameter
10. Salonikios, T. N.; Kappos, A. J.; Tegos, I.; and Penelis, G. G., “Cyclic Kinematic Theory for Shear-Dominated Reinforced Concrete Walls,”
Load Behavior of Low-Slenderness Reinforced Concrete Walls: Failure Journal of Structural Engineering, ASCE, V. 142, No. 7, 2016, p. 04016041.
Modes, Strength and Deformation Analysis, and Design Implications,” ACI doi: 10.1061/(ASCE)ST.1943-541X.0001489
Structural Journal, V. 97, No. 1, Jan.-Feb. 2000, pp. 132-142. 33. Alarcon, C.; Hube, M. A.; and De la Llera, J. C., “Effect of Axial
11. Hidalgo, P. A.; Ledezma, C. A.; and Jordan, R. M., “Seismic Behavior Loads in the Seismic Behavior of Reinforced Concrete Walls with
of Squat Reinforced Concrete Shear Walls,” Earthquake Spectra, V. 18, Unconfined Wall Boundaries,” Engineering Structures, V. 73, Aug. 2014,
No. 2, 2002, pp. 287-308. doi: 10.1193/1.1490353 pp. 13-23. doi: 10.1016/j.engstruct.2014.04.047
12. Palermo, D.; Vecchio, F. J.; and Solanki, H., “Behavior of Three-Di- 34. ACI Committee 318, “Building Code Requirements for Structural
mensional Reinforced Concrete Shear Walls,” ACI Structural Journal, Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
V. 99, No. 1, Jan.-Feb. 2002, pp. 81-89. Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
13. Bouchon, M.; Orbovic, N.; and Foure, N., “Tests on Reinforced 35. Cheng, M.-Y.; Hung, S.-C.; Lequesne, R. D.; and Lepage, A.,
Concrete Low-Rise Shear Walls under Static Cyclic Loading,” Proceed- “Earthquake-Resistant Squat Walls Reinforced with High-Strength Steel,”
ings, Thirteenth World Conference on Earthquake Engineering, Vancouver, ACI Structural Journal, V. 113, No. 5, Sept.-Oct. 2016, pp. 1065-1076. doi:
BC, Canada, 2004, pp. 1-10. 10.14359/51688825
14. Massone, L. M.; Orakcal, K.; and Wallace, J. W., “Modeling of Squat 36. Maier, J., and Thürlimann, B., “Bruchversuche an Stahlbeton-
Structural Walls Controlled by Shear,” ACI Structural Journal, V. 106, No. scheiben (in German),” Institut für Baustatik und Konstruktion, Eidgenös-
5, Sept.-Oct. 2009, pp. 646-655. sische Technische Hochschule (ETH) Zürich, Switzerland, 1985, 130 pp.
15. Athanasopoulou, A., and Parra-Montesinos, G., “Experimental Study 37. Nazé, P.-A., and Sidaner, J.-F., “Presentation and Interpretation of
on the Seismic Behavior of High-Performance Fiber-Reinforced Concrete Safe Tests: Reinforced Concrete Walls Subjected to Shearing,” Proceed-
Low-Rise Walls,” ACI Structural Journal, V. 110, No. 5, Sept.-Oct. 2013, ings, Sixteenth International Conference on Structural Mechanics in
pp. 767-777. Reactor Technology, Anaheim, CA, 2001, pp. 1-12.
16. Park, H.-G.; Baek, J.-W.; Lee, J.-H.; and Shin, H.-M., “Cyclic 38. Kono, S.; Tanaka, H.; and Watanabe, F., “Interface Shear Transfer
Loading Tests for Shear Strength of Low-Rise Reinforced Concrete for High Strength Concrete and High Strength Shear Friction Reinforce-
Walls with Grade 550 MPa Bars,” ACI Structural Journal, V. 112, No. 3, ment,” Proceedings, High Performance Materials in Bridges, Kona, HI,
May-June 2015, pp. 299-310. doi: 10.14359/51687406 2003, pp. 319-328.
17. Tran, T. A., and Wallace, J. W., “Cyclic Testing of Moderate-As- 39. Baek, J.-W.; Park, H.-G.; Lee, B.-S.; and Shin, H.-M., “Shear-Fric-
pect-Ratio Reinforced Concrete Structural Walls,” ACI Structural Journal, tion Strength of Low-Rise Walls with 550 MPa (80 ksi) Reinforcing Bars
V. 112, No. 6, Nov.-Dec. 2015, pp. 653-665. doi: 10.14359/51687907 under Cyclic Loading,” ACI Structural Journal, V. 115, No. 1, Jan. 2018,
18. Baek, J.-W.; Park, H.-G.; Lee, J.-H.; and Bang, C.-J., “Cyclic pp. 65-77.
Loading Test for Walls of Aspect Ratio 1.0 and 0.5 with Grade 550 MPa 40. ASCE/SEI 43-05, “Seismic Design Criteria for Structures, Systems,
(80 ksi) Shear Reinforcing Bars,” ACI Structural Journal, V. 114, No. 4, and Components in Nuclear Facilities,” American Society of Civil Engi-
July-Aug. 2017, pp. 969-982. doi: 10.14359/51689680 neers, Reston, VA, 2005, 15 pp.
19. Kolozvari, K.; Arteta, C.; Fischinger, M.; Gavridou, S.; Hube, M.; 41. Bentz, E. C.; Vecchio, F. J.; and Collins, M. P., “Simplified Modified
Isaković, T.; Lowes, L.; Orakcal, K.; Vásquez, J.; and Wallace, J., “Compar- Compression Field Theory for Calculating Shear Strength of Reinforced
ative Study of State-of-the-Art Macroscopic Models for Planar Reinforced Concrete Elements,” ACI Structural Journal, V. 103, No. 4, July-Aug.
Concrete Walls,” ACI Structural Journal, V. 115, No. 6, Nov. 2018, 2006, pp. 614-624.
pp. 1637-1657. doi: 10.14359/51710835 42. Grant, F.; Hardy, G.; and Short, S., “Seismic Fragility and Seismic
20. Luna, B. N., and Whittaker, A. S., “Peak Strength of Shear-Critical Margin Guidance for Seismic Probabilistic Risk Assessments (Product ID
Reinforced Concrete Walls,” ACI Structural Journal, V. 116, No. 2, Mar. 3002012994),” Electric Power Research Institute (EPRI), Palo Alto, CA,
2019, pp. 257-266. doi: 10.14359/51712280 2018, pp. 4-25-4-31.
21. Kim, J.-H., and Park, H.-G., “Shear and Shear-Friction Strengths of 43. ACI Committee 374, “Guide for Testing Reinforced Concrete Struc-
Squat Walls with Flanges,” ACI Structural Journal, V. 117, No. 6, Nov. tural Elements under Slowly Applied Simulated Seismic Loads (ACI
2020, pp. 269-280. doi: 10.14359/51728075 374.2R-13),” American Concrete Institute, Farmington Hills, MI, 2013, 5
22. Wood, S. L., “Shear Strength of Low-Rise Reinforced Concrete pp.
Walls,” ACI Structural Journal, V. 87, No. 1, Jan.-Feb. 1990, pp. 99-107. 44. Hiraishi, H., “Evaluation of Shear and Flexural Deformations of
23. Gulec, C. K.; Whittaker, A. S.; and Stojadinovic, B., “Shear Strength Flexural Type Shear Walls,” Bulletin of the New Zealand Society for
of Squat Rectangular Reinforced Concrete Walls,” ACI Structural Journal, Earthquake Engineering, V. 17, No. 2, 1984, pp. 135-144. doi: 10.5459/
V. 105, No. 4, July-Aug. 2008, pp. 488-497. bnzsee.17.2.135-144
24. Gulec, C. K., and Whittaker, A. S., “Empirical Equations for Peak 45. Vecchio, F. J., and Collins, M. P., “The Modified Compression-Field
Shear Strength of Low Aspect Ratio Reinforced Concrete Walls,” ACI Theory for Reinforced Concrete Elements Subjected to Shear,” ACI Journal
Structural Journal, V. 108, No. 1, Jan.-Feb. 2011, pp. 80-89. Proceedings, V. 83, No. 2, Mar.-Apr. 1986, pp. 219-231.
25. Eom, T.-S.; Park, H.-G.; Kim, J.-Y.; and Lee, H.-S., “Web Crushing 46. Vecchio, F. J., and Collins, M. P., “Compression Response
and Deformation Capacity of Low-Rise Walls Subjected to Cyclic Loading,” of Cracked Reinforced Concrete,” Journal of Structural Engi-
ACI Structural Journal, V. 110, No. 4, July-Aug. 2013, pp. 575-584. neering, ASCE, V. 119, No. 12, 1993, pp. 3590-3610. doi: 10.1061/
(ASCE)0733-9445(1993)119:12(3590)

Title No. 119-S41
Side-Face Blowout Strength of Two-Layer Headed Bars

Embedded in Exterior Beam-Column Joints
by Hye-Jung Sim and Sung-Chul Chun
The side-face blowout area of two-layer headed bars in an exte-  f y ψ e ψ p ψ o ψ c  1.5  f y ψ e ψ p ψ o ψ c  1.5
rior beam-column joint could be overlapped by each headed bar ldt , ACI =   db (MPa) =   db (psi)
and, therefore, the anchorage strength of the individual headed bar  31 f c′   75 f c′ 
could be reduced. However, ACI 318-19 and previous studies did
(1)
not consider the effects of layer-to-layer spacing. To evaluate the
anchorage strength of two-layer headed bars in the exterior beam-
column joint, a simulated exterior beam-column joint test was f dt , ACI =  31ldt  
f c′(MPa) = 
75ldt 
f c′(psi)
1.5  1.5 
conducted with 43 mm (No. 14) headed bars of 550 MPa (80 ksi) of  ψ e ψ p ψ o ψ c db   ψ e ψ p ψ o ψ c db 
design yield strength. The effects of layer-to-layer spacing, embed-
(2)
ment length, and side cover on the anchorage strength of two-layer
headed bars were experimentally estimated. Degradations of the
anchorage strength due to overlapping failure area are very similar where ldt,ACI is the development length of the headed bar; fc′ is
to the degradation of the side-face blowout strength of the group the compressive strength of concrete; fy is the yield strength
anchor subjected to tension, and the side-face blowout strength of of the headed bar; db is the bar diameter; ψe, ψp, ψo, and ψc
two-layer headed bars can be predicted by using a spacing factor. are factors of reinforcement coating, parallel tie reinforce-
A design equation for the development length of two-layer headed ment, side cover and confinement, and concrete strength,
bars is proposed by applying the spacing factor developed in this respectively (refer to Table 25.4.4.3 of ACI 318-19)2; and
study into a design equation for the single-headed bar developed fdt,ACI is the anchorage strength of the headed bars in ACI
from a previous study. 318-192 rewritten from ldt,ACI.
Chun et al.3 performed simulated exterior beam-column
Keywords: exterior beam-column joint; headed bar; side-face blowout
failure; two layers. joint tests to evaluate the side-face blowout strengths of
single-headed bars of maximum diameter 57 mm (No. 18).
INTRODUCTION The test setup is notably similar to that of the hooked bar
A headed bar is mainly used for multi-layer bars, such tests conducted by Marques and Jirsa,4 from which the
as beam-column joints and slab-wall joints. At the beam- ACI 318-145 provisions for hooked bars were developed. A
column joint, the upper reinforcing bars are mainly designed tensile force is applied to the headed bars, which is the longi-
as two layers, and the strain of the outer reinforcing bars tudinal beam reinforcement terminated in the column. The
is larger than that of the inner reinforcing bar. However, in compression zone of the beam is simulated with a steel plate
the ultimate state, the strains of both reinforcing bars greatly bearing against the face of the column. Test variables include
exceed the yield strain, so the anchorage of both bars should the bar diameter of 43 or 57 mm (No. 14 or 18), embedment
be designed for the same yield strength. As shown in Fig. 1, length of 7 to 16db, side cover of 1 or 2db, concrete compres-
the side-face blowout strength of two-layer headed bars on sive strength of 42 or 70 MPa (6 or 10 ksi), and the presence
the exterior beam-column joint is smaller than twice the or absence of transverse reinforcement. The specimens were
strength of a single-headed bar due to the overlapping failure designed to induce side-face blowout failure of the exterior
area of the individual headed bars. Based on the simulated beam-column joint and prevent joint shear failure, and the
beam-column joint tests by Shao et al.,1 ACI 318-192 regu- flexural or shear failure of columns. From the tests, Eq. (3)
lates the design equation Eq. (1) of the development length was proposed for predicting the side-face blowout strength
for headed bars. Design parameters for the development of the single-headed bar, including the effects of embedment
length include the yield strength of headed bars, concrete length, side cover, and transverse reinforcement. Equa-
compressive strength, bar diameter, location of headed bars, tion (3) also does not consider the effects of layer-to-layer
presence or absence of bar coating, and the effects of trans- spacing of two or more layered headed bars
verse reinforcement within joints. However, Eq. (1) does not
consider the effects of layer-to-layer spacing of two or more
layered headed bars. Equation (2), rewritten from Eq. (1),
shows the relationship between the expected strength of ACI Structural Journal, V. 119, No. 2, March 2022.
headed bars and given conditions MS No. S-2021-107.R1, doi: 10.14359/51734338, received August 21, 2021, and

Fig. 1—Side-face blowout failure of two-layer headed bars within exterior beam-column joint.
 l 
f dt ,1 =  3 dt + 30 ψ1 f c′(MPa)
 db 
(3)
 l 
=  36 dt + 360 ψ1 f c′(psi))
 d b 
cso K
ψ1 = 0.75 + 0.25 + 0.4 tr
db db
where ldt is the embedment length of a headed bar; fc′ cannot
be greater than 80 MPa (11.6 ksi); cco is the clear cover, and
cannot be less than 1.0db, nor greater than 3.0db; Ktr is the
transverse reinforcement index defined in Eq. (25.4.2.4b)
of ACI 318-19,2 and cannot be greater than 1.0db; and ψ1
cannot be greater than 1.5.
In Chapter 17, Anchoring to Concrete, of ACI 318-19,2 the Fig. 2—Equivalent single-bar method.
side-face blowout strength of a group of anchor subjected to
tension includes the effects of overlapping of the failure area RESEARCH SIGNIFICANCE
of each individual anchor. When the failure areas overlap, In the exterior beam-column joint with two-layer headed
the strength does not increase in proportion to the number of bars, the side-face blowout area of each headed bar may be
anchors. The side-face blowout strength of the group headed overlapped. The overlap of the failure area causes a reduction
anchor is determined by multiplying the side-face blowout in the side-face blowout strength of the individual headed
strength of the single anchor by a spacing factor if the anchor bar. Twenty-four simulated exterior beam-column joints
center-to-center spacing is less than 6ca1. The capacity of were tested for two-layer headed bars. The test results show
each individual anchor is less than that of the single anchor. that the layer-to-layer spacing of headed bars affects the
In two-layer headed bars, the side-face blowout strength of side-face blowout strength, and the failure area is larger than
the individual bar may also be reduced due to the overlap- that of a group anchor under the same conditions. The equa-
ping failure area. However, the amount of overlapped failure tion including the layer-to-layer spacing factor of headed
area can be different between headed anchors and headed bars was developed by analyzing test data compared with
bars, and it is necessary to investigate the effects of layer-to- the equation of a previous study.3 The proposed equation
layer spacing of two-layer headed bars. provides a safe anchorage design for a two-layer headed bar.
In this study, the effects of the layer-to-layer spacing of
headed bar were examined, which has not previously been METHODS FOR EVALUATING FAILURE AREA
reported. To evaluate anchorage strengths with varying OVERLAPS IN TWO-LAYER HEADED BARS
embedment length, side cover, and layer-to-layer spacing of To predict the side-face blowout strength of two-layer
headed bars in the joint, a simulated exterior beam-column headed bars, two methods (the equivalent single-bar method
joint test was conducted for two-layer headed bars with a and the spacing factor method) are suggested.
design yield strength of 550 MPa (80 ksi). The structural
behavior and strengths of two-layer headed bars were inves- Equivalent single-bar method
tigated, and the side-face blowout strength of two-layer The equivalent single-bar method was suggested with
headed bars was predicted by applying the spacing factor of reference to Section 25.6.1.6 on bundle reinforcement of
group anchor to the proposed equation of the previous study3 ACI 318-19.2 Figure 2 shows that two-headed bars can be
and the design equation of ACI 318-19.2 Finally, a design assumed to be a single bar having twice the cross-sectional
equation for the development length of two-layer headed area of a headed bar. The equivalent diameter db,eq becomes
bars is proposed. 2db, and the equivalent side cover cso,eq can be calculated

Fig. 3—Concrete side-face blowout failure of two head anchors in tension.
by subtracting the equivalent radius db,eq/2 from the edge the single-headed bar by the spacing factor given by Eq. (5).
distance cso + db/2 measured at the center of the single bar. The strength of the individual headed bar can be calcu-
The strength of two-layer headed bars can be estimated by lated by dividing the whole strength by two, the number of
replacing db and cso by the equivalent diameter db,eq and headed bars. Therefore, the side-face blowout strength of the
equivalent side cover cso,eq in Eq. (2) or (3), respectively. individual headed bar from the two-layer headed bars can
be calculated by multiplying Eq. (2) or Eq. (3) by Eq. (7),
Spacing factor method which is obtained by dividing Eq. (5) by 2
The spacing factor method allows the evaluation of the
side-face blowout strength of two-layer headed bars using s
ψ s1 = 0.5 + (7)
a spacing factor based on the failure area of the two-layer 12ca
headed bars, compared to that of the single-headed bar. This
is a typical spacing factor method that uses the equations of
TEST PROGRAM
side-face blowout strength for the group anchor subjected
Test variables and specimen design
to tension. According to Chapter 17 of ACI 318-19,2 the
Table 1 shows a specimen list. Grade 550 MPa (80 ksi)
nominal side-face blowout strength of the anchor group
headed bars of 43 mm (No. 14) diameter were used. The test
Nsbg is given by Eq. (4), which is based on the test results
variables include the embedment length, concrete compres-
of Furche and Eligehausen,6 if an anchor spacing is shorter
sive strength, side cover, layer-to-layer spacing, and trans-
than six times the edge distance. The anchorage strength of
verse reinforcement index. The head dimension is shown in
the group anchor can be calculated by multiplying a nominal
Fig. 4. The ranges of each variable were designed so that
side-face blowout strength Nsb of a single anchor to the
the highest expected strength was over 550 MPa (80 ksi),
anchor spacing factor of Eq. (5). Equation (5) assumes that
the design yield strength of the headed bars. The develop-
the affected distance of side-face blowout failure is 3ca from
ment length of the headed bar is measured from the column
the center of anchor6; therefore, if the anchor spacing is less
face corresponding to the critical section to the bearing face
than 6ca, the failure areas are overlapped, as shown in Fig. 3
of the head. Figure 5 shows the details of specimens simu-
lating the headed bars used as beam bars embedded in the
Nsbg = ψs,ACINsb (4)
joint. Table 1 shows a specimen list including the expected
strengths by applying the equivalent single-bar method
and the spacing factor method into Eq. (2) and (3). It was
 s 
ψ s , ACI = 1 + (5) intended that only side-face blowout failure would occur,
 6ca  and other failures would be prevented.3 Longitudinal and
transverse reinforcing bars of the specimens were placed in

( )
N sb = 13ca Abrg λ f c′(MPa)
(6)
accordance with ACI Committee 3527 to prevent both flex-
ural and shear failures of the column. The breadth (B) of
= (160c a )
Abrg λ f c′(psi) specimens was set so as not to interfere with the two center-
hole-hydraulic rams. The width (H) was minimized as the
where ca is the distance from the center of the anchor to embedment length plus the side cover and thickness of the
the edge of concrete in one direction; s is the center-to- head, so that a compressive strut at the joint restrained the
center spacing of the anchor; Abrg is the net bearing area of headed bars.
the anchor bolt; and λ is a modification factor to reflect the
reduced mechanical properties of lightweight concrete, rela- Test procedure
tive to normal weight concrete. Figure 6 shows the test setup, which was the same as that
If the layer-to-layer spacing of the two-layer headed bars of the study by Chun et al.3 For test convenience, the column
is narrow, the failure areas of the individual headed bars was laid down. The exterior beam-column joint was simu-
overlap, and the side-face blowout strength of the individual lated by applying a tensile force to the headed bars termi-
headed bar decreases. If the side-face blowout behavior of nated on the column and forming a compression region
the two-layer headed bars is similar with that of the group induced by the moment of the beam. The beam concrete was
anchor, the whole side-face blowout strength of two-layer not cast so as to directly apply a tensile force to the headed
headed bars can be obtained by multiplying the strength of bars. Two headed bars were terminated at each front and

Table 1—Test matrix
Expected strength, MPa (ksi)
Transverse Equivalent Spacing factor
reinforce-
S2 S3 S4
fc′, MPa ment
Specimen ID* ldt cso/db s (ksi) B x H x L, mm (in.) (Ktr/db) †
Eq. (2) Eq. (3) Eq. (2) Eq. (3) Eq. (2) Eq. (3) Eq. (2) Eq. (3)
42 544 x 639 x 2091 192 332 197 273 233 323
D43-L13-C1-42 1db — — —
(6.1) (21.4 x 25.2 x 82.3) (27.8) (48.2) (28.6) (39.6) (33.8) (46.8)
42 630 x 639 x 2091 192 398 183 317 204 354
D43-L13-C2-42 — — —
2db (6.1) (24.8 x 25.2 x 82.3) (27.8) (57.7) (26.5) (46.0) (29.6) (51.3)
13db
4db 70 630 x 639 x 2091 248 514 236 409 264 457
D43-L13-C2-70 2db — — —
(10.2) (24.8 x 25.2 x 82.3) (36.0) (74.5) (34.2) (59.3) (38.3) (66.3)
42 630 x 639 x 2091 HD19 @ 192 547 183 418 204 467
D43-L13-C2-42-C — —
(6.1) (24.8 x 25.2 x 82.3) 267 (1.0) (27.8) (79.3) (26.5) (60.6) (29.6) (67.7)
42 544 x 768 x 2435 236 369 243 309 265 337 287 365
D43-L16-C1-42 1db —
(6.1) (21.4 x 30.2 x 95.9) (34.2) (53.5) (35.2) (44.8) (38.4) (48.9) (41.6) (52.9)
42 630 x 768 x 2435 236 442 225 358 238 379 251 400
D43-L16-C2-42 2db —
(6.1) (24.8 x 30.2 x 95.9) (34.2) (64.1) (32.6) (51.9) (34.5) (55.0) (36.4) (58.0)
16db 3db
4db 70 630 x 768 x 2435 305 571 290 462 308 489 325 517
D43-L16-C2-70 2db —
(10.2) (24.8 x 30.2 x 95.9) (44.2) (82.8) (42.1) (67.0) (44.7) (70.9) (47.1) (75.0)
42 630 x 768 x 2435 HD19 @ 236 608 225 473 238 500 251 528
D43-L16-C2-42-C
(6.1) (24.8 x 30.2 x 95.9) 267 (1.0) (34.2) (88.2) (32.6) (68.6) (34.5) (72.5) (36.4) (76.6)
42 544 x 940 x 2893 295 418 303 356 358 421
D43-L20-C1-42 1db — — —
2db (6.1) (21.4 x 37.0 x 113.9) (42.8) (60.6) (43.9) (51.6) (51.9) (61.1)
20db
4db 42 630 x 940 x 2893 295 501 281 413 314 462
D43-L20-C2-42 2db — — —
(6.1) (24.8 x 37.0 x 113.9) (42.8) (72.7) (40.8) (59.9) (45.5) (67.0)
*
L–C––:  is embedded length from surface of specimen to head face normalized by bar diameter;  is side cover normalized by bar diameter;  is design compressive
strength of concrete (MPa); ‘C’ in  represents headed bars confined by hairpin-type transverse reinforcement.
†
Transverse reinforcement index, Ktr = 40Atr/sn (according to Eq. (25.4.2.4b) of ACI 318-19).
rear side of the column, and a total of four headed bars were
loaded. Two headed bars of 43 mm (No. 14) at each side
were connected to the upper flange of the connecting frame.
Two connecting frames were installed and a rod of 57 mm
(2.24 in.) diameter was connected to the lower flange of
each connecting frame, and a tensile force was applied with
a center-hole-hydraulic ram of 3000 kN (674 kip) capacity.
The two hydraulic rams were connected to a pump, and the
same load was applied. The applied loads were measured
with center-hole load cells placed under the hydraulic rams.
If one of the two sets of front and rear headed bars failed,
the oil of the ram connected to the failed set was cut off,
and only the other side of the headed bars was loaded. In
addition, if the 57 mm (2.24 in.) rods break during the test,
it is very dangerous. For test safety, when the load reached
1400 kN (315 kip) of the actual yield load of the 57 mm
(2.24 in.) rod, the test was stopped.
TEST RESULTS
Crack propagation and failure mode
Concrete compressive strengths were tested on the
test days. A 100 x 200 mm (4 x 8 in.) cylinder was used
for the concrete compressive test. Table 2 summarizes Fig. 4—Head dimension.
the compressive strengths by specimen series. The yield for safety reasons, after the applied load reaching 1400 kN
strength, tensile strength, and elastic modulus of the headed (315 kip). The structural behavior of the side-face blowout
bars were 620 MPa (89.92 ksi), 782 MPa (113.42 ksi) and failure of the two-layer headed bars was the same as that of a
204,421 MPa (29,648 ksi), respectively. previous study for single-headed bars.3 Figure 7(a) shows that
The side-face blowout failure occurred in all specimens initial cracks along the headed bars occurred from the column
except for three specimens for which testing was stopped face, and those cracks progressed towards the heads of the

headed bars. As the load increased, as shown in Fig. 7(b), stresses fdt,e in Tables 3 and 4 were calculated by dividing
the cracks widened and inclined cracks occurred due to the the measured loads in each hydraulic ram by twice of the
compressive strut in the joint. At the maximum loading, the cross-sectional area. The failure mode of the three speci-
side-face blowout failure occurred by spalling of the side mens for which testing was stopped for safety is indicated
cover concrete, as shown in Fig. 7(c). There were no signif- by Y in Tables 3 and 4. Figure 8 shows the measured bar
icant differences in crack propagation and failure mode for stresses with varying each variable. The bar stresses increase
specimens showing side-face blowout failure. The three spec- as the compressive strength, the embedment length, and the
imens that were stopped for safety were confined specimens side cover increase, which are included in Eq. (2) and (3).
with transverse reinforcement. The confined specimens had As the layer-to-layer spacing of the headed bars increases,
enhanced resistance by delaying spalling of the side cover the bar stresses also increase. However, as the layer-to-layer
concrete. The initial crack propagation was identical to that of spacing becomes twice from 2db to 4db, their increases are
the other specimens showing side-face blowout failure. If the very small. Comparison of the linear slopes of Fig. 8 shows
load had been increased continuously, the side-face blowout the embedment length has the most effect on the bar stresses.
failure would have occurred at higher loads. The expected strengths are calculated by applying the
equivalent single-bar method or the spacing factor method,
Bar stresses to Eq. (2) and (3); and Fig. 9 compares these calculated
The measured bar stresses fdt,e of the headed bars can values with the measured values. The average and COV of
be calculated by dividing the applied loads by the cross- the test-to-prediction ratios of each prediction are shown
sectional area of the bar. Because two headed bars were in Fig. 9. Because Eq. (2) has a safety factor as a design
loaded in each hydraulic ram at the same time, the bar equation, Eq. (2) gives conservative values, regardless of
using the equivalent single-bar method or the spacing factor
method. Therefore, the development length of the two-layer
headed bar can be safely designed by applying Eq. (2) to
Eq. (7). The values obtained with Eq. (3) multiplied by
Eq. (7) precisely estimated the measured values. Moreover,
the coefficient of variation (COV) of the test-to-prediction
ratios is only 6.58%, implying very high reliability for the
predicted values. For the two methods, the spacing factor
method yields better results with the average test-to-pre-
diction ratio close to 1.0, and the procedure to obtain the
factor is simpler than the equivalent single-bar method. The
Table 2—Test results of concrete

Series fce, MPa (ksi)
42 MPa 47.24 (6.85)
L16
70 MPa 78.97 (11.45)
L13 42 MPa 43.23 (6.26)

L20 70 MPa 86.81 (12.59)
Fig. 5—Details of specimen.
Fig. 6—Test setup.

means that the safety level of high-strength headed bars is
higher than the safety level of normal-strength headed bars.
Figure 11 shows the test-to-prediction ratios for side cover
according to embedment length. The linear slopes are all
almost horizontal, which means that the safety levels are
almost constant.
Figure 12 shows the test-to-prediction ratios for bar
spacing according to embedment lengths. The averages of
the test-to-prediction ratios for layer-to-layer spacing of were
2db, 3db, and 4db were 1.01, 0.98, and 0.95, respectively. As
the layer-to-layer spacing increases, the slope shows nega-
tive (–) values. As increasing the layer-to-layer spacing, the
application of the spacing factor of Eq. (7) may overesti-
mate the anchorage strength. Because overlapping failure
areas is reduced by increasing the layer-to-layer spacing,
the anchorage strength is expected to increase. However,
the increase of layer-to-layer spacing of headed bars causes
smaller strengths than the predicted value with the spacing
factor of Eq. (7). Figure 3 shows that the spacing factor in
Eq. (7) is based on the assumption that the affected distance
is 3ca from the center of anchor. The decrease of safety level
means that there is a difference in the failure area overlaps
between the headed bars and anchors.
Overlapping failure area of headed bars

Figure 13 shows the difference of failure area for side-
face blowout failures of anchors and headed bars. In the
case of the anchor with no deformation on the rod surface,
the anchorage strength is developed only by bearing resis-
tance in the anchor head. Therefore, when side-face blowout
Fig. 7—Crack propagation and side-face blowout of spec- failure occurs, failure areas overlap only near the anchor
imen D43-L20-C2-S4-42. head. However, in the case of headed bars, the anchorage
strength is developed not only by the bearing resistance of
equivalent single-bar method has the average test-to-pre-
the head, but also by the bond resistance along the embed-
diction ratio much lower than 1.0. Because this method is
ment length.8,9 For headed bars, the failure areas overlap both
not conservative, it is not recommended to apply to Eq. (3).
around the head and along the embedment length. Therefore,
If a modified spacing factor based on the test results of the
the overlapping failure area of headed bars is larger than that
headed bars is developed, the strengths of the two-layer
of anchors.
headed bars can be more precisely and reliably predicted.
Proposed spacing factor for headed bars
MODEL FOR SIDE-FACE BLOWOUT STRENGTH
To develop a new spacing factor for headed bars, a regres-
OF TWO-LAYER HEADED BARS
sion analysis of the test results was conducted with the test
The predicted strengths with Eq. (3) and (7) are compared
results. The regression analysis gives a new spacing factor
with the measured bar stresses more precisely to modify the
of 0.5 + s/(14.62ca1). The coefficient of 14.62 is greater
spacing factor.
than the value of 12 in Eq. (7). The increase in the coeffi-
cient is caused by the large overlapping failure area of the
Effects of test variables
headed bars. The factor of Eq. (8) is obtained by simplifying
To investigate whether the spacing factor of Eq. (7) can
the coefficient. Incorporating the spacing factor of Eq. (8)
accurately evaluate the effects of embedment length and
into Eq. (3) yields Eq. (9) to predict the side-face blowout
side cover, test-to-prediction ratios are plotted in Fig. 10 and
strength of individual bars in two-layer headed bars.
11 with varying embedment length and side cover, respec-
tively. The test-to-prediction ratio represents the degree of
conservatism for a given condition and will be expressed as s
ψ 2 = 0.5 + (8)
a safety level. Figure 10 shows the test-to-prediction ratios 15ca
for embedment length according to the bar layer-to-layer
spacing. There were no differences in safety level according  l 
to the bar layer-to-layer spacing, but as the embedment length f dt , 2 =  3 dt + 30 ψ1ψ 2 f c′(MPa)
 db 
increases, the level increases. As the embedment length (9)
increases, the developed bar stress also increases. Figure 10  l 
=  36 dt + 360 ψ1ψ 2 f c′(psi)
 d b 

Table 3—Test results of S2 and S3 series
Test-to-prediction
Equivalent single-bar method Spacing factor method
Failure
Series *
Specimen ID mode† P, kN (kip) fdt,e, MPa (ksi) fdt,equiv,ACI fdt,equiv,Chun fdt,factor,ACI fdt,factor,Chun
SF 790 (178) 272.0 (39.5) 1.35 0.78 1.32 0.95
D43-L13-C1-S42
SF 791 (178) 272.4 (39.5) 1.35 0.78 1.32 0.95
SF 921 (207) 317.1 (46.0) 1.58 0.76 1.66 0.96
D43-L13-C2-S42
SF 919 (207) 316.5 (45.9) 1.57 0.76 1.65 0.95
SF 1285 (289) 442.5 (64.2) 1.60 0.77 1.68 0.97
D43-L13-C2-S70
SF 1264 (284) 435.3 (63.1) 1.58 0.76 1.66 0.96
SF 1304 (293) 449.0 (65.1) 2.23 0.78 2.34 1.02
D43-L13-C2-S42-C
SF 1287 (289) 443.2 (64.3) 2.20 0.77 2.31 1.01
SF 1010 (227) 347.8 (50.4) 1.39 0.89 1.35 1.06
D43-L16-C1-S42
SF 1032 (232) 355.4 (51.5) 1.42 0.91 1.38 1.08
S2
SF 1192 (268) 410.5 (59.5) 1.64 0.88 1.72 1.08
D43-L16-C2-S42
SF 1195 (269) 411.6 (59.7) 1.64 0.88 1.72 1.08
SF 1396 (314) 480.9 (69.7) 1.49 0.79 1.56 0.98
D43-L16-C2-S70
SF 1384 (311) 476.5 (69.1) 1.47 0.79 1.54 0.97
Y 1457 (328) 501.9 (72.8) 2.00 0.78 2.10 1.00
D43-L16-C2-S42-C
Y 1465 (329) 504.5 (73.2) 2.02 0.78 2.11 1.01
SF 1098 (247) 378.1 (54.8) 1.29 0.92 1.26 1.08
D43-L20-C1-S42
SF 1093 (246) 376.4 (54.6) 1.29 0.92 1.25 1.08
SF 1204 (271) 414.6 (60.1) 1.42 0.85 1.49 1.02
D43-L20-C2-S42
SF 1210 (272) 416.7 (60.4) 1.43 0.85 1.50 1.03
SF 967 (217) 333.0 (48.3) 1.33 0.85 1.19 0.93
D43-L16-C1-S42
SF 944 (212) 325.1 (47.1) 1.30 0.83 1.16 0.91
SF 1142 (257) 393.3 (57.0) 1.57 0.84 1.56 0.98
D43-L16-C2-S42
SF 1142 (257) 393.3 (57.0) 1.57 0.84 1.56 0.98
S3
SF 1483 (333) 510.7 (74.1) 1.58 0.84 1.56 0.98
D43-L16-C2-S70
SF 1452 (326) 500.0 (72.5) 1.54 0.82 1.53 0.96
Y 1608 (361) 553.7 (80.3) 2.21 0.86 2.19 1.04
D43-L16-C2-S42-C
Y 1579 (355) 543.7 (78.9) 2.17 0.84 2.15 1.02
*
S–(number):  is bar spacing.
†
SF is side-face blowout failure; Y indicates test stopped after headed bars yielded.
Note: P is measured force of total bar; fdt,e is measured strength of total bar; fdt,equiv,ACI is predicted strength by applying equivalent single-bar method to ACI provision1; fdt,equiv,Chun
is predicted strength by applying equivalent single-bar method to Chun et al. model4; fdt,factor,ACI is predicted strength by applying anchor spacing factor method to ACI provision1;
fdt,factor,Chun is predicted strength by applying anchor spacing factor method to Chun et al. model.4
Figure 14 compares Eq. (9) with the tests, and the average is compared with Eq. (9) in Fig. 14, and gives the average
and COV of the test-to-prediction ratios are 1.02 and 6.3%, and COV of test-to-predictions of 1.01% and 8.72%, respec-
respectively. Figure 15 shows the test-to-prediction ratios tively. For design purpose, a safety factor of 5% fractile
for bar spacing according to embedment lengths. As the bar coefficient is introduced,10 which corresponds to a 5% prob-
spacing increased, the safety level looks a little decrease. ability of non-exceedance with a confidence of 90%. The 5%
However, the average test-to-prediction ratios of the speci- fractile coefficient using these data is 0.84. Incorporating the
mens with 2db, 3db, and 4db spacing are 1.04, 1.01, and 1.00, 5% fractile coefficient into Eq. (9), and solving for the devel-
respectively, meaning that the safety level is maintained for opment length of headed bars, yields Eq. (10)
the large spacing condition.
ldt , p fy
DEVELOPMENT LENGTH = − 10(MPa)
db 2.5ψ1ψ 2 f c′
A set of a total of 107 data, including a previous study (10)
on a single bar3 as well as the test presented in this study, fy
= − 10(psi)
30ψ1ψ 2 f c′

Table 4—Test results of S4 series
Test-to-prediction
Equivalent Anchor factor
Failure
Series Specimen ID mode P, kN (kip) fdt,e, MPa (ksi) fdt,equiv,ACI fdt,equiv,Chun fdt,factor,ACI fdt,factor,Chun
SF 879 (198) 302.7 (43.9) 1.51 0.87 1.24 0.89
D43-L13-C1-S42
SF 861 (193) 296.5 (43.0) 1.47 0.85 1.21 0.88
SF 953 (214) 328.2 (47.6) 1.63 0.79 1.53 0.88
D43-L13-C2-S42
SF 954 (214) 328.5 (47.6) 1.63 0.79 1.53 0.89
SF 1326 (298) 456.6 (66.2) 1.66 0.80 1.55 0.90
D43-L13-C2-S70
SF 1378 (310) 474.5 (68.8) 1.72 0.83 1.62 0.93
SF 1457 (327) 501.7 (72.8) 2.50 0.87 2.34 1.02
D43-L13-C2-S42-C
SF 1442 (324) 496.6 (72.0) 2.47 0.87 2.32 1.01
SF 1054 (237) 362.8 (52.6) 1.45 0.93 1.19 0.94
D43-L16-C1-S42
SF 1062 (239) 365.6 (53.0) 1.46 0.93 1.20 0.94
S4
SF 1291 (290) 444.7 (64.5) 1.78 0.95 1.67 1.05
D43-L16-C2-S42
SF 1254 (282) 431.9 (62.6) 1.72 0.92 1.62 1.02
SF 1330 (299) 458.1 (66.4) 1.42 0.76 1.33 0.83
D43-L16-C2-S70
SF 1303 (293) 448.5 (65.0) 1.39 0.74 1.30 0.82
Y 1615 (363) 556.1 (80.6) 2.22 0.86 2.09 0.99
D43-L16-C2-S42-C
Y 1568 (352) 539.8 (78.3) 2.16 0.84 2.02 0.96
SF 1236 (278) 425.6 (61.7) 1.46 1.04 1.20 1.03
D43-L20-C1-S42
SF 1245 (280) 428.7 (62.2) 1.47 1.05 1.21 1.04
SF 1319 (296) 454.2 (65.9) 1.55 0.93 1.46 1.00
D43-L20-C2-S42
SF 1363 (306) 469.4 (68.1) 1.61 0.96 1.51 1.04
Fig. 8—Bar stresses with varying variables.

Fig. 9—Comparison of measured bar stress with predicted stresses.
Fig. 10—Ratio of test-to-predictions with varying embed- Fig. 11—Ratio of test-to-predictions with varying side cover.
ment length.
cso K s
ψ1 = 0.75 + 0.25 + 0.4 tr , ψ 2 = 0.5 + (11)
db db 15ca1
where ldt,p is the proposed development length of the headed
bar; fc′ cannot be greater than 80 MPa (11,600 psi); cso/db
cannot be less than 1.0 nor greater than 3.0; Ktr/db cannot
be greater than 1.0; ψ1 cannot be greater than 1.5; and ψ2 is
used only for multi-layer headed bars and cannot be greater
than 1.0.
CONCLUSIONS
To evaluate the anchorage strength of two-layer headed
bars in the joint, simulated exterior beam-column joint
tests were performed with headed bars of 550 MPa (80 ksi)
design yield strength. The conclusion can be summarized
as follows: Fig. 12—Ratio of test-to-predictions with varying bar
1. In two-layer headed bars, the failure area of the side- spacing.
face blowout of each bar overlaps and therefore, the side-
face blowout strength of each bar decreases compared to the increase, the bar stress increases, which are included in the
strength of a single-headed bar. design code2 and the previous model.3 While increasing
2. The results of 24 simulated exterior beam-column joint the layer-to-layer spacing of headed bars also results in an
tests show that as the embedment length and side cover increase in the bar stress, the increase is not significant.

Fig. 13—Difference in overlapping failure area between anchors and headed bars.
Fig. 15—Ratio of test-to-prediction of proposed equation

with varying bar spacing.
the equivalent single-bar method to the previous model3
Fig. 14—Comparison of measured bar stresses with
developed for the single-headed bar, the average and coef-
predicted stress of proposed model. (Note: Values in paren-
ficient of variation (COV) of the test-to-prediction ratios are
theses are number of data.)
0.85 and 8.32%, respectively. Also, by applying the spacing
3. Using the equivalent single-bar method, assuming that factor method, the average and COV of the test-to-prediction
the cross-sectional area of an equivalent headed bar is equal ratios are 0.98 and 6.58%, respectively. The spacing factor
to twice the cross-sectional area of a headed bar, and the method has a lower COV than the equivalent single-bar
spacing factor method, the strength degradation of individual method, and provides a relatively accurate estimation of the
bars in two-layer headed bars was evaluated. By applying experimental values.

4. In side-face blowout failures, while the head anchors 408, Bond and Development of Steel Reinforcement. His research interests
include steel anchorage to concrete, composite structures, and the rehabil-
with no deformation on the rod surface resist the applied itation of reinforced concrete structures.
tension only by the head bearing, the headed bars resist the
applied tension by not only the head bearing, but also the ACKNOWLEDGMENTS
bond along the embedment length. Therefore, the affected This research was supported by the National Research Foundation
area of the side-face blowout for headed bars is larger than of Korea (NRF) grant funded by the Korea government (MSIT) (No.
2021R1A2C2010863).
that for anchors. With the spacing factor developed from
the tests on anchors, the safety level could be degraded
by increasing the layer-to-layer spacing of headed bars. A REFERENCES
1. Shao, Y.; Darwin, D.; O’Reilly, M.; Lequesne, R. D.; Ghimire, K.;
new spacing factor for headed bars is developed from the and Hano, M., “Anchorage of Conventional and High-Strength Headed
regression analysis of the test results. With the new spacing Reinforcing Bars,” SM Report No. 117, University of Kansas Center for
Research, Lawrence, KS, 2016, 234 pp.
factor, a model to predict the side-face blowout strength of 2. ACI Committee 318, “Building Code Requirements for Structural
the two-layer headed bar is proposed. The average and COV Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
of the test-to-prediction ratios for the proposed model are Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
3. Chun, S.-C.; Choi, C.-S.; and Jung, H.-S., “Side-Face Blowout Failure
1.02% and 6.3%, and a comparison of tests with predic- of Large-Diameter High-Strength Headed Bars in Beam-Column Joints,”
tion shows that for large bar spacing, the proposed equation ACI Structural Journal, V. 114, No. 1, Jan.-Feb. 2017, pp. 161-172. doi:
yields a safe result for even 4db of spacing between layers. 10.14359/51689161
4. Marques, J. L. G., and Jirsa, J. O., “A Study of Hooked Bar Anchor-
5. A set of a total of 109 data, including a previous study3 on ages in Beam-Column Joints,” ACI Journal Proceedings, V. 72, No. 5, May
a single bar, as well as this test, is compared with the proposed 1975, pp. 198-209.
model, and gives the average and COV of test-to-predictions 5. ACI Committee 318, “Building Code Requirements for Structural
of 1.00% and 8.7%, respectively. For design purposes, an Concrete Institute, Farmington Hills, MI, 2014, 519 pp.
equation for the development length of two-layer headed bars 6. Furche, J., and Eligehausen, R., “Lateral Blow-Out Failure of Headed
as well as a single bar, is developed by incorporating a 5% Studs Near a Free Edge,” Anchors in Concrete-Design and Behavior,
SP-130, American Concrete Institute, Farmington Hills, 1991, pp. 235-252.
fractile coefficient into the proposed model. 7. Joint ACI-ASCE Committee 352, “Recommendations for Design of
Beam-Column Connections in Monolithic Reinforced Concrete Structures
AUTHOR BIOS (ACI 352-02) (Reapproved 2010),” American Concrete Institute, Farm-
Hye-Jung Sim is a PhD Candidate in the Division of Architecture at ington Hills, MI, 2002, 38 pp.
Incheon National University, Incheon, Korea. She received her BS from 8. Thompson, M. K.; Jirsa, J. O.; and Breen, J. E., “Behavior and
Mokpo National University, Jeollanam-do, Korea, in 2015 and her MS from Capacity of Headed Reinforcement,” ACI Structural Journal, V. 103, No. 4,
Incheon National University in 2017. July-Aug. 2006, pp. 522-530.
9. Chun, S.-C., “Components of Side-Face Blowout Strengths of Headed
ACI member Sung-Chul Chun is a Professor in the Division of Architec- Bars in Exterior Beam-Column Joints,” ACI Structural Journal, V. 116,
ture and Urban Design at Incheon National University. He received his No. 2, Mar. 2019, pp. 159-169. doi: 10.14359/51711141
BS, MS, and PhD from Seoul National University, Seoul, Korea, in 1994, 10. Natrella, M. G., Experimental Statistics—National Bureau of Stan-
1996, and 2007, respectively. He is a member of Joint ACI-ASCE Commit- dards Handbook 91, 1996, pp. 1-14.
tees 352, Joints and Connections in Monolithic Concrete Structures, and

NEW
Symposium
Publications from ACI
An ACI Technical Publication An ACI Technical Publication
SYMPOSIUM VOLUME SYMPOSIUM VOLUME
Materials, Analysis, Structural Design Field Applications of Non-Conventional

and Applications of Textile Reinforced Reinforcing and Strengthening Methods
Concrete/Fabric Reinforced for Bridges and Structures
Cementitious Matrix
SP-345
SP-346
Editors:
Yail J. Kim, Steven Nolan, and Antonio Nanni
Editors:
Barzin Mobasher and Flávio de Andrade Silva
SP-345: Materials, Analysis, Structural SP-346: Field Applications of

Design and Applications of Textile Non-Conventional Reinforcing and
Reinforced Concrete/Fabric Reinforced Strengthening Methods for Bridges
Cementitious Matrix and Structures
Several state-of-the-art sessions on textile- A sustainable built environment requires a
reinforced concrete/fabric-reinforced comprehensive process from material selection
cementitious matrix (TRC/FRCM) were organized through to reliable management. Although
by ACI Committee 549 in collaboration with traditional materials and methods still dominate the
RILEM TC MCC during the ACI Fall 2019 design and construction of our civil infrastructure,
Convention in Cincinnati, OH, and the ACI nonconventional reinforcing and strengthening
Virtual Technical Presentations in June 2020. The methods for concrete bridges and structures can
forum provided a unique opportunity to collect address the functional and economic challenges
information and present knowledge in the field facing modern society.
of TRC and FRCM as sustainable construction
materials. Available in PDF format: $69.50
(FREE to ACI members)
Available in PDF format: $69.50
(FREE to ACI members)
+1.248.848.3700 • www.concrete.org    
Title No. 119-S42
Cyclic Behavior of Ultra-High-Performance Concrete Shear

Walls with Different Axial-Load Ratios
by Rui Hu, Zhi Fang, and Baodan Xu
This paper presents a comprehensive study on the cyclic behavior (Li et al. 2013). The shear-span ratio of the shear wall will
of ultra-high-performance concrete (UHPC) shear walls with be reduced, which leads to limited ductility and energy-
different axial-load ratios. A total of four full-scale UHPC shear dissipation capacity (Cao et al. 2006; Kuang and Ho 2018;
walls with test axial-load ratios of 0.1, 0.3, 0.35, and 0.4 were Mohamed et al. 2014). Given the advantages of UHPC,
tested. The seismic behavior of UHPC shear walls was evaluated,
the seismic behavior of UHPC shear walls is a promising
and the influences of the axial-load ratio on the cumulative energy
research direction.
dissipation, strength degradation, and stiffness degradation of the
UHPC shear walls were analyzed. The test results indicated that While UHPC material has the advantages of high ductility
the UHPC shear walls exhibited high energy-dissipation capacity and energy-dissipation capacity, few studies have inves-
and favorable ductility. Based on the test data and code require- tigated the seismic behavior of UHPC shear walls (Hu et
ments, the upper limit of the design axial-load ratio of the UHPC al. 2021). Hung et al. (2017) studied the seismic behavior
shear walls is suggested to be taken as 0.7. The authors proposed a of UHPC squat shear walls with a shear-span ratio of 0.73
seismic damage model and damage-level classification standard to without axial load. They noted that adding steel fibers
evaluate the damage level of the UHPC shear walls. enhanced the strength, confinement, and crack-width control
of UHPC squat shear walls, allowing the walls to exhibit
Keywords: high axial-load ratio; seismic behavior; shear wall;
ultra-high-performance concrete (UHPC).
ductile flexural-dominant behavior. Hung and Hsieh (2020)
also studied the shear behavior of squat UHPC shear walls
INTRODUCTION under cyclic lateral loads. They found that the addition of
Ultra-high-performance concrete (UHPC) is character- steel fibers transformed the damage pattern from significant
ized by good mechanical properties, high energy dissipation, concrete spalling to localized diagonal cracks. Tong et al.
favorable ductility, and excellent durability. Due to its high (2020) investigated the seismic behavior of UHPC shear
compressive strength, UHPC’s ratio of compressive strength walls with different shear-span ratios (1.0, 1.5, and 2.0) at an
to density is approximately four times that of normal concrete axial-load ratio of 0.11. They found that all the UHPC shear-
(Shi et al. 2015). UHPC can be used to form lighter and wall specimens exhibited high ductility, and the ductility
thinner components, especially in the case of compressive of the shear walls increased with an increase in the shear-
components with high axial-load ratios. The self-weight of span ratio. Dacanay (2016) analyzed the behavior of UHPC
UHPC structures would be significantly lower, which would shear walls and found that UHPC shear walls have equal
reduce the earthquake action accordingly. Thus, UHPC uncracked stiffness and superior post-cracking capacity at a
structures can be expected to achieve excellent seismic thickness 58% and a weight 59% that of high-performance
behavior (Yang and Okumus 2017; Wang et al. 2018). More- concrete (HPC) shear walls. Hung and El-Tawil (2011)
over, UHPC with high strength and large ultimate strain analyzed the seismic behavior of a coupled wall system with
gives an alternative solution for undesirable damage modes, high-performance fiber-reinforced concrete (HPFRC) mate-
such as bond failure, and excessively widening cracks, rials in critical regions and found that the use of HPFRC
which could appear when high-strength steel bars are used in in place of normal concrete led to a good overall seismic
normal-concrete structures (Hung and Chueh 2016). It can response of the structure.
be expected that UHPC is suitable to be used in high-rise From the aforementioned works, it can be found that the
buildings, especially in earthquake-prone areas (Zohrevand existing experimental data are still insufficient to provide
and Mirmiran 2013; Marchand et al. 2019). references for the design of UHPC shear walls. The
Shear walls are commonly used in high-rise buildings to conducted shear-wall specimens were evaluated with low
resist lateral loads and earthquake action (Mohamed et al. axial-load ratios or even without axial-load ratios (Dang
2014). However, holes are normally arranged on the shear et al. 2016). Studies on the seismic behavior of UHPC shear
walls considering the architectural function or equipment walls at high axial-load ratios have never been reported, and
layout. Based on their structural behavior, shear walls can be specific seismic design suggestions still need to be elabo-
categorized according to the height and width of the holes rated based on the experiments. Therefore, to investigate
(Peng et al. 2006). When the height of the hole is small and the seismic behavior of UHPC shear walls, four full-scale
the width is large, a shear wall can be categorized as a wall ACI Structural Journal, V. 119, No. 2, March 2022.
MS No. S-2021-126.R2, doi: 10.14359/51734339, received September 24, 2021, and
frame. In such cases, the stiffness of the coupling beam will reviewed under Institute publication policies. Copyright © 2022, American Concrete
be close to or even higher than that of the wall. The inflec- Institute. All rights reserved, including the making of copies unless permission is
tion point will occur at the shear wall in most of the floors closure, if any, will be published ten months from this journal’s date if the discussion

Fig. 1—Design details of test specimens. (Note: Units in mm: 1 mm = 0.0394 in.)
UHPC shear walls at different axial-load ratios underwent of 850 x 120 mm (33.5 x 4.7 in.). All the specimens were
quasi-static cyclic loading. The seismic behavior of the vertically reinforced with four steel bars 10 mm (0.4 in.)
UHPC shear walls was evaluated, and the influences of the in diameter in the web region (ρw = 0.58%), and six steel
axial-load ratio were analyzed. A design axial-load ratio bars 10 mm (0.4 in.) in diameter in the boundary element
limit was suggested based on the evaluation of the deforma- (ρb = 1.96%). The horizontal reinforcements were steel
tion capacity of UHPC shear walls, and a seismic damage bars 10 mm (0.4 in.) in diameter with the space of 200 mm
model and damage-level classification standard were estab- (7.9 in.) (ρh = 0.65%), and the stirrups in the boundary
lished to evaluate the damage level of the UHPC shear walls. element were steel bars 8 mm (0.3 in.) in diameter with the
Note that all the tested UHPC shear walls exhibited flexural- space 200 mm (7.9 in.) (ρsv = 1.07%). Note that no coarse
dominated failure, and the shear behavior of the UHPC shear aggregates were contained in the UHPC. It can be applied
walls is not discussed in this study. Thus, further experi- to structural members with dense reinforcement distribution
mental and finite element studies on the seismic behavior of (Shi et al. 2015; Hu et al. 2021).
the UHPC shear walls with different parameters still need to Each specimen included a UHPC wall element, reinforced
be carried out, and more design suggestions for the UHPC concrete base, and loading stub. The UHPC wall element
shear walls need to be proposed. was prefabricated with a height of 2660 mm (105 in.). Each
end of the precast wall element was inserted 30 mm (1.2 in.)
RESEARCH SIGNIFICANCE into the mold of the reinforced concrete (RC) base and
Although UHPC structures are expected to achieve good loading stub. For the shear-wall specimen at an axial-load
seismic behavior, existing experiments are still insufficient ratio of 0.1 and 0.3, the concrete with the design strength of
to provide references for the design of UHPC shear walls. In 55 MPa (8 ksi) was placed into the mold of the RC base and
this paper, full-scale UHPC shear walls with different axial- loading stub. However, considering the bearing capacity of
load ratios were tested. A design axial-load ratio limit was the concrete base and loading stub, UHPC was used to cast
suggested, and a seismic damage model was established to the concrete base and loading stub of the shear-wall spec-
evaluate the seismic damage level of the UHPC shear walls. imen at an axial-load ratio of 0.35 and 0.4. The concrete base
This study is expected to provide foundations for further and loading stub were demolded when the strength of the
studies on the seismic behavior of UHPC shear walls and concrete reached 75% of its design strength.
contribute to the application of UHPC in building structures. The test axial-load ratio (nt) of the shear wall can be deter-
mined with Eq. (1)
WALL SPECIMENS
Specimen details Nt
nt = (1)
A total of four full-scale shear-wall specimens were fabri- f cm A
cated and tested. Figure 1 shows the dimensions and rein-
forcement layout of the tested specimens. The UHPC shear in which Nt is the axial load; fcm is the tested axial compres-
walls had a height of 2600 mm (102.4 in.) and a cross section sive strength of UHPC, obtained from compression tests of
UHPC prisms; and A is the sectional area of the shear wall.

The design axial load (N) can be written as Table 1—Weight mixture proportions of UHPC
Water-
N = γFNt (2) binder
(cement +
in which γF is the partial safety factor for dead load, taken as 42.5R Fly Silica Quartz Water silica fume
1.35 according to fib Model Code 2010 (fib 2013). Material cement ash fume Slag sand reducer + slag)
The design axial compressive strength of UHPC can be Mixture
1.00 0.1 0.25 0.15 1.1 0.025 0.17
written as proportions
fcd = fcm/1.1γm (3) Table 2—Mechanical properties of UHPC

fcu, MPa (ksi) fc, MPa (ksi) Ec, MPa (ksi) ft0, MPa (ksi) ftu, MPa (ksi)
in which fcd is the design value of axial compressive strength;
172 (25) 140 (20) 49,200 (7134) 11.6 (1.68) 12.0 (1.74)
and γm is the partial safety factor, taken as 1.5 according to
Note: fcu is compressive strength of standard 100 mm (3.9 in.) cubes; fc is compressive
NF P 18-710 2016 (AFNOR 2016). strength of 100 x 100 x 300 mm (3.9 x 3.9 x 11.8 in.) prisms; ft0 is initial cracking
Thus, the relationship between the design axial-load ratio strength or matrix cracking strength of UHPC under uniaxial tension; ftu is tensile
(n) and test axial-load ratio (nt) can be written as strength of UHPC under uniaxial tension.
Based on the mixture proportions listed in Table 1, the raw

n = γnnt (4)
material was mixed in a blender for 20 minutes and then
cast into the mold. The UHPC was cured at room tempera-
where γn is the factor for the design axial-load ratio, taken as
ture for 48 hours, followed by steam curing at a tempera-
2.2 for UHPC shear walls.
ture of 90 ± 5°C for 48 hours. The steam was generated by
The test axial-load ratios of the shear-wall specimens were
a steam engine and transported by a pipeline. During the
set as 0.1, 0.3, 0.35, and 0.4. Therefore, the corresponding
steam curing, the UHPC wall element was covered with an
design axial-load ratios of the shear-wall specimens were
insulation film, and thermometers were used to monitor the
0.22, 0.66, 0.77, and 0.88, respectively.
temperature. The heating and cooling rate were controlled
Considering UHPC shear walls are thinner compared
at 12°C/h, to prevent thermal cracking (T/CBMF 37-2018
to normal-concrete shear walls, the critical buckling force
2018). After the steam curing, the UHPC wall element was
should be considered in the design of the UHPC shear-wall
demolded when the temperature difference between the
specimens. The critical buckling force can be calculated
surface of the specimen and the environment was less than
with Eq. (5) (Li 2001)
20°C. The UHPC cubes and prisms for determining the
properties of the material were cured under the same condi-
2 EI tions as the wall element.
N cr (5)
H
2
Table 2 lists the mechanical properties of the UHPC. The
cubic compressive strengths fcu of the UHPC were measured
where EI is the elastic stiffness of the shear-wall speci-
from six cubes of 100 x 100 x 100 mm (3.9 x 3.9 x 3.9 in.).
mens; H is the height of the shear-wall specimens; and μ
The axial compressive strength fc and compressive elastic
is the length coefficient, taken as 1.0 for the out-of-plane
modulus E0 of the UHPC were measured from three 100 x
stability calculation.
100 x 300 mm (3.9 x 3.9 x 11.8 in.) prisms in compres-
The critical buckling force of the shear-wall specimens was
sion. The initial cracking strength or elastic limit strength
calculated as 10,474 kN (2355 kip). The maximum applied
ft0 and tensile strength ftu of the UHPC were measured from
axial force was 5712 kN (1284 kip), which is less than the
the three axial tension tests. Figures 2(a) and (b) give the
calculated critical buckling force of the specimens. Thus, the
specimens of the tension tests and the installation of the
buckling of the shear-wall specimens can be avoided.
uniaxial tension test, respectively. Figures 3(a) and (b) give
the measured compressive and tensile stress-strain curves of
Material properties
the UHPC, respectively.
Table 1 gives the mixture proportions of the UHPC: 42.5
Table 3 provides the measured mechanical properties of
portland cement in China; fly ash with an activity index
the reinforcement bars. The yield strength of the steel bars
exceeding 80 and 110% at 7 and 28 days, respectively;
10 and 8 mm (0.4 and 0.3 in.) in diameter were 644 and
slag with a specific surface exceeding 800 m2/kg (3906 ft2/
534 MPa (93.4 and 77.4 ksi), respectively, measured from
lb); silica fume with an average grain size of 0.1 μm (3.9 ×
three tensile tests. All the values presented in Tables 2 and 3
10–5 in.); quartz sand with grain sizes from 0.38 to 0.83 mm
are the average of the corresponding test results.
(0.015 to 0.033 in.); and high-range water-reducing admix-
ture (HRWRA) with a water-reduction ratio above 30%.
TEST SETUP AND PROCEDURE
Straight steel fibers with a length of 13 mm (0.51 in.) and
Quasi-static tests of all the specimens were conducted
a diameter of 0.22 mm (0.009 in.) were used in the UHPC.
on the Hunan University Multi-Usage Structural Testing
The tensile strength of the steel fibers was higher than
System (HNU-MUST). The testing system included four
2000 MPa (290 ksi). The steel-fiber fraction of the UHPC
vertical actuators and two lateral actuators. HNU-MUST can
was 2% by volume.
provide a maximum vertical load of 20,000 kN (4496 kip)
in compression and 10,000 kN (2248 kip) in tension. It can

provide a maximum lateral load of 4000 kN (8992 kip) in reinforcement, the lateral reinforcement, and the stirrups in
both directions. The maximum vertical and lateral displace- the bottom region of the shear wall, as shown in Fig. 1.
ments of the testing system were 500 mm (19.7 in.) and During the test, a constant vertical load was applied first,
±400 mm (15.7 in.), respectively. Figures 4(a) and (b) give followed by a lateral drift ratio, as shown in Fig. 6. The drift
the experimental setup. The concrete loading stub of the ratio was defined as the ratio of the lateral displacement at
shear-wall specimen was fixed to the loading platform of the top of the shear wall to the height of the shear wall. Each
HNU-MUST with eight steel rods. The concrete base was target drift ratio was applied three times to investigate the
firmly fastened to a composite foundation with four steel strength degradation. The loading was stopped when the
rods. The composite foundation was specially designed to lateral force dropped to less than 85% of the maximum lateral
support the specimen and restrain any horizontal sliding. The force or the specimen could not sustain the vertical load.
foundation was fixed to the reaction floor with 16 steel rods.
Moreover, four more steel rods were fastened to the reac- EXPERIMENTAL RESULTS
tion floor on the side of the composite foundation to restrain SW-0.1 was tested under a cyclic lateral load and a
horizontal sliding. Note that section rotation at both ends of constant vertical load of 1428 kN (321 kip). Figure 7(a)
the shear-wall specimen was restrained fully. The inflection presents the failure mode of SW-0.1, in which the black
point is expected to appear at the midheight of the specimen. lines were the cracks that occurred at the positive displace-
Linear variable differential transformers (LVDTs) No. 1 ment, and the red lines were the cracks that occurred at the
to 5 detected the lateral displacement of the shear wall, as negative displacement. Figure 8(a) presents the hysteretic
shown in Fig. 5. The shear displacement was detected with response of specimen SW-0.1. At a drift ratio of 0.1% (lateral
LVDT No. 6 and 7 (Hung and Chueh 2016). The sectional force of 457 kN [103 kip]), the initial flexural cracks were
curvature at the bottom of the specimen was detected with observed. At a drift ratio of 0.6% (lateral force of 620 kN
dial indicators No. 8 and 9. The out-of-plane displacement [140 kip]), more cracks occurred, and the maximum crack
was measured with dial indicators No. 10 and 11. Strain width increased to 0.18 mm (0.007 in.). Four diagonal cracks
gauges were used to measure the strain in the vertical appeared on the web region of the specimen. When the drift
ratio reached 0.86% (lateral force of 678 kN [153 kip]),
two flexural cracks—that occurred at the positive and the
negative drift ratio of this loading stage, respectively—were
connected. At a drift ratio of 1.3% (maximum lateral force
of 794 kN [179 kip]), local concrete crushing appeared at
the corners of the shear wall, and the maximum crack width
was 0.22 mm (0.009 in.). When the drift ratio reached 3.0%
(lateral force of 671 kN [151 kip]), the maximum crack
width was 0.4 mm (0.016 in.), the lateral load decreased to
85% of the maximum load, and the loading was stopped.
SW-0.3 was tested under a cyclic lateral load and a constant
vertical load of 4284 kN (964 kip). Figures 7(b) and 8(b) give
Table 3—Mechanical properties of bars

Steel bars d, mm (in.) fy, MPa (ksi) fu, MPa (ksi) E0, MPa (ksi)
D10 10 (0.4) 644 (93) 819 (119) 212,000 (30,740)
D8 8 (0.3) 534 (77) 694 (101) 187,000 (27,115)
Fig. 2—Axial tensile test of UHPC. (Note: Units in mm: Note: d is diameter of bars; fy is yield strength of steel bars; fu is tensile strength of bars;
1 mm = 0.0394 in.) E0 is elastic modulus of bars.
Fig. 3—Measured stress-strain curve of UHPC.

Fig. 5—Measurement of specimens. (Note: Units in mm:
1 mm = 0.0394 in.)
Fig. 4—Test setup of shear walls.

the failure mode and hysteretic response of SW-0.3, respec-
tively. Initial cracks were observed at a drift ratio of 0.17%
(lateral force of 819 kN [184 kip]). At a drift ratio of 0.93%
(lateral force of 1478 kN [333 kip]), the cracks extended into Fig. 6—Loading system.
the web region of the shear wall, and the maximum crack
concrete crushing was observed at four corners of the spec-
width increased to 0.07 mm (0.003 in.). At a drift ratio of
imen. The width of the crushing section was approximately
1.42%, the specimen reached its maximum lateral force of
200 mm (7.9 in.). With the continuous increase of the drift
1509 kN (340 kip), local concrete crushing was observed at
ratio, the width of the crushing section increased. When the
the shear-wall corners, and the maximum crack width was
drift ratio reached 1% (lateral force of 1591 kN [358 kip]),
0.11 mm (0.004 in.). At a drift ratio of 2.1%, severe concrete
the whole cross section at the top of the specimen was
crushing was observed at the top left corner of the specimen,
damaged suddenly with a huge sound of concrete crushing.
the maximum crack width was 0.21 mm (0.008 in.), and
A large piece of concrete at the top of the specimen was
the lateral force decreased to 1270 kN (286 kip), 84% of
stripped, but still connected to the shear wall by steel fibers.
the maximum lateral load. Because of the higher axial load,
The vertical and lateral loads decreased rapidly, while the
much fewer cracks appeared on specimen SW-0.3, and the
drift ratio increased. Finally, the lateral load dropped to zero
cracks were much shorter than those on specimen SW-0.1.
when the drift ratio reached 1.3%. It can be observed that
SW-0.35 was tested under a cyclic lateral load and a
the failure mode of the shear wall has been changed when
constant vertical load of 4998 kN (1125 kip). Figures 7(c)
the axial-load ratio increased to 0.35. The failure of spec-
and 8(c) give the failure mode and hysteretic response of
imen SW-0.35 was sudden, and the concrete damage was
SW-0.35, respectively. At a drift ratio of 0.22% (lateral force
far more serious compared to the shear wall at a lower axial-
of 1013 kN [228 kip]), the initial cracks occurred on the
load ratio. The test was finally stopped because specimen
bottom right and the top left corner of the specimen. At a drift
SW-0.35 could not continue to withstand the vertical load.
ratio of 0.63% (lateral force of 1511 kN [340 kip]), the flex-
SW-0.4 was tested under a cyclic lateral load and a
ural cracks extended to the length of 60 mm (2.4 in.) long,
constant vertical load of 5712 kN (1285 kip). Figures 7(d)
and minor concrete crushing was observed at the corner of
and 8(d) present the failure mode and hysteretic response
the specimen. At a drift ratio of 0.80%, the specimen reached
of SW-0.4, respectively. At a drift ratio of 0.22% (lateral
its maximum lateral load of 1608 kN (362 kip), the sound
force of 1057 kN [238 kip]), the initial cracks occurred on
of the concrete crushing appeared constantly, and severe
the bottom right and the top left corner of the specimen. At

Fig. 7—Failure patterns of specimens.
a drift ratio of 0.39% (lateral force of 1415 kN [318 kip]), displacements of SW-0.1, SW-0.3, SW-0.35, and SW-0.4
local concrete crushing was observed at four corners of the were 13.2 mm (0.52 in.), 10.2 mm (0.4 in.), 2.9 mm (0.11 in.),
specimen. At a drift ratio of 0.58%, the specimen reached and 1.1 mm (0.04 in.), accounting for 17%, 19%, 11%, and
its maximum lateral load of 1635 kN (369 kip), the concrete 6% of the total lateral displacement at the top of the spec-
damage at four corners of the specimen aggravated. At a imen, respectively. The concrete damage aggravated as the
drift ratio of 0.65% (lateral force of 1629 kN [367 kip]), axial-load ratio increased. When the test axial-load ratio is
severe concrete damage initiated from the top right corner less than 0.3, only local concrete crushing was observed at
of the specimen, and a large piece of concrete at the top of the corners of the shear-wall specimen. When the test axial-
the specimen was stripped suddenly with a huge sound of load ratio increased to 0.35 and 0.4, the whole cross section
concrete crushing. The top part of the concrete piece was at the top of the specimen was damaged, and the shear-wall
separated from the shear wall, but the bottom part of the specimen failed suddenly. Note that none of the specimens
concrete piece was still connected to the shear wall by steel had out-of-plane displacement greater than 2 mm (0.08 in.),
fibers. After the sudden crushing, a number of horizontal and no out-of-plane instability was observed during any of
cracks suddenly appeared at the bottom right corner of the the tests.
shear wall. The vertical and lateral loads rapidly dropped to
zero. The specimen could not continue to bear the vertical DISCUSSION
load, and the test was stopped. The failure mode of specimen Ductility
SW-0.4 was similar to that of specimen SW-0.35, but the Figure 9 gives the strength envelopes of the shear walls.
concrete damage of specimen SW-0.4 was more serious. Table 4 gives the characteristic points of the shear-wall spec-
In general, all the tested UHPC shear walls exhibited imens. The yielding point of the shear walls was determined
flexural-dominated failure. The measured maximum shear based on the yield of the outermost steel bars. The ultimate

Fig. 8—Hysteretic responses of specimens. (Note: 1 kN = 0.225 kip.)
displacement was determined as the maximum lateral
displacement when the lateral load declined to less than 85%
of the maximum lateral load or before the sudden drop of the
lateral load. The ductility factor is calculated by the ratio of
the ultimate drift ratio to the yielding drift ratio.
The UHPC shear-wall specimens can be compared to
normal-concrete shear walls with similar conditions. Note
that several normal-concrete shear walls tested by Zhang et
al. (2008; 2009) had similar geometric dimensions and rein-
forcement ratios as the tested UHPC shear-wall specimens.
They found that the ductility factor of the normal-concrete
shear wall (shear-span ratio of 1.88) at an axial-load ratio of
0.158 was 3.7 (Zhang et al. 2008). Another normal-concrete
shear wall (shear-span ratio of 1.5) at an axial-load ratio of
0.3 had a ductility factor of 3.4 (Zhang et al. 2009). The ulti-
mate drift ratio of these two normal-concrete shear walls was
1.8% and 0.6%, respectively. Nevertheless, at an axial-load Fig. 9—Strength envelopes of specimens. (Note: 1 kN =
ratio of 0.3, the tested UHPC shear walls (shear-span ratio 0.225 kip.)
of 1.53) had a ductility factor of 4.2 and an ultimate drift
ratio of 2.1%. The UHPC shear walls had better deformation 131%, respectively, and the maximum load of the shear wall
capacity than the normal-concrete shear walls. increased by 90%, 103%, and 106%, respectively. However,
The cracking load and maximum load of the shear walls the deformation capacity of the shear wall decreased as
increased when the axial-load ratio increased. When the the axial-load ratio increased. When the axial-load ratio
axial-load ratio increased from 0.1 to 0.3, 0.35, and 0.4, the increased from 0.1 to 0.3, 0.35, and 0.4, the ultimate
cracking load of the shear wall increased by 79%, 122%, and displacement of the shear wall decreased by 30%, 67%, and

Table 4—Characteristic points of specimens
Cracking Yielding Maximum load Ultimate displacement
Loading
Specimens direction Fcr, kN Δcr, mm Fy, kN Δy, mm Fm, kN Δm, mm Fu, kN Δu, mm Ductility factor
Positive 457 2.5 617 11.5 794 35.0 671 78.3 6.8
SW-0.1
Negative 472 2.6 554 10.8 748 37.4 644 67.2 6.2
Positive 819 4.5 1362 13.1 1509 37.0 1270 54.7 4.2
SW-0.3
Negative 827 4.8 1230 12.0 1424 35.7 1378 46.3 3.9
Positive 1013 5.7 1444 13.2 1608 20.9 1591 26.1 2.0
SW-0.35
Negative 983 5.6 1319 13.0 1551 20.2 1551 20.2 1.6
Positive 1057 5.8 1614 14.6 1635 15.0 1629 17.0 1.2
SW-0.4
Negative 1021 5.9 1481 14.3 1547 16.5 1547 16.5 1.2
Note: 1 kN = 0.225 kip; and 1 mm = 0.0394 in.
Fig. 10—Reinforcement strain.

78%, respectively, and the ductility factor of the shear wall the maximum tensile strain in the vertical steel bars in
decreased by 38%, 71%, and 82%, respectively. the boundary element decreased by 58%, 61%, and 70%,
respectively. At an axial-load ratio of 0.4, the maximum
Reinforcement strain tensile strain was 3563 με, higher than the yield strain of the
Figure 10(a) shows the maximum strains in the outer- steel bars (3038 με). The maximum compressive strains in
most vertical steel bars in the boundary element. It can be the vertical steel bars of all the tested shear walls varied from
found that the maximum tensile strains in the vertical steel 4500 to 5600 με.
bars decreased as the axial-load ratio increased. When the Figure 10(b) shows the maximum strains in the outermost
axial-load ratio increased from 0.1 to 0.3, 0.35, and 0.4, vertical steel bars in the web region. The maximum tensile

Table 5—Evaluation of shear-wall specimens
Cumulative Self- Strength- Stiffness-
energy dissipa- centering degradation degradation
Specimens tion, kN·m coefficient coefficient coefficient
SW-0.1 420 0.56 0.82 0.05
SW-0.3 193 0.72 0.95 0.12
SW-0.35 113 0.71 0.97 0.30
SW-0.4 73 0.71 0.95 0.47
Note: 1 kN·m = 0.738 ft·kip.
strains in the vertical steel bars decreased as the axial-

load ratio increased. When the axial-load ratio increased
from 0.1 to 0.3 and 0.35, the maximum tensile strain in the
vertical steel bars in the web region decreased by 73% and
94%, respectively. At an axial-load ratio of 0.4, the vertical
steel bars in the web region were in a compression state Fig. 11—Cumulative energy dissipation of specimens. (Note:
throughout the test. Except for SW-0.1, the vertical steel bars 1 kN·m = 0.738 ft·kip.)
in the web region of all the other shear walls were still in the
elastic region.
Figure 10(c) shows the strains in the horizontal steel bars
at the bottom of the shear walls. The strain in the horizontal
steel bars consisted of two parts. First, once the axial load was
applied, the concrete would expand in the horizontal direc-
tion, which caused the initial strain in the horizontal steel
bars. Second, when the lateral displacement was applied,
the shear distortion would increase the strain in the hori-
zontal steel bars. The maximum strain in the horizontal steel
bars increased as the axial-load ratio increased. When the
axial-load ratio increased from 0.1 to 0.3, 0.35, and 0.4, the
maximum strain in the horizontal steel bars at the ultimate
drift ratio increased by 52%, 110%, and 122%, respectively.
Figure 10(d) shows the strains in the stirrups of the
boundary element. The strain in the stirrups also consisted
of two parts, which were caused by concrete expansion and
Fig. 12—Strength degradation of specimens.
lateral load, respectively. When the lateral load was applied,
the boundary element on one side of the shear wall with- crack development, and concrete crushing. As the axial-load
stood more vertical load than the boundary element on the ratio increased, the crack development of the shear wall
other side of the shear wall. Thus, the concrete expansion in slowed down and the tensile strain in the vertical steel bars
the boundary element on one side of the shear wall was more decreased significantly, leading to a lower energy dissipa-
serious than that on the other side of the shear wall, leading tion. When the axial-load ratio increased from 0.1 to 0.3,
to an asymmetry of the strain increase in the stirrups. The 0.35, and 0.4, the cumulative energy dissipation decreased
maximum strain in the stirrups increased as the axial-load by 54%, 73%, and 83%, respectively.
ratio increased, because of more serious concrete expansion
and higher lateral load at a higher axial-load ratio. When Strength degradation
the axial-load ratio increased from 0.1 to 0.3, 0.35, and 0.4, During the test, each lateral displacement was applied three
the maximum strain in the stirrups at the ultimate drift ratio times, and the lateral load at the same drift ratio decreased
increased by 118%, 178%, and 195%, respectively. as the loading cycles increased. To evaluate such behavior, a
strength-degradation coefficient λ was determined as Eq. (6)
Cumulative energy dissipation (JGJ/T 101-2015 2015)
Cumulative energy dissipation is a common index to
describe the ability of a structure to dissipate imposed Fj3
seismic energy. The cumulative energy dissipation was j (6)
Fj1
calculated by summing up the dissipated energy values in
consecutive load-displacement loops throughout the test. in which λj is the strength-degradation coefficient at the j-th
Figure 11 and Table 5 show the cumulative energy dissipa- loading level; and Fj1 and Fj3 are the maximum load at the first
tion of the tested UHPC shear walls. The energy dissipation and third loading cycle of the j-th loading level, respectively.
of the UHPC shear walls was mostly contributed by the steel Figure 12 and Table 5 show the strength-degradation coef-
bars yielding in tension, steel bars yielding in compression, ficients of the tested UHPC shear walls. The shear walls at a

low axial-load ratio had a lower strength-degradation coef- direction of the j-th loading level, respectively; and +Δj and
ficient compared to those at a high axial-load ratio. This is –Δj are the maximum displacement in the positive direction
mainly contributed by more crack propagation of the shear and negative direction of the j-th loading level, respectively.
wall at a lower axial-load ratio. When the axial-load ratio The stiffness-degradation coefficient is calculated as the
increased from 0.1 to 0.3 and 0.35, the strength degradation ratio of the secant stiffness to the initial stiffness. Figure 13
of the shear walls at ultimate displacement increased by 16% and Table 5 present the stiffness-degradation of the spec-
and 18%, respectively. However, when the axial-load ratio imens. The secant stiffness of the shear wall decreased
exceeded 0.35, the concrete damage of the shear wall aggra- rapidly at first and then slowed down as the drift ratio
vated, which increased the strength degradation. The UHPC increased. The secant stiffness of the shear wall at a high
shear wall at an axial-load ratio of 0.4 had a strength-degra- axial-load ratio was higher than that of the shear wall at a
dation coefficient 2% lower than that at an axial-load ratio low axial-load ratio. As the axial-load ratio increased from
of 0.35, but still 16% higher than that at an axial-load ratio 0.1 to 0.3, 0.35, and 0.4, the ultimate stiffness-degradation
of 0.1. coefficient of the shear wall increased by 140%, 500%, and
840%, respectively.
Stiffness degradation
Specimen stiffness decreased as the lateral displacement Displacements along height of shear walls
increased. The secant stiffness of the shear wall was deter- The lateral displacements of the shear walls at the heights
mined with Eq. (7) (JGJ/T 101-2015 2015) of 650, 1300, 1950, and 2600 mm (25.6, 51.2, 76.8, and
102.4 in.) were measured. Figure 14(a) presents the lateral
Fj Fj drift ratio distribution along the height of the shear wall
Kj (7) when the initial crack was observed. When the axial-load
j j ratio increased from 0.1 to 0.3, 0.35, and 0.4, the cracking
in which Kj was the secant stiffness at the j-th loading level, displacement at the top of the shear wall increased by 80%,
+Fj and –Fj are the maximum load in the positive and negative 128%, and 132%, respectively. The lateral displacement at
the midheight of the shear walls was basically half of that at
the top of the shear walls. According to the loading system
of the experiment, the sectional rotation was restricted at
the top and bottom end of the shear walls. The inflection
point can be observed at the midheight of the shear walls.
Figure 14(b) presents the lateral drift ratio distribution along
the height of the shear walls at the ultimate state. When the
axial-load ratio increased from 0.1 to 0.3, 0.35, and 0.4, the
ultimate displacement at the top of the shear wall decreased
by 30%, 67%, and 78%, respectively. After yielding, the
increase in the lateral displacement was mainly contributed
by the sectional rotation in the plastic hinge area. The defor-
mation curves were then almost linear along the height of the
shear walls at the ultimate state.
Fig. 13—Stiffness degradation of specimens.
Fig. 14—Lateral drift ratio along height of shear walls. (Note: 1 mm = 0.0394 in.)

Moment-curvature relationship remain unchanged after yielding, due to the plastic behavior
The sectional curvature at the bottom of the shear walls of steel bars. At an axial-load ratio of 0.3, the moment at the
was detected with dial indicators No. 8 and 9, as shown in bottom of the shear walls increased slower after yielding and
Fig. 5. Figure 15 presents the moment-curvature curve of the decreased by 9% at the ultimate drift ratio, caused by the
shear walls. Note that the contribution of axial load on the concrete crushing at the corners of the shear-wall specimens.
moment at the bottom of the shear walls should be consid- At axial-load ratios of 0.35 and 0.4, the shear-wall specimens
ered. In light of the boundary condition of the specimens, failed suddenly, and no descending section was observed on
the moment at the bottom of the shear walls should be deter- the moment-curvature curves. It can be observed that the
mined as ductility of the shear wall decreased as the axial-load ratio
increased. When the axial-load ratio increased from 0.1 to
H 0.3, 0.35, and 0.4, the ultimate curvature at the bottom of the
M F N (8) shear wall decreased by 35%, 62%, and 64%, respectively.
2 2
in which F is the lateral force; H is the height of the shear DESIGN AXIAL-LOAD RATIO LIMIT
wall; N is the vertical force; and δ is the lateral displacement It can be found that the ductility of the UHPC shear
at the top of the shear wall. walls decreased as the axial-load ratio increased. Thus, the
The maximum moment of the shear wall increased as upper limit of the design axial-load ratio can be proposed to
the axial-load ratio increased. When the axial-load ratio ensure the seismic performance of the UHPC shear walls.
increased from 0.1 to 0.3, 0.35, and 0.4, the maximum According to the code ASCE/SEI 41-17 (2017), for the
moment at the bottom of the shear wall increased by 93%, tested shear walls with a life-safety level (LS), the plastic
103%, and 105%, respectively. At an axial-load ratio of hinge rotation should exceed 1.5% when the design axial-
0.1, the moment at the bottom of the shear walls tended to load ratio is less than 0.1 and exceed 0.9% when the design
axial-load ratio is higher than 0.25. For the tested shear walls
with a collapse-prevention level (CP), the plastic hinge rota-
tion should exceed 2.0% when the design axial-load ratio is
less than 0.1 and exceed 1.2% when the design axial-load
ratio is higher than 0.25. Linear interpolation is permitted
between these values. Figure 16(a) presents the plastic hinge
rotation of the tested shear walls and the corresponding code
requirement. It can be found that the plastic hinge rotation
of 1.2 and 0.9% corresponded to the design axial-load ratio
of 0.7 and 0.73. According to the requirement on the plastic
hinge rotation, the design axial-load ratio limit of the UHPC
shear walls is suggested to be taken as 0.7.
Similar requirements on the deformation capacity of the
RC shear walls are also used in the Chinese codes GB 50010-
2010, “Code for Design of Concrete Structures” (MOHURD
2010a) and GB 50011-2010, “Code for Seismic Design of
Buildings” (MOHURD 2010b). It is suggested that the
Fig. 15—Moment-curvature relationship of shear walls. ductility factor of the RC compressive members should be
(Note: 1 kN·m = 0.738 ft·kip; and 1 mm = 0.0394 in.)
Fig. 16—Design axial-load ratio limit.

Fig. 17—Seismic damage index of UHPC shear-wall specimens.
not less than 3.0 (Tong et al. 2020). Figure 16(b) presents of the specimens at different drift ratios can be plotted as
the ductility factor of the tested shear walls and the corre- Fig. 17. It was found that the damage index of the shear wall
sponding code requirement. Note that based on the exper- at a high axial-load ratio increased faster than that of the
imental results, the plastic hinge rotation of 1.2 and 0.9% shear wall at a low axial-load ratio.
corresponded to the ductility factor of 3.4 and 2.8. Two kinds By reference to seismic damage levels of reinforced
of requirements on the deformation capacity of the RC shear normal-concrete shear walls (Ying et al. 2010; Su and Wong
walls yield similar results. Considering both requirements, 2007), the seismic damage levels of the UHPC shear walls
the design axial-load ratio limit of the UHPC shear walls can can be classified based on the experimental observation
be taken as 0.7. as follows:
1. Fully Operational (0 ≤ D < 0.1): The wall behavior was
SEISMIC DAMAGE MODEL nearly elastic. No cracks or only tiny cracks appeared on the
To evaluate the seismic damage level of the concrete shear walls. The component could continue to be functional.
shear walls at each loading stage, the formula for the seismic 2. Immediate Occupancy (0.1 ≤ D < 0.25): The wall
damage index with double variables was proposed by Park et behavior became nonlinear due to yielding of the steel bars
al. (1985). The seismic damage index (D) was calculated as or nonlinear behavior of the concrete in compression. More
the linear combination of deformation damage and energy- cracks appeared on the shear walls at a low axial-load ratio.
dissipation damage, as presented in Eq. (9) (Ying et al. 2010) The component could still meet most of the requirements.
3. Life safety (0.25 ≤ D < 0.55): Cracks were fully devel-
dE oped in this stage. Several cracks on the two sides of the
D (9)
u Qy u y shear walls would be connected at a low axial-load ratio.
Slight cover-concrete crushing was observed. The compo-
in which δy and δu are the yielding and ultimate displacement nent was easy to repair.
obtained from experiments, respectively; Qy is the yielding 4. Collapse Prevention (0.55 ≤ D ≤ 0.85): Severe concrete
load obtained from experiments; E is the corresponding damage was observed in this stage, and the maximum width
energy dissipation, calculated as the area inside the hyster- of the cracks exceeded the code requirements. The compo-
esis curve obtained from the experiments; and α and β are nent was difficult to repair.
the coefficients. 5. Collapse (D > 0.85): The lateral bearing capacity of
The seismic damage index increased when the damage of the shear walls declined to less than 85% of the maximum
the shear walls increased. When δ in Eq. (9) reached the ulti- lateral load, or the shear walls could not continue to with-
mate drift ratio, the seismic damage index should be close to stand the axial load. The main function of the shear walls
1. Substituting δ = δu into Eq. (9), the relationship between was deprived. The component was in danger of collapsing.
two coefficients can be written as
SUMMARY AND CONCLUSIONS
1 Eu The experimental study was conducted to investigate
1= (10)
Qy u y the seismic behavior of ultra-high-performance concrete
(UHPC) shear walls. The following conclusions can
in which Eu is the energy dissipation at the ultimate be drawn:
displacement. 1. UHPC shear walls exhibited better seismic performance
Substituting the experimental data into Eq. (10), it is compared to normal-concrete shear walls with similar
obtained that α and β are taken as 1.25 and 0.2, respectively, conditions. At a test axial-load ratio of 0.1 and 0.3, the tested
for the UHPC shear walls. Thus, the seismic damage index

UHPC shear wall had a displacement ductility factor of 6.8 GB 50010-2010, 2010a, “Code for Design of Concrete Structures,”
Ministry of Housing and Urban-Rural Development of the People’s
and 4.2, respectively, and an ultimate drift ratio of 3.0% and Republic of China (MOHURD), Beijing, China, 390 pp.
2.1%, respectively. GB 50011-2010, 2010b, “Code for Seismic Design of Buildings,”
2. As the axial-load ratio increased, the energy-dissipation Ministry of Housing and Urban-Rural Development of the People’s
Republic of China (MOHURD), Beijing, China, 66 pp.
capacity and deformation capacity of the UHPC shear walls Hu, R.; Fang, Z.; Shi, C.; Benmokrane, B.; and Su, J., 2021, “A Review
decreased. When the test axial-load ratio increased from 0.1 on Seismic Behavior of Ultra-High Performance Concrete Members,”
to 0.4, the cumulative energy coefficient and displacement Advances in Structural Engineering, V. 24, No. 5, Apr., pp. 1054-1069. doi:
10.1177/1369433220968451
ductility factor of the UHPC shear wall decreased by 83% Hung, C.-C., and Chueh, C.-Y., 2016, “Cyclic Behavior of UHPFRC Flex-
and 82%, respectively. The strength- and stiffness-degrada- ural Members Reinforced with High-Strength Steel Rebar,” Engineering
tion coefficient of the shear wall at a high axial-load ratio was Structures, V. 122, Sept., pp. 108-120. doi: 10.1016/j.engstruct.2016.05.008
Hung, C.-C., and El-Tawil, S., 2011, “Seismic Behavior of a Coupled
higher than that of the shear walls at a low axial-load ratio. Wall System with HPFRC Materials in Critical Regions,” Journal of
3. The design axial-load ratio of the UHPC shear walls Structural Engineering, ASCE, V. 137, No. 12, Dec., pp. 1499-1507. doi:
should be less than 0.7. It is normally required that the 10.1061/(ASCE)ST.1943-541X.0000393
Hung, C.-C., and Hsieh, P.-L., 2020, “Comparative Study on Shear
displacement ductility factor of the concrete shear wall Failure Behavior of Squat High-Strength Steel Reinforced Concrete Shear
should be not less than 3.0. When the test axial-load ratio Walls with Various High-Strength Concrete Materials,” Structures, V. 23,
reached 0.3 and 0.35, the ductility factor of the shear wall Feb., pp. 56-68. doi: 10.1016/j.istruc.2019.11.002
Hung, C.-C.; Li, H.; and Chen, H.-C., 2017, “High-Strength Steel Rein-
was 4.7 and 2.1, respectively. forced Squat UHPFRC Shear Walls: Cyclic Behavior and Design Impli-
4. A seismic damage model was established to evaluate cations,” Engineering Structures, V. 141, June, pp. 59-74. doi: 10.1016/j.
the damage level of the UHPC shear walls. According to engstruct.2017.02.068
JGJ/T 101-2015, 2015, “Specification for Seismic Test of Buildings,”
the existing formula for the seismic damage index, variables China Academy of Building Research (CABR), Beijing, China, 15 pp.
in the formula were determined on the basis of the exper- Kuang, J. S., and Ho, Y. B., 2018, “Seismic Behavior and Ductility of
imental results. Based on the experimental observation, a Squat Reinforced Concrete Shear Walls with Nonseismic Detailing,” ACI
Structural Journal, V. 105, No. 2, Mar., pp. 225-231. doi: 10.14359/19738
classification standard of the seismic damage level of UHPC Li, Q. S., 2001, “Stability of Tall Buildings with Shear-Wall Structures,”
shear walls was proposed. Engineering Structures, V. 23, No. 9, Sept., pp. 1177-1185. doi: 10.1016/
S0141-0296(00)00122-X
Li, X.; Yang, F.; and Zheng, S., 2013, “Categorization Research of Short-
AUTHOR BIOS Pier Shear Wall in the Structure,” Sichuan Building Science, V. 39, No. 4, pp.
Rui Hu is a PhD Candidate in the College of Civil Engineering at Hunan
43-46. (in Chinese) doi: 10.3969/j.issn.1008-1933.2013.04.00910.3969/j.
University, Changsha, Hunan, China. He received his BS and MS from the
issn.1008-1933.2013.04.009
School of Civil Engineering at Wuhan University, Wuhan, Hubei, China,
Marchand, P.; Baby, F.; Khadour, A.; Rivillon, P.; Renaud, J.-C.; Baron,
in 2015 and 2018, respectively. His research interests include ultra-high-
L.; Généreux, G.; Deveaud, J.-P.; Simon, A.; and Toutlemonde, F., 2019,
performance concrete (UHPC) structures and concrete structures with
“Response of UHPFRC Columns Submitted to Combined Axial and Alter-
hybrid fiber-reinforced polymer (FRP)/steel reinforcement bars.
nate Flexural Loads,” Journal of Structural Engineering, ASCE, V. 145,
No. 1, Jan., p. 04018225. doi: 10.1061/(ASCE)ST.1943-541X.0002209
Zhi Fang is a Professor in the College of Civil Engineering at Hunan
Mohamed, N.; Farghaly, A. S.; Benmokrane, B.; and Neale, K. W.,
University, where he received his BS, MS, and PhD in 1984, 1987, and
2014, “Drift Capacity Design of Shear Walls Reinforced with Glass Fiber-
1990, respectively. His research interests include UHPC structures and
Reinforced Polymer Bars,” ACI Structural Journal, V. 111, No. 6,
FRP composites.
Nov.-Dec., pp. 1397-1406. doi: 10.14359/51687099
Mohamed, N.; Farghaly, A. S.; Benmokrane, B.; and Neale, K. W.,
Baodan Xu is a Master’s Student in the College of Civil Engineering at
2014, “Flexure and Shear Deformation of GFRP-Reinforced Shear
Hunan University. He received his BS from the School of Civil Engineering
Walls,” Journal of Composites for Construction, ASCE, V. 18, No. 2, Apr.,
and Communication at North China University of Water Resources and
pp. 785-793. doi: 10.1061/(ASCE)CC.1943-5614.0000444
Electric Power, Zhengzhou, Henan, China. His research interests include
NF P 18-710, 2016, “National Addition to Eurocode 2—Design of
the seismic behavior of UHPC structures.
Concrete Structures: Specific Rules for Ultra-High Performance Fibre-
Reinforced Concrete,” Association Française de Normalisation (AFNOR),
ACKNOWLEDGMENTS Paris, France, 136 pp.
The authors gratefully acknowledge the financial support of the National Park, Y.-J.; Ang, A. H.-S.; and Wen, Y. K., 1985, “Seismic Damage
Natural Science Foundation of China (Grant No. 51878262), the Ministry Analysis of Reinforced Concrete Buildings,” Journal of Structural
of Science and Technology (Project No. 2017YFC0703008), and the Post- Engineering, ASCE, V. 111, No. 4, Apr., pp. 740-757. doi: 10.1061/
graduate Scientific Research Innovation Project of Hunan Province (Grant (ASCE)0733-9445(1985)111:4(740)
No. CX20200401). Peng, F.; Cheng, W.; and Fu, Y., 2006, “The Study of the Opening Size
of Short Pier Shear Walls and the Definition of Short Pier Shear Wall Struc-
ture,” Industrial Construction, V. 36, No. 3, pp. 41-43. (in Chinese) doi:
REFERENCES 10.13204/j.gyjz2006.03.01110.13204/j.gyjz2006.03.011
ASCE/SEI 41-17, 2017, “Seismic Evaluation and Retrofit of Existing
Shi, C.; Wu, Z.; Xiao, J.; Wang, D.; Huang, Z.; and Fang, Z., 2015, “A
Buildings,” American Society of Civil Engineers, Reston, VA, 550 pp.
Review on Ultra High Performance Concrete: Part I. Raw Materials and
Cao, W. L.; Zhang, J. W.; Xue, S. D.; and Chang, W. H., 2006, “Seismic
Mixture Design,” Construction and Building Materials, V. 101, Part 1,
Performance of RC Coupled Short-Pier Shear Walls with Concealed
Dec., pp. 741-751. doi: 10.1016/j.conbuildmat.2015.10.088
Bracing,” Advances in Structural Engineering, V. 9, No. 4, Aug.,
Su, R. K. L., and Wong, S. M., 2007, “Seismic Behaviour of Slender
pp. 577-589. doi: 10.1260/136943306778812732
Reinforced Concrete Shear Walls under High Axial Load Ratio,” Engi-
Dacanay, T. C., 2016, “Ultra-High Performance Concrete Shear Walls
neering Structures, V. 29, No. 8, Aug., pp. 1957-1965. doi: 10.1016/j.
in Tall Buildings,” master’s thesis, Department of Civil and Environmental
Engineering, Virginia Tech, Blacksburg, VA, 111 pp.
T/CBMF 37-2018, 2018, “Fundamental Characteristics and Test
Dang, Z.; Liang, X.; and Deng, M., 2016, “Cyclic Behavior of Shear
Methods of Ultra-High Performance Concrete,” China Building Material
Walls with HPFRCCs in the Inelastic Deformation Critical Region,” The
Federation (CBMF), Beijing, China, 9 pp.
Structural Design of Tall and Special Buildings, V. 25, No. 17, pp. 886-903.
Tong, X.; Fang, Z.; Luo, X.; and Gong, L., 2020, “Study on Shear
doi: 10.1002/tal.1288
Capacity of Ultra-High Performance Concrete Squat Shear Walls,” Case
Fédération internationale du béton, 2013, “fib Model Code for Concrete
Studies in Construction Materials, V. 12, June, p. e00314. doi: 10.1016/j.
Structures 2010,” fib, Lausanne, Switzerland, 434 pp.
cscm.2019.e00314
Wang, Z.; Wang, J.-Q.; Tang, Y.-C.; Liu, T.-X.; Gao, Y.-F.; and Zhang, J.,
2018, “Seismic Behavior of Precast Segmental UHPC Bridge Columns with

Replaceable External Cover Plates and Internal Dissipaters,” Engineering Engineering Journal, V. 42, No. 4, pp. 10-16. (in Chinese) doi: 10.15951/j.
Structures, V. 177, Dec., pp. 540-555. doi: 10.1016/j.engstruct.2018.10.012 tmgcxb.2009.04.00710.15951/j.tmgcxb.2009.04.007
Yang, C., and Okumus, P., 2017, “Ultrahigh-Performance Concrete for Zhang, H.; Lu, X.; Yang, X.; and Zhang, S., 2008, “Influences of Boundary
Posttensioned Precast Bridge Piers for Seismic Resilience,” Journal of Stirrup on Seismic Behavior of Reinforced Concrete Shear Walls,” Struc-
Structural Engineering, ASCE, V. 143, No. 12, Dec., p. 04017161. doi: tural Engineers, V. 24, No. 5, pp. 100-104. (in Chinese) doi: 10.3969/j.
10.1061/(ASCE)ST.1943-541X.0001906 issn.1005-0159.2008.05.01910.3969/j.issn.1005-0159.2008.05.019
Ying, Y.; Jiang, H.; Wang, B.; and Zhang, Y., 2010, “Study on Seismic Zohrevand, P., and Mirmiran, A., 2013, “Effect of Column Parameters
Damage Model with Double Variables for Reinforced Concrete Shear Walls,” on Cyclic Behavior of Ultra-High-Performance Concrete-Filled Fiber-
Structural Engineers V. 26, No. 5, pp. 61-65. (in Chinese) doi: 10.3969/j. Reinforced Polymer Tubes,” ACI Structural Journal, V. 110, No. 5,
issn.1005-0159.2010.05.01110.3969/j.issn.1005-0159.2010.05.011 Sept.-Oct., pp. 823-832. doi: 10.14359/51685835
Zhang, S.; Lu, X.; and Zhang, H., 2009, “Experimental and Analytical
Studies on the Ultimate Displacement of RC Shear Walls,” China Civil

Title No. 119-S43
Unified Strength Model for Spiral-Steel Confined Concrete

Columns
by Muhammad Junaid Munir, Syed Minhaj Saleem Kazmi, and Yu-Fei Wu
Existing strength models of spiral-steel confined natural aggregate ITZ between the adhered mortar and the aggregate.15 As a
concrete (NAC) are based on the Richart et al. model. Furthermore, result, when recycled aggregates are used to produce RAC,
scant work related to the strength model of spiral-steel confined the ITZ increases and results in inferior/lower performance
recycled aggregate concrete (RAC) is available in the literature. of RAC than NAC.16 Approximately 30 to 40% reduction
This paper focuses on developing a unified strength model for
in compressive strength of RAC is reported in the literature
spiral-steel confined circular NAC and RAC columns based on
compared to NAC.7
the Hoek-Brown peak strength criterion. The Hoek-Brown peak
strength criterion applies to the strengthening of concrete members Different methods are adopted by researchers to enhance
with existing damage and can be extended to the performance the performance of RAC. In this context, the porous mortar
enhancement of RAC through spiral-steel confinement. The perfor- attached to recycled aggregates was removed through
mance of existing and proposed models is evaluated through a mechanical,17 chemical,18 and thermal19 treatment tech-
test database comprising the wide-ranging experimental results niques. Similarly, strengthening of porous adhered mortar
of spiral-steel confined circular concrete columns. Results show of recycled aggregates through self-healing,20 surface
that the proposed model can effectively predict the peak strength coating,21 and carbonation22 techniques was also explored
of spiral-steel confined circular NAC and RAC columns, leading to by the researchers. Several mixing approaches (including
structural applications of RAC. sand-enveloped mixing, single mixing, double mixing,
Keywords: concrete columns; confinement; Hoek-Brown peak strength
two-stage mixing, and mortar mixing) and mineral admix-
criterion; recycled aggregate concrete; steel spirals; strength. tures were also investigated to enhance the RAC perfor-
mance.23 Although these methods improve the performance
INTRODUCTION of recycled aggregates and RAC, the handling of the high
Being a highly consumed material, 25 billion tons of amount of materials, required infrastructure, associated cost,
fresh concrete are produced annually worldwide.1,2 On the carbon dioxide emissions, and quality control are the short-
other side, many countries are dealing with the landfilling comings to adopting the treatment techniques on an indus-
issues of construction and demolition waste. Annually, 77, trial scale.24
325, and 510 million tons of construction and demolition Generally, strength enhancement due to confinement by
waste are generated in Japan, the United States, and Europe, lateral reinforcement is not considered in the current design
respectively.1,3 To consume such massive waste material, practice of concrete structures.7,25 Recently, Munir et al.25
many countries recycle the construction and demolition explored the use of strength enhancement through confine-
waste as aggregates to produce recycled aggregate concrete ment by spiral-steel reinforcement to overcome the strength
(RAC).4,5 In 2019, the municipal administration level of reduction of RAC. Results showed that confinement by
Shanghai, China, made it mandatory to use 15 to 30% of lateral reinforcement can significantly improve the perfor-
recycled aggregates to produce concrete for new construc- mance of RAC, and the increase in the confinement level
tion projects.6,7 RAC has also been used in the construction increases the peak stress of RAC. The strength reduction
of the Waldspirale complex in Germany; a housing project related to the use of recycled aggregates is also lower for
in Madrid, Spain; building Council House 2 in Melbourne, confined RAC samples than unconfined RAC samples.7 The
Australia; and Zürich’s largest school in Birch, Switzerland.3 lateral steel reinforcement is commonly used in the construc-
Although the demand for recycling construction and tion of reinforced concrete structures. Therefore, the idea of
demolition waste is increasing day by day, RAC applications using the confinement by spiral-steel reinforcement to over-
are still quite limited.8,9 The biggest hurdle in using RAC is come the strength reduction of RAC is not only effective but
its inferior performance compared to traditional concrete.10 is also practically possible for recycling the construction and
The poorer properties of RAC compared with natural aggre- demolition waste in the construction of concrete compres-
gate concrete (NAC) are owed to the inferior properties of sion members.
recycled aggregates compared with natural aggregates11,12;
the porous mortar adhered to the recycled aggregates results
in the poorer characteristics of recycled aggregates.13 The MS No. S-2021-132.R1, doi: 10.14359/51734340, received August 16, 2021, and
interfacial transition zone (ITZ) is the link between the reviewed under Institute publication policies. Copyright © 2022, American Concrete
aggregate and cement paste and is considered the weakest obtained from the copyright proprietors. Pertinent discussion including author’s
point of concrete.14 The recycled aggregates already have an is received within four months of the paper’s print publication.

RESEARCH SIGNIFICANCE the longitudinal reinforcement; and fs is the strength of steel
Various strength models of spiral-steel confined NAC are spirals.
available in the literature. However, limited studies are avail-
able related to the strength model of spiral-steel confined HOEK-BROWN PEAK STRENGTH CRITERION
RAC. This paper focuses on developing a unified strength Considering wide-ranging test data on intact rocks, jointed
model for spiral-steel confined circular NAC and RAC rock masses, and rock discontinuities, Hoek and Brown26
columns based on the Hoek-Brown26 peak strength criterion. developed a nonlinear empirical peak strength criterion for
The proposed model can accurately predict the peak strength rocks and rock masses
of spiral-steel confined circular NAC and RAC columns,
leading to structural applications of RAC. 1 3
m 3 s (4)
c c c
EXISTING MODELS
Based on detailed experimental tests on spiral-steel where σc is the uniaxial peak strength of the intact rock
confined NAC samples, Richart et al.27 proposed the material; σ1 and σ3 are the major and minor principal stresses
following relationship at failure, respectively; m is the material parameter related
to the rock type; and s is the rock discontinuities param-
fcc = fco + kfl (1) eter ranging from 0 to 1. For various types of rocks having
different discontinuity levels, the values of parameters m and
where fl is the lateral confinement pressure; fco and fcc are s are provided by Hoek and Brown.26
the compressive strengths of unconfined and confined NAC, As concrete can be considered a rock type, the Hoek-
respectively; and k is the confinement coefficient, which was Brown26 peak strength criterion can be applied to the
reported equal to 4.1 by Richart et al.27 based on the exper- concrete. The uniaxial compressive strength of concrete can
imental results. Equation (2) shows the normalized form of be determined by putting σ3 = 0 in Eq. (4).
the aforementioned relationship (Eq. (1)), which has been
adopted by numerous researchers to develop the strength σ1 = σc√s (5)
models for spiral-steel confined circular NAC columns. In
this context, the value of the coefficient k has been derived For NAC with no considerable defects, s = 1.0; therefore,
as a function of fl/fco, a function of the lateral confinement σ1 = σc as anticipated. Similarly, s = 0 for completely crushed
pressure (fl), or as a constant by fitting the experimental data. concrete, resulting in σ1 = 0.
Generally, the value of the coefficient k reduces with the rise The addition of coarse recycled aggregates (CRAs) in
in confinement pressure.27 concrete results in its inferior performance.14 RAC can be
considered concrete with existing damage or cracks. For
f cc f RAC columns, the parameter s can be calculated by rear-
1 k l (2) ranging Eq. (5)
f co f co
Through spiral-steel confinement, the concrete of the 2

columns undergoes triaxial compressive stresses.25 The s 1 (6)
compressive strength of spiral-steel confined circular NAC c
columns increases with the increase in confinement pres- where σ1 is the peak strength of RAC; and σc is the peak
sure.7 Using Eq. (3), the confinement pressure of spiral-steel strength of the NAC.
confined circular NAC columns can be calculated The value of the parameter m can be determined for any
concrete type by putting σ1 = 0 in Eq. (4). In this scenario, σ3
2 f y As is the tensile strength of concrete (σt), which gives
fl = (3)
Dp
where p and fy are the center-to-center spacing and yield
strength of steel spirals, respectively; As is the cross-

t 1

m m 2 4 s (7)
c 2

sectional area of steel spirals; and D is the diameter of spiral- For NAC with no considerable defects, s = 1.0; therefore,
steel confined circular NAC samples. the value of the parameter m can be determined for known
Table 1 presents the existing peak strength models compressive and tensile strengths of concrete using Eq. (7).
for spiral-steel confined circular NAC columns. All the The Hoek-Brown26 peak strength criterion applies to
presented models are empirical, which were regressed from strengthening or repairing concrete members with existing
experimental results. Moreover, most of these models were damage or defects. Therefore, this criterion can be extended
developed based on the results of the researchers. There- to the performance enhancement of RAC through spiral-
fore, those models can accurately predict their own results. steel confinement, which is the advantage of using the
In Table 1, s′and dsp are the clear spacing and diameter of Hoek-Brown peak strength criterion compared to previously
steel spirals, respectively; Al is the cross-sectional area of existing models for spiral-steel confined NAC.

Table 1—Peak strength models for spiral-steel confined circular NAC columns
Model Strength Parameters
f cc f 2 f y As
Richart et al.27 = 1 + 4.1 l fl =
f co f co Dp
f cc  21  f l 2 f y As
Shah et al.28 = 1 + 1.15 + fl =
f co  f co  f co Dp
s′
1−
f le = 0.5keρs f y ; ke = 2D
f cc f f
Mander et al.29 = 2.254 1 + 7.94 le − 2 le − 1.254 1 − ρcc
f co f co f co 4A 4 Al
ρs = s ; s ′ = p − d sp ; ρcc =
Dp πD 2
ρs f y  p 
f l ′= 1−
El-Dash and f cc f′ 2  1.25 D 
= 1 + k1 l
Ahmad30 f co f co
0.5
 f   d sp 
0.25
4 As
k1 = 5.1 co   ρ  ; ρs =
 fy  s Dp
f cc ρs f y 4 As
Hoshikuma et al.31 = 1 + 3.83 ρs =
f co f co Dp
Razvi and f cc f 2 f s As
= 1 + k1e l fl = ; k = 6.7 f l −0.17
Saatcioglu32 f co f co Dp 1e
 
f cc f rp 16 f y As 2 f y As 
Assa et al.33 = 1 + 3.36 f rp = max.  ,
f co f co  6p
Dp 
 Dpf co 0.34 e D 
s′
1−
f le = 0.5keρs f y ; ke = 2D
1 − ρcc
4 Al 4A
f cc f f ρcc = ; ρs = s ; s ′ = p − d sp
Bing et al.34 = 2.254 1 + 7.94α s le − 2α s le − 1.254 πD 2 Dp
f co f co f co f le
α s = 3.1 for f co > 52 MPa
f co
f
α s = ( 21.2 − 0.35 f co ) le for f co ≤ 52 MPa
f co
Wei and Wu35; f cc f 2 f s As

= 1 + k1′ l fl = ; k ′ = 5.35 fl −0.14
Munir et al.36 f co f co Dp 1
UNIFIED STRENGTH MODEL BASED ON HOEK- where ft is the tensile strength of NAC. It is well established
BROWN PEAK STRENGTH CRITERION in the literature that ft is a function of fco. Table 2 shows the
Model for spiral-steel confined circular NAC relationships between the tensile and compressive strengths
columns of NAC presented by different building standards and
Previously, Wu and Zhou37 developed a unified strength researchers. Based on these relationships, Eq. (9) can be
model for fiber-reinforced polymer (FRP) confined square written as
and circular NAC columns based on the Hoek-Brown26 peak
strength criterion. A similar methodology is adopted in this 2 n
m m2 4 f co (10)
study to develop the unified strength model for spiral-steel C
confined circular NAC columns. For NAC with no consid-
where n and C are the coefficients to be determined. The
erable defects, s = 1.0, and to evaluate the peak strength of
parameter m value can be determined by simplifying Eq. (10)
spiral-steel confined circular NAC columns, Eq. (4) can be
written as
f co n C
m n (11)
f cc f f C f co
l m l 1 (8)
f co f co f co After considering a wide range of experimental data on
confined circular NAC columns, Girgin et al.51 also reported
Similarly, Eq. (8) can be written as the dependency of the parameter m of the Hoek-Brown26
peak strength criterion on the compressive strength of NAC.

ft
1
m m 2 4 (9)
f co 2
The findings of Girgin et al.51 are in agreement with Eq. (11).
Considering Eq. (11), Eq. (8) can be written as

to assess models by Wei and Wu35 and Munir et al.25 To
f cc f f n C f
l co n l 1 (12) ensure the relevancy of the database with the current work,
f co f co C f co f co it is updated by adding new experimental data and reevalu-
ated carefully in this study. It is worth mentioning that the
The coefficients (n and C) of Eq. (12) can be determined by
test database includes only the studies focusing on the pure
minimizing the deviation between the experimental results
confinement effect—that is, either the tested specimens have
and model predictions through regression analysis. Based on
no longitudinal reinforcement or longitudinal reinforcement
regression analysis, the coefficients n and C are determined
has a negligible impact on the test results. Currently, limited
to be –0.725 and 0.011, respectively. It should be noted that
experimental data is available in the literature regarding
the regression coefficients n and C are based on the existing
spiral-steel confined circular NAC columns with no/negli-
test database (Table 3) and can be improved further with the
gible influence of longitudinal reinforcement. Therefore, the
availability of new data points. The details of the test data-
database includes the test results of 137 spiral-steel confined
base and the assessment criteria used for regression analysis
circular NAC columns. All the specimens were cylindrical
are discussed later. Therefore, the peak strength model for
with a diameter ranging between 145 to 500 mm and were
spiral-steel confined circular NAC columns is given by
examined under a uniaxial compression test. The database
covers wide-ranging values of various parameters related
f cc f f 0.725 0.011 f l to spiral-steel confined circular NAC columns, such as fl
l co 0.725 1 (13)
f co f co 0.011 f co f co varying from 1.0 to 28.3 MPa, fco varying from 15 to 85 MPa,
fcc/fco varying from 1.0 to 4.0, and fl/fco varying from 0.03 to
ASSESSMENT OF MODEL 0.83. Besides regression analysis, the database is also used
Database to evaluate the performance of existing and proposed models
For regression analysis, a test database is used in this in this study.
study, as shown in Table 3, which comprises the wide-
ranging experimental results of spiral-steel confined circular Assessment criteria
NAC columns. The same database has been used recently The performance of existing and proposed models is
evaluated by the deviation between experimental results
Table 2—Relationships between tensile and and predictions of the models. This deviation is calculated
compressive strengths of NAC through the error-index (EI) presented in Eq. (14)35
Model Relationship
f f f
Carino and Lew38 ft = 0.272(fco)0.71 cc cc / cc
f co exp. f co ana. f co exp.
Raphael39 ft = 0.313(fco)0.667
EI (14)
Shah and Ahmad40 ft = 0.46(fco)0.55 N
Gardner41 ft = 0.33(fco)0.667 where (fcc/fco)ana. and (fcc/fco)exp.are the theoretical and exper-
Oluokun et al.42 ft = 0.294(fco)0.69 imental values of the peak strength of spiral-steel confined
concrete columns, respectively; and N is the total number
Mokhtarzadeh and French43 ft = 0.32(fco)0.63
of concrete columns/specimens. EI is a normalized statis-
Eurocode 2:200444 ft = 0.3(fco)0.667 tical parameter, which is extremely susceptible to the varia-
Arɩoglu et al.45 ft = 0.387(fco)0.63 tion between the experimental results and predictions of the
NZS 3101:200646 and JSCE:200747 ft = 0.44(fco)0.5 models. As previously discussed, the coefficients n and C
are determined in this study through regression analysis by
AS 3600-2009 48
ft = 0.4(fco)0.5
minimizing the EI value.
ACI 318-1149 ft = 0.56(fco)0.5 The performance of models is also evaluated by a typical
Chhorn et al.50 ft = 0.47(fco)0.511 statistical analysis of the ratio (fcc/fco)ana./(fcc/fco)exp.. In this
context, the mean ((fcc/fco)ana./(fcc/fco)exp.)Avg., coefficient of
Table 3—Test database of spiral-steel confined circular NAC columns

Concrete
Confining pressure strength fco, Peak stress ratio Confinement ratio
Researchers Number Diameter, mm fl, MPa MPa fcc/fco fl/fco
Richart et al.52 18 254 1.0 to 10.4 14.7 1.20 to 3.97 0.07 to 0.71
Mander et al.53 15 500 1.0 to 4.3 24.0 to 32.0 1.24 to 1.80 0.03 to 0.15
Sheikh and Toklucu54 27 203 to 356 1.7 to 6.8 29.7 to 30.5 1.21 to 1.70 0.06 to 0.22
Assa et al.33
24 145 2.5 to 28.3 25.0 to 85.0 1.00 to 3.82 0.03 to 0.83
Bing et al.34
14 240 1.8 to 20.2 52.0 to 82.5 1.01 to 2.50 0.03 to 0.39
Wei and Wu 35
12 150 3.2 to 12.9 36.4 1.48 to 2.93 0.09 to 0.36
Munir et al.36 27 150 1.9 to 3.8 26.3 to 61.8 1.02 to 2.03 0.03 to 0.14

Table 4—Evaluation of strength models for spiral-steel confined circular NAC columns
Model EI, % ((fcc/fco)ana./(fcc/fco)exp.)Avg. SD COV, % ((fcc/fco)ana./(fcc/fco)exp.)Max. ((fcc/fco)ana./(fcc/fco)exp.)Min.
Richart et al. 27
11.2 1.00 0.13 12.7 1.35 0.78
Shah et al. 28
18.7 0.80 0.17 13.5 1.09 0.55
Mander et al. 29
11.3 1.05 0.12 12.2 1.40 0.78
El-Dash and Ahmad30 12.5 0.88 0.21 18.7 1.16 0.51
Hoshikuma et al.31 34.9 1.30 0.14 18.1 2.15 0.95
Razvi and Saatcioglu 32
12.5 1.08 0.11 11.4 1.45 0.87
Assa et al.33
14.5 0.85 0.11 9.7 1.03 0.69
Bing et al.34 18.9 1.01 0.37 36.8 1.81 0.31
Wei and Wu35 9.8 1.01 0.11 11.3 1.34 0.79
Munir et al. 36
9.8 1.01 0.11 11.3 1.34 0.79
Proposed 10.0 0.97 0.12 11.7 1.29 0.74
variation (COV), variance (SD), maximum ((fcc/fco)ana./

(fcc/fco)exp.)Max., and minimum ((fcc/fco)ana./(fcc/fco)exp.)Min. values
are calculated and compared for each model.
Discussion of results
The existing and proposed peak strength models of spiral-
steel confined circular NAC columns are evaluated through
the aforementioned statistical indexes, and the results
are presented in Table 4. For comparison purposes, EIs
of proposed and existing peak strength models for spiral-
steel confined circular NAC columns are also presented in
Fig. 1. The performance of the models is categorized into
three different groups: Group I (EI ≤ 15%), Group II (15% <
EI ≤ 30%), and Group III (EI > 30%), as presented in Fig. 1.
Results show that most of the strength models have good Fig. 1—EIs of strength models for spiral-steel confined
accuracy and are within Group I. circular NAC columns.
To better understand the error distributions of the proposed model is based on the Hoek-Brown26 peak strength
proposed and existing models, the comparison between the criterion.
experimental results and predictions of models for spiral-
steel confined circular NAC columns is presented in Fig. Model evaluation for application to spiral-steel
2. The Hoshikuma et al.31 model from Group III signifi- confined circular RAC columns
cantly underestimates the peak strength of spiral-steel As previously discussed, the Hoek-Brown26 peak strength
confined circular NAC columns, with 68% of the test data criterion applies to strengthening concrete members with
having EI > 15%. However, the Shah et al.28 model belongs existing damage. Therefore, the developed model based on
to Group II and overestimates the peak strength results of the Hoek-Brown26 peak strength criterion can be extended
spiral-steel confined circular NAC columns, with 61% of the to predict the peak strength of spiral-steel confined circular
test data having EI > 15%. Both the Hoshikuma et al.31 and RAC columns. To evaluate the developed model for appli-
Shah et al.28 models are developed based on their own exper- cation to spiral-steel confined circular RAC columns, a
imental data, which is why these models are not very accu- test database including a total of 153 test columns is used
rate in predicting the peak strength of spiral-steel confined in this study, as shown in Table 5, which comprises the
circular NAC columns. wide-ranging experimental results of spiral-steel confined
Figures 1 and 2 show that the model predictions of Wei circular RAC columns. It covers different types and replace-
and Wu,35 Munir et al.,36 and the proposed model are more ment ratios of CRA, fl varying from 1.8 to 3.8 MPa, and fco
accurate compared to the other models. Moreover, these varying from 26 to 62 MPa. The damage index s varies from
models show the uniform distributions of the error for the 0.33 to 0.94, which is determined through Eq. (6) using the
whole range of results. Based on Fig. 1 and 2, it is consid- peak strength of the NAC and RAC from the original refer-
ered that the proposed model based on the Hoek-Brown26 ences. The lower value of the damage index s represents the
peak strength criterion is suitable to predict the peak strength RAC samples having peak strength significantly reduced by
of spiral-steel confined circular NAC columns. It is worth adding RCA and vice versa.
mentioning that both the Wei and Wu35 and Munir et al.36 Recently, Munir et al.36 developed a model to predict the
models are based on the Richart et al.27 model, whereas the peak stress of spiral-steel confined NAC and RAC columns,

Fig. 2—Performance of strength models for spiral-steel confined circular NAC columns: (a) Richart et al.27; (b) Shah et al.28;
(c) Mander et al.29; (d) El-Dash and Ahmad30; (e) Hoshikuma et al.31; (f) Razvi and Saatcioglu32; (g) Assa et al.33; (h) Bing
et al.34; (i) Wei and Wu35; (j) Munir et al.36; and (k) proposed model.
Table 5—Test database of spiral-steel confined 2 f s As
circular RAC columns fl = (17)
Dp
Confining Concrete Damage
Diameter, pressure fl, strength level
Researchers Number mm MPa fco, MPa s A = –0.69WA2 + 8.69WA – 26.70 (18)
0.59 to
Munir et al.25 18 150 1.8 to 3.5 49.7
0.70
B = 0.43WA2 – 5.30WA + 16.13 (19)
26.3 to 0.33 to
Munir et al.7
81 150 1.9 to 3.8
61.8 0.94
26.3 to 0.67 to
Munir et al.36 27 150 1.9 to 3.8
61.8 0.83
a = 0.83WA2 – 10.32WA + 32.77 (20)
26.3 to 0.66 to
Munir et al.55 27 150 1.9 to 3.8 The performance of the Munir et al.36 model and Eq. (9)
61.8 0.93
in determining the peak strength of spiral-steel confined
presented as Eq. (15) to (20). The Munir et al.36 model is circular RAC columns for the 153 specimens is shown in
based on the Richart et al.27 model and involves the water Fig. 3. The EIs between the test results and the predictions
absorption (WA) and replacement ratio (R) of RCA to consider are 6.5 and 6.8% for Munir et al.36 and the proposed models,
the effect of RCA addition on the peak strength of concrete. respectively, as shown in Table 6. It should be noted that the
Basically, the Munir et al.36 model uses the water absorption Munir et al.36 model involves the material characteristics of
values of RCA to consider the impact of the quality of RCA RCA and was regressed using the test database presented in
on the peak stress of spiral-steel confined RAC columns. Table 5. Hence, the slightly lower EI of the Munir et al.36
model compared to the proposed model is expected.
Figure 4 shows the uniform distribution of error for the
f cc f
1 k1 l (15) complete range of the damage index s. Still, currently, no
f co f co data is available in the literature related to s value < 0.33.
Thus, the proposed model can be further evaluated in the
k1 = 5.35fl–0.14 – (ARa + BR)fcofl–1 (in MPa) (16) future for RAC columns severely affected by the CRA addi-
tion having s value < 0.33. However, s value < 0.33 means
that the compressive strength of RAC is lower than NAC
by more than 43%. Generally, the maximum strength loss

of 40% is reported in the literature for RAC compared to
NAC.56,57 Therefore, s value < 0.33 is unusual in practice.
Based on the previous discussion, the unified strength
model proposed in this study based on the Hoek-Brown26
peak strength criterion can be effectively used to predict
the peak strength of spiral-steel confined circular NAC and
RAC columns.
CONCLUSIONS
Various strength models of spiral-steel confined natural
aggregate concrete (NAC) are available in the literature.
However, limited studies are available related to the strength
model of spiral-steel confined recycled aggregate concrete
(RAC). Furthermore, most of the existing strength models
of spiral-steel confined NAC are based on the Richart et al.27
model. This paper focuses on developing a unified strength
model for spiral-steel confined circular NAC and RAC
columns based on the Hoek-Brown26 peak strength crite-
rion. The Hoek-Brown26 peak strength criterion applies to
strengthening or repairing concrete members with existing
damage or defects. Therefore, this criterion can be extended
to the performance enhancement of RAC through spiral-
steel confinement. A test database comprising the wide-
ranging experimental results of spiral-steel confined circular
concrete columns is used to evaluate the performance of
existing and proposed models in this study. Results show
that the proposed model can effectively predict the peak
strength of spiral-steel confined circular NAC and RAC
columns, leading to structural applications of RAC.
AUTHOR BIOS
Fig. 3—Performance of strength models for spiral-steel Muhammad Junaid Munir is a Research Assistant in the School of Engi-
neering at RMIT University, Melbourne, VIC, Australia. He received his
confined circular RAC columns: (a) Munir et al.33 model; BSc from Mirpur University of Science and Technology, Mirpur, Azad
and (b) proposed model. Kashmir, Pakistan; his MSc from the University of Engineering and Tech-
nology, Lahore, Punjab, Pakistan; and his PhD from RMIT University. His
research interests include concrete materials and structures.
Syed Minhaj Saleem Kazmi is a Research Assistant in the School of Engi-

neering at RMIT University. He received his BSc from Mirpur University
of Science and Technology; his MSc from the University of Engineering
and Technology; and his PhD from RMIT University. His research interests
include construction materials and sustainable structures.
Yu-Fei Wu is a Professor at the Guangdong Provincial Key Laboratory of

Durability for Marine Civil Engineering at Shenzhen University, Shenzhen,
China, and the School of Engineering at RMIT University. He received his
BSc from Zhejiang University, Hangzhou, China; his MSc from the National
University of Singapore, Singapore; and his PhD from The University of
Adelaide, Adelaide, SA, Australia. His research interests include concrete
structures and structural rehabilitation.
ACKNOWLEDGMENTS
This work was supported by the Australian Research Council
(DP200100631) and the Victoria-Jiangsu Program for Technology and
Innovation R&D by the Department of Economic Development, Jobs,
Transport and Resources, the state of Victoria, Australia. All the data and
models used in this study appear in the submitted paper.
Fig. 4—Distribution of error.
Table 6—Evaluation of strength models for spiral-steel confined circular RAC columns
Model EI, % ((fcc/fco)ana./(fcc/fco)exp.)Avg. SD COV, % ((fcc/fco)ana./(fcc/fco)exp.)Max. ((fcc/fco)ana./(fcc/fco)exp.)Min.
Munir et al. 36
6.5 0.96 0.10 7.6 1.12 0.70
Proposed 6.8 1.04 0.07 7.0 1.21 0.86

NOTATION Recycled Aggregate Concrete,” Cement and Concrete Composites, V. 97,
Al = cross-sectional area of longitudinal reinforcement Mar. 2019, pp. 341-356. doi: 10.1016/j.cemconcomp.2019.01.005
As = cross-sectional area of steel spirals 9. Kazmi, S. M. S.; Munir, M. J.; and Wu, Y.-F., “Application of Waste
C = coefficient related to relationship between ft and fco Tire Rubber and Recycled Aggregates in Concrete Products: A New
D = diameter of spiral-steel confined circular concrete Compression Casting Approach,” Resources, Conservation and Recycling,
samples V. 167, Apr. 2021, p. 105353. doi: 10.1016/j.resconrec.2020.105353
dsp = diameter of steel spirals 10. Thomas, C.; de Brito, J.; Cimentada, A.; and Sainz-Aja, J. A.,
EI = error-index “Macro- and Micro- Properties of Multi-Recycled Aggregate Concrete,”
fcc = compressive strength of confined concrete samples Journal of Cleaner Production, V. 245, Feb. 2020, p. 118843. doi: 10.1016/j.
(fcc/fco)ana. = theoretical peak strength values of spiral-steel confined jclepro.2019.118843
concrete columns 11. Kazmi, S. M. S.; Munir, M. J.; Wu, Y.-F.; Patnaikuni, I.; Zhou, Y.;
((fcc/fco)ana./ and Xing, F., “Effect of Different Aggregate Treatment Techniques on the
((fcc/fco)exp.)Avg. = mean of ratio of theoretical and experimental peak Freeze-Thaw and Sulfate Resistance of Recycled Aggregate Concrete,”
strength values of spiral-steel confined concrete Cold Regions Science and Technology, V. 178, Oct. 2020, p. 103126. doi:
columns 10.1016/j.coldregions.2020.103126
((fcc/fco)ana./ 12. Munir, M. J.; Kazmi, S. M. S.; Wu, Y.-F.; Lin, X.; and Ahmad, M. R.,
((fcc/fco)exp.)Max. = maximum value of ratio of theoretical and experi- “Axial Stress-Strain Performance of Recycled Aggregate Concrete Rein-
mental peak strength values of spiral-steel confined forced with Macro-Polypropylene Fibres,” Sustainability, V. 13, No. 10,
concrete columns May 2021, p. 5741. doi: 10.3390/su13105741
((fcc/fco)ana./ 13. Matias, D.; de Brito, J.; Rosa, A.; and Pedro, D., “Mechanical Proper-
((fcc/fco)exp.)Min. = minimum value of ratio of theoretical and experimental ties of Concrete Produced with Recycled Coarse Aggregates – Influence of
peak strength values of spiral-steel confined concrete the Use of Superplasticizers,” Construction and Building Materials, V. 44,
columns July 2013, pp. 101-109. doi: 10.1016/j.conbuildmat.2013.03.011
(fcc/fco)exp. = experimental peak strength values of spiral-steel 14. Kazmi, S. M. S.; Munir, M. J.; Wu, Y.-F.; and Patnaikuni, I., “Effect
confined concrete columns of Macro-Synthetic Fibers on the Fracture Energy and Mechanical Behavior
fco = compressive strength of unconfined NAC samples of Recycled Aggregate Concrete,” Construction and Building Materials,
fl = lateral confinement pressure V. 189, Nov. 2018, pp. 857-868. doi: 10.1016/j.conbuildmat.2018.08.161
fs = strength of steel spirals 15. Kazmi, S. M. S.; Munir, M. J.; Wu, Y.-F.; Patnaikuni, I.; Zhou, Y.;
ft = tensile strength of NAC samples and Xing, F., “Influence of Different Treatment Methods on the Mechanical
fy = yield strength of steel spirals Behavior of Recycled Aggregate Concrete: A Comparative Study,” Cement
k = confinement coefficient and Concrete Composites, V. 104, Nov. 2019, Article No. 103398. doi:
m = material parameter related to rock type 10.1016/j.cemconcomp.2019.103398
N = total number of concrete columns/specimens 16. Kazmi, S. M. S.; Munir, M. J.; Wu, Y.-F.; Patnaikuni, I.; Zhou, Y.;
n = coefficient related to relationship between ft and fco and Xing, F., “Effect of Recycled Aggregate Treatment Techniques on
p = center-to-center spacing of steel spirals the Durability of Concrete: A Comparative Evaluation,” Construction
R = replacement ratio of RCA and Building Materials, V. 264, Dec. 2020, p. 120284. doi: 10.1016/j.
SD = variance of results conbuildmat.2020.120284
s = rock discontinuities parameter ranging from 0 to 1 17. Montgomery, D. G., “Workability and Compressive Strength Prop-
s′ = clear spacing of steel spirals erties of Concrete Containing Recycled Concrete Aggregate,” Sustain-
WA = water absorption of RCA able Construction: Use of Recycled Concrete Aggregate, R. K. Dhir, N.
σc = peak strength of NAC/uniaxial peak strength of intact A. Henderson, and M. C. Limbachiya, eds., Thomas Telford Ltd., London,
rock material UK, 1998, pp. 287-296.
σt = tensile strength of concrete 18. Kim, Y.; Hanif, A.; Kazmi, S. M. S.; Munir, M. J.; and Park, C., “Prop-
σ1 = peak strength of RAC/major principal stress at failure erties Enhancement of Recycled Aggregate Concrete through Pretreatment
σ3 = minor principal stress at failure of Coarse Aggregates – Comparative Assessment of Assorted Techniques,”
Journal of Cleaner Production, V. 191, Aug. 2018, pp. 339-349. doi:
10.1016/j.jclepro.2018.04.192
REFERENCES 19. Al-Bayati, H. K. A.; Das, P. K.; Tighe, S. L.; and Baaj, H., “Evalua-
1. Kazmi, S. M. S.; Munir, M. J.; Wu, Y.-F.; Lin, X.; and Ahmad, M. R., tion of Various Treatment Methods for Enhancing the Physical and Morpho-
“Investigation of Thermal Performance of Concrete Incorporating Different logical Properties of Coarse Recycled Concrete Aggregate,” Construction
Types of Recycled Coarse Aggregates,” Construction and Building Mate- and Building Materials, V. 112, June 2016, pp. 284-298. doi: 10.1016/j.
rials, V. 270, Feb. 2021, p. 121433. doi: 10.1016/j.conbuildmat.2020.121433 conbuildmat.2016.02.176
2. Wu, Y.-F.; Kazmi, S. M. S.; Munir, M. J.; Zhou, Y.; and Xing, F., 20. Wang, J.; Vandevyvere, B.; Vanhessche, S.; Schoon, J.; Boon, N.; and
“Effect of Compression Casting Method on the Compressive Strength, De Belie, N., “Microbial Carbonate Precipitation for the Improvement of
Elastic Modulus and Microstructure of Rubber Concrete,” Journal Quality of Recycled Aggregates,” Journal of Cleaner Production, V. 156,
of Cleaner Production, V. 264, Aug. 2020, p. 121746. doi: 10.1016/j. July 2017, pp. 355-366. doi: 10.1016/j.jclepro.2017.04.051
jclepro.2020.121746 21. Saravanakumar, P.; Abhiram, K.; and Manoj, B., “Properties of
3. World Business Council for Sustainable Development, “The Cement Treated Recycled Aggregates and Its Influence on Concrete Strength
Sustainability Initiative - Recycling Concrete,” World Business Council for Characteristics,” Construction and Building Materials, V. 111, May 2016,
Sustainable Development, Geneva, Switzerland, 2009, 42 pp. pp. 611-617. doi: 10.1016/j.conbuildmat.2016.02.064
4. Munir, M. J.; Kazmi, S. M. S.; Wu, Y.-F.; and Patnaikuni, I., “Influ- 22. Tam, V. W. Y.; Butera, A.; Le, K. N.; and Li, W., “Utilising CO2 Tech-
ence of Concrete Strength on the Stress-Strain Behavior of Spirally nologies for Recycled Aggregate Concrete: A Critical Review,” Construc-
Confined Recycled Aggregate Concrete,” IOP Conference Series: Mate- tion and Building Materials, V. 250, July 2020, p. 118903. doi: 10.1016/j.
rials Science and Engineering, V. 829, No. 1, 2020, p. 012004. doi: conbuildmat.2020.118903
10.1088/1757-899X/829/1/012004 23. Wang, R.; Yu, N.; and Li, Y., “Methods for Improving the Micro-
5. Kazmi, S. M. S.; Munir, M. J.; Wu, Y.-F.; and Patnaikuni, I., structure of Recycled Concrete Aggregate: A Review,” Construction
“Mechanical and Post-Cracking Performance of Recycled Aggregate and Building Materials, V. 242, May 2020, p. 118164. doi: 10.1016/j.
Concrete Incorporating Synthetic Fibers,” IOP Conference Series: Mate- conbuildmat.2020.118164
rials Science and Engineering, V. 829, No. 1, 2020, p. 012003. doi: 24. Hanif, A., “Influence of the Pretreatment of Recycled Aggregates,”
10.1088/1757-899X/829/1/012003 Advances in Construction and Demolition Waste Recycling: Management,
6. Sohu, Inc., “Heavy on January 1, 2019, Shanghai Forced Use of Recy- Processing and Environmental Assessment, F. Pacheco-Torgal, Y. Ding, F.
cled Sand Aggregate,” Sohu, Inc., Beijing, China, 2019, https://m.sohu. Colangelo, R. Tuladhar, and A. Koutamanis, eds., Woodhead Publishing,
com/a/284790687_806515. (last accessed Jan. 31, 2022) (in Chinese) Sawston, UK, 2020, pp. 159-180.
7. Munir, M. J.; Kazmi, S. M. S.; Wu, Y.-F.; Patnaikuni, I.; Zhou, Y.; 25. Munir, M. J.; Wu, Y.-F.; Kazmi, S. M. S.; Patnaikuni, I.; Zhou, Y.;
and Xing, F., “Stress Strain Performance of Steel Spiral Confined Recycled and Xing, F., “Stress-Strain Behavior of Spirally Confined Recycled Aggre-
Aggregate Concrete,” Cement and Concrete Composites, V. 108, Apr. 2020, gate Concrete: An Approach Towards Sustainable Design,” Resources,
Article No. 103535. doi: 10.1016/j.cemconcomp.2020.103535 Conservation and Recycling, V. 146, 2019, pp. 127-139. doi: 10.1016/j.
8. Kazmi, S. M. S.; Munir, M. J.; Wu, Y.-F.; Patnaikuni, I.; Zhou, Y.; and resconrec.2019.03.043
Xing, F., “Axial Stress-Strain Behavior of Macro-Synthetic Fiber Reinforced

26. Hoek, E., and Brown, E. T., “Empirical Strength Criterion for Rock 41. Gardner, N. J., “Effect of Temperature on the Early-Age Properties
Masses,” Journal of the Geotechnical Engineering Division, ASCE, V. 106, of Type I, Type III, and Type I/Fly Ash Concretes,” ACI Materials Journal,
No. 9, Sept. 1980, pp. 1013-1035. doi: 10.1061/AJGEB6.0001029 V. 87, No. 1, Jan.-Feb. 1990, pp. 68-78.
27. Richart, F. E.; Brandtzæg, A.; and Brown, R. L., “A Study of the 42. Oluokun, F. A.; Burdette, E. G.; and Deatherage, J. H., “Splitting
Failure of Concrete under Combined Compressive Stresses,” Engineering Tensile Strength and Compressive Strength Relationships at Early Ages,”
Experiment Station, Bulletin No. 185, University of Illinois at Urbana- ACI Materials Journal, V. 88, No. 2, Mar.-Apr. 1991, pp. 115-121.
Champaign, Urbana, IL, V. 26, No. 12, Nov. 1928, 110 pp. 43. Mokhtarzadeh, A., and French, C., “Mechanical Properties of High-
28. Shah, S. P.; Fafitis, A.; and Arnold, R., “Cyclic Loading of Strength Concrete with Consideration for Precast Applications,” ACI Mate-
Spirally Reinforced Concrete,” Journal of Structural Engineering, rials Journal, V. 97, No. 2, Mar.-Apr. 2000, pp. 136-148.
ASCE, V. 109, No. 7, July 1983, pp. 1695-1710. doi: 10.1061/ 44. EN 1992-1-1:2004, “Eurocode 2: Design of Concrete Structures –
(ASCE)0733-9445(1983)109:7(1695) Part 1-1: General Rules and Rules for Building,” European Committee for
29. Mander, J. B.; Priestley, M. J. N.; and Park, R., “Theoretical Standardization, Brussels, Belgium, 2004, 227 pp.
Stress-Strain Model for Confined Concrete,” Journal of Structural Engi- 45. Arɩoglu, N.; Girgin, Z. C.; and Arɩoglu, E., “Evaluation of Ratio
neering, ASCE, V. 114, No. 8, Sept. 1988, pp. 1804-1826. doi: 10.1061/ between Splitting Tensile Strength and Compressive Strength for Concretes
(ASCE)0733-9445(1988)114:8(1804) up to 120 MPa and its Application in Strength Criterion,” ACI Materials
30. El-Dash, K. M., and Ahmad, S. H., “A Model for Stress–Strain Journal, V. 103, No. 1, Jan.-Feb. 2006, pp. 18-24.
Relationship of Spirally Confined Normal and High-Strength Concrete 46. NZS 3101:2006, “The Design of Concrete Structures,” Standards
Columns,” Magazine of Concrete Research, V. 47, No. 171, June 1995, New Zealand, Wellington, New Zealand, 2006.
pp. 177-184. doi: 10.1680/macr.1995.47.171.177 47. JSCE:2007, “Standard Specification for Concrete Structures,” Japan
31. Hoshikuma, J.; Kawashima, K.; Nagaya, K.; and Taylor, A. W., Society of Civil Engineers, Tokyo, Japan, 2007.
“Stress-Strain Model for Confined Reinforced Concrete in Bridge Piers,” 48. AS 3600-2009, “Concrete Structures,” Standards Australia, Sydney,
Journal of Structural Engineering, ASCE, V. 123, No. 5, May 1997, Australia, 2009, 199 pp.
pp. 624-633. doi: 10.1061/(ASCE)0733-9445(1997)123:5(624) 49. ACI Committee 318, “Building Code Requirements for Structural
32. Razvi, S., and Saatcioglu, M., “Confinement Model for High-Strength Concrete (ACI 318-11) and Commentary (ACI 318R-11),” American
Concrete,” Journal of Structural Engineering, ASCE, V. 125, No. 3, Mar. Concrete Institute, Farmington Hills, MI, 2011, 503 pp.
1999, pp. 281-289. doi: 10.1061/(ASCE)0733-9445(1999)125:3(281) 50. Chhorn, C.; Hong, S. J.; and Lee, S. W., “Relationship between
33. Assa, B.; Nishiyama, M.; and Watanabe, F., “New Approach for Compressive and Tensile Strengths of Roller-Compacted Concrete,”
Modeling Confined Concrete. I: Circular Columns,” Journal of Structural Journal of Traffic and Transportation Engineering, V. 5, No. 3, June 2018,
Engineering, ASCE, V. 127, No. 7, July 2001, pp. 743-750. doi: 10.1061/ pp. 215-223. (English edition)
(ASCE)0733-9445(2001)127:7(743) 51. Girgin, Z. C.; Arɩoglu, N.; and Arɩoglu, E., “Evaluation of Strength
34. Bing, L.; Park, R.; and Tanaka, H., “Stress-Strain Behavior of High- Criteria for Very-High-Strength Concretes under Triaxial Compression,”
Strength Concrete Confined by Ultra-High- and Normal-Strength Trans- ACI Structural Journal, V. 104, No. 3, May-June 2007, pp. 278-284.
verse Reinforcements,” ACI Structural Journal, V. 98, No. 3, May-June 52. Richart, F. E.; Brandtzæg, A.; and Brown, R. L., “The Failure of
2001, pp. 395-406. Plain and Spirally Reinforced Concrete in Compression,” Engineering
35. Wei, Y., and Wu, Y.-F., “Compression Behavior of Concrete Columns Experiment Station, Bulletin No. 190, University of Illinois at Urbana-
Confined by High Strength Steel Wire,” Construction and Building Mate- Champaign, Urbana, IL, V. 26, No. 31, Apr. 1929, 80 pp.
rials, V. 54, Mar. 2014, pp. 443-453. doi: 10.1016/j.conbuildmat.2013.12.083 53. Mander, J. B.; Priestley, M. J. N.; and Park, R., “Observed Stress-
36. Munir, M. J.; Kazmi, S. M. S.; Wu, Y.-F.; Patnaikuni, I.; Wang, J.; and Strain Behavior of Confined Concrete,” Journal of Structural Engi-
Wang, Q., “Development of a Unified Model to Predict the Axial Stress– neering, ASCE, V. 114, No. 8, Sept. 1988, pp. 1827-1849. doi: 10.1061/
Strain Behavior of Recycled Aggregate Concrete Confined through Spiral (ASCE)0733-9445(1988)114:8(1827)
Reinforcement,” Engineering Structures, V. 218, Sept. 2020, p. 110851. 54. Sheikh, S. A., and Toklucu, M. T., “Reinforced Concrete Columns
doi: 10.1016/j.engstruct.2020.110851 Confined by Circular Spirals and Hoops,” ACI Structural Journal, V. 90,
37. Wu, Y.-F., and Zhou, Y.-W., “Unified Strength Model Based on No. 5, Sept.-Oct. 1993, pp. 542-553.
Hoek-Brown Failure Criterion for Circular and Square Concrete Columns 55. Munir, M. J.; Kazmi, S. M. S.; Wu, Y.-F.; and Lin, X., “Axial Stress-
Confined by FRP,” Journal of Composites for Construction, ASCE, V. 14, Strain Performance of Steel Spiral Confined Acetic Acid Immersed and
No. 2, Apr. 2010, pp. 175-184. doi: 10.1061/(ASCE)CC.1943-5614.0000062 Mechanically Rubbed Recycled Aggregate Concrete,” Journal of Building
38. Carino, N. J., and Lew, H. S., “Re-examination of the Relation Engineering, V. 34, Feb. 2021, p. 101891. doi: 10.1016/j.jobe.2020.101891
Between Splitting Tensile and Compressive Strength of Normal Weight 56. Ahmed, W., and Lim, C. W., “Production of Sustainable and Struc-
Concrete,” ACI Journal Proceedings, V. 79, No. 3, May-June 1982, tural Fiber Reinforced Recycled Aggregate Concrete with Improved Frac-
pp. 214-218. ture Properties: A Review,” Journal of Cleaner Production, V. 279, Jan.
39. Raphael, J. M., “Tensile Strength of Concrete,” ACI Journal Proceed- 2021, p. 123832. doi: 10.1016/j.jclepro.2020.123832
ings, V. 81, No. 2, Mar.-Apr. 1984, pp. 158-165. 57. Tam, V. W. Y.; Wattage, H.; Le, K. N.; Buteraa, A.; and Soomro, M.,
40. Shah, S. P., and Ahmad, S. H., “Structural Properties of High “Methods to Improve Microstructural Properties of Recycled Concrete
Strength Concrete and its Implications for Precast Prestressed Concrete,” Aggregate: A Critical Review,” Construction and Building Materials,
PCI Journal, V. 30, No. 6, Nov.-Dec. 1985, pp. 92-119. doi: 10.15554/ V. 270, Feb. 2021, p. 121490.
pcij.11011985.92.119

NEW
Symposium
Publications from ACI
An ACI Technical Publication An ACI Technical Publication
SYMPOSIUM VOLUME SYMPOSIUM VOLUME
Recent Developments in High Strain Rate

Mechanics and Impact Behavior of Concrete
Foundations for Dynamic
Equipment
SP-347
SP-348
Editor:
Carl A. Nelson
Editors:
Eric Jacques and Mi G. Chorzepa
SP-347: Recent Developments in High SP-348: Foundations for Dynamic

Strain Rate Mechanics and Impact Equipment
Behavior of Concrete
This special publication grew out of the Technical
This Symposium Volume reports on the latest Session titled “Application of ACI 351-C Report on
developments in the field of high strain rate Dynamic Foundations,” held at the ACI Spring 2019
mechanics and behavior of concrete subject to Convention in Québec City, Québec. Following this
impact loads. This effort supports the mission event, Committee 351 decided to undertake a special
of ACI Committee 370 “Blast and Impact Load publication with contributions from those session
Effects” to develop and disseminate information participants willing to develop their presentations
on the design of concrete structures subjected to into full-length papers. Three papers included in
impact, as well as blast and other short-duration the current publication were contributed by these
dynamic loads. presenters and their coauthors, with six additional
papers provided by others.
Available in PDF format: $69.50
(ACI members: $39.00) ($30.50 savings) Available in PDF format: $69.50
(ACI members: $39.00) ($30.50 savings)
+1.248.848.3700 • www.concrete.org    
Title No. 119-S44
Experiments on Partially Coupled Concrete Wall System

with Perforated Steel Beam
by Woo-Young Lim, Thomas H.-K. Kang, Donghyuk Jung, and Sung-Gul Hong
A series of cyclic loading tests was carried out for five half-scale fully coupled walls (DC ≥ 0.66) and partially coupled walls
hybrid steel coupling beams to evaluate the seismic performance of (DC < 0.66).
a partially coupled wall system with different design parameters. During the past decades, efforts have been made to
This system comprises the structural concrete wall, embedded steel improve the seismic performance of the coupling beams.
beam, steel coupling beam with bolted connection, and top-seat
As part of such efforts, experimental and analytical studies
angles. The two primary test parameters were the web opening size
for various sectional details of reinforced concrete (RC)
of the coupling beam that determines its shear-to-moment capacity
ratio, and the presence of the top-seat angles. Test results showed coupling beams have been performed (Paulay and Binney
that the shear-to-moment capacity ratio of the coupling beam 1974; Tassios et al. 1996; Galano and Vignoli 2000; Canbolat
greatly influences their deformation and energy dissipation capaci- et al. 2005; Harries et al. 2005; Barbachyn et al. 2012; Naish
ties, as well as the degree of concrete wall damage. Top-seat angles et al. 2013). These studies have found that the diagonally
were also found to increase the lateral stiffness and energy dissipa- reinforced coupling beams have excellent energy dissipa-
tion of the coupled shear wall system. Creation of the web opening tion and deformation capacities. Hence, current concrete
leads to enhanced seismic performance of the beam as a replace- codes (Section 18.10.7 of ACI 318-19 [ACI Committee
able structural fuse. Based on the test results, a design procedure 318 2019]) provide two different details of the diagonally
of the hybrid steel coupling beam system with bolted connections reinforced coupling beam section with respect to the aspect
is proposed.
ratio. However, installation of the diagonal reinforcements
Keywords: bolted connections; cyclic tests; design procedure; hybrid steel on a construction site is considered to be quite demanding.
coupling beam; seismic behavior; shear-to-moment capacity; web opening. Harries et al. (2005) reported that if the shear demand of the
steel coupling beam exceeds the value of 0.5√fc′ (in MPa),
INTRODUCTION installation of the diagonal reinforcement in the RC walls is
Coupling beams placed between two shear walls transfer almost impossible due to interferences between reinforcing
loads between the walls and integrate their independent bars and diagonal reinforcement.
structural responses. Figure 1 shows the typical geometric To overcome this issue, a hybrid coupling beam system
configuration and load-transfer mechanism of the coupling that uses a steel beam or a steel-reinforced concrete (SRC)
beam-shear wall system. When the coupled shear walls are beam has been developed (Clarke and Symons 1978;
subjected to large lateral load, plastic hinges are first formed Marcakis and Mitchell 1980; Mattock and Gaffar 1982;
at the coupling beam before the concrete walls reach their Shahrooz et al. 1993; Harries et al. 1993, 2000; Gong and
ultimate limit state. Therefore, the coupling beam should be Shahrooz 2001; Lam et al. 2005; Park and Yun 2005; Lim
designed to have sufficient strength, stiffness, and deforma- et al. 2016, 2018). In general, this system consists of coupled
tion capacities because the response of the coupling beam concrete walls and H-shaped steel coupling beam. Table 1
substantially affects the overall seismic performance of the provides various shear resistance models of steel coupling
coupled shear wall system (Paulay and Priestley 1992). beams proposed by several more researchers.
In general, it is known that the lateral responses of the Previous studies (Harries et al. 1993; Park and Yun 2005)
coupled shear walls are governed by the ratio of the stiff- showed that hybrid coupling beams exhibited better seismic
ness between the coupling beam and the walls, named as the performance than that of the RC coupling beams with
degree of coupling (DC) (Chaallal et al. 1996). Depending diagonal reinforcement. However, in this system, concrete
on this ratio, the coupled shear wall can be categorized as the cover spalling occurs at a junction between concrete wall
wall with high, low, or intermediate coupling. The classifica- face and steel coupling beam, and it is considered one of
tion of the coupled shear wall is determined by the stiffness the important factors affecting the shear resistance of
of the coupling beam as well as the boundary condition at the the coupled wall system. Harries et al. (1993) stated that
coupling beam-wall connection. The high, low, and interme- concrete cover spalling is a typical failure mode in a mono-
diate coupled shear wall system defined by the DC can be lithic coupling beam system and can reduce embedment
presumed to have the fixed, pinned, and partially restrained length of the steel beam. Therefore, one of the crucial issues
boundary conditions, respectively. In Canadian Standards with the hybrid steel coupling beam system is how to deal
Association (CSA) A23.3 (2019), the DC of a coupled ACI Structural Journal, V. 119, No. 2, March 2022.
shear wall building is represented as the ratio of the total MS No. S-2021-134.R1, doi: 10.14359/51734341, received October 2, 2021, and
base overturning moment of the building resisted by push- Institute. All rights reserved, including the making of copies unless permission is
pull action of the coupled axial forces in the walls. Based on obtained from the copyright proprietors. Pertinent discussion including author’s
the calculated DC, CSA distinguishes the types system as is received within four months of the paper’s print publication.

Fig. 1—Coupled shear walls-coupling beam system.
Table 1—Existing shear resistance models
Authors Shear resistance Research subjects
Marcakis and Mitchell 0.85 f cble

Vn For embedded steel members
(1980) 1 3.6e1 / le
0.58 0.221
Vn 0.85 f c1ble
0.88 a / le
Mattock and Gaffar For embedded steel sections as
0.58 0.221
(1982) Vn f b1ble brackets
0.88 a / le
where f b 4.5 f c twall / b

0.66
0.85 f cbef leff

Vn For monolithic hybrid steel coupling
Harries et al. (1993) 3.6e
1 beams
leff
PCI Design Handbook 0.85 f cble

Vn For steel haunches
(2010) 3.67 4a / le
Vn 0.85 f cbef 1 2 le for le ≥ le _ min

Vp
le _ min
0.85 f cbef 1 2
For hybrid steel coupling beams with a
Lim et al. (2016)
X 2 0.5 X 2 X wide-flange section

X 4
X V p / 0.85 f cbef a

t
0.66
0.58 0.22
ANSI/AISC 341-16 For monolithic steel coupling beams
Vn 4.04 f c wall 1b f le 1
(2016) b 0.88 Lcs with a wide-flange section
f
2le

Note: fc′ is compressive strength of concrete; b is width of steel coupling beam; le is embedded length of steel beam; le_min is required minimum embedded length of steel beam; a
is shear span; e is equal to l/2 + leff/2 + cc; leff is equal to le – cc; cc is spalled cover concrete; e1 is equal to le/2+a; twall is thickness of wall; β1 is ratio of depth equivalent rectangular
stress distribution to depth of flexural compression zone as specified in ACI 318-19; fb is bearing strength of concrete; Vu is ultimate shear strength; Vp is plastic shear strength of steel
coupling beam; be is effective width of steel coupling beam; bf is width of flange; and Lcs is clear span of coupling beam.
with the concrete damage. According to Lim et al. (2016), a wall system. Also, premature concrete wall failure, such
reduction of the embedment length might cause the pullout as concrete cover spalling, can bring about a sudden drop
failure of the embedded steel beam. This failure mode could of the strength, and this leads to a low energy dissipation
lead to degradation of the strength, stiffness, energy dissi- capacity (Park and Yun 2005). To prevent such performance
pation, and/or deformation capacity of the coupled shear degradation caused by the premature wall failure in the early

stage during seismic actions, various reinforcement details coupled structural walls. Based on the test results, appro-
were proposed for concrete walls (Gong and Shahrooz 2001; priate design guides for the hybrid steel coupling beam
Harries et al. 1993; Lim et al. 2018). system were proposed.
Several researchers suggested an alternative design
concept which adopted a replaceable shear link in the TEST PROGRAM
steel coupling beam (Fortney et al. 2007; Farsi et al. 2016; Details of test specimens
Ji et al. 2017a,b). When the coupled shear wall system is Five half-scale specimens include four hybrid coupling
subjected to lateral load, this weak link accommodates most beams (HCB1A, HCBA, HCB3A, and HCB3) with web
of the inelastic deformations as a structural fuse and mini- opening and one control coupling beam (EL600A), which
mizes damages in other structural components. Despite had no punctuated web opening but the same embedment
these efforts, however, some of those proposed systems still length of structural steel in the wall. All the specimens had
suffer from significant concrete damage, and/or may require top and seat angles (top-seat angles), except for HCB3 that
complicated connection details between the shear link and was identical to HCB3A but only without top-seat angles.
steel coupling beams. The ratio (ln/H) of net distance (1440 mm [59.7 in.]) to the
The current study proposes a new type of hybrid steel height of the coupling beam (378 mm [14.9 in.]) is 3.81.
coupling beam which incorporates the feature of structural Figure 2 shows the details of the test specimens. The primary
fuse by providing a circular web opening. Figure 1 illus- test parameters are: 1) shear-to-moment capacity ratio of the
trates the design philosophy of the proposed steel coupling coupling beam controlled by the web opening size; and 2)
beam. The web opening is devised to reduce shear strength the presence of the top-seat angles.
(Vn) of the coupling beam and induce web shear yielding The test specimen consisted of the structural concrete wall,
before plastic moment capacity (Mp) is reached. The steel a steel coupling beam, an embedded steel beam, and top-seat
coupling beam is designed such that its shear capacity (that angles. The overall height, width, and the thickness of the
is, the maximum shear developed in the beam) becomes wall were 1900, 1800, and 300 mm (74.8, 70.9, and 11.8 in.),
less than shear (VMp = 2Mp/ln) required to cause plastic respectively. The design of concrete wall was carried out
moment, where ln is clear span of coupling beam. As this as per Section 18.10 of ACI 318-19. The wall satisfied the
shear-to-moment capacity ratio decreases, the web opening minimum longitudinal and transverse reinforcement ratio
acts as a structural fuse, preventing transfer of excessive requirements (at least 0.0025) of special structural walls.
force demands to the walls and accordingly alleviating the D19 bars that had the diameter (db) of 19 mm (0.75 in.) were
concrete damage. The steel coupling beam equipped with placed for longitudinal reinforcement of the wall with spac-
web opening is expected to have considerably reduced stiff- ings of 200 mm (7.9 in.) along the distance corresponding to
ness compared to the one without web opening, and there- the length of the embedded steel beam and 270 mm (10.6 in.)
fore the proposed coupling beam-shear wall system can be for the other lengths. For better confinement adjacent to the
categorized as the partially coupled walls. The hybrid steel embedded beam, the authors provided three additional D16
coupling beam is bolt-connected to steel beams embedded in (db = 16 mm [0.63 in.]) closed hoops perpendicular to the
the concrete walls, enabling the replacement of the damaged embedded steel beam. D22 (db = 22 mm [0.87 in.]) bars
coupling beam after a seismic event. The seismic responses were horizontally installed at 225 mm (8.9 in.) intervals.
of the new system are experimentally investigated through Additional horizontal reinforcement (D22) was provided to
quasi-static cyclic loading tests on half-scale coupling beam- fix the closed hoops along the embedded steel beam at the
wall subassemblies with various shear-to-moment capacity distance of 100 mm (3.9 in.) from the edges of the embedded
ratios. Furthermore, a design guideline for the hybrid steel steel beam. Also, D19 (db = 19 mm [0.75 in.]) U-shaped
coupling beam with bolted connections is proposed based horizontal reinforcing bars were installed at the midheight
on the test results. of the embedded steel beam to resist local buckling of the
extreme longitudinal reinforcements, which may be induced
RESEARCH SIGNIFICANCE by the pullout failure of the embedded steel beam.
Despite considerable research and development efforts, All the steel members were manufactured at a factory.
conventional coupling beams still have areas of improve- For both steel coupling beam and embedded steel beam, the
ment in terms of construction feasibility and damage- same welded wide-flange section was used with dimensions
induced performance degradation. To deal with those issues, of H-400 x 175 x 5 x 11 mm (H-15.7 x 6.9 x 0.2 x 0.4 in.)
this study proposed a new hybrid coupling beam that has (height x flange width x web thickness x flange thickness).
an opening in the web. Simply by creating the web opening Design yield strength (Fy) and tensile strength (Fu) of all the
without complex connection details, the hybrid beam plays steel members used in this study were 300 and 400 MPa
a role as a replaceable structural fuse and can improve the (43.5 and 58 ksi), respectively. Based on the nominal yield
seismic performance of the coupled wall system without strength, all the coupling beams were designed as the
requiring additional mechanical devices. Besides the struc- shear-critical member as per AISC’s Steel Construction
tural purposes, the web opening can produce an aesthetic Manual (AISC 2011). According to ANSI/AISC 341-16
effect for the buildings and/or practical advantage as a path (2016), when the length (ln) of the coupling beam is equal to
for service equipment. Large-scale experimental testing or less than the value of 1.6Mp/Vp, the steel coupling beam
was conducted to explore how various design parameters can be considered to be shear-critical, where Mp and Vp are
of the hybrid beam affect the seismic responses of partially the plastic moment and plastic shear strengths of the steel

Fig. 2—Details of test specimens. (Note: Units in mm; 1 mm = 0.0394 in.)
coupling beam, respectively. On the other hand, if the length (0.8 in.) thick triangular stiffener was additionally installed
is equal to or larger than 2.6Mp/Vp, the coupling beam can to prevent plastic deformation of the angles. In each connec-
be assumed to be a flexural-critical member. After the mate- tion, high-tension bolts, M24 (F10T), with a yield strength
rial testing, however, it was revealed that the coupling beam of 900 MPa (130.5 ksi) and tensile strength of 1000 MPa
of EL-600A fell within an intermediate member (that is, (145 ksi) were used. The details of the test specimens are
1.6Mp/Vp ≤ ln ≤ 2.6Mp/Vp) due to unexpectedly high yield summarized in Table 2. More detailed explanations on the
strength of the web plate. All other coupling beams (HCB1A, design process of the test specimens can be found from Lim
HCBA, HCB3A, and HCB3) were found to be shear-critical et al. (2018).
as intended and desirable due to yielding of short shear links
(that is, ln ≤ 1.6Mp/Vp). Material properties
The embedded steel beam had dimensions of H-400 x Tensile tests of steel plates and reinforcing bars were
175 x 5 x 11 mm (H-15.7 x 6.9 x 0.2 x 0.4 in.) and its length conducted according to ASTM E8/E8M (2016). As indicated
was 600 mm (23.6 in.). To prevent unexpected pullout earlier in this paper, the specified design yield strengths of
failure of the embedded beam, four stiffeners with a thick- the steel plate and reinforcing bar used in manufacturing
ness of 10 mm (0.4 in.) were placed at an interval of 200 mm the test specimens were 300 and 400 MPa (43.5 and 58.0
(7.9 in.) on each side of the beam. ksi), respectively. Table 3 shows stress and strain values
The top-seat angles were manufactured to be strong enough measured at yield and ultimate states. It was found that the
to induce the plastic limit state of the coupling beam and measured yield strengths of the web (5 mm) and flange (11
remain in the elastic state until the end of the test. The height mm) used in the fabrication of the couple beam specimens
and length of the top-seat angle were 200 mm (7.9 in.), and were 426.3 and 305.1 MPa (61.8 and 44.3 ksi), respectively.
thickness was 20 mm (0.8 in.). At top-seat angles, a 20 mm Also, compression tests were performed for concrete used

Table 2—Details of test specimens
Web opening Top-seat Shear capacity Shear force (VMp) at Shear-to-moment
Specimens Concrete wall, mm Steel beam, mm diameter dw, mm angles (Vpo or Vp), kN moment capacity, kN capacity ratio (Vp/VMp)
HCB1A 200 Y 227.6 371.8 0.61
HCB2A 268 Y 140.7 355 0.4
HCB3A 1900 x 1800 x 300 H-400 x 175 x 5 x 11 300 Y 99.8 345.3 0.29
HCB3 300 N 99.8 345.3 0.29
EL600A — Y 483.4 393 1.23
Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.
Table 3—Material properties of reinforcing bars and steel plates

Diameter or thickness, mm, (in.) Yield stress, MPa (ksi) Yield strain, mm/mm Tensile stress, MPa (ksi) Tensile strain, mm/mm
13 mm (0.51) 355.8 (51.6) 0.002 594.0 (86.2) 0.132
16 mm (0.63) 354.3 (51.4) 0.002 645.4 (93.6) 0.086
Reinforcement
19 mm (0.75) 418.0 (60.6) 0.002 630.1 (91.4) 0.136
22 mm (0.87) 363.3 (52.7) 0.002 607.5 (88.1) 0.145
5 mm (0.20) 426.3 (61.8) 0.002 459.6 (66.7) 0.072
7 mm (0.28) 425.6 (61.7) 0.0025 610.4 (88.5) 0.082
10 mm (0.39) 304.3 (44.1) 0.002 417.5 (60.6) 0.072
Steel plates
11 mm (0.43) 305.1 (44.3) 0.0019 463.3 (67.2) 0.054
12 mm (0.47) 311.6 (45.2) 0.0016 435.3 (63.1) 0.085
20 mm (0.79) 243.5 (35.3) 0.0017 336.4 (48.8) 0.049
for the walls as per ASTM C39/C39M (2018). The design ratios of HCB1A, HCB2A, HCB3A, HCB3, and EL600A
compressive strength of concrete was 35 MPa (5.1 ksi) and were 0.47, 0.29, 0.21, 0.21, and 1, respectively.
the strength of concrete measured on the testing day was 38
MPa (5.5 ksi). Design of bolted connections between embedded
beam and steel coupling beam
Shear strength ratio In this study, double shear simple connections were used
The shear strength ratio, one of the primary test param- to connect the steel coupling beam and the embedded steel
eters, was determined by intentionally creating the web beam (Fig. 2). For manufacturing the connections, steel
opening in the steel coupling beam and indicates how much plates (Fy = 300 MPa [43.5 ksi] and Fu = 400 MPa [58.0 ksi])
the shear capacity was reduced in comparison to the full with a thickness of 20 mm (0.8 in.) were used. The overall
capacity. The shear strength ratio (Vpo/Vp) was defined as the height and width of the connection were 400 and 120 mm
ratio of the plastic shear strength (Vpo = 0.6Fywtw(h – dw)) (15.7 and 4.7 in.), respectively.
of the coupling beams with the web opening to the plastic The bolted connection was designed as a slip-critical
shear strength (Vp = 0.6Fywtwh) of EL600A (control spec- connection in accordance with Section J3 of ANSI/AISC
imen without web opening). Here, Fyw is the specified yield 360-16 (2016). The design strength of the slip-critical connec-
strength of the web; tw is the thickness of the web; h is the tion is defined as the minimum value among the slip resis-
net height of the web (= H – 2tf); H is the overall height of tance (= ϕμhfT0Ns) of the high-tension bolts, design shear
the beam section; tf is the thickness of the flange; and dw is strength (= ϕnbFnvAbNs) of the bolts, and design bearing
the diameter of the web opening. The reason for adopting the strength (= ϕ1.2LctFu ≤ 2.4dtFu) of the bolt holes. Here, ϕ is
net distance between the flanges in the calculation of shear the strength reduction factor (0.75); μ is the mean slip coef-
strength is that the built-up steel beam was used. ficient; hf is the factor for fillers; T0 is a minimum fastener
For HCB1A, HCB2A, and HCB3A, the diameters of tension; Ns is the number of shear planes required to permit
web openings were 200, 268, and 300 mm (7.9, 10.6, and the connection to slip; nb is the number of bolts; Fnv is the
11.8 in.), respectively. The web opening diameter of HCB3 nominal shear strength of the high-tension bolts; Ab is the
was the same as that of HCB3A, but HCB3 did not have nominal area of the bolt; Lc is the net distance between the
the top-seat angles. The web openings were located at a end of the hole and the end of the material to be connected or
distance of 382 mm (15.0 in.) from the wall surface. The the end of the adjacent hole; t is the thickness of the connec-
plastic shear strengths of HCB1A, HCB2A, HCB3A, HCB3, tion; Fu is the tensile strength, and d is the nominal diameter
and EL600A based on the measured material strengths of the bolt. The design strength of the bolted connection was
(Table 3) were calculated as 227.6, 140.7, 99.8, 99.8, and taken as the design slip resistance of the high-tension bolts,
483.4 kN (51.2, 31.6, 22.4, 22.4, and 108.7 kip), respectively which was equal to 948 kN (213.3 kip).
(Table 2). Accordingly, the corresponding shear strength

Fig. 3—Cyclic loading schedule and test setup. (Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.)
Because the bolted connection should be maintained in the Test setup
elastic state until the coupling beam reaches the plastic state, Figure 3 shows the test setup and loading protocol for
the required tensile strength (Treq) of the connection can be the half-span cantilevered hybrid steel coupling beam. To
obtained as follows apply reversed cyclic loading, an actuator with a capacity
of 2000 kN (450 kip) was installed at the distance (ln/2)
Mp of 720 mm (28.3 in.) from the side face of the wall. The
Treq (1) lateral displacement at the point of loading was measured
H tf
using a linear variable differential transducer (LVDT).
where Mp is the plastic moment strength (= ZpFyw) of the Displacement-controlled loading was repeatedly applied three
steel coupling beam; and Zp is the plastic section modulus. times per each target drift ratio until the post-peak strength
For the beam with a web opening, the plastic section reached 80% of the peak strength (Kang and Wallace 2008).
modulus can be calculated by applying the maximum height
of the web opening (Darwin and Lucas 1990) as follows TEST RESULTS
Damage and crack patterns
Z p Bt f H t f tw H 2t f Figure 4 presents damage status and crack patterns of test
1
d w2 (2)
2

4 specimens at the end of the test. Figure 4(a) shows damages
where B and H are the width and overall height of the beam of the steel coupling beam. In Fig. 4(b), pictures of the wall
section, respectively. are provided except for HCB3A, which did not have visible
The plastic moment strengths (Mp) of HCB1A, HCB2A, damage, and the crack patterns of all walls are illustrated in
HCB3A, HCB3, and EL600A based on the measured yield Fig. 4(c). In the case of HCB1A, vertical cracks initiated at
strength of the steel plates (Table 3) were 267.7, 255.6, the joint between the wall and steel coupling beam, and then
248.6, 248.6, and 283 kN-m (197.4, 188.5, 183.4, 183.4, and propagated to the inside of the wall until a drift ratio of 2%.
208.7 kip-ft), respectively. Therefore, the required tensile Horizontal cracks occurred at the portion where the 20 mm
strengths of each specimen obtained from Eq. (1) were (0.8 in.) thick stiffener was placed at the drift ratio of 3%.
688.2, 657, 639.2, 639.2, and 727.4 kN (154.7, 147.7, 143.7, Also, the periphery of the web opening was distorted, but
143.7, and 163.5 kip), respectively. Because the design slip no tearing occurred. For HCB2A, vertical cracking started
resistance (948 kN [213.3 kip]) of the high-tension bolts was to occur from the drift ratio of 1% along the perimeter of the
higher than the required tensile strength, the strength of the embedded steel beam and developed to the inside of the wall
bolted connection complied with the design requirements of up to a drift ratio of 3%. The distortion of the web opening
the ANSI/AISC 360-16 (2016). was very similar to that of HCB1A. On the other hand, for
In the test setup considered in this study, it was assumed HCB3A, the periphery of the web opening was torn at a drift
that an external force was applied to the center of the span ratio of 4%, and the test was terminated. In this specimen, no
(ln) of the coupling beam (refer to the “Test setup” section). crack was observed in the concrete wall. The failure mode
Shear force (VMp = 2Mp/ln) needed to cause the plastic of HCB3 (without top-seat angles) was similar to that of the
moment capacity (Mp) in all the specimens can be accord- HCB3A, but no fracture occurred at the perimeter of the web
ingly computed as shown in Table 2. From the shear capacity opening. Instead, sliding of the steel beam increased slightly
(Vp) estimated in the previous section, the shear-to-moment due to bearing failure of bolt holes at the bolted connection.
capacity ratios (Vp/VMp) of HCB1A, HCB2A, HCB3A, Finally, in EL600A, both vertical and horizontal cracking
HCB3, and EL600A were 0.61, 0.4, 0.29 0.29, and 1.23, initiated at a 2% drift ratio along the flanges and stiffeners of
respectively. These ratios indicate that the specimens with the embedded steel beam, respectively. Furthermore, diag-
web opening would yield in shear, whereas EL600A would onal cracks were developed and propagated deep inside of
yield in flexure. the wall at a 2.5% drift ratio. Consequently, in the speci-
mens with the web opening (HCB1A, HCB2A, HCB3A,

Fig. 4—Damage and crack patterns of specimens at end of tests.
and HCB3), no significant wall damage occurred under the [28.3 in.]) between the side face of the wall and the loading
lateral cyclic loading. However, in the case of EL600A, point. Test results are summarized in Table 4.
substantial cracks were observed around the top-seat angles Figure 6 illustrates how yield point was defined in this
and embedded steel beam. The minimal wall damage of study (Lim et al. 2016, 2018): 1) draw a straight line from
the specimens with the web opening is clearly contrasted the origin passing through the point corresponding to 75% of
with severe cracks of a number of wall specimens coupled the peak strength in the load-displacement envelope curve;
conventional steel beams in the previous experimental 2) draw a horizontal line from the peak strength point toward
studies (Lim et al. 2016, 2018). the strength (vertical) axis; 3) find an intersection point
between those two straight lines; 4) draw a vertical line from
Load-drift ratio relationship the intersection point to the displacement (horizontal) axis;
Figure 5 shows the load-drift ratio relationship of each and 5) find another intersection point between the vertical
specimen subjected to the cyclic loading. The drift ratio line and envelope curve. The intersection point obtained in
was obtained by dividing the lateral displacement measured Step 5 was defined as the yield point.
by the LVDT (LV1 in Fig. 3) by the distance (= 720 mm

Fig. 5—Load-drift ratio relationship and envelope curves. (Note: 1 kN = 0.225 kip.)
At peak At failure
Positive (+) Negative (–) Positive (+) Negative (–)
−
Specimens
+
V peak , kN

@Vpeak , mm @Vpeak ,% V peak , kN @Vpeak , mm @Vpeak ,% V f+, kN f , mm f , % V f−, kN f , mm f , %
HCB1A 250.9 25.6 3.5 –239.3 –25.0 –3.4 176.1 35.3 4.8 –140.8 –35.8 –4.9
HCB2A 174.0 17.8 2.4 –168.0 –23.1 –3.2 110.1 32.8 4.5 –110.3 –33.4 –4.6
HCB3A 132.0 13.4 1.8 –123.8 –13.6 –1.9 88.8 34.2 4.7 –76.9 –35.2 –4.8
HCB3 131.2 25.8 3.5 –125.0 –26.5 –3.6 102.7 39.3 5.4 –90.7 –41.0 –5.6
EL600A 403.1 31.2 4.3 –398.6 –33.6 –4.7 306.1 39.1 5.3 –301.0 –40.7 –5.7
At yielding
Positive (+) Negative (–)
Specimens Vy+, kN y, mm y, % k y+, kN/mm Vy−, kN y , mm y , % k y−, kN/mm
HCB1A 248.9 21.8 3.0 11.4 –239.3 –24.5 –3.4 9.8
HCB2A 149.3 13.1 1.8 11.4 –153.6 –18.0 –2.5 8.5
HCB3A 102.1 4.1 0.6 25.0 –102.0 –7.0 –1.0 14.6
HCB3 111.6 20.2 2.8 5.5 –112.8 –20.4 –2.8 5.5
EL600A 352.7 21.4 3.0 16.5 –374.2 –28.4 –3.9 13.2
Vpeak/Vp Mpeak/Mp μ, Δf/Δy Ω, Vpeak/Vd Cd, Δf/Δd
Specimens (+) (–) (+) (–) (+) (–) (+) (–) (+) (–)
HCB1A 1.1 1.05 0.67 0.64 1.6 1.5 1.56 1.49 3.24 2.61
HCB2A 1.23 1.19 0.49 0.47 2.5 1.9 1.75 1.69 14.86 2.60
HCB3A 1.32 1.24 0.38 0.36 8.4 5.0 1.88 1.76 15.49 12.15
HCB3 1.31 1.25 0.38 0.37 1.9 2.0 1.86 1.78 4.57 3.57
EL600A 0.83 0.82 1.03 1.01 1.8 1.4 1.18 1.17 1.86 1.65
Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip; 1 kN/mm = 5.71 kip/in.

with top-seat angles, and from 5.4 to 5.6% for the one
without angles (HCB3). In the case of EL600A, the drift
ratio at failure was 5.3% for positive and –5.7% for negative
loading. This result indicates that the strength of the hybrid
steel coupling beam was maintained above 80% of the peak
strength at least until the drift ratio of 4.5% was reached.
Deformation capacity of the specimens was also compared
by estimating ductility (μ = Δy/Δf). Ductility of HCB1A,
HCB2A, HCB3A, HCB3, and EL600A was approximately
1.6, 2.5, 8.4, 1.9, and 1.8 for positive loading and 1.5, 1.9,
5.0, 2.0, and 1.4 for negative loading, respectively. Overall,
the specimens with web opening exhibited more ductile
performance compared to EL600A. What should be noted
here is that the ductility of HCB3A was exceptionally high
among all the specimens. The ductility of HCB3A was
approximately four times higher than that of HCB3.
For readers’ information, other seismic parameters such as
Fig. 6—Definition of yielding point and ductility. overstrength factor (Ω) and deflection amplification factor
(Cd) were also evaluated from the test results. As illus-
Strength at failure (Vf) was defined as the post-peak
trated in Fig. 6, overstrength factor (Ω) was obtained with
strength corresponding to 80% of the peak strength.
respect to shear by calculating the ratio of the measured
Displacements at yielding (Δy) and ultimate state (Δf) were
peak shear force (Vpeak) to design shear strength (Vd), which
defined as the displacements corresponding to the yield
was computed using the nominal yield strength of the web
and failure strength, respectively. If the strength has not
(300 MPa [43.5 ksi]). Deflection amplification factor (Cd)
dropped below 80% of the peak strength, the failure strength
was determined as Δf/Δd where Δd is the displacement corre-
was determined to be the strength at the end of the test.
sponding to design shear strength (Vd). As was the case of
The initial stiffness (ky) was obtained from the slope of the
Vpeak/Vp, the specimens with the web opening showed much
straight line connecting the origin and yield point. Ductility
higher overstrengths (Ω ≈ 1.5 to 1.9) than that of EL600A
(μ) was defined as the ratio between the displacement (Δf/Δy)
(Ω ≈ 1.2) because of the shear yielding at the web, and the
at failure and yielding, as shown in Fig. 6.
values tended to increase with the size of the web opening
The ratios (Vpeak/Vp) of the measured peak shear force
(refer to Table 4). On the other hand, values of Cd varied
(Vpeak) to the calculated plastic shear strength (Vpo or Vp) for
greatly (Cd ≈ 1.7 to 15.5) due to high variability of Δd esti-
HCB1A, HCB2A, HCB3A, HCB3, and EL600A were 1.1,
mated from the test results, especially for the perforated
1.23, 1.32, 1.31, and 0.83 for positive loading and 1.05, 1.19,
specimens. In ASCE 7-16 (2016), values of Ω and Cd for
1.24, 1.25, and 0.82 for negative loading, respectively. For
the case of reinforced concrete ductile coupled walls (the
the specimens with Vpeak/Vp greater than 1, it is presumed
most similar system) are suggested as 2.5 and 8, respec-
that strain hardening and stress redistribution possibly
tively. Compared to those standard values, the values of
contributed to the higher peak shear strengths of coupling
Ω of the hybrid coupling beam-walls are 61 to 73% (with
beams. The ratios of Vpeak/Vp revealed that all the coupling
opening) and 47% (without opening). The values of Cd are
beams experienced web shear yielding, except for EL600A,
31 to 173% (with opening) and 22% (without opening) of
as expected. This judgment is also clearly supported by
the standard values.
ratios (Mpeak/Mp) of the measured peak moment (Mpeak) to
the plastic moment strength (Mp). The ratios of Mpeak/Mp of
Strain variations
HCB1A, HCB2A, HCB3A, HCB3, and EL600A were 0.67,
Strain gauges were attached to the web and flange of the
0.49, 0.38, 0.38, and 1.03 for positive loading and 0.64, 0.47,
steel coupling beam in the longitudinal direction. Strain
0.36, 0.36, and 1.01 for negative loading, respectively, indi-
variations on the web and flange are shown in Fig. 7(a) and
cating that only EL600A yielded in flexure. The specimens
(b), respectively, along with locations of the strain gauges.
with web opening indeed yielded in shear before their plastic
In Fig. 7(a), web strains were measured at locations 250 mm
moment capacities without opening were reached according
(9.8 in.) away from the edge of the web in HCB1A, HCB3A,
to the design philosophy.
and EL600A. For the HCB2A and HCB3 specimens, the
Furthermore, the hybrid steel coupling beam system
strains were measured at a distance of 145 mm (5.7 in.) from
showed a good deformation capacity. Drift ratios at the peak
the loading point and at a distance of 50 mm (2.0 in.) from
strength of HCB1A, HCB2A, HCB3A, HCB3, and EL600A
the edge of the bolted connection, respectively.
were 3.5, 2.4, 1.8, 3.5, and 4.3% for positive loading direc-
Test results showed that the web yielding of the speci-
tion and –3.4, –3.2, –1.9, –3.6, and –4.7% for negative direc-
mens with the opening generally occurred when the strength
tion, respectively. Here, the positive drift ratio is taken as
reached the peak value (Table 4). For HCB1A, HCB2A,
the drift ratio when the actuator pushes the steel coupling
and HCB3A, the web yielded at the drift ratios of 3.49%,
beam (refer to Fig. 3). Drift ratios of each specimen at the
2.48%, 3.18%, and –3.05%, –2.95%, and –2.49% for posi-
ultimate state ranged from 4.5 to 4.9% for the specimens
tive and negative loading directions, respectively. In the case

Fig. 7—Strain variations on web and flange of coupling beam. (Note: 1 kN = 0.225 kip.)
of HCB3, the web yielded at 3.92% drift ratio. On the other –5.08%, respectively. However, other specimens did not
hand, in EL600A, the web yielded when the failure strength experience flange yielding. The reason the flange yielding
(Vf) was reached. In this case, the web yielding was reported did not occur in other specimens is that the premature defor-
at the drift ratios of 5.3% and –4.7% for positive and nega- mation arose at the flange due to the shear distortion of the
tive loading, respectively. web opening before the flexural yielding.
The strain on the flange was measured by strain gauge Test results indicated that if the coupling beam had been
(BF5) attached to the surface of the flange at the distance designed as the shear-dominant member and the top-seat
of 250 mm (9.8 in.) from the beam end (Fig. 7(b)). Flange angles had been mounted, the shear yielding of the coupling
yielding occurred only in the HCB1A and EL600A spec- beam would have been induced before the flexural yielding.
imens and was observed at the drift ratios of –5.22% and In particular, when a coupling beam section with the shear

Fig. 8—Stiffness degradation. (Note: 1 kN/mm = 5.71 kip/in.)
capacity reduced by more than half of the full capacity is used,
shear failure of the beam can be induced more efficiently.
Stiffness
For a lateral load-resisting system subjected to cyclic loads,
stiffness is a crucial factor to describe the inelastic defor-
mations of the system. For the coupled shear wall system
studied in this paper, stiffness is highly dependent upon the
boundary condition at the bolted connection between the
coupling beam and the side face of the wall. According to
previous research (Lim et al. 2016, 2018), the stiffness of the
hybrid steel coupling beam with top-seat angles is suggested
to be 75% of the stiffness (= 0.6ke) of the monolithic coupling
beam system, where ke is the elastic stiffness obtained from
the effective moment inertia (Ie = 0.6Ix/[1 + 36(Ix/Awe2)]),
which was suggested by Harries et al. (2000). Here, Ix is the Fig. 9—Energy dissipation. (Note: 1 kN-m = 0.738 kip-ft.)
moment of inertia of the coupling beam section, Aw is the web dissipation was defined as the enclosed area of the load-
area of the coupling beam section, and e is the clear span (ln) displacement curve obtained from the third cycle for each
plus the concrete cover (cc). The clear span represents the target displacement (ACI T1.1-01 2001). Figure 9(a)
length of the coupling beam between both walls. compares the cumulative energy dissipation up to 4.5% drift
Figure 8 depicts secant stiffness degradation, which was ratio, which was obtained by adding the absorbed energy at
obtained using the strength and displacement at the first each target displacement.
cycle of each target drift ratio (ACI T1.1-01 2001). For Test results revealed that the cumulative energy dissipation
comparisons, only test data in the positive loading direction is highly dependent on the shear-to-moment capacity ratio of
obtained from 0.5 to 5.5% drift ratio were used. At a 0.5% the steel coupling beam as indicated through the compar-
drift ratio, the stiffnesses of HCB1A, HCB2A, and HCB3A ison with the load-drift ratio relationship in Fig. 5. For the
specimens were 65.3%, 77.1%, and 65.3% of the stiffness of specimens with top-seat angles, the energy dissipations were
EL600A, respectively. However, when top-seat angles were quite comparable up to the drift ratio of 2%, but after 2.5%
not provided in the hybrid coupling beam system, the initial drift ratio, the energy dissipation of HCB1A increased more
stiffness was approximately 43% of that of EL600A. This rapidly than that of the others. At 4.5% drift ratio, the cumu-
result indicates that top-seat angles play an important role in lative energy dissipation of HCB1A was approximately 23%
increasing the stiffness of the bolted hybrid coupling beam. and 38% higher than that of HCB2A and HCB3A, respec-
The stiffness of the specimen with top-seat angles was up to tively. On the other hand, HCB3 (without top-seat angles)
50% higher than that without angles (HCB3: 17.9 kN/mm showed low energy dissipation capacity. At 4.5% drift ratio,
[102.2 kip/in.]), even though the stiffness of the coupling beam the energy dissipation of HCB3 was approximately 30%
with web opening (HCB1A: 27.2 kN/mm [155.3 kip/in.], lower compared to HCB3A although both specimens had
HCB2A: 32.1 kN/mm [183.3 kip/in.], HCB3A: 27.2 kN/mm similar shear capacities.
[153.3 kip/in.] at 0.5% drift ratio) was much lower than the The effects of web opening and top-seat angles become
stiffness (= 0.6ke = 69 kN/mm [394 kip/in.]) proposed by more prominent in normalized energy dissipation presented
Harries et al. (2000). in Fig. 9(b), in which the cumulative energy dissipation of
each specimen was divided by that of EL600A. For the spec-
Cumulative energy dissipation capacity imens with the web opening and top-seat angles (HCB1A,
To evaluate the capability to absorb the seismic energy, HCB2A, and HCB3A), the maximum value of the normal-
the cumulative energy dissipation capacity of the coupled ized energy dissipation ranged from 29 to 34% of that of
shear wall system was calculated. Herein, the energy

reduced as low as 15% at 3% drift ratio and remained below
25% until the end of testing in Fig. 10.
The shear strengths of the coupling beams with bolted
connection were less than half of that of the monolithic
coupling beam but the hybrid systems with web opening
behaved like the monolithic structure. The presence of web
opening had an impact of improving the equivalent energy
dissipation of the steel coupling beam. Among the speci-
mens with web opening, the damping ratios tended to grow
as the diameter of opening increased. In addition, as was the
case in Fig. 9, the top-seat angle enabled the coupled shear
wall system to absorb more energy.
Fig. 10—Equivalent viscous damping ratio. DESIGN OF SHEAR-DOMINANT HYBRID STEEL
the specimen without opening (EL600A). In the case of the COUPLING BEAM WITH BOLTED CONNECTIONS
specimen without angles (HCB3), the maximum normalized Based on the test results, this study suggests the design
energy dissipation was less than 15% of that of EL600A. procedure of shear-dominant hybrid steel coupling beam
This result demonstrates that top-seat angles contributed system with bolted connections as follows: It is noted that
greatly to improving the energy dissipation capacity of the the following design procedure includes the experimental
coupled shear wall system. Both HCB3A and HCB3 had the and analytical results of the previous research (Mattock and
same diameter of web opening but the energy dissipation of Gaffar 1982; Lim et al. 2016, 2018).
HCB3A increased up to twice that of HCB3. Step 1: Determine the required shear strength (Vu) using
load combinations in accordance with ASCE 7-16 (2016).
Equivalent viscous damping ratio Step 2: Choose the steel coupling beam section that
To estimate the energy dissipation through the nonlinear complies with the width to thickness ratios recommended by
hysteretic behavior, equivalent viscous damping coefficient ANSI/AISC 341-16 (2016).
(ξeq) was calculated. Equivalent viscous damping ratio was Step 3: Determine the design shear strength (φvVn) of
defined as the ratio of the area of load hysteresis curve to the steel coupling beam for a section with a full web area,
the enclosed area of trapezoid connecting the target load assuming that this section remains elastic subjected to Vu. In
and displacement. fib (2003) suggests the equivalent viscous this study, the nominal shear strength (Vn) is assumed to be
damping ratio as follows equal to the plastic shear strength (Vp) of the coupling beam.
Therefore, the relation between the required shear strength
2 Aloop and design shear strength is as follows
eq (3)
Arect
Vu ≤ φvVn = φvV p (4)
where Aloop represents the area enclosed by the hysteresis
loop; and Arect is the area of the rectangle circumscribing the where ϕv is the strength reduction coefficient for shear (=0.9).
hysteresis loop. When a hot-rolled H-sectional beam is used for the
According to fib (2003), when a structural system exhibits steel coupling beam, the plastic shear strength of the steel
equivalent viscous damping ratio of up to 25%, its hyster- coupling beam can be obtained from Vp = 0.6FywtwH because
etic response is regarded as that of an equivalent monolithic the area of the web is obtained assuming the overall height
structure composed of non-prestressed members. Addition- of the beam section. On the other hand, when a built-up
ally, hybrid and rocking systems display equivalent viscous steel coupling beam is used, the plastic shear strength can be
damping ratios of up to 18%. obtained from Vp = 0.6Fywtwh using the net distance between
Figure 10 shows the equivalent viscous damping ratio of the flanges.
each specimen with respect to its drift ratios. In the early Step 4: Determine the design shear strength (ϕvVn) of
loading stage—for example, at 1% drift ratio—all speci- the steel coupling beam for a section with web opening.
mens showed similar damping ratios of approximately 30% Because the section is supposed to act as the structural fuse,
but the values deviated as the drift ratio increased. The spec- it should be designed to yield in shear (that is, ϕvVn = ϕvVpo ≤
imens with web opening (HCB1A, HCB2A, and HCB3A) Vu) with the plastic shear strength calculated from Vpo =
showed relatively high equivalent viscous damping ratios 0.6Fywtw(h – dw). The diameter (dw) of web opening can be
mostly above 25%. Especially, HCB3A showed the excellent chosen depending on the desired level of shear strength ratio
hysteretic damping capability, which increased up to 43% at (Vpo/Vp). Herein, for the sake of minimized wall damage
4% drift ratio. On the other hand, in HCB3, damping ratio and sufficient equivalent viscous damping capability, the
values showed a decreasing tendency and maintained below authors recommend to apply Vpo/Vp = 0.2 to 0.3 on the basis
25% at a large portion of the drift ratios. Although EL600A of the experimental results (that is, the cases of HCB2A
showed the exceptionally high cumulative energy dissipa- and HCB3A). Furthermore, the coupling beam with web
tion in Fig. 9(a), its equivalent damping ratio continuously opening as part of a partially coupled wall system is allowed

Fig. 11—Stress distributions on embedded steel beam.
to have relatively lower stiffness (50% or less) compared to coupled shear wall system. The web opening which operated
monolithic steel coupling beam (= 0.6ke). as a structural fuse induced early shear yielding of the web
Step 5: Determine the plastic moment strength (Mp) of the and allowed the system to exhibit the excellent deformation
coupling beam using the equation of Mp = FyZp, where Zp is and energy dissipation capacities with minimal damage in
the plastic section modulus of the beam. the concrete wall under reversed cyclic deformation.
Step 6: Determine the clear length (Lcs) of the steel coupling 2. As anticipated, top-seat angles substantially increased
beam under ANSI/AISC 341-16, where the coupling beam the lateral stiffness and deformation capacity of the coupled
with the length of Lcs ≤ 1.6Mp/Vpo can be considered to be shear wall system. When top-seat angles were provided,
shear-dominant. Here, the plastic section modulus for the the specimen showed much higher lateral stiffness (up to
reduced section by web opening is considered. 50%) compared to the one without angles. Furthermore, the
Step 7: Determine the embedment length (le) of the steel ductility of HCB3A is approximately four times higher than
beam as per the previous research (Mattock and Gaffar that of HCB3, although both specimens had the same web
1982; Lim et al. 2016). Figure 11 shows two cases of stress opening size.
distribution available in determining the embedment length. 3. All the specimens maintained considerable lateral resis-
The former (Case I) corresponds to the case that only tance, more than 80% of its peak strength at least, until the
minor cracks develop at the front part of the wall (that is, drift ratio of 4.5% was reached.
Vpo/Vp ≈ 0.2 to 0.5). In this case, the stress distribution can 4. The proposed steel coupling beams showed sufficient
be assumed such that only the concrete under the embedded energy dissipation capability, and it was highly dependent on
steel beam reached its ultimate state as shown in Fig. 11(a), shear-to-moment capacity ratio as well as the presence of the
and an analytical study proposed by Mattock and Gaffar top-seat angles. In terms of the equivalent viscous damping
(1982) can be adopted as a part of an iterative procedure. ratio, HCB3A with the smallest web opening showed the
In the latter case (Case II), on the other hand, provided that best performance among all the specimens, which was
severe cracks occur along the whole embedment length (that approximately 85% higher than that of EL600A at 4% drift
is, Vpo/Vp ≈ 0.5 to 1.0), the uniform stress distribution can ratio. This structural fuse coupling beam was effective in
be used assuming that the concrete has reached its ultimate minimizing the damage in the concrete wall, indicating that
limit state (refer to Fig. 11(b) and Lim et al. 2016). it can be simply replaced after substantial damage.
Step 8: Finally, determine amounts and details of rein- 5. A design procedure for the proposed steel coupling
forcing bars adjacent to embedded beam as per the previous beam system with and without web opening was proposed
research by Lim et al. (2018). based upon the test results and existing code provisions
and research results. The steel coupling beam with bolted
CONCLUSIONS connection is designed as a shear-dominant beam, and shear
In this study, a new type of hybrid steel coupling beam yielding of structural steel is preferable to ensure sufficient
with web opening was proposed, and cyclic loading tests ductility and prevent severe cracking in walls and resulting
for five half-scale, half-span coupling beams were carried stiffness degradation in the nonlinear range of deforma-
out. The primary test parameters were the shear-to-moment tions. Top-seat angles are also needed for more efficient
capacity ratio of the steel coupling beam controlled by the web yielding.
web opening and the presence of the top-seat angles. Find-
ings of the experimental study can be drawn as follows: AUTHOR BIOS
1. The shear-to-moment capacity ratio of the coupling Woo-Young Lim is an Associate Professor of Structural Engineering
at Wonkwang University, Iksan, Korea. He received his PhD from Seoul
beam significantly affected the seismic behavior of the

National University, Seoul, Korea, in 2012. His research interests include Darwin, D., and Lucas, W. K., 1990, “LRFD for Steel and Composite Beams
reinforced and precast concrete members in shear. with Web Openings,” Journal of Structural Engineering, ASCE, V. 116,
No. 6, pp. 1579-1593. doi: 10.1061/(ASCE)0733-9445(1990)116:6(1579)
Thomas H.-K. Kang, FACI, is a Professor of structural engineering and Farsi, A.; Keshavarzi, F.; Pouladi, P.; and Mirghaderi, R., 2016,
Director of Engineering Innovation Center at Seoul National University. “Experimental Study of a Replaceable Steel Coupling Beam with an
He is a member of Joint ACI-PTI Committee 320, Post-Tensioned Concrete End-plate Connection,” Journal of Constructional Steel Research, V. 122,
Code; Joint ACI-ASCE Committees 352, Joints and Connections in Mono- pp. 138-150. doi: 10.1016/j.jcsr.2016.03.018
lithic Concrete Structures, and 423, Prestressed Concrete; Joint ACI-ASME fib, “Seismic Design of Precast Concrete Building Structures,” Fédéra-
Committee 359, Concrete Containments for Nuclear Reactors; and ACI tion Internationale du Béton, Lausanne, Switzerland, 2003, 262 pp.
Subcommittee 318-T, Post-Tensioned Concrete (Structural Concrete Fortney, P. J.; Shahrooz, B. M.; and Rassati, G. A., 2007, “Large-Scale
Building Code). He received the ACI Wason Medal for Most Meritorious Testing of a Replaceable ‘Fuse’ Steel Coupling Beam,” Journal of Struc-
Paper in 2009. His research interests include the design and behavior of tural Engineering, ASCE, V. 133, No. 12, pp. 1801-1807. doi: 10.1061/
reinforced, prestressed, and precast concrete structures. (ASCE)0733-9445(2007)133:12(1801)
Galano, L., and Vignoli, A., 2000, “Seismic Behavior of Short Coupling
ACI member Donghyuk Jung is an Assistant Professor at Pusan National Beams with Different Reinforcement Layouts,” ACI Structural Journal,
University, Busan, Korea. Before that, he was a Postdoctoral Researcher at V. 97, No. 6, Nov.-Dec., pp. 876-885.
Seoul National University. He received his BS from Korea University, Seoul, Gong, B., and Shahrooz, B. M., 2001, “Steel-Concrete Composite
Korea; and his MS and PhD from the University of Illinois at Urbana- Coupling Beams-Behavior and Design,” Engineering Structures, V. 23,
Champaign, Champaign, IL. His research interests include the design of No. 11, pp. 1480-1490. doi: 10.1016/S0141-0296(01)00042-6
reinforced concrete structures using advanced materials such as shape Harries, K. A.; Fortney, P. J.; Shahrooz, B. M.; and Brienen, P. J., 2005,
memory alloys. “Practical Design of Diagonally Reinforced Concrete Coupling Beams—
Critical Review of ACI 318 Requirements,” ACI Structural Journal, V. 102,
ACI member Sung-Gul Hong is a Professor in the Department of Archi- No. 6, Nov.-Dec., pp. 876-882.
tecture & Architectural Engineering at Seoul National University. He Harries, K. A.; Gong, G.; and Shahrooz, B. M., 2000, “Behavior and
received his BS and MS from Seoul National University in 1981 and Design of Reinforced Concrete, Steel, and Steel-Concrete Coupling Beams,”
1983, respectively, and his PhD from Lehigh University, Bethlehem, PA, Earthquake Spectra, V. 16, No. 4, pp. 775-799. doi: 10.1193/1.1586139
in 1994. His research interests include strut-and-tie models for bond Harries, K. A.; Mitchell, D.; Cook, W. D.; and Redwood, R. G., 1993,
transfer, shear strength of reinforced concrete members, shear friction with “Seismic Response of Steel Beams Coupling Concrete Walls,” Journal
creep, anchorage of multiple bars, and deformation of reinforced concrete of Structural Engineering, ASCE, V. 119, No. 12, pp. 3611-3629. doi:
columns in shear. 10.1061/(ASCE)0733-9445(1993)119:12(3611)
Ji, X.; Liu, D.; Sun, Y.; and Molina Hutt, C., 2017a, “Seismic Perfor-
ACKNOWLEDGMENTS mance Assessment of a Hybrid Coupled Wall System with Replaceable
The work presented in this paper was sponsored by the National Steel Coupling Beams versus Traditional RC Coupling Beams,” Earth-
Research Foundation of Korea (NRF) grants (No. 2018R1C1B5045037 and quake Engineering & Structural Dynamics, V. 46, No. 4, pp. 517-535. doi:
2021R1A5A1032433). The options, findings, and conclusions in this paper 10.1002/eqe.2801
are those of the authors and do not necessarily represent those of the sponsors. Ji, X.; Wang, Y.; Ma, Q.; and Okazaki, T., 2017b, “Cyclic Behavior of
Replaceable Steel Coupling Beams,” Journal of Structural Engineering, ASCE,
V. 143, No. 2, p. 04016169. doi: 10.1061/(ASCE)ST.1943-541X.0001661
REFERENCES Kang, T. H.-K., and Wallace, J. W., 2008, “Seismic Performance of Rein-
ACI Innovation Task Group 1, “Acceptance Criteria for Moment Frames forced Concrete Slab-Column Connections with Thin Plate Stirrups,” ACI
Based on Structural Testing (ACI T1.1-01),” American Concrete Institute, Structural Journal, V. 105, No. 5, Sept.-Oct., pp. 617-625.
Farmington Hills, MI, 2001, 10 pp. Lam, W. Y.; Su, R. K.-L.; and Pam, H.-J., 2005, “Experimental Study
ACI Committee 318, “Building Code Requirements for Structural on Embedded Steel Plate Composite Coupling Beams,” Journal of Struc-
Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American tural Engineering, ASCE, V. 131, No. 8, pp. 1294-1302. doi: 10.1061/
Concrete Institute, Farmington Hills, MI, 2019, 623 pp. (ASCE)0733-9445(2005)131:8(1294)
AISC, “Steel Construction Manual,” 14th edition, American Institute of Lim, W. Y.; Kang, T. H.-K.; and Hong, S. G., 2016, “Cyclic Testing of
Steel Construction, Chicago, IL, 2011, 2192 pp. Bolted Steel Coupling Beams in Fast-Track Precast Concrete Construc-
ANSI/AISC 341-16, “Seismic Provisions for Structural Steel Build- tion,” ACI Structural Journal, V. 113, No. 6, Nov.-Dec., pp. 1289-1300.
ings,” American Institute of Steel Construction, Chicago, IL, 2016, 430 pp. doi: 10.14359/51689250
ANSI/AISC 360-16, “Specification for Structural Steel Buildings,” Lim, W. Y.; Kang, T. H.-K.; and Hong, S. G., 2018, “Effect of Reinforce-
American Institute of Steel Construction, Chicago, IL, 2016, 620 pp. ment Details on Seismic Behavior of Precast Concrete Wall-Steel Coupling
ASCE, “Recommendations for Seismic Design of Hybrid Coupled Wall Beam Systems,” ACI Structural Journal, V. 115, No. 6, Nov., pp. 1751-
Systems,” American Society of Civil Engineers, Reston, VA, 2010, 70 pp. 1763. doi: 10.14359/51702414
ASCE/SEI 7-16, “Minimum Design Loads and Associated Criteria for Marcakis, A. H., and Mitchell, D., 1980, “Precast Concrete Connections
Buildings and Other Structures,” American Society of Civil Engineers, with Embedded Steel Member,” PCI Journal, V. 25, No. 4, pp. 88-116. doi:
Reston, VA, 2016. 10.15554/pcij.07011980.88.116
ASTM C39/C39M-18, “Standard Test Method for Compressive Strength Mattock, A. H., and Gaffar, G. H., 1982, “Strength of Embedded Steel
of Cylindrical Concrete Specimens,” ASTM International, West Consho- Sections as Brackets,” ACI Journal Proceedings, V. 79, No. 2, Feb.,
hocken, PA, 2018. pp. 83-93.
ASTM E8/E8M-16a, “Standard Test Methods for Tension Testing of Naish, D.; Fry, A.; Klemencic, R.; and Wallace, J., 2013, “Reinforced
Metallic Materials,” ASTM International, West Conshohocken, PA, 2016. Concrete Coupling Beams—Part I: Testing,” ACI Structural Journal,
Barbachyn, S. M.; Kurama, Y. C.; and Novak, L. C., 2012, “Analyt- V. 110, No. 6, Nov.-Dec., pp. 1057-1066.
ical Evaluation of Diagonally Reinforced Concrete Coupling Beams Park, W. S., and Yun, H. D., 2005, “Seismic Behaviour of Coupling
under Lateral Loads,” ACI Structural Journal, V. 109, No. 4, July-Aug., Beams in a Hybrid Coupled Shear Walls,” Journal of Constructional Steel
pp. 497-508. Research, V. 61, No. 11, pp. 1492-1524. doi: 10.1016/j.jcsr.2005.04.006
Canbolat, B. A.; Parra-Montesinos, G. J.; and Wight, J. K., 2005, “Exper- Paulay, T., and Binney, J. R., 1974, “Diagonally Reinforced Coupling
imental Study on Seismic Behavior of High-Performance Fiber-Reinforced Beams of Shear Walls,” Shear in Reinforced Concrete, SP-42, American
Cement Composite Coupling Beams,” ACI Structural Journal, V. 102, Concrete Institute, Farmington Hills, MI, pp. 579-598.
No. 1, Jan.-Feb., pp. 159-166. Paulay, T., and Priestley, M. J. N., 1992, Seismic Design of Reinforced
Chaallal, O.; Gauthier, D.; and Malenfant, P., 1996, “Classifica- Concrete and Masonry Structures, John Wiley & Sons, NY.
tion Methodology for Coupled Shear Walls,” Journal of Structural Precast/Prestressed Concrete Institute, 1999, PCI Design Handbook:
Engineering, ASCE, V. 122, No. 12, pp. 1453-1458. doi: 10.1061/ Precast and Prestressed Concrete, 5th Edition, Chicago, IL.
(ASCE)0733-9445(1996)122:12(1453) Shahrooz, B. M.; Remmetter, M. E.; and Qin, F., 1993, “Seismic
Clarke, J. L., and Symons, R. M., “Tests on Embedded Steel Billets Design and Performance of Composite Coupled Walls,” Journal of Struc-
for Precast Concrete Beam-Column Connections,” Technical Report tural Engineering, ASCE, V. 119, No. 11, pp. 3291-3309. doi: 10.1061/
No. 42.523, Cement and Concrete Association, London, UK, 1978, 12 pp. (ASCE)0733-9445(1993)119:11(3291)
CSA A23.3:2019, “Design of Concrete Structures,” Canadian Standards Tassios, T. P.; Moretti, M.; and Bezas, A., 1996, “On the Behavior and
Association, Mississauga, ON, Canada, 2019. Ductility of Reinforced Concrete Coupling Beams of Shear Walls,” ACI
Structural Journal, V. 96, No. 6, Nov.-Dec., pp. 711-720.

Title No. 119-S45
Glass Fiber-Reinforced Polymer-Reinforced Beam-Column

Connections with Novel Connection Details
by Muhammad Safdar, M. Neaz Sheikh, and Muhammad N. S. Hadi
This paper proposes two novel reinforcement details for glass and shear reinforcement in BCCs under cyclic loading was
fiber-reinforced polymer (GFRP) reinforced concrete (RC) exte- investigated in a few research studies.13-17 These studies
rior beam-column connections. Four RC exterior beam-column concluded that FRP-RC BCCs are capable of sustaining load
connections with different connection details (SH-bend, GH-bend, reversals without degradation of the tensile strength of FRP
Z-bend, and U-bend) were tested under reversed cyclic loading.
bars. The plastic deformation of the FRP-RC BCCs was low
The two conventional connection details (SH-bend and GH-bend)
due to the linear elastic behavior of FRP bars. Moreover,
comprised three 90-degree hooked bar anchorages for longitudinal
reinforcement of the beam and four rectangular stirrups (steel or the FRP-RC BCCs provided sufficient strength, ductility,
GFRP) within the connection region. In the two novel connection and energy dissipation before the failure of the connection.
details (Z-bend and U-bend) proposed in this study, U-shaped bars The current design codes of FRP-RC BCCs such as CSA
were used as anchorage at the end of the longitudinal reinforce- S806-12(R2017)18 and ACI 440.1R-1519 are based on strong
ment of the beam. The Z-bend detail contained four extra Z-shaped column-weak beam concepts, which ensures a balance
bars and one rectangular stirrup, and the U-bend detail contained between strength and ductility of BCCs by allowing the flex-
four additional U-bars and four rectangular stirrups at the connec- ural failure due to the crushing of concrete at the compres-
tion. Experimental results revealed that the U-bend detail achieved sion face of the beam before other failure mechanisms.
higher load capacity, ductility, and energy dissipation than those of This design approach prevents brittle tension failure due to
the Z-bend and conventional connection details.
the rupture of FRP bars in the beam and helps to achieve
Keywords: beam-column connection; cyclic performance; diagonal bars; adequate performances of FRP-RC BCCs. Due to the lack
drift ratio; glass fiber-reinforced polymer (GFRP) bars; reinforced concrete. of research data, the design codes18,19 of FRP-RC structures
do not provide complete guidelines for the design of BCCs
INTRODUCTION under cyclic loading. For example, flexural strength ratio,
Harsh exposure conditions adversely affect the dura- design connection shear stress, and connection reinforce-
bility of reinforced concrete (RC) structures and require ment detail are critical parameters in the design of exterior
repairing and rehabilitation during the service life of RC BCCs. However, design provisions on connection reinforce-
structures. One such harsh exposure condition is chloride ment detail have not been addressed in current design codes
attack in marine environments, which initiates the corrosion of FRP-RC BCCs such as CSA S806-12(R2017)18 and ACI
of steel reinforcement in RC structures.1-3 As a result, the 440.1R-15.19 Therefore, it is important to investigate the
service life of RC structures is reduced due to the loss of effects of connection reinforcement detail on the strength and
structural capacity and durability.4,5 In this regard, sustain- ductility of FRP-RC BCCs under cyclic loading to develop
able RC structures are designed to minimize maintenance design guidelines for FRP-RC beam-column connections.
costs by replacing steel reinforcement with fiber-reinforced The major factors influencing the cyclic behavior of
polymer (FRP) bars.6 This replacement eliminates the FRP-RC beam-column connections are confinement, design
corrosion problem in the harsh environment owing to the connection shear stress, axial load level on the column,
strong resistance of FRP bars against chloride and chemical concrete strength, anchorage type at the end of the longitu-
attacks.6,7 FRP bars are anisotropic in nature, having linear dinal bar of the beam, and connection details. The confine-
elastic behavior with low modulus of elasticity, high tensile ment effect of adjoining beams on the behavior of exterior
strength, and low compressive strength. Previous studies8-12 GFRP-RC BCCs under reversed cyclic loading was studied
on RC members reinforced with FRP bars revealed that glass by Ghomi and El-Salakaway20 and found that confined
fiber-reinforced polymer (GFRP) bar reinforced members connection could resist connection shear stresses equal to
achieved adequate strength and ductility under monotonic 1.1√fc′ and sustain 6.0% drift ratio. Hasaballa and El-Sal-
loading. However, the application of FRP bars as internal akawy21 tested six exterior GFRP-RC BCCs by varying the
reinforcement in moment-resisting frames (MRFs) demands design connection shear stress level and concrete strength.
investigation of the performance of beam-column connec- The BCCs tested in Hasaballa and El-Salakawy21 were
tions (BCCs) under cyclic loading.
The performance of BCCs of RC members reinforced MS No. S-2021-146.R, doi: 10.14359/51734342, received August 22, 2021, and
with FRP bars (FRP-RC) is an important design consider- reviewed under Institute publication policies. Copyright © 2022, American Concrete
ation to ensure the structural integrity of MRFs under cyclic obtained from the copyright proprietors. Pertinent discussion including author’s
loading. The performance of using FRP bars as flexural closure, if any, will be published ten months from this journal’s date if the discussion

designed with concrete of compressive strength of 30 and connection details capable of improving the ductility and
60 MPa (4.35 and 8.70 ksi) and connection shear strength energy dissipation of FRP-RC BCCs. This study proposes
of 0.7√fc′, 0.85√fc′, and 1.0√fc′. To avoid connection shear and investigates the performance of the proposed connection
failure, it was recommended that the connection shear stress details for GFRP-RC BCCs under cyclic loading.
should not exceed 0.85√fc′. In another study, Hasaballa and
El-Salakawy22 examined the influence of end anchorage EXPERIMENTAL OUTLINE
shape (90-degree-hooked and headed bars) and surface Specimen configuration
condition (sand coated and deformed) of GFRP bars on the Four full-scale specimens representing RC exterior BCCs
cyclic performance of exterior BCCs. The study found that of MRFs were constructed and tested under reversed cyclic
both anchorage types were able to achieve 4.0% drift ratio loading. The cross-sectional dimensions of the beam and
safely and surface texture had an insignificant effect on the column were the same in all the specimens, as shown in
energy dissipation and deformability of the BCCs. Fig. 1. The column height and beam length were selected
Most of the past tests on FRP-RC BCCs under cyclic from a previous research study on steel-RC exterior BCCs by
loading were conducted by adopting the conventional Hadi and Tran30 because the same test setup was used in the
connection reinforcement details of BCCs. In practice, a current study. The column height was 2800 mm (110.2 in.)
large quantity of stirrups for the confinement of connection and beam length was1400 mm (55.1 in) from the face of the
and 90-degree hooks at the end of the beam longitudinal column. The cross section of both beam and column was
bars are used in conventional connection reinforcement rectangular, having a width of 250 mm (9.8 in.) and a depth
details. As a result, reinforcement congestion and improper of 350 mm (13.8 in.).
concrete compaction problems can arise due to the limited The first specimen was used as a control specimen and
space available at the connection core, leading to inadequate was reinforced with steel bars in the longitudinal and trans-
performance of BCCs. On the other hand, a few research verse directions of beam and column. Six N16 (deformed
studies23-27 on steel-RC BCCs investigated various connec- steel bars of nominal diameter 16 mm [0.63 in.]) bars were
tion reinforcement details by using the combination of hori- placed in the longitudinal direction of the column. The trans-
zontal stirrups and diagonal bars to overcome the problem verse reinforcement of the column was made of R10 (plain
of reinforcement congestion. These studies concluded that steel bar of nominal diameter 10 mm [0.39 in.]) ties, which
diagonal bars enhanced the strength, ductility, and energy were distributed symmetrically about the longitudinal axis
absorption of BCCs. Kotsovou and Mouzakis28 found that of the beam. The first tie was placed at 50 mm (1.97 in.) from
the replacement of the horizontal stirrups up to 80% with the face of the beam and continued at 60 mm (2.4 in.) inter-
the diagonal bars reduced the connection distortions of vals for 470 mm (18.5 in), at 120 mm (4.7 in.) intervals for
steel-RC BCCs. Rajagopal and Prabavath29 examined the the next 360 mm (14.2 in.), and at 60 mm (2.4 in.) intervals
cyclic performance of different connection details by incor- for the remaining 395 mm (15.6 in.). Three N16 bars were
porating hair clips and diagonal bars. The results indicated used as flexural reinforcement at both the top and bottom of
that the inclusion of hair clips and diagonal bars enhanced the beam. The 90-degree hooks were provided at the end of
the shear strength, ductility, and stiffness of steel-RC BCCs. each N16 bar. The R10 stirrups were distributed along the
However, the effectiveness of using diagonal bars on the length of the beam according to CSA A23.3-19.31 The first
cyclic behavior of FRP-RC BCCs has not been investigated stirrup was located at 50 mm (1.97 in.) from the face of the
yet. column. The spacing of stirrups was kept at 75 mm (3 in.)
In the present study, two innovative connection details centers in the plastic hinge zone for 650 mm (25.6 in.) from
using the diagonal bar and connection stirrups have been the face of the column, 150 mm (5.9 in.) centers for the next
proposed for GFRP-RC exterior BCCs. The effectiveness of 450 mm (17.7 in.), and 83 mm (3.3 in.) centers for the last
different reinforcement configurations was investigated by 300 mm (11.8 in.). The reinforcement details of the control
testing four BCCs under reversed cyclic loading. The cyclic specimen are shown in Fig. 1(a).
performance of GFRP-RC BCCs was evaluated using the The remaining three specimens were reinforced with GFRP
experimental results in terms of ductility, stiffness degrada- reinforcing bars in the longitudinal and transverse direc-
tion, energy dissipation, and drift contributions. The scope tions. Six G18 (ribbed GFRP bar of 17.2 mm [0.67 in.] root
of this study is to evaluate the influence of connection details diameter32) and G12 (ribbed GFRP bar of 11 mm [0.43 in.]
on the performance GFRP-RC BCCs under cyclic loading. root diameter32) bars were used as the longitudinal and trans-
verse reinforcement of the column, respectively. Three 12M
RESEARCH SIGNIFCANCE (No. 4 sand-coated GFRP bar of 13.6 mm [0.54 in.] effective
Exterior BCCs in MRFs are critical regions due to unsym- diameter33) bars were placed as flexural reinforcement at the
metrical loading and availability of limited length for the top and bottom of the beam, while G12 stirrups were used as
anchorage of longitudinal bars of the beam at the connection. shear reinforcement. The transverse reinforcement spacing
Connection detail with the conventional 90-degree hooked in beam and column was kept the same as that of the control
bar anchorage in combination with horizontal stirrups often specimen, as shown in Fig. 1(b).
causes reinforcement congestion. Moreover, inadequate The specimens were identified as SS-H-CD0, GG-H-
anchorage length of longitudinal bars of the beam leads CD0, GG-U-CD1, and GG-U-CD2 based on the type of
to a sudden drop of strength, ductility, and energy dissipa- reinforcement and connection details. The SS and GG repre-
tion under cyclic loading. Hence, it is important to develop sent the steel and GFRP reinforced specimens, respectively.

Fig. 1—Beam-column connection details: (a) reinforcement for steel specimen; (b) reinforcement for all GFRP specimens; (c)
specimen SS-H-CD0 and GG-H-CD0; (d) specimen GG-U-CD1; and (e) specimen GG-U-CD2.
The connection details of the specimens were represented connection details are shown in Fig. 2. All bent bars were
by H-CD0, U-CD1, and U-CD2, as displayed in Fig. 1(c), specially manufactured.33
(d), and (e), respectively. The H-CD0 was referred to as
the conventional connection detail, which constituted a Specimen design
90-degree hook (H) for the anchorage of the longitudinal The experimental program of this study was designed
bars of the beam and 4G12 horizontal stirrups at the spacing based on the design compressive strength of concrete of
of 70 mm (2.75 in.) in the connection region. The U-CD1 32 MPa (4.64 ksi) at 28 days. For steel-RC specimens, the
and U-CD2 details comprised three U-shaped (U) anchorage nominal yield strengths for N12 and R12 bars were 500 and
by connecting the top and bottom reinforcement of the beam. 250 MPa (72.5 and 36.3 ksi), respectively. For GFRP-RC
The U-CD1 detail contained four extra Z-shaped (4-12M) specimens, the guaranteed ultimate tensile strengths for G18,
bars in the diagonal directions of the connection region, 12M and G12 were 891, 1100, and 904 MPa (129.2, 159.5,
which were extended by 500 mm (19.7 in.) into the upper and 131.1 ksi), respectively. The specimens were designed
and lower column. One G12 horizontal stirrup was placed based on the strong column-weak beam concept according
at the midheight of the connection to examine the possibility to CSA A23.3-1931 and CSA S806-12(R2017).18
of replacing the horizontal stirrups with the Z-bars. The Table 1 presents the design properties of the specimens
U-CD2 detail contained G12 horizontal stirrups spaced at based on measured material properties during the testing
70 mm (2.75 in.) centers with 2-12M diagonal bars (U-bend- period. Flexural capacities of the beams of each specimen
135) from the bottom of the beam and extended to the outer were calculated according to CSA A23.3-1931 and CSA
top face of the column and vice versa. The photos of the S806-12(R2017).18 The flexural capacity of the beam of

Fig. 2—Photos of connection details.
Table 1—Design properties of specimens
Specimen ID
Design property SS-H-CD0 GG-H-CD0 GG-U-CD1 GG-U-CD2
Average compressive strength (fcʹ), MPa 51.1 51.1 55.0 55.0
ρf/ffb* 0.21 1.37 1.30 2.20
Beam flexural capacity (Mnb), kN.m 88.8 118.7 122.8 130.5
Column flexural capacity (Mnc), kN.m 126.5 132.0 137.0 137.0
Flexural strength ratio (FR) 2.85 2.22 2.23 2.10
Maximum connection shear stress (τcs,max), MPa 2.90 3.96 4.08 4.85
*
Ratio between provided to balance reinforcement ratio.
Note: All design properties are based on measured material properties during testing period; 1 MPa = 0.145 ksi; 1 kN.m = 8.851 kip.in.
Specimen SS-H-CD0 was 88.8 kN.mm (786 kip.in). For the Flexural strength ratio (FR) is an important design param-
GFRP-reinforced specimens, the ratio between the tensile eter, which is the ratio of the sum of the flexural capacities
reinforcement ratios (ρf) and the balanced reinforcement of column sections to the sum of the flexural capacity of the
ratio (ρfb) was greater than 1 to avoid brittle tension failure beam section adjoining at a connection. To satisfy the strong
of the beam. The flexural capacities of Specimens GG-H- column and weak beam concept for the design of connec-
CD0, GG-U-CD1, and GG-U-CD2 were 118.7, 122.8, and tions, the CSA S806-12(R2017)18 specifies that FR ≥ 1.0.
130.5 kN.m (1050.6, 1086.9, and 1155.1 kip.in), respec- In the design of specimens, the flexural strength ratio was
tively. Specimen GG-U-CD2 had additional diagonal bars kept at approximately 2.0 for steel and GFRP specimens.
from column to the beam at the interface (Fig. 1(e)) that leads However, the flexural strength ratio of steel and GFRP
to higher moment capacity. The longitudinal reinforcement reinforced specimens was 2.9 and 2.2, respectively, using
ratio of the GFRP column was kept greater than the balance the measured material properties during the testing period.
reinforcement ratio to prevent the rupture of the bars. The Maximum connection shear stresses (τcs,max) were obtained
axial force-bending moment (Pc-Mc) curve of GFRP column by dividing the maximum horizontal connection shear force
section was computed according to the procedure presented (Vcs,max) with the effective connection area (Ac). The Vcs,max
in previous studies.34-37 To compute a point on the Pc-Mc was calculated from Eq. (1)
curve, the depth of the neutral axis is assumed to get the
strain distribution over the cross-section of the column. Then  L L + 0.5hc 
Vcs , max = Pnb  b − b  (1)
tensile and compressive forces are found by using the stress-  Jb Hc
strain relations. The axial load and moment are obtained by
applying the equilibrium equations. The depth of the neutral where Pnb is the nominal design load of the beam; Lb is the
axis is varied to get the different points of the axial force- length of beam from the face of the column to the loading
bending moment interaction curve for the column section. point; hc is the depth of column; Jb is the moment arm
The flexural capacity of the column was calculated from between tensile and compressive forces of the beam section;
the axial force-bending moment interaction diagram corre- and Hc is the height of column between the lateral supports.
sponding to the applied axial load of 280 kN (63 kip).

Fig. 3—Preparation of specimen: (a) formwork; and (b) cast specimen.
Table 2—Tensile properties of reinforcing bars
Bar designation
Material property R10 N16 G12 G18 12M
Bar type and surface condition Steel-plain Steel-deformed GFRP-deformed GFRP-deformed GFRP-sand coated
Nominal diameter (db), mm 10.0 16.0 11.0 17.2 13.6
Nominal bar area (Ab), mm 2
78.5 201.0 95.0 232.4 129.0
Elastic modulus (E), GPa 197.8 182.8 59.0 58.6 56.8
Yield strength (fy), MPa 320 494 — — —
Yield strain (εy), % 0.162 0.270 — — —
Ultimate tensile strength (fu), MPa 482 565 930 1246 1256
Ultimate strain (εu), % 20.3 14.2 1.6 2.1 2.2
Note: 1 mm = 0.039 in.; 1 mm = 0.0155 in. ; 1 GPa = 145 ksi; 1 MPa = 0.145 ksi.
2 2
Specimen preparation of concrete for Specimens SS-H-CD0, GG-H-CD0, GG-U-

The reinforcement cages of all specimens were assem- CD1, and GG-U-CD2 were 35.2, 35.2, 42.2, and 42.2 MPa
bled in a horizontal plane. Two formworks were made with (5.1, 5.1, 6.1, and 6.1 ksi), respectively. During the testing
plywood and were joined by timber strips using screws. The of Specimens SS-H-CD0 and GG-H-CD0, the compressive
T-shaped steel hooks were attached to the reinforcement strength of the concrete was 51.1 MPa (7.4 ksi), while the
cages at different locations to maintain the concrete cover of compressive strength of the concrete of Specimens GG-U-
20 mm. The reinforcement cage of the specimen placed into CD0 and GG-U-CD1 was 55 MPa (8.0 ksi). Table 1 reports
the formwork before casting is shown in Fig. 3(a). Ready the average compressive strength of the tested concrete
mix normal-strength concrete was placed into the formworks cylinders on the first and last day of testing.
and was compacted using a vibrator. The surface of the spec- The mechanical properties of steel bars (R10 and N16)
imens was leveled with a trowel. The finished surface of a were obtained by performing tension tests on three samples
cast specimen is shown in Fig. 3(b). The curing of the spec- of each bar in accordance with AS 1391-07(17).39 The
imens was done by covering the surface of the specimen average yield tensile strengths of R10 and N16 were 320
with soaked burlaps. The specimens were demolded from and 494 MPa (46.4 and 71.7 ksi), respectively. The average
the formworks after 2 weeks and stored in the lab. Speci- elastic modulus of R10 and N16 was 197.8 and 182.8 GPa
mens GG-U-CD1 and GG-U-CD2 were cast using the same (28,688 and 26,513 ksi), respectively. The tensile properties
formworks. of G18 and 12M bars were obtained experimentally using
the procedure given in ASTM D7205.40 The tensile proper-
Material properties ties of G12 (stirrup) were provided by their manufacturer.
Ready mix normal-strength concrete with a maximum The ultimate tensile strengths of G18, G12, and 12M were
aggregate size of 10 mm (0.39 in.) was used in the casting 1246, 930, and 1256 MPa (181, 135, and 182 ksi), respec-
of specimens. The compressive strength of concrete (fcʹ) was tively. Table 2 lists the tensile properties of reinforcing bars
measured at 7 days, 28 days, the first day of testing, and the used in this study.
last day of testing. The compression test on three concrete
cylinders of size 100 mm (3.93 in.) diameter and 200 mm Test setup and instrumentation
(7.87 in.) height was conducted in accordance with AS The specimens were tested under reversed cyclic loading
1012.9-14.38 The measured 28-day compressive strengths in the vertical position using a testing frame, as shown in

Fig. 4—Test setup: (a) schematic drawing; and (b) photo of test setup before testing.
Fig. 4. The testing frame constituted a supporting frame and section at 50 mm (1.95 in.) from the face of the column and
a cantilever beam. The supporting frame was made of four the other LVDT was attached to the column surface. Simi-
vertical steel I-section, which were connected at the top and larly, two LVDTs were fixed at the bottom face of the beam,
bottom with strong supporting steel beds. The supporting spaced 405 mm (15.79 in.) from the top face LVDTs. Two
frame was used to hold the specimen in the vertical posi- LVDTs were attached at the back face of the column; each
tion and resist the vertical and lateral loads. Two supporting LVDT was located at 300 mm (11.70 in.) from the center of
plates were fabricated and provided at the top and bottom the beam. The LVDTs worked in pairs to measure the plastic
end of the column. The column ends were constrained hori- hinge rotations, beam rotations relative to the column, and
zontally with two roller supports, which were 1100 mm column rotations. Rotations are calculated by dividing the
(43.3 in.) from the longitudinal axis of the beam. The combi- difference in the readings of LVDTs by the distance between
nation of horizontal and vertical roller supports produced the the LVDTs.
hinge mechanism to simulate the point of inflections. The The connection distortion was measured by mounting the
strong cantilever beam was fabricated from I-sections and two LVDTs along the diagonals of a square (150 x 150 mm
was connected to the top of the vertical supporting frame. [5.9 x 5.9 in.]), formed at the center of the connection,
A hydraulic cylinder was used to apply the constant axial as shown in Fig. 6(d). The connection distortion (γ) was
load on the top of the column, as shown in Fig. 4(b). The computed by measuring the elongation and shortening of the
cantilever beam resisted the constant axial load produced by connection diagonals using the readings of diagonal LVDTs
the hydraulic cylinder. Two loading plates were placed at the with Eq. (2)
loading points of the beam to distribute the load uniformly.
A hydraulic actuator with a capacity of 600 kN (135 kip) was δ1 − δ 2
used to apply cyclic loading at 1200 mm (46.8 in.) from the γ= ( tan α + cot α ) (2)
l
face of the column.
where δ1 and δ2 denote elongation and shortening in the
The strain evolutions at critical locations of longitudinal
length of the diagonals, respectively; l is the initial length of
and transverse reinforcement were monitored using strain
the diagonal; and α is the inclination of the diagonal LVDT
gauges, as displayed in Fig. 5. The connection distortion and
with the horizontal.
rotations of beam and column were measured using eight
linear variable displacement transducers (LVDTs) and three
Test procedure
inclinometers, as shown in Fig. 6. One inclinometer was
At the start of testing, an axial load (Pc) of 0.063Ag fc′ (Ag
attached at the beam-connection interface and the other two
is the gross area of the column section) was applied onto
were attached at the interface of column and connection.
the column, as adopted in previous studies,41-44 to simulate
Two LVDTs were placed horizontally at the top face of the
the gravity load. During testing, Pc fluctuated in the range
beam. One LVDT was attached to the beam surface by angle
of 280 kN (63 kip) ± 5%. The cyclic loading pattern of all

Fig. 5—Location of strain gauges for: (a) specimen SS-H-CD0 and GG-H-CD0; (b) specimen GG-U-CD1; and (c) specimen
GG-U-CD2.
Fig. 6—Instrumentation: (a) inclinometers and LVDTs for measuring beam relative rotation; (b) top view of LVDT attachment
at beam surface; (c) LVDT set for measuring column rotation; and (d) LVDT for measuring connection distortion.
specimens consisted of two phases of loading, as adopted in reinforced specimens was calculated by considering their
a few research studies.45-47 In the first phase, a single cycle serviceability limit states. According to CSA A23.3-19,31
of estimated cracking and service loads was applied at 200 the serviceability limit of the steel-reinforced specimen is
mm from the free end of the beam under load control, as assumed to be reached when the tensile strain of the longitu-
shown in Fig. 7(a). The cracking load cycle amplitude for dinal bar is 60% of the yield strain. The concrete strain limit
all the specimens was 15 kN (3.4 kip), which was calculated of 0.001 at the top compression fiber was considered as the
using the transformed section properties of the beam. The serviceability limit for GFRP specimens.48 The loading rate
service loading cycle amplitude for GFRP and steel speci- was 5 kN/min (1.1 kip/min). The aim of the first phase was
mens was 25 and 35 kN (5.6 and 7.9 kip), respectively. The to predict crack initiation load and check the proper working
amplitude of the service loading cycle of steel and GFRP of the instruments.

Fig. 7—Cyclic loading pattern: (a) load-controlled phase (phase-1); and (b) displacement-controlled phase (phase-2).
The second phase of beam loading was constituted of the research studies conducted by Mady et al.16 and Hasaballa
reversed cyclic load under displacement control. The ACI and El-Salakawy.17 The response of all the specimens turned
374.1-05(19)49 report recommendations on the acceptance nonlinear after reaching the peak load due to the formation
criteria for moment frames based on the structural testing of the plastic hinge, cracking, and slippage of bars.
was adopted to select cyclic loading patterns. The cyclic The hysteretic curve of control Specimen SS-H-CD0 is
load started with a low-level drift ratio (within the elastic shown in Fig. 8(a). The first flexural crack initiated around
range) and increased in subsequent cycles until the failure 14 kN (3.2 kip) and propagated in the beam and the beam-
of the specimen. The displacement and drift levels selected column connection as the drift ratio was increased during the
in the loading sequence of Phase two are shown in Fig. 7(b). experiment. The yielding of tensile reinforcing bar started at
The drift ratio (θ) is computed by dividing the displacement an approximate load of 60 kN and the maximum load of 81 kN
level (Δ) by the distance from the loading point to the center (18.2 kip) was reached at the first cycle of 2.33% drift ratio
of the column. Three cycles of constant displacement were (32 mm [1.3 in] displacement level). Then, the load started to
applied at the quasi-static rate of 0.01 Hz for each drift ratio. reduce and the pinching length increased in subsequent drift
The displacement level was increased from 10 to 90 mm (0.4 ratios until the failure of the specimen. At 5.09% drift ratio, a
to 3.5 in.) such that the increase in displacement should be sudden drop in load was observed during the second cycle of
in the range of 1.25 to 1.5 times of the previous displace- loading in the pulling direction due to the concrete crushing
ment. One cycle of service loading was applied after four and rupture of the longitudinal tensile bars.
loading steps to estimate the stiffness degradation due to The hysteretic behavior of Specimen GG-H-CD0 is shown
cyclic loading. in Fig. 8(b). The cracking of the specimen started at 10 kN
(2.3 kip) and then there was a sudden change in the slope
EXPERIMENTAL RESULTS AND DISCUSSIONS of the hysteretic curve. The slope then remained constant
Hysteretic response until reaching the maximum load of 78 kN (17.5 kip) at
The relationships between load and the beam tip displace- 5.09% drift ratio (70 mm [2.6 in.] beam tip displacement).
ment for all the specimens are shown Fig. 8. In general, Residual deformations of the specimens could be estimated
hysteresis curves indicated the linear-elastic response at low by analyzing the slope (stiffness) of the unloading curve in
displacement levels followed by the degradation of strength a hysteretic loop at each drift ratio. The hysteresis loops of
and stiffness with increasing beam tip displacements. The Specimen GG-H-CD0 indicated lower residual deforma-
strength of all the specimens in the pulling (negative) directions to that of Specimen SS-H-CD0 but with the absence
tion of loading was higher than the strength of the pushing of pinching effect and less area under the loops. This was
(positive) direction of loading. Under reversed cyclic attributed to the linear elastic behavior of the GFRP bars.
loading, a concrete section of the beam is subjected to repet- The specimen sustained only the first cycle of 6.55% drift
itive tensile-compressive stress cycles that caused the degra- ratio (90 mm [3.5 in.]) in the pushing direction and failed in
dation in the tensile strength of concrete due to the prop- the pulling direction. At the beam tip displacement of 80 mm
agation cracks at the top or bottom face of the beam. The (3.2 in.), the load significantly dropped to 32 kN (7.2 kip)
strength of specimens in pushing (positive) direction loading due to concrete crushing and excessive opening and closing
reduced due to the first formation of cracks at the top face of of cracks in the virtual plastic hinge region of the beam.
the beam. The strength decreased in the second and the third The hysteretic curves of Specimens GG-U-CD1 and
cycles compared to the strength of the first cycle due to the GG-U-CD2 exhibited similar behavior to that of Specimen
bridging stress degradation under reversed cyclic loading. GG-H-CD0, as shown in Fig. 8(c) and (d). The crack initi-
The hysteretic response of all specimens was consistent with ation loads of Specimens GG-U-CD1 and GG-U-CD2 were
the cyclic behavior of GFRP-RC BCCs observed in previous approximately 11 and 12 kN (2.5 and 2.7 kip), respectively.

Fig. 8—Hysteresis curves: (a) SS-H-CD0; (b) GG-H-CD0; (c) GG-U-CD1; and (d) GG-U-CD2.
Specimen GG-U-CD1 reached its maximum load of 86 kN summary of the experimental results and observations made
(19.3 kip) at 5.09% drift ratio and failed at a drift ratio of during the test. The cracking (Pcr) and maximum loads (Pmax)
6.1% (84 mm [3.3 in.]) due to the anchorage failure of the were noted from the experimental observations.
longitudinal bars. On the other hand, the maximum load of
Specimen GG-U-CD2 was approximately 103 kN (23.1 kip) Cracking pattern and failure modes
at a drift ratio of 6.55% (90 mm [3.5 in.]) without any signif- The cracking pattern and damage level of all the speci-
icant drop of the load. The increase in the maximum load mens at ultimate drift ratios are displayed in Fig. 10. The
and sustained drift ratio of Specimen GG-U-CD2 was due to crack propagation was marked after each drift ratio by visual
the addition of diagonal bars from the column to the beam. inspection of the specimen. For the control Specimen SS-H-
The hysteresis envelopes of all the specimens are shown CD0, the first crack initiated in the beam at 60 mm from
Fig. 9, which indicated stiffness (slope of the load-displace- the column face. Flexural cracks propagated in the beam as
ment envelope curve) of Specimens GG-H-CD0, GG-U- the loading increased from 0.73% to 2.33% drift ratio. The
CD1, and GG-U-CD2 before the cracking was comparable major crack at the beam-column interface appeared at 0.73%
to the stiffness of Specimen SS-H-CD0. After the cracking, drift ratio and the opening and closing of the major crack
a significant drop in the stiffness of the GFRP-RC specimens caused spalling of concrete at 3.05% drift ratio. Small hair-
(GG-H-CD0, GG-U-CD1, and GG-U-CD2) was observed line cracks initiated in the connection region at 1.75% drift
compared to Specimen SS-H-CD0 until the end of the tests. ratio and extended with increasing drift ratios. The observed
Specimen GG-H-CD0 with conventional connection detail failure mode was concrete crushing at 4.0% drift ratio
sustained lower peak loads than Specimen SS-H-CD0 under followed by the excessive opening of major crack, concrete
the same applied drift ratio. The peak loads of Specimen spalling, and rupture of longitudinal bars at 5.09% drift ratio.
GG-U-CD2 with U-bend detail were higher than the peak The damage state and cracking pattern of the connection at
load of Specimens GG-U-CD1 (Z-bend) and GG-H-CD0 the ultimate drift ratio are shown in Fig. 10(a).
(GH-bend) at each applied drift ratio. Table 3 provides a

Table 3—Experimental results
Cracking load Service load Maximum load
Specimen ID (Pcr), kN (Ps), kN (Pmax), kN Mode of failure
Concrete crushing, excessive crack opening followed by rupture of
SS-H-CD0 14.5 35.6 81.2
beam reinforcing bars
GG-H-CD0 10.5 24.1 78.0 Concrete crushing and beam bar slippage
Connection degradation followed by anchorage failure of beam
GG-U-CD1 11.2 25.8 86.2
reinforcing bars
Diagonal cracking at the connection region followed by slight
GG-U-CD2 12.8 37.8 103.7
concrete crushing in the virtual plastic hinge region
Note: 1 kN = 0.225 kip.
at the final stage of loading (6.55% drift ratio). No substan-

tial damage and anchorage failure of the longitudinal bars of
the beam was observed in the connection, as shown in Fig.
10(d). This was because the extra U-bar from the outer face
of the column to the beam provided additional anchorage
resistance to the longitudinal bars of the beam. Hence, the
proposed U-bend detail is capable of avoiding the anchorage
failures of the longitudinal bars of the beam.
Displacement ductility
One of the main parameters to assess the cyclic perfor-
mance is the ductility of the connections. The displacement
ductility represents the capability of a structural element to
sustain inelastic deformations without substantial loss of
its strength before failure. For a steel-reinforced connec-
tion, ductility can be calculated by dividing the ultimate
displacement (Δu) by the equivalent yield displacement (Δy).
However, this definition is not valid for an FRP-reinforced
connection due to the absence of yielding in the tensile
Fig. 9—Hysteresis envelopes for all specimens. behavior of FRP bars. To compute the displacement ductility
of the FRP connection, the virtual elastic displacement (Δe)
The cracking patterns of Specimens GG-H-CD0, GG-U- can be considered as a transition point between the elastic
CD1 and, GG-U-CD2 were similar to that of Specimen and inelastic behavior, which is the serviceability limit state
SS-H-CD0, as shown in Fig. 10(b) to (d). The cracks initi- at which concrete reaches the compressive strain of 0.001.48
ated in the virtual plastic hinge region of the beam and prop- Under cyclic loading, connection yielding and ultimate
agated along the length of the beam with increasing drift displacements are difficult to observe due to the cracking or
levels. For GFRP-RC specimens, the virtual plastic hinge yielding of connection parts. The procedure to estimate the
can be defined as the top or bottom region of the beam near ductility index for a steel-reinforced connection as proposed
the column face where large elastic deformation of the longi- by Park50 is shown in Fig. 11(a). The yield displacement (Δy)
tudinal bars of the beam occurs. Small shear cracks started is found by drawing a perpendicular on the horizontal axis
to appear in the connection at 2.33% drift ratio for all the through the intersection point of the secant line at 75% of
GFRP specimens. The failure of Specimen GG-H-CD0 initi- the maximum load (Pmax) and horizontal line at the Pmax. The
ated with concrete crushing and spalling of concrete chunks ultimate displacement (Δu) is the deflection corresponding to
near the bottom edges of the beam at 4.0% drift ratio. The 20% loss of Pmax. For the GFRP connection, the displacement
concrete spalling continued at the top and bottom of the corresponding to the intersection point of the secant stiffness
beam in subsequent drift ratios and the major crack at the passing through service loading point (Ps) and the horizontal
beam-connection interface propagated across the depth of line at Pmax is Δe, as adopted in previous studies,51,52 and
the beam at 6.55% drift ratio. The observed failure modes deflection corresponding to Pmax is considered Δu, as shown
of Specimen GG-U-CD1 with proposed Z-bend detail in Fig. 11(b).
were concrete crushing, major crack opening, connection The computed displacement ductility indexes for all the
concrete spalling, as shown in Fig. 10(c). During the first specimens in the pushing and pulling direction are listed in
cycle of 6.55% drift ratio, the anchorage failure of the beam Table 4. The average ductility index of the control Specimen
longitudinal bars with explosive sounds was observed. Spec- SS-H-CD0 was 2.93, while the GFRP-reinforced specimens
imen GG-U-CD2 with the proposed U-bend detail showed ranged between 1.98 and 2.60. The average ductility indexes
diagonal cracks at the connection region and followed by of Specimens GG-H-CD0 with conventional connection
concrete crushing in the virtual plastic hinge region of beams detail and GG-U-CD1 with Z-bend detail were 2.15 and

Fig. 10—Cracking pattern and failure of specimens: (a) SS-H-CD0; (b) GG-H-CD0; (c) GG-U-CD1; and (d) GG-U-CD2.
2.14, respectively, because their load-displacment envelope For Specimen SS-H-CD0, the average initial stiffness was
curves were almost similar, as indicated in Fig. 9. However, 1.15 times the average initial stiffness (ki,avg) of Specimen
Specimen GG-U-CD2 with U-bend detail exhibited a higher GG-H-CD0, as reported in Table 4. However, the secant
average ductility index of 2.55 as compared to that of Spec- stiffness of the control Specimen SS-H-CD0 was higher, and
imens GG-H-CD0 and GG-U-CD1. This was due to the the rate of stiffness degradation was lower with increasing
ability of Specimen GG-U-CD2 to sustain a high drift ratio drift ratio as compared to all the GFRP reinforced specimens
(6.55%) without any significant loss of load. due to the higher elastic modulus of steel bars compared to
GFRP bars (Fig. 12). At the yielding stage of loading, the
Stiffness degradation stiffnesses of Specimens SS-H-CD0 and GG-H-CD0 were
The degradation of secant stiffness in terms of drift ratios 0.60ki,avg and 0.40ki,avg, respectively. The secant stiffness at
for all the specimens is shown in Fig. 12. The secant stiff- the ultimate drift ratios of Specimens SS-H-CD0 and GG-H-
ness (ksec,i) is defined as the slope of the line connecting the CD0 were 0.13ki,avg and 0.15ki,avg, respectively.
peak loads in positive and negative directions of loading The stiffness degradation of Specimens GG-H-CD0
in the given cycle i, as illustrated in Fig. 13. The ksec,i was and GG-U-CD1 occurred at almost the same rate with an
computed by Eq. (3). increase of the drift ratio, as shown in Fig. 12. Neverthe-
less, the initial stiffness of Specimen GG-U-CD1 was higher
Pmax ,i + − Pmax ,i − than that of Specimen GG-H-CD0. At the service loading
k sec ,i = (3) stage, the secant stiffnesses of Specimens GG-H-CD0 and
∆i+ − ∆i−
GG-U-CD1 reduced by 60% and 64% of the initial stiff-
Overall, the secant stiffness of the specimens decreased nesses, respectively. At the ultimate stage of loading, the
with the consecutive cycles of increasing drift ratios due to secant stiffnesses of Specimens GG-H-CD0 and GG-U-CD1
the bridging stress degradation across the cracks. The slope were lower than that of Specimen GG-U-CD2. Hence, the
of the load-displacement (P-∆) envelope curve before the cyclic performance of Specimen GG-U-CD2 with U-bend
crack initiation in the positive or negative directions of detail was better than the cyclic performance of Specimens
cyclic loading is referred to as the initial stiffness (ki) of the GG-H-CD0 and GG-U-CD1.
connection, as illustrated in Fig. 11 and 13. The yield stiff-
ness (ky) of the connection is the slope of the line connecting Energy dissipation
the yield point on the P-∆ envelope curve with the origin Energy dissipation is one of the important parameters to
in the positive or negative directions of cyclic loading, as assess the cyclic performance of RC BCCs, which is the
shown in Fig. 11. The average of the stiffnesses in positive measure of the absorbed energy before failure. The energy
and negative directions of the cyclic loading at the initial and dissipation per cycle (Ei) is determined by computing the
yield point were termed as the average initial stiffness (ki,avg) area of the hysteresis loop at each drift ratio, as illustrated
and average yield stiffness (ky,avg), respectively. The compar- in Fig. 13. The cumulative energy was obtained by the
ison of the connection stiffness at the initial, yielding, and summation of energy dissipation in successive cycles of drift
ultimate drift ratio is given in Table 4. ratio. The comparison of cumulative energy dissipation for

Table 4—Displacement ductility index and connection stiffness
Load ki, ki,avg, kN/ k*, k*avg, ksec,u,
Specimen ID direction Pmax, kN Δ*, mm Δu, mm μ μavg kN/mm mm kN/mm kN/mm kN/mm
+ 80.1 18.3 54.1 2.95 7.20 4.37
SS-H-CD0 2.93 6.78 4.02 0.88
– 81.2 22.2 64.4 2.90 6.35 3.66
+ 75.2 34.4 72.2 2.10 5.92 2.25
GG-H-CD0 2.15 5.89 2.36 0.89
– 78.0 31.3 68.8 2.20 5.85 2.46
+ 80.6 36.7 84.3 2.30 7.15 2.19
GG-U-CD1 2.14 6.40 2.33 1.00
– 86.2 35.0 69.3 1.98 5.64 2.46
+ 103.1 34.4 89.4 2.60 7.86 3.00
GG-U-CD2 2.55 7.30 2.98 1.16
– 103.7 35.0 87.5 2.50 6.73 2.96
*
Δy, ky and Δe, ke are for steel- and GFRP-reinforced connection, respectively.
Note: Pmax is maximum load; Δy is yield displacement; Δe is virtual elastic displacement; Δu is ultimate displacement; μ is displacement ductility index; μavg is average displacement
ductility index; ki is initial stiffness; ki,avg is average initial stiffness; ky is yield stiffness; ke is virtual elastic stiffness; ky,avg is average yield stiffness; ke,avg is average virtual elastic
stiffness; ksec,u is ultimate secant stiffness; 1 kN = 0.225 kip; 1 mm = 0.039 in.; 1 kN/mm = 5.710 kip/in.
Fig. 11—Definition of ductility index: (a) control specimen; and (b) GFRP specimens.
higher than the cumulative energy dissipation of GFRP spec-
imens due to the yielding of steel bars, which increased the
pinching length and area of the hysteresis loops. Specimen
GG-H-CD0 with conventional connection detail showed
an increase in the cumulative energy dissipation up to 5%
drift ratio. At 5% drift ratio, the cumulative energy dissipa-
tion of Specimen GG-H-CD0 was 30% higher than Spec-
imen GG-U-CD1 with the proposed Z-bend detail. Then,
a sudden increase in the cumulative energy was observed
for Specimens GG-U-CD1 and GG-H-CD0 at 6% drift ratio
due to excessive damage caused by concrete crushing and
anchorage failure. On the other hand, Specimen GG-U-CD2
with the proposed U-bend detail showed a higher cumulated
energy dissipation up to 5% drift ratio compared to Speci-
mens GG-H-CD0 and GG-U-CD1 before failure, as shown
in Fig. 13. Specimen GG-U-CD2 was able to dissipate more
energy up to 6.55% drift ratio without any sudden collapse
Fig. 12—Degradation of secant stiffness. of the connection. The high cumulative energy dissipation
is desirable for MRFs before the collapse of the connection.
all the specimens is shown in Fig. 14. In general, the slope Hence, the cyclic performance of Specimen GG-U-CD2 in
of the cumulative energy dissipation curves increases with terms of cumulative energy dissipation is better than Speci-
the damage level towards high drift ratios. The cumulative mens GG-H-CD0 and GG-U-CD2.
energy dissipation of Specimen SS-H-CD0 was significantly

Fig. 13—Definition of secant stiffness and energy dissipation according to ACI 374.1-05(19).48
Strain evolutions
The evolution of the maximum strains in the longitudinal
reinforcement of the beam at the column face was recorded
by a strain gauge at each drift ratio and is plotted in Fig. 15(a).
The yield strain of the reinforcing bars (2702 με) of the
control Specimen SS-H-CD0 is also marked in Fig. 15(a).
The measured strains of the control Specimen SS-H-CD0
were within the elastic range up to 1.31% drift ratio. Then,
the measured strain exceeded the yield strain (2702 με) of
the reinforcing bar at 1.75% drift ratio, and the maximum
strain reached 7568 με at 2.33% drift ratio. Thereafter, the
measured strains reduced to 5684 με at the ultimate loading
drift ratio due to slippage and plastic hinge formation.
The measured strain of all the GFRP specimens increased
linearly with the drift ratio until the failure of the specimens.
The maximum strains recorded for Specimen GG-H-CD0
Fig. 14—Cumulative energy dissipation.
was 7452 με at 1.75% drift ratio and then the strain gauge
stopped working at higher drift ratios due to cracking and the maximum measured strains in the connection stirrups
bending. The evolutions of maximum strain for Specimens of Specimens GG-H-CD0, GG-U-CD1 and, GG-U-CD2
GG-U-CD1 and GG-U-CD2 were similar up to 1.02% drift showed were 3627, 2352, and, 3218 με, respectively, as
ratio. Afterwards, Specimen GG-U-CD2 exhibited a slightly shown in Fig. 15(b). The measured strains are less than
higher strain until 3.0% drift ratio, as shown in Fig. 15(a). the design strain limit (6000 με) provided by CSA S806-
The maximum achieved strain values were 10,746 με and 12(R2017),18 which confirms that the shear cracks in the
11,519 με at 3.0% drift ratio for Specimens GG-U-CD1 and connection region were not excessively widened as observed
GG-U-CD2, respectively. The strain values at higher loading during the experiment.
drift ratios were not captured due to the malfunctioning of
the strain gauge. Drift contribution
The development of maximum strains in the central The drift ratio can be decomposed into different compo-
connection stirrup are shown in Fig. 15(b). Overall, the nents such as plastic hinge rotation, main crack at column
maximum strains increased gradually up to 1.75% drift face, column rotation, and connection distortion. In this
ratio. Then, an abrupt increase in the measured maximum study, rotation attributed to flexural cracking and shear defor-
strains was observed because the connection stirrup started mation in the beam was not considered. The percentage of
to confine the connection after the appearance of diagonal contribution to the total drift angle for each component was
shear cracks in all specimens. The maximum measured determined by using LVDT readings and plotted in Fig. 16.
strains in the control Specimen SS-H-CD0 exceeded the The location of LVDTs and calculation rotations from LVDTs
yielding strain (2746 με) at 4.0% drift ratio and reached reading is explained in section “Test setup and instrumenta-
to peak strain of 3342 με at 5.09% drift ratio. At failure, tion” of this paper. The term virtual plastic hinge was used for

Fig. 15—Evolution of maximum strains: (a) beam longitudinal bar; and (b) connection stirrup.
Fig. 16—Decomposition of total drift angle: (a) SS-H-CD0 (b); GG-H-CD0; (c) GG-U-CD1; and (d) GG-U-CD2.
GFRP reinforced specimens due to the absence of yielding bar. The contribution to the total drift angle of column rota-
and plastic hinge for the steel reinforced specimen. tion and connection distortion was approximately 6 to 8%
For the control Specimen SS-H-CD0, the percentage of because of small deformations in the connection and column.
contribution to total drift angle due to plastic hinge rota- Specimen GG-H-CD0 with conventional connection
tion was approximately 36 to 45% throughout the loading, detail and Specimen GG-U-CD1 with Z-bend detail showed
as shown in Fig. 15(a). However, rotation due to the main similar behavior in regards to the contribution to total drift
crack at the column face remained approximately at 12% angle at all levels of loading, as shown in Fig. 16(b) and
of the total drift angle up to 2% drift ratio. Afterwards, the (c). The percentage contribution to the total drift angle of
percentage contribution to the total drift angle increased to the virtual plastic hinge and main crack at column face was
19% until 5% drift ratio due to slipping and yielding of the approximately 48 to 62% and 52 to 70% for Specimens

GG-H-CD0 and GG-U-CD1, respectively. However, rota- and 54%, respectively, of the design strain limit of 6000 µɛ
tions attributed to column rotation and connection distortion set by CSA S806-12(R2017).19 This indicates that the
were in the range of 6 to 8% for both specimens. On the connection shear failure did not occur in Specimens GG-U-
other hand, Specimen GG-U-CD2 with the proposed U-bend CD2 up to 6.55% drift ratio.
details exhibited less contribution to total drift angle at all 6. The proposed U-bend detail was able to attain a higher
levels of loading, as shown in Fig. 16(d). At 1.0% drift ratio, sustained drift ratio, avoid anchorage failure of the longitu-
the combined rotation due to virtual plastic hinge and main dinal bars of the beam, and reduce the beam rotation. This
crack at column face was approximately 30% of the total was due to the better anchorage of the longitudinal bars of
drift angle. After this drift ratio, the percent contribution the beam inside the connection.
of the virtual plastic hinge and main crack at the column
face went up by approximately 10% at 5.0% drift ratio. The AUTHOR BIOS
contribution of the virtual plastic hinge and main crack at Muhammad Safdar is a PhD Candidate in the School of Civil, Mining
and Environmental Engineering, University of Wollongong, Wollongong,
column face (beam rotations) for Specimen GG-U-CD2 with Australia. He received his BSc in civil engineering from University of Engi-
proposed U-bend detail was lower than that of the contribu- neering and Technology Lahore, Lahore, Pakistan, in 2011, and his MSc
tions of Specimens GG-H-CD0 with conventional connec- from Hokkaido University, Sapporo, Japan, in 2016. His research interests
include finite element analysis of concrete structures.
tion detail and GG-U-CD1 with the Z-bend detail. This was
due to the less damage caused by cracking and slippage in M. Neaz Sheikh is an Associate Professor in the School of Civil, Mining
the specimen. and Environmental Engineering, University of Wollongong. He received
his BSc in civil engineering from Chittagong University of Engineering
and Technology, Chittagong, Bangladesh, and his MPhil and PhD from
CONCLUSIONS the University of Hong Kong, Hong Kong. His research interests include
An experimental study was carried out to investigate the concrete structures, composite structures, and structural dynamics.
behavior of the GFRP reinforced exterior beam-column ACI member Muhammad N. S. Hadi is a Professor in the School of Civil,
connections under reversed cyclic loading with different Mining and Environmental Engineering, University of Wollongong. He
connection details. Two novel connection details were received his BS and MS from the University of Baghdad, Baghdad, Iraq,
in 1977 and 1980, respectively, and his PhD from the University of Leeds,
developed for the GFRP-RC beam-column connections. Leeds, UK, in 1989. His research interests include analysis and design of
The cyclic performance of the novel connection detail was concrete structures.
evaluated and compared with the conventional connection
details. Based on the test results, the following conclusions ACKNOWLEDGMENTS
can be drawn regarding the cyclic behavior of GFRP-RC The authors would like to express their graduate to J. Wallace, Senior
Technical Officer of Smart Engineering Lab, for his support and help during
beam-column connections. the experimental work and testing. The first author would like to acknowl-
1. Specimen GG-H-CD0 with conventional connection edge to Higher Education Commission (HEC) of Pakistan and the Graduate
detail exhibited adequate strength, lower stiffness, and Research School of the University of Wollongong for providing the PhD
research scholarship.
plastic deformation, which resulted in lower energy dissipa-
tion as compared to that of steel-reinforced Specimen SS-H-
REFERENCES
CD0 under cyclic loading. 1. Dousti, A.; Moradian, M.; Taheri, S. R.; Rashetnia, R.; and
2. Specimen GG-U-CD1 with the proposed Z-bend detail Shekarchi, M., “Corrosion Assessment of RC Deck in a Jetty Structure
showed similar performance in terms of strength, ductility, Damaged by Chloride Attack,” Journal of Performance of Constructed
Facilities, ASCE, V. 27, No. 5, 2013, pp. 519-528. doi: 10.1061/(ASCE)
and stiffness degradation to Specimen GG-H-CD0 with the CF.1943-5509.0000348
conventional connection detail. Specimens GG-H-CD0 and 2. Costa, A., and Appleton, J., “Case Studies of Concrete Deterioration
GG-U-CD1 exhibited the slippage and anchorage failure of in a Marine Environment in Portugal,” Cement and Concrete Composites,
V. 24, No. 1, 2002, pp. 169-179. doi: 10.1016/S0958-9465(01)00037-3
longitudinal bars of the beam at approximately 5.0% drift 3. Bertolini, L.; Gastaldi, M.; Pedeferri, M.; and Redaelli, E., “Preven-
ratio due to the availability of limited anchorage length at tion of Steel Corrosion in Concrete Exposed to Seawater with Submerged
the connection region. Sacrificial Anodes,” Corrosion Science, V. 44, No. 7, 2002, pp. 1497-1513.
doi: 10.1016/S0010-938X(01)00168-8
3. Specimen GG-U-CD2 with the proposed U-bend detail 4. Jung, S. H.; Yang, H. M.; Yang, K. H.; and Kwon, S. J., “Maintenance
exhibited higher strength, higher ductility, and a slower rate of for Repaired RC Column Exposed to Chloride Attack Based on Probability
stiffness degradation than Specimen GG-H-CD0 with conven- Distribution of Service Life,” International Journal of Concrete Structures
and Materials, V. 12, No. 1, 2018, pp. 1-9. doi: 10.1186/s40069-018-0259-2
tional connection detail and Specimen GG-U-CD1 with Z-bend 5. Dousti, A.; Shekarchi, M.; Alizadeh, R.; and Taheri-Motlagh, A.,
detail. This was due to the additional U-bar placed from the “Binding of Externally Supplied Chlorides in Micro Silica Concrete Under
outer face of the column to the beam, which enhanced the flex- Field Exposure Conditions,” Cement and Concrete Composites, V. 33,
No. 10, 2011, pp. 1071-1079. doi: 10.1016/j.cemconcomp.2011.08.002
ural capacity of the beam and drift capacity of BCCs. 6. Sharbatdar, M. K., “Concrete Columns and Beams Reinforced with
4. Specimen GG-U-CD2 with the proposed U-bend detail FRP Bars and Grids under Monotonic and Reversed Cyclic Loading,” PhD
experienced slightly higher cumulative energy dissipation dissertation, University of Ottawa, Ottawa, ON, Canada, 2003.
7. Sugita, M., “NEFMAC Grid Type Reinforcement,” Altenative
up to 5.0% drift ratio compared to Specimens GG-H-CD0 Materials for The Reinforcement and Prestressing of Concrete, Blackie
and GG-U-CD1. However, the percentage increase in the Academic & Professional, 1993, pp. 55-82.
cumulative energy dissipation of Specimen GG-U-CD2 8. Vijay, P. V., and GangaRao, H. V., “Bending Behavior and Deforma-
bility of Glass Fiber-Reinforced Polymer Reinforced Concrete Members,”
from 5.0% to 6.55% drift ratio was 94% without a sudden ACI Structural Journal, V. 98, No. 6, Nov.-Dec. 2001, pp. 834-842.
drop in strength. 9. Soric, Z.; Kisicek, T.; and Galic, J., “Deflections of Concrete Beams
5. The maximum strain achieved by Specimens GG-H- Reinforced with FRP Bars,” Materials and Structures, V. 43, No. 1, 2010,
pp. 73-90. doi: 10.1617/s11527-010-9600-1
CD0, GG-UCD1, and GG-U-CD2 at failure was 60%, 39%,

10. Tobbi, H.; Farghaly, A. S.; and Benmokrane, B., “Concrete Columns University-Engineering Sciences, V. 27, No. 2, 2015, pp. 142-152. doi:
Reinforced Longitudinally and Transversally with Glass Fiber-Reinforced 10.1016/j.jksues.2013.09.002
Polymer Bars,” ACI Structural Journal, V. 109, No. 4, July-Aug. 2012, 30. Hadi, M. N. S., and Tran, T. M., “Retrofitting Nonseismically
pp. 551-558. Detailed Exterior Beam–Column Joints Using Concrete Covers Together
11. Hadi, M. N., and Youssef, J., “Experimental Investigation of with CFRP Jacket,” Construction and Building Materials, V. 63, 2014, pp.
GFRP-reinforced and GFRP-Encased Square Concrete Specimens Under 161-173. doi: 10.1016/j.conbuildmat.2014.04.019
Axial and Eccentric Load, and Four-point Bending Test,” Journal of 31. CSA A23.3-19, “Design of Concrete Structures,” Canadian Stan-
Composites for Construction, ASCE, V. 20, No. 5, 2016, p. 04016020. doi: dards Association, Toronto, ON, Canada, 2019.
10.1061/(ASCE)CC.1943-5614.0000675 32. Mateenbar, “MateenBar-Technical Data,” Mina Jebel Ali UAE,
12. Afifi, M. Z.; Mohamed, H. M.; and Benmokrane, B., “Axial Capacity Accessed December, 2018, https://www.mateenbar.com/products/
of Circular Concrete Columns Reinforced with GFRP Bars and Spirals,” mateenbar/.
Journal of Composites for Construction, ASCE, V. 18, No. 1, 2014, 33. Pultrall, Inc., “V-ROD, Technical Data Sheets,” Thetford Mines, QC,
p. 04013017. doi: 10.1061/(ASCE)CC.1943-5614.0000438 Canada, 2017.
13. Fukuyama, H.; Masuda, Y.; Sonobe, Y.; and Taniagki, M., “Structural 34. Choo, C. C.; Harik, I. E.; and Gesund, H., “Strength of Rectangular
Performances of Concrete Frame Reinforced with FRP Reinforcement,” Concrete Columns Reinforced with Fiber-Reinforced Polymer Bars,” ACI
Non-Metallic (FRP) Reinforcement for Concrete Structures: Proceedings Structural Journal, V. 103, No. 3, May-June 2006, pp. 452-459.
of the Second International RILEM Symposium, Chapman & Hall, London, 35. Karim, H.; Sheikh, M. N.; and Hadi, M. N. S., “Load and Moment
UK, 1995, pp. 275-286. Interaction Diagram for Circular Concrete Columns Reinforced with GFRP
14. Said, A. M., and Nehdi, M. L., “Use of FRP for RC Frames in Seismic Bars and GFRP Helices,” Journal of Composites for Construction, ASCE, V.
Zones: Part II. Performance of Steel-Free GFRP-Reinforced Beam-Column 21, No. 1, 2017, p. 04016076. doi: 10.1061/(ASCE)CC.1943-5614.0000737
Joints,” Applied Composite Materials, V. 11, No. 4, 2004, pp. 227-245. doi: 36. Sharbatdar, M. K., “Concrete Columns and Beams Reinforced with
10.1023/B:ACMA.0000035480.85721.b5 FRP Bars and Grids under Monotonic and Reversed Cyclic Loading,” PhD
15. Sharbatdar, M. K.; Saatcioglu, M.; Benmokrane, B.; and El-Sal- thesis, University of Ottawa, Ottawa, ON, Canada, 2003.
akawy, E., “Behaviour of FRP Reinforced Concrete Beam-Column Joints 37. Zadeh, H. J., and Nanni, A., “Design of RC Columns Using Glass FRP
under Cyclic Loading,” 3rd International Conference on Durability and Reinforcement,” Journal of Composites for Construction, ASCE, V. 17,
Field Applications of Fibre Rein-forced Polymer (FRP) Composites for No. 3, 2013, pp. 294-304. doi: 10.1061/(ASCE)CC.1943-5614.0000354
Construction, CDCC-07, Québec, QC, Canada, 2007, pp. 541-548. 38. AS 1012.9-14, “Methods of Testing Concrete—Method 9: Deter-
16. Mady, M.; El-Ragaby, A.; and El-Salakawy, E., “Seismic Behavior of mination of the Compressive Strength of Concrete Specimens,” Standards
Beam-Column Joints Reinforced with GFRP Bars and Stirrups,” Journal of Australia, Sydney, NSW, Australia, 2014.
Composites for Construction, ASCE, V. 15, No. 6, 2011, pp. 875-886. doi: 39. AS 1391-07(R2017), “Metallic Materials: Tensile Testing at Ambient
10.1061/(ASCE)CC.1943-5614.0000220 Temperature,” Standards Australia, Sydney, NSW, Australia, 2017.
17. Hasaballa, M., “GFRP-Reinforced Concrete Exterior Beam-Column 40. ASTM D7205/D7205M-16, “Standard Test Method for Tensile Prop-
Joints Subjected to Seismic Loading,” PhD dissertation, University of erties of Fiber Reinforced Polymer Matrix Composite Bars,” ASTM Inter-
Manitoba, Winnipeg, MB, Canada, 2014. national, West Conshocken, PA, 2016.
18. CSA S806-12(R2017), “Design and Construction of Building Struc- 41. Gil-Martín, L. M.; Rodríguez-Suesca, A. E.; Fernández-Ruiz, M. A.;
tures with Fibre-Reinforced Polymers,” Canadian Standards Association, and Hernández-Montes, E., “Cyclic Behavior of RC Beam-Column Joints
Toronto, ON, Canada, 2017. with Epoxy Resin and Ground Tire Rubber as Partial Cement Replace-
19. ACI Committee 440, “Guide for the Design and Construction of ment,” Construction and Building Materials, V. 211, 2019, pp. 659-674.
Structural Concrete Reinforced with Fiber-Reinforced Polymers (FRP) doi: 10.1016/j.conbuildmat.2019.03.216
Bars (ACI 440.1R-15),” American Concrete Institute, Farmington Hills, 42. Ghobarah, A.; Aziz, T. S.; and Biddah, A., “Rehabilitation of Rein-
MI, 2015, 88 pp. forced Concrete Frame Connection Using Corrugated Steel Jacketing,” ACI
20. Ghomi, S. K., and El-Salakawy, E., “Seismic Performance of Structural Journal, V. 94, No. 3, May-June 1997, pp. 492-499.
GFRP-RC Exterior Beam–Column Joints with Lateral Beams,” Journal of 43. Ehsani, M. R., and Weight, J. K., “Exterior Reinforced Concrete
Composites for Construction, ASCE, V. 20, No. 1, 2016, p. 04015019. doi: Beam-to-Column Connections subjected to Earthquake-Type Loading,”
10.1061/(ASCE)CC.1943-5614.0000582 ACI Structural Journal, V. 82, No. 4, July-Aug. 1997, pp. 282-294.
21. Hasaballa, M., and El-Salakawy, E., “Shear Capacity of Exterior 44. Ismail, M. K.; Abdelaleem, B. H.; and Hassan, A. A. A., “Effect of
Beam-Column Joints Reinforced with GFRP Bars and Stirrups,” Journal of Fiber Type on the Behavior of Cementitious Composite Beam-Column
Composites for Construction, ASCE, V. 20, No. 2, 2016, p. 04015047. doi: Joints under Reversed Cyclic Loading,” Construction and Building Mate-
10.1061/(ASCE)CC.1943-5614.0000609 rials, V. 186, 2018, pp. 969-977. doi: 10.1016/j.conbuildmat.2018.08.024
22. Hasaballa, M., and El-Salakawy, E., “Anchorage Performance of 45. Chun, S. C.; Lee, S. H.; Kang, T. H.; Oh, B.; and Wallace, J. W.,
GFRP Headed and Bent Bars in Beam-Column Joints Subjected to Seismic “Mechanical Anchorage in Exterior Beam-Column Joints Subjected to
Loading,” Journal of Composites for Construction, ASCE, V. 22, No. 6, Cyclic Loading,” ACI Structural Journal, V. 104, No. 1, Jan.-Feb. 2007,
2018, p. 04018060. doi: 10.1061/(ASCE)CC.1943-5614.0000888 pp. 102-112.
23. Shen, X.; Li, B.; Chen, Y. T.; and Tizani, W., “Seismic Performance 46. Ghobarah, A., and El-Amoury, T., “Seismic Rehabilitation of
of Reinforced Concrete Interior Beam-column joints with Novel Rein- Deficient Exterior Concrete Frame Joints,” Journal of Composites for
forcement Detail,” Engineering Structures, V. 227, 2021, p. 111408. doi: Construction, ASCE, V. 9, No. 5, 2005, pp. 408-416. doi: 10.1061/
10.1016/j.engstruct.2020.111408 (ASCE)1090-0268(2005)9:5(408)
24. Chalioris, C. E.; Favvata, M. J.; and Karayannis, C. G., “Reinforced 47. Hakuto, S.; Park, R.; and Tanaka, H., “Seismic Load Tests on Interior
Concrete Beam–Column Joints with Crossed Inclined bars under Cyclic and Exterior Beam-Column Joints with Substandard Reinforcing Details,”
Deformations,” Earthquake Engineering & Structural Dynamics, V. 37, ACI Structural Journal, V. 97, No. 1, Jan.-Feb. 2000, pp. 11-25.
No. 6, 2008, pp. 881-897. doi: 10.1002/eqe.793 48. CSA S6-06(R2014), “Canadian Highway Bridge Design Code
25. Tsonos, A. G.; Tegos, I. A.; and Penelis, G. G., “Seismic Resistance (CHBDC),” Canadian Standards Association (CSA), Mississauga, ON,
of Type 2 Exterior Beam Column Joints Reinforced with Inclined bars,” Canada, 2014, pp. 1078.
ACI Structural Journal, V. 89, No. 1, Jan.-Feb. 1993, pp. 3-12. 49. ACI Committee 374, “Acceptance Criteria for Moment Frames
26. Lu, X.; Urukap, T. H.; Li, S.; and Lin, F., “Seismic Behavior of Inte- Based on Structural Testing (ACI 374.1-05(19)) and Commentary (Reap-
rior RC Beam-Column Joints with Additional Bars Under Cyclic Loading,” proved 2019),” American Concrete Institute, Farmington Hills, MI, 2019,
Earthquakes and Structures, V. 3, No. 1, 2012, pp. 37-57. doi: 10.12989/ 9 pp.
eas.2012.3.1.037 50. Park, R., “Evaluation of Ductility of Structures and Structural Assem-
27. Au, F. T.; Huang, K.; and Pam, H. J., “Diagonally-Reinforced blages from Laboratory Testing,” Bulletin of the New Zealand Society for
Beam–Column Joints Reinforced Under Cyclic Loading,” Proceedings of Earthquake Engineering, V. 22, No. 3, 1989, pp. 155-166. doi: 10.5459/
the Institution of Civil Engineers. Structures and Buildings, V. 158, No. 1, bnzsee.22.3.155-166
2005, pp. 21-40. doi: 10.1680/stbu.2005.158.1.21 51. Eladawy, M.; Hassan, M.; Benmokrane, B.; and Ferrier, E., “Lateral
28. Kotsovou, G., and Mouzakis, H., “Exterior RC Beam–Column Cyclic Behavior of Interior Two-Way Concrete Slab–Column Connections
Joints: New Design Approach,” Engineering Structures, V. 41, 2012, Reinforced with GFRP Bars,” Engineering Structures, V. 209, 2020, p.
pp. 307-319. doi: 10.1016/j.engstruct.2012.03.049 109978 doi: 10.1016/j.engstruct.2019.109978
29. Rajagopal, S., and Prabavathy, S., “Investigation on the 52. Mohamed, N.; Farghaly, A.; Benmokrane, B.; and Neale, K. W.,
Seismic Behavior of Exterior Beam–Column Joint Using T-Type “Drift Capacity Design of Shear Walls Reinforced with Glass Fiber-Re-
Mechanical Anchorage with Hair-Clip Bar,” Journal of King Saud inforced Polymer Bars,” ACI Structural Journal, V. 111, No. 6, Nov.-Dec.
2014, pp. 1397-1406. doi: 10.14359/51687099

Title No. 119-S46
Effect of Slenderness Ratio on Glass Fiber-Reinforced

Polymer-Reinforced High-Strength Concrete Columns
by Mu’taz Almomani, Karam Mahmoud, and Ehab F. El-Salakawy
This paper presents the results of eight large-scale high-strength capacity (ACI Committee 318 2019). Early studies, such as
concrete (HSC) columns reinforced internally with glass fiber- Mirmiran et al. (2001), suggest that fiber-reinforced polymer
reinforced polymer (GFRP) bars and spirals. The effects of the (FRP)-RC columns are more susceptible to the slenderness
slenderness ratio and the eccentricity-to-diameter ratio (e/D) on effect due to the lower modulus of elasticity of FRP reinforce-
the behavior of HSC columns that meet the minimum code require-
ment. Therefore, it was suggested that the slenderness ratio
ments are evaluated. Additionally, the column behavior was inves-
limit for FRP-RC short columns be reduced to 17, compared
tigated under pure flexural loading. Test results indicated that
increasing the e/D or the slenderness ratio resulted in a decrease to 22 for steel-RC columns. Numerous codes and guidelines
in the axial and lateral stiffness and the axial capacity of the HSC are currently available providing a reference for designers
columns. All tested columns exhibited a material-type failure, to incorporate FRP into structural elements. Many of these
which is characterized by the crushing of concrete. Furthermore, guidelines have maintained a high level of conservativeness
compressive strains measured in the GFRP bars indicated their regarding compression members. Among other codes, the
contribution to the column axial capacity. In addition, an interac- Canadian Highway Bridge Design Code (CSA S6-19 2019)
tion diagram was developed and compared to the predictions of the allows the use of FRP in members subjected to combined
available codes and guidelines. flexural and axial loads while limiting the strains in the FRP
bars in compression to 0.002. On the other hand, the ACI
Keywords: bending; eccentric loading; glass fiber-reinforced polymer
(GFRP) bars and spirals; high-strength concrete (HSC); short columns; 440.1R-15 guideline (ACI Committee 440 2015) provided
slender columns. recommendations not to consider FRP bars in compression
members or compression zones in flexural members. Further
INTRODUCTION limitations regarding slender compression members were
The effectiveness of glass fiber-reinforced polymer introduced by CSA S806-12 (2017), prohibiting the use of
(GFRP) reinforcement has gained increasing recogni- FRP entirely in such members.
tion over the past few decades. Researchers were able to In efforts to alleviate code restrictions, significant research
verify their viability as internal reinforcement for concrete has been carried out to map and define the behavior of
structures in comparison to their traditional steel counter- FRP columns. The behavior of FRP-RC short columns has
parts (Ali and El-Salakawy 2016; Ghomi and El-Salakawy been studied in recent literature (De Luca et al. 2010; Afifi
2016; El-Gendy and El-Salakawy 2016; Mahmoud and El- et al. 2014; Hadhood 2017; Barua and El-Salakawy 2020;
Salakawy 2016; Hadhood 2017; Rahman et al. 2017; Elchalakani et al. 2020). These studies showed similarities
Abdelazim 2020). In addition, the corrosion-resistant in the behavior of the FRP-RC columns and their steel-RC
nature of GFRP bars makes their use more favorable in counterparts. Additionally, the studies presented consistent
structures exposed to harsh conditions (Bakht et al. 2004; data showing a small increase in axial capacity and a more
ACI Committee 440 2015). The inevitability of corrosion pronounced increase in ductility with increasing FRP longi-
in steel-reinforced concrete (RC) means more costs are tudinal and transverse reinforcement under different eccen-
incurred in mitigating or repairing affected members (ACI tricities. However, recent studies have focused on slender
Committee 222 2019). Thus, eliminating the corrodible FRP-RC members due to the increasing need for FRP mate-
component of the structural element increases its service life rials in a wider range of structural applications. Abdelazim
and, consequently, reduces life-cycle costs. (2020), Khorramian and Sadeghian (2020), and Barua
Circular columns, compared to rectangular columns, et al. (2021) investigated normal-strength concrete (NSC)
provide uniform confinement and a more aesthetic appear- slender columns reinforced with GFRP bars. They found
ance. This makes their use in structures, such as bridges and that the mode of failure of columns with lower eccentricities
parking garages, more common. More critical to the perfor- showed a material-type failure characterized by the crushing
mance of a compression member is the slenderness ratio of the concrete. However, larger eccentricities caused the
(λ), which is defined as the ratio of the member’s effective columns to fail due to the development of excessive cracks
length to its radius of gyration (kℓ/r), where k, ℓ, and r are the
effective length coefficient, unbraced length, and radius of MS No. S-2021-150.R2, doi: 10.14359/51734343, received September 21, 2021, and
gyration, respectively. As the slenderness ratio increases, the reviewed under Institute publication policies. Copyright © 2022, American Concrete
column is subjected to larger lateral deflections. This creates obtained from the copyright proprietors. Pertinent discussion including author’s
secondary moments, which directly affect the column axial closure, if any, will be published ten months from this journal’s date if the discussion

on the tension side. As expected, the studies showed that ones. This study aims to experimentally investigate HSC
for the same eccentricity-to-diameter ratio (e/D), the lateral short and slender columns reinforced with GFRP bars and
displacements in the slender columns are consistently higher spirals that meet the minimum code requirements under an
than those observed in the short columns. These conclusions axial load with different eccentricities.
were similar for rectangular and circular columns. Barua
et al. (2021) compared slender columns (λ = 28) and short EXPERIMENTAL PROGRAM
columns (λ = 20), as defined by current codes. The study Specimen configuration and details
found that changing the slenderness ratio from 20 to 28 In this study, a total of eight large-scale concrete columns
showed a greater difference for results of steel-RC columns were constructed and tested to failure. The columns were
than GFRP-RC columns. Barua et al. (2021) suggested that reinforced with sand-coated GFRP bars and spirals. The
this difference was due to the steel-RC columns exhibiting columns were designed in accordance with the Canadian
distinct short and slender behavior, respectively. However, standard CSA S806-12 (2017). The columns were cast
the GFRP-RC columns were similar, as both slenderness with HSC and had a diameter of 355 mm. The height of
ratios exhibited slender behavior, suggesting the limit for the columns was either 1250 or 2450 mm, corresponding
GFRP-RC columns is lower than the 22 in codes, which to λ of 14 and 28, respectively. According to CSA S806-12
agrees with the findings of Mirmiran et al. (2001). (2017), short columns are defined as those with a λ ≤ 22.
Very limited research has been conducted to investigate the However, as mentioned earlier, recent literature recom-
behavior of high-strength concrete (HSC) slender columns mended reducing the limit of the short column from 22 to 18
reinforced with GFRP bars (Hales et al. 2016; Abdelazim (Abdelazim 2020). Consequently, the columns presented in
2020). Hales et al. (2016) studied HSC circular columns this study are designed to investigate slenderness ratios that
measuring 305 mm in diameter and 760 and 3730 mm in are defined as short and slender in accordance with both the
height (λ = 10 and 49, respectively) and with reinforce- current definition of slender and short compression members
ment ratios of 1.65 and 2.71%. Abdelazim (2020) tested and the recommendations in the literature. Throughout
specimens with the same diameter, however, with a column this paper, the columns with λ = 14 will be referred to as
length of 2500 mm (λ = 33) and reinforcement ratios of 3.28 “short” and those with λ = 28 will be referred to as “slender.”
and 4.66%. Hales et al. (2016) stated that for low e/D (up to Furthermore, CSA S6-19 (2019) specifies that the smallest
8.3%), both short and slender columns exhibited a failure permissible length-to-diameter ratio of a column is 2.5,
that was initiated by crushing of the concrete, followed by which is satisfied by the specimen dimensions selected. The
the rupture and buckling of the spirals and the compression cross-sectional diameter of the columns is greater than 300
bars, respectively. In contrast, a higher e/D (33%) in slender mm, and the longitudinal and transverse reinforcement ratios
columns showed much higher deflection at midheight and meet the minimum requirements for the reinforcement ratio,
exhibited a buckling (stability-type) failure. Conversely, number of bars, and spiral pitch specified by CSA S806-12
Abdelazim (2020) found that slender columns showed no (2017).
signs of stability-type failure. All columns failed due to The specimens were longitudinally reinforced with six
concrete crushing, which is a material-type failure. No. 16 GFRP bars and transversely with No. 10 GFRP
Available literature has provided a solid foundation for spirals spaced at 85 mm, as shown in Fig. 1. The axial load
the use of FRP in RC columns. However, due to the brittle eccentricities were 60, 90, 120, and 150 mm corresponding
nature of HSC, many aspects, such as ductility, the effect of to e/D of 0.17, 0.26, 0.34, and 0.43, respectively. In addi-
slenderness and eccentricity of the applied axial load, and tion to the axial loading, a column tested under pure flex-
the reinforcement ratio still need to be further investigated. ural loading was investigated in the study. The name of the
In addition, columns designed for the minimum reinforce- specimen consists of two parts. The first part is denoted by
ment requirements are common in buildings. Broms and S28 or S14 for slender (λ = 28) and short (λ = 14) columns,
Viest (1961) stated that the minimum reinforcement ratio is respectively. The second part represents the type of loading,
more critical for slender columns where a decrease in the where “FL” corresponds to flexural loading and “17,” “26,”
proportion of the load carried by the reinforcement leads to “34,” and “43” represent the e/D of 0.17, 0.26, 0.34, and
a less stable column. Therefore, further research is required 0.43, respectively, for axially loaded columns. The details of
to fully understand the behavior of such columns and to ease the specimens are summarized in Table 1.
the level of conservativeness of related provisions in the
codes and guidelines. Material properties
The columns were cast with ready mixed concrete with
RESEARCH SIGNIFICANCE a target 28-day strength of 60 MPa. The nominal aggregate
Despite the studies conducted on NSC columns reinforced size used was 20 mm. Standard cylinders, 100 x 200 mm,
with GFRP bars, gaps are still present in areas pertaining were cast and tested in accordance with CSA A23.1-19/CSA
to HSC columns due to the different nature of the material. A23.2-19 (2019) to determine the strength of the concrete
Additionally, previous literature has pointed to key differ- on the same day the columns were tested. The concrete
ences between short and slender columns with recommen- strength obtained for each specimen is listed in Table 1. The
dations of reducing the slenderness ratio limit to 17. Further- average concrete strength obtained for all the columns was
more, FRP design codes and guidelines remain cautious 63.7 MPa. The columns were reinforced with pultruded,
when approaching compression members, especially slender sand-coated GFRP bars and spirals. The properties of the

Fig. 1—Details of columns and GFRP reinforcement. (Note: 1 mm = 0.0394 in.)
Table 1—Test matrix
Longitudinal reinforcement Transverse reinforcement
Specimen ID fc′, MPa No. of bars ρf, % Pitch, mm ρfs, % λ e, mm e/D
S14-17 57.1 60 0.17
S14-26 64.3 14 90 0.26
S14-34 56.0 120 0.34
S28-17 64.0 60 0.17
Six No. 16 1.21 85 1.11
S28-26 64.3 90 0.26
S28-34 65.4 28 120 0.34
S28-43 66.4 150 0.43
S28-FL 72.0 ∞ ∞
Note: 1 mm = 0.0394 in.; 1 MPa = 0.14504 ksi.
GFRP reinforcement were provided by the manufacturer, as Table 2—Mechanical properties of GFRP
shown in Table 2. The manufacturer obtained the properties reinforcement
through certified tests that were carried out according to the Area, mm2 Elastic
CSA S807-19 (2019). tensile Tensile Ultimate
Nominal Annex modulus, strength, strain,
Test setup and procedure Bar size diameter Nominal A* GPa MPa %
The columns were tested using a 5000 kN capacity No. 16
hydraulic machine. The load was applied at a displace- (Straight 15.9 199 235 64 ± 1.49 1558 ± 47 2.40
ment-controlled rate of 1.5 mm/min. The columns tested bar)
under axial loading were fitted with two heavy steel collars No. 10
9.5 71 83 58 ± 1.50 667 ± 41 1.14
and were grouted to ensure confinement of the column ends. (Spiral)
A pin-pin boundary condition was simulated by welding a *
Area according to test method in CSA S806 Annex A.
pin and socket to collars and the loading machine, respec- Note: 1 mm = 0.0394 in.; 1 GPa = 145.04 ksi; 1 MPa = 0.14504 ksi; 1 mm2 =
tively, to ensure no transfer of moment. The location of the 0.00155 in2.

Fig. 2—Test setup and external instrumentation. (Note: 1 mm = 0.0394 in.)
pin on the collars varied to achieve the required load eccen- with the crushing of the concrete on the compression side.
tricity. A 25% drop in the column axial load marked the Following the crushing of the concrete, S14-17 was the
end of the test. The flexural setup consisted of two supports only specimen that showed buckling of the GFRP bar and
and two points of loading. The four loading points were rupture of the confining spiral at the same location on the
fitted with semicircular saddles ensuring full contact to the compression side. This was due to the additional compres-
column. The load was applied through a spreader beam to sion stresses on the confined core as the cover spalled. As
which the loading saddles were connected. The testing setup the eccentricity increased, the columns began to show earlier
is depicted in Fig. 2. signs of cracking attributed to the increased lateral deflec-
tions. The remaining short specimens showed cracks at 550
Instrumentation and 290 kN for specimens S14-26 and S14-34, respectively.
The behavior of the columns was monitored using various The slender columns showed the first crack at much lower
internal and external instrumentation. The GFRP bars and loads (390, 170, and 150 kN for specimens S28-26, S28-34,
spirals were fitted with several strain gauges to measure the and S28-43, respectively). Those observations also show that
strains on the tension and compression sides. In addition, a for the same e/D, the slender columns experienced tension
strain gauge and a PI gauge were mounted on the concrete cracks earlier than their short column counterparts. This was
surface on the compression side to measure the compressive due to tension cracks being influenced predominantly by the
strains. The deflected shape of the column was measured e/D for short columns; however, for the “S28” series, this
using five linear variable displacement transducers (LVDTs). was influenced by the e/D and the additional moment due
One LVDT was placed at midheight and the remaining were to the slenderness effect. The slenderness effect, introduced
evenly distributed along the length of the column. Further- by large eccentricities, presented no stability-type failure or
more, two LVDTs were also placed to measure the axial tensile rupture in the FRP bars. Furthermore, no debonding
displacement of the columns. Similar instrumentation was of the GFRP bars from the concrete was observed and the
used for the column under pure flexure. However, only three bond was maintained throughout the loading until failure. It
LVDTs were used to measure the deflection. The instrumen- is worth noting that the bond strength in HSC predicted by
tation for the two test setups is shown in Fig. 2. the design codes was very conservative for both low- and
high-modulus GFRP bars (Hossain et al. 2014).
EXPERIMENTAL RESULTS AND DISCUSSION
General behavior and modes of failure Effect of eccentricity
The observed modes of failure for all the specimens were The eccentricity, at which the load was applied, varied
material-type failure. As the specimens reached their peak from 60 to 150 mm, which corresponds to an e/D of 0.17 to
loads, the concrete cover began to spall, and the load imme- 0.43. As expected, when the eccentricity increased, the axial
diately dropped. Figure 3 shows the tested specimens at capacity decreased, which is a direct consequence of the
failure. For specimens S28-17 and S14-17, no cracks were substantial increase in the moment on the column. Because the
observed until failure. Tension cracks formed simultaneously concrete strength varied between columns, the experimental

Fig. 3—Mode of failure for all test specimens.
axial loads were normalized using a factor of 63.7/fc′, S14-34, respectively. The effect of changing the e/D, for both
where 63.7 MPa represents the average of the concrete values of λ, showed a very similar decrease in the overall
strength of all the columns. The test results are summarized capacity of the column; however, at greater loads for the
in Table 3. The normalized axial capacities for the slender short columns in general. This is due to the increased applied
columns were 3247, 2383, 1470, and 1023 kN for the moment as the e/D increased for both column slenderness
columns with e/D of 0.17, 0.26, 0.34, and 0.43, respectively. ratios. The effect on the capacity was similar for both series;
This represents a decrease of 26.6, 54.7, and 68.5% in the however, the disparity in the overall capacity was due to the
axial capacity compared to that of column S28-17. Similarly, slenderness effect, which reduced the axial capacity of the
the short columns achieved normalized axial capacities of slender columns greatly. As the axial capacity decreased, the
4079, 3047, and 1890 kN for columns S14-17, S14-26, and moment increased, indicating the change of e/D from 0.17

Table 3—Test results
Mu1*, kN∙m Mu2*, kN∙m
Specimen ID fc′, MPa e, mm δ, mm Pu, kN Pu*, kN [Pu × e] [Pu* × (e + δ)]
S14-17 57.1 60 2.36 3657 4079 244.7 255.3
S14-26 64.3 90 3.75 3076 3047 274.2 285.7
S14-34 56.0 120 5.30 1662 1890 226.8 236.1
S28-17 64.0 60 8.90 3264 3247 194.8 223.6
S28-26 64.3 90 9.80 2406 2383 214.5 237.8
S28-34 65.4 120 15.60 1510 1470 176.4 199.3
S28-43 66.4 150 24.50 1056 1023 152.0 176.7
S28-FL 72.0 ∞ 71.30 — — — 133.1
*
Normalized ultimate load and moments using average concrete strength.
Note: 1 kN = 0.2248 kip; 1 mm = 0.0394 in.
Fig. 4—Strains in longitudinal bars in axially loaded columns. (Note: 1 kN = 0.225 kip.)
to 0.26 was still above the theoretical balance point found 0.17, 0.26, 0.34, and 0.43, respectively. This was due to
on a typical axial load-bending moment interaction diagram. the reduced depth of the compression zone in the concrete
Further decrease in the axial capacity (or increase in e/D) with increasing the eccentricity. The increase in the moment
resulted in a consistent decrease in the moment at failure. because of increasing the eccentricity means the neutral axis
Barua et al. (2021) tested NSC columns with identical shifted further toward the compression side, thus increasing
dimensions and λ = 28. It was observed that increasing the the tensile strains in the tension side. However, the compres-
e/D from 0.0 to 0.085, and further to 0.17 and 0.34, caused sive strains in the compression side bars were very close in
a decrease in the specimen axial load capacity by 29, 37, value and approached the maximum design compressive
and 69%. In addition, increasing the e/D from 0.17 to 0.34 strains of the outer compression fibers of the concrete. The
resulted in a 46% reduction in the axial capacity, which is measured strains on the bar on the compression side were
less than the 54.7% observed for HSC columns in the current –3020, –2400, –2480, and –2640 με for columns with e/D
study. On the other hand, for slender columns (λ = 33) with of 0.17, 0.26, 0.34, and 0.43, respectively. Similarly, for
80 MPa concrete strength, Abdelazim (2020) reported a the “S14” series, an increase was observed in the strain
similar decrease in the axial capacity of 51 and 54% when measured in the bar on the tension side when increasing
the e/D increased from 0.16 to 0.33 and 0.66, respectively. the e/D. These tensile strains were 550, 1390, and 3130 με
This means that the axial capacity of GFRP-RC columns is in columns with e/D of 0.17, 0.26, and 0.34, respectively.
significantly affected by the concrete strength. Correspondingly, the compressive strains in the GFRP
Figure 4 shows the measured strains in the GFRP bars. bars were –3050, –2880, and –2860 με in columns S14-17,
Strains in the outermost bars on the tension side increased S14-26, and S14-34, respectively.
linearly up to the peak load as the e/D increased. As observed Concrete compressive strains (Fig. 5) were close to or
in Fig. 4, the tensile strains in the bars for the “S28” series exceeded the design strains of –3000 με specified in ACI
at peak load were 480, 2020, 3820, and 7490 με for e/D of 440.1R-15 (ACI Committee 440 2015) and –3500 με

Fig. 5—Concrete compressive strains in axially loaded columns. (Note: 1 kN = 0.225 kip.)
Fig. 6—Spiral strains in axially loaded specimens. (Note: 1 kN = 0.225 kip.)

specified in CSA S806-12 (2017). The concrete strains on the with an e/D of 0.17. This was due to the larger increase in
compression side measured –2890, –3120, and –3430 με for axial capacity for specimens with a lower e/D.
specimens S14-17, S14-26, and S14-34, respectively, while As the load increased, the spiral strains increased almost
they were –3430, –2530, –3150, and –3300 με for specimens linearly until approximately 80 to 95% of the peak load;
S28-17, S28-26, S28-34, and S28-43, respectively. However, however, the increase in strain became nonlinear prior to the
it is worth noting that some of the readings correspond to peak load. The strains increased rapidly when the concrete
loads less than the peak load as those strain gauges malfunc- cover spalled after the peak load was reached (Fig. 6). This
tioned prior to failure. The NSC columns tested by Barua shows that the spirals continued to confine the concrete core
et al. (2021) with e/D of 0.17 and 0.34 showed compression even after the peak load was reached. In general, columns
bar strains of –1530 and –2030 με and tension bars strains of having a high e/D (0.34 and 0.43) have a lower load-strain
100 and 3200 με, respectively. On the other hand, their HSC slope; therefore, the rate of increase of spiral strain is greater
counterparts (S28-17 and S28-34) had strains of –3020 and for those specimens. Increasing the e/D from 0.17 to 0.26
–2480 με in compression and 480 and 3820 με in tension, and further to 0.34 in columns with λ = 14 decreased the
respectively. This corresponds to an increase of 97 and 22% spiral strains at the peak load by 36.3 and 57.4%, respec-
in compressive strain, while the increase in tensile strains tively, compared to that with e/D of 0.17. Similarly, for
was 380 and 19.5% for e/D 0.17 and 0.34, respectively. The slender columns (λ = 28), increasing the e/D from 0.17
increased strains in the HSC columns are due to the larger to 0.26, and further to 0.34 and 0.43, decreased the spiral
column capacity. Moreover, the increase in strains associated strains by 29.0, 50.0, and 55.2%, respectively, compared to
with increasing concrete strength was greater for columns the specimen with e/D = 0.17.

Fig. 7—Lateral displacement of axially loaded columns. (Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.)
Fig. 8—Axial displacement of axially loaded columns. (Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.)
Moreover, as the e/D increased, the lateral displacement further to 0.43, respectively. This behavior was expected, as
of the columns increased. As shown in Fig. 7, the maximum increasing the eccentricity was shifting the response of the
displacement for column S28-17 (e/D = 0.17) was 8.87 mm. column to a more flexural behavior. Because there was more
Increasing the e/D to 0.26, 0.34, and 0.43 resulted in an applied bending moment, the column showed much larger
increase in the lateral displacement by 10.0, 74.0, and deformations.
175.0%, respectively, with reference to column S28-17. The Similarly, the load-axial displacement relationship was
shorter specimens, S14-17, S14-26, and S14-34, showed linear up to failure, as shown in Fig. 8. The axial displace-
a lateral displacement of 2.36, 3.75, and 5.30 mm, respec- ment at the peak load decreased as the e/D increased. The
tively. Accordingly, the measured lateral stiffness decreased axial stiffness of the short columns decreased by 13.0 and
with increasing the e/D. For short columns, the lateral stiff- 32.0% when the e/D increased from 0.17 to 0.26 and 0.34,
ness decreased by 8.3 and 17.0% as the e/D increased from respectively. For slender columns, increasing the e/D from
0.17 to 0.26 and further to 0.34, respectively. The decrease in 0.17 to 0.26, 0.34, and further to 0.43 resulted in a reduction
the lateral stiffness for slender columns was 15.0, 57.0, and in the axial stiffness by 10.0, 43.0, and 60.0%, respectively.
81.0% when the e/D increased from 0.17 to 0.26, 0.34, and

Effect of slenderness ratio (e/D = 0.17) experienced a lateral displacement of 2.36 mm
The increase of λ from 14 to 28 had a very significant effect at the peak load, whereas the counterpart slender column
on the behavior of the columns. Such an increase in the slen- (S28-17) achieved 2.86 mm at only 50% of the peak capacity.
derness ratio caused a decrease in the peak axial capacity Thereafter, column S28-17 reached a lateral displacement
of the columns by 20.4, 21.0, and 22.2% for columns with of 8.87 mm at the peak load, which represents a 275.0%
e/D of 0.17, 0.26, and 0.34, respectively. This is due to the increase. Similarly, for e/D of 0.26 and 0.34, the increase in
slender columns exhibiting higher lateral displacements lateral displacements due to doubling the slenderness ratio
throughout the loading, which translate to a higher secondary were 161.0 and 194.0%, respectively, at the peak load.
moment, thus reducing the axial capacity of the specimens.
The decrease in the axial capacity was accompanied by a Flexural test
decrease in the total moment capacity of 12.4, 16.8, and In this study, a specimen was tested under pure flexural
15.6% for e/D of 0.17, 0.26, and 0.34, respectively. Because loading in a four-point bending setup, as shown in Fig. 2.
the primary moment capacity is a direct product of the axial The specimen showed initial cracks at a load of 32 kN.
capacity and load eccentricity, the primary moment would, With the increasing load, more cracks formed. When the
therefore, decrease with increasing slenderness due to the load reached 80 kN, the cracks began to slightly widen.
low axial load achieved by the slender columns. However, The load increased steadily until the concrete cover in the
slender columns with e/D of 0.17, 0.26, and 0.34 showed compression zone of the column spalled. The lateral stiff-
an increase in the secondary moment by 172.0, 102.6, and ness of the column reduced because of the formed cracks.
146.0% compared to that of their short counterparts. This Nevertheless, the column continued to carry the load until
shows that the increased lateral displacement of the column reaching a maximum load of 318 kN, at which the deflec-
caused additional moments on the column and thus reduces tion was 71.3 mm. The peak load represented a moment
the axial capacity. carrying capacity of 150 kN∙m. After the peak load, the spec-
Despite the change in the slenderness ratio, the strain that imen lost its capacity gradually with increasing deflection
developed in the bar on the compression side at peak loads until the test was halted. As shown in Fig. 9, the strain in
for different e/D was very similar (Fig. 4). These strains the compression concrete before the initial spalling of the
were in good agreement with the concrete compressive concrete cover was –3750 με. After the concrete spalled, the
strain measured during the test. The values ranged between concrete strain gauge malfunctioned. However, the strain
–2400 and –3050 με, which are slightly lower for the slender in the GFRP bar in the compression side kept increasing to
columns compared to their short counterparts. The developed a strain of –12,385 με, while a tensile strain of more than
compressive strains suggest they have an active contribu- 15,000 με was measured on the tension side. The bars and
tion to the axial capacity in both short and slender columns. spirals showed no sign of failure or rupture.
On the other hand, short columns with higher eccentricities
experienced lower tensile strains in the bar on the tension Deformability of axially loaded GFRP-RC columns
side of the column compared to their slender counterparts. Elchalakani et al. (2020) introduced a procedure to calcu-
However, columns with e/D of 0.17 experienced relatively late the deformability of GFRP-RC columns tested under
low lateral displacements compared to those in columns with eccentric axial loads. This method is based on the energy
a higher e/D; the measured tensile strain in the bars was very absorption of the column specimens after the first peak
small. The strain in the tension side bar in specimens S28-17 load, which was a modified version of a method developed
and S14-17 was 480 and 550 με, respectively. However, for by Foster and Attard (1997). The deformability index for
higher e/D, the difference became larger, which was consis- GFRP-RC columns was calculated as the area under the
tent with the increased lateral displacement and, conse- load-axial displacement curve up to 3Δ75 or up to the ultimate
quently, the secondary moments in the slender columns. As axial displacement divided by the area under the load-axial
more moment was exerted on the columns with e/D 0.26 and displacement curve up to Δ75, where Δ75 is the corresponding
0.34, the deflected shape became more exaggerated. axial displacement obtained by extending the linear-elastic
In addition, the spiral strain was reduced as the slender- range (from 0 to 75% of the peak load) to the peak load. More
ness ratio increased. Increasing the slenderness ratio from details about this method can be found in Elchalakani et al.
14 to 28 reduced the spiral strain by 28.7, 20.5, and 16.4% (2020). In this study, the ultimate axial displacement was
in columns with e/D of 0.17, 0.26, and 0.34, respectively. considered in the calculations of the deformability index.
Again, this is due to the higher axial loads achieved by the For short columns, S14-17, S14-26, and S14-34, the deform-
short columns that in turn increased the confining pressure ability index was 1.52, 2.11, and 2.31, respectively. Simi-
on the spiral. larly, the deformability index was 1.03, 1.07, 1.2, and 1.31
The axial stiffness across the three e/D was significantly for slender columns S28-17, S28-26, S28-34, and S28-43,
impacted when λ increased from 14 to 28. For the same respectively. It should be noted that the deformability index
e/D, columns with e/D of 0.17, 0.26, and 0.34 experienced a increased as the e/D increased, which can be attributed to
reduction in axial stiffness by 38.0, 37.0, and 45.0%, respec- the low axial load capacity and the reduced stiffness due to
tively. Similarly, the lateral displacement of the columns was the developed tension cracks in the columns as the eccen-
affected by the increased slenderness ratio. The lateral stiff- tricity increased in both the short and slender columns. On
ness decreased by 20.0, 29.0, and 58.0% for columns with the other hand, increasing the slenderness ratio from 14 to 28
e/D of 0.17, 0.26, and 0.34, respectively. Column S14-17 resulted in a reduction in the deformability index by 32, 49,

Fig. 9—Flexural specimen: (a) extreme tension and compression bar strains; (b) concrete strain; (c) midspan displacement;
and (d) spiral strain. (Note: 1 kN = 0.225 kip; 1 mm = 0.0394 in.)
and 48% in columns with e/D of 0.17, 0.26, and 0.34, respec- model developed by Thorenfeldt et al. (1987) for unconfined
tively. Again, this was due to the low axial load capacity and HSC was used. This model was also employed in previous
the reduced stiffness because of the higher intensity of the literature, such as Hadi et al. (2017), to accurately determine
tension cracks as the slenderness ratio increased. the stresses in the HSC section. The analysis of the column
section assumes there is a perfect bond between the concrete
EXPERIMENTAL AND CODE-PREDICTED and reinforcement, the concrete in tension is neglected, and
INTERACTION DIAGRAMS the plane sections remain plane after bending (Park and
The specimens in this study were designed to satisfy the Paulay 1975). Therefore, the strain along the cross section
provisions of the Canadian standard CSA S806-12 (2017). and the strain in the reinforcement layers are proportional
The columns were subjected to different loading eccen- to the depth of the natural axis. To increase the accuracy of
tricities in addition to a specimen tested under pure flex- the integration, a 1 mm thickness of the strips was used. The
ural loading. Using the different loading conditions, an stress in each concrete strip, f sn in Fig. 10(c), is defined as
axial load-moment interaction diagram was developed.
The capacities of the columns according to Canadian stan- f c c xr
dard CSA S806-12 (2017) and American guideline ACI f sn (1)
r 1 x kr
440.1R-15 (ACI Committee 440 2015) were calculated
using a strip analysis of the column section. This method
consisted of dividing the column cross section into small c
x (2)
rectangular strips to integrate using a finite number of strips 0
(Fig. 10(a)). Thereafter, assuming a linear strain distribu- where εc is the corresponding strain to a concrete stress fc;
tion, the stress in each concrete strip was calculated. After and ε0 is that corresponding to the maximum concrete stress
calculating the width of the individual strips and using fc′. The remaining factors r and k are curve-fitting and slope
a predetermined thickness for the strips, the force in each control factors for the stress-strain curve, respectively, and
strip and the corresponding force in the reinforcement can can be defined as follows.
be determined and used to calculate the axial and moment
capacity of the section. Due to the difference between NSC
f c r
and HSC, the stress-strain relationship differs. Therefore, a 0 (3)
Ec r 1

Fig. 10—Strip section analysis.
Fig. 11—Normalized load-moment interaction diagram.
f predicted interaction diagrams and the experimental interac-

r 0.8 c (4) tion diagrams plotted for each slenderness ratio. The interac-
17 tion diagrams were normalized, where the y-axis represents
the normalized axial load (Kn) and the x-axis represents the
c normalized bending moment (Rn)
for 1.0 , k = 1.0, and for c 1.0 ,
0 0
Pn
f Kn (6)
k 0.67 c 1.0 (5) Ag f c
62
The modulus of elasticity, Ec, was calculated according Mn
Rn (7)
to the respective code or guideline. This method was used Ag f c D
to calculate the interaction diagrams predicted by ACI
440.1R-15 (ACI Committee 440 2015), CSA S6-19 (2019),
M n Pn (e ) (8)
and CSA S806-12 (2017). Differences in each code were
considered, such as the maximum concrete compressive
where Pn is the peak load achieved by the column; and δ
strain (0.003 for the ACI guideline and 0.0035 for the CSA
is the maximum lateral displacement corresponding to
standards). Additionally, because CSA S806-12 (2017) and
the peak load. The nominal moment Mn is defined as the
ACI 440.1R-15 (ACI Committee 440 2015) do not allow for
peak load multiplied by the applied eccentricity in addi-
the compressive capacity of the GFRP bars to be considered,
tion to the maximum lateral displacement; and Ag, fc′, and
their contribution has been excluded for the corresponding
D are the gross area of the section, the concrete compres-
interaction diagram. On the other hand, CSA S6-19 (2019)
sive strength, and the diameter of the column cross section,
considers the contribution of the GFRP bars with a limited
respectively. Both the slender and short columns exhibited a
compressive strain in the bars of 0.002. Figure 11 shows the
“knee”-shaped interaction diagram, which is similar to that

of the code predictions in the range of the e/D used in this reinforced internally with glass fiber-reinforced polymer (GFRP) bars and
spirals.
study. In addition, the experimental Kn and Rn were higher
than the code predictions for all e/D despite neglecting any Karam Mahmoud is a Postdoctoral Fellow in the Department of Civil
reductions in terms of the slenderness effect in the predicted Engineering at the University of Manitoba. He is an Assistant Professor (on
leave) in the Department of Civil Engineering at Assiut University, Assiut,
diagrams. The slenderness effect is represented by the Egypt. His research interests include the design, experimental testing, and
disparity between the interaction diagrams for the slender finite element modeling of concrete structures reinforced/strengthened with
and short columns. On both accounts, the predictions of FRP materials.
different codes are conservative, with ACI 440.1R-15 (ACI Ehab F. El-Salakawy, FACI, is a Professor in the Department of Civil Engi-
Committee 440 2015) giving the closest predictions at the neering at the University of Manitoba. He is a member of ACI Committee
tested e/D near the balance point in the interaction diagram. 440, Fiber-Reinforced Polymer Reinforcement. He is the recipient of the
2020 ACI Mete A. Sozen Award for Excellence in Structural Research. His
However, CSA S6-19 (2019) yielded better predictions research interests include the design, construction, durability, large-scale
toward the pure moment (flexure) side of the interaction testing, and rehabilitation of reinforced concrete structures using FRP
diagram because the compressive strains in the GFRP bars reinforcement.
are the determining factor of the capacities in this case.
ACKNOWLEDGMENTS
The authors would like to extend their deepest gratitude to the Natural
CONCLUSIONS Sciences and Engineering Research Council of Canada (NSERC) for their
The following conclusions can be drawn based on the test financial support. The GFRP material was generously provided by Pultrall
Inc. The technical staff at the W. R. McQuade Structures Laboratory at the
results of the eight full-size high-strength concrete (HSC) University of Manitoba are acknowledged for their tremendous effort and
circular columns reinforced with glass fiber-reinforced assistance.
polymer (GFRP) bars and spirals:
1. The mode of failure in all the specimens can be cate- NOTATION
gorized as material-type failures. The columns failed by Ag = gross area of column cross section
D = diameter of column cross section
spalling and crushing of the concrete on the compression dN.A = depth of neutral axis
side. With lower eccentricities, cracks on the tension side dsn = depth of centroid of concrete strip “n”
were observed near failure. However, as the eccentricity e = load eccentricity
Ffrp = force in FRP layer “n”
increased, tension cracks formed at earlier stages due to the Fs
n
= force in concrete strip “n”

increased bending moment and excessive lateral deflections. n
fc = stress in concrete
2. As the eccentricity-to-diameter ratio (e/D) increased, the fc′ = maximum concrete stress obtained from testing standard
concrete cylinders
axial load capacity was greatly reduced. This was observed fs = stress in concrete strip “n”
for both the short and slender columns. n
Kn = normalized axial load
3. Based on the measured strains, the longitudinal bars k = effective length factor (controlled by column boundary
conditions)
have actively contributed to the axial capacity of the column. ℓ = unbraced length of column
The strains in the GFRP bar on the compression side reached Mn = nominal moment
approximately the same values observed in the concrete. Mu = experimental ultimate moment
Pn = nominal axial load
Furthermore, only the short column under e/D = 0.17 showed Pu = experimental ultimate axial load
failure of the GFRP bar and spiral on the compression side. Rn = normalized bending moment
4. As the slenderness ratio increased, the axial capacity r = radius of gyration
Δ75 = corresponding axial displacement obtained by extending linear-
of the column decreased for all e/D. Although the primary elastic range (from 0 to 75% of peak load) to peak load
moment capacity decreased in the slender columns, the δ = lateral deflection of column
secondary moment showed a significantly large increase that εc = strain in concrete
ε0 = strain in concrete corresponding to fc′
could reach 172%. λ = slenderness ratio, defined as kℓ/r
5. The column under pure flexural loading reached a ρf = GFRP longitudinal reinforcement ratio
maximum load of 318 kN, which represents a moment of ρfs = GFRP transverse reinforcement ratio
150 kN∙m. No signs of failure were observed in the GFRP
bars or spirals. The GFRP bars were able to develop substan- REFERENCES
Abdelazim, W. M. A., 2020, “Behavior of Slender Concrete Columns
tial compressive and tensile strains, indicating active contri- Reinforced with GFRP-Bars And Spirals Under Concentric and Eccentric
bution to the specimen capacity. Loads,” PhD thesis, Department of Civil Engineering, Université de Sher-
6. The interaction diagrams showed that the capacity of brooke, Sherbrooke, QC, Canada, 243 pp.
ACI Committee 222, 2019, “Guide to Protection of Metals in Concrete
both the short and slender columns exceeded the predictions Against Corrosion (ACI 222R-19),” American Concrete Institute, Farm-
of the Canadian standards (CSA S806-12 [2017] and CSA ington Hills, MI, 60 pp.
S6-19 [2019]) and the American guidelines (ACI Committee ACI Committee 318, 2019, “Building Code Requirements for Struc-
tural Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
440 2015), with CSA S6-19 (2019) and ACI 440.1R-15 Concrete Institute, Farmington Hills, MI, 624 pp.
(ACI Committee 440 2015) giving the closest predictions at ACI Committee 440, 2015, “Guide for the Design and Construction of
different regions of the interaction diagram. Structural Concrete Reinforced with Fiber-Reinforced Polymer (FRP) Bars
(ACI 440.1R-15),” American Concrete Institute, Farmington Hills, MI,
88 pp.
AUTHOR BIOS Afifi, M. Z.; Mohamed, H. M.; and Benmokrane, B., 2014, “Axial
ACI member Mu’taz Almomani is an MSc Student at the University of Capacity of Circular Concrete Columns Reinforced with GFRP Bars and
Manitoba, Winnipeg, MB, Canada. He received his BSc in civil engineering Spirals,” Journal of Composites for Construction, ASCE, V. 18, No. 1, Feb.,
from the Jordan University of Science and Technology, Irbid, Jordan. His p. 04013017. doi: 10.1061/(ASCE)CC.1943-5614.0000438
research interests include the behavior of reinforced concrete columns

Ali, M. A., and El-Salakawy, E., 2016, “Seismic Performance of Ghomi, S. K., and El-Salakawy, E., 2016, “Seismic Performance of
GFRP-Reinforced Concrete Rectangular Columns,” Journal of Composites GFRP-RC Exterior Beam–Column Joints with Lateral Beams,” Journal of
for Construction, ASCE, V. 20, No. 3, June, p. 04015074. doi: 10.1061/ Composites for Construction, ASCE, V. 20, No. 1, Feb., p. 04015019. doi:
(ASCE)CC.1943-5614.0000637 10.1061/(ASCE)CC.1943-5614.0000582
Bakht, B.; Mufti, A. A.; and Tadros, G., 2004, “Discussion of ‘Fibre- Hadhood, A., 2017, “Behavior, Strength and Flexural Stiffness of
Reinforced Polymer Composite Bars for the Concrete Deck Slab of Wotton Circular Concrete Columns Reinforced with FRP Bars and Spirals/Hoops
Bridge’,” Canadian Journal of Civil Engineering, V. 31, No. 3, June, under Eccentric Loading,” PhD thesis, Department of Civil Engineering,
pp. 530-531. Université de Sherbrooke, Sherbrooke, QC, Canada, 302 pp.
Barua, S., and El-Salakawy, E., 2020, “Performance of GFRP-Rein- Hadi, M. N. S.; Hasan, H. A.; and Sheikh, M. N., 2017, “Experimental
forced Concrete Circular Short Columns under Concentric, Eccentric, and Investigation of Circular High-Strength Concrete Columns Reinforced with
Flexural Loads,” Journal of Composites for Construction, ASCE, V. 24, Glass Fiber-Reinforced Polymer Bars and Helices under Different Loading
No. 5, Oct., p. 04020044. doi: 10.1061/(ASCE)CC.1943-5614.0001058 Conditions,” Journal of Composites for Construction, ACSE, V. 21, No. 4,
Barua, S.; Mahmoud, K.; and El-Salakawy, E., 2021, “Slender GFRP-RC Aug., p. 04017005. doi: 10.1061/(ASCE)CC.1943-5614.0000784
Circular Columns under Concentric, Eccentric, and Flexural Loads: Experi- Hales, T. A.; Pantelides, C. P.; and Reaveley, L. D., 2016, “Exper-
mental Investigation,” Journal of Bridge Engineering, ASCE, V. 26, No. 7, imental Evaluation of Slender High-Strength Concrete Columns with
July, p. 04021033. doi: 10.1061/(ASCE)BE.1943-5592.0001729 GFRP and Hybrid Reinforcement,” Journal of Composites for Construc-
Broms, B., and Viest, I. M., 1961, “Long Columns: Ultimate Strength tion, ASCE, V. 20, No. 6, Dec., p. 04016050. doi: 10.1061/(ASCE)
of Hinged Columns,” Transactions of the American Society of Civil Engi- CC.1943-5614.0000709
neers, ASCE, V. 126, No. 2, Jan., pp. 308-400. Hossain, K. M. A.; Ametrano, D.; and Lachemi, M., 2014, “Bond
CSA A23.1-19/CSA A23.2-19, 2019, “Concrete Materials and Methods Strength of Standard and High-Modulus GFRP Bars in High-Strength
of Concrete Construction/Test Methods and Standard Practices for Concrete,” Journal of Materials in Civil Engineering, ACSE, V. 26, No. 3,
Concrete,” Canadian Standards Association, Toronto, ON, Canada, 882 pp. Mar., pp. 449-456.
CSA S6-19, 2019, “Canadian Highway Bridge Design Code,” Canadian Khorramian, K., and Sadeghian, P., 2020, “Experimental Investigation
Standards Association, Toronto, ON, Canada, 1182 pp. of Short and Slender Rectangular Concrete Columns Reinforced with
CSA S806-12 (R2017), 2017, “Design and Construction of Building GFRP Bars under Eccentric Axial Loads,” Journal of Composites for
Structures with Fibre-Reinforced Polymers,” Canadian Standards Associ- Construction, ASCE, V. 24, No. 6, Dec., p. 04020072. doi: 10.1061/(ASCE)
ation, Toronto, ON, Canada, 206 pp. CC.1943-5614.0001088
CSA S807-19, 2019, “Specification for Fibre-Reinforced Polymers,” Mahmoud, K., and El-Salakawy, E., 2016, “Size Effect on Shear Strength
Canadian Standards Association, Toronto, ON, Canada, 67 pp. of Glass Fiber-Reinforced Polymer-Reinforced Concrete Continuous
De Luca, A.; Matta, F.; and Nanni, A., 2010, “Behavior of Full-Scale Beams,” ACI Structural Journal, V. 113, No. 1, Jan.-Feb., pp. 125-134. doi:
Glass Fiber-Reinforced Polymer Reinforced Concrete Columns under Axial 10.14359/51688065
Load,” ACI Structural Journal, V. 107, No. 5, Sept.-Oct., pp. 589-596. Mirmiran, A.; Yuan, W.; and Chen, X., 2001, “Design for Slenderness
El-Gendy, M. G., and El-Salakawy, E., 2016, “Effect of Shear Studs in Concrete Columns Internally Reinforced with Fiber-Reinforced Polymer
and High Moments on Punching Behavior of GFRP-RC Slab–Column Bars,” ACI Structural Journal, V. 98, No. 1, Jan.-Feb., pp. 116-125.
Edge Connections,” Journal of Composites for Construction, ASCE, V. 20, Park, R., and Paulay, T., 1975, Reinforced Concrete Structures, John
No. 4, Aug., p. 04016007. doi: 10.1061/(ASCE)CC.1943-5614.0000668 Wiley & Sons, Inc., Hoboken, NJ.
Elchalakani, M.; Dong, M.; Karrech, A.; Sadakkathulla, M. A.; and Huo, Rahman, S. M. H.; Mahmoud, K.; and El-Salakawy, E., 2017, “Behavior
J.-S., 2020, “Circular Concrete Columns and Beams Reinforced with GFRP of Glass Fiber–Reinforced Polymer Reinforced Concrete Continuous
Bars and Spirals under Axial, Eccentric, and Flexural Loading,” Journal of T-Beams,” Journal of Composites for Construction, ASCE, V. 21, No. 2,
Composites for Construction, ASCE, V. 24, No. 3, June, p. 04020008. doi: Apr., p. 04016085. doi: 10.1061/(ASCE)CC.1943-5614.0000740
10.1061/(ASCE)CC.1943-5614.0001008 Thorenfeldt, E.; Tomaszewicz, A.; and Jensen, J. J., 1987, “Mechanical
Foster, S. J., and Attard, M. M., 1997, “Experimental Tests on Eccentri- Properties of High Strength Concrete and Application to Design,” Proceed-
cally Loaded High-Strength Concrete Columns,” ACI Structural Journal, ings, Utilization of High-Strength Concrete: Symposium, Stavanger,
V. 94, No. 3, May-June, pp. 295-302. Norway, pp. 149-159.

in
Your
Classroom
Integrate into
your classroom!
To support future leaders, ACI has launched several initiatives to engage
students in the Institute’s activities and programs – select programs that
may be of interest to Educators are:
• Free student membership – encourage • Scholarships and fellowships – students
students to sign up who win awards are provided up to $15,000
• Special student discounts on ACI 318 and may be offered internships and paid
Building Code Requirements for Structural travel to attend ACI’s conventions
Concrete, ACI 530 Building Code Require- • ACI Award for University Student Activities –
ments and Specification for Masonry receive local and international recognition
Structure, & Formwork for Concrete manual. for your University’s participation in concrete-
• Access to Concrete International – free to related activities
all ACI student members • Free access to the ACI Collection of Con-
• Access to ACI Structural Journal and ACI crete Codes, Specifications, and Practices –
Materials Journal – free to all ACI student in conjunction with ACI’s chapters, students
members are provided free access to the
• Free sustainability resources – free copies online ACI Collection
of Sustainable Concrete Guides provided to • ACI online recorded web sessions and
universities for use in the classroom continuing education programs – online
• Student competitions – participate in ACI’s learning tools ideal for use as quizzes or
written and/or team-based competitions in-class study material
https://www.concrete.org/educatorsandresearchers/aciinyourclassroom.aspx
Title No. 119-S47
Moment-Curvature-Deformation Response of Post-

Tensioned Ultra-High-Performance Concrete Beams
by Mehmet Dogu and Fatmir Menkulasi
A mechanics-based analytical method for obtaining the complete others15-20 provide guidance for how to obtain the complete
moment-curvature-deformation response of ultra-high-perfor- flexural response. While the need to predict the complete
mance concrete (UHPC) beams post-tensioned with internal flexural response of PT NSC members has been met, there
unbonded tendons is presented. The proposed procedure does not is currently no efficient methodology to predict the complete
rely on empiricism other than what is included in the assumed
flexural response of PT UHPC members whose behavior is
material constitutive models and provides the means to determine
significantly different from that of their NSC counterparts.
the variation of curvature and deflection as the beam is loaded
to failure, thus providing an avenue to quantify ductility at the Dogu and Menkulasi21 recently presented a flexural design
cross section and member level. The influence of various consti- methodology for PT UHPC beams. However, the goal of the
tutive models for the compressive and tensile domain of UHPC, study was limited to offering an approach for predicting
prestressed and non-prestressed reinforcement ratio, loading the ultimate flexural capacity of such members, which is of
configuration, and tendon profile on the complete beam flexural interest for collapse prevention but does not provide insight
behavior is quantified. The most influential parameters for cross in terms of their complete flexural response from the onset of
section- and member-level ductility are the ultimate UHPC strain loading to failure. Nonlinear finite element analysis (NFEA)
in tension and loading configuration, respectively. can be used to obtain the complete flexural response of PT
UHPC beams; however, the creation and analysis of reliable
Keywords: ductility; moment-curvature-deformation; post-tensioning;
ultra-high-performance concrete (UHPC); unbonded tendons. models even for simple beams requires considerable time.
The goal of the research provided in this paper is to present
INTRODUCTION an efficient analytical method for obtaining the complete
Post-tensioned (PT) concrete flexural members have moment-curvature-deformation response of PT UHPC
the ability to offer enhanced rentable space, architectural beams by taking into consideration the lack of bond between
freedom, and adaptable structures by offering floor systems tendons and concrete and the unique domains of UHPC in
that feature very few columns. This characteristic has been tension and compression. The proposed method can capture
well explored in the construction industry, and PT concrete the variation of strand stress, curvature, and deflection at
floors with unbonded tendons are the system of choice in various stages of loading without requiring the definition of
many midrise and high-rise building structures. The ability to empirically obtained elastic and inelastic design parameters
offer slender structures in general and slender floor systems in such as bond reduction coefficients and plastic hinge length.
particular can be further enhanced if the PT flexural member However, it can facilitate the use of approaches that employ
is constructed with ultra-high performance fiber-reinforced these parameters by helping formulate them without having
concrete, herein referred to as UHPC, which in addition to to rely on extensive experimental testing or time-con-
its high compressive strength also offers considerable tensile suming finite element analysis. This study addresses simply
strength. According to the Federal Highway Administration supported PT UHPC beams with a rectangular cross section
(FHWA),1 “UHPC-class materials are cementitious based that features straight and draped tendons.
composite materials with discontinuous fiber reinforcement,
compressive strengths above 22 ksi (150 MPa), tensile RESEARCH SIGNIFICANCE
strengths above 0.7 ksi (5 MPa), and enhanced durability via PT UHPC members can offer span-depth ratios that are
their discontinuous pore structure.” In this study, the FHWA1 beyond the capabilities of their NSC counterparts. There is
definition for UHPC is adopted, although the variability of currently no efficient approach to predict the complete flex-
key parameters in the constitutive model is considered to ural response of UHPC beams post-tensioned with internal
evaluate their impact on PT beam behavior. unbonded tendons to gravity-induced loading. An analytical
The determination of the complete flexural response of procedure capable of supplying this response is presented.
a PT member is related to the determination of unbonded The proposed procedure is non-empirical and can be used
strand stress at various stages of loading. Various approaches to characterize cross section and member level behavior in
have been used to determine this stress and consequently the
complete flexural response for normal-strength concrete MS No. S-2021-151.R2, doi: 10.14359/51734344, received September 16, 2021, and
(NSC) members. Some2-14 include the determination of reviewed under Institute publication policies. Copyright © 2022, American Concrete
strand stress at discrete points and limited ranges such obtained from the copyright proprietors. Pertinent discussion including author’s
as ultimate limit state and service, respectively, whereas is received within four months of the paper’s print publication.

Fig. 1—Typical PT UHPC beam detail considered in moment-curvature and load-deformation response analysis.
terms of moment-curvature and load-deformation response moment, Mcracking. This is a straightforward task when the
from the onset of loading to the attainment of peak load tendon is straight; however, for draped configurations, it
provided that material behavior is characterized through the requires an iterative procedure because the location of the
specification of key parameters. cracking moment depends on the effective depth as well as
the applied loading configuration. Additionally, the demar-
PROCEDURE FOR COMPUTING MOMENT cation between cracked and uncracked regions will shift as
CURVATURE DEFORMATION RESPONSE the applied moment intensity increases.
The rectangular UHPC beam shown in Fig. 1 is taken as The proposed procedure requires the specification of
a prototype to develop the proposed procedure. The beam key parameters to supply the stress-strain relationship for
contains bonded compression and tension mild steel, As′ and all materials. The assumed stress-strain relationship for
As, respectively, and unbonded PT tendons, Aps. The beam each material is shown in Fig. 2. The stress-strain curve in
features a span-depth ratio of 28, which is within the typical compression for UHPC up until the attainment of peak stress
range of 26 to 30 for PT NSC members, even though PT is based on recommendations by Graybeal.22 Cracking stress
UHPC beams have the potential to provide much higher and modulus of elasticity for UHPC are based on guide-
span-depth ratios. The size and spacing of transverse rein- lines provided by Russel and Graybeal.1 The key parame-
forcement is assumed to not affect the flexural response, and ters used to define the compressive and tensile domains of
therefore, the provision of this information is not required. In UHPC can be adjusted to reflect the use of various UHPC
addition, the beams are assumed to be sufficiently reinforced formulations including different fiber types and character-
for shear such that the governing failure mode is flexure. istics. The stress-strain curve for the prestressing strands is
The proposed procedure includes analysis conducted based on recommendations by Delavapura and Tadros,23 and
at the member level as well as at the cross section level. mild steel is assumed to be elastic perfectly plastic. Several
Member-level analysis is required due to the lack of bond controls are introduced to distinguish beams that exhibit
between the tendons and the surrounding concrete, which a fiber tension-controlled failure from those that exhibit a
does not allow the application of a cross section-level UHPC compression-controlled failure. Effective prestress,
analysis based on strain compatibility. Therefore, compati- fpe, is considered as an input; thus, it is assumed that the
bility is enforced at the member level by equating the total designer has a methodology available to estimate short-term
change in length in the concrete fiber at the centroid of and long-term losses. Guidance on how to estimate losses
tendons to the total change in tendon length. This member- in pretensioned UHPC beams is provided by Graybeal24
level condition of compatibility is enforced before the beam and John et al.25 The proposed methodology outlines how
cracks as well as after the beam cracks. to calculate the history of the change in strand strain, Δεps,
The proposed procedure covers pre- and post-cracking so that this change can be added to the effective prestrain,
behavior. Naturally, a cracked beam will have portions that εpe, to calculate the total strand strain εps, and consequently,
have cracked as well as portions that have not cracked. the force in the tendons Ftendon, internal moment Mn, cross-
The challenge in this case is to demarcate the cracked and sectional curvature φ, total load, and midspan displace-
uncracked regions so that the appropriate analysis approach ments. The determination of long-term deformations, which
can be followed in each case. This demarcation involves the is typically addressed using multiplier-based methods or
determination of the point along the length of the member more advanced incremental time step analysis,26 is outside
where the applied moment, Mapplied, exceeds the cracking the scope of work.

Fig. 2—Stress-strain relationship for UHPC in: (a) compression; (b) tension; (c) prestressing steel; and (d) mild steel.
Before cracking prestrain, εpe, to calculate the total strain in the strands, εps_total
The flowchart for obtaining the moment-curvature and (Eq. (12)). Using the stress-strain curve for prestressing
load-deformation response of an uncracked PT UHPC strands (Fig. 2(c)), the total strain is used to obtain the total
beam is provided in Fig. 3. Half of the beam is taken as one strand stress, fps (Eq. (13)). The effective prestress, fpe, is then
section. No further discretization or meshing is required subtracted from this total stress to determine the change in
because numerical integration using the Gaussian quadrature strand stress, Δfps,calculated (Eq. (14)), which is compared to the
rule27 is used to account for the variation of moment along value assumed at the beginning (Eq. (15)). If the two values
the span when calculating strand strain, εps, and midspan match, then moment, curvature, total load, and midspan
beam displacement. A change in strand stress, Δfps,assumed, is deflection are reported, and the procedure is terminated.
initially assumed and added to the effective prestress, fpe, to Total load is calculated using the loading configuration in
calculate the total strand stress, fps (Eq. (1)), and the total question and the magnitude of the moment at midspan; while
force in the tendon, Ftendon (Eq. (2)). Using this tendon force, midspan deflection is calculated using the obtained curvature
the cracking moment, Mcracking (Eq. (3)), at midspan is calcu- values at Gauss points and the second moment area theorem
lated, and the magnitude of the moment at the rest of the (Eq. (16)). If the assumed and calculated changes in tendon
Gauss points is computed by scaling the midspan moment stress do not match, the iteration continues with an assumed
considering the loading configuration and its magnitude at change set equal to that calculated in the previous iteration
midspan. until convergence occurs.
For every Gauss point, the curvature, ϕ (Eq. (4) to (6)),
and the change in strain in concrete at the centroid of After cracking
tendons, Δεconcrete_elastic (Eq. (7)), is computed. Then, the The flowchart for obtaining the moment-curvature and
strain in concrete at the centroid of tendons is numerically load-deformation response of a cracked PT UHPC beam is
integrated along the length of the beam to calculate the total provided in Fig. 4. First, the section at midspan is considered
change in fiber length at the centroid of tendons, ΔLelastic and is assigned a number (that is, S = 1 at midspan). Then
(Eq. (8)). This change is set equal to the total change in the maximum usable strain in tension and compression for
tendon length to enforce member-level deformation compat- all materials considered are stored as this information will be
ibility. The change in tendon length is then divided by the used to validate the failure mode, which initially is assumed
original tendon length, L, to calculate the change in tendon to be a fiber tension-controlled failure.21 A maximum
strain, Δεps_elastic (Eq. (9)), which is added to the decompres- compressive strain for UHPC at midspan is then assumed to
sion strain, Δεdecompression (Eq. (10) to (11)), and effective determine the point, up until which the procedure is desired

Fig. 3—Flowchart used to obtain moment-curvature-deformation response in an uncracked PT UHPC beam.
to be implemented. Then, an initial value for the top (Eq. (2)). Knowing the force in the tendon and an anchor
fiber compressive strain in UHPC at midspan, εc,s=1, is point in the strain diagram, force equilibrium can be used
selected together with an assumed change in strand stress, to iteratively determine the depth to the neutral axis, c, and
Δfps,assumed. The change in tendon stress is used to determine internal moment at midspan. The location of the first crack in
the total stress, fps (Eq. (1)), and the force in the tendon, Ftendon the beam is subsequently determined, and this information is

Fig. 4—Flowchart used to obtain moment-curvature-deformation response in a cracked PT UHPC beam.
used to distinguish cracked regions, Lcracked, from uncracked section, the combination of concrete top fiber strain, εtop,c,
regions (Eq. (17)), Lelastic. The cracked region (Eq. (18)) is and neutral axis depth c that results in force and moment
divided into an odd number of sections, n, to facilitate coef- equilibrium is iteratively determined. This information is
ficient symmetry when implementing Simpson’s rule for then used to calculate curvature, ϕ, at every section as well
numerical integration. This creates an even number (n-1) as the change in strain in concrete at the centroid of tendons,
of beam elements (segments), dLcracked (Eq. (19)). For each Δεconcrete_cracked (Eq. (20)), and decompression strain,

Δεdecompression (Eq. (21)). The uncracked region is consid- to correctly capture the behavior of UHPC beams prestressed
ered as one section. The change in concrete fiber length at with bonded tendons as well as the behavior of NSC beams
the centroid of tendons due to decompression, ΔLdecompression with unbonded tendons should be able to correctly capture
(Eq. (10)), and due to subsequent loading, ΔLcracked (Eq. (22)), the behavior of UHPC beams posttensioned with unbonded
is calculated by integrating the strain in the concrete at tendons. Figure 5 shows the comparison between results
the level of tendons along the length of the beam. While obtained using the proposed procedure and those obtained
Gauss-Legendre quadrature is used to integrate strain in the using validated nonlinear finite element analysis. The results
uncracked region, Simpson’s 1/3 rule is used to conduct the agree well at various levels. Figure 5(a) illustrates the
numerical integration in the cracked region because Gauss behavior at the member level in terms of total load versus
quadrature cannot be used since the method requires the mid-span displacement. Figure 5(b) illustrates cross-sec-
integrated function to be continuous rather than discrete to tional level behavior in terms of moment versus curvature
facilitate integration at the predetermined Gauss points. at midspan, and Fig. 5(c) and (d) illustrate the variation of
Once the total change in fiber length at the centroid of tendon stress with respect to mid-span displacement. The
tendons is computed, a similar procedure to that used prior variation in tendon stress with respect to midspan displace-
to cracking is used to calculate the change in tendon strain ment is illustrated in terms of the total tendon stress (Fig.
due to decompression (Eq. (11)) and subsequent loading 5(c)) as well as the change in tendon stress (Fig. 5(d)). The
(Eq. (23)), total strand strain (Eq. (24)) and stress (Eq. (13)), close agreement between NFEA results and those obtained
and change in strand stress (Eq. (14)). If the change in strand using the proposed procedure in the latter case is noteworthy
stress matches the assumed change at the beginning of the because the change in tendon stress is rather small. The
procedure, then the midspan moment, midspan curvature, results shown in Fig. 5 demonstrate that the proposed proce-
total reaction, and midspan displacement are reported. dure can be used to reliably simulate the complete flexural
Midspan displacement is computed by using the second behavior of PT UHPC beams.
moment area theorem, and numerical integration based on
Simpson’s 1/3 rule (Eq. (25)) and Gauss-Legendre quadra- DEFINITION OF CROSS-SECTIONAL AND
ture (Eq. 16) for cracked and uncracked regions, respectively. MEMBER-LEVEL DUCTILITY
This entire procedure is then repeated for different levels of The proposed procedure for computing the moment-
strain at the top concrete fiber until the full moment-curva- curvature-deformation response of PT UHPC beams is used
ture-deformation response is obtained. Procedure termina- to evaluate ductility at the cross section and member level,
tion criteria is established to ensure that the analysis stops and to provide insight about the complete flexural behavior
once flexural failure occurs. Flexural failure is defined as the of PT UHPC members. A typical moment-curvature-defor-
attainment of maximum usable concrete strain in tension or mation relationship for an PT UHPC beam obtained using
compression, or strand rupture. the proposed approach is provided in Fig. 6 and is character-
It should be noted that the proposed procedure does not ized by four distinct phenomena: the application of prestress,
rely on any sort of empiricism other than the one included the cracking of concrete, the yielding of reinforcement, and
in the assumed material constitutive models. The proposed the attainment of ultimate moment capacity. In this paper,
procedure relies on several iterations to produce the curvature ductility, μφ, is defined as the ratio of the curvature
moment-curvature-deformation response. The selection of at the maximum calculated flexural resistance, φu, over the
the number of sections required to produce accurate results curvature at the flexural resistance corresponding to yielding
depends on the loading configuration and should be deter- of the reinforcement in the tension zone, φy. This definition is
mined such that key changes in the moment diagram are consistent with that provided in Annex 8.1 of Canadian Stan-
captured. A total of 13 sections for half of the beam should dards Association (CSA) S6 Fibre Reinforced Concrete,28
be sufficient to produce reliable results. which requires a minimum curvature ductility ratio μmin of
2.0 for members designed for the applied moments under
VALIDATION OF PROPOSED PROCEDURE an elastic analysis without any force redistribution. ACI
Provided that there are currently no experimental tests 318-1929 and AASHTO LRFD Bridge Design Specifi-
on PT UHPC beams, the moment-curvature-deformation cations30 use a strain-based definition for cross section-
response of PT UHPC beams obtained using the proposed level ductility of prestressed and nonprestressed concrete
procedure was compared with that obtained from validated members constructed with NSC. A tension-controlled
nonlinear finite element analysis. The full validation as well section is defined as a cross section in which the net tensile
as the details of the numerical modeling protocol used to strain, εt, in the extreme tension steel at nominal strength is
create the finite element models, including the modeling of greater than or equal to εty + 0.003, where εty is equal to 0.002
the anchorage zones, is provided by Dogu and Menkulasi21 for all prestressed reinforcement and Grade 60 (414 MPa)
and was based on a submodel validation approach. In this deformed reinforcement. This definition for εty is used to
approach, finite element models of NSC beams posttensioned determine the curvature corresponding to yielding of the
with unbonded tendons, and UHPC beams prestressed with reinforcement in the tension zone, φy. The net tensile strain,
bonded tendons were created and the results were compared εt, is the tensile strain calculated in the extreme tension rein-
with those obtained from physical tests. Good correlation forcement at nominal strength exclusive of strains due to
between computed and measured results were obtained. It prestress, creep, shrinkage, and temperature.29 If the limiting
was therefore deduced that a modeling protocol that is able strain at the ultimate limit state for a tension-controlled

Fig. 5—Validation of proposed approach to compute moment-curvature-deformation response of UHPC beams post-tensioned
with draped internal unbonded tendons.
is outside the scope of work for this study. Member-level
ductility is expressed using two metrics. The first is the
ratio of displacement when the beam reaches its maximum
load-carrying capacity, Δu, over the displacement corre-
sponding to yielding of the reinforcement in the tension
zone, Δy. The second is the ratio of span length, L, over
deflection at incipient failure Δu31 (that is, when the beam
reaches its maximum load-carrying capacity). In addition,
in cases when the characterization of cross section- and
member-level ductility using a single numerical value may
be insufficient, the proposed procedure may be used to
examine the complete cross-sectional and member response
from the onset of loading to the attainment of peak load. In
this study, all examined beams exhibited a noticeable soft-
Fig. 6—Definition of cross-sectional and member-level ening response as illustrated in Fig. 5(a) and (b).
ductility.
PARAMETRIC ANALYSIS FOR FLEXURAL
member (0.005) is divided with the strain at first yield
STRENGTH AND DUCTILITY
(0.002), a strain-based cross section-level ductility of 2.5 is
A parametric analysis was conducted to understand the
obtained, which is slightly higher than the curvature based
influence of various parameters on flexural strength and
ductility of 2.0 specified in Canadian standard28 for fiber-
ductility of UHPC beams post-tensioned with unbonded
reinforced concrete (FRC). Both the curvature-based defi-
tendons. Table 1 shows the geometrical and material prop-
nition provided in the Canadian standard28 for FRC and
erties of the beam used for parametric analysis. This beam
the strain-based definition provided in ACI 318-1929 and
is considered to represent a baseline case. The beam is 6 in.
AASHTO LRFD Specifications30 for NSC are used to
(152 mm) wide (bw) and 12 in. (305 mm) deep (h) with a
quantify cross section-level ductility for PT UHPC beams.
span length of 336 in. (8.53 m) (L), and it features two 0.5 in.
Because these North American standards do not require
(13 mm) diameter draped tendons with zero eccentricity at
the obtention of the post-peak response, its determination
ends (e) and 4 in. (102 mm) eccentricity at midspan (em).

Table 1—Properties of the benchmark PT UHPC beam
Beam properties
Tendon Astension, Ascomp.,
Beam bw, mm h, mm Aps, mm2 configura- em, mm mm2 mm2 fc′, MPa fcr, MPa fy, MPa fpu, MPa fpe, MPa
ID (in.) (in.) (in.2) tion (in.) (in.2) (in.2) (ksi) (ksi) (ksi) (ksi) (ksi) fpy/fpu
152 305 198 102 205 63 152 6.89 414 1862 1262
B-1 Draped 0.9
(6) (12) (0.31) (4) (0.32) (0.1) (22) (1) (60) (270) (183)
Table 2—Parameters for various UHPC constitutive models considered

UHPC—compression UHPC—tension
Model fc′, fc1, fc2, fcr, ft′, ft1, ft2,
ID ksi ksi ksi εco εc1 εc2 ksi ksi ksi ksi εcr εto εt1 εt2
Benchmark
B 22 1.3 1.3 0.0038 0.0039 0.01 1 1 1 0.05 0.00015 0.005 0.01 0.0105
Variation of compressive domain
C1 22 10.9 1.3 0.0038 0.006 0.01 1 1 1 0.05 0.00015 0.005 0.01 0.0105
C2 22 10.9 1.3 0.0038 0.007 0.01 1 1 1 0.05 0.00015 0.005 0.01 0.0105
Variation of tensile domain
T1 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.5 0.05 0.00015 0.005 0.01 0.0105
T2 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.5 0.05 0.00015 0.004 0.01 0.0105
T3 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.5 0.05 0.00015 0.006 0.01 0.0105
T4 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.5 0.05 0.00015 0.003 0.01 0.0105
T5 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.5 0.05 0.00015 0.002 0.01 0.0105
T6 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.5 0.05 0.00015 0.007 0.01 0.0105
T7 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.6 0.05 0.00015 0.005 0.01 0.0105
T8 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.7 0.05 0.00015 0.005 0.01 0.0105
T9 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.4 0.05 0.00015 0.005 0.01 0.0105
T10 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.3 0.05 0.00015 0.005 0.01 0.0105
T11 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.8 0.05 0.00015 0.005 0.01 0.0105
T12 22 1.3 1.3 0.0038 0.0039 0.01 1 1.5 0.2 0.05 0.00015 0.005 0.01 0.0105
Note: 1 ksi = 6.895 MPa.
Compressive strength of UHPC (fc′) is 22 ksi (152 MPa). The in many of these cases, the peak load was achieved prior to
prestressing strands are assumed to be low relaxation steel the attainment of the maximum usable tensile strain, εt1, for
with 270 ksi (1862 MPa) ultimate tensile strength (fpu). The UHPC. Six out of the 12 additional constitutive models for
beam features two 0.24 in. (6 mm) and two 0.4 in. (10 mm) the tensile domain of UHPC did not meet strain-based or
diameter mild steel bars in tension, and two 0.24 in. (6 mm) curvature based minimum ductility requirements provided
diameter mild steel bars in compression. The yield stress of in North American codes for NSC and FRC. Therefore,
mild steel is assumed to be 60 ksi (414 MPa). for these beams, adjustments must be made in the amount
of unbonded or bonded reinforcement to elevate cross-
Effect of constitutive model for UHPC on flexural sectional ductility to the required level. In general, the influ-
strength and ductility ence of the tensile domain on flexural strength and ductility
The impact of various UHPC constitutive models (Fig. 2 is a function of the slope of the strain hardening branch and
and Table 2) on flexural strength and ductility is illustrated in the slope of the descending branch after the peak tensile
Table 3. The baseline cases are the beams designated as B-T stress is achieved.
and B-C for the tensile and compressive domains, respec- The consideration of the additional two stress-strain curves
tively. It is worth noting that the impact of the 12 additional for UHPC in compression (C1-C2) resulted in up to a 17.5%
constitutive models (T1 to T12) for the tensile domain of change in flexural capacity (Table 3). It should be noted that
UHPC on flexural strength was contained to at most 8.6% the failure mode for the beam specimen with the benchmark
change in nominal moment capacity compared to the bench- stress-strain curve in compression was selected intentionally
mark (bilinear) tensile curve. The use of UHPC formulations to be a compression-controlled failure, and therefore, the
with strain hardening characteristics in the tensile domain beam exhibits no ductility (that is, ductility < 1). The inclu-
(that is, trilinear model) generally resulted in a decrease in sion of descending branches in the compression domain
ductility compared to the benchmark bilinear model because switched the failure mode from compression-controlled to

Table 3—Influence of various constitutive models for UHPC on flexural strength and ductility
Cross section ductility Member ductility
Model Tendon Loading Mn, m-kN
ID* configuration configuration (in.-kip) %Δ (Mn) με = εu/εy μϕ = ϕu/ϕy μΔ = Δu/Δy L = Δu
B-C †
Draped One point 279 (2469) Baseline 0.42 <1.0 <1.0 276
B-T ‡
Draped One point 125 (1106) Baseline 4.48 3.30 1.65 109
Variation of compressive domain
C1§ Draped One point 325 (2876) 16.64 2.62 2.01 1.27 140
C2 §
Draped One point 328 (2903) 17.51 2.81 1.88 1.25 136
Variation of tensile domain
T1 Draped One point 133 (1177) 5.97 2.43 1.95 1.51 112
T5 Draped One point 125 (1106) –0.04 1.63 1.41 1.07 153
T12 Draped One Point 132 (1168) 5.27 2.30 1.86 1.47 115
*
All properties are as shown in Tables 1 and 2 unless otherwise noted.
†
B-C is benchmark model for evaluating variation in compressive domain with ρstension = 0.49%; ρps = 3.06%; ρtotal = 3.55%.
‡
Benchmark model for evaluating variation in tensile domain (same as B in Table 2).
§
ρstension = 0.49%; ρps = 3.06%; ρtotal = 3.55%.
steel tension-controlled, because the reinforcement yielded in Table 4 and Fig. 7(a) suggest that ±50% change in εt1
prior to the crushing of UHPC in compression. As a result, causes only a marginal change in flexural capacity, because
the inclusion of residual branches in the compressive domain a significant increase in εt1 causes only a small decrease
resulted in an increase in ductility in addition to the increase in the neutral axis depth (Table 4), which in turn causes a
in flexural strength. The benchmark constitutive models for marginal increase in the tension force provided by UHPC
the tensile and compressive domains were used to further (Fig. 8(a)). In addition, UHPC is not the only component
investigate the influence of certain parameters in the consti- providing tensile resistance to an applied external moment,
tutive model as well as the influence of loading configura- therefore a certain percentage increase in εt1 cannot have a
tion, tendon profile, and prestressed and non-prestressed proportional increase in the moment capacity of the beam.
reinforcement ratio on flexural strength and ductility. This is illustrated in Fig. 8(a), which shows the tensile force
provided by UHPC, tendons, and mild steel as εt1 increases.
Effect of εt1 on flexural strength and ductility Table 4 and Fig. 7(a1) to (a3) suggest that εt1 has a direct
It has been previously demonstrated21 that PT UHPC influence on cross-sectional and member-level ductility,
members exhibit a tension-controlled failure, which is although the degree of this influence varies from cross-
characterized by fiber pullout in tension before concrete sectional to member level. At the cross-sectional level,
crushes in compression. Fiber pullout in tension is charac- when ACI 318-19’s29 and AASHTO’s30 strain-based defi-
terized by a reduction in UHPC’s ability to sustain tensile nition for ductility is used, εt1 has a proportional influence
stresses accompanied by an increase in strain. The tensile as illustrated by the lines ε_Point, ε_Two point, and ε_distributed. A
strain at incipient fiber pullout is hereby called maximum 100% increase in εt1 typically results in a 100% increase in
usable tensile strain, εt1, and is considered to be a parameter με. When the Canadian28 curvature-based definition is used,
worthy of investigation because its attainment represents εt1 has a slightly smaller influence on curvature ductility
the point when the beam reaches its maximum load-car- but still a very strong one as illustrated by the lines φ_Point,
rying capacity, especially in UHPC formulations whose φ_Two point, and φ_distributed. Out of the nine beams consid-
tensile domain can be idealized using a bilinear model. The ered in Table 4, only the ones that featured an εt1 equal to
influence of εt1 on flexural strength and ductility of UHPC 0.005 failed to meet minimum cross-section level ductility
beams post-tensioned with unbonded tendons is shown in requirements provided in American29,30 and Canadian stan-
Table 4 and Fig. 7(a). The magnitude of εt1 was varied from dards28 for NSC and FRC, respectively. This suggests that
0.005 to 0.015 in increments of 0.005. The results shown either the design of the UHPC formulation should be chosen

Table 4—Influence of εt1 on flexural strength and ductility
Beam Loading Mn, m-kN
ID configuration εt1 c, mm (in.) (in.-kip) %Δ (Mn) με = εu/εy μϕ = ϕu/ϕy μΔ = Δu/Δy L = Δu
B-1-ETU1-P One point 0.005 84 (3.31) 119 (1053) –4.57 2.19 1.80 1.26 143
B-1-ETU2-P One point 0.010 63 (2.48) 125 (1106) Baseline 4.48 3.30 1.65 109
B-1-ETU3-P One point 0.015 53 (2.09) 126 (1115) +0.47 6.38 4.45 1.73 104
B-1-ETU1-TP Two point 0.005 85 (3.35) 119 (1053) –7.21 2.23 1.81 1.41 74
B-1-ETU2-TP Two point 0.010 54 (2.13) 128 (1133) Baseline 4.29 3.27 2.20 47
B-1-ETU3-TP Two point 0.015 64 (2.52) 129 (1142) +0.98 6.74 5.15 2.37 44
B-1-ETU1-DP Distributed 0.005 85 (3.35) 126 (1115) –4.68 2.23 1.80 1.73 60
B-1-ETU2-DP Distributed 0.010 65 (2.56) 132 (1168) Baseline 3.91 3.03 2.24 46
B-1-ETU3-DP Distributed 0.015 55 (2.17) 139 (1230) +5.65 6.31 4.77 3.74 28
such that there is a sufficient amount of fibers so that εt1 is tendon profiles exhibit slightly higher ultimate tendon stress
at least equal to 0.006, or the area of strands and tension compared to beams with draped tendon profiles since the
mild steel should be adjusted accordingly. It is worth noting overall tendon deformation is larger when the tendon is
how the influence of εt1 on member-level ductility (Δ_Point, closer to the most extreme tension fiber. This member-level
Δ_Two point, and Δ_distributed) is not nearly as strong as that on dependency results in slightly larger moment capacities at
cross-sectional-level ductility, especially for beams that the ultimate limit state for beams with straight tendon profiles
feature one point and two-point loading when the ratio of (Fig. 7(c1)). However, these differences are marginal. The
Δu/Δy is used to quantify ductility. For a distributed loading greatest influence of tendon profile on moment capacity is
configuration, the influence of εt1 on member-level ductility exhibited in beams subject to a two-point loading configu-
remains strong, suggesting that ductility at the member level ration because in these beams, the weakest section may be
is highly dependent on the loading configuration. This obser- at the location of point loads rather than at midspan due to
vation is further corroborated when member-level ductility the draping of the tendons and the reduced effective depth.
is evaluated using the L/Δu ratio, as shown in Table 4 and Tendon profile appears to have some effect on cross section-
Fig. 7(a3). Beams with two-point loading and distributed level ductility, with beams featuring a straight tendon profile
loading configurations exhibit much larger deflections at typically featuring higher ductility values compared to those
incipient failure than beams with a point load at midspan. with draped tendons.
Table 6 and Fig. 7(c2) and (c3) suggest that the influence
Effect of fcr on flexural strength and ductility of loading configuration on ductility varies from mild to
Table 5 and Fig. 7(b) illustrate the influence of UHPC strong at the cross section and member level, respectively.
cracking stress, fcr, on flexural capacity and ductility. The At the cross section level (Fig. 7(c2)), curvature ductility
investigated beams featured a range of fcr from 0.15√fc′ is barely influenced by loading configuration or tendon
(0.39√fc′) to 0.32√fc′ (0.84√fc′) where fc′ is 22 ksi (152 MPa). profile. Strain-based cross-sectional ductility appears to be
A certain percent change in the fcr causes a proportional slightly influenced by loading configuration when draped
change in the magnitude of the tension force provided by tendon profiles are considered and barely influenced when
UHPC (Fig. 8(b)). The influence of fcr on flexural capacity beams with straight tendon configuration are considered. It
is more pronounced than that of εt1 because of the more is interesting how at the member level (Fig. 7(c2) and (c3)),
significant impact that fcr has on the tension force provided loading configuration appears to have a marked influence
by UHPC (Fig. 8(b)). Table 5 and Fig. 7(b) suggest that fcr with beams featuring a two-point and distributed loading
has almost no marked effect on cross section- and member- configurations possessing higher ductility than those with
level ductility. All 12 investigated beams exhibited similar one point-loading configuration.
levels of ductility and they all met minimum cross section-
level ductility requirements provided in American29,30 Effect of ρps on flexural strength and ductility
and Canadian standards28 for NSC and FRC, respectively. Table 7 and Fig. 7(d1) to (d3) show the influence of longitu-
Figure 7(b3) suggests that when the ratio L/Δu was used to dinal unbonded reinforcement ratio, ρps, on flexural strength
evaluate member-level ductility, an increase in fcr corre- and ductility. The investigated total longitudinal reinforce-
sponded with a slight increase in ductility (that is, a lower ment ratio ρtotal (or ρt) was varied from 1.00 to 1.47%,
L/Δu). while the unbonded reinforcement ratio ρps was varied from
0.51 to 0.98%. The bonded tensile reinforcement ratio,
Effect of loading configuration and tendon profile ρstension, was kept constant at 0.49%, which is greater than
on flexural strength and ductility the 0.22% required minimum ratio specified in ACI 318-1929
Table 6 and Fig. 7(c) illustrate the influence of loading for bonded reinforcement in NSC members post-tensioned
configuration and tendon profile on flexural strength and with unbonded tendons. While an increase in ρps causes a
ductility of PT UHPC beams. The beams with straight significant increase in flexural capacity (Fig. 7(d1)), this

Fig. 7—Influence of various parameters on flexural strength and ductility: (a) εt1; (b) fcr; (c) loading configuration and tendon
profile; (d) ρps; and (e) ρstension.
increase is not directly proportional to that induced in the provided by the tendons and their contribution to flexural
ρps because tendons are not the only component providing capacity. Figure 8(c) shows that an increase in ρps causes
tensile resistance, as illustrated in Fig. 8(c). Also, an increase a slight decrease in the tension force provided by UHPC
in ρps results in a decrease in the tendon stress at ultimate, due to a shift in the neutral axis, thus further dampening the
fps,21 thus affecting the magnitude of the tension force increase in flexural capacity caused by an increase in ρps.

Table 5—Influence of cracking stress fcr on flexural strength and ductility
fcr = a√fc′ Cross section ductility Member ductility
ID configuration fcr, MPa (ksi) a, SI (Imp.) (in.-kip) %Δ (Mn) με = εu/εy μϕ = ϕu/ϕy μΔ = Δu/Δy L = Δu
B-1-fcr1 One point 4.83 (0.7) 0.39 (0.15) 114 (1009) –8.74 4.36 3.24 1.63 115
B-1-fcr2 One point 6.89 (1.0) 0.56 (0.21) 125 (1106) Baseline 4.48 3.30 1.65 109
B-1-fcr3 One point 8.27 (1.2) 0.68 (0.26) 132 (1168) +5.66 4.41 3.08 1.69 109
B-1-fcr4 One point 10.34 (1.5) 0.84 (0.32) 143 (1266) +14.67 4.44 3.19 1.75 102
B-1-fcr1 Two point 4.83 (0.7) 0.39 (0.15) 118 (1044) –8.15 4.32 3.27 2.13 50
B-1-fcr2 Two point 6.89 (1.0) 0.56 (0.21) 128 (1133) Baseline 4.29 3.27 2.20 47
B-1-fcr3 Two point 8.27 (1.2) 0.68 (0.26) 136 (1204) +5.92 4.49 3.27 2.20 46
B-1-fcr4 Two point 10.34 (1.5) 0.84 (0.32) 147(1301) +14.57 4.48 3.28 2.24 44
B-1-fcr1 Distributed 4.83 (0.7) 0.39 (0.15) 122 (1080) –7.7 4.11 3.19 2.21 46
B-1-fcr2 Distributed 6.89 (1.0) 0.56 (0.21) 132 (1168) Baseline 3.91 3.03 2.24 46
B-1-fcr3 Distributed 8.27 (1.2) 0.68 (0.26) 139 (1230) +5.76 4.22 3.21 2.28 45
B-1-fcr4 Distributed 10.34 (1.5) 0.84 (0.32) 150 (1328) +13.85 4.22 3.15 2.30 44
Fig. 8—Variation in the tension force provided by each component as function of: (a) εt1; (b) fcr; (c) ρps; and (d) ρstension.
Table 7 shows that the influence of ρps on flexural capacity and Canadian28 standards for NSC and FRC, respectively.
is stronger than the influence of fcr for a given percentage Figures 7(d2) and (d3) suggest that for beams with point and
change in these parameters from the baseline values. Table 7 two-point loading configurations, ductility decreased as ρps
and Fig. 7(d2) and (d3) suggest that an increase in ρps typi- increased, whereas the ductility of the beams with a distrib-
cally results in a decrease in cross section- and member- uted loading remained essentially unaffected by the increase
level ductility, although all nine investigated beams met in ρps.
minimum ductility requirements provided in American29,30

Table 6—Influence of loading configuration and tendon profile on flexural strength and ductility
Tendon Loading Mn, m-kN
Beam ID configuration configuration (in.-kip) %Δ (Mn) με = εu/εy μϕ = ϕu/ϕy μΔ = Δu/Δy L = Δu
B-1S Straight One point 125 (1106) +0.04 4.62 3.40 1.71 114
B-1S Straight Two point 133 (1177) +6.77 4.42 3.17 2.56 40
B-1S Straight Distributed 132 (1168) +5.57 4.55 3.23 2.70 48
B-1D Draped One point 125 (1106) Baseline 4.48 3.30 1.65 109
B-1D Draped Two point 128 (1133) +2.48 4.29 3.27 2.20 47
B-1D Draped Distributed 132 (1168) +5.49 3.91 3.03 2.24 46
Table 7—Influence of ρps on flexural strength and ductility

ρ, % Cross section ductility Member ductility
Loading Mn, m-kN
Beam ID configuration ρtotal ρps (in.-kip) %Δ (Mn) με = εu/εy μϕ = ϕu/ϕy μΔ = Δu/Δy L = Δu
B-1-Aps1 One point 1.00 0.51 125 (1106) Baseline 4.48 3.30 1.65 109
B-1-Aps2 One point 1.21 0.72 147 (1301) +17.28 4.27 3.13 1.57 113
B-1-Aps3 One point 1.47 0.98 170 (1505) +35.82 3.77 2.65 1.37 125
B-1-Aps1 Two point 1.00 0.51 128 (1133) Baseline 4.29 3.27 2.20 47
B-1-Aps2 Two point 1.21 0.72 149 (1319) +16.34 4.26 3.20 1.90 54
B-1-Aps3 Two point 1.47 0.98 171 (1513) +33.33 3.98 2.98 1.54 63
B-1-Aps1 Distributed 1.00 0.51 132 (1168) Baseline 3.91 3.03 2.24 46
B-1-Aps2 Distributed 1.21 0.72 159 (1407) +20.28 3.90 2.93 2.23 46
B-1-Aps3 Distributed 1.47 0.98 192 (1699) +45.69 3.77 2.92 2.22 47
Note: ρtotal = ρstension + ρps; ρstension = Astension/(bwds); ρps = Aps/(bwdps); ρmin_stension_ACI = 0.22% (Astension_min_ACI = 0.004Act).
Effect of ρstension on flexural strength and ductility beam was heavily reinforced and the longitudinal reinforce-
Table 8 and Fig. 7(e) show the influence of the bonded ment ratios were well beyond practical limits for concrete
longitudinal reinforcement ratio in tension, ρstension, on flex- beams posttensioned with unbonded tendons.
ural capacity and ductility. The investigated total longitu-
dinal reinforcement ratio ρtotal varied from 1.00 to 3.08%, CONCLUSIONS
while the reinforcement ratio for bonded tension steel, ρsten- 1. The proposed procedure is based on structural mechanics
sion, varied from 0.49 to 2.57%. The unbonded reinforcement and is capable of reliably and efficiently computing the
ratio, ρps, was kept constant at 0.51%. Similar to the influ- complete moment-curvature-deformation response of
ence of ρps on flexural capacity, while an increase in ρstension post-tensioned (PT) ultra-high-performance concrete
causes a proportional increase in the tension force provided (UHPC) beams and offers insight about their ductility at the
by mild steel, it does not cause a proportional increase in cross section and member level. This information is useful in
flexural capacity due to the contribution of tendons and cases when a performance-based design approach is selected
UHPC. In general, Fig. 8(d) illustrates how the contribution and a characterization of member behavior under different
of each component that provides tensile resistance varies for load intensities is desired. This includes cases when the
different scenarios. As ρstension is changing, there is no signif- determination of deflection under a given load and subse-
icant difference in the tensile resistance provided by UHPC quent determination of compliance with specified criteria is
and prestressing tendons. Table 8 and Fig. 7(e1) show that desired, or when the quantification of deflection at incipient
while the influence of ρstension on flexural capacity is not as failure is of interest. The latter has been recently reported31
strong as the influence of fcr and ρps for a given percentage as an important warning of failure metric.
change from the baseline values, ρstension is still an influential 2. The proposed method does not require the definition of
factor on flexural capacity. Table 8 and Fig. 7(e2) and (e3) empirically obtained elastic and inelastic design parameters
suggest that ρstension has a similar effect with ρps on cross-sec- such as bond reduction coefficients and plastic hinge length.
tional and member-level ductility, with higher ρstension values However, it can facilitate the use of approaches that employ
resulting in lower ductility. All 12 investigated beams these parameters by helping formulate them without having
met minimum cross section-level ductility requirements to rely on extensive experimental testing or time-consuming
provided in American29,30 and Canadian standards28 for finite element analysis.
NSC and FRC, respectively. One of the beams exhibited a 3. Cross section-level ductility was generally higher than
curvature-based cross section-level ductility of 2.08, which member-level ductility. Additionally, strain-based computa-
is close to the 2.0 Canadian limit28; however, this beam had a tions resulted in higher levels of cross section-level ductility
ρtotal of 3.08% and a ρstension of 2.57%, which suggest that the compared to curvature-based computations. Most beams

Table 8—Influence of ρstension on flexural strength and ductility
ρ, % Cross-section ductility Member ductility
ID configuration ρtotal ρstension (in.-kip) %Δ (Mn) με = εu/εy μϕ = ϕu/ϕy μΔ = Δu/Δy L = Δu
B-1-As1 One point 1.00 0.49 125 (1106) Baseline 4.48 3.30 1.65 109
B-1-As2 One point 1.27 0.76 136 (1206) +8.46 4.30 3.15 1.52 110
B-1-As3 One point 2.01 1.50 163 (1443) +30.56 4.24 2.88 1.35 123
B-1-As4 One point 3.08 2.57 198 (1752) +58.58 3.82 2.08 1.15 137
B-1-As1 Two point 1.00 0.49 128 (1132) Baseline 4.29 3.27 2.20 47
B-1-As2 Two point 1.27 0.76 138 (1221) +7.55 4.27 3.17 2.00 49
B-1-As3 Two point 2.01 1.50 163 (1442) +27.6 4.22 3.06 1.68 53
B-1-As4 Two point 3.08 2.57 199 (1761) +55.16 4.15 2.89 1.34 59
B-1-As1 Distributed 1.00 0.49 132 (1168) Baseline 3.91 3.03 2.24 46
B-1-As2 Distributed 1.27 0.76 143 (1266) +8.46 3.87 3.02 2.20 48
B-1-As3 Distributed 2.01 1.50 180 (1593) +36.2 3.84 2.98 1.98 50
B-1-As4 Distributed 3.08 2.57 215 (1903) +62.94 3.79 2.97 1.85 53
Note: ρtotal = ρstension + ρps; ρstension = Astension/(bwds); ρps = Aps/(bwdps); ρmin_stension_ACI = 0.22% (Astension_min_ACI = 0.004Act).
met minimum cross section-level ductility requirements State University. His research interests include the behavior of post-ten-
sioned ultra-high-performance concrete (UHPC) members.
specified in American and Canadian standards for normal-
strength concrete (NSC) and fiber-reinforced concrete ACI member Fatmir Menkulasi is an Assistant Professor at Wayne State
(FRC), respectively. University. He received his BS from Middle East Technical University and
his MS and PhD from Virginia Polytechnic Institute and State University,
4. The influence of the considered constitutive models for Blacksburg, VA. He is a member of ACI Committees 123, Research and
the tensile and compressive domains was contained to an Current Developments, and 239, Ultra High-Performance Concrete; ACI
8.6% and 17.5% change in flexural capacity when compared Subcommittee 239-C, Structural Design on UHPC; and Joint ACI-ASCE
Committee 423, Prestressed Concrete. His research interests include
to the benchmark stress-strain curves. Most members exhib- behavior of reinforced and prestressed UHPC structures.
ited a fiber tension-controlled failure.
5. The most influential parameter for cross section-level DATA AVAILABILITY STATEMENT
ductility is εt1. A minimum value of 0.006 is recommended All or part of the data including the algorithm for obtaining the
for εt1 to meet existing minimum ductility requirement set moment-curvature-deformation response can be obtained from the corre-
sponding author upon reasonable request.
forth for NSC and FRC members. Additionally, while εt1 is
an important parameter related to the failure mode of UHPC
beams post-tensioned with unbonded tendons, its variation REFERENCES
1. Russel, H. G., and Graybeal, B. A., “Ultra-High-Performance
does not cause a marked difference in flexural capacity. Concrete: State-of-the-Art Report for the Bridge Community,” Federal
6. The most influential parameter for member level Highway Administration, Washington, DC, 2013.
2. Balaguru, P., “Increase of Stress in Unbonded Tendons in Prestressed
ductility is the loading configuration, especially when Concrete Beams and Slabs,” Canadian Journal of Civil Engineering, V. 8,
the ratio L/Δu was used as the metric for evaluating such No. 2, 1981, pp. 262-268. doi: 10.1139/l81-033
ductility. The proposed methodology provides a framework 3. Naaman, A. E., and Alkhairi, F. M., “Stress at Ultimate in Unbonded
Prestressing Tendons Part 1: Evaluation of the State-of-the-Art,” ACI Struc-
for accurately determining the deflection at incipient failure, tural Journal, V. 88, No. 5, Sept.-Oct. 1991, pp. 641-651.
Δu. The influence of loading configuration on cross section- 4. Naaman, A. E., and Alkhairi, F. M., “Stress at Ultimate in Unbonded
level ductility was much smaller than that exhibited at the Prestressing Tendons Part 2: Proposed Methodology,” ACI Structural
Journal, V. 88, No. 6, Nov.-Dec. 1991, pp. 683-692.
member level. 5. Harajli, M. H., and Kanj, M. Y., “Service Load Behavior of Concrete
7. The most influential parameters in terms of increasing Members Prestressed with Unbonded Tendons,” Journal of Structural
flexural capacity are the longitudinal unbonded reinforce- Engineering, ASCE, V. 118, No. 9, 1992, pp. 2569-2589. doi: 10.1061/
(ASCE)0733-9445(1992)118:9(2569)
ment ratio, ρps, followed by the UHPC cracking stress, fcr, 6. Pannell, F. N., “The Ultimate Moment of Resistance of Unbonded
and the longitudinal bonded reinforcement ratio, ρstension, Prestressed Concrete Beams,” Magazine of Concrete Research, V. 21,
when the impact of these parameters for a given percentage No. 66, 1969, pp. 43-54. doi: 10.1680/macr.1969.21.66.43
7. Tam, A., and Pannell, F. N., “The Ultimate Moment of Resistance
change from the baseline case was considered. While of Unbonded Partially Prestressed Reinforced Concrete Beams,” Maga-
a change in fcr did not influence ductility at any level, an zine of Concrete Research, V. 28, No. 97, 1976, pp. 203-208. doi: 10.1680/
increase in ρps and ρstension resulted in a decrease in cross macr.1976.28.97.203
8. Au, F. T. K., and Du, J. S., “Prediction of Ultimate Stress in Unbonded
section- and member-level ductility. Prestressed Tendons,” Magazine of Concrete Research, V. 56, No. 1, 2004,
pp. 1-11. doi: 10.1680/macr.2004.56.1.1
AUTHOR BIOS 9. Harajli, M. H., “On the Stress in Unbonded Tendons at Ultimate: Crit-
ACI member Mehmet Dogu is a Structural Engineer at Emay Engineering ical Assessment and Proposed Changes,” ACI Structural Journal, V. 103,
and Consultancy, Istanbul, Turkey. He received his BS from Middle East No. 6, Nov.-Dec. 2006, pp. 803-812.
Technical University, Ankara, Turkey, and his MS and PhD from Wayne 10. Harajli, M. H., “Effect of Span-Depth Ratio on the Ultimate Steel
Stress in Unbonded Prestressed Concrete Members,” ACI Structural
Journal, V. 87, No. 3, May-June 1990, pp. 305-312.

11. Lee, L. H.; Moon, J. H.; and Lim, J. H., “Proposed Methodology for 21. Dogu, M., and Menkulasi, F., “A Flexural Design Methodology for
Computing of Unbonded Tendon Stress at Flexural Failure,” ACI Structural UHPC Beams Posttensioned with Unbonded Tendons,” Engineering Struc-
Journal, V. 96, No. 6, Nov.-Dec. 1999, pp. 1040-1048. tures, V. 207, 2020, p. 110193. doi: 10.1016/j.engstruct.2020.110193
12. Harajli, M. H.; Mabsout, M. E.; and Al-Hajj, J. A., “Response of 22. Graybeal, B. A., “Compressive Behavior of Ultra-High-Perfor-
Externally Post-Tensioned Continuous Members,” ACI Structural Journal, mance Fiber-Reinforced Concrete,” ACI Materials Journal, V. 104, No. 2,
V. 99, No. 5, Sept.-Oct. 2002, pp. 671-680. Mar.-Apr. 2007, pp. 146-152.
13. Au, F. T. K.; Du, J. S.; and Cheung, Y. K., “Service Load Analysis of 23. Devalapura, R. K., and Tadros, M. K., “Stress-Strain Modeling of
Unbonded Partially Prestressed Concrete Members,” Magazine of Concrete 270 ksi Low-Relaxation Prestressing Strands,” PCI Journal, V. 37, No. 2,
Research, V. 57, No. 4, 2005, pp. 199-209. doi: 10.1680/macr.2005.57.4.199 1992, pp. 100-106. doi: 10.15554/pcij.03011992.100.106
14. Vega, M., and Dotreppe, J. C., “Numerical Procedure for the Analysis 24. Graybeal, B. A., “Structural Behavior of Ultra-High-Performance
of the Ultimate Limit State Behavior of Prestressed Concrete Structures,” Concrete Prestressed I-Girders (No. FHWA-HRT-06-115),” Office of Infra-
Federation International de la Precontrainte, Lausanne, Switzerland, 1988. structure Research and Development, Federal Highway Administration,
15. Alkhairi, F. M., and Naaman, A. E., “Analysis of Beams Prestressed Washington, DC, 2006.
with Unbonded Internal or External Tendons,” Journal of Structural 25. John, E. E.; Ruiz, E. D.; Floyd, R. W.; and Hale, W. M., “Transfer
Engineering, ASCE, V. 119, No. 9, 1993, pp. 2680-2700. doi: 10.1061/ and Development Lengths and Prestress Losses in Ultra-High-Perfor-
(ASCE)0733-9445(1993)119:9(2680) mance Concrete Beams,” Transportation Research Record: Journal of
16. Knight, D.; Visintin, P.; Oehlers, D. J.; and Ali, M., “Simulating RC the Transportation Research Board, V. 2251, No. 1, 2011, pp. 76-81. doi:
Beams with Unbonded FRP and Steel Prestressing Tendons,” Compos- 10.3141/2251-08
ites. Part B, Engineering, V. 60, 2014, pp. 392-399. doi: 10.1016/j. 26. ACI Committee 435, “Control of Deflection in Concrete Structures
compositesb.2013.12.039 (ACI 435R-95) (Reapproved 2000),” American Concrete Institute, Farm-
17. Ariyawardena, N., and Ghali, A., “Prestressing with Unbonded ington Hills, MI, 1995, 77 pp.
Internal or External Tendons: Analysis and Computer Model,” Journal of 27. Hornbeck, R. W., Numerical Methods, Prentice-Hall, Inc., Engle-
Structural Engineering, ASCE, V. 128, No. 12, 2002, pp. 1493-1501. doi: wood Cliffs, NJ, 1975, pp. 293-294.
10.1061/(ASCE)0733-9445(2002)128:12(1493) 28. CSA S6:19, “Canadian Highway Bridge Design Code,” Canadian
18. Ozkul, O.; Nassif, H.; Tanchan, P.; and Harajli, M., “Rational Standards Association, Rexdale, ON, Canada, 2019.
Approach for Predicting Stress in Beams with Unbonded Tendons,” ACI 29. ACI Committee 318, “Building Code Requirements for Structural
Structural Journal, V. 105, No. 3, May-June 2008, pp. 338-347. Concrete (ACI 318-19) and Commentary (ACI 318R-19),” American
19. Vu, N. A.; Castel, A.; and François, R., “Response of Post-Tensioned Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
Concrete Beams with Unbonded Tendons Including Serviceability and Ulti- 30. AASHTO, “LRFD Bridge Design Specifications, 8th edition,” Amer-
mate State,” Engineering Structures, V. 32, No. 2, 2010, pp. 556-569. doi: ican Association of State Highway and Transportation Officials, Wash-
10.1016/j.engstruct.2009.11.001 ington, DC, 2017.
20. Kim, K. S., and Lee, D. H., “Nonlinear Analysis Method for 31. van Weerdhuizen, M., and Bartlett, F. M., “Deflection at Incipient
Continuous Post-Tensioned Concrete Members with Unbonded Tendons,” Failure as Warning-of-Failure Metric,” ACI Structural Journal, V. 117,
Engineering Structures, V. 40, No. 1, 2012, pp. 487-500. doi: 10.1016/j. No. 4, July 2020, pp. 233-241.

NOTES:

Title No. 119-S48
A Multi-Gene Genetic Programming Model for Predicting

Shear Strength of Steel Fiber Concrete Beams
by Mohamed K. Ismail, Ahmed Yosri, and Wael El-Dakhakhni
The superior performance of steel fibers (SFs) in enhancing the improve the shear strength of concrete and reach a possible
shear capacity of reinforced concrete (RC) beams supports their reduction in the stirrups’ volume ratio. ACI 318,1 for
use in regions that would otherwise require minimum shear rein- example, permits the use of deformed SFs as a replacement
forcement (stirrups). Subsequently, several researchers proposed for the minimum stirrups. The positive effect of SFs on
different predictive models to capture the contribution of SFs to
concrete’s shear strength is attributed to the fact that SFs act
the shear capacity of RC beams without stirrups. However, most of
as crack arrestors that delay the initiation and propagations
these models were empirically developed using respective limited
data sets, restricting their generalizability. Artificial intelligence- of diagonal cracks.2 Once diagonal cracks are developed, SFs
based models have shown high efficacy in imitating the behavior prevent the latter’s localized widening and help to transfer
of complex systems based on large data sets; however, such stress across crack faces through bridging mechanisms to
models are usually criticized for being impractical for design the surrounding concrete. SFs may also behave similarly to
and too complex to use. To address this issue, a new methodology elongated aggregates that form a special type of aggregate-
was adopted in this paper using multi-gene genetic program- interlock mechanism that contributes to increasing the fric-
ming (MGGP)—a class of artificial intelligence techniques—to tion along the diagonal cracks and preventing slippage.3
develop an elegant shear strength prediction model for steel fiber- Adding SFs to concrete in beams that already satisfy the
reinforced concrete (SFRC) beams. Guided by mechanics and design requirements for tensile and shear reinforcement
previous research findings, the governing parameters and their
can also boost their ductility, energy absorption capacity,4
key combinations were first identified and selected for the model
impact resistance,5 and fatigue life,6 and provide a margin of
development. Subsequently, the model was trained, validated,
and tested using observations from 752 experimental tests, and safety against accidental tensile force demands.
different measures were applied to assess its performance and Despite the structural benefits achieved through intro-
generalizability. The MGGP-based model developed in this study ducing SFs to RC beams, no procedure currently exists to
outperformed 15 shear strength prediction models developed in facilitate their adoption in relevant design standards, which
previous studies and is also presented in a standards-ready format limits the proliferation of SFs in the construction industry.
to facilitate adoption. The outcomes from this study demonstrate To close this knowledge gap, several researchers carried out
the promising capability of a mechanics-guided artificial intelli- experimental and analytical studies to propose models to
gence approach to develop interpretable models that can efficiently predict the shear strength of steel fiber-reinforced concrete
predict complex structural behaviors of other components and (SFRC) beams.7-15 The most simplified model was proposed
systems.
by Sharma7 to empirically estimate the shear strength of
Keywords: data-driven empirical models; genetic programming; sensi- SFRC beams based on the tensile strength of concrete only,
tivity analysis; shear strength prediction; steel fiber-reinforced concrete considering 48 SFRC beams. Although this model was
(SFRC) beams. recommended by ACI 544.4R-88,16 it can be argued that
the model ignored key influencing factors of relevant beam
INTRODUCTION shear strength, including the tensile reinforcement ratio, the
Shear failures in reinforced concrete (RC) beams present volume, aspect ratio, and shape of fibers. A function only
high risk as they occur abruptly with little to no advanced of the concrete tensile strength, the model is also unable to
warning. The serious consequences of such failures have differentiate between the shear strength of beams cast with
been attracting substantial concerns from practicing engi- various concrete mixture types.
neers as well as building code and design standards commit- Narayanan and Darwish9 considered the factors neglected
tees to carefully consider associated design provisions. As by Sharma7 in their model, in which the tensile reinforce-
such, adequate shear reinforcement (for example, stirrups) ment ratio, fiber-matrix interfacial bond strength, fiber
is usually provided to arrest diagonal cracks, compensate volume, aspect ratio, and shape were considered. In addi-
for the low diagonal tension strength of concrete, and safe- tion, a non-dimensional arching factor (e), similar to that in
guard against catastrophic failures. The use of congested Zsutty’s equation,17 was included in the model to account for
shear reinforcement in zones subjected to high stress levels
also presents a construction challenge as casting and consol- ACI Structural Journal, V. 119, No. 2, March 2022.
MS No. S-2021-165.R2, doi: 10.14359/51734345, received September 7, 2021, and
idating of concrete would then be difficult. In their attempts reviewed under Institute publication policies. Copyright © 2022, American Concrete
to find effective solutions for these issues, several studies obtained from the copyright proprietors. Pertinent discussion including author’s
identified steel fibers (SFs) as an effective alternative to closure, if any, will be published ten months from this journal’s date if the discussion

deep and slender SFRC beams. Narayanan and Darwish’s9 For example, Shahnewaz and Alam20,21 proposed GP-based
model was developed and validated based on a limited models to estimate the shear strength of deep and slender
data set of 33 specimen results. In their model, Narayanan SFRC beams without stirrups, using a data set of 358 tests
and Darwish9 suggested considering a fixed value for the to develop and verify their models. Similar approaches
fiber-matrix interfacial bond strength (4.15 MPa, similar were followed by Gandomi et al.,22 Sarveghadi et al.,23 and
to that proposed by Swamy et al.18) irrespective of the Chaabene and Nehdi24 using observations from 213, 208,
strength of the surrounding matrix. Such an assumption and 309 tests, respectively, to generate their shear predic-
may lead to inaccurate estimation for the contribution of tion models for deep and slender SFRC beams. The results
fibers if concretes with varied compressive strengths were reported by the aforementioned studies indicate that clas-
used. Similar to Narayanan and Darwish,9 but considering a sical GP presents a powerful tool that can be used to develop
larger data set of 139 tests, Kwak et al.14 developed a model efficient predictive models compared to those developed
based on Zsutty’s equation17 to account for the effect of the based on classical techniques (that is, regression analysis).
tensile strength on arching action with an additional term However, the GP models proposed in these studies may be
for the fibers. Kwak et al.14 suggested a new value for the criticized for their apparent lack of: 1) mechanics-based
arching factors (that is, taken as 1.0 for a/d greater than 3.4 interpretation, as they typically consist of random, dimen-
and equal to 3.4 d/a for a/d less than 3.4) but kept the fiber- sionally inconsistent combinations of influencing param-
matrix interfacial bond strength as constant, as assumed by eters; and 2) elegance, considering that unconstrained GP
Narayanan and Darwish.9 Al-Ta’an and Al-Feel15 included often produces complex, long mathematical expressions,
the fiber factor in their model to represent the effect of particularly when the behavior of the underlying system is
fibers and made no explicit reference to the interfacial bond highly nonlinear. Such disadvantages limit the adoption of
strength. Ashour et al.10 also proposed two models to esti- available GP models by relevant design standards and guide-
mate the shear strength of singly reinforced high-strength lines—a situation that highlights the need for a mechan-
SFRC beams without stirrups. The models were developed ics-guided GP approach. In addition, the GP is best applied
by conducting regression analyses on experimental results in a multi-gene fashion, where multiple short mathematical
obtained from 18 tests. The first model was suggested by expressions are generated simultaneously and subsequently
incorporating the fiber factor into Zsutty’s equation,17 combined through a linear regression analysis.25 The appli-
whereas the second model presented a modification of cation of multi-gene genetic programming (MGGP) thus
the ACI1 equation, in which a/d was introduced for both facilitates capturing the nonlinear behavior of the underlying
the concrete and the longitudinal reinforcement. The fiber system at a higher accuracy.25
factor and fiber-matrix interfacial bond strength (taken as Therefore, this study adopts a mechanics-guided MGGP
4.15 MPa) were also added to the ACI1 shear equation to approach to develop an elegant and reliable shear strength
account for the fiber effects. The ACI equation was also prediction model, trained, tested, and validated using a
modified by Mansur et al.,8 Swamy et al.,11 and Khuntia large SFRC beam data set. A sensitivity analysis was also
et al.13 Unlike previous models, the model by Khuntia et performed to identify the most important parameters and
al.13 does not assume a constant value for the fiber-matrix their key combinations. To assess the proposed model’s
interfacial bond strength, but the effect of the fiber is taken performance, its predictions were compared to those of other
proportional to the fiber factor and the concrete compressive models available in the literature.
strength. As such, the model considers the concrete strength
effects on improving the interfacial bond strength between RESEARCH SIGNIFICANCE
the fiber and surrounding matrix. However, such models were The objective of this research is to analyze the complex
developed based on regression analyses of a limited number interrelationships between the factors controlling the shear
of beam subsets that do not capture the interplay between behavior of SFRC beams and subsequently develop a related
all factors controlling the shear behavior of SFRC beams, reliable prediction model using a powerful evolutionary
thus limiting their generalizability and prediction accuracy. computing tool (that is, GP) guided by mechanics and
This situation highlights the need for developing more accu- previous research findings. Despite being applied by others,
rate models considering more comprehensive experimental existing GP-based mathematical models are highly complex,
data sets. In addition, the intrinsic diversity of experimental dimensionally inconsistent, and difficult to be interpreted by
data sets collected from different research programs hinders practitioners. In this respect, this study employed MGGP,
the use of classical mathematical and statistical techniques where a set of GP-based models are combined to analyze a
(for example, regression analysis) to describe the underlying large data set consisting of 752 test results. To the best of the
mechanics. As such, artificial intelligence (AI) techniques— authors’ knowledge, this data set has never been used in full
including, for example, artificial neural network and genetic to develop any of the existing models, and was used herein
programming (GP)—may present a better alternative to for the model development, validation, and testing. The
mechanics-based approaches in terms of capturing complex quantity and diversity in the data set facilitated capturing
behaviors without necessarily offering a plausible interpre- complex nonlinear relationships affecting the shear strength
tation of the underlying mechanics, creating what is known of SFRC beams. This in turn contributed to addressing the
as black-box models.19 The advances in AI techniques low accuracy of existing empirical shear models when used
encouraged researchers20-24 to use GP to analyze large data to predict experimental results other than those originally
sets to generate models with higher prediction capabilities. used to develop such models. The authors believe that the

procedures proposed in this study will be of general interest models—specifically, 15 models, as listed in Appendix A.*
to practicing structural engineers and standards committees, It should be noted that the development backgrounds of the
and within the scope of the current study, can result in accu- selected 15 models were briefly discussed in the previous
rate estimation of the shear strength of SFRC beams without sections. Finally, a sensitivity analysis was applied to iden-
stirrups, aiming to accelerate mainstream adoption of SFRC tify the key parameters controlling the shear behavior of
components in structural applications. SFRC beams based on the developed MGGP-based model.
METHODOLOGY EXPERIMENTAL DATA SET

To achieve the main objective of the present research, a A data set consisting of 752 tests (presented in Appendix B)
specific methodology was followed. First, a brief description conducted on deep and slender beams was collected from
of the collected data set used in this study was presented. several studies available in the literature to develop a new,
Informed by mechanics and previous research findings, key efficient model for predicting the ultimate shear capacity of
parameters and their key combinations affecting the shear SFRC beams constructed without stirrups. The parameters
strength of SFRC beams were subsequently identified and used in the experimental investigations include beam width
combined to guide the model and avoid random generation (b), effective depth (d), shear span-effective depth ratio
of uninterpretable and dimensionally inconsistent terms. (a/d), compressive strength of concrete (fc′), longitudinal
Following training, testing, and validation procedures, the reinforcement ratio (ρs), fiber volume (ρf), fiber aspect ratio
best MGGP-based model was identified through specific (lf/df), and fiber type. Table 1 shows a summary of the values
metrics and its performance was compared to that of existing of such parameters.
FACTORS AFFECTING SHEAR STRENGTH

Figure 1 shows the free body diagram for forces acting
along an inclined shear crack in an SFRC beam without
Table 1—Ranges of parameters used in this study
stirrups. Based on the figure, the applied shear force (Vu)
Parameter Minimum limit Maximum limit is carried by the compression zone (Vc), dowel action
Width (b), mm 40 610 of the main flexural reinforcement (Vd), aggregate inter-
Height, mm 100 1220 locking (Va), and fibers’ mechanisms along the inclined
crack (Vf). However, the shear strength of SFRC beams is
Effective depth (d), mm 80 1118
a more complex phenomenon that is influenced by other
Shear span-depth ratio (a/d) 0.46 6 parameters, such as shear span-depth ratio (that is, arch
Concrete compressive strength (fc′), action), size effect, and concrete type (that is, lightweight
13.5 172
MPa or normalweight). The governing parameters affecting the
Concrete density, kg/m3 1775 2606 shear strength of SFRC beams are discussed in the following
sections. It should be noted that the contribution of aggre-
Longitudinal reinforcement ratio
(ρs), %
0.38 5.8 gate interlock (Va) to the shear capacity of SFRC beams was
neglected in this study to expand the scope of the developed
Fiber volume (ρf), % 0.2 4.5
model to include, for example, high-performance cement-
Fiber aspect ratio (lf/df) 25 133 based composites that are typically produced without coarse
Straight, plain, flat, conical-end, aggregates.
recycled, crimped, wire,
Fiber type
corrugated, hooked, duoform, and
indented *
Note: 1 mm = 0.0394 in; 1 MPa = 145 psi; 1 kg/m3 = 0.06243 lb/ft3. from ACI headquarters for a fee equal to the cost of reproduction plus handling at the
time of the request.
Fig. 1—Forces along inclined shear crack in SFRC beam without stirrups.

Fig. 2—Scatter plot of affecting parameters versus ultimate shear strength.
Concrete compressive strength increases. This indicates a considerable influence that may
As shown in Fig. 2(a), the scatter plot indicates that come from the dowel action of longitudinal reinforcement.
the shear strength of SFRC beams generally exhibits an Taylor26 reported that the contribution of dowel action to the
increasing trend as the compressive strength (fc′) increased. beam shear strength ranged from 15 to 25% when no stirrups
According to Taylor,26 in normal concrete beams without were used. In some design equations and existing empirical
stirrups, the compression shear zone was approximately models, the influence of the longitudinal reinforcement ratio
accountable for 20 to 40% of the total shear strength. was commonly taken as
The proportional relationship between the concrete shear
capacity and its compressive strength is evident in most vu s 2 (2)
c
standards design equations. For example, ACI1 and CSA27

suggested that the shear strength of beams is proportional to where c2 ranged from 0.22 to 1.
the square root of their concrete’s compressive strength—
that is, (fc′)1/2—whereas Eurocode 2 (EC2)28 proposed the Contribution of steel fibers to shear strength
shear strength as proportional to (fc′)1/3. Therefore, the effect Considering Fig. 1, the maximum stress transferred
of compressive strength on the shear capacity of SFRC between the crack faces by a single fiber is governed by
beams (vu) can be generally expressed as follows either the tensile strength of the fiber or the fiber-matrix
interfacial bond strength (τ). Because the tensile strength of
vu f c 1 (1)
c
steel fibers is typically reported to be greater than 1150 MPa,
it is expected that the fiber would be pulled out rather than be
where the exponent c1 determines the rate of increases in allowed to yield. This is particularly relevant considering the
the shear capacity of beams with increasing the concrete fibers’ relatively short length. Consequently, the fiber-matrix
compressive strength. interfacial bond strength more likely controls the maximum
stress limit that can be bridged within the cracked section.
Longitudinal reinforcement ratio (ρs) The force required to pull out a single fiber (Tif-pullout) is
The scatter plot in Fig. 2(b) shows that the shear strength assessed as follows
generally increases as the longitudinal reinforcement ratio
Tif pullout Af 0 (3)

and subsequently, the total force required for all fibers of the surrounding matrix can significantly limit the capa-
(Tf-pullout) should be bility of the models proposed to predict the shear strength
for different concrete types. Other researchers considered
T f pullout 1n f Tif pullout 1n f Af (4) the influence of surrounding matrix strength on quantifying
the value of τ. For example, Naaman31 suggested the value
where φ1 is a factor for taking the variation of the fibers’ of τ for hooked steel fibers to be double the tensile strength
orientation into account, because all fibers are not typically of concrete. Fantilli and Vallini32 also proposed the τ value
oriented perpendicular to the crack surface; nf is the number as a proportion of the square root of the concrete compres-
of fibers along the crack path; and Af is the surface area of a sive strength. Therefore, the fiber’s shape and characteristics
single fiber. of the surrounding matrix must be considered to obtain a
reasonable expression for the fiber-matrix interfacial bond
Number of fibers (nf) strength. In this study, τ was assumed to be a function of both
To identify the nf, a strip with a width equal to the fiber the fibers’ shape and the compressive strength of concrete, as
length (lf) was assumed along the critical crack path (that is, follows
where the crack is the centerline of the strip). Then, the nf
that should be incorporated within the strip is D f f c 3 (8)
c
nf
volume of fibers in the taken strip 2 f vc (5) where Df is the shape factor taken in this study as 0.5, 0.75,
volume of a single fiber 2 and 1 for straight, deformed (for example, crimped and
d l
4 f f hooked), and indented fibers, respectively, similar to those
suggested by Narayanan and Darwish9; and c3 represents
where φ2 is a factor assumed to consider the nonunifor- the rate of growth or decay in the fiber-matrix interfacial
mity of fiber dispersion within the matrix; ρf is the volume bond strength with changing the compressive strength of the
percentage of fibers; df is the fiber diameter; lf is the fiber surrounding matrix. From Eq. (3) to (8),
length; vc is the volume of concrete strip taken and is calcu-
lated as vc = b × lc × lf; b is the beam width; lc is the length of
the crack path (= dv/sinθ); dv is taken as 0.9d; d is the effective 2 f bd v
3l f D f f c (9)
c3
shear depth; and θ is the crack’s angle, which is assumed in T f pullout 1n f Af 1
1
d f sin
different models as 45 degrees for simplification, but from
4
the literature, the θ value was found to have a wide range
from 20 to 40 degrees. Accordingly, nf can be evaluated as The total contribution of fibers to the shear strength of
SFRC beams can be summarized as

2 f bd v lf
D f f c 3 cot bd v
c
nf (6) V f T f pullout cos 412 3 f
d 2f sin df
4
(10)
Fiber surface area (Af)
To estimate the bridging capacity of fibers, the surface where ρf(lf/df)Df = F (fiber factor).
area of a single fiber (Af) must be identified. The value of Af In previous models, the influence of fiber factors (F) was
is taken as (πdfla), where df is the fiber diameter, and la is the taken as proportional to F c4, where c4 was commonly taken
anchorage length of the fiber. Fibers typically have different as 0.5 or 1. By taking C = 4φ1φ2φ3cotθ, the relationship
anchorage lengths at the cracked section. In the available between the fibers’ contribution and the shear strength of
literature, the anchorage length of fibers was commonly beams can be simplified as
assumed as one-quarter of the fiber length (lf).9 In this study,
however, la was not assumed as a specified value but instead vu Cf cc3 F c4 (11)
taken as a percentage of the fiber length—that is, la = φ3lf,
where φ3 is a factor assumed to consider the nonuniformity Shear span-effective depth ratio (a/d)
of the fibers’ anchorage length at the cracked section. As Several studies reported the high influence of the a/d on
such, each fiber’s surface area can be calculated as the shear capacity of SFRC beams. As shown in Fig. 2(c),
the shear strength appeared to exponentially decrease with
Af d f 3l f (7) increasing the a/d values. This is attributed to the arch
action that occurs when a load is applied at a relatively short
Fiber-matrix interfacial bond strength (τ) distance from a support such that a considerable portion
The available literature contains different values proposed of the load is directly transferred to the support through
for τ. For example, τ was suggested to be equal to 4.15, 5.12, fan-like diagonal compression. Therefore, the contribution
and 6.8 MPa for straight, crimped, and hooked fibers, respec- of arch action increases as the value of a/d decreases. The
tively.18,29,30 However, assuming such constant values for τ relationship between vu and a/d can be expressed as
(based on the fibers’ shape) and ignoring the characteristics

c5 MULTI-GENE GENETIC PROGRAMMING
d
vu (12) As mentioned earlier, GP is an AI-based technique that is
a typically used to uncover complex mathematical interrela-
where, in existing models, c5 varied between 0.25 and 1. tionships that cannot be explored using classical mathemat-
ical and statistical tools. Inspired by the theory of natural
Concrete density selection developed by Darwin40 that relies on the survival
As recommended by ACI,1 CSA,27 and EC2,28 the shear of the fittest, GP adopts a set of genetic operators that prevent
strength of concrete beams cast with semi-lightweight or the solution from being trapped in local optima. Due to its
lightweight mixtures must be modified to consider the effect powerful capability in identifying unknown mathematical
of lightweight aggregates on the shear strength of beams. relationships, GP has been extensively used in numerous
The model proposed in this study considered the modifica- civil engineering areas including soil mechanics,41 water
tion factor λ as suggested by ACI,1 as follows resources,42,43 transportation,44,45 and recently structural
engineering.20-23
0.85 for sand-lightweight concrete The application of GP starts by defining terminal and
ACI (13) function sets, where the former contains the model inputs
0.75 for all lightweigght concrete
and any constants suggested, whereas the latter encom-
passes the mathematical and logical operators. An initial
Size effects population of randomly generated mathematical expres-
Previous studies conducted by Kani,33 Bažant and Kim,34 sions is then assumed, where each expression represents a
and Shioya et al.35 demonstrated the effect of size on the candidate solution within the search space and is typically
shear strength of RC beams without stirrups. They reported referred to as an individual. A fitness value is subsequently
that the likelihood of shear stress-induced failure decreases assigned to each individual, reflecting its ability to replicate
as the beam depth increases and referred to that as the size the expected output. Such a fitness value is calculated based
effect (K). The results obtained from Shioya et al.35 showed on a prespecified objective function or a model performance
that large-scale beams with a depth greater than 1000 mm criterion (for example, correlation coefficient or root-mean-
failed at a shear stress equal to approximately one-third of square error [RMSE]). Genetic operations (that is, elitism,
that which caused failure in small-scale beams with a depth crossover, mutation) are then applied to the initial popula-
of less than 200 mm. The failure of the largest tested beam tion to produce newer generations, where: 1) elitism refers to
(depth of 3000 mm) occurred at a value less than half of replicating individuals with higher fitness values in the next
the value estimated based on the ACI design code. The generations; 2) crossover represents combining two indi-
size effect was attributed to the wider cracks that typically viduals to produce two offspring (Fig. 3); and 3) mutation
developed in large beams, where crack widths are greater refers to randomly changing single individuals to produce
than their roughness, and this in turn diminishes shear new ones (Fig. 3). New generations of candidate solutions
transmission across cracks.36,37 Such findings highlight the are continuously produced until reaching a predetermined
importance of considering the size effect in shear capacity termination criterion such as a specific number of genera-
prediction to avoid unexpected sudden failures. In this study, tions, allowable fitness value, certain computational cost, or
the size effect was taken according to Eq. (14), which was a combination of two or more criteria.
proposed by Walraven38,39 for short and slender members. When MGGP is employed, multiple population sets are
generated, and genetic operations are applied to each set sepa-
K 1
200
(14) rately.22 A high-level crossover can also be allowed, where
d individuals from different population sets are combined to
produce offspring. Such a crossover process increases the
General formula for shear strength diversity within each population set, which most often results
Considering the key parameters affecting the shear in individuals with higher fitness values. The best individual
strength of concrete presented through Eq. (1) to (14), the from each population set is obtained after termination, and
shear strength of SFRC beams constructed without stirrups, these best individuals are subsequently combined using a
can be generally formulized as follows linear regression analysis.25 This pseudo-linear combination
is similar to ensemble averaging applied in other AI tech-
niques (for example, decision tree), which has been shown
d
c5

vu f K , , f c 1 , s 2 , Cf cc3 F c4 ,
c c
(15) to produce high-performance models.46-48
a Table 2 shows the MGGP parameters employed in the

present study, where the fitness function used combines the
With no prior assumptions, MGGP was employed to
RMSE and the correlation coefficient (R) to represent an error
discover the constants [that is, c1, c2, c3, c4, c5, and C= f
estimator and a linear dependency measure, respectively.
(φ1φ2φ3cotθ)] and functions that relate the influencing
It should be noted that MGGP might be terminated before
parameters to the ultimate shear strength.
achieving the desired fitness value (that is, the absolute
minimum, as shown in Table 2), resulting in a local optimal
solution. As such, the MGGP can be applied multiple times
using different initial population sets to ensure the global

Fig. 3—Schematic of: (a) crossover; and (b) mutation used when GP is applied.
Table 2—MGGP parameters used in this study converged quickly without affecting the global optimality of
Parameter Value
the resulting solution.
Initialization
Final MGGP-based model
Terminal set fc′, ρs, a/d, and F Different trials were performed by changing the MGGP
Function set +, ×, ÷, square root, and power parameters, and thus different models were generated. During
Fitness function RMSE + 1/R2
each trial, 70% of the data set (shown in Appendix B) was
used for the model training, whereas the remaining 30% was
Population size 200
kept for testing. After each trail, the generated model was
Reproduction evaluated through its goodness of fit for both the training and
Elitism = 0.05 testing subsets, the number of important factors considered
Genetic operation ratio Crossover = 0.8 (satisfying all guidelines suggested in Eq. (1) to (15)), the
Mutation = 0.15 model dimensions, and the model simplicity. Equation (16)
Termination shows the best-performing model corresponding to MGGP
Maximum number of
parameters listed in Table 2
200
generations
da
0.75
Allowable fitness value 1.0 vu K 6.7s 0.5 F 0.25
f c (16)
Multi-gene related
Number of genes 3 where K, λ, ρs, d/a, and F are dimensionless, while fc′ is in
MPa units.
optimality of the resulting solution. It should also be noted Figures 4(a) and (b) show the correlation between the
that the convergence rate of MGGP is governed by the inter- MGGP-based model’s estimates against the experimental
play between the population size and the maximum number observations for the training and testing subsets. The figures
of generations specified. A large population size is typically show that the model successfully captures the complex rela-
used to increase the possibility of obtaining the optimal tionship between the governing parameters and the shear
solution at earlier generations49; however, model diversity strength of SFRC beams with an R2 value of 0.94. The model
introduced by crossover and mutation may increase the predictions were both accurate and precise, in which the vuexp/
computational time required.50,51 Therefore, the selection vuGP ratios had a mean (μ) of 1.04, standard deviation (SD)
of the population size and maximum number of generations of 0.30, and coefficient of variation (CoV) of 28.6%. The
is typically carried out iteratively such that the MGGP is model also showed an excellent performance for the testing

Fig. 4—Plot of MGGP predictions against experimental shear strengths: (a) training; and (b) testing.
Table 3—Evaluation measures for models’ performance
Rank
Overall
Model R 2
RMSE AAE μ SD CoV, % R 2
RMSE AAE μ SD CoV, % rank
Proposed model 0.94 1.06 0.80 1.02 0.28 27.6 1 1 1 2 1 1 1
Shahnewaz and Alam 21
0.93 1.23 0.86 1.09 0.33 30.3 2 2 2 4 3 3 2
Gandomi et al. 22
0.91 1.3 0.89 1 0.31 31.3 6 3 3 1 2 6 3
Kwak et al. 14
0.87 1.79 0.99 1.08 0.34 31.2 10 7 6 3 4 5 4
Sarveghadi et al.23 0.91 1.34 0.94 1.19 0.42 35 7 4 4 8 9 9 5
Shin et al.12
0.88 1.57 1.05 1.15 0.42 36.7 8 6 7 5 8 11 6
Narayanan and Darwish 9
0.77 2.24 1.11 1.25 0.37 29.9 14 10 9 9 6 2 7
Chaabene and Nehdi 24
0.92 1.5 0.97 1.29 0.53 41.2 3 5 5 10 12 15 8
Ashour et al. (b)10 0.83 1.91 1.13 1.18 0.41 35.2 12 8 10 7 7 10 9
Al-Ta’an and Al Feel 15
0.75 2.33 1.1 1.18 0.37 31.5 16 14 8 6 5 7 10
Khuntia et al. 13
0.92 2.25 1.72 1.8 0.55 30.8 4 11 16 16 13 4 11
Sharma 7
0.87 2 1.28 1.3 0.5 38.3 11 9 12 11 10 12 12
Shahnewaz and Alam20 0.92 2.31 1.13 1.55 1.19 76.7 5 12 11 13 16 16 13
Swamy et al.11 0.88 2.33 1.67 1.71 0.58 34.2 9 13 15 15 14 8 14
Ashour et al. (a) 10
0.76 2.34 1.34 1.3 0.5 38.4 15 15 13 12 11 13 15
Mansur et al. 8
0.83 2.37 1.58 1.62 0.66 41.1 13 16 14 14 15 14 16
subset (R2 of 0.96, as shown in Fig. 4(b)), even with a higher MAE, and CoV) and the score increases above 1 as the R2
precision as the vuexp/vuGP ratios had μ, SD, and CoV of 0.97, value decreases or the RMSE, MAE, and CoV increases,
0.24, and 24.4%, respectively. These measures support the indicating lower performance. An overall rank was then
predictability and generalizability of the model. assigned to each model based on the corresponding average
score. From Table 3, the MGGP model proposed in this
Model performance comparison study outperformed the existing models with R2, RMSE, and
The performance of the proposed MGGP-based model MAE values of 0.94, 1.06, and 0.78, respectively. In addi-
was compared to that of 15 existing models proposed by tion, the vexp/vpred ratios obtained by the proposed model had
other researchers (all models are listed in Appendix A). a mean value of 1.05 with the lowest SD and CoV values
For each model, the accuracy of predictions (vpred) against (0.29 and 27.6%, respectively) among all models, which
the experimental results (vexp) was evaluated by calculating indicate the high accuracy and the consistent prediction
the coefficient of determination (R2) and RMSE values, capability (supported by the lower dispersion around the
together with the mean absolute error (MAE). The values mean value). The superiority of the proposed model, even
of μ, SD, and CoV of the vexp/vpred ratios yielded by each when compared to the other two models developed using
model were also obtained. All these measures are presented classical GP, can be attributed to the comprehensive data set
in Table 3 for all models, along with the model performance used in the current study, which has never been employed in
ranking. The ranking system used a score of 1 for the best full to develop any of the existing models. In addition, the
performing model (that is, the highest R2 and lowest RMSE, mechanics-guided approach adopted in this study enabled

Fig. 5—Residual analysis results: (a) error frequency compared to theoretical PDF of corresponding normal distribution; and
(b) relationship between model errors and estimates.
the development of an elegant model that is dimensionally
consistent and conforms to mechanics and previous research
findings—facilitating adoptability by design engineers
and standards committees. It is important to recall that the
second- and third-ranked predictive models in Table 3 were
also developed using traditional GP (Shahnewaz and Alam21
and Gandomi et al.,22 respectively), which indicates the high
capability of GP and MGGP in predicting complex structural
behaviors.
Residual analysis
Because the shear strength model developed in the present
study (that is, Eq. (16)) resulted from regressively combining
three GP-based mathematical expressions, a residual analysis
is essential to ensure that the model residuals are indepen-
dent and identically distributed. This requires the model
errors to be independent and identically distributed (IID)
through following a normal distribution and being indepen- Fig. 6—Effect of parameters studied on shear capacity of
dent of the model estimates. The relationships presented SFRC beam, where V refers to any independent variable
in Fig. 5 support that the MGGP model residuals are IID, from Eq. (16).
and thus justifies the validity of the pseudo-linear relation-
ship presented in Eq. (16) as an SFRC beam shear strength contribution of compression zone in resisting shear force,
predictive model. but also implicitly accounts for the fiber-matrix interfacial
bonding strength, as explained in Eq. (3) to (11).
Parameter analysis
The relationships between the studied parameters and the Sensitivity analysis
ultimate shear strength of SFRC beams are shown in Fig. 6. Identifying the parameters governing the variability of
This figure was drawn by changing only a single parameter vu within the model predictions has been one of the key
within the considered range, while the other parameters objectives behind this study. Conducting a global sensi-
were set at their corresponding mean values. Figure 6 shows tivity analysis (GSA) is thus key to investigating the vari-
the positive correlation between vu and ρs, fc′, and F, where ability in the model output for the whole range of the input
increasing these parameters boosts the contribution of dowel parameters, especially considering the input-output highly
action, compression zone and fibers, and subsequently the nonlinear relationships. In the present study, a variogram-
beam’s overall shear resistance. On the other hand, the and a variance-based GSA were applied following the proce-
increase in a/d decreases the contribution of the arch action dures described in Razavi and Gupta52,53 and Saltelli et al.,54
in resisting the applied shear force, which yields a reduction respectively. The former enables assessing the output vari-
in the beam’s shear capacity. The increase in the beam depth ability at different perturbation scales of the inputs, whereas
(d) (corresponding to a decrease in the K value) also reduces the latter is concerned with the infinite perturbation of the
vu due to the size effect. It should be noted that the compres- inputs.53 It is important to note that both GSA approaches
sive strength in the developed model not only refers to the consider the interaction (not the interdependence) between

Fig. 7—Global senstivity analysis results based on: (a) variogram; and (b) variance, where γ(h) is variogram value at pertur-
bation scale h.
the input parameters; however, variance-based GSA facili- 3.855,56; and η is the standard deviation of vuexp/(a × vuGP).
tates segregating the net effect of each input parameter. Figure 8(a) shows that vuexp/vuGP can be fitted by a lognormal
Figure 7 shows the GSA analysis results based on the two distribution, and a comparison between vuexp and vuGP for the
approaches. It is observed that the variability in the shear 752 tests employed in the present study shows that a value
strength of SFRC beams is primarily controlled by that of a = 0.989 can be used (Fig. 8(b)). Based on the previous
of a/d, followed by fc, K, and the fiber properties (that is, discussion, a ψ value of 0.4 is suggested in the present study
F), as supported by the variogram, net effect (S), and total as a shear strength reduction factor.
effect (ST) of each input parameter. This justifies the unique
behavior of SFRC beams compared to concrete beams CONCLUSIONS
without steel fibers and highlights the need for different This study proposed a new approach to develop a
models to describe the behavior of each system. It is note- mechanics-guided multi-gene genetic programming
worthy that the contribution of fiber to the shear strength (MGGP)-based model that can efficiently predict the shear
of SFRC beams is primarily due to its interaction with the strength of steel fiber-reinforced concrete (SFRC) beams.
concrete properties (that is, fc), loading conditions (that is, An extensive data set of 752 test results was collected from
a/d), and other factors (for example, K) rather than the direct previous studies and used for model development and testing.
effects of its own characteristics, as supported by the posi- The performance of the model was compared to that of 15
tive ST-S values (Fig. 7). existing models that were developed either using empirical
approaches and classic regression analysis or using classical
Model adaptation for design genetic programming (GP). Statistical measures were used
Notwithstanding the efficacy of the proposed MGGP- for accuracy comparison and thus determining the best-per-
based model in reproducing the experimental shear strength forming model. The following conclusions can be drawn
observations as supported by the relatively high R2 values, based on the main findings:
the corresponding vuexp/vuGP ratios suggest that Eq. (16) may 1. The results of the present study demonstrated the
overestimate the shear capacity, resulting in possible beam power of MGGP in identifying nonlinear relationships
collapse under design loads. Such a situation is not unex- governing complex phenomena such as the shear strength
pected considering that data-driven models are developed of SFRC beams. This indicates promising potentials for GP
to best-fit the (experimental) observations. Therefore, the and MGGP in modeling other ill-defined behaviors in the
majority of design standards and guidelines adopt reduced field of structural engineering that are difficult to address
shear strength values (for example, vud) for design purposes. analytically.
Accordingly, vud corresponding to the MGGP-based model 2. The procedures followed in this study highlighted
developed in the present study can be expressed as the importance of producing GP models that conform to
mechanics and well-understood engineering knowledge.
d
0.75
Although the lack of mechanics guidance may yield models
vud K 6.7s 0.5 F 0.25 f c (17) with good prediction capability, such models may nonethe-
a less include illogical or uninterpretable combinations of the
where ψ is the strength reduction factor employed to governing parameters (evident by other GP-based models
decrease the likelihood of overestimating the shear strength developed in earlier studies).
and is estimated as ψ = a × exp(–αβη – 0.5η2) when vuexp/ 3. Data-driven models are highly dependent on the quality,
vuGP follows a lognormal distribution55,56; a is the slope of quantity, and diversity of the data set used for their training.
the relationship between vuexp and vuGP56; α is a sensitivity The diverse data set used in this study helped to develop
factor assumed as 0.855,56; β is a reliability index taken as a well-trained model, which was able to yield accurate

Fig. 8—Comparison between: (a) statistical distribution of vuexp/vuGP and corresponding theoretical lognormal distribution;
and (b) vuexp and vuGP.
predictions even against experimental results from other models to simulate the behavior of complex, nonlinear interdependent
systems.
data sets that were not used for training. This supports the
generalizability of the developed model. Wael El-Dakhakhni is a Professor in the Department of Civil Engineering
4. Among all existing models that were considered for and School of Computational Science and Engineering at McMaster Univer-
sity. His research interests include the behavior of concrete and masonry
comparison, the developed MGGP-based model not only structures under extreme loading, data analytics, and performance predic-
achieves the best prediction accuracy and consistency but tion models.
is also characterized by being simple, elegant, and easy
to apply. Such characteristics render the proposed model REFERENCES
a strong candidate, compared to other existing models, 1. ACI Committee 318, “Building Code Requirements for Structural Concrete
(ACI 318-19) and Commentary (ACI 318R-19),” American Concrete Institute,
for adoption by design engineers and code and standards Farmington Hills, MI, 2019, 624 pp.
committees. 2. Ismail, M. K., and Hassan, A. A. A., “Shear Behaviour of Large-Scale
5. The model parameter analysis showed a positive influ- Rubberized Concrete Beams Reinforced with Steel Fibres,” Construc-
tion and Building Materials, V. 140, 2017, pp. 43-57. doi: 10.1016/j.
ence of the ρs, fc′, and F on the shear strength of SFRC conbuildmat.2017.02.109
beams, whereas the increase of a/d (that is, decreased arch 3. Ismail, M. K.; Hassan, A. A. A.; and Hussein, A. A., “Structural
action) led to a reduction in the shear strength. The proposed Behaviour of Reinforced Concrete Beams Containing Crumb Rubber
and Steel Fibres,” Magazine of Concrete Research, V. 69, No. 18, 2017,
model also considers the size effect (K) on shear stress- pp. 939-953. doi: 10.1680/jmacr.16.00525
induced failure. 4. Ismail, M. K., and Hassan, A. A. A., “An Experimental Study on
6. The sensitivity analysis indicated that the variability Flexural Behaviour of Large-Scale Concrete Beams Incorporating Crumb
Rubber and Steel Fibres,” Engineering Structures, V. 145, 2017, pp. 97-108.
in the shear strength of SFRC beams is primarily governed doi: 10.1016/j.engstruct.2017.05.018
by that of a/d, followed by fc′, K, and the fiber properties 5. Yoo, D.-Y.; Banthia, N.; and Yoon, Y.-S., “Impact Resistance of Rein-
(that is, Df, ρf, and lf/df). These results conform with previous forced Ultra-High-Performance Concrete Beams with Different Steel Fibers,”
ACI Structural Journal, V. 114, No. 1, Jan.-Feb. 2017, pp. 113-124. doi:
research findings. 10.14359/51689430
7. To facilitate code adoption, the design-ready MGGP- 6. Isojeh, B.; El-Zeghayar, M.; and Vecchio, F. J., “Fatigue Resistance
based model proposed in this study can be used for reliable of Steel Fiber-Reinforced Concrete Deep Beams,” ACI Structural Journal,
V. 114, No. 5, Sept.-Oct. 2017, pp. 1215-1226. doi: 10.14359/51700792
shear strength predictions of SFRC beams without stirrups. 7. Sharma, A. K., “Shear Strength of Steel Fiber Reinforced Concrete
Beams,” ACI Journal Proceedings, V. 83, No. 4, July-Aug. 1986,
DATA AVAILABILITY STATEMENT pp. 624-628.
8. Mansur, M. A.; Ong, K. C. G.; and Paramasivam, P., “Shear Strength
The data set used in this study is available in Lantsoght57 of Fibrous Concrete Beams Without Stirrups,” Journal of Structural
and other references that are listed in Appendix B. Engineering, ASCE, V. 112, No. 9, 1986, pp. 2066-2079. doi: 10.1061/
(ASCE)0733-9445(1986)112:9(2066)
9. Narayanan, R., and Darwish, I. Y. S., “Use of Steel Fibers as Shear
AUTHOR BIOS Reinforcement,” ACI Structural Journal, V. 84, No. 3, May-June 1987,
Mohamed K. Ismail is a Postdoctoral Fellow in the Department of Civil
pp. 216-227.
Engineering at McMaster University, Hamilton, ON, Canada, and an
10. Ashour, S. A.; Hasanain, G. S.; and Wafa, F. F., “Shear Behavior of
Assistant Professor in the Department of Structural Engineering at Cairo
High-Strength Fiber Reinforced Concrete Beams,” ACI Structural Journal,
University, Giza, Egypt. His research interests include structural appli-
V. 89, No. 2, Mar.-Apr. 1992, pp. 176-184.
cations of sustainable construction materials, strengthening and rehabil-
11. Swamy, R. N.; Jones, R.; and Chiam, A. T. P., “Influence of Steel
itation of concrete structures, and modeling complex structural behaviors
Fibers on the Shear Resistance of Lightweight Concrete I-Beams,” ACI
using artificial intelligence and data-driven models.
Structural Journal, V. 90, No. 1, Jan.-Feb. 1993, pp. 103-114.
12. Shin, S. W.; Oh, J. G.; and Ghosh, S. K., “Shear Behavior of Laboratory-
Ahmed Yosri is a Postdoctoral Fellow in the Department of Civil Engi-
Sized High-Strength Concrete Beams Reinforced with Bars and Steel
neering at the INTERFACE Institute for Multi-Hazard and Systemic Risk
Fibers,” Fiber Reinforced Concrete: Developments and Innovations,
Studies at McMaster University, and an Assistant Professor in the Depart-
SP-142, J. I. Daniel and S. P. Shah, eds., American Concrete Institute, Farm-
ment of Irrigation and Hydraulics Engineering at Cairo University. His
ington Hills, MI, 1994, pp. 181-200.
research interests include the development of artificial intelligence-based

13. Khuntia, M.; Stojadinovic, B.; and Goel, S. C., “Shear Strength of 36. Collins, M. P., and Kuchma, D., “How Safe Are Our Large, Lightly
Normal and High-Strength Fiber Reinforced Concrete Beams without Stir- Reinforced Concrete Beams, Slabs, and Footings?” ACI Structural Journal,
rups,” ACI Structural Journal, V. 96, No. 2, Mar.-Apr. 1999, pp. 282-289. V. 96, No. 4, July-Aug. 1999, pp. 482-491.
14. Kwak, Y.-K.; Eberhard, M. O.; Kim, W.-S.; and Kim, J., “Shear 37. Reineck, K. H., “Mechanical Model for the Behavior of Reinforced
Strength of Steel Fiber-Reinforced Concrete Beams without Stirrups,” ACI Concrete Members in Shear,” PhD thesis, University of Stuttgart, Stuttgart,
Structural Journal, V. 99, No. 4, July-Aug. 2002, pp. 530-538. Germany, 1990, 273 pp.
15. Al-Ta’an, S. A., and Al-Feel, J. R., “Evaluation of Shear Strength 38. Walraven, J. C., “Size Effects: Their Nature and Their Recognition in
of Fibre-Reinforced Concrete Beams,” Cement and Concrete Composites, Building Codes,” Studi e Ricerche, V. 16, 1995, pp. 113-134.
V. 12, No. 2, 1990, pp. 87-94. doi: 10.1016/0958-9465(90)90045-Y 39. Walraven, J., and Lehwalter, N., “Size Effects in Short Beams
16. ACI Committee 544, “Design Considerations for Steel Fiber Rein- Loaded in Shear,” ACI Structural Journal, V. 91, No. 5, Sept.-Oct. 1994,
forced Concrete (ACI 544.4R-88),” American Concrete Institute, Farm- pp. 585-593.
ington Hills, MI, 1988, 18 pp. 40. Darwin, C., The Origin of Species by Means Of Natural Selection,
17. Zsutty, T., “Shear Strength Prediction for Separate Categories of Random House Inc., 1859.
Simple Beam Tests,” ACI Journal Proceedings, V. 68, No. 2, Feb. 1971, 41. Parasuraman, K.; Elshorbagy, A.; and Si, B. C., “Estimating Satu-
pp. 138-143. rated Hydraulic Conductivity Using Genetic Programming,” Soil Science
18. Swamy, R. N.; Mangat, P. S.; and Rao, C. V. S. K., “The Mechanics Society of America Journal, V. 71, No. 6, 2007, pp. 1676-1684. doi:
of Fiber Reinforcement of Cement Matrices,” Fiber Reinforced Concrete, 10.2136/sssaj2006.0396
SP-44, American Concrete Institute, Farmington Hills, MI, 1974, pp. 1-28. 42. Chadalawada, J.; Havlicek, V.; and Babovic, V., “A Genetic Program-
19. Rudin, C., “Stop Explaining Black Box Machine Learning Models ming Approach to System Identification of Rainfall-Runoff Models,” Water
for High Stakes Decisions and Use Interpretable Models Instead,” Nature Resources Management, V. 31, No. 12, 2017, pp. 3975-3992. doi: 10.1007/
Machine Intelligence, V. 1, No. 5, 2019, pp. 206-215. doi: 10.1038/ s11269-017-1719-1
s42256-019-0048-x 43. Yosri, A.; Siam, A.; El-Dakhakhni, W.; and Dickson-Anderson, S., “A
20. Shahnewaz, M., and Alam, M. S., “Improved Shear Equations Genetic Programming–Based Model for Colloid Retention in Fractures,”
for Steel Fiber-Reinforced Concrete Deep and Slender Beams,” ACI Ground Water, V. 57, No. 5, 2019, pp. 693-703. doi: 10.1111/gwat.12860
Structural Journal, V. 111, No. 4, July-Aug. 2014, pp. 851-860. doi: 44. Das, S.; Raju, N.; Kumar Maurya, A.; and Arkatkar, S., “Evalu-
10.14359/51686736 ating Lateral Interactions Of Motorized Two-Wheelers Using Multi-Gene
21. Shahnewaz, M., and Alam, M. S., “Genetic Algorithm for Predicting Symbolic Genetic Programming,” Transportation Research Record:
Shear Strength of Steel Fiber Reinforced Concrete Beam with Parameter Journal of the Transportation Research Board, V. 2674, No. 9, 2020,
Identification and Sensitivity Analysis,” Journal of Building Engineering, pp. 1120-1135. doi: 10.1177/0361198120934476
V. 29, 2020, p. 101205. doi: 10.1016/j.jobe.2020.101205 45. Patnaik, A. K.; Agarwal, L. A.; Panda, M.; and Bhuyan, P. K.,
22. Gandomi, A. H.; Alavi, A. H.; and Yun, G. J., “Nonlinear Modeling “Entry Capacity Modelling of Signalized Roundabouts under Hetero-
of Shear Strength of SFRC Beams Using Linear Genetic Programming,” geneous Traffic Conditions,” Transportation Letters, V. 12, No. 2, 2020,
Structural Engineering and Mechanics, V. 38, No. 1, 2011, pp. 1-25. doi: pp. 100-112. doi: 10.1080/19427867.2018.1533160
10.12989/sem.2011.38.1.001 46. Murray, S. A., “The Importance of Ensemble Techniques for Oper-
23. Sarveghadi, M.; Gandomi, A. H.; Bolandi, H.; and Alavi, A. H., ational Space Weather Forecasting,” Space Weather, V. 16, No. 7, 2018,
“Development of Prediction Models for Shear Strength of SFRCB Using pp. 777-783. doi: 10.1029/2018SW001861
a Machine Learning Approach,” Neural Computing & Applications, V. 31, 47. Tegegne, G.; Melesse, A. M.; and Worqlul, A. W., “Develop-
No. 7, 2019, pp. 2085-2094. doi: 10.1007/s00521-015-1997-6 ment of Multi-Model Ensemble Approach for Enhanced Assessment of
24. Chaabene, W. B., and Nehdi, M. L., “Genetic Programming Based Impacts of Climate Change on Climate Extremes,” The Science of the
Symbolic Regression for Shear Capacity Prediction of SFRC Beams,” Total Environment, V. 704, 2020, Article No. 135357. doi: 10.1016/j.
Construction and Building Materials, V. 280, 2021, p. 122523. doi: scitotenv.2019.135357
10.1016/j.conbuildmat.2021.122523 48. Chen, C.-H.; Tanaka, K.; Kotera, M.; and Funatsu, K., “Comparison
25. Searson, D.; Leahy, D.; and Willis, M., “GPTIPS: An Open-Source and Improvement of the Predictability and Interpretability with Ensemble
Genetic Programming Toolbox For Multigene Symbolic Regression,” Learning Models in QSPR Applications,” Journal of Cheminformatics,
Proceedings of the International MultiConference of Engineers and V. 12, No. 1, 2020, Article No. 19, 16 pp. doi: 10.1186/s13321-020-0417-9
Computer Scientists, V. 1, 2010, pp. 17-20. 49. Gotshall, S., and Rylander, B., “Optimal Population Size and the
26. Taylor, H. P. J., “The Fundamental Behavior of Reinforced Concrete Genetic Algorithm,” World Scientific and Engineering Academy and
Beams in Bending and Shear,” Shear in Reinforced Concrete, SP-42, Amer- Society (WSEAS) International Conference on Soft Computing, Optimiza-
ican Concrete Institute, Farmington Hills, MI, 1974, pp. 43-77. tion, Simulation, and Manufacturing Systems (SOSM 2002), 2002.
27. CSA A23.3-04, “Design of Concrete Structures,” CSA Group, 50. Koljonen, J., and Alander, J. T., “Effects of Population Size and Rela-
Toronto, ON, Canada, 2004. tive Elitism on Optimization Speed and Reliability of Genetic Algorithms,”
28. EN 1992-1-1, “Eurocode 2 – Design of Concrete Structures – Part 1– Proceedings of the Ninth Scandinavian Conference on Artificial Intelli-
1: General Rules and Rules for Buildings,” European Committee for Stan- gence (SCAI 2006), Espoo, Finland, 2006.
dardization, Brussels, Belgium, 2004, 227 pp. 51. Williams, E. A., and Crossley, W. A., “Empirically-Derived Popu-
29. Lim, T. Y.; Paramasivam, P.; and Lee, S. L., “Shear and Moment lation Size and Mutation Rate Guidelines for a Genetic Algorithm with
Capacity of Reinforced Steel-Fibre-Concrete Beams,” Magazine of Uniform Crossover,” Soft Computing in Engineering Design and Manu-
Concrete Research, V. 39, No. 140, 1987, pp. 148-160. doi: 10.1680/ facturing, P. K. Chawdhry, R. Roy, and R. K. Pant, eds., Springer, London,
macr.1987.39.140.148 UK, 1998, pp. 163-172.
30. Swamy, R. N., and Bahia, H. M., “Effectiveness of Steel Fibers as 52. Razavi, S., and Gupta, H. V., “A New Framework for Compre-
Shear Reinforcement,” Concrete International, V. 7, No. 3, Mar. 1985, hensive, Robust, and Efficient Global Sensitivity Analysis: 1. Theory,”
pp. 35-40. Water Resources Research, V. 52, No. 1, 2016, pp. 423-439. doi:
31. Naaman, A. E., “High-Performance Fiber Reinforced Cement 10.1002/2015WR017558
Composites,” Proceedings of the IABSE Symposium on Concrete Structures 53. Razavi, S., and Gupta, H. V., “A New Framework for Compre-
for the Future, Paris, France, 1987, pp. 371-376. hensive, Robust, and Efficient Global Sensitivity Analysis: 2. Applica-
32. Fantilli, A. P., and Vallini, P., “A Cohesive Interface Model for tion,” Water Resources Research, V. 52, No. 1, 2016, pp. 440-455. doi:
the Pullout of Inclined Steel Fibers in Cementitious Matrixes,” Journal 10.1002/2015WR017559
of Advanced Concrete Technology, V. 5, No. 2, 2007, pp. 247-258. doi: 54. Saltelli, A.; Ratto, M.; Andres, T.; Campolongo, F.; Cariboni, J.;
10.3151/jact.5.247 Gatelli, D.; Saisana, M.; and Tarantola, S., Global Sensitivity Analysis: The
33. Kani, G. N. J., “How Safe Are Our Large Reinforced Concrete Primer, John Wiley & Sons, Inc., New York, 2008, 304 pp.
Beams?” ACI Journal Proceedings, V. 64, No. 3, Mar. 1967, pp. 128-141. 55. Eurocode 2, “Commentary (rev A 31-03-2017): Section 6,” European
34. Bažant, Z. P., and Kim, J.-K., “Size Effect in Shear Failure of Longi- Concrete Platform ASBL, 2008, Brussels, Belgium, 220 pp.
tudinally Reinforced Beams,” ACI Journal Proceedings, V. 81, No. 5, 56. Soetens, T., “Design Models for the Shear Strength of Prestressed
Sept.-Oct. 1984, pp. 456-468. Precast Steel Fibre Reinforced Concrete Girders,” PhD dissertation, Ghent
35. Shioya, T.; Iguro, M.; Nojiri, Y.; Akiyama, H.; and Okada, T., “Shear University, Ghent, Belgium, 2015, 360 pp.
Strength of Large Reinforced Concrete Beams,” Fracture Mechanics: 57. Lantsoght, E. O. L., “Database of Shear Experiments on Steel Fiber
Application to Concrete, SP-118, V. C. Li and Z. P. Bažant, eds., American Reinforced Concrete Beams without Stirrups,” Materials, V. 12, No. 6,
Concrete Institute, Farmington Hills, MI, 1989, pp. 259-279. 2019, pp. 1-36. doi: 10.3390/ma12060917

REVIEWERS IN 2021
In 2021, 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.
A, Selva Ganesa Moorthi Abdalqader, Ahmed

Indian Institute of Technology Madras Queen’s University Belfast
Chennai, Tamil Nadu, India Belfast, United Kingdom of Great Britain and Northern Ireland
A, Sofi Abdel Hadi, Nafeth

VIT University Al-Balqa Applied University
Vellore, India Al-Salt, Jordan
A, Suresh Kumar Abdel-Mohti, Ahmed

Kalasalingam Academy of Research and Education McNeese State University
Krishnan Kovil, Tamil Nadu, India Lake Charles, LA, United States
A. Abbas, Waleed Abdel-Wahab, Noran

University of Technology Houston, TX, United States
Al-Sinaa Street, Baghdad, Iraq
AbdelAleem, Basem
Abavisani, Iman St. John’s, NL, Canada
Semnan University
Sabzevar, Khorasan Razavi, Islamic Republic of Iran Abdelazim, Waseem
Université de Sherbrooke
Abbas, Aamer Sherbrooke, QC, Canada
Baghdad, Iraq
Abdelaziz, Gamal
Abbas, Abdelgadir Benha University Faculty of Engineering at Shoubra
Carleton University Cairo, Egypt
Ottawa, ON, Canada
Abdelbary, Ahmed
Abd El-Aleem, Saleh Alexandria University Faculty of Engineering
Fayoum University-Faculty of Science Alexandria, Egypt
El-Fayoum, Egypt
Abdelgader, Hakim
Abd Elrazek, Mostafa University of Tripoli
Assiut, Egypt Tripoli, Libyan Arab Jamahiriya
Abd Razak, Rafiza Abdellatef, Mohammed

Universiti Malaysia Perlis University of New Mexico
Kangar, Perlis, Malaysia Albuquerque, NM, United States
Abd-Elfatah, Sahar Abdul-Razzaq, Khattab

Higher Institute for Engineering and Technology University of Diyala
Alexandria, Egypt Diyala, Iraq
Abd-Elssamd, Ammar Abdulla, Nwzad

University of Tennessee Salahaddin University
Knoxville, TN, United States Erbil, Iraq
Abdalla, Hany Abdullah, Ahmad

College of Technological Studies Aswan University
Shuwaikh, Kuwait Sahary City, Aswan, Egypt
Abdallah, Amr Abdullah, Saman

University of Manitoba University of California, Los Angeles
Winnipeg, MB, Canada Los Angeles, CA, United States

REVIEWERS IN 2021
Abed, Mohammed Adamczewski, Grzegorz
Rutgers University Warsaw University of Technology
Piscataway, NJ, United States Warsaw, Poland
Abed, Ziyad Adekunle, Saheed

University of Technology King Fahd University of Petroleum and Minerals
Baghdad, Iraq Dhahran, Saudi Arabia
Abellan-García, Joaquin Adesina, Adeyemi

Universidad del Norte University of Windsor
Barranquilla, Colombia Windsor, ON, Canada
Abo-Shadi, Nagi Adil, Goran

Robert Englekirk Inc. University of Missouri–Kansas City
Santa Ana, CA, United States Kansas City, MO, United States
Abouhussien, Ahmed Adinkrah-Appiah, Kwadwo

University of Toronto Sunyani Polytechnic
Toronto, ON, Canada Sunyani, Ghana
Abousnina, Rajab Afif, Rahma

University of Southern Queensland Damascus University Faculty of Civil Engineering
Toowoomba, QLD, Australia Damascus, Syrian Arab Republic
Abraham, Sarah Mariam Agarwal, Pankaj

Birla Institute of Technology and Science, Pilani Indian Institute of Technology Roorkee
Pilani, Rajasthan, India Roorkee, Uttarakhand, India
Abu Obeidah, Adi Aghayari, Reza

Rutgers, The State University of New Jersey Razi University, Faculty of Engineering
Piscataway, NJ, United States Kermanshah, Islamic Republic of Iran
Abu Taqa, Ala Agrawal, Unnati

Qatar University Nagpur, Maharashtra, India
Doha, Qatar
Aguayo, Federico
Abu Yosef, Ali University of Washington Seattle Campus
Pivot Engineers Seattle, WA, United States
Austin, TX, United States
Agustiningtyas, Rudi
Abuzeid, Al-Tuhami Ministry of Public Works
Badr for Consulting and Retrofitting Bandung, Indonesia
Ciro, Nasr City, Egypt
Ahani, Elshan
Accornero, Federico Tabriz, East Azerbaijan, Islamic Republic of Iran
Politecnico di Torino
Torino, Italy Ahmad, Habil
University of Wollongong
Achillopoulou, Dimitra Wollongong, NSW, Australia
Democritus University of Thrace
Xanthi, Greece Ahmad, Junaid
National University of Sciences and Technology
Acun, Bora Islamabad, Pakistan
UBC Okanagan
Kelowna, BC, Canada Ahmad, Shamsad
King Fahd University of Petroleum and Minerals
Adak, Dibyendu Dhahran, Saudi Arabia
National Institute of Technology Meghalaya
Shillong, Meghalaya, India

REVIEWERS IN 2021
Ahmadi, Jamal Al-Azzawi, Adel
University of Science of Technology Al-Nahrain University
Tehran, Tehran, Islamic Republic of Iran Baghdad, Iraq
Ahmed, Afaf Al-Bahadli, Hussein

Sohar College of Applied Sciences Najaf, Iraq
Sohar, Oman
Al-Bukhaiti, Khalil
Ahmed, Ali Southwest Jiaotong University
Lahore, Punjab, Pakistan Chengdu, China
Ahmed, Mshtaq Al-Hadithi, Abdulkader

King Saud University University of Anbar
Riyadh, Riyadh, Saudi Arabia Ramadi, Al-Anbar, Iraq
Ahmed, Ramadan Al-Hamd, Rwayda Kh. S.

Dow Chemical Canada ULC Baghdad, Iraq
Fort Saskatchewan, AB, Canada
Al-Jaberi, Zuhair
Akakin, Tumer Missouri University of Science and Technology
Aggregate and Ready Mix Association of Minnesota Rolla, MO, United States
Eagan, MN, United States
Al-Kamal, Mustafa
Akalin, Ozlem Al-Nahrain University
PLUSTECHNO Ltd Baghdad, Iraq
İstanbul, Turkey
Al-Khafaji, Ali
Akbari, Reza Missouri University of Science and Technology
University of Tehran Rolla, MO, United States
Tehran, Islamic Republic of Iran
Al-Kheetan, Mazen
Akcaoglu, Tulin Mutah University
Eastern Mediterranean University Karak, Jordan
Magusa, Turkey
Al-Lebban, Yasir
Akhlaghi, Alireza University of Central Florida
Isfahan, Islamic Republic of Iran Orlando, FL, United States
Akiyama, Hiroshi Al-Nasra, Moayyad

The Zenitaka Corporation American University of Ras Al Khaimah School of Engineering
Tokyo, Japan Ras Al Khaimah, United Arab Emirates
Akkaya, Yildir Al-Qaralleh, Mohammad

İstanbul Teknik Üniversitesi Mutah University
Istanbul, Maslak, Sariyer, Turkey Karak, Jordan
Al Mansouri, Omar Al-Sadoon, Zaid

CSTB University of Sharjah College of Engineering
Marne-la-Vallée, France Sharjah, United Arab Emirates
Al Martini, Samer Al-Sodani, Khaled

Abu Dhabi University King Fahd University of Petroleum and Minerals
Abu Dhabi, United Arab Emirates Dhahran, Saudi Arabia
Al-Ani, Moustafa Al-Tuhami, Ahmed Yaseen

Auckland, New Zealand Ain Shams University Faculty of Engineering
Cairo, Egypt
Al-Attar, Tareq
University of Technology
Baghdad, Iraq

REVIEWERS IN 2021
Al-Zubaidi, Aseel Alkerwei, Rana
University of Technology Mustansiriyah University
Baghdad, Iraq Baghdad, Iraq
Al-Zuheriy, Ahmed Alkhairi, Fadi

University of Technology - Iraq Magna MEP
Baghdad, Iraq Dubai, United Arab Emirates
AlRikabi, Fouad Allena, Srinivas

Ohio University Cleveland State University
Athens, OH, United States Cleveland, OH, United States
Alam, A. K. M. Jahangir Allouzi, Rabab

Bangladesh University of Engineering and Technology (BUET) Amman, Jordan
Dhaka, Bangladesh
Almasabha, Ghassan
Alam, Md Shah The Hashemite University
University of Bahrain Sahab, Amman, Jordan
Sakhir, Bahrain
Almbaidheen, Khalil
Aldwaik, Mais Mapei SpA
Russeifa, Zarqa, Jordan Dubai, United Arab Emirates
Alexandre, Pierre Almomani, Mu’taz

Université de Cergy-Pontoise University of Manitoba Faculty of Engineering
Cergy-Pontoise, France Winnipeg, MB, Canada
Alfadhli, Nabilah Almottiri, Falah

Basrah, Iraq College of Technological Studies (PAAET)
Ardiya, Kuwait
Alfeehan, Ashraf
Almustansiriyah University / Faculty of Engineering Almuhsin, Bayrak
Baghdad, Iraq University of Technology
Karrada, Baghdad, Iraq
Alghazali, Hayder
University of Kufa Almusallam, Tarek
Kufa, Iraq King Saud University
Riyadh, Saudi Arabia
Ali, Majid
Capital University of Science and Technology Alnaggar, Mohammed
Islamabad, Pakistan Rensselaer Polytechnic Institute
Troy, NY, United States
Ali, Shehroze
University of Wollongong Alnuaimi, Hamad
Wollongong, NSW, Australia University of Arizona
Tucson, AZ, United States
Aliasghar-Mamaghani, Mojtaba
Virginia Polytechnic Institute and State University Alqam, Maha
Blacksburg, VA, United States University of Jordan
Amman, Jordan
Alimrani, Naser
Budapest University of Technology and Economics Alyousif, Ahmed
Budapest, Hungary Ryerson University Faculty of Engineering and Architectural
Science
Aljazaeri, Zena Toronto, ON, Canada
Al-Nahrain University
Baghdad, Iraq Amani Dashlejeh, Asghar
University of Mohaghegh Ardabili
Ardabil, Islamic Republic of Iran

REVIEWERS IN 2021
Amziane, Sofiane Asaad, Diler
University of Southern Britanny Gaziantep University
Lorient, France Gaziantep, Turkey
Andersson, Ronny Asadian, Alireza

Höllviken, Sweden Concordia University
Montreal, QC, Canada
Andrade, Jairo
Catholic University of Rio Grande do Sul Ashish, Deepankar Kumar
Porto Alegre, RS, Brazil Yonsei University
Seodaemun-gu, Seoul, Republic of Korea
Angst, Ueli
ETH Zürich Ashrafy, Mohammad
Zürich, Switzerland Islamic Azad University – Arak Branch
Kermanshah, Islamic Republic of Iran
Angulski da Luz, Caroline
Universidade Tecnológica Federal do Paraná Aslam, Fahid
Pato Branco, Paraná, Brazil Riyadh, Saudi Arabia
Antoni, Antoni Assaad, Joseph

Petra Christian University University of Balamand
Surabaya, Indonesia Balamand, Lebanon
Antunes, Rodrigo Assi, Lateef

University of Florida Al-Mustaqbal University College
Gainesville, FL, United States Hilla, Babylon, Iraq
Anwari, Qareeb Ullah Attia Ahmed, Mohamed

NUST Institute of Civil Engineering Ain Shams University Faculty of Engineering
Islamabad, Islamabad, Pakistan Cairo, Egypt
Arafa, Ahmed Attia, Walid

Université de Sherbrooke Cairo University
Sherbrooke, QC, Canada Giza, Egypt
Aras, Murat Attiyah, Ali

Bilecik Şeyh Edebali University University of Kufa
Bilecik, Turkey Kufa, Iraq
Araujo, Guilherme Aure, Temesgen

State University of Campinas Faculty of Technology Addis Ababa Science and Technology University
Limeira, Brazil Addis Ababa, Ethiopia
Arfiadi, Yoyong Avci, Eyübhan

Universitas Atma Jaya Yogyakarta Bursa Teknik Üniversitesi
Yogyakarta, DIY, Indonesia Bursa, Yıldırım, Turkey
Aristizabal-Ochoa, Jose Awida, Tarek

National University of Colombia SQC International Consultants
Medellín, Antioquia, Colombia Kuwait, Kuwait
Arito, Philemon Aycardi, Luis

University of Namibia Proyectistas Civiles Asociados
Ongwediva, Oshana, Namibia Miami, FL, United States
Arroudj, Karima Aydin, Ertug

University of Science and Technology Houari Boumediene Lefke Avrupa Üniversitesi
Algiers, Algeria Lefke, Cyprus

REVIEWERS IN 2021
Ayothiraman, Ramanathan Banjara, Nawal
Indian Institute of Technology Delhi CSIR-Structural Engineering Research Centre
New Delhi, India Chennai, Tamil Nadu, India
Azevedo, Antonio Banting, Bennett

Universidade do Porto Faculdade de Engenharia Canada Masonry Design Centre
Porto, Porto, Portugal Mississauga, ON, Canada
Aziz, Ali Bao, Chao

Al-Mustansiriyah University-College of Engineering Ningxia University
Baghdad, Iraq Yinchuan, China
Aziz, Tariq Barman, Manik

NED University of Engineering and Technology University of Minnesota Duluth
Karachi, Sindh, Pakistan Duluth, MN, United States
Azzam, A. Barreto, Matheus

University of Manitoba University of the State of Minas Gerais
Winnipeg, MB, Canada Divinópolis, Minas Gerais, Brazil
B C, Gayana Bartos, Peter

National Institute of Technology Karnataka University of the West of Scotland
Surathkal, Karnataka, India Paisley, United Kingdom of Great Britain and Northern Ireland
B S, Meenakshi Bassim, Raymond

KPRIET University of Illinois Chicago
Coimbatore, India Chicago, IL, United States
B S, Sindu Baty, James

CSIR-Structural Engineering Research Centre Concrete Foundations Association
Chennai, India Mount Vernon, IA, United States
Babafemi, Adewumi Bawono, Ali Aryo

Stellenbosch University TUM Create
Stellenbosch, Western Cape, South Africa Singapore, Singapore
Bahari, Ali Bedirhanoglu, Idris

Department of Physics, University of Mazandaran New York University - Abu Dhabi
Babolsar, Islamic Republic of Iran Diyarbakir, Turkey
Bahekar, Prasad Beglarigale, Ahsanollah

Visvesvaraya National Institute of Technology İstanbul Okan Üniversitesi
Nagpur, Maharashtra, India Istanbul, Turkey
Bahij, Sifatullah Behnoud, Ali

King Fahd University of Petroleum and Minerals Aurecon Australia Group Ltd Sydney
Dhahran, Dammam, Saudi Arabia Neutral Bay, NSW, Australia
Bai, Shaoliang Belagraa, Larbi

Chongqing University Bordj Bou Arréridj, University Center
Chonqqing, China Bordj Bou Arréridj, Est Algeria, Algeria
Balakrishnan, Bijily Belkowitz, Jon

Indian Institute of Technology Tirupati Stevens Institute of Technology
Tirupati, India Freehold, NJ, United States
Balouch, Sana Belleri, Andrea

University of Dundee University of Bergamo
Dundee, United Kingdom of Great Britain and Northern Ireland Dalmine, Italy

REVIEWERS IN 2021
Benliang, Liang Bidkar, Kisan
Shanghai, China Nasik, India
Benmokrane, Brahim Biernacki, Joseph

University of Sherbrooke Tennessee Technological University
Sherbrooke, QC, Canada Cookeville, TN, United States
Bentz, Evan Bilek, Vlastimil

University of Toronto ZPSV a.s.
Toronto, ON, Canada Brno, Czech Republic
Benítez, Pablo Bilir, Turhan

Universidad Nacional de Itapúa Istanbul University-Cerrahpaşa
Encarnación, Itapúa, Paraguay Istanbul, Turkey
Berenguer, Romildo Bilisik, Kadir

Universidade Federal de Pernambuco Centro de Ciências Exatas Erciyes University
e da Natureza Kayseri, Central Anatolia, Turkey
Recife, Brazil
Billah, Abu Hena
Bernard, Erik Stefan Lakehead University
TSE P/L Thunder Bay, ON, Canada
Penrith, NSW, Australia
Binici, Baris
Bernardo, Luis Middle East Technical University
University of Beira Interior Ankara, Turkey
Covilhã, Portugal
Biolzi, Luigi
Bescher, Eric Politecnico di Milano
Los Angeles, CA, United States Milan, Italy
Bettadapura, RaghuPrasad Black, Leon

Indian Institute of Science Sheffield Hallam University
Bangalore, Karnataka, India Sheffield, United Kingdom of Great Britain and Northern Ireland
Beyene, Mengesha Blandón Valencia, John

Turner-Fairbank Highway Research Center National University of Colombia Medellín
McLean, VA, United States Medellín, Antioquia, Colombia
Bharadwaj, Keshav Bonacci, John

Oregon State University Karins Engineering Group
Corvallis, OR, United States Sarasota, FL, United States
Bhattacharjee, Bishwajit Borzovič, Viktor

Indian Institute of Technology Delhi Slovak University of Technology, Faculty of Civil Engineering
New Delhi, Delhi, India Bratislava, Slovakia
Bhogayata, Ankur Boulifa, Ridha

Marwadi Education Foundation’s Group of Institutions Mohamed-Cherif Messaadia University Souk-Ahras
Rajkot, Gujarat, India Souk-Ahras, Algeria
Bhojaraju, Chandrasekhar Bouzid, Haytham

SJEC Université Ibn Khaldoun Tiaret
Mangalore, Karnataka, India Tiaret, Algeria
Bian, R. B. Bradberry, Timothy

Jiangsu Research Institute of Building Science TxDOT Bridge Division
Nanjing, Jiangsu, China Austin, TX, United States

REVIEWERS IN 2021
Braestrup, Mikael Carrette, Jordan
Ramboll Group A/S Crosier Kilgour & Partners Ltd.
København, Denmark Winnipeg, MB, Canada
Brahma, Abdelmalek Cascudo, Oswaldo

Université Saad Dahlab Blida Universidade Federal de Goiás
Blida, Algeria Goiânia, GO, Brazil
Braimah, Abass Castles, Bryan

Carleton University Western Technologies Inc.
Ottawa, ON, Canada Phoenix, AZ, United States
Bujnak, Jan Cattaneo, Sara

Peikko Group Corporation Politecnico di Milano
Kráľová nad Váhom, Slovakia Milan, Italy
Burgoyne, C. Cavunt, Yavuz

University of Cambridge Istanbul Technical University
Cambridge, United Kingdom of Great Britain and Northern Istanbul, Turkey
Ireland
Celik, Kemal
Buttignol, Thomaz Eduardo New York University - Abu Dhabi Campus Engineering Division
Universidade Estadual de Campinas - Campus Cidade Abu Dhabi, United Arab Emirates
Universitaria Zeferino Vaz
Campinas, SP, Brazil Chakkamalayath, Jayasree
Kuwait Institute for Scientific Research
Büyükkaragöz, Alper Safat, Kuwait
Faculty of Technology
Ankara, Turkey Chakraborty, Arun
Bengal Engineering and Science University
C, Vijayaprabha Howrah, West Bengal, India
Karaikudi, Tamil Nadu, India
Chandrasekaran, Vijayvenkatesh
Campos, Claudia Saint Joseph’s College of Engineering and Technology
Universidade Federal Fluminense Thanjavur, Tamil Nadu, India
Rio de Janeiro, Brazil
Chang, Wei
Canbay, Erdem Harbin, China
Middle East Technical University
Ankara, Çankaya, Turkey Chao, Shih-Ho
University of Texas at Arlington
Cannella, Francesco Arlington, TX, United States
Università degli Studi di Palermo
Palermo, Sicilia, Italy Chellappan, C. Selin Ravikumar
Malla Reddy Engineering College
Canpolat, Orhan Secunderabad, Telangana, India
Yildiz Technical University
Istanbul, Esenler, Turkey Chen, Tao
China Academy of Building Research
Cao, Qi Beijing, China
Dalian University of Technology
Dalian, Liaoning, China Chen, Wei
Wuhan University of Technology
Carloni, Christian Wuhan, Hubei, China
Case Western Reserve University
Cleveland, OH, United States Chen, Xia
Changjiang River Scientific Research Institute
Carpio, Víctor Wuhan, China
Universidad Central del Ecuador
Quito, Pichincha, Ecuador

REVIEWERS IN 2021
Chen, Xin Christen, Robert
Shenyang, Liaoning, China Amer Engr Test Inc.
Port Charlotte, FL, United States
Chen, Xu
University of Colorado Boulder Chu, S. H.
Boulder, CO, United States University of Hong Kong
Hong Kong, China
Cheng, Hu
Jiangnan University Chua, Teck
Wuxi, Jiangsu, China Herndon, VA, United States
Cheng, Jianwei Cifuentes, Gustavo

China University of Mining and Technology Universidade do Estado de Santa Catarina
Xuzhou, Jiangsu, China Joinville, Santa Catarina, Brazil
Cheng, Min-Yuan Cladera, Antoni

National Taiwan University of Science and Technology University of the Balearic Islands
Taipei, Taiwan, China Palma, Balearic Islands, Spain
Cherian, Philip Coelho, Jano

Indian Institute of Technology Madras AltoQi Informática
Chennai, Tamil Nadu, India Florianópolis, Santa Catarina, Brazil
Chiaia, Bernardino Cook, Dan

Politecnico di Torino Oklahoma State University–Stillwater
Torino, Piedmont, Italy Stillwater, OK, United States
Chidambaram, R. Siva Cordero Verge, Mariela

CSIR-Central Building Research Institute Mekano4
Roorkee, Uttarakhand, India Sant Joan Despí, Barcelona, Spain
Chitty, Francisco Coronelli, Dario

Florida International University College of Engineering and Politecnico di Milano
Computing Milano, Italy
Miami, FL, United States
Cueto, Jorge
Cho, Soon-Ho Universidad de La Salle
Gwangju University Bogotá, Colombia
Gwangju, Republic of Korea
Cui, Taotao
Choi, Oan Shenyang University of Technology
Soongsil University Shenyang, Liaoning, China
Seoul, Republic of Korea
da Cunha Jr, Byl Farney
Choong, Kokkeong Pontifícia Universidade Católica de Goiás
Universiti Sains Malaysia Goiânia, Goiás, Brazil
Pulau Pinang, Seberang Perai Selatan, Malaysia
Dang, Canh
Chorzepa, Migeum University of Arkansas
Park Ridge, IL, United States Fayetteville, AR, United States
Chougule, IrfanAhamad Dang, Yudong

Sangli, Maharashtra, India Tongji University
Shanghai, China
Chowdhury, Sharmin
Boğaziçi University Darwin, David
Istanbul, Turkey University of Kansas
Lawrence, KS, United States

REVIEWERS IN 2021
Das Adhikary, Satadru Den Uijl, Joop
National University of Singapore Delft University of Technology
Singapore Delft, the Netherlands
Daugevičius, Mykolas Dhole, Rajaram

Vilnius Gediminas Technical University Katy, TX, United States
Vilnius, Lithuania
Dhonde, Hemant
Dave, Shemal University of Houston
Marwadi Education Foundation’s Group of Institutions Houston, TX, United States
Jamnagar, Gujarat, India
Ding, Lujun
Davila, Jose Chengdu, China
University of Huelva, Escuela Técnica Superior de Ingeniería
Palos de la Frontera, Huelva, Spain Diniz, Sofia Maria
Universidade Federal de Minas Gerais
DC, Mitra Belo Horizonte, Brazil
College of Engineering Trivandrum
Thiruvananthapuram, Kerala, India Dogan, Unal
Istanbul Technical University
de Korte, Ariën Istanbul, Turkey
University of Twente
Enschede, the Netherlands Dolan, Charles
University of Wyoming
de Oliveira, Gabriel Laramie, WY, United States
UNICAMP
Campinas, São Paulo, Brazil Dong, Jiangfeng
Sichuan University
DeFord, Harvey Chengdu, China
Florida Department of Transportation
Gainesville, FL, United States Dongell, Jonathan
Pebble Technologies
Deb, Arghya Scottsdale, AZ, United States
Indian Institute of Technology, Kharagpur
Kharagpur, West Bengal, India Dontchev, Dimitar
University of Chemical Technology and Metallurgy
Debbarma, Solomon Sofia, Bulgaria
Texas State University
San Marcos, TX, United States Drangu, Sehu
Basel, Switzerland
Degtyarev, Vitaliy
Home Du, Lianxiang
Lexington, SC, United States University of Alabama at Birmingham
Birmingham, AL, United States
del Rey Castillo, Enrique
The University of Auckland Du, Yingang
Auckland, New Zealand Anglia Ruskin University
Chelmsford, United Kingdom of Great Britain and Northern
Delalibera, Rodrigo Ireland
University of São Paulo
São Carlos, São Paulo, Brazil Dufalla, Nicole
NCE
Delatte, Norbert Reno, NV, United States
Oklahoma State University
Stillwater, OK, United States Dundar, Cengiz
Çukurova University
Demis, Sotiris Adana, Turkey
University of Patras
Patras, Greece

REVIEWERS IN 2021
Dutta, Debabrata Elfgren, Lennart
Camellia Group Education Luleå University of Technology
Kolkata, West Bengal, India Luleå, Sweden
Duvallet, Tristana Eljadei, Abdelatee

University of Kentucky Personal
Lexington, KY, United States Pittsburgh, PA, United States
Dwairi, Hazim Elmasry, Mohamed

The Hashemite University Arab Academy for Science, Technology and Maritime Transport
Zarqa, Jordan Alexandria, Egypt
Eid, Fatma Eltahawy, Reham

Menoufia University Ain Shams University
Shebin El-Kom, Egypt Cairo, Egypt
Eid, Rami Emamy Farvashany, Firooz

SCE - Shamoon College of Engineering Perthpolis Pty Ltd
Be’er Sheva, Israel Perth, WA, Australia
El Maghraby, Yosra Eskandari, Hamid

The British University in Egypt Hakim Sabzevari University
Cairo, Egypt Sabzevar, Islamic Republic of Iran
El Shafey, Asmaa Eskandari, Rasoul

King Khalid University-Sarat Ebida Campus Kashmar, Razavi Khorasan, Islamic Republic of Iran
Sarat Abidah, Abha, Saudi Arabia
Esmaili, Omid
Elamin, Anwar University of California, Irvine
University of Nyala Irvine, CA, United States
Nyala, Sudan
Etman, Emad
Elaqra, Hossam El-Mahalla El-Kobra, Egypt
Gaza, State of Palestine
Evangelista, Ana
El-Dash, Karim Engineering Institute of Technology
College of Technological Studies Perth, WA, Australia
Kuwait, Kuwait
Ewis, Ahmed
El-Gendy, Mohammed Cairo, Egypt
Tetra Tech Canada
Winnipeg, MB, Canada Ezeldin, Samer
The American University in Cairo
El-Hassan, Hilal Cairo, Egypt
United Arab Emirates University Faculty of Engineering
Al Ain, United Arab Emirates Fadaee, Mostafa
Concordia University Department of Building, Civil and
El-Salakawy, Ehab Environmental Engineering
University of Manitoba Montreal, QC, Canada
Winnipeg, MB, Canada
Fahim, Andrew
El-Sayed, Ahmed Giatec Scientific Inc.
University of Sherbrooke Ottawa, ON, Canada
Sherbrooke, QC, Canada
Faleschini, Flora
Eldarwish, Aly University of Padova
Alexandria, Egypt Padova, Italy

REVIEWERS IN 2021
Fam, Amir Feys, Dimitri
Queen’s University Missouri University of Science and Technology
Kingston, ON, Canada Rolla, MO, United States
Fantilli, Alessandro Pasquale Fikry, Ahmed

Politecnico di Torino Cairo, Egypt
Torino, Italy
Fiset, Mathieu
Faria, Duarte Université du Québec à Chicoutimi
Faculdade de Ciências e Tecnologia Chicoutimi, QC, Canada
Caparica-Lisbon, Portugal
Fitik, Birol
Farshadfar, Omid Hochschule für Technik Stuttgart
Thornton Tomasetti Stuttgart, Germany
Lawrence, KS, United States
Folino, Paula
Farzadnia, Nima University of Buenos Aires
Missouri University of Science and Technology Buenos Aires, Argentina
Rolla, MO, United States
Foraboschi, Paolo
Farzam, Masood Università Iuav di Venezia-Dipartimento di Architettura
Structural Engineering Costruzione Conservazione
Tabriz, Islamic Republic of Iran Venice, Italy
Fatemi, Hassan Francüois, Buyle-Bodin

Montreal, QC, Canada University of Lille
Villeneuve-d’Ascq, France
Fattuhi, Nijad
Birmingham, United Kingdom of Great Britain and Northern French, Catherine
Ireland University of Minnesota
Minneapolis, MN, United States
Feldman, Steven
National Institute of Standards and Technology Fu, Chung
Gaithersburg, MD, United States Rockville, MD, United States
Feng, De-Cheng Fuentes, Jose Maria

Southeast University Polytechnic University of Madrid
Nanjing, China Madrid, Madrid, Spain
Feng, Muzai G D R N, Ransinchung

Walter P Moore Roorkee, Uttarakhand, India
Dallas, TX, United States
G., Dhinakaran
Fernandes, Fernando SASTRA University
ITPAC Porto Nacional Thanjavur, India
Porto Nacional, Tocantins, Brazil
G.S., Rampradheep
Fernandez Montes, David Kongu Engineering College
Madrid, Madrid, Spain Erode, Tamil Nadu, India
Fernández Ruiz, Miguel Gabrijel, Ivan

École Polytechnique Fédérale de Lausanne University of Zagreb, Faculty of Civil Engineering
Lausanne, Vaud, Switzerland Zagreb, Croatia
Ferron, Raissa Gadkar, Shubhada

The University of Texas at Austin Ontario, CA, United States
Austin, TX, United States
Gamble, William
University of Illinois at Urbana-Champaign
Urbana, IL, United States

REVIEWERS IN 2021
Gan, Dan Girardi, Ricardo
Chongqing, China Pontifícia Universidade Católica do Rio Grande do Sul
Porto Alegre, Brazil
Ganesan, N.
National Institute of Technology Girgin, Canan
Calicut, India Yıldız Technical University
Istanbul, Turkey
GangaRao, Hota
West Virginia University Goksu, Caglar
Morgantown, WV, United States Istanbul Technical University
Istanbul, Turkey
Gao, Peng
Hefei University of Technology Gong, Bill
Hefei, China Markham, ON, Canada
Gao, Xiangling González, Javier

Tongji University University of Basque Country
Shanghai, China Bilbao, Basque Country, Spain
Garcia-Taengua, Emilio González, María

University of Leeds Polytechnic University of Madrid
Leeds, United Kingdom of Great Britain and Northern Ireland Madrid, Spain
Gayed, Ramez González-Fonteboa, Belén

University of Calgary Universidade da Coruna Escuela Técnica Superior de Ingenieros
Calgary, AB, Canada de Caminos, Canales y Puertos
A Coruña, Spain
Geng, Yan
MCC Central Research Institute of Building and Construction Govan, G. Elangovan
Co. Ltd. Civil Engg
Beijing, China Chennai, Tamil Nadu, India
Ghali, Amin Guades, Ernesto

University of Calgary Technical University of Denmark
Calgary, AB, Canada Lyngby, Denmark
Ghiami Azad, Amir Reza Guan, Bowen

University of Tehran Xi’an, China
Boston, MA, United States
Guan, Garfield
Ghimire, Krishna University of Cambridge
Kansas State University Cambridge, United Kingdom of Great Britain and Northern
Manhattan, KS, United States Ireland
Ghoddousi, Parviz Guan, Junfeng

Iran University of Science and Technology North China University of Water Resources and Electric Power
Tehran, Islamic Republic of Iran Zhengzhou, Hehan, China
Gholamhoseini, Ali Guerrini, Gabriele

Auckland, Auckland, New Zealand City and County of San Francisco
San Francisco, CA, United States
Giaccio, Craig
AECOM Gunasekara, Chamila
Melbourne, VIC, Australia RMIT University City Campus
Melbourne, VIC, Australia
Gilbert, Raymond
University of New South Wales Guner, Serhan
Sydney, NSW, Australia University of Toledo
Toledo, OH, United States

REVIEWERS IN 2021
Guo, Honglei Han, Baoguo
Wuhan Polytechnic University Dalian University of Technology
Wuhan, Hubei, China Dalian, China
Guo, Liping Haneefa Kolakkadan, Mohammed

Southeast University SSN College of Engineering
Nanjing, Jiangsu Province, China Kalavakkam, Tamil Nadu, India
Guo, Zixiong Hanif, Asad

Huaqiao University Hong Kong University of Science and Technology
Quanzhou, Fujian, China Kowloon, Hong Kong, China
Guzmán, Andrés Hariri-Ardebili, Mohammad Amin

Universidad del Norte University of Colorado Boulder
Barranquilla, Atlántico, Colombia Boulder, CO, United States
Guzmán-Torres, José Harith, Iman

UMSNH Babylon, Iraq
Morelia, Michoacán, Mexico
Hasan, Noor
Gökçe, H. Süleyman The University of Adelaide Faculty of Engineering, Computer
Bayburt Üniversitesi and Mathematical Sciences
Bayburt, Turkey Adelaide, SA, Australia
Gör, Mesut Hashemi, Shervin

Firat University Engineering Faculty Seoul National University
Elazığ, Turkey Seoul, Republic of Korea
Habulat, Afifudin Hashim Abbas, Zainab

Universiti Teknologi MARA Babylon, Iraq
Permatang Pauh, Penang, Malaysia
Hasnat, Ariful
Hadad, Houman University of Asia Pacific
DeSimone Consulting Engineers Dhaka, Bangladesh
Miami, FL, United States
Hassan, Maan
Haddad Kolour, Hosain University of Technology
University of Maine Baghdad, Iraq
Orono, ME, United States
Hassan, Mohamed
Hadje-Ghaffari, Hossain University of Sherbrooke
John A. Martin & Associates, Inc. Sherbrooke, QC, Canada
Los Angeles, CA, United States
Hassan, Mostafa
Haggag, Hesham University of Connecticut
Haggag Consultancy for Construction Storrs, CT, United States
Cairo, Egypt
Hassanein, Ahmed
Hajibabaee, Amir Université de Sherbrooke
Chicago, IL, United States Sherbrooke, QC, Canada
Hamad, Ali Hassani Esgandani, Mohammadreza

Engineering Technical College of Mosul University of Technology Sydney
Mosul, Nineveh, Iraq Ultimo, NSW, Australia
Hamed, Ehab Hassani, Arash

University of New South Wales University of Arizona
Sydney, NSW, Australia Tucson, AZ, United States

REVIEWERS IN 2021
Hassanpour, Sina Hoult, Neil
Sharif University of Technology Toronto, ON, Canada
Hu, Bo
Hawileh, Rami University of Arizona
American University of Sharjah Tucson, AZ, United States
Sharjah, United Arab Emirates
Hu, Jiong
Hedegaard, Brock University of Nebraska-Lincoln
University of Minnesota Duluth Omaha, NE, United States
Duluth, MN, United States
Hu, Nan
Helmy, Huda South China University of Technology
Applied Science International Guangzhou, Guangdong, China
Durham, NC, United States
Huang, Bo-Tao
Henigal, Ashraf The Hong Kong Polytechnic University
Suez University Kowloon, Hong Kong, China
Suez, Egypt
Huang, Chang-Wei
Hernández-Montes, Enrique Chung Yuan Christian University
University of Granada Chung-li, Taiwan, China
Granada, Spain
Huang, Chung-Ho
Hijazi, Moheldeen National Taipei University of Technology
Istanbul Technical University Taipei, Taiwan, China
Istanbul, Turkey
Huang, Jianwei
Ho, Johnny Southern Illinois University Edwardsville
The University of Hong Kong Edwardsville, IL, United States
Hong Kong, Hong Kong, China
Huang, Wu-Jang
Hojati, Maryam Pingtung, Taiwan, China
University of New Mexico
Albuquerque, NM, United States Huo, Jingsi
Huaqiao University
Holly, Ivan Xiamen, Fujian, China
Slovak University of Technology in Bratislava Faculty of Civil
Engineering Hussain, Raja
Bratislava, Slovakia King Saud University
Horowitz, Bernardo
Recife, PE, Brazil Hussain, Safdar
MS Ramaiah University of Applied Sciences
Hossain, Khandaker Bangalore, Karnataka, India
Ryerson University
Toronto, ON, Canada Huts, Andriy
Politechnika Rzeszowska im. Ignacego Łukasiewicza
Hossain, Tanvir Rzeszów, Poland
Louisiana State University
Houston, TX, United States Hwang, Hyeon Jong
Seoul National University
Hosseini, Seyed Mohammad Seoul, Republic of Korea
University of Sherbrooke
Sherbrooke, QC, Canada Hwang, Shyh-Jiann
National Taiwan University
Hosseinpoor, Masoud Taipei, Taiwan, China
Université de Sherbrooke Faculté de génie

REVIEWERS IN 2021
Ibarra, Luis Jacintho, Ana
University of Utah Pontifical Catholic University of Campinas
Salt Lake City, UT, United States Campinas, São Paulo, Brazil
Ibrahim, Ayoob Jackson, Marie

University of Wollongong University of Utah
Wollongong, NSW, Australia Salt Lake City, UT, United States
Ibrahim, Izni Jahami, Ali

Universiti Teknologi Malaysia Beirut Arab University
Johor Bahru, Johor, Malaysia Beirut, Lebanon
Ibrahim, Taha Jang, Seung Yup

Faculty of Engineering, Benha University Korea National University of Transportation
Cairo, Egypt Uiwang, Gyeonggi-do, Republic of Korea
Ibrahim, Tamer Jansen, Daniel

Menofia, Egypt California Polytechnic State University
San Luis Obispo, CA, United States
Idrees, Maria
University of Engineering and Technology Lahore Janssen, Donald
Lahore, Punjab, Pakistan Seattle, WA, United States
Ilki, Alper Jarallah, Husain

Istanbul Technical University Al-Mustansiriyah University
Istanbul, Turkey Baghdad, Iraq
Ioannou, Socrates Jazaei, Robabeh

Military Technological College University of Wisconsin–Platteville
Muscat, Oman Platteville, WI, United States
Irassar, Edgardo Jiang, Hua

Dep. Ingeniería Civil - UNCPBA University of Georgia
Olavarria, Buenos Aires, Argentina Athens, GA, United States
Islam, Mohammad Momeen Ul Jiang, Liying

The University of Adelaide - North Terrace Campus Jensen Hughes Inc
Adelaide, SA, Australia Wakefield, MA, United States
Ismail, Kamaran Jing, Denghu

University of Sheffield Southeast University
Sheffield, United Kingdom of Great Britain and Northern Ireland Nanjing, Jiangsu Province, China
Ismail, Mohamed Jing, Gouqing

Memorial University of Newfoundland Beijing Jiaotong University
St. John’s, NL, Canada Beijing, China
Isufi, Brisid Jones, Scott

Universidade Nova de Lisboa Faculdade de Ciências e National Institute of Standards and Technology
Tecnologia Gaithersburg, MD, United States
Caparica, Portugal
Joshi, Buddhi
Iwaki, Ichiro Pokhara University
Nihon University Lekhnath, Nepal
Kōriyama, Japan
Jozić, Dražan
Izzet, Amer Faculty of Chemical Technology
University of Baghdad Split, Croatia
Baghdad, Iraq

REVIEWERS IN 2021
Juvandes, Luis Karakale, Vail
Fac de Eng da Univ do Porto - FEUP Marmara Üniversitesi
Porto, Portugal Istanbul, Turkey
Jędrzejewska, Agnieszka Karayannis, Christos

Silesian University of Technology Democritus University of Thrace
Gliwice, Poland Xanthi, Greece
K., Balaji Rao Kareem, Aseel

CSIR-Structural Engineering Research Centre University of Baghdad Al-Jaderyia Campus College of Science
Chennai, India Baghdad, Iraq
Kabir, Mohammad Z. Kawamura, Mitsunori

Amirkabir University of Technology Kanazawa, Ishikawa, Japan
Kawashima, Shiho
Kadima Lukanu Lwa Nzambi, Aaron Columbia University
Universidade Federal do Pará New York, NY, United States
Belém, Brazil
Kaya, Osman
Kaliyavaradhan, Senthil Muğla Sıtkı Koçman Üniversitesi
Hunan University Muğla, Turkey
Changsha, Hunan, China
Kazemi, Sadegh
Kalkan, Ilker University of Alberta
Kırıkkale University Edmonton, AB, Canada
Kırıkkale, Turkey
Kellouche, Yasmina
Kamble, Rohit Université Hassiba Benbouali de Chlef
Sanjay Ghodawat University Faculty of Engineering Chlef, Algeria
Kolhapur, India
Keskin, Süleyman
Kamde, Deepak Muğla Sıtkı Koçman Üniversitesi
Indian Institute of Technology Madras Muğla, Turkey
Chennai, TN, India
Kevern, John
Kanagaraj, Ramadevi University of Missouri–Kansas City
Kumaraguru College of Technology Kansas City, MO, United States
Coimbatore, Tamil Nadu, India
Khabaz, Amjad
Kandasami, Siva Hasan Kalyoncu University
Bristol, United Kingdom of Great Britain and Northern Ireland Gaziantep, Turkey
Kanemoto, Kiyo-omi Khalifa, Essam

Shimizu Corporation Higher Technological Institute
Tokyo, Japan Cairo, Egypt
Kang, Shao-Bo Khaliq, Wasim

Chongqing University National University of Sciences and Technology
Chongqing, China Islamabad, ICT, Pakistan
Kankam, Charles Khan, Sadaqat Ullah

Kwame Nkrumah University of Science & Technology NED University of Engineering and Technology
Kumasi, Ghana Karachi, Sindh, Pakistan
Kanta Rao, Velidandi Khatib, Jamal

Central Road Research Institute University of Wolverhampton
New Delhi, Delhi, India Wolverhampton, United Kingdom of Great Britain and Northern
Ireland

REVIEWERS IN 2021
Khayat, Kamal Kusbiantoro, Andri
Missouri University of Science and Technology Universiti Malaysia Pahang
Rolla, MO, United States Gambang, Pahang, Malaysia
Kim, Jae Hong Kwan, Albert

Korea Advanced Institute of Science and Technology The University of Hong Kong
Daejeon, Republic of Korea Hong Kong, China
Kim, Jang Hoon Kwon, Sukmin

Ajou University Tohoku University
Suwon, Republic of Korea Sendai, Miyagi, Japan
Kim, Taehwan Larbi, Kacimi

University of New South Wales University of Sciences and Technology of Oran
Sydney, NSW, Australia Oran, Algeria
Kirane, Kedar Liu, Chengqing

Stony Brook University Southwest Jiaotong University
Stony Brook, NY, United States Chengdu, Sichuan, China
Klemczak, Barbara Lai, Jianzhong

Silesian University of Technology Nanjing University of Science and Technology
Gliwice, Poland Nanjing, Jiangsu, China
Koehler, Eric Lakhani, Hitesh

Titan America Institute of Construction Materials
Norfolk, VA, United States Stuttgart, Germany
Kopiika, Nadiia Lange, David

Lviv Polytechnic National University University of Illinois at Urbana-Champaign
Lviv, Ukraine Urbana, IL, United States
Kotsovou, Gregoria Lapi, Massimo

National Technical University of Athens Università degli Studi di Firenze
Athens, Greece Firenze, Firenze, Italy
Kreiger, Eric Lawler, John

U.S. Army Corps of Engineers Construction Engineering Wiss, Janney, Elstner Associates, Inc.
Research Laboratory Northbrook, IL, United States
Champaign, IL, United States
Lazic, Zarko
Krem, Slamah University of Belgrade Faculty of Civil Engineering
University of Waterloo Belgrade, Serbia
Waterloo, ON, Canada
Lederle, Rita
Krstulovic-Opara, Neven University of St. Thomas
ExxonMobil Production Company Saint Paul, MN, United States
Houston, TX, United States
Lee, Chadon
Kumar, Rakesh Chung-Ang University
Central Road Research Institute Anseong, Kyungki-do, Republic of Korea
Delhi, India
Lee, Chang Hoon
Kunieda, Minoru Cornell University
Nagoya University Ithaca, NY, United States
Nagoya, Japan
Lee, Daniel
Kunyong, Zhang The Royal Danish Academy of Fine Arts, School of Architecture
Hohai University Copenhagen, Denmark
Nanjing, China

REVIEWERS IN 2021
Lee, Deuckhang Libre, Nicolas Ali
Chungbuk National University Missouri University of Science and Technology
Cheongju, Republic of Korea Rolla, MO, United States
Lee, Foo Wei Lignola, Gian Piero

Universiti Tunku Abdul Rahman Università Degli Studi di Napoli Federico II Scuola Politecnica
Kajang, Selangor, Malaysia e delle Scienze di Base
Napoli, Italy
Lee, Yoon-Si
Bradley University Lim, Malcolm
Peoria, IL, United States MLim Consulting Inc.
Buffalo Grove, IL, United States
Lee, Young Hak
Seoul, Republic of Korea Lima, Nathalia
Universidade Federal de Pernambuco Centro de Ciências Exatas
Leech, Craig e da Natureza
Arup Recife, PE, Brazil
Sydney, NSW, Australia
Lin, Wei-Ting
Leiva Fernández, Carlos Department of Civil Engineering
University of Seville Ilan, Taiwan, China
Seville, Andalucia, Spain
Liu, Kai-Wei (Victor)
Leone, Marianovella Texas A&M University Transportation Institute
University of Salento College Station, TX, United States
Lecce, Italy
Liu, Leping
Lequesne, Remy Nanning, China
University of Kansas
Lawrence, KS, United States Liu, Peng
Changsha, China
Leung, Christopher
Hong Kong University of Science and Technology Liu, Yanbo
Kowloon, Hong Kong, China Harbin Engineering University
Harbin, Heilongjiang, China
Li, Bo
Hong Kong Polytechnic University Liu, Yanhua
Kowloon, Hong Kong, China Yangtze University
Jingzhou, China
Li, Fumin
China University of Mining and Technology Liu, Yin
Xuzhou, Jiangsu, China Shandong University of Science and Technology
Qingdao, China
Li, Genfeng
Chongqing University of Arts and Sciences Lomboy, Gilson
Chongqing, China Rowan University
Glassboro, NJ, United States
Li, Zhengqi
Houston, TX, United States Loo, Yew-Chaye
Gold Coast, QLD, Australia
Li, Zhuguo
Yamaguchi University Looi, Daniel
Ube, Yamaguchi, Japan Swinburne University of Technology - Sarawak Campus
Kuching, Malaysia
Liang, Ninghui
Chongqing University Lopes, Anne
Chongqing, China Furnas Centrais Elétricas SA
Aparecida de Goiânia, Goiás, Brazil

REVIEWERS IN 2021
Lopes, Sergio Mari, Antonio
University of Coimbra Universitat Politécnica de Catalunya
Coimbra, Portugal Barcelona, Spain
Lotfy, Abdurrahmaan Markovic, Ivan

Lafarge Canada Inc. dsp Ingenieure & Planer AG
Toronto, ON, Canada Greifensee, Switzerland
Luo, Baifu Marques, Marília

Harbin, China Rio Paranaíba, Minas Gerais, Brazil
López-Almansa, Francisco Martin-Perez, Beatriz

Polytechnic University of Catalonia University of Ottawa
Barcelona, Spain Ottawa, ON, Canada
Ma, Zhongguo Martinez Andino, Marcos

University of Tennessee Stanley D. Lindsey and Associates
Knoxville, TN, United States Atlanta, GA, United States
Mabrouk, Rasha Maruyama, Ippei

Giza, Egypt Nagoya University
Nagoya, Aichi, Japan
MacDougall, Colin
Queen’s University Matta, Fabio
Kingston, ON, Canada University of South Carolina
Columbia, SC, United States
Macht, Jürgen
Kirchdorf, Austria Maximos, Hany
Maximos Engineering
Madani, Hesam Mississauga, ON, Canada
Kerman Graduate University of Technology
Kerman, Islamic Republic of Iran McCall, W.
Concrete Engineering Consultants
Mahajan, Milinda Charlotte, NC, United States
Vishwakarma Institute of Information Technology
Pune, Maharashtra, India Mechtcherine, Viktor
TU Dresden
Maheswaran, Chellapandian Dresden, Germany
Mepco Schlenk Engineering College
Sivakasi, Tamil Nadu, India Medeiros-Junior, Ronaldo
Universidade Federal do Paraná
Mahfouz, Ibrahim Curitiba, Paraná, Brazil
Cairo, Egypt
Megid, Wael
Mahinroosta, Mostafa TISEC Inc.
Tehran, Islamic Republic of Iran Morin-Heights, QC, Canada
Mander, John Mehanny, Sameh

Texas A&M University Cairo University
College Station, TX, United States Cairo, Egypt
Mardani-Aghabaglou, Ali Mehany, Shehab

Bursa Uludağ University University of Sherbrooke
Bursa, Turkey Sherbrooke, QC, Canada
Mardi, Sahand Meinheit, Donald

QIAU Wiss, Janney, Elstner Associates, Inc.
Qazvin, Islamic Republic of Iran Chicago, IL, United States

REVIEWERS IN 2021
Mejia, Luis Gonzalo Mohammed, Tarek
LGM & Cia Islamic University of Technology
Medellín, Colombia Gazipur, Dhaka, Bangladesh
Melo, José Moini, Mohamadreza

Faculty of Engineering of the University of Porto Princeton University
Porto, Portugal Princeton, NJ, United States
Meng, Chaomei Moradi, Hiresh

Xi’an, China AmirKabir University of Technology
Meng, Tao
Institution of Building Materials Moraes Neto, Bernardo
Hangzhou, Zhejiang, China Universidade Federal do Pará
Belém, Brazil
Meng, Weina
Missouri University of Science and Technology Morgese, Maurizio
Rolla, MO, United States University of Illinois Chicago
Chicago, IL, United States
Menkulasi, Fatmir
Wayne State University Mosavi, Hossein
Detroit, MI, United States University of Florida
Gainesville, FL, United States
Menu, Bruce
Université Laval Faculté des Sciences et de Génie Moser, Robert
Quebec, QC, Canada U.S. Army Engineer Research and Development Center
Vicksburg, MS, United States
Michael, Antonis
University of Florida Mostafa, Mostafa M. A.
Gainesville, FL, United States Al-Azhar University - Assiut Branch
Qena, Egypt
Mirrashid, Masoomeh
Semnan University Mostofinejad, Davood
Semnan, Islamic Republic of Iran Isfahan University of Technology
Isfahan, Islamic Republic of Iran
Mishra, Dhanada
KMBB College of Engineering and Technology Motaref, Sarira
Khordha, Odisha, India University of Connecticut
Storrs, CT, United States
Mock, Andrew
HP Engineering, Inc. Muciaccia, Giovanni
Oklahoma City, OK, United States Politecnico di Milano
Milan, Italy
Mogili, Srinivas
The Hong Kong University of Science and Technology Munoz, Jose
Hong Kong, China Federal Highway Administration
McLean, VA, United States
Mohamed, Ashraf
Alexandria University Nabavi, Seyed Esrafil
Alexandria, Egypt Rezvanshahr, Islamic Republic of Iran
Mohamed, Nayera Naganathan, Sivakumar

Assiut University Universiti Tenaga Nasional
Assiut, Egypt Kajang, Selangor, Malaysia
Mohammed, Ahmed Nair, Sriramya

Sulaimani, United States Ithaca, NY, United States

REVIEWERS IN 2021
Najimi, Meysam Nili, Mahmoud
Iowa State University Bu-Ali Sina University
Ames, IA, United States Hamedan, Islamic Republic of Iran
Nakarai, Kenichiro Nilsson, Lars-Olof

Hiroshima Daigaku Lund University
Higashi-Hiroshima, Japan Lund, Sweden
Nam, Jeongsoo Nishiyama, Minehiro

Chungnam National University Kyoto University
Daejeon, Republic of Korea Kyoto, Japan
Narayanan, Subramanian Nkinamubanzi, Pierre-Claver

Gaithersburg, MD, United States Institute for Research in Construction
Ottawa, ON, Canada
Naser, M. Z.
Clemson University Noor-E-Khuda, Sarkar
Clemson, SC, United States Central Queensland University
Perth, WA, Australia
Nassiri, Somayeh
Washington State University Noshiravani, Talayeh
Pullman, WA, United States EPFL
Lausanne, Switzerland
Naxine, Dhiraj
GH Raisoni University Olek, Jan
Amravati, Maharashtra, India Purdue University
West Lafayette, IN, United States
Nayel, Ashraf
Housing and Building National Research Center Omar, Mohamed
Dokki, Giza, Egypt Dubai, United Arab Emirates
Nazarimofrad, Ebrahim Ombres, Luciano

Bu-Ali Sina University University of Calabria
Hamedan, Islamic Republic of Iran Cosenza, Italy
Neithalath, Narayanan Omer, Alkailani

Clarkson University Dalhousie University
Potsdam, NY, United States Halifax, NS, Canada
Neuenschwander, Martin Orakcal, Kutay

University of California, Berkeley Boğaziçi University
Berkeley, CA, United States Istanbul, Turkey
Neves, Luís Orta, Luis

University of Coimbra ITESM
Coimbra, Portugal Zapopan, Jalisco, Mexico
Ng, P. L. Othman, Nor Hazurina

The University of Hong Kong Universiti Tun Hussein Onn Malaysia
Hong Kong, China Batu Pahat, Johor, Malaysia
Nguyen, Tan Oukaili, Nazar

Ton Duc Thang University University of Baghdad
Ho Chi Minh City, Viet Nam Baghdad, Iraq
Nguyen, Van-Luat Oyguc, Resat

Hanoi, Viet Nam İstanbul Teknik Üniversitesi
Istanbul, Turkey

REVIEWERS IN 2021
Ozturan, Turan Perez Caldentey, Alejandro
Boğaziçi University Universidad Politécnica de Madrid
Istanbul, Turkey Madrid, Spain
Özkal, Fatih Mehmet Perrot, Arnaud

Atatürk Üniversitesi Lorient, Bretagne, France
Erzurum, Turkey
Persson, Bertil
Pacheco, Alexandre Bara, Sweden
Universidade Federal do Rio Grande do Sul (UFRGS)
Porto Alegre, RS, Brazil Piccinin, Roberto
Hilti, Inc.
Paiva, Maria Tulsa, OK, United States
UFRJ
Rio de Janeiro, RJ, Brazil Pierce, James
Littleton, CO, United States
Palieraki, Vasiliki
National Technical University of Athens Pileggi, Rafael
Athens, Zografou, Greece University of São Paulo
São Paulo, Brazil
Palmisano, Fabrizio
Politecnico di Bari Pisani, Marco
Bari, Italy Politecnico di Milano
Milan, Italy
Pan, Wang Fook
SEGi University Pocesta, Ylli
Petaling Jaya, Selangor, Malaysia Debar, North Macedonia
Panda, Sushree Pourbaba, Masoud

KIIT University Islamic Azad University, Maragheh
Bhubaneswar, India Maragheh, East Azerbaijan, Islamic Republic of Iran
Pandit, Poornachandra Prakash, M N

Manipal Institute of Technology J.N.N. College of Engineering
Manipal, India Shimoga, Karnataka, India
Pantazopoulou, Stavroula Prieto, Francisco

York University Autonomous University of the State of Hidalgo
Toronto, ON, Canada Mineral de la Reforma, Hidalgo, Mexico
Parghi, Anant Putra Jaya, Ramadhansyah

University of British Columbia Universiti Teknologi Malaysia
Kelowna, BC, Canada Gambang, Pahang, Malaysia
Parsekian, Guilherme PV, Premalatha

UFSCar CARE School of Engineering
São Carlos, Brazil Tiruchirappalli, India
Pastor, José Qasrawi, Hisham

University of Alicante The Hashemite University
Alicante, Spain Zarqa, Jordan
Paulino, Glaucio Qi, Chengqing

Georgia Institute of Technology CEMEX
Atlanta, GA, United States Riverview, FL, United States
Perceka, Wisena Qian, Ye

Parahyangan Catholic University Faculty of Engineering The University of Hong Kong
Bandung, West Java, Indonesia Hong Kong, China

REVIEWERS IN 2021
Qiao, Chunyu Rashad, Alaa
Wiss, Janney, Elstner Associates, Inc. Housing and Building National Research Center
Austin, TX, United States Giza, Egypt
Qissab, Musab Rashed, Youssef F.

Al-Nahrain University Faculty of Engineering
Baghdad, Iraq Giza, Egypt
Quintana Gallo, Patricio Rasulo, Alessandro

Universidad de Valparaíso University of Cassino and Southern Lazio
Valparaíso, Chile Cassino, Italy
Quiroga, Pedro Reda Taha, Mahmoud

Escuela Colombiana de Ingeniería University of New Mexico
Bogotá, Colombia Albuquerque, NM, United States
Rafi, Muhammad Reknes, Kare

NED University of Engineering and Technology Sika Norge AS
Karachi, Sindh, Pakistan Skytta, Norway
Raghavendra, T. Rezaifar, Omid

R.V. College of Engineering, Visvesvaraya Technological Semnan, Islamic Republic of Iran
University
Bangalore, Karnataka, India Ricker, Marcus
Hochschule Biberach
Ragheb, Wael Biberach, Baden-Württemberg, Germany
Alexandria University
Windsor, ON, Canada Ridha, Maha (Lianne)
Melbourne, VIC, Australia
Rahal, Khaldoun
Kuwait University Riedel, Philipp
Safat, Kuwait Universität Siegen Naturwissenschaftlich-Technische Fakultät
Siegen, Germany
Raj, Bharati
NSS College of Engineering Rizwan, Syed Ali
Palakkad, India Superior University Lahore
Lahore, Pakistan
Rajbanshi, Soumi
Indian Institute of Technology Guwahati Rodhia, Rohit
Guwahati, Assam, India Varanasi, Uttar Pradesh, India
Ram, V. Rodríguez, Ángel

BITS Pilani Hyderabad Campus Polytechnic University of Burgos
Hyderabad, Andhra Pradesh, India Burgos, Spain
Ramaswamy, Ananth Rojas Aguero, Rosangel

Indian Institute of Science Universidade Federal do Rio Grande
Bangalore, Karnataka, India Rio Grande, Rio Grande do Sul, Brazil
Ramirez, Julio Rombach, Guenter

Purdue University Hamburg University of Technology
West Lafayette, IN, United States Hamburg, Germany
Ramirez-Garcia, Alberto Romero, Manuel

University of Arkansas Universitat Politécnica de Valéncia
Fayetteville, AR, United States Valencia, Spain
Ramos, António Ross, Brandon

Faculdade de Ciências e Tecnologia Clemson University
Monte de Caparica, Portugal Clemson, SC, United States

REVIEWERS IN 2021
Rughooputh, Reshma Saqan, Elias
University of Mauritius American University in Dubai
Moka, Mauritius Dubai, United Arab Emirates
Rupnow, Tyson Sargam, Yogiraj

Louisiana Transportation Research Center Iowa State University
Baton Rouge, LA, United States Ames, IA, United States
Russell, Bruce Sartaji, Parisa

Oklahoma State University Ardabil, Islamic Republic of Iran
Stillwater, OK, United States
Sato, Yuichi
S, Kumaravel Kyoto University
Annamalai University Kyoto, Japan
Cuddalore, Tamil Nadu, India
Sayed Ahmed, Mahmoud
Saeed, Yasir Ryerson University
Portland State University Toronto, ON, Canada
Portland, OR, United States
Sayyar Roudsari, Sajjad
Safan, Mohamed North Carolina Agricultural and Technical State University
Menoufia University Greensboro, NC, United States
Shebeen El-Kom, Menoufia, Egypt
Sayın, Barış
Safiuddin, Md. Engineering Faculty
George Brown College Istanbul, Turkey
Toronto, ON, Canada
Schackow, Adilson
Sahoo, Dipak Kumar Santa Catarina State University
Cochin University of Science and Technology Joinville, Santa Catarina, Brazil
Kochi, Kerala, India
Schindler, Anton
Said, Shwan Auburn University
Erbil, Iraq Auburn, AL, United States
Sakban, Haider Schlicke, Dirk

Fullerton, CA, United States Graz University of Technology
Graz, Austria
Salib, Sameh
Markham, ON, Canada Seco, Laura
Coimbra, Portugal
Sancak, Emre
Suleyman Demirel University Selvaraj, Kandasamy
Isparta, Turkey Teegala Krishna Reddy Engineering College
Hyderabad, Telangana, India
Santa Maria, Hernan
Pontificia Universidad Católica de Chile Seo, Junwon
Santiago, RM, Chile South Dakota State University
Brookings, SD, United States
Santos, Jose
University of Madeira Seręga, Szymon
Funchal, Portugal Cracow University of Technology
Kraków, Poland
Santos, Sérgio
SBS Engineering Consultancy Sha, Fei
Goiânia, Goiás, Brazil Shandong University
Jinan, Shandong, China
Sapp, James
ATMI Precast
Aurora, IL, United States

REVIEWERS IN 2021
Shabana, Islam Shi, Ting
University of Sherbrooke Ashburn, VA, United States
Shirkhorshidi, Seyed Masoud
Shafigh, Payam Sharif University of Technology
Kuala Lumpur, Malaysia Tehran, Islamic Republic of Iran
Shafiq, Nasir Shivali, Ram

Universiti Teknologi Petronas Central Soil and Materials Research Station
Tronoh, Perak, Malaysia New Delhi, India
Shah, Santosh Shoukry, Hamada

Dharmsinh Desai University Cairo, Egypt
Nadiad, Gujarat, India
Shrive, Nigel
Shaheen, Ehab University of Calgary
University of Calgary Calgary, AB, Canada
Calgary, AB, Canada
Siddique, Mohammad
Shahmansouri, Amir Ali Bangladesh University of Engineering and Technology
University of Mazandaran Dhaka, Bangladesh
Babolsar, Islamic Republic of Iran
Silfwerbrand, Johan
Shannag, M. Jamal KTH Royal Institute of Technology
King Saud University Stockholm, Sweden
Silva, John
Sharifi, Yasser Hilti North America
Vali-e-Asr University of Rafsanjan San Rafael, CA, United States
Rafsanjan, Islamic Republic of Iran
Silva, Marcos
Shariq, Mohd Campinas, Brazil
Civil Engineering Section
Aligarh, Uttar Pradesh, India Singh, DDN
New Delhi, India
Sharma, Akanshu
Institute of Construction Materials Singh, Harvinder
Stuttgart, Germany Guru Nanak Dev Engineering College
Ludhiana, Punjab, India
Shehata, Medhat
Ryerson University Singh, Navdeep
Toronto, ON, Canada Jalandhar, Punjab, India
Sheikh, Shamim Singh, Ran Bir

University of Toronto Central University of Haryana
Toronto, ON, Canada Mahendragarh, Haryana, India
Shen, Yin Smadi, Mohammad

Tongji University Jordan University of Science and Technology
Shanghai, China Irbid, Jordan
Sherif, Mohtady Smyl, Danny

Ryerson University Faculty of Engineering and Architectural University of Sheffield
Science Sheffield, United Kingdom of Great Britain and Northern Ireland
Toronto, ON, Canada
Snellings, Ruben
Shi, Caijun VITO
Hunan University Mol, Belgium
Changsha, Hunan, China

REVIEWERS IN 2021
Soares, Renan Gustavo Pacheco Sujjavanich, Suvimol
Universidade Federal de Pernambuco Kasetsart University
Recife, Pernambuco, Brazil Bangkok, Thailand
Sofi, Massoud Suksawang, Nakin

University of Melbourne Florida Institute of Technology
Melbourne, VIC, Australia Melbourne, FL, United States
Sokoli, Drit Sultan, Mohamed

Thornton Tomasetti Al-Azhar University
Austin, TX, United States Cairo, Egypt
Soliman, Ahmed Sun, Tao

Western University Wuhan, China
London, ON, Canada
Sun, Zhihui
Soltanzadeh, Fatemeh University of Louisville
ISISE Louisville, KY, United States
Guimarães, Minho, Portugal
Sunayana, Sushree
Sonebi, Mohammed Indian Institute of Technology Kharagpur
Queen’s University Belfast Kharagpur, West Bengal, India
Belfast, United Kingdom of Great Britain and Northern Ireland
Suraneni, Prannoy
Sonyal, Muhammad University of Miami
University of Engineering and Technology Coral Gables, FL, United States
Lahore, Punjab, Pakistan
Suryawanshi, Yogesh
Sotiriadis, Konstantinos JSPM’s Imperial College of Engineering and Research Wagholi
Ústav Teoretické a Aplikované Mechaniky Akademie věd České Pune, Maharashtra, India
Republiky
Praha, Czech Republic Söylev, Altug
Yeditepe University
Spinella, Nino Istanbul, Turkey
University of Messina
Messina, Italy T., Hemalatha
CSIR-Structural Engineering Research Centre
Spyridis, Panagiotis Chennai, Tamil Nadu, India
Institute for Structural Engineering
Vienna, Austria Tadayon, Mohsen
Iranian Concrete Institute
Sreenivasappa, Nandeesh Tehran, Islamic Republic of Iran
M S Ramaiah Institute of Technology
Bangalore, Karnataka, India Tadros, Maher
e.construct.US, LLC
Stivaros, Pericles Omaha, NE, United States
Feld, Kaminetzky & Cohen, P.C.
Jericho, NY, United States Tahmouresi, Behzad
Rasht, Islamic Republic of Iran
Stynoski, Peter
U.S. Army Corps of Engineers Waterways Experiment Station Tajaddini, Abbas
Champaign, IL, United States University of Bath Faculty of Engineering and Design
Bath, United Kingdom of Great Britain and Northern Ireland
Su, Junsheng
Tianjin University Tang, Chao-Wei
Tianjin, China Cheng Shiu University
Niaosong District, Kaohsiung City, Taiwan, China
Su, Yu-Min
National Kaohsiung University of Science and Technology
Kaohsiung, Taiwan, China

REVIEWERS IN 2021
Tao, Lei Toubia, Elias
Shell Oil Company University of Dayton
Houston, TX, United States Dayton, OH, United States
Tarafder, Nilanjan Trautner, Christopher

Siliguri, West Bengal, India University of California, San Diego
La Jolla, CA, United States
Tariq, Samia
University of Engineering and Technology, Lahore Tregger, Nathan
Lahore, Punjab, Pakistan GCP Applied Technologies Inc.
Cambridge, MA, United States
Tatiraju, Ramanjaneyulu
North Dakota State University Tuchscherer, Robin
Fargo, ND, United States Northern Arizona University
Flagstaff, AZ, United States
Tawfic, Yasser
Minia University Tuhin, Ishtiaque
Minia, Egypt South Dakota State University
Brookings, SD, United States
Tazarv, Mostafa
University of Nevada, Reno Turgut, Paki
Reno, NV, United States İnönü Üniversitesi
Malatya, Turkey
Tegos, Ioannis
Salonica, Greece Umipig, Jorge
Vista Land & Lifescapes Inc.
Tennis, Paul Mandaluyong City, Metro Manila, Philippines
Portland Cement Association
Lawrence, KS, United States Ungermann, Jan
RWTH Aachen University
Teo, Wee Aachen, Germany
Heriot-Watt University - Malaysia Campus
Putrajaya, Malaysia Vanhove, Yannick
Artois University - Technology Institute Béthune
Teomete, Egemen Béthune, France
Izmir, Turkey
Vasovic, Dejan
Thiagarajan, Ganesh University of Belgrade, Faculty of Architecture
University of Missouri–Kansas City Belgrade, Serbia
Kansas City, MO, United States
Velázquez Rodríguez, Sergio
Thokchom, Suresh Universidad Panamericana
Manipur Institute of Technology Zapopan, Jalisco, Mexico
Imphal, Manipur, India
Vercher, Jose
Thomas, Michael Polytechnic University of Valencia
University of New Brunswick Valencia, Valencia, Spain
Fredericton, NB, Canada
Vintzileou, Elizabeth
Tiwari, Ashok Kumar National Technical University of Athens
Ultratech India Limited Athens, Greece
New Mumbai, Maharashtra, India
Vollum, Robert
Tiznobaik, Mohammad Department of Civil and Environmental Engineering
University of British Columbia | Okanagan Campus London, United Kingdom of Great Britain and Northern Ireland
Kelowna, BC, Canada
Vosahlik, Jan
Tognonvi, Tohoue Monique ICON
Université de Sherbrooke Austin, TX, United States

REVIEWERS IN 2021
Vu, Ngoc Son Worley II, Robert
National University of Civil Engineering University of Vermont
Hanoi, Viet Nam Burlington, VT, United States
Wan-Wendner, Roman Woyciechowski, Piotr

Ghent University Warsaw University of Technology
Ghent, Belgium Warsaw, Poland
Wang, Dun Wu, Chenglin

Research Institute of Structural Engineering and Disaster Missouri University of Science and Technology
Reduction, Tongji University Rolla, MO, United States
Shanghai, China
Wu, Kai
Wang, Jia-Ji Nanjing, Jiangsu, China
Tsinghua University
Beijing, China Wu, Tao
Chang’an University
Wang, Junjie Xi’an, China
Tsinghua University
Beijing, China Wu, Zemei
Missouri University of Science and Technology
Wang, Kejin Rolla, MO, United States
Iowa State University
Ames, IA, United States Xi, Xun
University of Strathclyde
Wang, Shaojie Glasgow, United Kingdom of Great Britain and Northern Ireland
Shandong Agricultural University
Tai’an, China Xia, Jin
Zhejiang University
Wang, Xiao-Yong Hangzhou, Zhejiang, China
Kangwon National University
Chuncheon, Republic of Korea Xiang, Ping
City University of Hong Kong College of Science and
Wardhono, Arie Engineering
The State University of Surabaya Kowloon, Hong Kong, China
Surabaya, Jawa Timur, Indonesia
Xie, Tianyu
Wegner, Leon RMIT University
University of Saskatchewan College of Engineering Melbourne, VIC, Australia
Saskatoon, SK, Canada
Xu, Aimin
Wei, Ya ARRB Group
Tsinghua University Melbourne, VIC, Australia
Beijing, China
Yahia, Ammar
Wei-Jian, Yi Université de Sherbrooke
Changsha, China Sherbrooke, QC, Canada
Werner, Anne Yan, Peiyu

Southern Illinois University Edwardsville Tsinghua University
Edwardsville, IL, United States Beijing, China
Woju, Utino Yang, Deqiang

National Institute of Technology Karnataka Inner Mongolia University of Technology
Mangalore, Karnataka, India Hohhot, Inner Mongolia, China
Wong, Sook Fun Yang, Hee Jun

Nanyang Technological University Ansan-si, Gyeonggi-do, Republic of Korea
Singapore, Singapore

REVIEWERS IN 2021
Yang, Junlong Younis, Khaleel
Dalian University of Technology University of Sheffield
Dalian, China Sheffield, United Kingdom of Great Britain and Northern Ireland
Yang, Keun-Hyeok Yousefpour, Hossein

Kyonggi University The University of Texas at Austin
Suwon, Kyonggi-do, Republic of Korea Austin, TX, United States
Yang, Shuyan Youzhi, Liu

Yinchuan City, China China Institute of Water Resources and Hydropower Research
Beijing, China
Yang, Sung Chul
Hongik University Yu, Jing
Mapo-gu, Republic of Korea Sun Yat-sen University
Guangzhou, China
Yapa, Hiran
University of Peradeniya Faculty of Engineering Yu, Xiaoniu
Peradeniya, Sri Lanka Southeast University
Nanjing, China
Yassein, Mohamed
Abu Dhabi, United Arab Emirates Yuan, Qiang
Central South University
Yasso, Samir Changsha, China
University of Mosul
Mosul-Bartella, Nineveh, Iraq Yuniarsyah, Eko
Institut Teknologi Bandung
Yazdani, Nur Bandung, Indonesia
The University of Texas at Arlington
Arlington, TX, United States Yuye, Xu
Quanzhou, Fujian, China
Yehia, Sherif
American University of Sharjah Zaidi, S. Kaleem
Sharjah, United Arab Emirates Aligarh Muslim University
Aligarh, UP, India
Yekrangnia, Mohammad
Shahid Rajaee Teacher Training University Zapata, Luis
Tehran, Islamic Republic of Iran Universidad Industrial de Santander
Bucaramanga, Santander, Colombia
Yepez, Fabricio
Universidad San Francisco de Quito Zhang, Feng
Quito, Pichincha, Ecuador Nanjing Hydraulic Research Institute
Nanjing, China
Yerramala, Amarnath
University of Dundee Zhang, Hongen
Dundee, United Kingdom of Great Britain and Northern Ireland Tongji University
Shanghai, China
Yingjie, Shan
Central South University Zhang, Jiaolong
Changsha, Hunan, China Tongji University
Shanghai, China
YK, Guru
Indian Institute of Science Zhang, Peng
Bangalore, India Karlsruhe Institute of Technology (KIT)
Karlsruhe, Germany
Yost, Joseph
Villanova University Zhang, Wei
Villanova, PA, United States Tsukuba Daigaku
Tsukuba, Japan

REVIEWERS IN 2021
Zhang, Xingbin Zhong, Rui
Beijing, China Southeast University
Nanjing, China
Zhao, Haitao
Hohai University Zhou, Wei
Nanjing, China Harbin Institute of Technology
Harbin, China
Zhao, Xin-Yu
South China University of Technology Zhu, Yanping
Guangzhou, China Missouri University of Science and Technology
Rolla, MO, United States
Zheng, Herbert
Gammon Construction Limited Zych, Mariusz
Hong Kong, Hong Kong, China Politechnika Krakowska im. Tadeusza Kościuszki
Kraków, Poland
Zhenhong, Wang
China Institute of Water Resources and Hydropower Research
Beijing, China

NOTES:

ACI
STRUCTURAL
J O U R N A L
J O U R N
The American Concrete Institute (ACI) is a leading authority and

resource worldwide for the development and distribution of
consensus-based standards and technical resources, educational
programs, and certifications for individuals and organizations involved
in concrete design, construction, and materials, who share
a commitment to pursuing the best use of concrete.
Individuals interested in the activities of ACI are encouraged to

explore the ACI website for membership opportunities, committee
activities, and a wide variety of concrete resources. As a volunteer
member-driven organization, ACI invites partnerships and welcomes
all concrete professionals who wish to be part of a respected,
connected, social group that provides an opportunity for professional
growth, networking, and enjoyment.

SJ March 2022 V. 119 No. 2

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

SJ March 2022 V. 119 No. 2

Uploaded by

Copyright:

Available Formats

V. 119, NO.

A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE

Board of Direction 19 Maximum Shear Strength of Reinforced Concrete Beams, by Shyh-

Engineers 155 Uncertainty Modeling and Analysis of Punching Shear Resistance of

ACI Structural Journal/March 20221

2 ACI Structural Journal/March 2022

Effect of Noncontact Lap Splices in Reinforced Concrete

Current design codes specify the bar development length empiri-

Keywords: lap splice; moment distribution; noncontact lap splice; rein-

ACI Structural Journal/March 2022 3

4 ACI Structural Journal/March 2022

f  lm0 = max(0.7fydb)/4fbd, 15db, 200 mm; kd is coefficient for arrangement of transverse

αm3 = kd(Kmtr – αt/50) ≥ 0

ACI Structural Journal/March 2022 5

Note: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi; cl is spacing of spliced bars.

6 ACI Structural Journal/March 2022

ACI Structural Journal/March 2022 7

Fig. 4—Test setup. (Note: Units in mm.)

Fig. 5—Moment-deflection and load-deflection relationships of four-point loading test specimens.

8 ACI Structural Journal/March 2022

ACI Structural Journal/March 2022 9

10 ACI Structural Journal/March 2022

Fig. 8—Crack pattern and failure mode of three-point loading tests.

ACI Structural Journal/March 2022 11

Fig. 13—Reinforcing bar stress ratio of specimens under

12 ACI Structural Journal/March 2022

ACI Structural Journal/March 2022 13

14 ACI Structural Journal/March 2022

ACI Structural Journal/March 2022 15

16 ACI Structural Journal/March 2022

Table B1—Test parameters of existing noncontact bar-splice test specimens

Note: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi.

ACI Structural Journal/March 2022 17

Notes: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi.

18 ACI Structural Journal/March 2022

Maximum Shear Strength of Reinforced Concrete Beams

ACI Structural Journal/March 2022 19

RESEARCH SIGNIFICANCE Vmax, ACI  8(w1/ 3  1)bw d f c (fcꞌ in psi)

Av f yt d Vmax, Lee  (0.66w1/ 3 f c  0.2 f c)bw d (fcꞌ in MPa) (5)

20 ACI Structural Journal/March 2022

ACI Structural Journal/March 2022 21

22 ACI Structural Journal/March 2022

ACI Structural Journal/March 2022 23

MODEL COMPARISON ON MAXIMUM SHEAR

24 ACI Structural Journal/March 2022

Author No. Author No. Author No.

ACI Structural Journal/March 2022 25

26 ACI Structural Journal/March 2022

296 B-4 0.66 51 339 1.17 T (25°) — 1.41 0.76 0.75

ACI Structural Journal/March 2022 27

Fig. 6—Comparison of failure mode estimations with cracking photos.

28 ACI Structural Journal/March 2022

ACI Structural Journal/March 2022 29

30 ACI Structural Journal/March 2022

Long-Term Behavior of Glass Fiber-Reinforced Polymer-

ACI Structural Journal/March 2022 31

RESEARCH SIGNIFICANCE ANALYTICAL MODELING

Ec  0   57, 000 f c in psi (7a)

Ec  0   4730 f c in MPa (7b)

Fig. 1—Flowchart of modeling approach.

32 ACI Structural Journal/March 2022

 t   u (10) M x   ydA  M xc  M xf (17)

t00.5 M y   xdA  M yc  M yf (18)

Board of Direction 19 Maximum Shear Strength of Reinforced Concrete Beams, by Shyh-

Engineers 155 Uncertainty Modeling and Analysis of Punching Shear Resistance of

ACI Structural Journal/March 20221

2 ACI Structural Journal/March 2022

f lm0 = max(0.7fydb)/4fbd, 15db, 200 mm; kd is coefficient for arrangement of transverse

RESEARCH SIGNIFICANCE Vmax, ACI 8(w1/ 3 1)bw d f c (fcꞌ in psi)

Av f yt d Vmax, Lee (0.66w1/ 3 f c 0.2 f c)bw d (fcꞌ in MPa) (5)

Ec 0 57, 000 f c in psi (7a)

Ec 0 4730 f c in MPa (7b)

t u (10) M x ydA M xc M xf (17)

t00.5 M y xdA M yc M yf (18)

M x M 210 M 22 x M 23 y (27) QyT Qy n f 1 Afi x fi (42)