ACI Structural Journal Sept. - 2023 V. 120 No. 5

V. 120, NO.
5
SEPTEMBER 2023
ACI
STRUCTURAL J O U R N A L
A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE
@seismicisolation
@seismicisolation
CONTENTS
STATEMENT OF OWNERSHIP
Editorial Board
Michael Kreger, Editor-in-Chief
ACI Structural Journal
University of Alabama
Catherine French September 2023, V. 120, No. 5
University of Minnesota
Mary Beth Hueste a journal of the american concrete institute
Texas A&M University an international technical society
David Sanders
Iowa State University 3 Ultra-High-Toughness Concrete Retrofitted Boundary Column Shear
Gustavo Parra-Montesinos
University of Wisconsin–Madison
Walls: Tests and Capacity Prediction, by Hongmei Zhang, Giorgio
Monti, Yuanfeng Duan, Zhiyuan Chen, and Chen Gu
Board of Direction
19 Hysteretic Model of Coupler Box Assembly for Seismic Retrofitting of
President
Antonio Nanni
Severely Damaged Reinforced Concrete Buildings, by Naveen Kumar
Kothapalli, R. Siva Chidambaram, and Pankaj Agarwal
Vice Presidents
Maria Juenger 37 Seismic Performance of Five-Spiral Concrete Columns with Large-
Michael J. Paul Diameter Longitudinal Reinforcement, by Yu-Chen Ou and Brandon Li
Directors
Oscar R. Antommattei
49 Rehabilitation of Exterior Beam-Column Joint by Geopolymer Mortar
Peter Barlow under Quasi-Static Loading, by Arshad Hussain Choudhury and Aminul
Michael C. Brown Islam Laskar
Arturo Gaytan Covarrubias
Anthony R. DeCarlo Jr. 63 Strength and Behavior of Glass Fiber-Reinforced Polymer-Rein-
John W. Gajda forced Concrete Box Girders without Web Reinforcement under Pure
Carol Hayek Torsion, by Ibrahim T. Mostafa, Salaheldin Mousa, Hamdy M. Mohamed,
Kamal H. Khayat
Robert C. Lewis and Brahim Benmokrane
Anton K. Schindler
Matthew R. Sherman 77 Experimental Study on Drilled Shaft Reinforcing Bar Anchorage in
Lawrence L. Sutter Footings, by Yousun Yi, Hyunsu Kim, Jongkwon Choi, Juan Murcia-Delso,
and Oguzhan Bayrak
Past President Board Members
Jeffrey W. Coleman
Cary S. Kopczynski 93 Design of Glass Fiber-Reinforced Polymer-Reinforced Concrete
Charles K. Nmai Columns per ACI CODE-440.11-22, by Zahid Hussain and Antonio Nanni
Executive Vice President 109 Cover Spalling in Reinforced Concrete Beams Subjected to Pure
Frederick H. Grubbe Torsion, by Allan Kuan, Evan C. Bentz, and Michael P. Collins
Staff 121 Behavior of Concrete Bridge-Deck Slabs Reinforced with Basalt

Publisher Fiber-Reinforced Polymer and Steel Bars, by Yahia M. S. Ali, Xin Wang,
John C. Glumb Shui Liu, and Zhishen Wu
Managing Director, Engineering and
Professional Development 139
Failure Mode-Dependent Behavior of Carbon Fiber-Reinforced
Michael L. Tholen Polymer Prestressed Concrete Girders, by Yail J. Kim, Jun Wang,
Woo-Tai Jung, Jae-Yoon Kang, and Jong-Sup Park
Engineers
Will J. Gold 153 Deflection Control Methodologies for Curvilinear Concrete Members
Matthew R. Senecal Reinforced with Glass Fiber-Reinforced Polymer Bars, by Seyed
Michael L. Tholen Mohammad Hosseini, Salaheldin Mousa, Hamdy M. Mohamed, and
Gregory M. Zeisler
Brahim Benmokrane
Managing Editor 169 Seismic-Fire-Combined Loadings Applied to Carbon Fiber-Rein-
Lauren E. Mentz
forced Polymer-Confined Reinforced Concrete Columns, by Ju-Hyung
Associate Editor Kim, Yail J. Kim, and Jun Wang
Kimberly K. Olesky
Editors Contents continued on next page

Erin N. Azzopardi
Lauren C. Brown Discussion is welcomed for all materials published in this issue and will appear ten months from
Kaitlyn J. Dobberteen this journal’s date if the discussion is received within four months of the paper’s print publication.
Tiesha Elam Discussion of material received after specified dates will be considered individually for publication or
Angela R. Noelker private response. ACI Standards published in ACI Journals for public comment have discussion due
Kelli R. Slayden dates printed with the Standard.
ACI Structural Journal
Copyright © 2023 American Concrete Institute. Printed in the United States of America.
The ACI Structural Journal (ISSN 0889-3241) is published bimonthly by the American Concrete Institute. Publication
office: 38800 Country Club Drive, Farmington Hills, MI 48331. Periodicals postage paid at Farmington, MI, and at
additional mailing offices. Subscription rates: $192 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.
@seismicisolation
@seismicisolation
ACI Structural Journal/September 20231

Contributions to
CONTENTS
MEETINGS ACI Structural Journal
The ACI Structural Journal is an open
SEPTEMBER 28-30—2023 Concrete NZ Conference, forum on concrete technology and papers
Hamilton, New Zealand, https://confer. related to this field are always welcome.
17-20—17th International Congress on All material submitted for possible publi-
eventsair.com/concretenz-conference-2023 cation must meet the requirements of
Polymers in Concrete (ICPIC 2023),
the “American Concrete Institute Publi-
Warsaw, Poland, https://icpic23.org OCTOBER cation Policy” and “Author Guidelines
and Submission Procedures.” Prospective
17-20—4th Fiber Reinforced Concrete 6-9—2023 APA Annual Convention, authors should request a copy of the Policy
International ACI-fib-RILEM Joint Delray Beach, FL, www.archprecast.org/ and Guidelines from ACI or visit ACI’s
Workshop: from Design to Structural annual-convention website at www.concrete.org prior to
Applications, Tempe, AZ, www. submitting contributions.
rilem.net/agenda/4th-fiber-reinforced- 9-11—Dni Betonu 2023, Wisła, Poland,
Papers reporting research must include
concreteinternational-aci-fib-rilem-joint- www.dnibetonu.com/en a statement indicating the significance of
workshop-from-design-tostructural- the research.
applications-1537 12-14—National Academy of
Construction Annual Meeting, Boston, The Institute reserves the right to return,
MA, www.naocon.org/annual-meetings without review, contributions not meeting
18-22—16th International Congress on the the requirements of the Publication Policy.
Chemistry of Cement 2023 (ICCC 2023),
Bangkok, Thailand, www.iccc2023.org 12-14—NPCA 58th Annual Convention, All materials conforming to the Policy
Oklahoma City, OK, https://precast.org/ requirements will be reviewed for editorial
convention/ quality and technical content, and every
25-29—2nd Conference of the European effort will be made to put all acceptable
Association on Quality Control of Bridges papers into the information channel.
and Structures – EUROSTRUCT 2023, 18-19—BUILDEX Alberta, Calgary, AB,
Canada, www.buildexalberta.com However, potentially good papers may be
Vienna, Austria, https://eurostruct.org/ returned to authors when it is not possible
eurostruct-2023 25-27—2023 Decarbonization Conference to publish them in a reasonable time.
26-29—Greenbuild 2023: Building for the Built Environment, Washington,

Discussion
Transformation, Washington, DC, https:// DC, www.ashrae.org/conferences/
topical-conferences/2023-decarbonization- All technical material appearing in the
greenbuild.informaconnect.com/2023 ACI Structural Journal may be discussed.
conference-for-the-built-environment
If the deadline indicated on the contents
27-30—ASCC 2023 Annual Conference, page is observed, discussion can appear
Grand Rapids, MI, https://ascconline.site- in the designated issue. Discussion should
ym.com/event/ac2023 be complete and ready for publication,
including finished, reproducible illustra-
tions. Discussion must be confined to the
scope of the paper and meet the ACI Publi-
cation Policy.
Follow the style of the current issue.
Be brief—1800 words of double spaced,
typewritten copy, including illustrations
and tables, is maximum. Count illustra-
tions and tables as 300 words each and
submit them on individual sheets. As an
ACI CONCRETE CONVENTION: FUTURE DATES 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.
2023—Oct. 29-Nov. 2, Westin Boston Seaport District & Boston Convention and Exhibition margins, and include two good quality
Center, Boston, MA copies of the entire discussion. References
2024—March 24-28, Hyatt Regency New Orleans, New Orleans, LA should be complete. Do not repeat refer-
2024—November 3-7, Philadelphia Marriott Downtown, Philadelphia, PA ences cited in original paper; cite them
2025—Mar. 30-Apr. 3, Sheraton Centre Toronto Hotel, Toronto, ON, Canada by original number. Closures responding
to a single discussion should not exceed
For additional information, contact: 1800-word equivalents in length, and to
Event Services, ACI multiple discussions, approximately one
half of the combined lengths of all discus-
38800 Country Club Drive
sions. Closures are published together
Farmington Hills, MI 48331 with the discussions.
Telephone: +1.248.848.3795
Discuss the paper, not some new or
email: conventions@concrete.org
outside work on the same subject. Use
references wherever possible instead of
repeating available information.
ON COVER: 120-S79, p. 29, Fig. 11—Damaged model frame after preliminary test.
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.
@seismicisolation Journals.Manuscripts@concrete.org
@seismicisolation
2 ACI Structural Journal/September 2023

ACI STRUCTURAL JOURNAL TECHNICAL PAPER
Title No. 120-S78
Ultra-High-Toughness Concrete Retrofitted Boundary

Column Shear Walls: Tests and Capacity Prediction
by Hongmei Zhang, Giorgio Monti, Yuanfeng Duan, Zhiyuan Chen, and Chen Gu
The rapid growth of the construction industry in Asia and the conse- increases the initial dimensions (Woods et al. 2016); besides,
quent updating of design specifications put forward higher perfor- it presents the problem of transferring the bending action
mance requirements for structural components, which results in a from the existing to the new elements (Woods et al. 2016).
large number of existing shear walls that are not compliant with the Fiber-reinforced concrete (FRC) exhibited excellent
current seismic standards. A prospective retrofitting method, which
strength, toughness, and damage tolerance (Li et al. 2001;
is based on replacing the existing boundary concrete or attaching
Boshoff et al. 2009) due to crack prevention and strain-hard-
external boundary columns to nonconforming shear walls, is
experimentally studied. Four shear-wall specimens were designed ening effects. Fischer and Li (2002a) emphasized that the
according to the current Chinese design code: one using plain most important impact of high-performance FRC compos-
concrete boundary columns and three using ultra-high-toughness ites on the structural response is to maintain the integrity of
boundary columns (UHTBCs), adopting three different strength- and provide lateral stability to steel reinforcements, thereby
ening strategies relevant to the boundary size and the connection ensuring adequate cyclic inelastic deformation and energy
form. Cyclic performance, damage patterns due to UHTBCs, and dissipation. Experiments involving columns (Mirmiran
connection form are discussed based on the experimental results, and Shahawy 1997; Fischer and Li 2002a, 2003), beams
from which it was ascertained that shear walls with UHTBCs show (Li and Xu 2009; Prota et al. 2004), and beam-column
improved seismic performance, compatible with the requirements joints (Parra-Montesinos 2005; Prota et al. 2004) have
of the current seismic design code, even for the reduced-boundary
proved that structural elements with FRC exhibit improved
UHTBCs and non-connection specimens. The predictive equation
seismic resistance. The bridging effect helps the structural
for the sectional moment capacity of shear walls with UHTBCs
was discussed as a practical tool for retrofitting applications. This elements maintain the lateral bearing capacity (Naaman and
study highlights the most important features of a rapid retrofitting Najm 1991) and alleviate damage evolution (Fischer and
measure to improve the resilience of existing nonconforming shear- Li 2002b). Thanks to the high tensile-strength-to-weight
wall structures, while also proving to be an effective measure for ratio and durability, FRC is ideal for repairing and retro-
newly constructed structures. fitting infrastructures and historical buildings (Mustafaraj
et al. 2020). Other authors (Xu et al. 2012; Li et al. 2009)
Keywords: boundary columns; cyclic loading test; retrofitting; seismic proposed ultra-high-toughness cementitious composites
performance; shear wall; ultra-high-toughness.
(UHTCCs) containing high-strength short fibers, which
showed excellent performance in bending elements and
INTRODUCTION
proved to be a viable alternative to plain concrete, thanks to
Background
their strain-hardening behavior and simple applicability to
Reinforced concrete (RC) shear walls are widely used in
existing structural elements.
mid- and high-rise structures as seismic-resistant elements.
Applying FRC composites to improve the flexural
Under the combined action of flexure, shear, and torsion, the
capacity of shear walls has recently attracted research
damage tends to concentrate in the outer boundary zones of
interest (Parra-Montesinos 2005; Parra-Montesinos et al.
RC shear walls (Mutō 1969; Zhang et al. 2014). Cracked
2006; Dazio et al. 2008; Olsen and Billington 2011; Li et al.
concrete weakens its compression capacity, which in turn
2014), where the positive effects on the seismic performance
expedites concrete crushing, leading to the failure of the
were demonstrated. FRC composites are also considered to
shear walls (Fischer and Li 2002a; Zhang et al. 2014). This
apply in shear-wall retrofitting (Kesner and Billington 2005;
shear and flexure interaction combined with local damage
Liang et al. 2013; Dang et al. 2014; Lu et al. 2018; Zhang
makes retrofitting of such shear walls a challenging endeavor.
et al. 2020).
To improve the flexural and shear performance of existing
and newly constructed shear walls, different retrofit strate-
RESEARCH SIGNIFICANCE
gies have been conceived—for example, the insertion of RC
In most of the existing shear walls constructed before 2001,
boundary columns (Al-Gemeel and Zhuge 2019), attaching
due to material deterioration and old specification-based
a steel profile (Chrysanidis and Tegos 2020), or even
enlarging the boundary elements. Attaching a steel profile ACI Structural Journal, V. 120, No. 5, September 2023.
to an existing concrete structure involves serious prob- MS No. S-2021-064.R5, doi: 10.14359/51738829, received July 15, 2022, and
reviewed under Institute publication policies. Copyright © 2023, American Concrete
lems of effective collaboration and bond durability (Zhou Institute. All rights reserved, including the making of copies unless permission is
obtained from the copyright proprietors. Pertinent discussion including author’s
et al. 2010). Inserting additional jacketed concrete columns 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.
@seismicisolation
@seismicisolation
ACI Structural Journal/September 2023 3

Table 1—Specimen parameters
Boundary
Cast- columns Boundary columns Longitudinal Boundary
Section, in-place width, Axial longitudinal reinforcement in columns Transverse
Specimens mm concrete material compressive force reinforcement core wall stirrups reinforcement
200 mm,
SW-1 plain
concrete HRB400_D10 2 x 5ϕ6
2000 x
1000 x C40 200 mm, 600 kN ϕ6@80 ϕ6@125
UCSW-2
125 UHTCC
UCSW-3 150 mm,
HRB400_D10 2 x 6ϕ6
UCSW-4 UHTCC
designs, the bending capacity may not conform to current lengths of SW-1 and UCSW-2 were 200 mm, and those of
specifications. Besides, the application of strengthening UCSW-3 and UCSW-4 were 150 mm. The experiments were
measures is often impeded by limited available space. In conducted in the State Key Laboratory of Disaster Reduction
this study, an alternative seismic strengthening method in Civil Engineering at Tongji University, Shanghai, China.
using ultra-high-toughness boundary columns (UHTBCs) The specimen configurations and details are shown in
on noncompliant shear walls with weak moment capacity Fig. 1. The boundary columns of Specimens UCSW-2
is presented. With this scope, four shear walls—one of plain to UCSW-4 were cast with UHTCC. The RC core wall,
concrete representing the current seismic design code (as a top beam, and foundation were cast with plain concrete.
reference specimen), two UHTBC shear walls with equal Except for the boundary column type, UCSW-2 was iden-
and reduced boundary elements, respectively (as strength- tical to SW-1, which was tested to determine the effect of
ened specimens by inserting additional elements or replacing the UHTBCs on the seismic performance. UCSW-3 was
the boundary concrete), and one UHTBC shear wall without designed with reduced boundary column widths and longitu-
shear connections (representing a case of rapid retrofit- dinal bars to investigate the possibility of reduced boundary
ting)—were cyclically tested to validate the corresponding columns. For Specimens UCSW-2 and UCSW-3, there were
strengthening strategies. key slots and shear reinforcing bars for connection to the
This study provides in-depth considerations on this rapid inner core wall. Half-grouted sleeve connections were used
strengthening measure that proved to be effective in retro- to connect the longitudinal bars of the UHTBC elements
fitting slender shear walls, through the effective use of a and the RC foundation (JG/T 398-2012 [2012]). No connec-
high-performance material, with the significant advantage tion was set for Specimen UCSW-4, which was designed
that the overall wall configuration is only slightly affected. to explore the possibility of no-connection retrofitting for
Furthermore, it also improves the bending capacity without ancient and protected buildings.
damaging the inner portions. The UHTBCs can be prefab-
ricated and assembled on site, and even replaced at a later Material properties
stage if needed. The developed predictive equation of the The UHTCC used in the boundary columns of the tested
moment capacity also provides a reference for design. shear walls was composed of: PO 52.5 cement, fly ash, silica
fume, and fine sand; water-reducing additive; water; and
EXPERIMENTAL TESTS high-strength polyvinyl alcohol (PVA) fibers (fiber content:
Specimen design 2%). The literature demonstrates that, as multiple cracking
The experimental shear-wall specimens were designed phenomena occur, UHTCC exhibits tensile strain-hardening
referring to an existing high-rise building built in the 1990s in behavior and attains a maximum strain capacity larger than
Shanghai in a second-category seismic site. In this area, the 0.03, even with typical fiber volume fractions of 2% or less
earthquake intensity level is Grade 7, which corresponds to a (Xu and Cai 2010; Li et al. 2009). The main properties of the
design peak ground acceleration (PGA) = 0.35g (g is gravity high-strength PVA fibers include: length = 12 mm; diameter
acceleration) having an exceedance probability of 10% in = 39 μm; tensile strength ftc = 1620 MPa; tensile modulus of
50 years. A half-scale shear wall (SW-1) was designed and elasticity Ef = 42.8 GPa; ultimate strain = 0.07; and density
constructed with plain concrete as a reference wall, repre- = 1.3 g/cm3.
senting the shear wall satisfying the current seismic design Figure 2 shows the specimens and test views of dog-bone-
code (GB 50011-2010 [2010]). Three half-scale shear- shaped UHTCC specimens for tensile strength and cubic
wall specimens (UCSW-2, UCSW-3, and UCSW-4) were specimens of both plain concrete and UHTCC for compres-
designed with the same geometry and reinforcement as the sive strength. Figures 3(a) and (b) show the stress-strain
reference shear wall but strengthened with UHTBCs. For all curves of the tested material specimens. The main proper-
specimens, the 1:2 scale factor applies to all geometry quan- ties, including the elastic modulus and compressive strength
tities, while the materials are designed to be the same as in of the plain concrete and UHTCC, are presented in Table 2.
the original structure. Therefore, the load-bearing capacity Longitudinal reinforcement to the boundary columns was
is 1:4 of the original structure. Detailed descriptions of the formed from HRB400, and the main material properties of
specimens are presented in Table 1. The boundary column @seismicisolation
@seismicisolation

Fig. 1—Configuration details of tested shear walls.
Fig. 2—Dog-bone tensile and cubic compressive tests: (a) dog-bone-shaped specimen and tensile test; (b) UHTCC specimen;
and (c) plain concrete specimen.
Fig. 3—Tensile and compressive properties of plain concrete and UHTCC.

the reinforcing bars are presented in Table 3. The Young’s The dog-bone-shaped specimens are 250 mm in length,
modulus is between 205 and 210 GPa. 60 mm in width, and 13 mm in thickness, with an efficient
testing length and width of 80 mm and 30 mm, respectively.
@seismicisolation
@seismicisolation

As Fig. 3(a) shows, the three dog-bone-shaped speci- half-grouted sleeve connection effectively transferred tensile
mens show higher toughness. The tensile ultimate strain is force between the upper and lower steel bars.
over 0.03, which is significantly higher than that of plain The shear-wall specimens were manufactured based on
concrete, typically assumed as zero. The compressive stress- the material test results. Pictures of the specimen manufac-
strain curve was tested on cubic specimens of 150 mm turing process are shown in Fig. 5, including the main ingre-
sides, while the elastic modulus is obtained on 150 x 150 x dients, the mixed cement, boundary column casting, and the
300 mm prismatic specimens. This study confirms the typi- testing view of the specimens.
cally ascertained literature finding that fibers have a positive
effect on the tensile strength, thanks to the beneficial crack Test setup and loading program
prevention effect. It is also shown that larger fiber fractions The test setup was composed of a horizontal loading
may affect the compressive strength (Xu and Cai 2010), system, a vertical loading system, and a measurement
whereas the compressive toughness improves, thanks to the system. The lateral load was applied by a horizontal
fiber bridging effect shown in Fig. 3(b). servo-actuator (with a load capacity of 630 kN). The lateral
As shown in Fig. 4, steel sleeves were employed to load was applied to one end of the top beam by the actuator
connect the UHTBC elements and the bottom beam. This through four steel screw rods with two steel plates attached
method was confirmed to be reliable in a previous study to the loading side. The vertical load was applied to the top
(Dazio et al. 2008) and used to connect the retrofitted beam by three hydraulic jacks. Rollers were used at the top
boundary columns. The tube is made of Q235 steel, which is of the jacks to reduce the friction of the reaction frame, as
plain carbon structural steel with a yield strength of approxi- shown in Fig. 5(c). The lateral and vertical forces acting on
mately 235 MPa, and is filled with super-high-strength (≥60 the specimens were balanced by the reaction force of the
MPa) non-shrink cement-based grouting material, with a bottom beam anchored to the foundation.
grip strength of ≥4 MPa, to connect longitudinal C10 steel The loading scheme was designed according to the spec-
bars used in the boundary columns. Tensile tests of the three ifications of the Chinese standard JGJ/T 101-2015 (2015).
steel bar connection specimens were conducted before using Before the lateral load was applied, 40 to 60% of the designed
them in the boundary elements, as shown in Fig. 4. The three vertical load was applied and removed three times to elimi-
specimens showed ductile behavior by steel bar necking nate possible imperfections in the specimens. Subsequently,
outside the steel sleeve until breaking. This indicated that the
Table 3—Properties of HRB400 steel reinforcing
Table 2—Properties of concrete from tests bars from tests
Cubic Cylinder Yield Ultimate Elastic
compressive compressive Elastic Tensile Reinforcing strength, Yield strength, Ultimate modulus,
strength strength modulus strength bar MPa strain MPa strain GPa
Material fcu, MPa fc, MPa Ec, MPa ft, MPa
HRB400_
501 0.00235 539 0.0242 213
Plain D6
48.0 39.4 34,213 —
concrete
HRB400_
466 0.00224 598 0.1165 210
UHTCC 50.5 49.1 34,626 4.5 D6
Fig. 4—Half-grouted steel sleeve connection. @seismicisolation

@seismicisolation

Fig. 5—Construction sequence of boundary columns and testing view.
the designed vertical load was applied and kept constant, Bearing and deformation behavior
and then the lateral load was applied according to a cyclic The hysteresis curves (obtained through load cells applied
loading scheme. The loading sequence is designed as single on the actuator and through LVDTs for the displacements)
cycles with a 1 mm displacement increment before the spec- and crack development are shown in Fig. 7. In this case, the
imen yielded, followed by triplets of cycles with a 2 mm yield point was the cross point determined by the peak value
increment. Such a loading sequence allows for identifying and the secant line when the lateral load increased to 75%
the response envelope with good accuracy, and, at the same of the peak force; the failure was defined either at the 85%
time, ascertaining both the global pinching effect and the decrease of the lateral-force peak value or when the edge
local degradation effects of repeated cycles. The loading steel bars broke, whichever was less. The lateral loads and
was terminated either when significant concrete crushing the corresponding lateral displacements at the yield, peak,
occurred at the corner or when the reinforcing bar buckled. and ultimate states are presented in Table 4. The lateral
Linear variable differential transformers (LVDTs) were bearing forces at different drift ratios (1/500, 1/200, 1/100,
used to record the deformation of the shear walls, and strain and 1/50) are presented in Fig. 7.
gauges were attached to the reinforcing bars to monitor the As shown in Fig. 7 and Table 4, UCSW-2 showed a larger
strain development during the loading procedure. Forces capacity at failure than SW-1. With the contribution of the
were measured through load cells. All data were recorded by UHTBCs, the hysteresis loops were enlarged, exhibiting
an automatic data acquisition system. constant growth. During the cyclic loading procedure, the
lateral force and deformation are theoretically symmetric on
TEST RESULTS the pull and push sides. However, the actual response turns
Failure process and modes out to be not perfectly symmetrical because, after the first
The different patterns of concrete cracking and crushing at semi-cycle, the specimen is already damaged when pushed
failure are shown in Fig. 6. Failure was defined as an 85% toward the opposite direction. Thus, this slightly different
decrease in the lateral-force peak value. As shown in Fig. 6, behavior was regularized by averaging the values obtained
UCSW-2 exhibited much richer cracks than SW-1, and most on the two opposite semi-cycles. The lateral force and the
of them were horizontal orientation. Concrete crushing was corresponding lateral displacement of UCSW-2 were larger
observed at the corner of the plain core wall, which was finer than those of SW-1 by 14% and 8% at failure, respectively.
and denser than that in SW-1. Fewer cracks were observed The lateral force of UCSW-2 was higher than that of SW-1,
in UCSW-3 than in UCSW-2. The concrete-crushing areas particularly in the “serious damage” to “no collapse” states.
of these two specimens were significantly smaller than those The strength results at different drift ratios also indicate that
of SW-1. In contrast to UCSW-3, the cracks were horizontal UCSW-2 maintained a good resistance capacity even when
for UCSW-4; additionally, the concrete at the corner of the the lateral drift ratio was 1/50. Owing to the contribution of
core wall was seriously damaged at failure. the UHTBCs, the shear wall exhibited an enhanced lateral
bearing capacity and ductility.
@seismicisolation
@seismicisolation

Fig. 6—Failure modes and crack patterns of specimens.
Fig. 7—Hysteresis curves and evolving crack pattern.

UCSW-3 exhibited higher initial stiffness, larger peak UCSW-3 were still well rounded and developed more gradu-
lateral bearing force, and higher stiffness than SW-1, although ally compared with SW-1 and close to UCSW-2. The lateral
the confined length was smaller than that for SW-1. Addi- force of UCSW-3 in the peak state was increased by 12%
tionally, the number of longitudinal steel bars in the confined compared with SW-1. As shown in Fig. 7, the lateral bearing
boundary columns was reduced. The hysteresis loops of force was almost identical to that of UCSW-2 before the
@seismicisolation
@seismicisolation

Table 4—Critical points of lateral force and deformation characteristics
Yield Peak Ultimate
Specimen Loading direction Δy, mm Fy, kN Δm, mm Fm, kN Δu, mm Fu, kN
Push 17 240 32 285 38 270
SW-1 Pull 13 225 33 252 36 246
Average 15 232 32 269 37 258
Push 19 283 36 328 45 279
UCSW-2 Pull 14 282 28 328 38 279
Average 17 282 32 328 42 279
Push 13 241 30 298 37 254
UCSW-3 Pull 15 255 30 303 38 257
Average 14 248 30 301 38 256
Push 15 222 33 267 46 227
UCSW-4 Pull 19 226 31 260 43 221
Average 17 224 32 264 45 224
specimen was seriously damaged, and it decreased margin- bars’ dowel action, shear cannot be transferred between the
ally faster than that of UCSW-2 near the “no collapse” stage. boundary columns and core wall, which eventually end up
This indicates that the seismic performance of UCSW-3 carrying the horizontal force independently. To obtain an
satisfied the requirements of the current seismic design code. improved energy dissipation effect, connecting shear keys
Moreover, the specimen with a weak connection are essential (Park 2006).
(UCSW-4) exhibited high toughness and collapse-resistance
capacity in the experiment. Although the lateral force of Crack development
UCSW-4 was smaller than those of UCSW-2 and UCSW-3 The toughness of FRC composites is essentially due to
before the drift ratio reached 1/100, it was not smaller than the cracking prevention by the bridging effect of the fibers.
that of SW-1. The lateral bearing force of UCSW-4 at a For this reason, the crack growth during the experiment was
drift ratio of 1/50 was 34.3% and 5.6% larger than those of investigated. Figure 9 shows the crack distribution profile of
SW-1 and UCSW-3, respectively. The results indicate that each specimen at the yield, peak, and ultimate states and at
the seismic performance was significantly improved through drift ratios of 1/200 (~10 mm) and 1/100 (~20 mm), around
the connection-strengthening strategy involving UHTBCs, which is approximately “medium damage” and “serious
particularly in the large-deformation stage. damage.” The fractal index D (Mandelbrot 1982), as in
The hysteresis curves of Specimens UCSW-2, -3, and -4 Eq. (1), was used to evaluate the crack distribution as follows
show a pinching effect caused by slippage occurring at the
interface between the boundary columns and core wall. As ln N(r)
_
demonstrated by other authors (for example, Park [2006] D = − lr→0
im (1)
ln r
and Sørensen et al. [2017a,b]), shear keys can be prevented
from slipping by adopting concrete key slots and shear where r represents the size of a square grid cell; and N
reinforcing bars, which, however, show different damage represents the number of grids with cracks inside. The
modes while differently affecting the shearing capacity. vertical cracks appearing at the interface between the
In this study, Specimens UCSW-2 and -3 have shear key boundary column and the core wall are caused by a weak
slots and shearing reinforcing bars, as shown in Fig. 8(a). interface connection, which did not significantly impact
The shear reinforcing bars are placed at 160 mm spacing. the final failure and can be improved in actual use; thus,
Specimen UCSW-4 is constructed directly by precast and they were disregarded. In addition to the fractal index, the
on-site cast concrete, as shown in Fig. 8(b). In principle, crushing ratio, defined as the crushing area divided by the
under shear force, the concrete shear strength at the inter- total shear-wall surface area, was extracted from the images
face is firstly overcome, then the concrete key slots and for each specimen and was used to evaluate the concrete-
shear reinforcing bars continue to bear the shear force. In crushing performance at ultimate (Table 5).
the tests, the interfaces of Specimens UCSW-2 and UCSW-3 Figure 9 reveals smaller horizontal cracks in the UCSW-2
slightly cracked along the key, indicating that key slots and specimen than in SW-1. The fractal indexes of the three
shear reinforcing bars play an important role in the ultimate UHTBC shear walls were significantly larger than that of
state and that they are activated after concrete cracks. The SW-1, and the fractal-index growth rate of SW-1 was signifi-
pinching effect is more significant for UCSW-4 because cantly higher than those of the UHTBC shear walls before
there are no connecting bars to resist shear. After the shear the lateral drift ratio reached approximately 1/133. Subse-
strength is overcome and the concrete cracks, the aggregate quently, the fractal-index growth rate was lower than those
interlock contribution is weak and, without the reinforcing of the UHTBC shear walls. The crushing areas of UCSW-2
and
@seismicisolation
@seismicisolationUCSW-3 were significantly reduced compared with

Fig. 8—Schematic of key slots and shear reinforcing bars adopted to improve interface connection between boundary columns
and inner core wall, as opposed to simple friction connection.
Fig. 9—Fractal-index development.

that of SW-1. The small boundary column length specimen Table 5—Crushing zone in final stage
(UCSW-3) exhibited a reduced fractal index and growth rate
Specimen Crushing area, m2 Crushing percentage, %
compared with UCSW-2 but significantly higher values than
SW-1. The development rate of the fractal index following SW-1 0.0532 2.67
the early crack formation of UCSW-3 was higher than that UCSW-2 0.0146 0.73
of SW-1, indicating that most of the cracks in SW-1 formed UCSW-3 0.0113 0.56
earlier than those in UCSW-3. For UCSW-4, the fractal
UCSW-4 0.0609 3.05
index was larger than that of SW-1, and the growth rate
was significantly higher than those of UCSW-3 and SW-1
bars, the UHTBC shear wall (UCSW-3) exhibited better
after the lateral drift ratio exceeded 1/133. Subsequently,
cracking and a higher damage-prevention capacity than
the fractal-index growth rate remained higher than those of
the plain concrete shear wall (SW-1). The weak-connec-
the other specimens. The displacement of SW-1 was smaller
tion specimen (UCSW-4) exhibited a remarkable bending
than those of the other UHTBC specimens when the frac-
behavior and a high crack growth rate in the large-defor-
tal-index growth rate decreased to 0.01, whereas that of
mation stage. Thanks to the high-toughness columns, early
UCSW-4 was significantly larger.
crushing of the weak core wall was prevented. Prior to the
As shown in Fig. 9, the high toughness of the UHTBCs
failure of the corner of the core wall, the UHTBC elements
mitigated the damage concentration and prevented rapid
functioned efficiently.
crack development and local concrete crushing, which
resulted in the constant production of new local fine cracks.
Secant stiffness degradation
The reduction in concrete damage can also be explained by
The secant stiffness in the lateral direction was adopted as
the mitigation of the high toughness of the UHTBC elements.
the characteristic parameter to evaluate the shear-wall stiff-
Despite the smaller edge-confined length and fewer steel
ness, which was defined by Eq. (2) as
@seismicisolation
@seismicisolation

Fig. 10—(a) Secant stiffness; and (b) energy dissipation for single loops.
|Fi|+ |− Fi|
_
Ki = (2) particularly in the large-deformation stage. The energy dissi-
|δi|+ |− δi|
pations of UCSW-2 and SW-1 at a drift ratio of 1/100 were
5.62 and 8.85 kN∙m, respectively, and those at a drift ratio
where Fi represents the forward peak lateral force of the
of 1/50 were 19.12 and 22.73 kN∙m, respectively, as shown
i-th circle; δi represents the corresponding displacement; and
in Fig. 10(b), which were 57% and 19% higher than those of
–Fi and –δi represent the corresponding reverse values. The
SW-1 at drift ratios of 1/100 and 1/50.
variation in the lateral secant stiffness during the loading
The reduced boundary column length specimen (UCSW-3,
sequence is presented in Fig. 10(a). As the lateral load
confined length reduced by 25%) exhibited a higher energy
increased, the secant stiffness of each specimen decreased;
dissipation capacity than the plain concrete shear wall. The
the reduction was initially rapid and subsequently slowed
energy dissipation of UCSW-3 was 55% larger than that of
down. The rapid degradation is attributed to the rapid crack
SW-1 at the drift ratio of 1/100.
development, as indicated by Fig. 11 and 12.
The bending-mode specimen (UCSW-4, with a weak
The lateral secant stiffness values of the UHTBC shear
connection) exhibited better deformation and energy dissi-
walls were significantly higher than that of the plain concrete
pation capacity in the large-deformation stage than the plain
shear wall throughout the loading sequence. The secant stiff-
concrete shear wall. The hysteresis loop of the energy dissi-
ness of UCSW-2 was significantly higher than that of SW-1
pation of the UCSW-4 specimen expanded gradually, even
(by 51%, 14%, and 95% at drift ratios of 1/500, 1/100, and
when the experiment was terminated.
1/50, respectively). This indicates that the fiber-reinforced
When the lateral drift ratio was less than 1/100, the three
boundary column specimen had a significantly higher lateral
UHTBC-strengthened shear walls exhibited higher energy
stiffness than SW-1 owing to the resistance to the crack
dissipation capacity than the plain concrete boundary
development. For UCSW-3, the foregoing increases rela-
column specimens, which can be attributed to the fraction
tive to SW-1 were 53, 27, and 50%; the lateral secant stiff-
and cracking energy dissipated by the fibers. Furthermore,
ness values were also significantly higher than those of the
even with reduced boundary column length and longitu-
plain concrete shear wall. UCSW-4 exhibited a satisfactory
dinal bars, UCSW-3 exhibited enhanced energy dissipation
remaining lateral stiffness, particularly after the drift ratio
compared with the full-length plain concrete specimen, even
reached 1/50. The initial lateral secant stiffness of UCSW-4
close to the “no collapse” limit state. When the connection
was higher than that of SW-1 and marginally higher than
between the boundary column and the core wall was weak-
that of UCSW-3 after the drift ratio reached 1/50. This indi-
ened, although the energy-dissipation value was less than
cates that a higher lateral secant stiffness remained to resist
the well-connected specimen (UCSW-3), it kept increasing
collapse in the case of the UHTBC shear wall with a weak
throughout the loading procedure.
connection between the UHTBC elements and the core wall.
SECTIONAL BEARING CAPACITY CALCULATION
Energy dissipation
The strain development in the lower section (strain gauges
The dissipation capacity of the FRC composites improves
above the foundation of 200 mm) is presented in Fig. 11.
owing to fibers debonding and pullout from the matrix
The horizontal and vertical axes indicate the section length
(Parra-Montesinos 2005). In this study, the improvement of
of the shear wall and the strain values measured by the strain
FRC composites on dissipated energy is also observed for
gauges, respectively. The lateral displacements of 2, 8, 16,
the increasing lateral displacement.
and 32 mm, corresponding to drift ratios of 1/1000, 1/250,
The energy-dissipation results indicated that the UHTBC
1/125, and 1/62.5, respectively, were compared. The strain
elements increased the energy dissipation capacity,
gauge on the one side of UCSW-4 failed to get data.
@seismicisolation
@seismicisolation

Fig. 11—Section strain distribution along shear-wall length.
Fig. 12—Curvature and plastic hinge length, compared with model.
@seismicisolation
@seismicisolation

As indicated by the vertical strain behavior in Fig. 11, at column and inner core wall induced independent deforma-
the peak point state, strain within the confined column on the tion of the two parts, so that the yield curvature is less than
compression side reached the ultimate compressive stress, the cast-together shear walls. Moreover, the curvature distri-
and the longitudinal steel bars yielded. Similar to SW-1, the bution of UVSW-4 follows the typical bending deformation
outside longitudinal steel bar on the tension side reached the mode. At peak point, the shear walls’ base curvature with
ultimate strength, and the longitudinal bars in the compres- UHTBCs is lower than or equal to the model value, while
sive area yielded at the peak point state for the other shear- that of the plain concrete shear wall is significantly larger
wall specimens. The inner vertical strains of the tension-side than the calculated value. The plastic hinge length is concen-
longitudinal bars for UCSW-2 and UCSW-3 were margin- trated at the lower quarter of the wall height for SW-1, while
ally smaller than those of SW-1 at a lateral displacement the plastic zone is not so easily detectable for shear walls
of 32 mm (drift ratio of 1/62.5). The strains of UCSW-4 with UHTBCs. Thus, both curvature distribution and plastic
were larger than those of the other specimens owing to the hinge length prove the beneficial effect provided by the high
complete bending moment behavior. toughness of the boundary columns on the bearing capacity,
along with a significant reduction in damage.
Curvature and plastic hinge
Regardless of the bending-shear coupling effect, the Lateral bearing capacity estimation
flexural displacement can be obtained by integrating the The lateral bearing capacity stems from the series system
sectional curvature along the height of the wall. Bohl and arrangement of the two resisting mechanisms: shear and
Adebar (2011) proposed a plastic hinge model accounting bending.
for the shear aspect ratio. This model assumes a linear curva- According to ACI 318-19 (ACI Committee 318 2019), the
ture distribution before the shear-wall base yields. In the shear capacity is calculated as
post-yield stage, the plastic strain is assumed as uniformly _
distributed within the plastic hinge length. When the shear Vn = (αcλ√ fc + ρtfy)Acv
(5)
wall yields, the curvature is maximum at the wall base and
zero at its top. In this case, the top yield flexural displace- where αc = 2 for elements with an aspect ratio of 2.0; Acv is
ment (Δy) is the section area; λ = 1 for the concrete used in this study;
fc = 0.65fcu = 25.6 MPa is the concrete compressive strength
_1 (0.65 is the reduction coefficient for this case); ρt is the trans-

Δy = 3 φyH2 (3)
verse steel geometric ratio (in this study, ρt = 0.0036); and
fy = 501 MPa is the steel yield strength.
where φy is the yield sectional curvature.
In the Chinese code (GB 50010-2010), the shear capacity
When the lateral force reaches the maximum value, the
is calculated as
plastic hinge length is
lp = 1.25(0.2h + 0.044H) (4)

_
V =
1
λ0− 0 . 5 ( Aw
_
)
0.5ftbw h0 + 0.13NA + fyh
_Ash

s h0 (6)
where H is the wall height; and h is the section depth. where λ0 = 2 is the aspect ratio; ft = 4.5 MPa is the concrete
The curvatures at yield and at the peak lateral force are tensile strength; bw and h0 are the shear-wall section width
shown in Fig. 12, as compared to those calculated according and length; N = 600 kN is the axial force; Aw and A are the
to the model (Bohl and Adebar 2011) and those obtained shear web and entire section area, respectively; and Ash =
from the test results. The experimental yield and peak point 56 mm2, s = 125 mm, and fyh = 501 MPa are the trans-
curvatures are obtained from strain gauges applied on the verse reinforcement area, its vertical spacing, and its yield
outer steel bars along the wall height. The corresponding strength, respectively.
lateral displacements are approximately 14 to 17 mm and According to the aforementioned codes, the average
32 mm (the lateral top drift is 1/62.5), respectively. The yield shear capacity of the four specimens tested in this study is
curvature is calculated when the outer strain reaches the estimated as 895 kN and 445 kN, respectively. It is worth
yield value (0.002), while the peak point curvature is found noticing that these two values are significantly different, thus
when compressive concrete attains its strength according proving the large discrepancies in the shear capacity equa-
to the model. The peak point curvature depends on the tions of different codes (De Domenico et al. 2023). Nonethe-
boundary length, so the values obtained for UCSW-3 and less, because the shear capacity computed with both equa-
-4 are slightly lower than SW-1 and UCSW-2. The plastic tions is higher than the bending capacity exhibited in the
hinge length is obtained by Eq. (4) as 360 mm for all four tests, which is at most 400 kN, this check confirms that the
specimens, as shown in Fig. 12. shear walls were correctly designed to prevent shear failure.
In all four specimens, the yield curvature distribution is in As far as the flexural capacity is concerned, both the
good accordance with the model results for SW-1, UCSW-2, Chinese code (GB 50010-2010) and ACI 318-19 are based
and UCSW-3, which indicates that the model is reasonable on the cross-section equilibrium and stress-strain relation-
for shear walls with reliable shear keys. However, for spec- ship at nominal strength. In both, it is assumed that: 1) the
imen UCSW-4, the yield curvature is lower than the model section strain is linearly distributed; 2) the longitudinal rein-
results. The weak interface connection between the boundary forcing bars in the tension boundary column have yielded;
@seismicisolation
@seismicisolation

Fig. 13—Equilibrium diagram of RC shear-wall section at ultimate limit state.
3) the compressive concrete reaches its ultimate strain; stress along the inner core wall, from the neutral axis to
and 4) concrete tensile stress is neglected. According to the interface with the boundary column. According to GB
the strain-distribution diagram and the damage pattern, the 50011-2010, fcc = 0.88α1α2fcu, α1 = 0.76, and α2 = 1.0 are the
tension area covers an average of 70% of the section length, ratios of uniaxial/cubic strength and the reduction factor to
and the outside steel bars on the tension side reach the account for brittleness. The other terms Fs, Fs′, and fsw′Asw′
ultimate strength at the peak lateral bearing capacity. The are the resultants of the steel bars in the compression and
steel bar beside the outer longitudinal bars on the tension tension boundary columns and in the inner core wall, respec-
side yielded at the peak lateral bearing stage. Accordingly, tively. Given a section configuration, knowing the strength
the strain and stress distribution of the modified model are of concrete and steel bars, lc, can be obtained from Eq. (3)
shown in Fig. 13. The steel bars’ strain distribution still (assuming the compression zone is larger than l0).
meets the plane sections assumption. To find the influence of Because all tested specimens exhibited flexural-controlled
FRC in the boundary regions, two models are proposed by failure and the calculated shear capacity is larger than the
considering the tensile strength of the UHTCC material due tested lateral force, the maximum section moment determines
to the strain-hardening effect, as shown in Fig. 6(a), where fsy the lateral bearing capacity. For the rectangular-section shear
and fsw are the yield strength of the steel bars in the boundary walls, performing the moment equilibrium at the centroid of
columns and the yield strength of the distributed steel bars in the confined compression zone, the ultimate moment stems
the core wall, respectively; As′ and As are the corresponding from the two equilibrium equations
areas of the steel bars; l0 and lc are the boundary element
length and the length of the compression zone of the core N = Fs + Fc = αβ1fccbwl0 + 0.5β2fcbwlc + ΣfsyAs –
wall, respectively; bw is the thickness of the shear wall; fcc, Σfsy′As′ – fsw′Asw′ (8)
fc, and fct are the compressive and tensile strengths of the
UHTCC and plain concrete, respectively; and Fs′ and Fs
are the resultant forces of the longitudinal bars within the M = Ms + Mc + MN = fs′As′(hw0 – l0) + ΣfsyAsls + Σfsw′Asw′lsw′
boundary columns at the tension and compression region,
respectively.
For equilibrium, the conditions ΣF = 0 and ΣM = 0 must
_1
( _lc _
)
l0 _1
+ 2 β2fcbw lc 3 + 2 + 2 N(hw0
− l0 )
(9)
be satisfied, and the vertical force equilibrium can be stated Because the post-cracking effect of FRC shows a signif-
as icant impact on the seismic performance, two models are
discussed and compared with the currently used model. As
N = αβ1fccbwl0 + 0.5β2fcbwlc + Fs – Fs′ – fsw′Asw′ (7) stipulated in the codes, the post-cracking tensile strength of
concrete is not considered in the sectional bearing capacity
where α is the ratio of the peak stress to the concrete (ACI 318-19; GB 50011-2010), and the fiber-reinforcement
strength, equal to 1.0 for concrete strength lower than effect is not considered in Model 1. In Model 2, the boundary
50 MPa; β1 and β2 are the stress block shape factors of columns’ steel bars are assumed to reach the ultimate
compressive concrete, equal to 0.8 for concrete strength strength according to the actual steel bar strain. In Model 3,
lower than 50 MPa, according to GB 50011-2010; the first the post-cracking FRC tensile strength is also accounted for
term αβ1fccbwl0 represents the compression resultant of based on the assumption of Model 2. The descriptions of the
the assumed uniformly distributed concrete stress in the models are listed in Table 6.
boundary column; and the second term 0.5β2fcbwlc represents In Model 2, the section moment is
the compression resultant of the linearly distributed concrete
@seismicisolation
@seismicisolation

Table 6—Lateral bearing capacity prediction model description
Specimens Model 1* Model 2 Model 3
SW-1 The steel bars in boundary areas reached the
The steel bars in boundary areas reached the
UCSW-2 The steel bars in boundary areas yield, and ultimate strength, the tensile strain of inner
ultimate strength, and the tensile strain of
the concrete in compression-confined areas steel bars on the tension side is linearly
UCSW-3 inner steel bars on the tension side is linearly
reached ultimate strength. distributed, and the tension contribution of
distributed.
UCSW-4 the FRC is equivalent to γfct.
*
Post-cracking tensile strength is not considered.
Table 7—Calculated flexural resistance and lateral bearing force

Model 1 Model 2 Model 3 Tested specimens
γ = 0.5/0.75 Maximum Equivalent
lateral force maximum moment
Specimens M, kN∙m Error % M, kN∙m Error % M, kN∙m Error % F, kN M, kN∙m
SW-1 464,425 –13.54 545,255 1.51 — — 268.57 537.14
UCSW-2 447,647 –31.76 541,471 –17.46 585,922/608,147 –10.68/–7.29 328.00 656.00
UCSW-3 446,848 –25.53 524,494 –12.58 559,955/577,686 –6.67/–3.72 300.00 600.00
UCSW-4 446,848 –23.22 524,494 –9.88 559,955/577,686 –3.79/–0.74 291.00 582.00
M = fsu′As′(hw0 – l0) + ΣfsAsls + Σfsw′Asw′lsw′ the reduction factor of the fiber-reinforced concrete of post-
(10) cracking tensile strength is approximately 0.5 and the error
_1
( _lc _
)
l0 _1
+ 2 β2fcbw lc 3 + 2 + 2 N(hw0
− l0 ) is between 3.8 and 10%. When the reduction factor is 0.75,
the error is between 0.7 and 7.3%.
In Model 3, the equilibrium of vertical force gives As the calculation comparison based on different failure
patterns indicates, the tensile strength of plain concrete
N = αβ1fccbwl0 + 0.5β2fcbwlc + ΣfsyAs – Σfsu′As′ shows a minor contribution to the bearing capacity, which
– fsw′Asw′ – γfctbwl0 (11) meets with the current assumption. When the boundary
column concrete is replaced by the UHTCC, the strength of
The section moment is then steel bars in the boundary area and the post-cracking tensile
strength of UHTCC contribute significantly to the lateral

M = fsu′A
s′h w0
_l0
(
−2 + ∑fsA )
sls+ ∑fsw′A
sw′lsw′
(12)
bearing capacity. With the fiber confinement, the edge longi-
tudinal steel bars are more stable against buckling. Conse-
+ γfct bw l0 ( hw0
_1
( _
)
lc _
l0 _1
− l0 ) + 2 β2fcbw lc 3 + 2 + 2 N( hw0
− l0 ) quently, the boundary column steel bars are more likely to
reach the ultimate strength.
where fs and fsu′ are the yield and ultimate strengths of longi- DISCUSSION
tudinal steel bars in the boundary area on the tension and Strengthening effect of UHTBC
compression sides; As′ and As are the area of the boundary-area The experimental results indicated that replacing plain
steel bars on the tension and compression side, respectively; concrete in the boundary columns with the UHTCC can
ls′ is the distance from the rotation center; γ is the reduction significantly improve the seismic performance of shear
factor of the tensile strength of FRC in the confined boundary walls.
zone; and Σfsw′Asw′lsw′ represents the moment caused by The UHTBC used in this study demonstrates its effective-
distribution bars, which can be calculated according to the ness for strengthening the boundary elements of shear walls
plane sections assumption. by mitigating the local cracks, reducing their configuration,
Table 7 shows the calculated maximum section moment and preventing rapid crack development and local concrete
through Models 1 to 3. The sectional bearing moment and crushing. The shear walls with UHTBC elements exhibited
the corresponding lateral bearing force are also listed for improved lateral resistances and deformation capacities. The
comparison. In Model 3, the reduction factor γ of the tensile peak lateral force and the corresponding lateral displace-
strength of FRC in the confined boundary region is set as 0.5 ment of UCSW-2 were 22% and 19% larger, respectively,
and 0.75, separately. than those of SW-1. The UHTBC specimens maintained a
As Table 7 shows, the maximum lateral bearing capacity good resistance capacity even when the lateral bearing force
obtained from Model 1 is acceptable for plain concrete shear dropped to 70% of the peak value. The secant stiffness of
walls, but the error is 23 to 32% for the FRC-RC shear walls. UCSW-2 was significantly higher than that of SW-1 (by
When considering the ultimate strength of the outer-edge 51%, 13%, and 95% at drift ratios of 1/500, 1/100, and 1/50,
steel bars, the error is lower yet still 10 to 17%. In Model 3, respectively). The UHTBC elements increased the energy
considering the concrete tensile strength, for the two kinds dissipation capacity, particularly in the large-deformation
of confined column shear walls, the error is significantly stage. The energy dissipations of the single hysteresis loops
reduced. As the literature suggested (Zhang et al. @seismicisolation
2022),
@seismicisolation

for UCSW-2 were 57% and 19% higher than those of SW-1 significant during the strong shaking portion of the action in
at drift ratios of 1/100 and 1/50, respectively. In these cases, near-fault earthquakes.
it provides an alternate strengthening method for shear walls
not conforming to the seismic design code, leading to better CONCLUSIONS
seismic performance than that of walls designed according The seismic performance of nonconforming slender
to the current design code. shear walls strengthened by ultra-high-toughness boundary
columns (UHTBCs) was experimentally validated in this
Possibility of reducing boundary column size and study, and the main conclusions are as follows.
longitudinal bars 1. The shear walls strengthened by UHTBCs possess
Comparing UCSW-3 with SW-1 reveals that the shear wall improved bearing and deformation capacity. Compared
with a reduced boundary column configuration still exhib- to the plain concrete shear wall designed according to the
ited better seismic performance than the shear wall designed current code, the secant stiffness of the UHTBC-strengthened
according to the current standards. The lateral force and shear walls was higher by 51%, 13%, and 95% at drift ratios
displacement of UCSW-3 in the peak state were increased of 1/500, 1/100, and 1/50, respectively,
by 12% and 13%, respectively, and the strength degradation 2. The shear walls with reduced-boundary UHTBCs
of post-peak lateral force was significantly smaller. These conform with the bearing capacity of the specimen designed
characteristics contribute to the possibility of strengthening to the current specifications. The lateral force and displace-
the existing shear walls with limited space. ment of the reduced-boundary shear wall with UHTBCs at
the peak state were higher by 12% and 13%, respectively,
Feasibility of no-connection UHTBC compared to the standard specimen.
boundary elements 3. The slender shear wall strengthened by reduced no-
As the experimental results of UCSW-4 show, the shear connection UHTBCs exhibited a less disruptive interaction
wall strengthened with the UHTBC boundary elements with the existing shear wall, thus preventing its shear failure
satisfied the seismic demand of the current seismic design and at the same time maintaining an appreciable bearing
code GB 50011-2010. The non-connection UHTBC spec- capacity even until the “collapse prevention” state. The
imen (UCSW-4) had better deformation and energy dissi- peak lateral force and displacement were increased by 8%
pation capacity in the large-deformation stage than the plain and 20%, and the lateral force at a drift ratio of 1/50 was
concrete shear wall. The lateral force and displacement in increased by 34%, respectively, compared to the specimen
the peak state were increased by 8.4% and 20%, and the designed to the current code.
lateral force at a drift ratio of 1/50 was increased by 34%, 4. The feasible prediction equation for the lateral bearing
respectively. The strength degradation was significantly capacity of the UHTBC-strengthened shear walls is obtained
slower than that for the other specimens. The hysteresis loop by accounting for the tensile strength through a reduction
of the energy dissipation for UCSW-4 expanded gradually, factor of 0.75 and the ultimate strength of the boundary steel
even close to specimen failure. The test results indicate the bars.
feasibility of shear-wall retrofitting without connecting with This technique can be used both in noninvasive and
the existing structure, through the construction of an addi- no-damage retrofitting measures to improve the seismic
tional boundary element. That provides evidence for the performance of shear walls, particularly in severe
feasibility of the proposed method being applied in the rapid earthquake-prone areas, even using reduced-boundary
replacement of the damaged boundary columns or retrofit- UHTBCs, which ensure ample application feasibility for
ting protected structures. Additionally, the non-connection existing shear-wall structures.
boundary element shear wall exhibits a flexural failure
pattern of the three separate parts. AUTHOR BIOS
Hongmei Zhang is a Professor at Zhejiang University, Hang-
zhou, Zhejiang, China. Before that, she was an Associate Professor
Section design model at Tongji University, Shanghai, China. From 2014 to 2015, she
In contrast to the code-specified model, the models was a Visiting Professor at the University of Illinois at Urbana-
considering the ultimate strength of the edge steel bars on Champaign, Urbana, IL. She received her PhD in civil engineering from
Tongji University in 2007. Her research interests include disaster reduction
the tension side and the post-cracking tensile strength of the and seismic design of reinforced concrete structures.
fiber-reinforced concrete provide more reasonable prediction
results. The error of the predicted lateral bearing capacity to Giorgio Monti is a Full Professor at Sapienza University of Rome, Rome,
Italy, and a Visiting Professor at Zhejiang University. He graduated in
the test results is within 7% using the post-cracking tensile civil engineering from Sapienza University of Rome in 1986; received his
strength reduction factor γ = 0.75. Moreover, the steel bars Master of Science from the University of California, Berkeley, Berkeley,
at the boundary area of the tension side were more likely CA, in 1993; and received his PhD in structural engineering from Sapienza
University of Rome in 1994.
to reach the ultimate strength and contribute to the shear-
wall lateral bearing capacity, thanks to the fiber-reinforced Yuanfeng Duan is a Full Professor at Zhejiang University. From 2014 to
concrete protection. 2015, he was a Visiting Professor at the University of Illinois at Urbana-
Champaign. He received his PhD in civil engineering from The Hong Kong
Moreover, the energy dissipated by fiber pullout and Polytechnic University, Kowloon, Hong Kong, in 2004. His research inter-
debonding only takes place with increasing deformation ests include disaster reduction for structures.
amplitude. Thus, the contribution from UHTCC is especially
@seismicisolation
@seismicisolation

Zhiyuan Chen is a Graduate of Zhejiang University. He received his BS Li, M.; Luu, H. C.; Wu, C.; Mo, Y. L.; and Hsu, T. T. C., 2014, “Seismic
in civil engineering from the Zhejiang University of Technology, Hangzhou, Performance of Reinforced Engineered Cementitious Composite Shear
Zhejiang, China, in 2019. Walls,” Earthquakes and Structures, V. 7, No. 5, pp. 691-704. doi:
10.12989/eas.2014.7.5.691
Chen Gu is a Graduate of Zhejiang University. He received his BS in civil Li, Q. H., and Xu, S. L., 2009, “Experimental Investigation and
engineering from Henan Polytechnic University, Jiaozuo, Henan, China, Analysis on Flexural Performance of Functionally Graded Composite
in 2020. Beam Crack-Controlled by Ultrahigh Toughness Cementitious Compos-
ites,” Science in China Series E: Technological Sciences, V. 52, No. 6, June,
ACKNOWLEDGMENTS pp. 1648-1664. doi: 10.1007/s11431-009-0161-x
This project was financially supported by the National Natural Science Li, V. C.; Wang, S.; and Wu, C., 2001, “Tensile Strain-Hardening
Foundation of China (Grant No. 52078459 and U1709216), the National Behavior of Polyvinyl Alcohol Engineered Cementitious Composite (PVA-
Key R&D Program of China (2019YFE0112600), and the National Natural ECC),” ACI Materials Journal, V. 98, No. 6, Nov.-Dec., pp. 483-492.
Science Foundation of Zhejiang Province (Grant No. LZ22E080005). The Liang, X.-W.; Zheng, Y.; Deng, M.-K.; Kou, J.-L.; and Che, J.-L., 2013,
experimental data that support the findings of this study are available from “An Investigation of Deformation Behavior of the Shear Wall with Fiber-
the corresponding author upon reasonable request. Reinforced Concrete in Plastic Hinge Region,” Engineering Mechanics,
V. 30, No. 3, Mar., pp. 256-262.
Lu, X.; Zhang, Y.; Zhang, H.; Zhang, H.; and Xiao, R., 2018, “Exper-
REFERENCES imental Study on Seismic Performance of Steel Fiber Reinforced High
ACI Committee 318, 2019, “Building Code Requirements for Structural Strength Concrete Composite Shear Walls with Different Steel Fiber
Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved Volume Fractions,” Engineering Structures, V. 171, Sept., pp. 247-259. doi:
2022),” American Concrete Institute, Farmington Hills, MI, 624 pp. 10.1016/j.engstruct.2018.05.068
Al-Gemeel, A. N., and Zhuge, Y., 2019, “Using Textile Reinforced Mandelbrot, B. B., 1982, The Fractal Geometry of Nature, W. H.
Engineered Cementitious Composite for Concrete Columns Confine- Freeman and Company, New York City, NY.
ment,” Composite Structures, V. 210, Feb., pp. 695-706. doi: 10.1016/j. Mirmiran, A., and Shahawy, M., 1997, “Behavior of Concrete
compstruct.2018.11.093 Columns Confined by Fiber Composites,” Journal of Structural Engi-
Bohl, A., and Adebar, P., 2011, “Plastic Hinge Lengths in High-Rise neering, ASCE, V. 123, No. 5, May, pp. 583-590. doi: 10.1061/
Concrete Shear Walls,” ACI Structural Journal, V. 108, No. 2, Mar.-Apr., (ASCE)0733-9445(1997)123:5(583)
pp. 148-157. Mustafaraj, E.; Yardim, Y.; Corradi, M.; and Borri, A., 2020, “Polypro-
Boshoff, W. P.; Mechtcherine, V.; and van Zijl, G. P. A. G., 2009, pylene as a Retrofitting Material for Shear Walls,” Materials, V. 13, No. 11,
“Characterizing the Time-Dependent Behaviour on the Single Fibre June, Article No. 2503. doi: 10.3390/ma13112503
Level of SHCC: Part 1: Mechanism of Fibre Pull-Out Creep,” Cement Mutō, K., 1969, Newly-Devised Reinforced Concrete Shear Walls for
and Concrete Research, V. 39, No. 9, Sept., pp. 779-786. doi: 10.1016/j. High-Rise Building Structures, Muto Institute of Structural Mechanics,
cemconres.2009.06.007 Tokyo, Japan, 38 pp.
Chrysanidis, T., and Tegos, I., 2020, “Axial and Transverse Strength- Naaman, A. E., and Najm, H., 1991, “Bond-Slip Mechanisms of Steel
ening of R/C Circular Columns: Conventional and New Type of Steel and Fibers in Concrete,” ACI Materials Journal, V. 88, No. 2, Mar.-Apr.,
Hybrid Jackets Using High-Strength Mortar,” Journal of Building Engi- pp. 135-145.
neering, V. 30, July, Article No. 101236. doi: 10.1016/j.jobe.2020.101236 Olsen, E. C., and Billington, S. L., 2011, “Cyclic Response of Precast
Dang, Z.; Liang, X. W.; Deng, M. K.; Li, F.; and Yu, J., 2014, “Exper- High-Performance Fiber-Reinforced Concrete Infill Panels,” ACI Struc-
imental and Theoretical Studies on Seismic Behavior of Fiber Reinforced tural Journal, V. 108, No. 1, Jan.-Feb., pp. 51-60.
Concrete Shear Walls,” Journal of Building Structures, V. 35, No. 6, Park, K.-G., 2006, “Design and Analysis on the Connections of RC
pp. 12-22. Precast Large Panel,” Journal of Korean Association for Spatial Structures,
Dazio, A.; Buzzini, D.; and Trüb, M., 2008, “Nonlinear Cyclic Behaviour V. 6, No. 2, Serial No. 20, pp. 85-92.
of Hybrid Fibre Concrete Structural Walls,” Engineering Structures, V. 30, Parra-Montesinos, G. J., 2005, “High-Performance Fiber-Reinforced
No. 11, Nov., pp. 3141-3150. doi: 10.1016/j.engstruct.2008.03.018 Cement Composites: An Alternative for Seismic Design of Structures,” ACI
De Domenico, D.; Quaranta, G.; Zeng, Q.; and Monti, G., 2023, Structural Journal, V. 102, No. 5, Sept.-Oct., pp. 668-675.
“Shear Capacity of RC Elements with Transverse Reinforcement through Parra-Montesinos, G. J.; Canbolat, B. A.; and Jeyaraman, G. R., 2006,
a Variable-Angle Truss Model with Machine-Learning-Calibrated Coeffi- “Relaxation of Confinement Reinforcement Requirements in Structural
cients,” Artificial Intelligence and Machine Learning Techniques for Civil Walls through the Use of Fiber Reinforced Cement Composites,” Proceed-
Engineering, V. Plevris, A. Ahmad, and N. D. Lagaros, eds., IGI Global, ings of the 8th U.S. National Conference on Earthquake Engineering, San
Hershey, PA. Francisco, CA, pp. 4045-4054.
Fischer, G., and Li, V. C., 2002a, “Effect of Matrix Ductility on Deforma- Prota, A.; Nanni, A.; Manfredi, G.; and Cosenza, E., 2004, “Selective
tion Behavior of Steel-Reinforced ECC Flexural Members under Reversed Upgrade of Underdesigned Reinforced Concrete Beam-Column Joints
Cyclic Loading Conditions,” ACI Structural Journal, V. 99, No. 6, Using Carbon Fiber-Reinforced Polymers,” ACI Structural Journal, V. 101,
Nov.-Dec., pp. 781-790. No. 5, Sept.-Oct., pp. 699-707.
Fischer, G., and Li, V. C., 2002b, “Influence of Matrix Ductility on Sørensen, J. H.; Hoang, L. C.; Olesen, J. F.; and Fischer, G., 2017a,
Tension-Stiffening Behavior of Steel Reinforced Engineered Cementi- “Tensile Capacity of Loop Connections Grouted with Concrete or Mortar,”
tious Composites (ECC),” ACI Structural Journal, V. 99, No. 1, Jan.-Feb., Magazine of Concrete Research, V. 69, No. 17, Sept., pp. 892-904. doi:
pp. 104-111. 10.1680/jmacr.16.00466
Fischer, G., and Li, V. C., 2003, “Deformation Behavior of Fiber- Sørensen, J. H.; Hoang, L. C.; Olesen, J. F.; and Fischer, G., 2017b, “Test
Reinforced Polymer Reinforced Engineered Cementitious Composite and Analysis of a New Ductile Shear Connection Design for RC Shear
(ECC) Flexural Members under Reversed Cyclic Loading Conditions,” ACI Walls,” Structural Concrete, V. 18, No. 1, Feb., pp. 189-204. doi: 10.1002/
Structural Journal, V. 100, No. 1, Jan.-Feb., pp. 25-35. suco.201600056
GB 50011-2010, 2010, “Code for Seismic Design of Buildings,” Ministry Woods, J. E.; Lau, D. T.; and Cruz-Noguez, C. A., 2016, “In-Plane
of Housing and Urban-Rural Development of the People’s Republic of Seismic Strengthening of Nonductile Reinforced Concrete Shear Walls
China, Beijing, China. Using Externally Bonded CFRP Sheets,” Journal of Composites for
JGJ/T 101-2015, 2015, “Specification for Seismic Test of Buildings,” Construction, ASCE, V. 20, No. 6, Dec., p. 04016052. doi: 10.1061/(ASCE)
Ministry of Housing and Urban-Rural Development of the People’s CC.1943-5614.0000705
Republic of China, Beijing, China. Xu, S.-L., and Cai, X.-R., 2010, “Experimental Study and Theoretical
JG/T 398-2012, 2012, “The Grouting Coupler for Rebars Splicing,” Models on Compressive Properties of Ultrahigh Toughness Cementitious
Ministry of Housing and Urban-Rural Development of the People’s Composites,” Journal of Materials in Civil Engineering, ASCE, V. 22,
Republic of China, Beijing, China. No. 10, Oct., pp. 1067-1077. doi: 10.1061/(ASCE)MT.1943-5533.0000109
Kesner, K., and Billington, S. L., 2005, “Investigation of Infill Panels Xu, S. L.; Wang, N.; and Zhang, X. F., 2012, “Flexural Behavior of
Made from Engineered Cementitious Composites for Seismic Strengthening Plain Concrete Beams Strengthened with Ultra High Toughness Cemen-
and Retrofit,” Journal of Structural Engineering, ASCE, V. 131, No. 11, titious Composites Layer,” Materials and Structures, V. 45, No. 6, June,
Nov., pp. 1712-1720. doi: 10.1061/(ASCE)0733-9445(2005)131:11(1712) pp. 851-859. doi: 10.1617/s11527-011-9803-0
Li, H.; Xu, S.; and Leung, C. K. Y., 2009, “Tensile and Flexural Proper- Zhang, H.; Zhang, Y.; Lu, X.; Duan, Y.; and Zhang, H., 2020, “Influ-
ties of Ultra High Toughness Cementitious Composite,” Journal of Wuhan ence of Axial Load Ratio on the Seismic Behavior of Steel Fiber–
University of Technology-Materials Science Edition, V. 24, No. 4, Aug., Reinforced Concrete Composite Shear Walls,” Journal of Structural
pp. 677-683. doi: 10.1007/s11595-009-4677-5 @seismicisolation
@seismicisolation

Engineering, ASCE, V. 146, No. 1, Jan., p. 04019171. doi: 10.1061/(ASCE) High-Strength-Concrete Composite Shear Walls,” Structures, V. 35, Jan.,
ST.1943-541X.0002444 pp. 765-779. doi: 10.1016/j.istruc.2021.11.038
Zhang, H. M.; Lu, X. L.; Duan, Y. F.; and Zhu, Y., 2014, “Experi- Zhou, Y.; Lu, X.; and Dong, Y., 2010, “Seismic Behavior of Composite
mental Study on Failure Mechanism of RC Walls with Different Boundary Shear Walls with Multi-Embedded Steel Sections. Part I: Experiment,” The
Elements under Vertical and Lateral Loads,” Advances in Structural Engi- Structural Design of Tall and Special Buildings, V. 19, No. 6, pp. 618-636.
neering, V. 17, No. 3, Mar., pp. 361-379. doi: 10.1260/1369-4332.17.3.361 doi: 10.1002/tal.598
Zhang, Y.; Zhang, H.; and Lu, X., 2022, “Seismic Performance
Evaluation and Experimental Validation of Steel-Fiber-Reinforced
@seismicisolation
@seismicisolation

Title No. 120-S79
Hysteretic Model of Coupler Box Assembly for Seismic

Retrofitting of Severely Damaged Reinforced Concrete
Buildings
by Naveen Kumar Kothapalli, R. Siva Chidambaram, and Pankaj Agarwal
Cyclic tests are conducted on interlinked reinforcing bar resisting large bending moments, which may lead to bond-
coupler-box assemblies, adopted to retrofit buckled reinforcing slip failure. Welding can be an alternative technique for
bars at the plastic hinge locations of columns in multi-storied connecting broken reinforcing bars with new bars. However,
reinforced concrete building frames. The efficacy of the proposed it also requires removing a large amount of intact concrete
retrofitting technique is evaluated by comparing the hysteresis
on both sides of the damaged section to expose the required
behavior, computed parameters of performance index, and failure
weld length and an all-around clear space for functioning.
mechanism of the reconstructed frame with the original frame. An
energy-based strength deterioration hysteresis model is developed The length of lap weld for connecting reinforcing bars as per
on the basis of cyclic test results for analytically computing the strength, and codal requirements, IS 9417:2018 (2018), is
post-yield behavior of retrofitted reinforced concrete (RC) frame approximately 600 mm (23.62 in.), which is higher than the
with the proposed coupler-box assembly. The experimental test concrete removal area required for insertion of the proposed
results manifest that the coupler-box assembly can be a promising coupler sleeve, 500 mm (19.69 in.)—that is, two times the
futuristic approach for seismic retrofitting of severely damaged proposed coupler sleeve length. Further, welding causes
reinforced concrete buildings, where buckling of longitudinal rein- metallurgical changes with the recrystallization of micro-
forcing bars at the plastic hinge location of columns is inevitable, structures in particles and is not preferred in the potential
and the process of restoration is challenging under existing gravity hinge region of primary structural members. The study on
loads. The suggested retrofitting mechanism restrains the section
mechanical properties of welded deformed reinforcing steel
from any movement against rotation and helps in shifting the yield
bars conducted by Ahmed (2015) shows that the strength
location of reinforcing bars. The main advantage of adopting the
coupler-box is that there is no observed slip of reinforcing bar from and elongation of welded reinforcing bars decrease up to
the sleeve, and the entire retrofitted section remains intact even 40% and 60%, respectively.
after a lateral storey drift of 6%, which is larger than the collapse The use of reinforcing bar couplers or coupling sleeves
prevention drift level of 4% as per Federal Emergency Manage- in joining the ends of reinforcing bar segments is a preva-
ment Agency guidelines. lent practice in the modernized construction of buildings or
bridges, which safeguards the reinforcing materials at their
Keywords: coupler-box assembly; deterioration strength hysteresis model; lap-splice regions. The invention of grouted splice sleeves
energy dissipation; hysteresis behavior; reinforcing bar coupler sleeve;
seismic retrofitting.
dates back to the late 1960s by Yee (1970, 1986, 2009).
Several studies in the past on coupler/coupling sleeves are
INTRODUCTION based on their shape, diameter, length, material character-
Buckling/yielding of longitudinal reinforcing bars at the istic, bonding properties of the grout material, anchorage
plastic hinge locations of multi-storied building frames under capacity, and so on. Hybrid couplers with a combination
severe earthquake excitations is one of the most common of mechanical and grouting techniques were tested by
failure modes. As a result, the column member shortens Hope (1987). Threaded end reinforcing bars into grout-
and loses its capability to carry even its prescribed gravity filled coupler sleeves were suggested by Lancelot (1995).
load, which sometimes leads to the progressive collapse of Mechanical couplers in the form of tubular members with
buildings. Seismic retrofitting of buckled reinforcement at and without internal threading, along with a distinctive
the hinge location of columns in such multi-story buildings number of external bolts, were proposed by Holdsworth
proves to be challenging as there is no mechanism to realign (1997, 2006). Annular-shaped seals at the ends of steel
the buckled bars under high compressive loads. Under these sleeves for locking reinforcing bar elements along with
circumstances, removal of the buckled portion of rein- grout matter were recommended by Mochizuki and Nihei
forcing bar is inevitable, and new reinforcement is intro- (1998). High-strength concrete grout in addition to large-
duced either by welding, splicing, or through mechanical headed reinforcing bars through pipe couplers was suggested
couplers. The conventional method of lap splicing at hinge
ACI Structural Journal, V. 120, No. 5, September 2023.
locations is neither recommended nor practical as it requires MS No. S-2021-120.R7, doi: 10.14359/51738831, received March 4, 2023, and
large demolition of intact concrete in an existing structural reviewed under Institute publication policies. Copyright © 2023, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
member on either side of the cut reinforcement. Moreover, 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
splicing reinforcing bars at hinge locations is incompetent in is received within four months of the paper’s print publication.
@seismicisolation
@seismicisolation

by Dahl (2001). Taper-threaded reinforcing bar ends were damaged precast RC bridge columns were repaired by Parks
associated with threaded collars and linked with a central et al. (2016) using grouted splice sleeves and composite
circular sleeve in the research study by Allen et al. (2007). shells, which relocate the column’s plastic hinge.
The various failure modes in sleeve connections under The use of mechanical coupling devices for integrating
direct tensile load were evaluated and analysed by Hua et reinforcing bar segments in modern and retrofitting solu-
al. (2008). The cyclic behavior of beam-column joint sub- tions has been extensively reviewed. Hybrid couplers with
assemblies with spiral threaded couplers at the joint’s inter- external bolting and grouting are comparatively effective.
face was investigated by Ingham and Bai (2009). Rowell and Dimensions like development length and thickness of the
Hager (2010) studied the dynamic performance of various coupler sleeve also decide the failure pattern in structural
sleeves under higher strain rates, where the threaded rein- members. In practice, the commonly used standard couplers
forcing bar coupler system performs best with 99% ultimate are threaded, where the ends of reinforcing bars are mechan-
dynamic tensile strength. Damaged columns were retrofitted ically threaded and connected. Threaded couplers are gener-
by Huaco and Jirsa (2012), where the buckled reinforcing ally shorter in length (approximately 2 to 2.5 times the
bars were replaced and reconnected through mechanical diameter of connected reinforcing bar) and are used in new
sleeves with high-strength bolts torqued to a prescribed constructions in a staggered pattern. Their performance in
level. The cyclic test on reinforcing bars connected through bonding reinforcing bar segments is adequate under axial
the mechanical sleeves shows that the rupture of reinforcing loads, but is uncertain under flexural loads. In the process
bar is at its last bolt in shorter sleeves and is distant away of retrofitting damaged structures, the existing reinforcing
from the sleeve in longer ones. Column bridge bents with bars yield at a common section. Further, threading of the
couplers attached to their longitudinal reinforcing bars at embedded reinforcing bars is not practically feasible as it
staggered heights in the plastic hinge zone were tested by requires complete processing of reinforcing bar ends, such as
Phillippi and Hegemier (2013), where lower compressive enlargement of diameter and threading. Therefore, threaded
stress is experienced in the couplers outside the plastic hinge couplers are suitable for new constructions and are unfit or
zone. The effect of sleeve diameter and development length ineffective in retrofitting projects, particularly in the plastic
on the bonding performance of connectors was studied by hinge region of severely damaged structures.
Alias et al. (2014), where the connectors with an optimum The use of reinforcing bar coupling devices in an existing
development length of 200 mm (7.87 in.) along with smaller structural component, as seen in Fig. 1—that is, specifically
internal and external diameters provided a better bonding at the flexure-dominated plastic hinge regions—becomes
performance. Research on the use of coupler sleeves to join a complicated task as threading the existing reinforcing
precast elements like columns has been conducted in the bars is impractical. The present study focuses on a unique
past. The lateral load behavior of precast columns joined by coupler-box assembly which not only connects the rupture
corrugated steel sleeves was studied by Popa et al. (2015). A ends of reinforcing bars at the severely damaged hinge
sleeve-connected column had a similar hysteretic response locations of columns, but also significantly enhances the
with energy dissipation capacity as a cast-in-place column. performance of a standalone coupling sleeve. A coupler-box
A design methodology to increase the plastic rotation confinement technique is proposed in the present study,
capacity and ductility of precast columns by shifting their where standalone couplers are integrated through flat steel
plastic hinge locations above the grouted sleeves, using tran- bars to make a box section that helps resist high moment
sition splicing and high-strength steel dowels in the connec- shear along with torsional forces and confines the frame
tion footings, was proposed by Haber et al. (2017). Grouted section to withstand larger plastic deformations. The
splice sleeve connectors were adopted to connect precast proposed coupling sleeves used in box formation are inter-
columns with their footings at various locations by Ameli nally grooved, externally bolted, and grouted using high-
et al. (2016). A column with a grouted sleeve in the plastic strength epoxy mortar grout to avoid any bond-slip failure.
hinge region performs better than a specimen with the sleeve Experimental verification on the efficiency of coupler-box
located inside its footing. A novel methodology for precast association is conducted by testing a full-scale RC portal
erection of columns with one-touch couplers guided into frame. Comparative performance of standalone coupler and
corresponding column reinforcing bars through mild steel coupler-box assemblage is evaluated by performing a cyclic
plates was proposed by Nzabonimpa and Hong (2018). test on RC columns. The obtained cyclic test results of portal
Inelastic behavior in reinforced concrete (RC) members frame pave the way for development of energy-based hyster-
connected through compact threaded and slender threaded esis models, which can be used in the numerical modeling
sleeves was studied by Bompa and Elghazouli (2019). It of retrofitted structures with coupler-box assemblies and can
was noticed that a slender coupler alters the plastic hinge possibly be a vital contribution to the field of structural and
behavior by localising curvatures and reducing the rotational retrofitting engineering.
capacity, whereas the compact couplers perform identical to
specimens with continuous reinforcement. Bridge columns RESEARCH SIGNIFICANCE
with interlocking spirals and fractured longitudinal rein- Retrofitting or restoring damaged RC frame structures
forcing bars were repaired by mechanical splicing of rein- with buckled/yielded reinforcement proves complex under
forcing bars in the plastic hinge locations and external jack- existing load conditions. The proposed research work
eting by Yang et al. (2015), which is efficient in restoring provides a feasible solution to overcome this issue, where the
the lateral and torsional strength of the column. Severely recommended coupler sleeve slides freely over the existing
@seismicisolation
@seismicisolation

Fig. 1—Buckling failure of longitudinal reinforcement noticed during earthquake loading.
reinforcing bar without any cross-sectional modification epoxy grout and external bolt, and the computed strength
and is further interlinked with flat steel bars to form a should be greater than the yield strength of connected rein-
box-shaped assembly. The adaptability of the proposed tech- forcing bars. A factor of 1.25 is applied in Eq. (3), which
nique is verified by experimental testing, and energy-based considers the strain-hardening effect in connected rein-
hysteresis models are proposed for analytical modeling, forcing bars—that is, the UTS/YS ratio. The total number of
which is believed to be a significant contribution to the field bolts required to fix the coupler system is calculated based
of retrofitting. on the shear strength of each bolt, and the combined strength
for clamping each reinforcing bar segment should be greater
UNIAXIAL TENSILE BEHAVIOR OF PROPOSED than the ultimate tensile strength of the reinforcing bar. The
STANDALONE COUPLER SLEEVE additional sleeve bolts are provided for the interlinking
The proposed coupler/sleeve is particularly preferred in the purpose and their orientation is kept perpendicular to one
field of retrofitting buckled longitudinal reinforcing bars in another. This establishes a connection with adjoining sleeves
damaged columns of RC buildings subjected to severe earth- which are positioned at right angles to each other. The esti-
quake excitations. Under these circumstances, removing the mated geometrical dimensions for a proposed coupler sleeve
buckled portion of reinforcing bars is inevitable for retrofit- are calculated based on the reinforcing bar diameter (Ø)
ting the frame sections. The present study focuses on a prob- using the following simplified empirical equations.
able solution to connect two separated ends of reinforcing The inner bore diameter of coupling sleeve (Øinner) is
bars at the plastic hinge location of existing columns, and
their efficiency is evaluated under extreme forces or defor- Øinner = Ø + 5 mm (1)
mations that may arise in case of impending earthquakes.
The proposed coupler sleeve grips the connected rein- Thickness of the sleeve or wall thickness (t) is
forcing bar elements with external bolts and high-strength
epoxy grout, as shown in Fig. 2. The filler grout used to seal t(mm) = Ø(mm)/2 (2)
void spaces between reinforcing bar and sleeve is a dual-
component epoxy-based mortar with a mixture proportion Length of the coupler sleeve (L) is
of 3:1 (epoxy resin:amine hardener). The mixed resin main-
tains a viscosity of 23 Pa-s (0.0033 lbf-s/in.2), bond strength L(in. cm) = 1.25 × Ø(in. mm) (3)
of 16.4 MPa (2.38 ksi), compressive strength of 102 MPa
(14.79 ksi), and tensile strength of 45 MPa (6.53 ksi). The The diameter of the external bolt (Øbolt) is computed
coagulated resin mixture is injected against gravity through as Øbolt = Ø/2 with a standard center-to-center spacing of
one of the bottom bolt-holes of the coupler sleeve. To allow 20 mm (0.79 in.), and an estimated number of six bolts are
the free flow of resin and disseminate it around the rein- clamped over each reinforcing bar segment—that is, half-
forcing bar surface, the inner bore is provided with 5 mm length of the coupler sleeve.
(0.19 in.) additional gap than the reinforcing bar diameter, In the present study, a coupler sleeve with a reinforcing
which provides an annular gap of 2.5 mm (0.09 in.) around bar of 20 mm (0.79 in.) diameter is considered to retrofit the
the reinforcing bar surface. Further, this gap allows the free plastic hinge locations of damaged columns in a building
insertion of the sleeve onto the reinforcing bar segment. frame. It is specifically designed using a hollow circular
The reinforcing bars at connected plastic hinge locations mild steel pipe of 45 mm (1.77 in.) external diameter with
are assumed to be ineffective, and to maintain an equiva- inner walls grooved at a rate of 10 grooves per in. length.
lent shear area in the hinge location of structural member, Twelve numbers of 10 mm (0.39 in.) diameter high-strength
the thickness of sleeve or wall thickness is considered to 12.9 graded Allen bolts are used to clasp the reinforcing bar
be half the diameter of reinforcing bar. The bond strength elements against the sleeve walls, and four numbers of 8 mm
required for each coupled reinforcing bar segment is the (0.32 in.) diameter bolts are used to interconnect them later-
primary design criteria to determine the length of sleeves ally. The grooved surface/lining at the inner walls provides
and the diameter of external bolts. The bonding strength additional frictional resistance between the reinforcing bar
over each coupled reinforcing bar is a cumulative effect of and coupler sleeve. Photographs of the proposed coupler
@seismicisolation
@seismicisolation

Fig. 2—Typical details of proposed coupler sleeve.
sleeve, along with its bolting and reinforcing bar intercon- behavioral plot of coupled reinforcing bar specimen based
nection, are shown in Fig. 2. on the applied machine stroke. Low-cycle fatigue test of
Uniaxial direct tension testing on reinforcing bars with 10,000 cycles with stress variations from +173 to –173 MPa
and without coupler connection is carried out to examine (+25.1 to –25.1 ksi) and a high-cycle fatigue test of 2 million
the performance of proposed coupler sleeve, as shown in stress cycles with variations from 270 to 330 MPa (39.2
Fig. 3(a). There is no failure noticed in the coupled region to 47.9 ksi) are performed on the proposed coupler sleeve
of specimen, and yielding of the reinforcing bar gets shifted connection. Post completion of 10,000 cycles, no failure is
with a rupture pattern similar to that of the intact reinforcing noticed in the sleeve connection, and the load is statically
bar. The failure in coupler-connected reinforcing bars occurs increased until failure, as shown in Fig. 3(d). Similarly, after
at a comparatively low axial deformation than the original the completion of 2 million cycles, the sleeve connection
reinforcing bar. However, the computed stress-strain param- remains intact and is monotonically loaded until failure, as
eters of coupled reinforcing bars qualify the minimum codal shown in Fig. 3(e). In the high-cycle fatigue test, the axial
specifications of high-strength reinforcing bars for concrete stress reduces post the ultimate strength of the reinforcing
reinforcement, as per IS 1786:2008 (2008). The post-yield bar—that is, post 685 MPa (99.3 ksi), which is by the prob-
behavior in coupler-connected reinforcing bar is influenced able slip in connected reinforcing bar. However, the coupled
by the number of external bolts and the strength of epoxy- reinforcing bar achieves the minimum required reinforcing
grout material. The cumulative effect of external bolting, bar elongation percentage as per the codal requirements in IS
skin frictional resistance from the internal grooved surface, 1786:2008 (2008). The accomplished tests on coupler-con-
and epoxy-based grout provides additional gripping to the nected reinforcing bars comply with the requirements of
connected reinforcing bars, which constrains them to fail codal specifications for the use of reinforcement couplers
with slightly high initial stiffness and relatively low axial as mechanical splices to bars in concrete, IS 16172:2014
deformation. (2014), which authenticates the efficiency and competency
A slip test on the coupler sleeve connection is performed of the proposed coupler sleeve.
by loading the specimen with stress variation from 0.60fy to
20 MPa (2.9 ksi), and a linearly variable differential trans- COMPARATIVE HYSTERESIS PERFORMANCE OF
former (LVDT) is used to estimate the possible amount of STANDALONE COUPLER WITH COUPLER BOX
slip. Figure 3(b) shows the cyclic plot of coupled reinforcing ASSEMBLY
bar specimen subjected to slip test, where a minimum slip of The design basis for interlocking coupler sleeves is to
0.10 mm (0.0039 in.) is noticed. The proposed coupler sleeve apprehend the bond-slip or flexural failure of reinforcing
connection is subjected to 100 cyclic loading stress cycles bar from the sleeve, which may develop at regions of high
with stress variations from 0.05 to 0.95fy to examine its moment and rotation—that is, the plastic hinge locations.
efficiency under recurrent loading. The coupled connection Symmetric design of the box assembly is ensured as the
effectively resists the applied loading cycles without failure, direction of earthquake loading is reversible. The advan-
and the load is statically increased until failure post-comple- tage of interlinking is that it allows the collective participa-
tion of 100 cycles, where no loss in tensile strength capacity tion of all the coupler sleeves, similar to trusses. The truss
of the specimen is noticed. Figure 3(c) shows the cyclic ideology designed is as per the equilibrium (plasticity) truss
@seismicisolation
@seismicisolation

Fig. 3—Performance requirements of proposed coupler sleeve as per standard code of practice (IS 16172:2014).
model by Hsu and Mo (2010). The strut-and-tie model is in the case of more than four reinforcing bars is shown in
adopted in designing column regions with the concept of Fig. 4(a). In such a case, the length of flat steel bars varies
stress flows. Interaction relationship for shear-torsion- and remains subjective with the distance between adja-
bending is derived. The interconnecting flat steel bars cent sleeves. Further, the flat bar thickness reduces as per
undergo tension and compression based on the direction of the computed shear area at the plastic hinge. In columns of
loading, and the diagonal connection assists in the additional rectangular cross section, where the number of reinforcing
confinement of concrete enclosed between sleeves, which is bars are greater than four, the intercepting coupler sleeves of
expected to withstand higher axial loads without failure. The intermediate reinforcing bars are positioned with their inter-
process of interlinking further prevents the local buckling of linking bolt-holes oriented towards the column face and the
surrounding sections of reinforcing bars. The procedure for maximum number of interlinking flat bars that converge at
interlinking coupler sleeves is carried out using mild steel an intermediate coupler in any column configuration is four.
rectangular-shaped flat bars of 25 mm (0.98 in.) width and To compare the structural efficiency of coupler-box
5 mm (0.19 in.) thickness, while the dimensions are deter- assembly (interlinked coupler sleeve) over standalone
mined using equivalent cross-sectional area of shear rein- coupler (stirrup bound coupler, Fig. 4(b)), reverse cyclic tests
forcement in the plastic hinge region. The 8 mm (0.32 in.) are conducted on two-full scale RC columns of length 3.3 m
diameter inclined bolt holes of the sleeve are used for inter- (10.83 ft) and cross-sectional size 400 x 400 mm (15.75 x
linking, as shown in Fig. 4(a). The interlinking @seismicisolation
process 15.75 in.), as shown in Fig. 5. The coupler connections are
@seismicisolation

Fig. 4—Pictorial representation of recommended and existing coupler sleeve applications.
Fig. 5—Details and experimental test setup for RC columns.

maintained above the foundation level with the bars orig- higher drifts are not preferred to be developed in the system,
inally connected prior to casting. Four numbers of 20 mm as there is no mechanism to quantify them. The columns are
(0.79 in.) diameter reinforcing bars with a yield strength of tested under cyclic incremental lateral drifts until failure,
500 MPa (72.52 ksi) are provided as longitudinal reinforce- with the testing and loading protocols remaining similar to
ment with two-legged, 10 mm (0.39 in.) diameter reinforcing the model frame. Thus the reinforcement detailing and the
bars (500 MPa [72.52 ksi]) as shear reinforcement. Column test setup for columns can be inferred from the subsequent
casting is executed using ordinary portland cement (OPC) of section of frame testing.
Grade 43 (C), locally available river sand as fine aggregate The load-deflection behavior of tested columns in the
(FA), and well-crushed granite stone of 20 mm (0.79 in.) form of hysteresis loops and envelope plots are shown in
maximum size as coarse aggregate (CA). A conventional Fig. 6. Cyclic test results of the columns are detailed in
mixture ratio of Grade M20 concrete—that is, 1.00 (C):1.50 Table 1. There is a significant increase in the hysteretic
(FA):3.00 (CA) with a water-cement ratio (w/c) of 0.50—is performance of standalone couplers by interlinking them
adopted for column casting. Reinforcement detailing in the into a box-shaped confinement system. Performance of both
column specimens is kept identical to that of the columns of the columns is identical up to their yield point—that is, the
selected RC portal frame. The column specimens are tested hysteretic backbone curves follow a similar path until the
under a consistent vertical dead load of 30 kN (6.74 kip)— yield point, after which a minor difference in lateral load is
that is, the self-weight of the loading beam in the frame test noticed towards the negative half-cycle of the coupler-stirrup
setup. The column study aims to examine the efficiency of column, and a significant variation in the post-yield behavior
coupler sleeves under the worst possible mode of failure that of coupler-box column is noticed towards positive half-
occurs near the plastic hinge region of columns. Tension cycle. The column with box arrangement sustains a stable
failure of the column due to flexure is one such significant post-yield behavior up to a lateral drift ratio of 6.25%, which
failure mode, where the reinforcing bars may possibly slip is around two times more than the column with standalone
out from their respective sleeve connections. The proposed coupler. The integral confinement action of reinforcing bars
sleeve connections are strong in shear and compression, as provided by the coupler-box assembly favours a balanced
the effective cross section doubles near the connection. Thus, and symmetric hysteretic loop behavior in the box assem-
the columns are projected to fail in tension, where the axial blage column. The reinforcing bars of conventional coupler-
load must be in the range of zero to balanced load condition— stirrup column undergo stress reversals independently,
that is, approximately 20% of pure axial load capacity. The unlike the coupler-box column where the interlinking flat
axial load is kept minimum, so that tension failure prevails steel bars develop a truss mechanism inside the plastic hinge
under any test conditions. Further, the secondary moments at region and distribute the stresses over the interlinked group
@seismicisolation
@seismicisolation

Fig. 6—Comparative performance of interlinked coupler-box technique.
Table 1—Cyclic test results of tested RC specimens
Yield stage Maximum stage Ultimate stage Ductility
factor
Compres-
Positive Negative Positive Negative Positive Negative (μ)
sive Energy
strength on Py, ∆y, ∆y, Pm, ∆m, Pm, ∆m, ∆u, ∆u, dissipation,
testing day, kN mm Py, kN mm kN mm kN mm Pu, kN mm Pu kN, mm kN-mm
Specimen ID MPa (ksi) (kip) (in.) (kip) (in.) (kip) (in.) (kip) (in.) (kip) (in.) (kip) (in.) ∆u/∆y (kip-in.)
Coupler- 25.52 20.5 19.5 20.9 19 32.5 49.6 34.8 79.8 26 102.3 29.8 159.9 43,089.15
6.83
stirrup column (3.70) (4.6) (0.77) (4.7) (0.75) (7.3) (1.95) (7.8) (3.14) (5.8) (4.03) (6.7) (6.29) (381.36)
Coupler-box 25.14 17.4 15.5 19.9 15 36.5 99.8 32.4 79.3 30.9 199.3 25.9 154.9 145,824.05
11.59
column (3.65) (3.9) (0.61) (4.5) (0.59) (8.2) (3.93) (7.2) (3.12) (6.9) (7.85) (5.8) (6.09) (1290.60)
26.28 60.1 18.5 60.6 18 91.2 59.49 90.8 69.4 72.9 166.9 72.7 166.2 312,629.33
Control frame 9.12
(3.81) (13.5) (0.73) (13.6) (0.71) (20.5) (2.34) (20.4) (2.73) (16.4) (6.57) (16.3) (6.54) (2766.89)
Restored 31.71 66.9 20.5 62.4 19.7 108.6 59.52 99.4 59.2 86.4 141.8 79.5 132.7 267,472.44
6.83
frame (4.59) (15) (0.81) (14) (0.78) (24.4) (2.34) (22.4) (2.33) (19.4) (5.58) (17.9) (5.22) (2367.24)
of reinforcing bars. This leads to a conventional failure or coupler-stirrup column, which is absent in the coupler-box
rupture in reinforcing bars of the coupler-stirrup column column.
with an asymmetric behavior. The irregular behavior in Fig. Interlinking the proposed coupler sleeves essentially
6(a) is due to the rupture of reinforcing bars at the right edge unites the strength of the standalone coupler under flexure,
of the column face—that is, for classification, rupture occurs which minimizes the chance of flexural yielding and the
in reinforcing bars that experience tensile stresses under restored section is subjected to shear force where the
the negative half-cycle, which causes a sudden reduction strength of the coupler sleeve is resistive. Thus, the entire
in lateral strength post 3.2% lateral drift. This aspect is not box assembly functions as a rigid link section which shrinks
noticed during the negative half-cycle of testing. Moreover, the plastic hinge length and compels the adjacent rein-
the pinching effect in the hysteretic loops is noticed in the forcing bar segments to yield away from the coupled region.
Energy dissipation capacity of the box assembly column is
@seismicisolation
@seismicisolation

enhanced by 3.3 times, and the ductility factor is improved load through two rollers placed at a meter’s distance (3.28
by 1.7 times compared to the standalone coupler column. ft) from joints and is kept consistent throughout the test
The failure mode remains identical in both columns, where sequence (refer to Fig. 9). The deformation/deflection profile
the yielding of longitudinal reinforcing bars occurs at the of beam under uniformly distributed infill wall load is simu-
column-foundation interface at 3% lateral drift. Subse- lated by applying two-point loading on top of the beam. The
quently, with the increase in lateral drift, a sudden brittle axial load is preferred to be low to make the columns fail
failure is observed in the coupler-stirrup column, whereas a in flexure/tension and to minimize the secondary moments
ductile and balanced behavior, even at a drift of 6% (which is (P-∆) at large lateral displacements, as there is no mech-
higher than the prescribed codal provision of 4% by FEMA anism to monitor them in the test setup. The horizontal
356 [2000]) is observed in coupler-box column, as shown actuator associated with the top face of the loading beam
in Fig. 7. exercises lateral displacements in an incremental sinusoidal
wave pattern and the lateral force response obtained from it
HYSTERETIC PERFORMANCE EVALUATION is used in plotting hysteresis curves. The test is carried out
OF RC PORTAL FRAME RETROFITTED WITH until comprehensive damage to the model frame is achieved.
COUPLER-BOX ASSEMBLY Hence, the number of load cycles are high and the frame can
The primary model frame opted for this study is an RC withstand a higher lateral drift of 6%.
portal frame of height 3.35 m (10.99 ft), which is cast and The hysteresis behavior of the conventional portal frame
tested under the combined effect of vertical and lateral loads (hereby referred to as the CON frame) is plotted, as shown
in a large-scale structural test facility at the Department of in Fig. 10. A stable behavior with gradual post-peak perfor-
Earthquake Engineering, Indian Institute of Technology (IIT) mance is observed up to a lateral drift of 6%. Eventually,
Roorkee. Columns of cross-sectional size 400 x 400 mm buckling of longitudinal reinforcing bars occurs at the plastic
(15.75 x 15.75 in.) are connected by a 3.0 m (118.11 in.) hinge location of columns—that is, at the column base along
length beam of size 300 x 300 mm (11.81 x 11.81 in.). with shear cracks at upper beam-column joints at a higher
The frame assembly is supported by a 5.0 m (196.85 in.) lateral drift of 6%. The damaged CON frame model after its
length foundation beam of sectional size 550 x 600 mm preliminary test is shown in Fig. 11.
(21.65 x 23.62 in.). A minimum longitudinal reinforce- The damaged model frame after the preliminary test is
ment of 0.8%—that is, four numbers of 20 mm (0.79 in.) repaired at its hinge locations using the proposed retrofitting
diameter thermomechanical treatment (TMT) processed methodology of coupler-box assembly. The sequence for
reinforcing bars with an average yield strength of 500 MPa restoring a damaged column base is shown in Fig. 12. The
(72.52 ksi)—are used for reinforcing the beam and column fundamental procedure consists of six main steps, namely:
sections. Two-legged stirrups of 10 mm (0.39 in.) diameter (a) supporting the damaged frame by redistributing the
reinforcing bar with 500 MPa (72.52 ksi) yield strength are vertical dead loads through jacks and chipping of crushed
used to reinforce the sections for shear. Stirrup detailing concrete at hinge regions to expose the buckled reinforcing
of the frame is provided as per the codal specifications for bars (the amount of concrete that is demolished is two times
ductile detailing of reinforced concrete structures subjected the length of the sleeve, with the provision of one length
to seismic forces, IS 13920:2016 (2016). Section details, of sleeve on either side of the cut reinforcement; this deci-
along with the reinforcement detailing of the portal frame, sion remains subjective with the available length of concrete
are shown in Fig. 8. The mixture ratio adopted for frame removal); (b) rectification of reinforcing bars along with the
casting in the preliminary study is 1.00 (C):1.50 (FA):3.00 insertion of coupling sleeves on repositioned reinforcing
(CA) with a w/c of 0.50. bars; (c) proper positioning and fixing of sleeves with high-
The RC portal frame is tested under displacement- strength external bolts at a proportion of six bolts (half the
controlled reverse cyclic load as per interim test protocols total number of bolts) on each connected reinforcing bar
for determining seismic performance characteristics of segment; (d) interlinking the coupling sleeves in lateral
structural and nonstructural components (FEMA 461 2007). and diagonal directions, respectively, using flat steel bars
A constant loading frequency of 0.01 Hz is adopted with an to generate a box-shaped assembly; (e) injection of high-
initial loading rate of 0.2 mm/sec (0.0079 in./sec), which is strength epoxy mortar grout to fill the void spaces between
increased at a rate of 0.2 mm/sec (0.0079 in./sec) until 20 reinforcing bar and coupler sleeve and eventually reinstating
mm (0.79 in.) lateral displacement. After the 20 mm (0.79 the transverse reinforcement near adjacent frame sections;
in.) displacement cycle, the loading rate is increased at a and (f) laying freshly prepared concrete (mixture ratio: 1.00
pace of 0.4 mm/sec (0.016 in./sec) until the end of the test (C):1.40 (FA):2.20 (CA)) using letter-box shuttering. A
sequence. The test is executed using three servo-controlled resin-based concrete bonding agent is applied to the existing
hydraulic actuators (one horizontal: 500 kN [112.4 kip]; old concrete’s surface to create a better bond with the new
±250 mm [9.84 in.] and two vertical: 250 kN [56.2 kip]; concrete.
±250 mm [9.84 in.]), as shown in Fig. 9. All the three actu- The retrofitted (RET) frame is tested under identical
ators are integrally connected to the loading beam, which is loading conditions as of the CON frame, and the hysteresis
bracketed to the model frame through tie-rods. The vertical plot is shown in Fig. 13. The test results of model frames
actuators with swivel-end assemblies at both edges are are detailed in Table 1. The comprehensive performance of
fixed atop steel stands and placed in line with the columns the RET frame is monitored using extensive instrumenta-
of the model frame. The loading beam enforces vertical tion during the cyclic test. LVDTs are used to monitor the
@seismicisolation
@seismicisolation

Fig. 7—Failure patterns noticed in RC columns.
Fig. 8—Reinforcement detailing in tested RC model frames.

@seismicisolation
@seismicisolation

Fig. 9—Schematic diagram showing test setup for RC frames.
captured using the Vic-Snap software. Complementary
metal oxide semiconductor (CMOS) camera sensors of 8.9
megapixels are equipped with 10 mm (0.39 in.) lens, and a
calibration plank of 40 mm (1.57 in.) size is used along the
column height, which is pre-speckled with a 0.20 in. (5.08
mm) sized roller. Test images are captured at a rate of two
frames per second, and post-processing is performed using
the Vic-3D software, where the selected subset size is 45
pixels, step size is 11, and the strain computation filter size
is 29 pixels.
The comparative performance of CON and RET frame
models in the form of an envelope curve of hysteresis loops
is shown in Fig. 14. The initial strength and stiffness of RET
frame are similar to that of the CON frame, which signifies
the efficiency of the adopted retrofitting technique. A consid-
erable increase in the post-yield behavior of the RET frame
in comparison to the CON frame is observed. The enhance-
ment in gradual post-peak behavior of the RET frame is
noticed until a lateral drift ratio of 5%, which is higher than
Fig. 10—Hysteresis curves of CON model frame. the collapse limiting drift of 4% as prescribed by FEMA
356. This manifests the fact that the coupler-box assembly
lateral displacement of frames at various heights—that is, sustains higher strains without any slip of longitudinal rein-
(i) just above the foundation level at 150 mm (5.91 in.); forcing bars from their respective sleeves. An average incre-
(ii) mid-level of the column at 1525 mm (60.04 in.); and ment of 20% in lateral strength is noticed, which emphasizes
(iii) upper section of the column below bracket attachment that the coupler-box confinement resists the high amount of
at 2750 mm (108.27 in.). Strain variations in reinforcing shear and moment developed at the plastic hinge region. The
bars above and below the box assemblies of RET frame are box assembly induces rigidity to the frame section which
acquired by a 32-channel data acquisition (DAQ) system shifts the formation of failure hinges away from the coupled
through 10 mm (0.39 in.) strain gauges fixed at these loca- region—that is, near the foundation at column-foundation
tions. Further, a noncontact full-field deformation measure- junction.
ment technique known as digital image correlation (DIC) is The linear deflection profiles of column members in the
used to inspect the column surface of RET frame for vari- CON and RET frame models are determined using the
ation in superficial principal strains and lateral displace- recorded data of LVDT and DIC. The peak displacement
ments. This technique adopts a three-dimensional (3-D) DIC readings obtained from the three equidistant LVDTs are
system, which analyzes high-definition speckled images plotted against their corresponding peak loads, as shown in
@seismicisolation
@seismicisolation

Fig. 11—Damaged model frame after preliminary test.
Fig. 12—Schematic diagram depicting procedure for reestablishing damaged column.

images are shown in Fig. 15. Selective output images in
their left and right sway directions are presented at the
lateral displacement cycles of 10, 100, and 200 mm (0.39,
3.94, and 7.87 in.)—that is, at initial, midway, and end of
the test sequence, respectively. The variation in principal
strains along the column length helps in recognizing the
regions of high-strain concentrations—that is, the possible
location of cracks/damage. It is clearly manifested that the
percentage of principal strain variation is comparatively low
at the coupler-confined region of column, which indicates
the absence of concrete cracking. This emphasizes the suit-
ability of the employed retrofitting technique in confining
core concrete and providing better stiffness retention to the
coupler-confined column sections. In the eventual part of
the test run, flexural cracks are noticed at the base part of
columns—that is, away from the coupler-box section, which
enlarges with a gain in lateral drift ratio. Diagonally inclined
shear cracks and spalling of cover concrete at beam-column
joint locations of the frame are noticed at a higher lateral
Fig. 13—Hysteresis curves of RET model frame. drift of 6%.
Strain variations at possible locations of longitudinal
Fig. 14. The plotted displacement contours distinctly indi-
reinforcing bars placed above and below the box assem-
cate that the RET frame retained its initial stiffness along
blies in the columns of the RET frame are plotted against
with an enhanced lateral load capacity. A similar trend of
their corresponding lateral drifts, as shown in Fig. 16. Strain
variation in deflections is observed in both the model frames,
values in gauges below the box section show larger vari-
with a meagre variation in RET frame post 5% lateral drift.
ation than the gauges above it, which indicates the yield/
The damage patterns observed in the RET frame with
rotation of reinforcing bars below the assembly. The average
superficial crack propagations inferred from DIC @seismicisolation
output
@seismicisolation

Fig. 14—Comparative performance in tested RC model frames.
strain variation value in reinforcing bars below the assembly The damage tolerance capacities of tested frames are
is 0.015, whereas it is only 0.005 above it, which is much computed using the modified damage index (DI) proposed
lower than the yield strain of the adopted reinforcing bar. by Park and Ang (1985) (Eq. (4)), and the computed values
The cumulative amount of energy dissipated by the CON are plotted against corresponding lateral drifts, as shown in
and RET frame models is plotted against corresponding Fig. 19. A similar trend of variation in DI values of frame
lateral drifts, as shown in Fig. 17. The energy-dispersion models is noticed with the RET frame reaching its collapse
capacities of both the model frames are identical up to a stage at a lateral drift of 3.73% and the CON frame attaining
lateral drift of 5%, after which a meager drop in the capacity it at 4% drift.
of the RET frame is noticed in comparison with the CON
frame. The CON frame model undergoes a flexural mode STRENGTH DEGRADATION MODELS BASED ON
of failure and dissipates higher energy until the yielding of HYSTERETIC ENERGY DISSIPATION
reinforcing bars, which occurs at 6% drift. The coupler-box The peak-oriented cyclic deterioration modes of basic and
assembly of RET frame increases the local rigidity of post-capping strength for the CON and RET frame speci-
column section and consecutively transfers the lateral force mens are developed based on model equations proposed by
to an adjacent frame section near to column-foundation Ibarra and Krawinkler (2005). The backbone curve for the
junction, where the longitudinal reinforcing bars undergo a hysteretic model is defined by initial stiffness (Ke), hard-
repeated reversal of stresses until failure. This leads to the ening stiffness (Ks), post-capping stiffness (Kc), and residual
formation of a plastic hinge at column base, which reduces strength (Fr), as described in Fig. 20 (Ibarra and Krawinkler
the relative amount of energy dissipated post 5% drift. The 2005), where αs represents the strain-hardening coefficient;
equivalent viscous damping (EVD) coefficient/ratio ξhyst of αc denotes the post-capping stiffness coefficient; and λ is the
frame models is computed based on the elastic strain energy coefficient of residual strength. Strength-degradation modes
approach, and the percentage of variations is plotted against of the tested portal frames are stimulated when the lateral
the corresponding drift, as shown in Fig. 17. The calculated load value exceeds yield strength, and the degradation rate
value of the EVD ratio is directly proportional to the amount is computed based on hysteretic energy dissipated during
of energy dissipated and thus, a similar trend of deviation as cyclic loading (Rahnama and Krawinkler 1993). The deteri-
that of the dispersion capacity is noticed, where a variation oration parameter βi is defined as
in the damping ratio of RET frame occurs post 5% drift.
The degradations in strength and stiffness of model frames _ Ei C
( )
post their elastic state are computed and plotted against
βi = i (5)
corresponding lateral drifts, as shown in Fig. 18. The varia-
E
t −
∑
E
j
j=1
tion in stiffness degradation pattern is similar in both model
frames, with a marginally reduced rate in the RET frame. where Ei represents the hysteretic energy dissipated in the
The variation in strength degradation parameter of frame i-th cycle; ∑Ej indicates the summation of energy dissipated
models shows a deviation in the RET frame post 5% drift by in all previous cycles until i; and Et is the energy dissipation
virtue of its declination in lateral strength capacity. Thus, it capacity expressed as a function of twice the elastic strain
is affirmed that the RET frame regains strength and stiffness energy at yielding—that is, Fyδy
up to its original level by adopting the recommended retro-
fitting technique. Et = γFyδy (6)
δm− δy _ ∫E= dE

E=Em
Here, the factor γ is calibrated from the experimental test
_ E1

D = + β (4) results and is different for each deterioration parameter. The
δf− δy Fyδf
@seismicisolation
@seismicisolation

Fig. 15—Variation of superficial principal strains in RET frame monitored using DIC.
exponent factor C defines the rate of degradation in eval- equal to its reduced yield strength. This strength deteriora-
uated hysteresis parameters—that is, strength and stiffness, tion mode also includes the degradation of strain-hardening
with values ranging between 1.0 and 2.0. The quantified slope computed using parameter βs,i and rotated towards the
value of βi must be within the range of 0 < βi, ≤ 1, and the horizontal axis using corresponding equations. This degra-
hysteretic energy is assumed to be exhausted for conflicting dation mode is well-explained in Fig. 21, where the mode
values. initiates post the yield point (1). The reduced yield strength
in the negative direction Fi– (4, 10) is computed using Eq. (8)
Basic and post-capping strength degradation when the curve traverses the horizontal axis (3, 9). In succes-
modes sion, the strain-hardening slope of the curve Ks,i
− is modified
The basic strength degradation mode of a peak-oriented using Eq. (10), corresponding to which the coordinates (5,
hysteretic model is defined by translating the strain-hardening 11) are computed. The reduced yield strength in the positive
branch of the backbone curve towards origin by an @seismicisolation
amount direction Fi+ (7, 13) after every negative inelastic excursion
@seismicisolation

Fig. 16—Strain variations in longitudinal reinforcing bars of RET frame.
Fig. 17—Energy-based parametric comparison in tested frames.
Fig. 18—Comparative plots of strength and stiffness degra-

dations in tested model frames. Fig. 19—Damage index versus drift ratio plots of model
frames.
(6, 12) is computed using Eq. (7), and the strain-hardening
slope K
s,i
+ is altered using Eq. (9) to attain corresponding curves maintains a slope equal to the initial elastic curve
coordinates (8, 14). The downward branch of degradation through which the abscissa of points (3, 9, 6, and 12) on
the horizontal axis are computed. This computation of the
@seismicisolation
@seismicisolation

degradation model of the curve is plotted until the exhaus- hardening curve before expedition i; and βs,i is determined
tion of hysteretic energy—that is, until βs,i > 1 using the relevant γs factor in the aforementioned equations.
The post-capping strength degradation mode of the

Fi+ = (1 − βs,i) Fi−1
+ (7) peak-oriented hysteretic model translates the post-cap stiff-
ness branch of the backbone curve towards origin by an
amount equal to its reduced reference strength Fref, which

Fi− = (1 − βs,i) Fi−1
− (8) is estimated as the intersection ordinate of the vertical force
axis with the projection of the post-cap branch, as shown in
where Fi+/−
is the deteriorated yield strength after expedition/ Fig. 21. The reduced reference strengths in both positive and
cycle i; and Fi−1
+/−is the deteriorated yield strength before negative and directions, Fref,1
+ and Fref,1
− , are computed using
expedition/cycle i Eq. (11) and (12) when the curve crosses the horizontal axis
(3, 8, 6, and 10), respectively. The slope of post-capping

Ks,i
+ = (1 − βs,i) Ks,i−1
+ (9) branch is kept unaltered and the abscissa of points on the
horizontal axis is obtained by the intersection of downward
branch of the degradation curve, whose slope is identical

Ks,i
− = (1 − βs,i) Ks,i−1
− (10) to the initial elastic curve. Computation of the degradation
mode is carried until hysteretic energy exhaustion—that is,
where Ki+/−
is the deteriorated slope of strain-hardening curve until the value of βc,i exceeds unity
after expedition i; K i−1
+/−is the deteriorated slope of strain

Fref,i
+ = (1 − βc,i) Fref,i−1
+ (11)

Fref,i
− = (1 − βc,i) Fref,i−1
− (12)
where Fref,i
+/− is the deteriorated reference strength after expe-
dition i; Fref,i−1
+/− is the deteriorated reference strength before
expedition i; and βc,i is computed using the relevant γc factor
in the aforementioned equations.
The computed parametric values, along with the cyclic
degradation modes of basic and post-capping strength for
tested RC portal frames, are presented in Fig. 22, where
the value of degradation parameter γ is calibrated based
on the experimentally obtained hysteresis behavior of the
individual frame. The value of γ is high for the CON frame
specimen owing to its larger cumulative energy dissipation
capacity. The model ductility (δc/δy) is relatively high by 7%
for the CON frame, whereas the peak strength of the model
backbone is comparatively high by 7% for the RET frame.
The post-capping stiffness coefficient (αc) for the CON frame
model is half the value of the RET frame, which signifies the
Fig. 20—Backbone curve for hysteretic models (Ibarra and gradual post-peak behavior. However, the strain-hardening
Krawinkler 2005).
Fig. 21—Peak-oriented models with computed deterioration modes (Ibarra and Krawinkler 2005).
@seismicisolation
@seismicisolation

Fig. 22—Cyclic degradation effect on hysteretic response: peak-oriented models.
coefficient (αs) is high for the RET frame, which is indicated concrete, integrates the reinforcing bar action against lateral
by a sharp rise in the slope of curve post its yield point. The forces, and alters the critical flexure failure of reinforcing
basic strength cyclic degradation mode in CON and RET bar slip from the sleeve connection.
frame specimens reaches its collapse stage—that is, βi > 1— 3. A comparative hysteresis performance of the retrofitted
at normalized displacement ratios of 11 and 10, respectively. frame shows that it regains the initial strength and stiff-
The post-capping cyclic degradation mode depreciates the ness of the preliminary damaged frame with a better post-
peak point of model curve based on its modified reference yield behavior even at a lateral frame drift of 5.6%. This is
strength, and the declination rate is comparably high in the more than the collapse prevention drift level of 4% as per
RET frame due to its higher descent in the post-capping FEMA 356 guidelines. Concurrently, an average increase of
branch. These cyclic deterioration modes are computed 20% in lateral strength is observed in the retrofitted frame,
based on experimentally dissipated hysteretic energy, and which emphasizes the efficiency of coupler-box assembly in
the CON frame model shows better variation than the RET sustaining high moments and shear developed at the plastic
frame. hinge regions. The induced rigidity of coupler-box assembly
shrinks the plastic hinge length and allows its forma-
CONCLUSIONS tion below the box assemblage—that is, near the column-
Most of the past earthquake damage survey reports world- foundation junction, under larger flexural moments.
wide show that the buckling of reinforcing bars at the hinge 4. The propagation of cracks and failure pattern in the
location of frame members is a common failure mode during retrofitted frame, as inferred from the digital image correla-
earthquake loading. The current research focuses on seismic tion (DIC) technique, illustrates that the superficial principal
retrofitting such as severely damaged frame sections in strain at the coupled region of columns shows a lower range
reinforced concrete (RC) buildings using the coupler-box of values than the cracking strain of concrete, which indi-
confinement technique, which is validated by an experi- cates the absence of concrete cracks. Simultaneously, the
mental investigation. The efficiency of the proposed retrofit- average strain variation values recorded by gauges fixed on
ting scheme is evaluated by conducting reverse cyclic tests reinforcing bars above the box assembly are relatively lower
on RC columns and frames with four numbers of longitu- than those below. This authenticates the failure location in
dinal reinforcing bars. The test results are further used in devel- retrofitted frame—that is, yielding in reinforcing bars below
oping hysteresis models for RC frames retrofitted with the the column at the foundation connection.
recommended technique. The following conclusions are 5. The comparative evaluation of performance parame-
drawn from the study: ters computed using the hysteresis behavior of frames, such
1. The uniaxial tensile test on proposed coupler sleeve as energy dissipation, viscous damping, strength, and stiff-
reveals that its post-yield behavior depends on the diam- ness degradations, clearly interprets the retrofitted frame’s
eter of the connected reinforcing bar, and its performance behavior in its elastic and inelastic range, which is relatively
is enhanced by increasing its bond strength through: (a) at par with the original bare frame. The computed damage
external bolts; (b) high-strength epoxy filler material; and index (DI) values show a similar variation, with the frames
(c) a grooved surface. The proposed coupler sleeve accom- reaching their collapse stage at a nearly equal lateral drift
plishes the performance requirements of IS 16172:2014 ratio.
through static, slip, cyclic tensile, low-cycle, and high-cycle 6. Strength-degradation hysteresis models for tested frames
fatigue tests. are determined based on experimentally obtained backbone
2. The quasi-static cyclic test results of RC columns with curves and energies dissipated in their respective cycles.
standalone coupler and confined coupler-box assembly The deterioration models based on basic and post-capping
clearly manifest the significance of interlinking, which strength modes can reasonably capture the cyclic perfor-
improves the column’s load-deformation capacity by two mance of retrofitted frame with coupler-box assemblies, and
times. Further, the provided box assembly confines the core can further be used to evaluate existing structures.
@seismicisolation
@seismicisolation

AUTHOR BIOS Modes,” Malaysian Journal of Civil Engineering, V. 20, No. 1. doi:
ACI member Naveen Kumar Kothapalli is a Research Scholar in the 10.11113/mjce.v20.15759
Department of Earthquake Engineering, Indian Institute of Technology Huaco, G., and Jirsa, J., 2012, “Performance of Damaged Column Retro-
(IIT) Roorkee, Uttarakhand, India. He received his BTech from Acharya fitted with Innovative Materials and Devices,” 15th World Conference on
Nagarjuna University, Guntur, Andhra Pradesh, India, in 2012, and his Earthquake Engineering, 15WCEE, 10 pp.
MTech from IIT Roorkee in 2015. His research interests include repair, Ibarra, L. F., and Krawinkler, H., 2005, “Global Collapse of Frame
retrofitting, and rehabilitation of reinforced concrete structures. Structures Under Seismic Excitations,” Pacific Earthquake Engineering
Research Center, Berkeley, CA, pp. 29-51.
R. Siva Chidambaram is a Scientist at Central Building Research Insti- IS 1786:2008, 2008, “High Strength Deformed Steel Bars and Wires
tute, CSIR-CBRI Roorkee, India. He received his BE and ME from Anna for Concrete Reinforcement – Specification (Fourth Revision),” Bureau of
University, Chennai, Tamil Nadu, India, in 2007 and 2010, respectively, Indian Standards, New Delhi, India.
and his PhD from IIT Roorkee in 2016. His research interests include shear IS 9417:2018, 2018, “Welding of High Strength Steel Bars for Rein-
strengthening of beam-column joints with high-performance materials in forced Concrete Construction – Recommendations,” Bureau of Indian Stan-
precast and reinforced concrete structures. dards, New Delhi, India.
IS 13920:2016, 2016, “Ductile Design and Detailing of Reinforced
Pankaj Agarwal is a Professor in the Department of Earthquake Engi- Concrete Structures Subjected to Seismic Forces - Code of Practice (First
neering at IIT Roorkee, where he received his ME and PhD in 1994 and Revision),” Bureau of Indian Standards, New Delhi, India.
2000, respectively. He received his BTech from G.B. Pant University of IS 16172:2014, 2014, “Reinforcement Couplers for Mechanical Splices
Agriculture & Technology, Pantnagar, Uttarakhand, India, in 1989. His of Bars in Concrete – Specification,” Bureau of Indian Standards, New
research interests include earthquake-resistant design of structures, repair, Delhi, India.
and retrofitting of masonry and reinforced concrete buildings. Ingham, J. M., and Bai, A. S. H., 2009, “Seismic Performance of
Mechanically Coupled Reinforcing Bars,” Magazine of Concrete Research,
V. 61, No. 7, pp. 529-537. doi: 10.1680/macr.2008.00098
REFERENCES Lancelot, H. B. III, 1995, “Combination Mechanical/Grout Sleeve
Ahmed, G. H., 2015, “Mechanical Properties of Welded Deformed Rein-
Coupling for Concrete Reinforcement Bars,” U.S. Patent 5,383,740, filed
forcing Steel Bars,” ARO-The Scientific Journal of Koya University, V. 3,
Aug. 2, 1993, published Jan. 24, 1995.
No. 1, pp. 28-39. doi: 10.14500/aro.10059
Mochizuki, H., and Nihei, T., 1998, “Mortar Grout Splice Sleeve for
Alias, A.; Sapawi, F.; Kusbiantoro, A.; Zubir, M. A.; and Rahman, A. A.,
Reinforcing Bars,” U.S. Patent 5,732,525, filed Nov. 15, 1996, published
2014, “Performance of Grouted Splice Sleeve Connector Under Tensile
Mar. 31, 1998.
Load,” Journal of Mechanical Engineering Science, V. 7, pp. 1096-1102.
Nzabonimpa, J. D., and Hong, W. K., 2018, “Novel Precast Erection
Allen, P. B.; Colarusso, L. J.; Russwurm, D.; and Judson, W. M.,
Method of Interlocking Mechanical Joints Using Couplers,” Journal
2007, “Reinforcing Bar Splice with Threaded Collars,” U.S. Patent
of Construction Engineering and Management, ASCE, V. 144, No. 6,
20070175167A1, filed Jan. 10, 2007, published Aug. 2, 2007.
p. 04018036. doi: 10.1061/(ASCE)CO.1943-7862.0001482
Ameli, M. J.; Brown, D. N.; Parks, J. E.; and Pantelides, C. P., 2016,
Park, Y. J., and Ang, A. H. S., 1985, “Mechanistic Seismic Damage Model
“Seismic Column-To-Footing Connections Using Grouted Splice Sleeves,”
for Reinforced Concrete,” Journal of Structural Engineering, ASCE, V. 111,
ACI Structural Journal, V. 113, No. 5, Sept.-Oct., pp. 1021-1030. doi:
No. 4, pp. 722-739. doi: 10.1061/(ASCE)0733-9445(1985)111:4(722)
10.14359/51688755
Parks, J. E.; Brown, D. N.; Ameli, M. J.; and Pantelides, C. P., 2016,
Bompa, D. V., and Elghazouli, A. Y., 2019, “Inelastic Cyclic
“Seismic Repair of Severely Damaged Precast Reinforced Concrete Bridge
Behaviour of RC Members Incorporating Threaded Reinforcement
Columns Connected with Grouted Splice Sleeves,” ACI Structural Journal,
Couplers,” Engineering Structures, V. 180, pp. 468-483. doi: 10.1016/j.
V. 113, No. 3, May-June, pp. 615-626. doi: 10.14359/51688756
engstruct.2018.11.053
Phillippi, D. J., and Hegemier, G. A., 2013, “Use of Mechanical Couplers
Dahl, K. L., 2001, “High Strength Grouted Pipe Coupler,” U.S. Patent
in Concrete Columns,” AEI 2013: Building Solutions for Architectural
6,192,647, filed Apr. 15, 1999, published Feb. 27, 2001.
Engineering, pp. 715-724.
FEMA 356, 2000, “Prestandard and Commentary for the Seismic Reha-
Popa, V.; Papurcu, A.; Cotofana, D.; and Pascu, R., 2015, “Experimental
bilitation of Buildings,” Federal Emergency Management Agency, Wash-
Testing on Emulative Connections for Precast Columns Using Grouted
ington, DC.
Corrugated Steel Sleeves,” Bulletin of Earthquake Engineering, V. 13,
FEMA 461, 2007, “Interim Testing Protocols for Determining the
No. 8, pp. 2429-2447. doi: 10.1007/s10518-014-9715-9
Seismic Performance Characteristics of Structural and Non-structural
Rahnama, M., and Krawinkler, H., 1993, “Effects of Soft Soil and
Components,” Federal Emergency Management Agency, Washington, DC.
Hysteresis Model on Seismic Demands,” Report No. 108, John A. Blume
Haber, Z. B.; Mackie, K. R.; and Al-Jelawy, H. M., 2017, “Testing and
Earthquake Engineering Center, Standford, CA.
Analysis of Precast Columns with Grouted Sleeve Connections and Shifted
Rowell, S. P., and Hager, K. P., 2010, “Investigation of the Dynamic
Plastic Hinging,” Journal of Bridge Engineering, ASCE, V. 22, No. 10,
Performance of Large Reinforcement Bar Mechanical Couplers,” Struc-
p. 04017078. doi: 10.1061/(ASCE)BE.1943-5592.0001105
tures Congress, pp. 2059-2075.
Holdsworth, S., 2006, “Tubular Coupler for Concrete Reinforcing Bars,”
Yang, Y.; Sneed, L. H.; Morgan, A.; Saiidi, M. S.; and Belarbi, A., 2015,
U.S. Patent 5909980A, filed Sept. 8, 1997, published June 8, 1999.
“Repair of RC Bridge Columns with Interlocking Spirals and Fractured
Holdsworth, S. E., 1997, “Tubular Coupler for Concrete Reinforcing
Longitudinal Bars – An Experimental Study,” Construction and Building
Bars,” U.S. Patent 5,664,902, filed June 16, 1996, published Sept. 9, 1997.
Materials, V. 78, pp. 405-420. doi: 10.1016/j.conbuildmat.2015.01.010
Hope, P. F., 1987, Reinforcing Bar Coupling System, U.S. Patent
Yee, A. A., 1970, “Splice Sleeve for Reinforcing Bars,” U.S. Patent
4,666,326.
3,540,763, filed June 17, 1968, published Nov. 17, 1970.
Hsu, T. T., and Mo, Y. L., 2010, Unified Theory of Concrete Structures,
Yee, A. A., 1986, “Splice Sleeve for Reinforcing Bars with Cylindrical
John Wiley & Sons Inc., New York.
Shell,” U.S. Patent 4,627,212, filed Aug. 9, 1985, published Dec. 9. 1986.
Hua, L. J.; Rahman, A. B. A.; Mirasa, A. K.; and Hamid, Z. A., 2008,
Yee, A. A., 2009, “Rebar Splice Sleeve and Method of Splicing,” U.S.
“Performance of Cs-Sleeve Under Direct Tensile Load: Part I: Failure
Patent Application 12/081,539.
@seismicisolation
@seismicisolation

JOIN AN
ACIChapter!
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.
@seismicisolation
@seismicisolation   
www.concrete.org/chapters
Title No. 120-S80
Seismic Performance of Five-Spiral Concrete Columns

with Large-Diameter Longitudinal Reinforcement
by Yu-Chen Ou and Brandon Li
Large-diameter longitudinal reinforcement (D43 [No. 14]) and consists of one central large spiral and four small spirals
multi-spiral reinforcement were examined for use in the construc- at the corners of the column cross section. The reinforced
tion of concrete columns to reduce labor demand. Two full-scale concrete design code of Taiwan14 is the only design code
columns were tested to investigate their seismic performance. The globally that includes the five-spiral reinforcement. The code
test column was reinforced with large-diameter longitudinal rein-
requires that the maximum distance between the inner faces
forcement (D43 [No. 14]) and the control column with normal-
of small and large spirals in the overlapping region shall not
diameter longitudinal reinforcement (D36 [No. 11]). Both columns
were laterally confined by five-spiral reinforcement. Test results be less than 0.3 times the inner diameter of the small spiral
showed that the seismic performance of the test column was similar and 60 mm (2.36 in.) to ensure proper interlocking between
to the control column. Both columns showed ductile behavior with small and large spirals. Moreover, the volumetric ratio of
good energy dissipation. Based on the test observations, a buckling each spiral of five-spiral reinforcement in the plastic hinge
model was developed to predict the buckling behavior of longitu- region shall satisfy the requirements for conventional spiral
dinal bars laterally supported by five-spiral reinforcement. Based reinforcement.7,14
on the proposed model, it was found that for the same amount and The large-diameter longitudinal reinforcement used
within the practical spacing range of five-spiral reinforcement, the in this research has a diameter of 43 mm (1.69 in.) (D43
calculated buckling slenderness ratio and hence the compressive [No. 14]). Large-diameter longitudinal reinforcement can
stress-strain behavior of D43 (No. 14) bars is similar to that of
reduce the number of mechanical splices and hence decrease
D36 (No. 11) bars.
labor demand associated with splicing operations. Previous
Keywords: buckling; columns; cyclic test; five-spiral reinforcement; studies have investigated the bond behavior of large-diam-
large-diameter bars; multi-spiral reinforcement; reinforced concrete; eter bars. Ichinose et al.15 studied the bar size effect on the
seismic. bond strength of bars with a diameter up to 52 mm (2.05 in.).
Research results showed bond strength tended to decrease
INTRODUCTION with increasing bar diameter. However, this size effect
Due to the shortage of workers, the construction industry decreased with increasing confinement to the bars provided
in Taiwan has been developing methods to decrease labor by concrete or transverse reinforcement. Steuck et al.16 inves-
demand. In this research, two ways to reduce labor demand tigated the bond behavior of bars grouted in corrugated steel
are employed for the construction of reinforced concrete ducts for a bar diameter up to 57 mm (2.24 in.). Research
columns, including multi-spiral transverse reinforcement results showed that the bar size effect on bond strength was
and large-diameter longitudinal reinforcement. insignificant. Note that good confinement was provided for
Multi-spiral reinforcement is developed to allow non- bars tested in Steuck’s study. The bond failure was governed
circular columns to use and take advantage of spiral rein- by shearing along the bar-grout interface, which reduced
forcement. Several forms of multi-spiral reinforcement the size effect according to Ichinose’s study. The bond tests
have been developed and studied, including two-spiral rein- by Ichinose et al. and Steuck et al. were carried out using
forcement for oblong columns,1-6 five-spiral reinforcement monotonic loading. Murcia-Delso et al.17 studied the bond
for square columns,7-9 six-spiral reinforcement for rect- behavior of bars with a diameter up to 57 mm (2.24 in.) using
angular columns,10 seven-spiral reinforcement for oblong cyclic loading. The bars were embedded and well confined
columns,11-13 and 11-spiral reinforcement for rectangular in a large circular concrete member reinforced with a dense
columns.12 It has been demonstrated that multi-spiral rein- transverse reinforcement. This simulated the confined situa-
forcement can be fabricated by a machine and assembled tion for a column longitudinal bar embedded into an enlarged
rapidly on a construction site, significantly reducing labor pile shaft. A development length equation was proposed for
demand.7 Moreover, multi-spiral reinforcement shows large-diameter bars based on the test results and analytical
a better confinement effect than rectilinear tie reinforce- study.
ment. Even with a smaller amount of transverse reinforce-
ment, columns with multi-spiral reinforcement show better
seismic performance than columns with rectilinear trans- ACI Structural Journal, V. 120, No. 5, September 2023.
MS No. S-2021-347.R2, doi: 10.14359/51738834, received April 5, 2022, and
verse reinforcement. In this research, five-spiral reinforce- reviewed under Institute publication policies. Copyright © 2023, American Concrete
ment is used. Figure 1 shows the five-spiral column tested Institute. All rights reserved, including the making of copies unless permission is
in this research. It can be seen that five-spiral reinforcement closure, if any, will be published ten months from this journal’s date if the discussion
@seismicisolation
@seismicisolation

Fig. 1—Column with five-spiral transverse reinforcement and large-diameter longitudinal reinforcement tested in this research:
(a) column cross-sectional view; and (b) fabrication of reinforcement cage.
Table 1—Specimen design parameters
Axial load Transverse reinforcement Longitudinal reinforcement
ratio,
Specimen fc', MPa fca', MPa P/fca'Ag fyt, MPa fyta, MPa Size @ spacing, mm ρs , % fy, MPa fya, MPa Quantity-size ρg, %
420 538.8 Large: D14 @ 75
C36 35 41.2 0.0425 1.18 420 465.7 24-D36 2.98
420 494.9 Small: D9 @ 75
420 538.8 Large: D14 @ 75
C43 35 40.5 0.0432 1.18 420 461.5 16-D43 2.87
420 494.9 Small: D9 @ 75
Note: 1 MPa = 145.04 psi; 1 mm = 0.0394 in.
Murcia-Delso et al.18 tested two full-scale columns RESEARCH SIGNIFICANCE

supported by enlarged shafts using lateral cyclic loading. Very few test data exist for concrete columns reinforced
One column used longitudinal reinforcement with a diam- with large-diameter steel bars (bar diameter ≥ 43 mm
eter of 36 mm (1.42 in.), which is typically the maximum [1.69 in.]) as longitudinal reinforcement. This research
bar diameter used for construction, and the other one with provides valuable cyclic test data on a full-scale five-spiral
a diameter of 43 mm (1.69 in.). Test results showed the column with D43 (No. 14) longitudinal bars. A control
embedded lengths designed for both columns were adequate. column with D36 (No. 11) longitudinal bars is also provided
Moreover, both columns showed seismic performance with for comparison. Furthermore, a new buckling model is
good energy dissipation and ductility capacities. However, provided for longitudinal bars laterally supported by five-
because the two columns were designed with different spiral reinforcement.
column heights and different ratios of longitudinal and trans-
verse reinforcement, it is difficult to conclude the perfor- EXPERIMENTAL PROGRAM
mance difference between columns with normal-diameter Specimen design and test setup
(36 mm [1.42 in.]) and large-diameter (43 mm [1.69 in.]) Two full-scale rectangular columns with cross-sectional
longitudinal bars. dimensions of 900 x 900 mm (35.43 x 35.43 in.) and a height
The objective of this research is to investigate the differ- of 5020 mm (197.64 in.) were designed and tested in this
ence in seismic performance of multi-spiral columns with research. The column was from the column of a prototype
normal-diameter (36 mm [1.42 in.]) and large-diameter elevated expressway located in Taiwan. One column (C36)
(43 mm [1.69 in.]) diameter longitudinal bars. Two full- was designed with D36 (No. 11) reinforcing bars for longitu-
scale column specimens were designed with the same height dinal reinforcement and served as a control specimen, repre-
and same amount of transverse reinforcement, and a similar senting a conventional column. The other column (C43)
longitudinal reinforcement ratio. A new bar buckling model was designed with D43 (No. 14) bars for longitudinal rein-
was developed for bars laterally supported by multi-spiral forcement and served as a test specimen to investigate the
reinforcement and used to investigate the difference in buck- seismic performance of columns with large-diameter longi-
ling behavior of normal-diameter (36 mm [1.42 in.]) and tudinal reinforcement. The design details of both columns
large-diameter (43 mm [1.69 in.]) longitudinal bars. are listed in Table 1 and shown in Fig. 1. Both columns had
the same specified material properties. Column C36 was
designed to have a nominal flexural strength of 3923 kN-m
(2893.46 kip-ft), close to the nominal flexural strength of
@seismicisolation
@seismicisolation

Fig. 2—Dimensions and reinforcement details for: (a) column C36; (b) column C43; and (c) column cross section. (Note:
1 mm = 0.0394 in.)
column C43—that is, 3827 kN-m (2822.65 kip-ft). As height of the loading point from the column base (3600 mm
shown in Fig. 2, 24 longitudinal bars were used for column [141.73 in.]). Loading to each drift ratio was repeated three
C36. The number was significantly reduced to 16 for column times. The test setup was used to simulate the loading condi-
C43. The reinforcement ratio ρg for columns C36 and C43 tion of a column subjected to axial gravity and lateral seismic
was 2.98% and 2.87%, respectively. loading. During testing, crack development was monitored
Both columns were designed with the same type of five- by visual inspection. Strains of reinforcement were recorded
spiral transverse reinforcement consisting of a large spiral by strain gauges installed on reinforcement during specimen
with an outer diameter of 820 mm (32.28 in.) and four small fabrication. Deformation of the column was tracked using an
spirals with an outer diameter of 240 mm (9.45 in.). The optical motion tracking system.
maximum distance between the inner faces of small and
large spirals in the overlapping region was 97 mm (3.81 in.), Damage process and hysteresis behavior
larger than the smaller value of 0.3 times the inner diameter During testing, flexural cracks first appeared at the 0.25%
of the small spiral (0.3 × 240 = 72 mm [2.83 in.]) and 60 mm drift ratio for both columns. As the drift ratio increased,
(2.36 in.) as required by the reinforced concrete design code more flexural cracks occurred. Moreover, the cracks grad-
of Taiwan.14 D9 and D14 steel bars were used for small ually propagated towards the neutral axis of the column
and large spirals, respectively. The vertical center-to-center cross section. During drift ratios of 0.5 to 0.75%, flexural
spacing of small and large spirals was 75 mm (2.95 in.). The cracks started to turn diagonally into flexural-shear cracks
volumetric ratio of transverse reinforcement was 1.181%, for both columns. At the 1.5% drift ratio, most of the flexural
which was calculated based on the confinement requirement cracks had been developed for both columns. Figures 4(a)
for the plastic hinge region of a column.7,14,19 The columns and 5(a) showed one of the column faces perpendicular to
were designed to fail in flexure. The shears corresponding the loading direction at the 1.5% drift ratio for C36 and
to the nominal flexural capacities of columns C36 and C43 C43, respectively. At the 3% drift ratio, spalling concrete
were 1189 and 1160 kN (267.3 and 260.78 kip), respec- at the column toe appeared, as shown in Fig. 4(b) and 5(b)
tively—much lower than the nominal shear capacity9 of for C36 and C43, respectively. As the drift ratio increased,
2315 kN (520.43 kip) of both columns. the extent of concrete spalling increased. Until the 5% drift
The columns were tested with an axial load (Table 1) ratio, the damage progress of C36 was similar to that of C43
applied vertically to the top of the column first. The axial except for the crack spacing. The average spacing of cracks
load was maintained constant throughout the testing. Then, on the column faces perpendicular to the loading direction
displacement-controlled cyclic loading was applied laterally of C43 was approximately 1.25 times that of C36. This is
to the top of the column with the bottom of the column fixed because the maximum spacing of longitudinal bars in C43
to the strong floor. The columns were tested in a single- was 500 mm (19.69 in.), two times larger than in C36. The
curvature fashion, as shown in Fig. 3. The lateral cyclic cracking spacing has been shown to increase with increasing
loading contained drift ratios of 0.25, 0.375, 0.5, 0.75, 1.0, spacing of longitudinal bars.20 However, the difference in the
1.5, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0, and 9.0%. The drift ratio crack spacing appeared not to have a significant effect on the
is defined as the relative horizontal displacement between seismic performance of the column.
the column base and the point of lateral load divided@seismicisolation
by the
@seismicisolation

Fig. 3—(a) Test setup; (b) loading protocol; and (c) photo of test setup. (Note: 1 mm = 0.0394 in.)
Fig. 4—C36 damage pattern at: (a) 1.5% drift ratio; (b) 3% drift ratio; (c) buckling of longitudinal bars (7% drift ratio); and
(d) fracture of longitudinal bars (8% drift ratio).
At the 7% drift ratio, for C36, buckling of longitudinal strength dropped to 47% of the peak strength during the
bars occurred during the third cycle of loading, as shown 8% drift loading. The testing was terminated at the 9% drift
in Fig. 4(c). The maximum lateral strength of the column ratio when the lateral strength dropped to 38% of the peak
dropped to 83% of the peak strength during the 7% drift strength.
loading. At the 8% drift ratio, fracture of longitudinal bars For C43, buckling of longitudinal reinforcement also
occurred, as shown in Fig. 4(d). The maximum lateral occurred at the 7% drift but during the second cycle, slightly
@seismicisolation
@seismicisolation

Fig. 5—C43 damage pattern at: (a) 1.5% drift ratio; (b) 3% drift ratio; (c) buckling of longitudinal bars (7% drift ratio); and
(d) fracture of longitudinal bars (8% drift ratio).
Fig. 6—Force and displacement diagram for: (a) C36; and (b) C43. (Note: 1 kN = 0.22481 kip; 1 mm = 0.0394 in.)
earlier than C36. The maximum lateral strength decreased The relationship between the lateral force and drift ratio
to 84% of the peak strength during the 7% drift loading. (hysteretic behavior) of C36 and that of C43 are shown in
Figure 5(c) shows the damage condition of the column at Fig. 6. Note that the axial loading system tilted as the column
the 7% drift ratio. At this drift ratio, fracture of five-spiral deformed laterally during testing, causing an additional
reinforcement was observed. This was not found in C36. moment at the column base. This additional moment was
Fracture of longitudinal bars also occurred at the 8% drift removed from the force presented in Fig. 6. Both columns
ratio, causing the lateral strength to drop to 34% of the peak showed ductile behavior with drift capacities significantly
strength during the 8% drift loading. The testing was stopped higher than 3%, a seismic performance target used in the
at the 9% drift ratio when the lateral strength dropped to literature for columns.21 The difference in behavior between
23% of the peak strength. the two columns was not significant until the 7% drift ratio,
at which the longitudinal bars started to buckle. After the
@seismicisolation
@seismicisolation

Table 2—Displacement and moment capacities
Specimen Vy, kN ∆y, mm ∆u, mm μ Mtest, kN-m Mn, kN-m Mtest/Mn
C36 1243.09 37.58 282.99 7.53 4533 3907 1.16
C43 1279.56 41.31 267.11 6.47 4618 3832 1.21
Note: 1 kN = 0.22481 kip; 1 m = 3.28 ft; 1 mm = 0.0394 in.
Fig. 7—(a) Equivalent viscous damping ratio; (b) strain of longitudinal reinforcement; (c) strain of large spirals; and (d) strain
of small spirals.
7% drift ratio, the lateral strength of C43 degraded faster material properties,19 and Mtest/Mn are also listed in Table 2.
than that of C36. The number of longitudinal bar fractures C43 exhibited a slightly higher overstrength (Mtest/Mn) than
for C36 was the same as that for C43. However, because C36. In general, the performance difference between the two
the area of one longitudinal bar of C43 was larger than that columns was not significant.
of C36, the strength degradation for C43 related to each bar
fracture was higher than C36. Energy dissipation and reinforcement strain
The envelope of the hysteretic behavior was idealized The hysteretic energy dissipation in terms of the equiv-
by a bilinear relationship based on FEMA 365.22 The first alent viscous damping ratio for both columns is shown in
segment of the bilinear relationship intersected the envelope Fig. 7(a). C43 exhibited an energy dissipation capacity
response at 60% idealized yield force Vy, which is the force similar to C36. The use of D43 (No. 14) longitudinal bars
at the idealized yield point. The second segment ended at did not significantly alter the energy dissipation capacity of
the ultimate drift ratio ∆u, defined as the drift ratio when the the column. Fig. 7(b), (c), and (d) show the strain responses
envelope response drops to a point with a lateral force equal of the longitudinal reinforcement, the large spiral, and the
to 80% of the peak lateral strength. The intersection of the small spiral, respectively, from the plastic hinge region of
two linear segments, which defines the idealized yield point, the columns. The locations of the strain gauges used in these
was selected so that the area below the envelope response plots can be found in Fig. 2. The longitudinal reinforcement
and below the idealized bilinear relationship were similar. of C43 generally showed a smaller strain response than C36
The displacement ductility μ of each column is defined as for the same drift ratio. The longitudinal reinforcement of
the ratio of ∆u to the displacement of the idealized yield C43 yielded at the 1% drift ratio, later than C36, which
point ∆y. The Vy, ∆y, ∆u, and μ of the bilinear relationship occurred at the 0.75% drift ratio. However, at the 1.5% drift
for the two columns are listed in Table 2. C36 had a slightly ratio, the longitudinal reinforcement of both columns showed
larger displacement capacity (∆u and μ) than C43. The similar strain responses. The strain gauges of both columns
maximum column moment measured from testing Mtest, the failed later at this drift ratio. The strain responses of the large
nominal moment strength Mn calculated based on measured and small spirals of C43 were similar to those of C36. The
@seismicisolation
@seismicisolation

Fig. 8—Relationship between buckling length and spacing of transverse reinforcement in: (a) buckled bar restrained by spiral;
(b) Dhakal and Maekawa’s model; and (c) proposed model.
large and small spirals yielded early at drift ratios from 2 to applied compression P to the buckled bar. This is described
3% for both columns. The spirals were effectively mobilized by Eq. (1), in which the function for the deflected shape of
to confine the core concrete. The large spirals yielded earlier the buckled bar v is assumed to be Eq. (2)
than the small spirals. This is because the volumetric ratio
of the large spiral was smaller than the small spiral. This is U = U buckled bar + U spring − U P
typically the case for five-spiral reinforcement.
1  δ  n −1  2πi 
2 2
1 L 1 L
= ∫0 EI (v ′′ ) dx + k   ∑ 1 − cos − P ∫0 (v ′ ) 2 dx
2
   

PROPOSED BUCKLING MODEL FOR 2 2 2 i =1 n 2
LONGITUDINAL BARS (1)
As observed from testing and shown in Fig. 6, buckling
δ 2πx 
of longitudinal bars occurred during the 7% drift ratio. v= 1 − cos  (2)
The buckled bars started to fracture when stretched again 2 L 
in tension during the following drift ratio (8%), causing the where U is total energy of the buckling system; Ubuckled bar is
strength of the column to degrade significantly. This obser- energy stored in the buckled bar; Uspring is energy stored in
vation demonstrates that buckling of longitudinal bars has the equivalent springs; Up is potential energy due to work
an important effect on the ultimate drift ratio and displace- done by the applied compression to the buckled bar; E is
ment ductility of the columns. To investigate the buckling effective modulus of elasticity of longitudinal steel bars,
behavior of D36 (No. 11) and D43 (No. 14) longitudinal taken as 2.5% initial modulus of elasticity of longitudinal
bars constrained by five-spiral reinforcement, a buckling steel bars to consider the effect of inelastic behavior24; I is
model to determine the buckling length was developed in moment inertia of the buckled bar; v is shape function of
this research and presented as follows. buckling deflection; k is equivalent spring stiffness of a spiral
and will be discussed in the section “Stiffness of equivalent
Proposed buckling model spring for five-spiral reinforcement”; δ is maximum deflec-
The proposed buckling model was developed based tion of the buckled bar; n is number of buckling spacings;
on Dhakal and Maekawa’s model23 with modifications P is applied axial compressional force; and L is buckling
to remove the limitation that the buckling length has to length. As stated previously, the longitudinal bars of both
be integral multiples of transverse reinforcement spacing columns yielded early at the 0.75 to 1% drift ratios, much
and consider the unique constraint behavior of five-spiral earlier than the buckling of the reinforcement at the 7% drift
reinforcement. ratio. This supports the use of the effective modulus of elas-
ticity for longitudinal bars.
Dhakal and Maekawa’s model According to the principle of stationary total potential
Figure 8(b) shows Dhakal and Maekawa’s buckling energy, the stable equilibrium of the model is achieved
model, consisting of a laterally displaced bar supported by when the total potential energy of the system is minimum.25
equivalent springs. The deflected bar represents a longitu- Therefore, by taking the partial derivative of the total poten-
dinal bar buckled in compression, and the equivalent springs tial energy with respect to the maximum deflection of the
represent transverse reinforcement. The buckling length L is buckled bar δ and setting the derivative to be zero, the equa-
assumed to be an integral multiple of transverse reinforce- tion for buckling load P can be established.
ment spacing s—that is, L = n × s, where n is an integer.
Based on Dhakal and Maekawa’s model, the total poten- ∂U 4 EI π 2 kL n −1  2πi 
2
tial energy of the model consists of the energy stored in the = 0 ⇒ P = + ∑  1 − cos  (3)
∂δ L3 2π 2 i =1  n 
buckled bar, the energy stored in the equivalent springs, and
a decrease in the potential energy due to work done@seismicisolation
by the
@seismicisolation

As can be observed from Eq. (3), P varies with L. The The shape function v is the same as Eq. (2). Following the
L that produces the minimum P is the governing buckling same derivation as that for Dhakal and Maekawa’s model,
length. Knowing L = n × s, by taking the partial derivative Eq. (6) and (7) can be obtained
of P with respect to n and setting the derivative to be zero,
∂U 4 EI π 2 kL n +1  2π( sa + (i − 1) s ) 
2
Eq. (4) is obtained.24
=0⇒ P= + 2 ∑ 1 − cos 
∂δ L2
2π i =1  L
∂P EI π 4 16
=0⇒k = 3 (6)
∂n s n −1
 2πi   2πi  4πi 2πi 
n3 ∑ 1 − cos  1 − cos − sin
i =1  n   n  n n  ∂P
=0⇒
(4) ∂L
16 EI π 4
k=
As stated previously, n should be an integer. For each   2π ( sa + (i − 1) s )  
value of n, Eq. (4) can be used to determine the stiffness  L  cos − 1 + 4π ( sa + (i − 1) s ) × 
n +1   L  
of the equivalent spring k required for this buckling length L2 ∑  
L = n × s. The stiffness of the equivalent spring provided in i =1
  2π ( sa + (i − 1) s )   2π ( sa + (i − 1) s )  
a column is compared with the required k for various values  sin   cos  − 1 
  L   L 
of n. If the provided k is between the required k for L =
(n – 1) × s and that for L = n × s, the provided k is suffi- (7)
cient to restrain a buckling length of n × s but not enough
to restrain a buckling length of (n – 1) × s. Because n is an With the k and n already known, sa can be solved; and the
integer, the L is conservatively assumed to be n × s in Dhakal buckling length L can be obtained as (n – 1) × s + 2sa.
and Maekawa’s model. However, the actual buckling length
is between (n – 1) × s and n × s. To consider this, in the Stiffness of equivalent spring for five-spiral
proposed model, the locations of the equivalent springs are reinforcement
rearranged, as shown in Fig. 8(c). The L is set equal to (n – 1) As shown in Fig. 9(a), the cross section of C36 is used
× s + 2sa and has a value range of (n – 1) × s < L ≤ n × s. This to illustrate that in a five-spiral column, longitudinal bars
makes the buckling length to be more precise. can be either constrained by a large spiral, a small spiral, or
an overlapping region. According to Dhakal and Su,26 longi-
Proposed model tudinal bars that buckle first in a spiral/circular hoop are
Before using the proposed model, Dhakal and Maeka- those in the first layer from the extreme compression fiber,
wa’s model is used first to determine the value of n. Then, and those overlap with the first layer of bars on a projected
by resetting L = (n – 1) × s + 2sa with sa unknown, the total straight line along the buckling direction. Based on this
potential energy of the proposed model is concept, the longitudinal bars of the five-spiral column
U = U buckled bar + U spring − U P
shown in Fig. 9(a) are projected on a line along the loading
2
direction (buckling direction). It can be seen that the two
1  δ  n +1  2π( sa + (i − 1) s ) 
2
=
1 L
∫ EI (v ′′ ) dx + k   ∑  1 − cos
2 1 L
 − 2 P ∫0 (v ′ ) dx
2 bars (shown in blue color) in each of the two small spirals
2 0 2  2  i =1  L in the compression side and the one bar (shown in red color)
(5) in the large spiral are prone to buckle. Bars in overlapping
regions are assumed not to buckle because they are further
where sa is additional buckling length beyond (n – 1) × s. away from the extreme compression fiber and are subjected
Fig. 9—(a) Locations of buckled bars; (b) deformed parts of small and large spirals; and (c) equilibrium to determine pushing
force from buckled bar. (Note: Full-color PDF can be found at www.concrete.org.)
@seismicisolation
@seismicisolation

Table 3—Buckling parameters of specimens
Specimens k, MPa-mm S, mm n sa, mm L, mm Ltest, mm L/db
C36 7 18.85 487.70 487.5 13.547
284.15 75
C43 8 28.80 582.59 562.5 13.549
Note: 1 MPa-mm = 5.7101 psi-in.; 1 mm = 0.0394 in.
to strong confinement from a large and a small spiral. More- sectional area of a spiral. For the large spiral of C36 and C43,
over, because the overlapping region is subjected to strong nb = 1 and 0, respectively. For the small spiral of C36 and
confinement and is typically designed with longitudinal bars C43, nb = 2. As stated previously, the spirals of the columns
to enhance the capability of the region to interlock the large tested yielded at the 2 to 3% drift ratios, earlier than the
and small spirals, the overlapping region is assumed to act as buckling of the longitudinal bars, which occurred at the 7%
a stiff column and hence provides a strong anchorage for the drift ratio. This supports the use of the post-yield modulus of
small spiral anchored to the region. As a result, buckling of elasticity for spirals.
the two longitudinal bars in each small spiral is assumed not The stiffness of the equivalent spring k for the large or
to affect the large spiral. Due to different constraints from the small spiral can be obtained by the following equation with
large or small spiral and different numbers of buckled bars F from Eq. (9)
within the large or small spiral, the stiffness of equivalent
θ 
springs for the large spiral and that for the small spiral need
to be calculated and compared. The bar within a spiral with F
( )
Et At cos ( π − θ m ) + 1  m 
 2π  (10)
k= =
a smaller stiffness buckles first. δ Dnb
When the bar in the large or small spiral buckles, the large
where k is equivalent spring stiffness of a spiral. For the large
or small spiral is pushed to deform outwardly, as shown in
spiral of C36, k = 563.34 MPa-mm (3216.75 psi-in.) based
Fig. 9(b). Due to the constraint by concrete, only a portion
on Eq. (10). The large spiral of C43 has no longitudinal bars
of the spiral is deformed. The deformed part of the large
for buckling consideration (nb = 0). For the small spiral of
spiral is assumed to be one-half of the large spiral and has
C36 and C43, k = 284.15 MPa-mm (1622.54 psi-in.) based
a central angle θm of 180 degrees. This assumption is the
on Eq. (10). Therefore, the governing value of k for both C36
same as that used by Dhakal and Su.26 The effect of the small
and C43 is 284.15 MPa-mm (1622.54 psi-in.). And, for both
spiral is ignored. The deformed part of the small spiral is
columns, buckling of longitudinal bars in small spirals is the
assumed to be one-half of the small spiral, excluding the
governing buckling behavior.
overlapping region, as shown in Fig. 9(b). The portion of the
small spiral within the overlapping region is assumed not
Comparison with test observation
to deform due to the strong confinement effect. The central
Table 3 lists the values of important parameters for the
angle of the deformed part is smaller than 180 degrees. With
analytical buckling behavior of the D36 (No. 11) and D43
these assumptions, the tensile strain developed in the large
(No. 14) bars of C36 and C43, respectively. As stated previ-
or small spiral due to the push from the buckled bar can be
ously, the governing value of k is from the small spiral.
calculated as
This means buckling of longitudinal bars within the small
spiral governs the buckling behavior. This is consistent with
θm ( D + δ) − θm D θmδ the experimental observation, as shown in Fig. 4 and 5 for
ε = = (8)
2πD 2πD C36 and C43, respectively. The longitudinal bars within the
large spiral of C36 did not show significant buckling until
where ε is strain of a spiral due to bar buckling; θm is angle
the end of testing. Note that for C43, no longitudinal bars
corresponding to the deformed part of a spiral; and D is
were designed within the large spiral close to the extreme
diameter of a spiral. For the large spiral of C36 and C43,
compression fiber of the section (Fig. 2). Experimental
θm = 180 degrees and D = 820 mm (32.28 in.). For the small
observation also showed that longitudinal bars within the
spiral of C36 and C43, θm = 143 degrees and D = 240 mm
overlapping regions did not expose and were still well
(9.45 in.).
confined by surrounding concrete and spirals until the end
From the free-body diagram shown in Fig. 9(c), the
of testing. This is consistent with the assumption used in
pushing force from the buckled bar F can be determined by
the proposed buckling model that the overlapping regions
equilibrium with ε from Eq. (8)
provide a strong anchorage for small spirals to restrain buck-
ling of longitudinal reinforcement.
θm δ
( ) ( )
nb F = Et At ε cos ( π − θ m ) + 1 = Et At cos ( π − θ m ) + 1
2πD
The calculated buckling lengths are 477.70 and 582.59 mm
(18.81 and 22.94 in.) for C36 and C43, respectively, close
(9) to measured buckling lengths—that is, 487.5 and 562.5 mm
(19.19 and 22.14 in.), respectively—from the experiment.
where nb is number of buckled bars; Et is post-yield Because the test data are still limited, more tests are needed
modulus of elasticity of spiral steel, taken as 1.5% the to further validate the applicability of the proposed model to
initial modulus of elasticity of spiral steel ; and At is cross-
27 the buckling behavior of longitudinal bars laterally supported
by five-spiral reinforcement. According to Dhakal and
@seismicisolation
@seismicisolation

Moreover, a buckling model was developed for longitu-
dinal bars laterally supported by five-spiral reinforcement.
Conclusions are summarized as follows.
1. Both columns showed similar damage progress,
force-displacement behavior, and energy dissipation until
the 7% drift ratio. Buckling of longitudinal reinforcement
of both columns started at the same drift ratio (7%). D43
(No. 14) bars buckled slightly earlier than D36 (No. 11) bars.
Because the total cross-sectional area of D43 (No. 14) bars
of column C43 in the buckling critical region (within the
small spiral) was higher than that of column C36, column
C43 showed a faster strength degradation after buck-
ling than column C36. As a result, column C36 showed a
slightly larger displacement capacity than column C43. Both
columns exhibited flexural-dominated behavior with good
displacement capacities and energy dissipation. The perfor-
Fig. 10—Relationships between buckling slenderness ratio mance difference between the two columns was, in general,
and center-to-center spacing of spiral. (Note: 1 mm = not significant.
0.0394 in.) 2. The average spacing of flexural cracks of C43 was
approximately 1.25 times that of C36. This is because the
Maekawa,28 the compressive stress-strain behavior of longi- maximum spacing of longitudinal bars of C43 was two
tudinal bars in compression mainly depends on the buck- times larger than C36. However, the difference in the crack
ling slenderness ratio, defined as the buckling length divided spacing appeared not to have a significant effect on the
by the diameter of the longitudinal bar L/db. As shown in seismic performance of the column.
Table 3, the calculated L/db for D43 (No. 14) bars in C43 is 3. Based on the proposed buckling model, the buckling
similar to that for D36 (No. 11) bars in C36. This is consis- of longitudinal bars within the small spiral governs both
tent with the experimental observation that both columns columns’ buckling behavior. This is consistent with the
showed buckling of longitudinal bars at the same drift ratio experimental observation. The calculated buckling length
(7%). Note that the lateral strength of C43 degraded faster for D36 (No. 11) and that for D43 (No. 14) bars are close
after buckling than that of C36. This is mainly because for to the measured buckling lengths from the experiment. The
C43, the cross-sectional area of longitudinal bars within the calculated buckling slenderness ratio of D36 (No. 11) bars
small spiral was higher than that for C36. is similar to that of D43 (No. 14) bars, which means both
With the proposed buckling model and the same equiva- bars are expected to have a similar buckling behavior. This
lent spring stiffness as used for C36 and C43, the relation- is consistent with the experimental observation that buckling
ship between L/db and the center-to-center spacing of the of longitudinal reinforcement started at the same drift ratio
spiral s for D36 (No. 11) bars and that for D43 (No. 14) bars (7%). For the same amount and within the practical spacing
were calculated and shown in Fig. 10. The value of s ranges range of five-spiral reinforcement, the calculated buckling
from 30 to 200 mm (1.18 to 7.87 in.). According to ACI slenderness ratio of D43 (No. 14) bars is similar to that of
318-19,19 the clear spacing of spirals should not be greater D36 (No. 11) bars.
than 76.2 mm (3 in.), which corresponds to s = 85.2 mm
(3.35 in.) for the case of the small spiral used in C36 and AUTHOR BIOS
C43. From Fig. 10, it can be seen that with the same amount Yu-Chen Ou is a Distinguished Professor of Civil Engineering at National
and spacing of five-spiral reinforcement, the buckling slen- Taiwan University, Taipei, Taiwan. He received his PhD from the State
University of New York at Buffalo, Buffalo, NY. His research interests
derness ratios of D36 and D43 (No. 11 and No. 14) bars are include reinforced concrete structures and earthquake engineering. He is
very similar for the values of s within the upper limit by the President of the ACI Taiwan Chapter.
ACI 318-19. The difference becomes notable only when the
Brandon Li received his BS and MS from the Department of Civil Engi-
value of s is larger than approximately 190 mm (7.48 in.), far neering, National Taiwan University. His research interests include rein-
beyond the upper limit set by ACI 318-19 for spirals. This forced concrete structures and buckling behavior of longitudinal bars.
means D43 (No. 14) bars have a similar compressive stress-
strain behavior to D36 (No. 11) bars for the practical range ACKNOWLEDGMENTS
of s of five-spiral reinforcement. The authors would like to thank the financial support from the Ministry of
Science and Technology of Taiwan under Contract No. 109-2923-M-002-
006-MY3 and from Ruentex Engineering & Construction Co.
CONCLUSIONS
Two full-scale five-spiral columns were tested to investi- NOTATION
gate the difference in seismic performance between a column Ag = gross area of concrete cross section
with normal-diameter longitudinal bars (D36 [No. 11]) bars At = area of an individual transverse reinforcement
D = outside diameter of longitudinal bar
with a bar diameter of 36 mm (1.42 in.) (column C36) and db = diameter of longitudinal bar
that with large-diameter longitudinal bars (D43 [No. 14]) E = Young’s modulus of longitudinal bar
bars with a bar diameter of 43 mm (1.69 in.) (column C43). Et = Young’s modulus of transverse reinforcement
fc'
@seismicisolation= specified compressive strength of concrete
@seismicisolation

fca' = actual compressive strength of concrete 10. Wu, T.-L.; Ou, Y.-C.; Yen-Liang Yin, S.; Wang, J.-C.; Wang, P.-H.;
fy = specified yield strength of longitudinal reinforcement and Ngo, S.-H, “Behavior of Oblong and Rectangular Bridge Columns
fya = actual yield strength of longitudinal reinforcement with Conventional Tie and Multi-Spiral Transverse Reinforcement under
fyt = specified yield strength of transverse reinforcement Combined Axial and Flexural Loading,” Zhongguo Gongcheng Xuekan,
fyta = actual yield strength of transverse reinforcement V. 36, No. 8, 2013, pp. 980-993. doi: 10.1080/02533839.2012.747047
I = moment inertia of longitudinal bar 11. Ou, Y. C.; Ngo, S. H.; Yin, S. Y.; Wang, J. C.; and Wang, P. H., “Shear
i = order of spring Behavior of Oblong Bridge Columns with Innovative Seven-Spiral Trans-
k = equivalent stiffness of transverse reinforcement verse Reinforcement,” ACI Structural Journal, V. 111, No. 6, Nov.-Dec.
kLS = equivalent stiffness of large spiral reinforcement 2014, pp. 1339-1349. doi: 10.14359/51686873
kMSR = equivalent stiffness of multi-spiral reinforcement 12. Ou, Y. C.; Ngo, S. H.; Roh, H.; Yin, S. Y.; Wang, J. C.; and
kSS = equivalent stiffness of small spiral reinforcement Wang, P. H., “Seismic Performance of Concrete Columns with Innovative
kc = equivalent stiffness of circular/spiral reinforcement Seven- and Eleven-Spiral Reinforcement,” ACI Structural Journal, V. 112,
L = buckling length calculated from proposed model No. 5, Sept.-Oct. 2015, pp. 579-592. doi: 10.14359/51687706
Ltest = buckling length obtained from experiment 13. Ou, Y. C., and Ngo, S. H., “Discrete Shear Strength of Two- and
Mn = nominal moment strength Seven-Circular-Hoop and Spiral Transverse Reinforcement,” ACI
Mtest = maximum moment strength obtained from experiment Structural Journal, V. 113, No. 2, Mar.-Apr. 2016, pp. 227-238. doi:
n = number of buckling spacing 10.14359/51688058
nb = number of buckling participant longitudinal bars 14. CPAMI, “Design Specifications for Reinforced Concrete Structures,”
P = applied axial compressional force Construction and Planning Agency, Ministry of the Interior, R.O.C., 2019.
s = spacing of transverse reinforcement 15. Ichinose, T.; Kanayama, Y.; Inoue, Y.; and Bolander, J. E., Jr.,
sa = additional buckling length beyond (n – 1) × s “Size Effect on Bond Strength of Deformed Bars,” Construction and
U = total energy of buckling system Building Materials, V. 18, No. 7, 2004, pp. 549-558. doi: 10.1016/j.
Ubuckled bar = energy stored in buckled bar conbuildmat.2004.03.014
Up = potential energy due to work done by applied compression to 16. Steuck, K. P.; Eberhard, M. O.; and Stanton, J. F., “Anchorage of
buckled bar Large-Diameter Reinforcing Bars in Ducts,” ACI Structural Journal,
Uspring = energy stored in equivalent springs V. 106, No. 4, July-Aug. 2009, pp. 506-513. doi: 10.14359/56616
Vy = idealized yield strength 17. Murcia-Delso, J.; Stavridis, A.; and Shing, P. B., “Tension
v = shape function of buckling deformation Development Length of Large-Diameter Bars for Severe Cyclic Loading,”
∆u = ultimate displacement ACI Structural Journal, V. 112, No. 6, Nov.-Dec. 2015, pp. 659-669. doi:
∆y = yield displacement 10.14359/51687937
δ = maximum deflection of buckled bar 18. Murcia-Delso, J.; Liu, Y.; and Shing, P. B., “Required Development
μ = member ductility Length of Column Reinforcement Extended into Enlarged Pile Shafts,”
ρg = ratio of longitudinal reinforcement area to gross area Seventh National Seismic Conference on Bridges & Highways, Oakland,
ρs = volumetric ratio of transverse reinforcement CA, May 2013.
19. ACI Committee 318, “Building Code Requirements for Structural
Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved
REFERENCES 2022),” American Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
1. Tanaka, H., and Park, R., “Seismic Design and Behavior of Reinforced 20. Frosch, R. J., “Another Look at Cracking and Crack Control in Rein-
Concrete Columns with Interlocking Spirals,” ACI Structural Journal, forced Concrete,” ACI Structural Journal, V. 96, No. 3, May-June 1999,
V. 90, No. 2, Mar.-Apr. 1993, pp. 192-203. pp. 437-442.
2. McLean, D. I., and Buckingham, G. C., “Seismic Performance of 21. Elwood, K. J.; Maffei, J.; Riederer, K. A.; and Telleen, K., “Improving
Bridge Columns with Interlocking Spiral Reinforcement,” Report No. Column Confinement Part 1: Assessment of Design Provisions,” Concrete
WA-RD 357.1, Washington State Transportation Center, Seattle, WA, 1994. International, V. 31, No. 11, Nov. 2009, pp. 32-39.
3. Otaki, T., and Kuroiwa, T., “Test of Bridge Columns with Interlocking 22. FEMA 356, “Prestandard and Commentary for the Seismic Rehabil-
Spiral Reinforcement and Conventional Rectangular Hoop with Ties,” itation of Buildings in Rehabilitation Requirements,” Federal Emergency
Reports of the Technological Research Institute, Construction Co. Ltd, Management Agency, Washington, DC, 2000.
Tokyo, Japan, No. 25, 1999, pp. 33-38. 23. Dhakal, R. P., and Maekawa, K., “Reinforcement Stability and
4. Shito, K.; Igase, Y.; Mizugami, Y.; Ohasi, G.; Miyagi, T.; and Fracture of Concrete Cover in Reinforced Concrete Members,” Journal of
Kuroiwa, T., “Seismic Performance of Bridge Columns with Interlocking Structural Engineering, ASCE, V. 128, No. 10, 2002, pp. 1253-1262. doi:
Spiral/Hoop Reinforcements,” First fib Congress, Osaka, Japan, 2002. 10.1061/(ASCE)0733-9445(2002)128:10(1253)
5. Kawashima, K., “Enhancement of Flexural Ductility of Reinforced 24. Su, J.; Wang, J.; Bai, Z.; Wang, W.; and Zhao, D., “Influence of Rein-
Concrete Bridge Columns,” First International Conference on Urban forcement Buckling on the Seismic Performance of Reinforced Concrete
Earthquake Engineering, Tokyo Institute of Technology, Tokyo, Japan, Columns,” Engineering Structures, V. 103, 2015, pp. 174-188. doi:
2004, pp. 85-95. 10.1016/j.engstruct.2015.09.007
6. Correal, J. F.; Saiidi, M. S.; Sanders, D.; and El-Azazy, S., “Shake 25. Chen, W. F., and Lui, E. M., Structural Stability Theory and Imple-
Table Studies of Bridge Columns with Double Interlocking Spirals,” ACI mentation, Elsevier Science Publishing Co., Inc., New York, 1987.
Structural Journal, V. 104, No. 4, July-Aug. 2007, pp. 393-401. 26. Dhakal, R. P., and Su, J., “Design of Transverse Reinforcement to
7. Yin, S. Y. L.; Wu, T. L.; Liu, T. C.; Sheikh, S. A.; and Wang, R., “Inter- Avoid Premature Buckling of Main Bars,” Earthquake Engineering &
locking Spiral Confinement for Rectangular Columns,” Concrete Interna- Structural Dynamics, V. 47, No. 1, 2018, pp. 147-168. doi: 10.1002/
tional, V. 33, No. 12, Dec. 2011, pp. 38-45. eqe.2944
8. Yin, S. Y. L.; Wang, J. C.; and Wang, P. H., “Development of Multi- 27. Bae, S.; Mieses, A. M.; and Bayrak, O., “Inelastic Buckling of Rein-
Spiral Confinements in Rectangular Columns for Construction Automa- forcing Bars,” Journal of Structural Engineering, ASCE, V. 131, No. 2,
tion,” Zhongguo Gongcheng Xuekan, V. 35, No. 3, 2012, pp. 309-320. doi: 2005, pp. 314-321. doi: 10.1061/(ASCE)0733-9445(2005)131:2(314)
10.1080/02533839.2012.655528 28. Dhakal, R. P., and Maekawa, K., “Modeling for Postyield Buckling of
9. Ou, Y. C.; Li, J. Y.; and Roh, H., “Shear Strength of Reinforced Reinforcement,” Journal of Structural Engineering, ASCE, V. 128, No. 9,
Concrete Columns with Five-Spiral Reinforcement,” Engineering Struc- 2002, pp. 1139-1147. doi: 10.1061/(ASCE)0733-9445(2002)128:9(1139)
tures, V. 233, 2021, p. 111929. doi: 10.1016/j.engstruct.2021.111929
@seismicisolation
@seismicisolation

We’re Building the Future
Mission: We make strategic investments in ideas, research, and
people to create the future of the concrete industry
Through its councils and programs, the ACI Foundation helps to keep
the concrete industry at the forefront of advances in material composition,
design, and construction. Our focus:
Our Concrete Innovation Council

identifies technologies and innovation
 that are aligned with ACI and industry

strategies and helps facilitate their use
when appropriate
Our Concrete Research Council advances

the knowledge and sustainable aspects
 of concrete materials, construction,
and structures by soliciting, selecting,
financing, and publishing research
Our Scholarship Council supports our

future concrete innovators and leaders
by administering fellowships and scholar-
ships to help bridge the financial gap for
students
Our Veterans Rebate for ACI Certification

program helps honorably discharged
veterans and increases skills in the
industry’s workforce. ACICertification.org/
veteranrebate
@seismicisolation
@seismicisolation
Building the Future
Title No. 120-S81
Rehabilitation of Exterior Beam-Column Joint by

Geopolymer Mortar under Quasi-Static Loading
by Arshad Hussain Choudhury and Aminul Islam Laskar
Most of the studies conducted on the rehabilitation of reinforced to establish a strength hierarchy between members framing
concrete (RC) beam-column joints are on pre-1970 structures. into the joint, thereby preventing brittle joint failure in the
Recently, it was reported that seismically designed beam-column column and promoting ductile beam failure (strong-column,
joints might also suffer damage under lateral loading. On the other weak-beam).8 Several techniques such as concrete jacketing,
hand, there is an increasing interest among researchers to study
steel jacketing, haunch solution, fiber-reinforced polymer
the effectiveness of geopolymer as an alternative repair material.
(FRP) wrapping, removal and replacement technique, and so
To date, no study has been conducted to examine the performance
of geopolymer for the rehabilitation of seismically detailed beam- on are available to rehabilitate damaged beam-column joint.
column joints following the removal and replacement method under Concrete jacketing has been found successful in restoring
cyclic loading. In the present investigation, two groups of exterior the cyclic performance of damaged beam-column joints.1
beam-column joints with different flexural strength ratios were However, concrete jacketing is a time-intensive process that
rehabilitated with geopolymer mortar. For comparison, another set increases the structural weight and size of the members.
of beam-column joints (one from each group) were rehabilitated Construction time decreases due to the use of prefabricated
with cement mortar following the same rehabilitation technique steel plates and steel haunch retrofit systems. Both tech-
and testing. Test results indicated that geopolymer rehabilitated niques change the brittle joint shear of control specimens
specimens exhibited 20 to 21% higher initial stiffness, 19 to 22% to ductile beam failure of rehabilitated specimens.9-11 Along
higher displacement ductility, 24 to 37% higher cumulative energy
with increased weight and size of repaired joint, corrosion
dissipation, 14 to 17% higher initial equivalent viscous damping
is a major concern. Moreover, drilling through slabs poses
ratio, 21 to 26% higher ultimate equivalent viscous damping ratio
at failure, and 10 to 14% lower damage index compared to spec- a practical implementation problem for haunch retrofit
imens rehabilitated with cement mortar. However, irrespective systems. FRP has shown exemplary performance in the
of repair material, removal and replacement technique was only rehabilitation of damaged beam-column joints.1 Neverthe-
able to partially restore the cyclic performance of rehabilitated less, high cost, debonding due to incompatibility between
specimens. the epoxy and concrete substrate, decreased performance in
high temperatures, and wet environments are some of the
Keywords: beam-column joint; cyclic loading; geopolymer; removal and demerits associated with FRP.
replacement method; seismically detailed.
FEMA 30812 suggests that the removal and replacement
technique should be carried out for rehabilitation of highly
INTRODUCTION
damaged specimens subjected to severe earthquakes. In
An abundant number of literatures are available on the
this technique, damaged concrete is removed and replaced
repair and strengthening of beam-column joints damaged
with high-strength, low-shrinkage repair material.1,13-16 Lee
under earthquake loading. Most of the studies are limited to
et al.13 adopted a removal and replacement method to repair
substandard pre-1970 structures with inadequate detailing.1
extensively damaged exterior beam-column joints under
With the advent of the seismic code of practice, beam-
severe earthquake loading. Concrete of the damaged joint
column joints are being constructed with special confining
was removed followed by subsequent replacement with
hoop reinforcement and adequate anchorage. Such seismi-
higher strength concrete (compressive strength 48 MPa
cally detailed beam-column joints exhibit better strength,
[6.96 ksi]). Experimental results showed that the removal
ductility, and higher energy dissipation capacity.2 Recently,
and replacement method was able to restore structural integ-
some studies revealed that reinforced concrete (RC) struc-
rity of the rehabilitated specimens up to control specimens.
tures designed as per seismic code of practice exhibited
Karayannis et al.14 removed the fragmented joint concrete
damage in the joint region when maximum force acting
of exterior beam-column subassemblies. Voids were filled
on the joint exceeded the code expectation level.3-5 Only a
with high-strength (83 MPa [12.04 ksi]), rapid-hardening,
limited number of studies are available on the rehabilita-
and low-shrinkage cement paste. When tested under cyclic
tion of seismically designed and detailed structures.3,6,7 The
loading, a significant increase in ultimate load, stiffness,
failure of such structures opened new research avenues in
the field of rehabilitation of ductile beam-column joints. ACI Structural Journal, V. 120, No. 5, September 2023.
Rehabilitation of damaged structures provides an econom- MS No. S-2021-494.R2, doi: 10.14359/51738835, received December 2, 2022, and
ically viable alternative to demolition and reconstruction. Institute. All rights reserved, including the making of copies unless permission is
The aim of seismic rehabilitation of beam-column joints is closure, if any, will be published ten months from this journal’s date if the discussion
@seismicisolation
@seismicisolation

and energy dissipation was observed in repaired speci- The present study was undertaken to investigate the effec-
mens. Tsonos15 repaired two exterior beam-column joints tiveness of the removal and replacement method in reha-
by complete removal of joint core concrete, followed by bilitation of seismically detailed beam-column joints using
replacement with non-shrink, higher-compressive-strength geopolymer. Results were compared with another set of the
(70 MPa [10.15 ksi]) mortar. Results indicated the rehabil- same specimens repaired with cement mortar adopting same
itation scheme was successful in enhancing cyclic perfor- repair technique.
mance of the rehabilitated test specimens. Marthong et al.16
removed concrete entirely from the affected joint panel of a RESEARCH SIGNIFICANCE
beam-column subassembly damaged under severe loading. Though the removal and replacement method of rehabil-
Damaged concrete was replaced by commercially available itation has performed satisfactorily for substandard beam-
micro-concrete (compressive strength 50 MPa [7.25 ksi]). column joints, the performance of this method is unknown
Repaired specimens exhibited higher load-carrying capacity, for seismically designed and detailed beam-column joints.
displacement ductility, energy dissipation, and lower stiff- In the last decade, geopolymer has shown immense poten-
ness degradation. It is worthwhile to mention that all exte- tial as an alternate repair material for structures subjected to
rior beam-column joints repaired and strengthened by the monotonic loading, but repair studies concerning structures
removal and replacement technique were substandard under cyclic loading are scarce. The findings of this paper
in nature and were lacking ductile detailing provisions, will provide insight into the performance of the removal
according to seismic code of practice. and replacement technique for rehabilitation of seismically
In late 1970, Davidovits17 developed a cementless inor- detailed beam-column joints subjected to cyclic loading
ganic binder known as geopolymer. Geopolymer is formed using geopolymer.
when aluminosilicate-sourced materials react with alka-
line activator, resulting in an amorphous alkali aluminos- MATERIALS AND METHODS
ilicate with a three-dimensional polymeric chain of sialate Materials for beam-column subassembly
(–Si–O–Al–O–). Use of industrial by-products such as Grade M20 concrete22 was prepared by mixing Grade
ground-granulated blast-furnace slag (GGBS) and fly ash 43 OPC, crushed coarse aggregate 10 mm (0.39 in.) down,
as source material contributed to lower CO2 emissions and locally available river sand, and potable water. Grade Fe 550
energy consumption during the production stage, making reinforcement of was used in the longitudinal direction and
geopolymer an eco-friendly green material. Numerous Grade Fe 250 reinforcing bar was provided in the transverse
research has been conducted on geopolymer since its direction. Upon laboratory testing as per IS:1608-2005,23 the
development. Geopolymer has better mechanical properties yield stress of Grades Fe 550 and Fe 250 was found to be
including high early strength, low shrinkage, acid resistance, 590 MPa (85.55 ksi) and 300 MPa (43.5 ksi). All the ingre-
durability, and thermal stability when compared to ordinary dients of concrete were machine mixed in a 50 L (13.2 gal.)
portland cement (OPC).18 This makes geopolymer a poten- pan-type concrete mixer.
tial alternative repair material to cement-based materials.
Vasconcelos et al.19 worked out a cheap metakaolin-based REPAIR MATERIALS
repair material to retrofit concrete slab and beam. The Geopolymer mortar (GM)
metakaolin geopolymer mortar displayed high mechanical GGBS, in combination with Class F FA, was used as
resistance and relevant adhesion to concrete substrate. Duan a binder in the geopolymer mixture. The proportion of
et al.20 proposed a geopolymeric material having low setting GGBS to FA was fixed at 80:20 for all trial mixtures. A
time, high compressive strength, high bond strength, and higher proportion of GGBS was used as it is the primary
better water resistance capacity to repair marine structures binding agent and can contribute to strength grain in ambient
and concrete roads. Phoo-Ngernkham et al.21 observed that temperature.24 FA was used as an additive to regulate setting
compared to commercial repair binders, fly ash (FA)-based time. The chemical composition and physical properties
geopolymer mortar exhibited a denser interface transition of GGBS and Class F FA are shown in Table 1. A sodium
zone with higher bond, shear, and bending strength when hydroxide (NaOH) pellet (specific gravity 2.13 and purity of
used as repair binder. Most geopolymer-based repair was 97%) was used as an alkali activator.
associated with structures subjected to monotonic loading. For selecting a suitable geopolymer mixture to carry out
It may be summarized from the literature review that beam- rehabilitation, three sets of trial mixtures (T1, T2, and T3)
column subassemblies rehabilitated so far by the removal
and replacement technique were substandard in nature and Table 1—Chemical composition of GGBS and FA
lacked ductile detailing provisions, according to the seismic
Chemical composition GGBS, % by mass FA, % by mass
code of practice. Seismically detailed beam-column joints
suffered joint damage under lateral loading in some cases. Silicon dioxide (SiO 2) 35 66.39
No study was undertaken to explore the effectiveness of the Aluminium oxide (Al2O3) 20 22.63
removal and replacement technique in the rehabilitation of Ferric oxide (Fe2O3) 2 5.30
seismically detailed beam-column joints. Further, perfor-
Calcium oxide (CaO) 34 0.67
mance of geopolymer mortar as repair material under cyclic
loading for seismically detailed beam-column joint is yet to Sulfur oxide (SO3) 0.8 0.41
be investigated. Magnesium oxide (MgO) 8 0.16
@seismicisolation
@seismicisolation

Table 2—Properties of geopolymer trial mixtures
Compressive strength, MPa
Trial Initial setting time, Final setting time,
mixture Molarity 3 days 7 days 28 days min min
T1 8M 28 35 43 24 60
T2 10M 30 40 46 20 50
T3 12M 33 42 48 15 42
Note: 1 MPa = 0.145 ksi.
with various molarities—8M, 10M, and 12M—of activator strength ratio, test specimens were grouped into two catego-
solution were prepared. The binder (GGBS + FA) to sand ries: Group 1 (flexural strength ratios = 1.2); and Group 2
proportion was maintained at 1:3. For rapid strength gain (flexural strength ratios = 1.4). The flexural strength ratio of
and high compressive strength, the ratio of alkali to binder a joint is defined as the ratio of summation of nominal design
was kept constant at 0.65.25 The codal provisions of IS strength of all the columns to the summation of nominal
1727-195726 were followed to prepare trial mixtures in stan- design strength of all the beams framing into the joint.32
dard 50 cm (7.75 in.) mortar cubes and to evaluate setting Such flexural strength ratios were achieved by varying beam
times. Test results of trial mixtures are provided in Table 2. and column reinforcement. The structural design of beam
According to Duan et al.20 and Yun and Choi,27 desired prop- and column components of specimens was done as per
erties for suitable repair materials are early strength, rapid provisions laid down in IS 456-2000.33 The joint panel of the
setting (low initial and final setting time), and high compres- specimens was designed and detailed as per codal provisions
sive strength. Among all the trial mixtures, T3 displayed the of IS 13920-2016.32 Special confinement reinforcement in
earliest strength gain, least initial and final setting time, and the form of rectangular hoop reinforcement was provided in
highest compressive strength. Therefore, T3 was selected as and around the joint region by taking into the provision of
the most suitable repair material to carry out the rehabilita- codal requirement of IS:13920-2016.32 The cross-sectional
tion work. area of the bar (Ash) of rectangular hoop reinforcement as
per IS:13920-201632 is calculated by the following equation
Cement mortar (CM)
[ fy ]
= maximum of 0 . 18 svh [A − 1]or ⌊0.05svh   ⌋
Grade 43 OPC was used to prepare mortar mixture for fck
_ _ Ag fck
_
repair work.28 Cement to sand ratio was maintained at Ash
fy k
1:3 for mortar preparation, and water-cement ratio (w/c) (1)
was fixed at 0.35. For higher workability, a high-range
water-reducing admixture (HRWRA) at the rate of 1.5% by where sv is center-to-center spacing of stirrup; h is longer
weight of cement was used in the mixture. All these ingredi- length of rectangular stirrup measured up to outer face; Ag is
ents were mixed to prepare 50 cm (7.75 in.) mortar cubes to gross cross-sectional area of column; Ak is area of concrete
determine the compressive strength. Cubes were demolded confined in the core; fck is characteristic compressive strength
after 24 hours and cured under water. Compressive strength of concrete; and fy is 0.02% proof strength of transverse steel
was determined as per IS:1727-196726 and was found to be reinforcement bars.
25 MPa (3.63 ksi), 34 MPa (4.93 ksi), and 52 MPa (7.54 ksi) Adequate anchorage length was provided following
at the third, seventh, and 28th days, respectively. IS:13920-201632 to prevent pullout failure of beam rein-
forcement. In all specimens, both top and bottom bars of
Test specimens beams were bent at 90 degrees heading towards joint core.
In the present study, one-third scaled-down beam-column As per IS:13920-2016,32 the total anchorage length was
joint specimens were used. ACI 374.1-0529 allows scaling determined by Eq. (2)
down of specimens up to one-third of full-size specimens,
as the scaled-down specimens are large enough to represent Anchorage length =
load transfer mechanism along with all the real material (Ld + 10Ø – allowance for 90-degree bend) (2)
complexities and behavior of full-scaled specimens. The
size of columns and beam components were kept uniform where Ø is bar diameter; and Ld is development length given
at 150 x 120 mm (4.72 x 5.91 in.) for all specimens. Beam- by Eq. (3) as per IS:456-2000
column joints were designed on the assumption that the
point of contraflexure occurs at the beam midspan and _Ø σs
column midheight.30 Column length measured from one
Ld = 4 τ (3)
bd
contraflexure point to another was 1400 mm (55.12 in.). The
length of the beam measured from inner face of column to where Ø is bar diameter; σs is stress in longitudinal
the point of contraflexure in beam was 690 mm (27.17 in.). reinforcing bars; and τbd is design bond stress.
Beam-column subassemblies were designed following the Dimensions and reinforcement details of test specimens
strong-column, weak-beam concept.31 Based on the flexural are provided in Fig. 1.
@seismicisolation
@seismicisolation

Fig. 1—Reinforcement details and dimensions of test specimens under study. (Note: 1 mm = 0.0394 in.)
Fig. 2—Schematic representation of cyclic test setup. (Note: 1 kN = 0.2248 kip; 1 mm = 0.0394 in.)
Test setup and experimental procedure experiment started by applying a displacement of magnitude
The experimental setup to test beam-column joints 1 mm (0.039 in.) to the beam-end. Displacement progressed
under cyclic loading is shown in Fig. 2. Quasi-static cyclic with an increment of 1 mm (0.039 in.) having 0.25 Hz
loading was applied to beam-column joints by means of a frequency until failure of the joint. The failure was marked
servo-controlled hydraulic actuator having a maximum by a drop in peak load by 25% as per ACI 374.1-05.29 Each
capacity of ±100 kN (22.48 kip) and peak displacement of displacement level is comprised of three cycles of push and
±90 mm (3.54 in.). The ends of the columns were mounted pull loading. Typical loading history is presented in Fig. 3.
horizontally on a roller steel support to simulate hinged- Similar loading protocol was also adopted by Mukherjee
boundary condition, and the beam was placed in the vertical and Joshi,35 Chidambaram and Agarwal,36 and Kheni et al.,37
plane, as shown in Fig. 2. The beam was hinged connected to among others. The displacement applied by the actuator
the actuator through swivel base assembly. A uniform axial was gauged with linear variable differential transformers
load of 10% gross capacity of the column was applied to (LVDTs) fitted in the actuator, and the corresponding load
the column with a hydraulic jack to replicate gravity loading was recorded by a coaxially mounted load cell. A data
condition.34 acquisition system collected all test data and stored it in the
In this study, the loading protocol was selected following computer connected to the actuator.
codal provision mentioned in ACI 374.1-05.29 The
@seismicisolation
@seismicisolation

Rehabilitation procedure Group 2 rehabilitated with GM. The finished specimen is
The control specimens of Groups 1 and 2 were tested shown in Fig. 4(b).
under cyclic loading as per loading protocol, discussed in To evaluate the performance of geopolymer with respect
the previous section. Once control specimens were tested, to OPC under cyclic loading, the study was further extended
fractured concrete from the joint region was removed with to include cement-based repaired beam-column joints. For
an air hammer to expose the joint reinforcements. A similar this purpose, another two specimens (one from each group)
procedure for concrete removal was followed by Ghobarah were tested under cyclic loading. The damaged specimens
and Said38 during the rehabilitation of beam-column joints were rehabilitated with CM (compressive strength 52 MPa
with glass fiber-reinforced polymer (GFRP). Attention was [7.54 ksi]26) following the same rehabilitated technique as
paid to the removal of any concrete fragments from the rein- discussed in the preceding paragraph. Rehabilitated speci-
forcement. For exposed surface preparation, a chisel and mens were named 1CM and 2CM, where numeric 1 or 2
hammer were used to make the interface rough up following implied group number and CM implied cement mortar. After
the provisions of ACI 318-08 to ensure proper shear transfer repair, specimens 1CM and 2CM were wrapped with a wet
between old, hardened concrete and new repair materials. jute bag to cure for 28 days.
Subsequently, compressed air was used to remove any
debris, dust, or loose materials. Cleaned specimens were RESULTS AND DISCUSSION
then placed on wooden formwork that was constructed for Crack formation and failure mode
rehabilitation, as shown in Fig. 4(a). Geopolymer mixture T3 The typical failure crack patterns of control specimens 1C
was used to fill the voids left from the removal of damaged and 2C are shown in Fig. 5. In specimen 1C with flexural
joint concrete. The placed geopolymer mixture was thor- strength ratio 1.2, the first crack was observed at the 2 mm
oughly compacted to filled up all cavities. The wooden (0.079 in.) displacement level. With an increase in flexural
formwork was removed the next day, and specimens were strength ratio to 1.4 (specimen 2C), formation of the first
ambient cured from 28 days. Rehabilitated specimens were crack was delayed to a displacement of 3 mm (0.118 in.).
termed as 1GM and 2GM. 1GM means Group 1 specimen However, the first visible crack started to appear in the corner
rehabilitated with GM. Similarly, 2GM implied specimen of of beam-column connection and expanded towards the entire
depth of the beam with progressive loading in both speci-
mens. This ultimately led to the formation of flexural crack
in the beam. With repeated opening and closing of flexural
crack due to load reversal, minor diagonal cracks started to
appear in the joint core. The appearance of diagonal crack in
the joint would mean that transverse steel crossing diagonal
crack began to yield.39-41 With further increase in displace-
ment, the flexural crack in the beam widened and resulted in
failure of control specimens by formation of plastic hinge in
the beam. Such failure is in accordance with “strong-column,
weak-beam” design philosophy. No concrete bulging was
observed at any stage of loading in the joint region of control
specimens.
The crack patterns of specimens 1GM, 2GM, 1CM, and
2CM are shown in Fig. 6. Irrespective of repair material and
flexural strength ratio, the first visible crack in all rehabili-
Fig. 3—Loading history applied to all test specimens. (Note: tated specimens was observed in the corner of the beam at
1 mm = 0.0394 in.)
Fig. 4—(a) Typical specimen after removal of damage concrete from joint region; and (b) typical beam-column joint after
rehabilitated by GM. @seismicisolation
@seismicisolation

Fig. 5—Typical crack pattern of control specimens at failure.
a displacement of 1 mm (0.039 in.). With initiation of crack substrate. Such improvement was resultant of the reaction
in the beam corner, the crack started to penetrate across between the surface product of concrete substrate and alkali
the entire depth of the beam with increasing displacement. activator.43,44 Higher bond strength of GM resisted degra-
After the flexural cracking of beam, major diagonal cracks dation of bond between GM and old concrete due to tensile
in the joint panels of all four rehabilitated specimens began stress, which decreased intensity of crack formation at the
to appear. Diagonal cracks in the joint regions of 1GM and cold joint in 1GM and 2GM.
1CM and 2GM and 2CM were higher in number than that When control specimens were tested under cyclic loading,
of respective control specimens. Moreover, bulging of the joint hoop reinforcement would yield and develop residual
repair material leading to spalling was observed in rehabil- deformation.45 In the removal and replacement technique,
itated specimens. The effect of the cold joint between the only the damaged concrete was removed and replaced
old concrete of beam-column subassembly and new repair with suitable repair material. Pre-yielded hoop reinforce-
material was also observed. The cold joint is the plane of low ment remained inside rehabilitated specimens after repair.
bond strength.42 Therefore, when the magnitude of tensile Therefore, when rehabilitated specimens were subjected to
stress becomes high, repair materials from damaged areas cyclic loading, yielded hoop reinforcements having residual
tend to get separated from old concrete substrate of beam- deformation were reloaded. Reloading exaggerated the
column subassembly. This led to debonding at the interface force-deformation characteristics of such hoops, and hence,
of old concrete and repair material resulting in cracking small amount of stress caused large amounts of strain in the
formation at cold joints at the column. Such cracks in the joint hoops.45 GM and CM, being brittle in nature, were not
column are undesirable as per strong-column, weak-beam able to sustain such large strain, thus resulting in the forma-
design philosophy. However, it may be seen that effect of tion of many major cracks in the joint core. Furthermore,
cold joints is less prominent in case of 1GM and 2GM spec- splitting cracks at the outer face of column were observed in
imens than that of 1CM and 2CM, respectively. This was the case of all such specimens. This was due to the inability
due to better bond between geopolymer and old concrete of pre-yielded hoop reinforcement to confine the joint core
@seismicisolation
@seismicisolation

Fig. 6—Typical crack pattern of rehabilitated specimens at failure.
effectively, resulting in column reinforcement buckling. All interface, and loss of confinement due to the presence of
these factors contributed to the failure of rehabilitated spec- yielded hoop reinforcement in rehabilitated specimens. Initi-
imens by beam failure accompanied by brittle joint failure. ation of load at the beam-end led to the formation of cracks
at cold joint interfaces, resulting in a decrease in lateral
Hysteresis response and envelope curve confinement of the joint core.46 With progressive loading,
The typical hysteresis response of Group 1 specimen is such cracks became a secondary source for new joint crack
plotted in Fig. 7, and load-displacement envelope curves generation. Due to the higher compressive strength of GM
all specimens are plotted in Fig. 8. Peak loads of test spec- and CM, repair materials resisted crack formation momen-
imens were tabulated in Table 3. An increase of 14 and tarily, thereby increasing the maximum peak load of such
12% was observed in 1GM and 2GM from their respective specimens. However, after attaining peak load when the
controls. A similar increase in peak load of 17 and 18% was magnitude of shear stress was generated in the joint signifi-
also observed in 1CM and 2CM, respectively. The increase cantly increased, GM and CM failed to resist further crack
in peak load of rehabilitated specimens was due to higher development due to its brittle nature. Moreover, yielded
compressive strength of repair materials. 1CM and 2CM hoops could not confine the joint core effectively.39 Loss
exhibited an even higher peak load than 1GM and 2GM due in confinement at the joint core leads to easier crack devel-
to higher compressive strength of CM (52 MPa [7.54 ksi]) opment. All these factors combinedly caused significant
compared to GM (48 MPa [6.96 ksi]), respectively. cracking of joint core resulting in rapid loss of load-carrying
However, it may be observed from Fig. 8 that rehabilitated capacity, as observed in Fig. 8.
specimens exhibited brittle behavior with respect to control.
This was due to a sudden decrease in the load-carrying Joint stresses
capacity of rehabilitated specimens after attaining peak The design shear capacity of the exterior beam-column
load. This can be attributed to the brittle nature of the repair joint as per IS 13920:201632 is as follows
materials (GM and CM), debonding failure at the cold joint
@seismicisolation
@seismicisolation

Fig. 7—Typical hysteresis curve of Group 1 test specimens. (Note: 1 kN = 0.2248 kip; 1 mm = 0.0394 in.)
Fig. 8—Load-displacement curve of test specimens. (Note: 1 kN = 0.2248 kip; 1 mm = 0.0394 in.)
Table 3—Peak load and joint shear stress of shear force; Lb and db are length and effective depth of beam,
specimens respectively; Lc is column length; and Dc is the overall
Specimens Peak load, kN τIS, MPa τjh, MPa τjh/τIS
column depth. The value of τjh is presented in Table 3. The
value of designed shear stress τIS is dependent on compres-
1C 9.10 5.37 2.31 0.43
sive strength of joint concrete. As in rehabilitated speci-
1GM 10.40 8.31 2.64 0.32 mens, joint core was strengthened with higher compressive
1CM 10.66 8.65 2.70 0.31 strength repair materials (fck = 48 MPa [6.96 ksi] for GM and
2C 10.27 5.37 2.60 0.49
fck = 52 MPa [7.54 ksi] for CM) compared to control spec-
imens (fck = 20 MPa [2.9 ksi]); the value of τIS for rehabili-
2GM 11.51 8.31 2.92 0.35
tated specimens increased significantly. On the other hand,
2CM 12.16 8.65 3.08 0.36 due to the same geometrical dimensions of all specimens, τjh
Note: 1 kN = 0.2248 kip; 1 MPa = 0.145 ksi.
is dependent primarily on P. The τjh of rehabilitated spec-
_ imens also increased with an increase in peak load owing

τIS = 1.2√fck
(4) to the use of high-strength GM and CM. A large increase
in τIS compared to τjh decreased the value of τjh/τIS so that
where fck is the characteristic strength of concrete. it is less than 1. Apparently, the ratio of τjh/τIS for all reha-
For the exterior beam-column joint, horizontal shear bilitated specimens indicates that specimens have adequate
force attained in the joint during loading is taken as follows shear strength capacity and enabled adjoining beam to reach
(Murty et al.47) its ultimate capacity. However, rehabilitated specimens
failed due to beam failure accompanied by joint failure, as
τjh = h ( − L )
_ P _ Lb _
Lb + 0.5Dc observed from Fig. 6. This may be because Eq. (5) does not
(5)
A core db c reflect the effect of cold joints formed due to rehabilitation
and the presence of yielded hoop reinforcement in the joint.
where P is the peak load of the test specimens; Ahcore is hori-
zontal cross-sectional area of the joint resisting horizontal
@seismicisolation
@seismicisolation

Fig. 9—Stiffness degradation curve of all specimens. (Note: 1 kN/mm = 5.71 kip/in.)
Stiffness degradation
Stiffness degradation is used to quantify the loss of lateral
stiffness in reinforced concrete structures subjected to
reversed cyclic loading.45 In this investigation, secant stiff-
ness is used to represent stiffness of test specimens. Stiff-
ness at an i-th displacement is calculated by the following
expression
|F+i
_ |
− F−i

Ki = (6)
|δ+i− δ−i|
where F+i and F–i are maximum positive and negative load at
corresponding displacement δ+i and δ–I; respectively.
The stiffness calculated from Eq. (6) is plotted against
drift angle to obtain stiffness degradation curve, as shown in Fig. 10—Typical stress-strain graph of hoop reinforcement.
Fig. 9. Control specimens 1C and 2C displayed initial stiff- (Note: 1 MPa = 0.145 ksi.)
ness of 2.561 and 2.722 kN/mm (14.62 and 15.83 kip/in.),
and 2CM were able to resist diagonal tensile cracks forma-
respectively. GM and CM were only able to restore 79 to
tion at lower drift levels. Nevertheless, with increasing
80% and 65 to 67% of initial stiffness in rehabilitated speci-
drift, cracks at the cold joint interface expanded and caused
mens in comparison to control. This was due to the presence
development of new cracks in the joint. On the other hand,
of the cold joint interface between new repair mortars and
yielded hoop reinforcement also contributed to formation of
old concrete. When load was applied on rehabilitated speci-
crack in the joint. When yielded hoops were reloaded under
mens, the cold joint being a weak interface between concrete
reversed cyclic loading, yielded hoops yield at lower stress
and the repair material, a small amount of force opened
level and sustain lower ultimate strain. This is evident from
the interface. This led to earlier crack formation at lower
the stress-versus-strain graph of fresh hoops (tested before
displacement levels, as observed in the previous section
cyclic loading) and yielded hoops (taken out from control
where crack was first observed in rehabilitated specimens
specimens after cyclic loading), as shown in Fig. 10. Similar
at 1 mm (0.039 in.), irrespective of repair materials. On the
stress-strain curve of fresh and yielded reinforcing bar was
other hand, initial crack at 1C and 2C were observed at 2 and
also reported by Mukherjee and Jain.48 Yielding of yielded
3 mm (0.079 and 0.118 mm) displacement levels, respec-
hoops at a lower stress level produced loss of confinement
tively. Thus, earlier crack at lower displacement lowered
and deterioration of bond at lower beam-tip load compared
initial stiffness of all rehabilitated specimens. Higher initial
to control specimens. As a result, confinement of joint
stiffness of 1GM and 2GM than 1CM and 2CM was due to
hoops deteriorates significantly and becomes ineffective
higher bond strength of geopolymer with concrete substrate
against volumetric expansion of joint core. Such expansion
compared to OPC. As a result, a higher amount of force
44
of joint core cannot be resisted by the brittle nature of GM
is required to form first crack in 1GM and 2GM thereby
and CM, resulting in a large number of cracks in the joint.
leading to 20 to 21% increase in initial stiffness compared to
These factors contributed to higher crack development in
1CM and 2CM, respectively.
joint, thereby initiating higher stiffness degradation of 1GM,
Figure 9 reveals that 1GM, 2GM, 1CM, and 2CM exhib-
1CM, 2GM, and 2CM at higher drift.
ited lower initial stiffness, yet they displayed lower stiffness
degradation at lower drift level compared to controls. This
Pinching width ratio (PWR)
can be attributed to higher compressive strength of repair
The pinching width ratio (PWR) is a dimensionless param-
materials which increases the strength of diagonal strut,
eter developed by Mostofinejad and Akhlaghi49 to measure
as performance of diagonal strut depends on compressive
pinching associated with hysteresis loops under cyclic
strength of joint core material. Hence, 1GM, 2GM,@seismicisolation
1CM,
@seismicisolation

loading. The higher the value of PWR, the lower will be the stiffness degradation of rehabilitated specimens caused rapid
pinching and vice-versa. Pinching of the hysteresis loops is decrease in PWR at a higher drift level.
largely influenced by stiffness degradation, characterized by
crack formation. It is defined as follows Displacement ductility
Ductility of the specimens was quantified by displacement
PWR = Pa/Pi (7) ductility, which is defined as the ratio of ultimate displace-
ment (δu) to yield displacement (δy). Ultimate displacement
where Pa is actual pinching of a hysteretic loop at a partic- (δu) represents post-peak deformation corresponding to 75%
ular loading cycle; and Pi is the pinching width of the ideal- peak load.31 Yield displacement (δy) is taken as displacement
ized hysteresis loops at same displacement. These parame- corresponding to secant stiffness at 0.75 times of the design
ters are illustrated in Fig. 11. Pa is the least distance between load of the equivalent elasto-plastic curve.50 Displacement
the positive and negative loading path of actual hysteresis ductility of all the test specimens is tabulated in Table 4.
loop of that loading cycle at approximately zero displace- Geopolymer mortar and cement mortar were able to restore
ment level. Pi is the distance between two sides of parallel- 74 to 78% and 61 to 67% ductility for rehabilitated spec-
ogram at approximately zero displacement. K1 and K2 are imens with respect to control specimens. Geopolymer
initial uncracked stiffness of positive and negative loading repaired specimens exhibited 19 to 22% better ductility
cycle. Using all the parameters, PWR was determined for all compared to CM repaired specimens. The increased ductility
specimens against displacement (Fig. 12). Lower stiffness of 1GM and 2GM was predominantly due to the increase
degradation of rehabilitated specimens at initial displace- in ultimate displacement (Table 4). Higher bond strength
ment levels contributed to higher PWR, indicating lower of GM with concrete substrate decreased crack formation
pinching of the hysteresis curve then control specimens intensity at cold joint interface. Better bonding ability of GM
initially. However, with increasing displacement, rehabil- with embedded steel reinforcement prevented reinforcement
itated specimens experienced higher degradation of PWR slippage and decreased intensity of concrete wedge separa-
value than control specimens such that towards the ends, it tion from the outer face of column. Higher tensile strength
was observed that the PWR of 1GM and 1CM, and 2GM of geopolymer reduced diagonal tensile crack formation in
and 2CM, was lower than 1C and 2C, respectively. Higher GM repaired specimens. As a result, the crack resistance
ability of specimens 1GM and 2GM significantly increased,
enabling specimens to sustain loading to higher displace-
ment level compared to CM-repaired specimens.
From Table 4, the ultimate displacement of the rehabili-
tated specimens was smaller compared to control specimens.
This was mainly due to presence cold joint and yielded
hoops in rehabilitated specimens. With progressive loading,
cracks in the cold joint interface eventually penetrated the
entire depth of columns and beam, reaching the level of
longitudinal reinforcement crossing cold joint interfaces.
At that instant, the total shear force acting at the cold joint
interfaces is resisted entirely by longitudinal reinforcements
at that interface,42,46 causing early yielding of longitudinal
reinforcements in the case of repaired specimens. Such early
yielding of reinforcement reduced post-yield performance
of such specimens. Also, cracks at the cold joint interface
led to the development of new cracks in the joint with
Fig. 11—Illustration of parameters used to measure PWR.
Fig. 12—PWR of specimens under study. (Note: 1 mm = 0.0394 in.)

@seismicisolation
@seismicisolation

increased loading. Due to their brittle characteristics, CM The maximum amount of energy is dissipated in the inelastic
and GM could not arrest new crack formation. Moreover, as range of deformation.35 Therefore, a larger inelastic zone
mentioned earlier when rehabilitated specimens were tested resulted in 24 to 37% higher energy dissipation by 1GM and
under cyclic loading, hoop reinforcements yielded at lower 2GM compared to 1CM and 2CM, respectively.
stress levels, resulting in bond degradation and confinement Full enhancement of cumulative energy dissipation in
loss of joint core at lower loads. These contributed to easier rehabilitated specimens was hindered due to the presence of
crack formation in the repaired joint, subsequently causing cold joints and yielded hoops in the joint. Due to the low
larger cracks in the joint core. These factors initiated early initial stiffness and higher stiffness degradation of reha-
failure as rehabilitated specimens could not sustain load up bilitated specimens at higher drift level, energy dissipated
to higher displacement levels, resulting in partial enhance- though viscous damping decreased. Also, early yielding of
ment of ductility. longitudinal reinforcement due to the formation of cracks at
the cold joint interface42 and yielding of pre-yielded hoops
Cumulative energy dissipation at lower stress level (Fig. 10) reduced the performance of
Energy dissipation is a measure to evaluate the capacity rehabilitated specimens in the post-yield range of loading
of a structure to endure an earthquake through dissipation of thereby causing failure at lower displacement. This resulted
energy from ground motion in the post-elastic deformation in a smaller inelastic zone in rehabilitated specimens
phase.34 The area of load-deformation hysteresis loops gives compared to controls (Table 4), which reduced the energy
an estimate of energy dissipation capacity of specimens dissipation capacity through hysteretic damping. There-
under cyclic loading. A plot of cumulative energy dissipation fore, reduced energy dissipation by viscous and hysteretic
of specimens against displacement is shown in Fig. 13. It damping combinedly reduced the total energy dissipation
may be seen that GM restored 74 to 80% of total cumulative capacity of rehabilitated specimens.
energy dissipation in rehabilitated specimens with respect to
control specimens. On the other hand, 1CM and 2CM dissi- Equivalent viscous damping ratio
pated 60% and 58% of total cumulative energy dissipation of The equivalent viscous damping ratio represents the ability
specimens 1C and 2C, respectively. Higher ductility enabled of a structure to dampen peak excitation that arises due to
1GM and 2GM to maintain load-carrying capacity to higher inelastic deformation during earthquake loading. Cumula-
deformation in the post-elastic range of loading. This tive energy dissipation gives an idea about combined energy
resulted in a larger inelastic zone of deformation (Table 4). dissipated by the structure up to a loading cycle, whereas the
equivalent viscous damping ratio is associated with energy
Table 4—Displacement ductility of test specimens dissipated by the structure during a particular loading cycle.
It is determined as follows51
Yield displace- Ultimate displace- Ductility
Specimens ment (δy), mm ment (δu), mm (δu/δy)
_ Ei
1C 7.85 41.50 5.29
ξequi = 2π F D (8)
mi mi
1GM 7.94 32.94 4.15
1CM 8.12 28.37 3.49 where ξequi is equivalent viscous damping ratio; Ei is the
energy dissipation; Fmi is the peak load; and Dmi is corre-
2C 4.42 43.89 9.93
sponding displacement at the i-th loading cycle, respec-
2GM 4.74 34.84 7.35 tively. The equivalent viscous damping ratio for the test
2CM 4.82 29.13 6.04 specimens is plotted against displacement in Fig. 14. Due to
Note: 1 mm = 0.0394 in.
the higher compressive strength of GM and CM, initial crack
Fig. 13—Comparison of cumulative energy dissipation. (Note: 1 kNm = 0.0088 kip-in.; 1 mm = 0.0394 in.)
@seismicisolation
@seismicisolation

Fig. 14—Equivalent viscous damping ratio of specimens. (Note: 1 mm = 0.0394 in.)
formation in rehabilitated specimens released higher energy. ultimate displacement δu was approximately estimated from
This contributed to an increase in ξequ by 64 to 73% and 44 the load-deformation envelope.53 Here, δu was taken as the
to 48% in GM and CM rehabilitated specimens at initial post-peak displacement when load comes down to 75% of
displacement compared to control specimens, respectively. peak load. Damage indexes for all the beam-column joints
Also, the higher initial stiffness of GM rehabilitated speci- were calculated from Eq. (9) and plotted against displace-
mens contributed to higher energy dissipation leading to 14 ment in Fig. 15. 1GM and 2GM suffered 10 to 14% lower
to 17% higher initial ξequ than CM rehabilitated specimens. damage than 1CM and 2CM. This is also evident from
Nevertheless, with increasing displacement, a dip in ξequ was Fig. 15, showing damage index values of the test specimens
observed in rehabilitated specimens. This was due to crack at 25 mm (1 in.) displacement level. The damage index of
propagation in cold joint interfaces during the subsequent the beam-column joint depends on cumulative energy dissi-
loading cycle, which lowered energy dissipation per cycle. pation and ultimate deformation.52 The higher the amount of
When cracks started to penetrate the joint core, the higher deformation that a beam-column joint could sustain and the
compressive strength of GM and CM resisted crack forma- higher the energy dissipation, the lower will be the damage
tion, leading to higher energy dissipation per loading cycle index and vice-versa. 1GM and 2GM exhibited higher
until failure. Release in higher energy per loading cycle ultimate deformation (Table 4) and higher energy dissipa-
increased ξequ value such that 1CM and 2CM, and 1GM and tion than 1CM and 2CM (Fig. 13); hence, 1GM and 2GM
2GM, displayed 13 to 16% and 36 to 42% higher ξequ than sustained lower damage than 1CM and 2CM, respectively.
control specimens, respectively. Higher energy dissipation at On the contrary, control specimens showed the highest
failure contributed to 21 to 26% higher ξequ in GM rehabil- ultimate deformation and the highest energy dissipation in
itated specimens compared to CM rehabilitated specimens. Group 1 and Group 2 specimens, respectively. This resulted
in least damage suffered by such specimens under cyclic
Damage index loading as shown in Fig. 15.
Park and Ang’s52 damage index model is used in the
current investigation to quantify damage suffered by the CONCLUSIONS
specimens in the inelastic phase of deformation under cyclic The present work was undertaken to study the effective-
loading. The following formula proposed by Park and Ang52 ness of removal and replacement techniques in the reha-
was used to determine the damage index in the present bilitation of seismically detailed beam-column joints. For
investigation this purpose, two groups of beam-column specimens with
different flexural strength ratios were selected. High-strength
δM _
_ β geopolymer mortar (GM) was used as a repair material to

D = + ∫ dE (9) rehabilitate two beam-column joints of each group. The
δu Qyδu
test results were compared with another set of specimens
where δM is the maximum deformation at the i-th displace- from the same group, rehabilitated by cement mortar (CM)
ment; δu is the maximum displacement sustained by a adopting the same technique. Based on the test results, the
beam-column joint under monotonic loading; β is a non- following general conclusions may be drawn:
dimensional coefficient related with strength degradation 1. Control specimens failed due to beam failure only,
of beam-column joint; in the current study, the value of whereas both GM- and CM-rehabilitated specimens failed
β is taken as 0.15;53 Qy is the yield force calculated from due to beam failure accompanied by joint failure. The effect
load-deformation envelope; and dE is the cumulative energy of cold joints and splitting cracks at the outer face of the
dissipation considered up to the i-th displacement. Because column was less prominent in GM-rehabilitated specimens
monotonic loading was not applied on the tested specimens, than CM-rehabilitated specimens.
@seismicisolation
@seismicisolation

Fig. 15—Damage index of test specimens. (Note: 1 mm = 0.0394 in.)
2. GM- and CM-repaired specimens exhibited 12 to fulfill the objective of seismic rehabilitation for seismically
14% and 17 to 18% higher peak load than their respective detailed beam-column joints. Irrespective of repair mate-
controls. rial, the removal and replacement technique could partially
3. GM and CM could restore 79 to 80% and 65 to 67% restore the cyclic performance of rehabilitated specimens.
of the initial stiffness of rehabilitated specimens. Specimens Nevertheless, GM exhibited better performance as a repair
rehabilitated with GM and CM displayed lower stiffness material than CM for rehabilitation of seismically detailed
degradation during initial drift levels compared to control beam-column joints using the removal and replacement
specimens, while GM and CM displayed higher stiffness technique under cyclic loading.
degradation compared to control specimens at higher drift.
4. Rehabilitated specimens initially exhibited higher AUTHOR BIOS
pinching width ratio (PWR) than control specimens, but at Arshad Hussain Choudhury is an Assistant Professor in the Department
of Civil Engineering, Chaitanya Bharathi Institute of Technology (CBIT),
failure, the PWR of rehabilitated specimens was lower than Hyderabad, India. He received his BE (civil) from Assam Engineering
control. College, India, and his MTech and PhD from the National Institute of Tech-
5. GM was able to restore ductility and cumulative energy nology (NIT) Silchar, Silchar, India.
dissipation in rehabilitated specimens up to 74 to 78% and Aminul Islam Laskar is a Professor in the Department of Civil Engi-
72 to 74%, respectively, while CM restored 61 to 67% of neering, NIT Silchar. He received his BE (civil) from NIT Silchar; his
control specimens’ ductility and 58 to 64% of control speci- MTech from the Indian Institute of Technology (IIT) New Delhi, New Delhi,
India; and his PhD from IIT Guwahati, Guwahati, India.
mens’ cumulative energy dissipation.
6. The equivalent viscous damping ratio of GM- and CM- REFERENCES
rehabilitated specimens were 64 to 73% and 44 to 48% 1. Engindeniz, M.; Kahn, L. F.; and Zureick, A.-H., “Repair and Strength-
higher than control specimens, respectively, during initial ening of Reinforced Concrete Beam-Column Joints: State of the Art,” ACI
Structural Journal, V. 102, No. 2, Mar.-Apr. 2005, pp. 187-197.
loading. The ultimate equivalent viscous damping ratio at 2. Shafaei, J.; Hosseini, A.; Marefat, M. S.; Ingham, J. M.; and Zare,
failure was 36 to 42% higher for GM-rehabilitated speci- H., “Experimental Evaluation of Seismically and Non-Seismically Detailed
mens and 13 to 16% higher for CM-rehabilitated specimens External RC Beam-Column Joints,” Journal of Earthquake Engineering,
V. 21, No. 5, 2017, pp. 776-807. doi: 10.1080/13632469.2016.1185052
7. Control specimens sustained the least damage, and 3. Shafaei, J.; Hosseini, A.; Marefat, M. S.; and Ingham, J. M., “Rehabil-
CM-rehabilitated specimens sustained the highest damage itation of Earthquake Damaged External RC Beam-Column Joints by Joint
in terms of damage index in each group. Enlargement Using Prestressed Steel Angles,” Earthquake Engineering
& Structural Dynamics, V. 46, No. 2, 2016, pp. 291-316. doi: 10.1002/
8. GM-rehabilitated specimens exhibited 20 to 21% higher eqe.2794
initial stiffness, 19 to 22% higher displacement ductility, 24 4. Alavi, B., and Krawinkler, H., “Behavior of Moment-Resisting Frame
to 37% higher cumulative energy dissipation, 14 to 17% Structures Subjected to Near-Fault Ground Motions,” Earthquake Engi-
neering & Structural Dynamics, V. 33, No. 6, 2004, pp. 687-706. doi:
higher initial equivalent viscous damping ratio, 21 to 26% 10.1002/eqe.369
higher ultimate equivalent viscous damping ratio at failure, 5. Alavi, B., and Krawinkler, H., “Effects of Near-Fault Ground Motions
and 10 to 14% lower damage index compared to CM- on Frame Structures,” Report No. 138, John A. Blume Earthquake Engi-
neering Center, Stanford, CA, 2001, 311 pp.
rehabilitated specimens. 6. Vatani-Oskouei, A., “Repairing of Seismically Damaged RC Exte-
The removal and replacement technique was only able rior Beam—Column Connection Using CFRP,” Journal of Reinforced
to enhance peak load, PWR at initial displacement, and the Plastics and Composites, V. 29, No. 21, 2010, pp. 3257-3274. doi:
10.1177/0731684410371407
equivalent viscous damping ratio of rehabilitated specimens 7. Zamani Beydokhti, E., and Shariatmadar, H. “Strengthening and
up to the level of control specimens. Other parameters such Rehabilitation of Exterior RC Beam–Column Joints Using Carbon-FRP
as initial stiffness, PWR at failure, cumulative energy dissi- Jacketing,” Materials and Structures/Materiaux et Constructions, V. 49,
No. 12, 2016, pp. 5067-5083.
pation, displacement ductility, and damage index were only 8. Pantelides, C. P.; Okahashi, Y.; and Reaveley, L. D., “Seismic Reha-
restored partially. Furthermore, rehabilitated specimens bilitation of Reinforced Concrete Frame Interior Beam-Column Joints with
failed due to beam failure accompanied by joint failure. FRP Composites,” Journal of Composites for Construction, ASCE, V. 12,
No. 4, 2008, pp. 435-445. doi: 10.1061/(ASCE)1090-0268(2008)12:4(435)
Thus, the removal and replacement technique was not@seismicisolation
able to
@seismicisolation

9. Arzeytoon, A.; Hosseini, A.; and Goudarzi, A., “Seismic Rehabil- 31. Park, R., and Paulay, T., Reinforced Concrete Structures, John
itation of Exterior RC Beam-Column Joints Using Steel Plates, Angles, Wiley & Sons, Inc., New York, 1975.
and Posttensioning Rods,” Journal of Performance of Constructed Facil- 32. IS 13920:2016, “Ductile Design and Detailing of Reinforced
ities, ASCE, V. 30, No. 1, 2016, p. 04014200. doi: 10.1061/(ASCE) Concrete Structures Subjected to Seismic Forces - Code of Practice,”
CF.1943-5509.0000721 Bureau of Indian Standards, New Delhi, India, 2016.
10. Marchisella, A.; Muciaccia, G.; Sharma, A.; and Eligehausen, R., 33. IS 456:2000, “Plain and Reinforced Concrete - Code of Practice,”
“Experimental Investigation of 3D RC Exterior Joint Retrofitted With Bureau of Indian Standards, New Delhi, India, 2000.
Fully-Fastened-Haunch-Retrofit-Solution,” Engineering Structures, V. 239, 34. Choudhury, A. H., and Laskar, A. I., “Rehabilitation of Substandard
2021, p. 112206. doi: 10.1016/j.engstruct.2021.112206 Beam-Column Joint Using Geopolymer,” Engineering Structures, V. 238,
11. Sharma, A.; Reddy, G. R.; Eligehausen, R.; Genesio, G.; and 2021, p. 112241. doi: 10.1016/j.engstruct.2021.112241
Pampanin, S., “Seismic Response of Reinforced Concrete Frames with 35. Mukherjee, A., and Joshi, M., “FRPC Reinforced Concrete Beam-
Haunch Retrofit Solution,” ACI Structural Journal, V. 111, No. 3, May-June Column Joints Under Cyclic Excitation,” Composite Structures, V. 70,
2014, pp. 673-684. doi: 10.14359/51686625 No. 2, 2005, pp. 185-199. doi: 10.1016/j.compstruct.2004.08.022
12. FEMA 308. “Repair of Earthquake Damaged Concrete and Masonry 36. Chidambaram, R. S., and Agarwal, P., “Seismic Behavior of Hybrid
Wall Buildings,” Federal Emergency Management Agency, Washington, Fiber Reinforced Cementitious Composite Beam–Column Joints,” Mate-
DC, 1998, 80 pp. rials & Design, V. 86, 2015, pp. 771-781. doi: 10.1016/j.matdes.2015.07.164
13. Lee, D. L. N.; Wight, J. K.; and Hanson, R. D., “Repair of Damaged 37. Kheni, D.; Scott, R. H.; Deb, S. K.; and Dutta, A., “Ductility
Reinforced Concrete Frame Structures,” Proceedings of the Sixth World Enhancement in Beam-Column Connections Using Hybrid Fiber-
Conference on Earthquake Engineering, New Delhi, India, 1977. Reinforced Concrete,” ACI Structural Journal, V. 112, No. 2, Mar.-Apr.
pp. 2486-2491. 2015, pp. 167-178. doi: 10.14359/51687405
14. Karayannis, C. G.; Chalioris, C. E.; and Sideris, K. K., “Effectiveness 38. Ghobarah, A., and Said, A., “Seismic Rehabilitation of Beam-
of RC Beam-Column Connection Repair Using Epoxy Resin Injections,” Column Joints Using FRP Laminates,” Journal of Earthquake Engineering,
Journal of Earthquake Engineering, V. 2, No. 2, 1998, pp. 217-240. doi: V. 5, No. 1, 2001, pp. 113-129. doi: 10.1080/13632460109350388
10.1080/13632469809350320 39. Pantazopoulou, S., and Bonacci, J., “Considerations of Questions
15. Tsonos, A. G., “Seismic Rehabilitation of Reinforced Concrete of Beam-Column Joints,” ACI Structural Journal, V. 89, No. 1, Jan.-Feb.
Joints by the Removal and Replacement Technique,” International Journal 1992, pp. 27-36.
of European Earthquake Engineering and Engineering Seismology, V. 3, 40. Hwang, S. J.; Lee, H. J.; and Liao, T. F., “Role of Hoops on Shear
2001, pp. 29-43. Strength of Reinforced Concrete Beam-Column Joints,” ACI Structural
16. Marthong, C.; Dutta, A.; and Deb, S. K., “Seismic Rehabilitation Journal, V. 102, No. 3, May-June 2005, pp. 445-453.
of RC Exterior Beam-Column Connections Using Epoxy Resin Injection,” 41. Lee, J.-Y.; Kim, J.-Y.; and Oh, G.-J., “Strength Deterioration of
Journal of Earthquake Engineering, V. 17, No. 3, 2013, pp. 378-398. doi: Reinforced Concrete Beam–Column Joints Subjected to Cyclic Loading,”
10.1080/13632469.2012.738284 Engineering Structures, V. 31, No. 9, 2009, pp. 2070-2085. doi: 10.1016/j.
17. Davidovits, J., “Geopolymers - Inorganic Polymeric New Mate- engstruct.2009.03.009
rials,” Journal of Thermal Analysis, V. 37, No. 8, 1991, pp. 1633-1656. doi: 42. Roy, B., and Laskar, A. I., “Cyclic Behavior of In-Situ Exterior
10.1007/BF01912193 Beam-Column Subassemblies with Cold Joint in Column,” Engineering
18. Zhang, P.; Zheng, Y.; Wang, K.; and Zhang, J., “A Review on Proper- Structures, V. 132, 2017, pp. 822-833. doi: 10.1016/j.engstruct.2016.12.001
ties of Fresh and Hardened Geopolymer Mortar,” Composites. Part B, Engi- 43. Hu, S.; Wang, H.; Zhang, G.; and Ding, Q., “Bonding and Abra-
neering, V. 152, 2018, pp. 79-95. doi: 10.1016/j.compositesb.2018.06.031 sion Resistance of Geopolymeric Repair Material Made with Steel Slag,”
19. Vasconcelos, E.; Fernandes, S.; Barroso De Aguiar, J. L.; and Cement and Concrete Composites, V. 30, No. 3, 2008, pp. 239-244. doi:
Pacheco-Torgal, F., “Concrete Retrofitting Using Metakaolin Geopolymer 10.1016/j.cemconcomp.2007.04.004
Mortars and CFRP,” Construction and Building Materials, V. 25, No. 8, 44. Wang, Y.-S.; Peng, K.-D.; Alrefaei, Y.; and Dai, J.-G., “The Bond
2011, pp. 3213-3221. doi: 10.1016/j.conbuildmat.2011.03.006 Between Geopolymer Repair Mortars and OPC Concrete Substrate:
20. Duan, P.; Yan, C.; and Luo, W., “A Novel Waterproof, Fast Strength and Microscopic Interactions,” Cement and Concrete Composites,
Setting and High Early Strength Repair Material Derived from Metaka- V. 119, 2021, p. 103991. doi: 10.1016/j.cemconcomp.2021.103991
olin Geopolymer,” Construction and Building Materials, V. 124, 2016, 45. Choudhury, A. H., and Laskar, A. I., “Effect of Hoop Reinforcement
pp. 69-73. doi: 10.1016/j.conbuildmat.2016.07.058 Yielding on the Cyclic Behavior of Beam-Column Joint,” Journal of Earth-
21. Phoo-Ngernkham, T.; Sata, V.; Hanjitsuwan, S.; Ridtirud, C.; quake Engineering, V. 26, 2020, pp. 1-18.
Hatanaka, S.; and Chindaprasirt, P., “High Calcium Fly Ash Geopolymer 46. Roy, B., and Laskar, A. I., “Beam–Column Subassemblies with
Mortar Containing Portland Cement for Use as Repair Material,” Construc- Construction Joint in Columns Above and Below the Beam,” Maga-
tion and Building Materials, V. 98, 2015, pp. 482-488. doi: 10.1016/j. zine of Concrete Research, V. 70, No. 2, 2018, pp. 71-83. doi: 10.1680/
conbuildmat.2015.08.139 jmacr.17.00155
22. IS 516:1959, “Methods of Tests for Strength of Concrete,” Bureau of 47. Murty, C. V. R.; Rai, D. C.; and Bajpai, K. K., “Effectiveness of Rein-
Indian Standards, New Delhi, India, 1959. forcement Details in Exterior Reinforced Concrete Beam-Column Joints for
23. IS 1608:3005, “Metallic Materials - Tensile Testing at Ambient Earthquake Resistance,” ACI Structural Journal, V. 100, No. 2, Mar.-Apr.
Temperature,” Bureau of Indian Standards, New Delhi, India, 2005. 2003, pp. 149-156.
24. Laskar, S. M., and Talukdar, S., “Preparation and Tests for Workability, 48. Mukherjee, A., and Jain, K. K., “Performance of the FRPC Reha-
Compressive and Bond Strength of Ultra-Fine Slag Based Geopolymer as bilitated RC Beam-Column Joints Subjected to Cyclic Loading BT -
Concrete Repairing Agent,” Construction and Building Materials, V. 154, Advances in Structural Engineering,” Springer, New Delhi, India, 2015,
2017, pp. 176-190. doi: 10.1016/j.conbuildmat.2017.07.187 pp. 2025-2042.
25. Singhi, B.; Laskar, A. I.; and Ahmed, M. A., “Investigation on Soil– 49. Mostofinejad, D., and Akhlaghi, A., “Experimental Investigation
Geopolymer with Slag, Fly Ash and Their Blending,” Arabian Journal for of the Efficacy of EBROG Method in Seismic Rehabilitation of Deficient
Science and Engineering, V. 41, No. 2, 2016, pp. 393-400. doi: 10.1007/ Reinforced Concrete Beam–Column Joints Using CFRP Sheets,” Journal
s13369-015-1677-y of Composites for Construction, ASCE, V. 21, No. 4, 2017, p. 04016116.
26. IS 1727:1967, “Methods of Test for Pozzolanic Materials,” Bureau of doi: 10.1061/(ASCE)CC.1943-5614.0000781
Indian Standards, New Delhi, India, 1967. 50. Park, R., “State of the Art Report-Ductility Evaluation from Labo-
27. Yun, K.-K., and Choi, P., “Causes and Controls of Cracking at ratory and Analytical Testing,” Proceedings of the 9th World Conference
Bridge Deck Overlay with Very-Early Strength Latex-Modified Concrete,” on Earthquake Engineering, V. 3, Science Council of Japan, Tokyo, Japan,
Construction and Building Materials, V. 56, 2014, pp. 53-62. doi: 10.1016/j. 1988, pp. 605-616.
conbuildmat.2014.01.055 51. Blandon, C. A., and Priestley, M. J. N., “Equivalent Viscous Damping
28. IS 8112:2013, “Ordinary Portland Cement, 43 Grade - Specifica- Equations for Direct Displacement Based Design,” Journal of Earthquake
tion,” Bureau of Indian Standards, New Delhi, India, 2013. Engineering, V. 9, No. Sup2, 2005, pp. 257-278.
29. ACI Committee 374, “Acceptance Criteria for Moment Frames 52. Park, Y. J., and Ang, A. H. S., “Mechanistic Seismic Damage Model
Based on Structural Testing (ACI 374.1-05) and Commentary (Reapproved for Reinforced Concrete,” Journal of Structural Engineering, ASCE, V. 111,
2019),” American Concrete Institute, Farmington Hills, MI, 2005, 9 pp. No. 4, 1985, pp. 722-739. doi: 10.1061/(ASCE)0733-9445(1985)111:4(722)
30. Ghobarah, A., and El-Amoury, T., “Seismic Rehabilitation of Beam – 53. Karayannis, C. G.; Chalioris, C. E.; and Sirkelis, G. M., “Local
Column Joint Using GFRP Sheets,” Engineering Structures, V. 24, No. 11, Retrofit of Exterior RC Beam–Column Joints Using Thin RC Jackets—An
2002, pp. 1397-1407. doi: 10.1016/S0141-0296(02)00081-0 Experimental Study,” Earthquake Engineering & Structural Dynamics,
V. 37, No. 5, 2008, pp. 727-746. doi: 10.1002/eqe.783
@seismicisolation
@seismicisolation

Title No. 120-S82
Strength and Behavior of Glass Fiber-Reinforced

Polymer-Reinforced Concrete Box Girders without
Web Reinforcement under Pure Torsion
by Ibrahim T. Mostafa, Salaheldin Mousa, Hamdy M. Mohamed, and Brahim Benmokrane
The torsional behavior of solid reinforced concrete (RC) members deterioration, particularly in North America, resulting in
reinforced with fiber-reinforced polymer (FRP) bars has been the costly repairs, rehabilitation, and a considerable reduction in
subject of several experimental studies. No experimental research, the service life span. Fiber-reinforced polymer (FRP) rein-
however, seems to have focused on RC box girders reinforced with forcing bars, on the other hand, can be used as an alternative
FRP bars under a pure torsional moment. This paper reports the
to steel reinforcement to combat corrosion, thereby lowering
results of an experimental investigation on the torsional strength
maintenance costs and extending service life. There are
and behavior of full-scale RC box girders reinforced with longi-
tudinal glass FRP (GFRP) bars. All specimens measured 380 mm numerous advantages to using FRP bars instead of steel
(15 in.) in height, 380 mm (15 in.) in width, 100 mm (4 in.) wall bars, such as higher tensile strength, weight one-quarter to
thickness, and 4000 mm (157.48 in.) in length. They were tested one-fifth lighter than steel, noncorroding nature, and longer
under pure torsional loading over a clear span of 2000 mm service life than steel (ACI 440.1R-15 [ACI Committee 440
(78.74 in.). The test specimens consisted of four RC box girders 2015]). The last two decades have seen a rapid increase in
with longitudinal GFRP bars and one RC box girder with longitu- the use of FRP reinforcing bars in many applications, such
dinal steel bars as a reference. All the specimens were constructed as bridges, piles, parking garages, marine structures, water
without web reinforcement to study the contribution of the longi- tanks, and tunnels (El-Salakawy et al. 2004; Eladawy et al.
tudinal reinforcement to torsional strength. The test variables 2019; Mohamed and Benmokrane 2014; Mousa et al. 2018;
included the longitudinal reinforcement ratio (ranging between
Mohamed et al. 2020).
1.10 and 2.74%) and the type of longitudinal reinforcement (GFRP
Due to the scarcity of experimental studies on torsion
or steel). The test results indicate that increasing the GFRP longi-
tudinal reinforcement ratio increased the torsional strength after members reinforced with FRP bars, the FRP-RC design
the initiation of the first diagonal crack, especially for specimens guidelines (ACI 440.1R-15; JSCE 1997) do not include
with a high reinforcement ratio. In addition, theoretical torsional any provisions pertaining to torsion. In addition, the torsion
moment-twist curves were developed and gave predictions consis- provisions in FRP-RC design codes (CSA S806-12 [2017];
tent with the experimental test results. Lastly, the ultimate torsional CSA S6 [2019]; AASHTO 2018) are mainly modifications
strength of the GFRP-RC box girders without web reinforcement of steel-RC codes. These modifications include the mechan-
was estimated with the CSA S806-12 (R2017) design equation with ical properties of FRP reinforcement (modulus of elasticity
a modification related to the GFRP tensile strain limit. and stirrup bending strength). The design of torsion in
FRP-RC codes (CSA S806-12; CSA S6:19; AASHTO 2018)
Keywords: cracking pattern and modes of failure; design codes; glass
fiber-reinforced polymer (GFRP) bars; reinforced concrete (RC) box is based on a thin-walled tube space-truss analogy. Once RC
girders; reinforcement ratio; theoretical modeling; torsional crack width members have cracked under torsion, the torsional strength
and toughness; torsional loading; torsional moment-twist response; is mainly provided by closed stirrups and longitudinal bars.
torsional strength. CSA S806-12 provides an equation to estimate the torsional
strength based on the contribution of the GFRP longitudinal
INTRODUCTION reinforcement; this equation is a function of the total area
Reinforced concrete (RC) box girders can be found in and tensile strength of the GFRP longitudinal bars. The
many civil engineering applications. The most notable are experimental program of the current study was intended to
cable-supported bridges, pedestrian bridges, curved bridges, evaluate the accuracy of the CSA S806-12 design equation.
and modern elevated structures for light rail transport (Rahal Valuable research work has been carried out in the last
and Collins 1995). There are economic and structural bene- decade to investigate the torsional behavior of the FRP-RC
fits to using box girders, such as low self-weight of struc- members with a solid cross section (Deifalla et al. 2014;
tures and higher torsional stiffness (Jeng et al. 2013). Torsion Hadhood et al. 2020; Mohamed and Benmokrane 2015;
in bridges could occur as a consequence of the geometric Shehab et al. 2009; Zhou et al. 2017). These studies
complexities of horizontally curved bridges and/or the large
eccentricity of gravity loads. Such structures are usually ACI Structural Journal, V. 120, No. 5, September 2023.
exposed to aggressive and/or harsh marine environments, MS No. S-2022-137.R3, doi: 10.14359/51738836, received November 28, 2022, and
leading to corrosion of the steel reinforcement. This type of Institute. All rights reserved, including the making of copies unless permission is
corrosion in concrete structures is a prime cause of structural closure, if any, will be published ten months from this journal’s date if the discussion
@seismicisolation
@seismicisolation

considered a wide range of parameters such as reinforce- girders and to extend the design and code provisions related
ment type (glass or carbon), concrete type (normal-strength to GFRP reinforcement for concrete structures.
or fiber-reinforced concrete), stirrup configurations (ties or
spirals), transverse reinforcement ratio, and cross-section EXPERIMENTAL PROGRAM
configurations (rectangular or L-shaped). Mohamed and Materials
Benmokrane (2016) conducted an experimental investiga- Reinforcement—The GFRP bars employed in this study
tion to study the torsional behavior of full-scale rectangular were manufactured and developed by a manufacturer in
members reinforced with carbon FRP (CFRP), glass FRP Thetford Mines, QC, Canada (Pultrall Inc. 2019). The GFRP
(GFRP), and steel reinforcement with and without trans- bars had a sand-coated surface, as shown in Fig. 1(a), to
verse torsional reinforcement. The results revealed that the improve the bond performance and force transfer between
torsional failure was controlled by concrete splitting for the bars and the surrounding concrete. The GFRP bars and
the specimens only reinforced with CFRP, GFRP, or steel stirrups were made of continuous boron-free glass longi-
longitudinal bars. The torsional behavior and strength of all tudinal fibers impregnated in a vinyl-ester resin with fiber
specimens without transverse reinforcement were similar. contents of 83.10% and 79.50%, respectively, in accordance
Hadhood et al. (2020) demonstrated the torsional strength with CSA S807 (2019). Number 5 Grade III (nominal diam-
of large-scale rectangular members reinforced with GFRP eter of 15.90 mm) GFRP bars (CSA S807:19) were used as
bars and spirals. They concluded that the ultimate torsional longitudinal reinforcement in this study. Number 3 Grade II
strength of the specimen with no web reinforcement coin- (nominal diameter of 9.50 mm) GFRP stirrups (CSA
cided with the cracking torsional strength. In addition, the S807:19) were used at 1 m from both sides of the box girder
failure of this specimen was controlled by concrete splitting. as stiffeners. The stress-strain relationship of the GFRP bars
In contrast, Khagehhosseini et al. (2013) conducted an exper- was linear elastic in tension up to failure. The ultimate tensile
imental and numerical investigation to evaluate the effect of strength ffu and modulus of elasticity Ef of the GFRP bars
the longitudinal reinforcement ratio on the torsional capacity and straight portions of the GFRP stirrups were determined
of RC members reinforced with steel reinforcement and according to ASTM D7205/D7205M (2021), as reported by
without transverse torsional reinforcement. Their test results the manufacturer. In addition, the ultimate tensile strength
showed that increasing the longitudinal reinforcement ratio of the GFRP stirrups at the bent portions ffu,bent was deter-
resulted in a consistent increase in the post-cracking strength mined according to ASTM D7914/D7914M (2021). The
and torsional rigidity. steel control specimen was reinforced in the longitudinal
So far, the torsional strength and behavior of GFRP-RC direction with deformed M15 (nominal diameter of 16 mm)
box girders have not been investigated. In addition, exper- steel bars and at 1 m from both sides with M10 (nominal
imental results have yet to demonstrate the influence of the diameter of 9.50 mm) stirrups. Table 1 gives the mechanical
longitudinal reinforcement ratio on the torsional behavior properties of the GFRP and steel reinforcement.
of GFRP-RC box girders. This study discusses the exper- Concrete—The specimens were cast from the same
imental results from five full-scale concrete box girders concrete batch with normal-strength, normalweight concrete
reinforced internally with longitudinal GFRP and steel bars provided by a local ready mixed supplier; all specimens
without web reinforcement to study the contribution of the were cast at the University of Sherbrooke, Sherbrooke, QC,
longitudinal reinforcement to the torsional strength of the Canada. Fifteen concrete cylinders measuring 100 x 200 mm
specimens. (4 x 8 in.) were cast and cured under the same conditions
as the test specimens. Twelve were tested in compression
RESEARCH SIGNIFICANCE according to ASTM C39/C39M (2021), including six after
Valuable experimental work has been conducted to inves- 28 days and six cylinders on the day of specimen testing. The
tigate the torsional behavior of solid concrete sections rein- remaining three cylinders were tested in tension according
forced with FRP. The torsional behavior and strength of to ASTM C496/C496M (2017) by performing split cylinder
FRP-RC box girders, however, have not been addressed. The tests. Table 2 provides the average compressive (fc′) and
effect of different test parameters on the cracking patterns, tensile (ft′) strength of the concrete on the testing day of the
modes of failure, cracking and ultimate strengths, measured specimens.
crack widths, and strain behavior of the box girders was
investigated. Furthermore, the torsional moment-twist Specimen fabrication
curves were predicted theoretically. This study also exam- Before the concrete was cast, the cages were placed in the
ined the accuracy of the available theories and design forms; spacers were used to keep the concrete cover equal on
provisions for the cracking torsional strength of GFRP-RC all sides during casting. All specimens were cast at the same
box girders. In addition, the ultimate torsional strength of time from the same concrete batch. Two electrical needle
the GFRP-RC box girders without web reinforcement was vibrators were used to ensure adequate concrete quality and
estimated with the CSA S806-12 design equation with a prevent segregation. The cages were expected to rise during
modification related to the GFRP tensile strain. The results the casting process for two reasons: the expanded polysty-
reported in this paper represent a significant contribution to rene rigid insulation panel representing the void, and the
the relevant literature and provide end users, engineers, and weight of the GFRP bars. Consequently, two threaded rods
code committees with much-needed data and recommenda- were attached to the top of the cages and to a wooden plate
tions to advance the use of GFRP reinforcement in RC box that was fastened to the formwork with two steel angles, as
@seismicisolation
@seismicisolation

Fig. 1—(a) Sand-coated No. 5 GFRP bars; (b) casting; (c) storage; and (d) curing.
Table 1—Mechanical properties of GFRP and steel reinforcement
Elastic tensile
Bar size Bar type Diameter, mm Nominal area, mm2 Immersion area, mm2 modulus, GPa Tensile strength, MPa Tensile strain, %
ffu = 1225
No. 3 GFRP stirrups 9.50 71 88 58.20 2.10
ffu,bent = 671
No. 5 GFRP bars 15.90 198 230 62.50 ffu = 1500 2.40
M10 Steel stirrups 9.50 71 — 200 fy = 460 0.23
M15 Steel bars 16 200 — 200 fy = 460 0.23
Note: 1 mm = 0.0394; 1 mm2 = 0.00155 in.2; 1 MPa = 0.145 ksi.
Table 2—Test matrix and specimen details

Longitudinal reinforcement
No. of bars
Specimen ID fc′, MPa ft′, MPa Bar type Uniform distribution Reinforcement ratio ρL, %
BSW-12 39.84 3.33 Steel 12 M15 1.66
BGW-8 39.84 3.33 GFRP 8 No. 5 1.10
BGW-12 39.84 3.33 GFRP 12 No. 5 1.66
BGW-16 39.84 3.33 GFRP 16 No. 5 2.20
BGW-20 39.84 3.33 GFRP 20 No. 5 2.74
Note: ρL = AL/Ac, where AL is total area of longitudinal reinforcement and Ac is area of outer perimeter of concrete cross section. 1 MPa = 0.145 ksi.
shown in Fig. 1(b). Once the concrete was cast, the spec- box girders reinforced with longitudinal GFRP bars but
imens were covered with wet burlap and plastic sheets to without transverse reinforcement. A total of five large-scale
prevent moisture loss and were cured for 7 days. All speci- RC box girders—including four reinforced with longitudinal
mens were stored in the laboratory until the day of testing. GFRP bars and one reinforced with longitudinal steel bars
Figure 1 shows the casting, curing, and storage process of as a control specimen—were tested under pure torsional
the test specimens. loading. The test matrix was designed to study the effect of
longitudinal reinforcement on the torsional behavior of RC
Specimen design box girders reinforced with GFRP bars. The test parame-
The experimental program of this study was designed to ters included the type of longitudinal reinforcement (GFRP
provide experimental data on the torsional behavior@seismicisolation
of RC versus steel) and the longitudinal reinforcement ratio. The
@seismicisolation

Fig. 2—Reinforcement details and dimensions of RC box girders. (Note: Dimensions in mm; 1 mm = 0.0394 in.)
specimens were 4000 mm (157.48 in.) long, 380 mm (15 in.) hollow with a 180 mm (7 in.) square void that consisted of
wide, and 380 mm (15 in.) deep, with a wall thickness of thick sheets of expanded polystyrene rigid insulation panel.
100 mm (4 in.). The cross-sectional dimensions were chosen Three 60 mm (2.36 in.) thick layers of expanded polystyrene
to be similar to those used by Mitchell and Collins (1974) rigid insulation panel were combined to create the square
(beams PT5 and PT6). The test region was kept constant void measuring 180 mm wide (7 in.) by 2000 mm (78.74 in.)
at 2000 mm (78.74 in.) to ensure that each specimen long. The concrete cover was kept clear at 40 mm (1.57 in.)
developed at least one complete helical crack. All girders in all specimens. Figure 2 shows the dimensions and rein-
had a 1000 mm (40 in.) solid section overhang past the forcement details of the test specimens. The GFRP-RC
supports as an anchorage length on each side to prevent box girders were reinforced longitudinally with eight, 12,
any premature failure before torsional failure. Moreover, 16, and 20 No. 5 (15.9 mm) GFRP bars. The GFRP bars
the solid section was reinforced with transverse reinforce- were uniformly distributed around the perimeter of the cross
ment spaced at 150 mm (6 in.). The test region was made section, with reinforcement ratios of 1.10%, 1.66%, 2.20%,
@seismicisolation
@seismicisolation

Fig. 3—Instrumentation details of test specimens. (Note: Dimensions in mm; 1 mm = 0.0394 in.)
and 2.74%, respectively. The longitudinal reinforcement of a microscope and measuring scale) was used to measure
ratio (ρ) was calculated by dividing the total nominal area the initial crack width, which ranged from 0.10 to 0.15 mm.
of longitudinal reinforcement by the total cross-sectional Subsequently, two high-accuracy LVDTs were installed to
area of the girder. The control steel-reinforced specimen was measure the crack width electronically with load increase.
reinforced longitudinally with 12 M15 (16 mm) deformed The instrumentations for measuring the load, twist, concrete
steel bars with a reinforcement ratio of 1.66%. The speci- surface strain, longitudinal bar strains, and the crack width
mens were designated with this pattern: the letter B refers of the girder were recorded on a data acquisition system
to the girder specimen, the letters G and S stand for GFRP connected to a computer. The loading procedures of the actu-
or steel as longitudinal reinforcement, and the letter W ator prescribed a displacement-control rate of 0.5 mm/min.
indicates specimens without transverse reinforcement. The A torsion test setup was designed and fabricated at the
number stands for the total number of longitudinal bars in University of Sherbrooke’s structural laboratory for testing
each specimen. Table 2 provides the test matrix and details the RC box girders under pure torsional loading. It was
of the test specimens. similar to that used by Koutchoukali and Belarbi (2001).
The torsional moment was applied to the RC box girder with
Instrumentation and test setup a servo-controlled, 1000 kN (224.8 kip) hydraulic actuator
Electrical-resistance strain gauges with a gauge length attached to a rigid steel arm fastened to the girder. The load
of 6 mm (0.24 in.) and a gauge factor of 2.09% were used had a 715 mm (28.15 in.) lever arm from the centroidal axis
to measure the strain in the longitudinal reinforcing bars. of the girder, giving the test rig a 715 kN∙m (527.40 kip∙ft)
Five strain gauges were mounted on longitudinal bars in the torque capacity. All specimens were supported on a fixed
middle section of the test region, as shown in Fig. 3(a). To hinge 2000 mm (78.74 in.) apart. The hinge support was
measure the concrete strains, four strain gauges with a gauge designed to allow rotation about the longitudinal axis of the
length of 60 mm (2.36 in.) and a gauge factor of 2.06% were girder. The fixed support was a rigid steel beam resting on
attached to the concrete front and top surfaces at 45 and the girder and attached to the laboratory’s rigid floor with
135 degrees measured from the longitudinal axis. Further- two steel bars. The girders were allowed to slide and elon-
more, three linear variable differential transformers (LVDTs) gate freely to avert any longitudinal strain and subsequent
were used to measure the concrete surface strain as well as compression after cracking. This was achieved by resting the
specimen longitudinal elongation. The three LVDTs were girder on rollers at the fixed end. Figure 4 shows the details
mounted on the concrete front surface in a rosette format of the test setup.
to measure average concrete strains in three directions at
the midspan of the test region, as shown in Fig. 3(b). Two TEST RESULTS AND DISCUSSION
LVDTs were placed at 45 and 135 degrees measured from Cracking pattern and modes of failure
the longitudinal axis of the girder; the other one was placed Crack formation and propagation in each girder and the
along the longitudinal axis of the girder. To measure the rela- corresponding loads were marked and recorded during
tive rotation of the cross section, four potentiometers were testing. When the applied principal tensile stresses reached
placed at two different locations within the test region. In the concrete tensile strength, the first crack initiated in all
addition, one potentiometer was placed under the applied specimens at the middle of the front face of the test region.
vertical load. Two potentiometers were placed at the fixed The cracking torsional strength of girders BGW-8, BGW-12,
end to make sure that no vertical movement was induced BGW-16, BGW-20, and BSW-12 was 36.1, 36.6, 36.9, 36.2,
during specimen testing. Once the first crack appeared, a and 36.9 kN∙m (26.64, 27.01, 27.23, 26.72, and 27.23 kip∙ft),
crack comparator (an optical measurement device consisting respectively. As the applied torsional loading increased,
@seismicisolation
@seismicisolation

Fig. 4—Test setup schematic.
more cracks appeared on each side and then linked with the cross sections and no transverse reinforcement (Mohamed
crack on the front face, forming a spiral shape along the and Benmokrane 2016; Hadhood et al. 2020). The mode of
periphery of the specimen. The cracking pattern was affected failure observed in these studies was concrete splitting, with
by the longitudinal reinforcement ratio. Increasing the longi- a failure plane angle that ranged between 27 and 30 degrees
tudinal reinforcement ratio increased the number of cracks in the longitudinal direction. Figure 5 shows the observed
and decreased crack inclination in the longitudinal direction. cracking patterns and modes of failure of the test specimens.
The cracking patterns of specimens BGW-12 and BSW-12
were almost similar in terms of the failure plane angle and Torsional moment-twist response
the number of cracks. The failure plane angle of specimens This section presents the torsional moment-twist curves
BGW-8, BGW-12, BGW-16, BGW-20, and BSW-12 was for the tested specimens in two groups to show the effect of
approximately 55, 49, 40, 37, and 47 degrees, respectively, test parameters on the torsional behavior of the GFRP-RC
with respect to the longitudinal direction. Generally, the box girders, as depicted in Fig. 6. From initial loading until
principal stress distribution in RC members subjected to the formation of the first diagonal concrete crack, all spec-
pure torsional loading consists of diagonal compression and imens displayed almost identical linear torsional moment-
tension. Therefore, all cracks appeared diagonally, similar to twist responses, regardless of their reinforcement ratio and
what was observed in past studies on GFRP-RC members type, representing the uncracked condition governed by the
under torsion (Mohamed et al. 2015). The cracks observed properties of the concrete section, as shown in Fig. 6(a)
on the face of the specimens with a high longitudinal rein- and (b). The uncracked torsional stiffness of all the spec-
forcement ratio (BGW-16 and BGW-20) were diagonal. imens ranged between 11,650 and 14,470 kN∙m2 (28,193
Unlike in the case of girders reinforced with a relatively and 35,017 kip∙ft2). As the load increased, torsional cracking
lower longitudinal reinforcement ratio (BGW-8, BGW-12, eventually occurred within the torsion test region. At this
and BSW-12), the cracks started diagonally, followed by stage, all the specimens exhibited a very low twisting
more longitudinal cracks that propagated from the diagonal angle (ranging from 0.0026 to 0.0031 rad/m [0.00079 to
cracks. These longitudinal cracks were due to dowel action 0.00095 rad/ft]), which reflects the concrete’s gross section
in the longitudinal bars, which were distributed uniformly stiffness. After cracking, the GFRP-RC box girders rein-
throughout the box girder. The dramatic increase in the forced with a relatively lower longitudinal reinforcement
dowel action in these bars produced vertical tensile stress in ratio (1.10 and 1.66%) experienced a significant loss of
the concrete around the bars. In general, all specimens failed strength and stiffness. In contrast, the specimens reinforced
by concrete splitting due to a lack of transverse torsional with a higher longitudinal reinforcement ratio (2.20 and
reinforcement. A major diagonal spiral crack developed 2.74%) exhibited an increase in torsional strength after
within the formed crushed surface at the midheight of cracking. Table 3 presents the cracking torsional strength
the front face. A similar mode of failure was observed in and the corresponding twist and stiffness, the post-cracking
previous studies on FRP-RC beams with solid rectangular @seismicisolation
@seismicisolation

Specimens BGW-8 and BGW-12, with relatively lower rein-
forcement ratios (1.10 and 1.66%), exhibited a sudden drop
in the torsional moment with the appearance of the first diag-
onal crack, as shown in Fig. 6(a). The reduction in torsional
strength was 36.5% and 32.9% of the cracking torsional
strength for specimens BGW-8 and BGW-16, respectively.
This reduction could be attributed to the absence of the trans-
verse reinforcement and the fact that such a low longitudinal
reinforcement ratio did not provide enough confinement to
the concrete hollow core to allow the redistribution of the
internal forces developed in the specimens. Subsequently,
the specimens (BGW-8 and BGW-12) showed a slight
increase in torsional strength after their strength reduction
by almost 9% and 12%, respectively. After that, increasing
the angle of twist resulted in a drop in stiffness and strength.
Consequently, it can be concluded that increasing the GFRP
longitudinal reinforcement ratio of the RC box girders from
1.10% (BGW-8) to 1.66% (BGW-12) had an insignificant
impact on the post-cracking torsional stiffness strength.
Moreover, the ultimate capacity of these specimens was
controlled by concrete splitting and coincided with the corre-
sponding cracking torque. This finding is in good agreement
with the test results of Mohamed and Benmokrane (2016)
and Hadhood et al. (2020). The ultimate torsional capacity of
the specimens with GFRP longitudinal reinforcement ratios
of 1.33 and 0.75% but no web reinforcement coincided with
the cracking strength. Figure 6(a) indicates that, for the spec-
imens with relatively higher GFRP longitudinal reinforce-
ment ratios (BGW-16 and BGW-20), a significant increase
in the torsional strength was observed after the first diag-
onal crack appeared. BGW-16 and BGW-20 had enhanced
strength of approximately 8% and 14%, respectively, after
the initiation of the first diagonal crack. This is in good
agreement with the experimental studies reported by Hsu
(1968) for specimens with only longitudinal reinforcement.
The test results revealed that increasing the longitudinal
reinforcement ratio increased the torsional strength up to
15%. Subsequently, BGW-16 and BGW20 exhibited gradual
reductions in their strength of 26% and 14%, respectively,
with increased angles of twist. After that, a torque-twist
plateau formed, and degradation in stiffness and strength
resulted from increasing the angle of twist.
Increasing the reinforcement ratio by approximately 100%
and 150% (from 1.10 to 2.20% and from 1.10 to 2.74%)
increased the torsional strength by 10% and 14.2%, respec-
tively. In addition, increasing the reinforcement ratio by
33% and 67% (from 1.66 to 2.20% and from 1.66 to 2.74%)
enhanced the torsional strength by 8.5% and 12.8%, respec-
tively. The increase in the torsional strength was more notice-
Fig. 5—Cracking patterns at failure of test specimens. able in the specimens with relatively higher reinforcement
torsional strength and the corresponding twist and stiffness, ratios. The torsional strengths of all GFRP-RC box girders
and the failure mode. were compared versus the longitudinal reinforcement ratio
at 0.01 rad/m (0.003 rad/ft) (approximately four times the
Effect of longitudinal reinforcement ratio on angle of rotation at cracking) to evaluate their post-cracking
torque-twist response torsional strength, as shown in Fig. 7. This figure indicates
Figure 6(a) shows the effect of the longitudinal rein- that increasing the longitudinal reinforcement ratio had a
forcement ratio on the torsional moment-twist behavior. significant effect on the post-cracking torsional strength of
the GFRP-RC box girders.
@seismicisolation
@seismicisolation

Table 3—Experimental test results
Specimen ID Tcr, kN∙m φcr, rad/m Kun(exp), kN∙m2 Tu, kN∙m φu, rad/m Kcr(exp), kN∙m2 Toughness, kN Failure mode
BSW-12 36.88 0.0026 14,185 39.30 0.0030 6050 1.04 Concrete splitting
BGW-8 36.10 0.0031 11,650 36.10 — — 0.43 Concrete splitting
BGW-12 36.55 0.0029 12,600 36.55 — — 0.93 Concrete splitting
BGW-16 36.90 0.0030 12,725 39.65 0.0039 3055 1.21 Concrete splitting
BGW-20 36.18 0.0025 14,470 41.20 0.0037 4190 1.82 Concrete splitting
Note: Tcr is torsional moment corresponding to first diagonal crack; Tu is maximum torsional moment; Kun(exp) is uncracked torsional stiffness (Tcr/φcr); and Kcr(exp) is cracked
torsional stiffness ([Tu – Tcr]/[φu – φcr]). 1 kN∙m = 0.7376 kip∙ft; 1 kN∙m2 = 2.42 kip∙ft2; 1 kN = 0.225 kip; 1 rad/m = 0.305 rad/ft.
Fig. 6—Effect of test parameters on torsional moment-twist behavior. (Note: 1 kN∙m = 0.7376 kip∙ft; 1 rad/m = 0.305 rad/ft.)
Effect of type of longitudinal reinforcement on
torque-twist response
Figure 6(b) shows the effect of longitudinal reinforce-
ment type on the torsional moment-twist behavior. Spec-
imen BGW-12 and its counterpart steel-reinforced specimen
(BSW-12) were designed to have the same longitudinal
reinforcement ratio (1.66%). The figure indicates that the
cracking strength and the pre-cracking torsional stiffness
were almost the same for both specimens. Table 3 indicates
that the torsional cracking strength of BGW-12 and BSW-12
was 36.55 and 36.88 kN∙m (26.97 and 27.22 kip∙ft), respec-
tively. After cracking occurred, the specimens’ torsional
performance was highly dependent on the axial stiffness
of the longitudinal reinforcing bars, which is a function of
the area A and modulus of elasticity E of the longitudinal
reinforcement. Figure 6(b) indicates that the steel-reinforced
specimen (BSW-12) experienced less of a drop in torsional
strength than its counterpart GFRP-reinforced specimen
(BGW-12) after the first diagonal crack appeared. In contrast,
BSW-12 achieved a slight increase in torsional strength of
Fig. 7—Ratio of post-cracking to cracking strength versus
7% after the first crack, followed by a reduction of 10% in
reinforcement ratio.
ultimate torsional strength. The reduction in the torsional
strength of the GFRP-RC box girder (BGW-12) was almost (200 GPa [29,000 ksi]) to that of the GFRP bars (62.5 GPa
3.29 times that of its steel counterpart (BSW-12). This value [9063 ksi]). Afterward, the two specimens showed degrada-
is similar to the ratio of the modulus of elasticity of the@seismicisolation
steel tion in stiffness and strength as the angle of twist increased.
@seismicisolation

Fig. 8—Torsional moment-longitudinal strain behavior.
Fig. 9—Torsional moment-concrete strain behavior. (Note:
(Note: 1 kN∙m = 0.7376 kip∙ft.)
1 kN∙m = 0.7376 kip∙ft.)
Torsional moment-longitudinal strain behavior readings are significantly lower than the concrete crushing
Figure 8 plots the measured applied torsional moment on strain of 3000 με specified in ACI 318-19 (ACI Committee
the specimens versus the strain behavior relationships for the 318 2019) and 3500 με specified in CSA S806-12 (2017)
internal GFRP and steel longitudinal bars. All the RC box and CSA A23.3 (2019), indicating that the specimens failed
girders had negligible strain readings and similar behavior by concrete splitting. Generally, the diagonal concrete strain
from the initial loading up to the first diagonal crack, regard- dropped by approximately 46% with a 150% increase in the
less of the reinforcement ratio and type. After cracking longitudinal reinforcement ratio (from 1.10% in BGW-8
occurred, a redistribution of the internal forces from the to 2.74% in BGW-20). In contrast, BSW-12 showed lower
concrete section to the longitudinal reinforcement resulted strain values compared to all GFRP-RC box girders at the
in a significant increase in the longitudinal reinforcement same torque level. This can be attributed to the effect of
strains. The specimens reinforced with longitudinal GFRP the axial stiffness of the longitudinal reinforcement on the
bars exhibited higher bar strain after concrete cracking than induced strain deformation of the tested specimens.
the steel-reinforced specimen at the same torque level. This
could be attributed to the variation in the modulus of elas- Torsional crack width
ticity of the two materials. In contrast, increasing the longi- The crack widths for all specimens were measured with
tudinal reinforcement ratio from 1.10 to 1.66, 2.20, and two LVDTs and recorded on a data acquisition system
2.74% decreased the measured strain value at all load levels. throughout the testing. Figure 10 shows the measured crack
The maximum recorded strains in the GFRP bars were 2400, width versus the applied torsional moment for all specimens.
2160, 1840, and 1580 με, representing 10%, 9%, 7.7%, Each curve started with the cracking torsional moment and
and 6.6% of the ultimate tensile GFRP bar strain for spec- terminated with specimen failure. The experimental investi-
imens BGW-8, BGW-12, BGW-16, and BGW-20, respec- gation and numerical analysis conducted by Park et al. (2001)
tively. Generally, these low strain values emphasize that the revealed that the maximum crack width was controlled
torsional failure was not controlled by the rupture of GFRP by the relative amounts of torsional reinforcement in the
longitudinal bars. Furthermore, the test results indicate that longitudinal direction. As shown in Fig. 10, increasing the
no slippage or anchorage problems occurred throughout longitudinal reinforcement ratio decreased the crack widths
testing. In contrast, the maximum recorded strain value of at specimen failure. In addition, specimen BGW-12 exhib-
BSW-12 was 1060 με, which is less than the yield strain of ited wider crack widths than BSW-12, which was reinforced
the steel reinforcement. with steel. The maximum recorded crack width at the end of
the test was 3.70, 2.90, 2.10, 1.50, and 1.80 mm (0.15, 0.11,
Torsional moment-diagonal compressive 0.083, 0.059, and 0.071 in.) for BGW-8, BGW-12, BGW-16,
concrete strain BGW-20, and BSW-12, respectively.
Figure 9 gives the measured diagonal compressive strains
versus the applied torsional moment for the GFRP- and Torsional toughness
steel-reinforced specimens. The figure indicates that the diag- Toughness refers to the material’s capacity to absorb energy
onal compressive strain decreased by increasing the GFRP and deform plastically without fracturing. The torsional
longitudinal reinforcement ratio. In addition, all specimens toughness of the examined specimens can be estimated as
exhibited a minimal value of diagonal compressive concrete the area under torsional moment-twist curves, as illustrated
strain up to failure. The recorded diagonal compressive in Fig. 6, because the cross-sectional dimensions of all the
strain corresponding to the maximum torsional moment was specimens were constant. Table 3 shows the torsional tough-
–280, –175, –160, –150, and –135 με for BGW-8, BGW-12, ness of all the test specimens. As can be seen, increasing
BGW-16, BGW-20, and BSW-12, respectively. @seismicisolation
These
@seismicisolation

higher GFRP longitudinal reinforcement ratios (2.20 and
2.74%) achieved torsional strength higher than the cracking
strength. Consequently, the torsional moment-twist curve
for these specimens can be predicted theoretically until the
second zone. The following sections discuss the prediction
of the torsional moment-twist curve theoretically for the first
and second zones.
Uncracked torsional behavior (first zone)

Pre-cracking torsional stiffness—Saint-Venant’s theory
can be used to calculate the torsional stiffness of RC members
before cracking with reasonable accuracy (Hsu 1973). The
torsional stiffness can be calculated as
_T

Kun(Theo)
= φ (1)
Fig. 10—Torsional moment-crack width. (Note: 1 kN∙m = where T is the applied torque to the girder; φ is the twist
0.7376 kip∙ft; 1 mm = 0.0394 in.) per unit length; and Kun(Theo) is the theoretical pre-cracking
the longitudinal reinforcement ratio from 1.10% in BGW-8 torsional stiffness (Kun(Theo) = RGC); R is the reduction factor
to 2.74% in BGW-20 increased the torsional toughness by (R ≈ 0.60); G is the shear modulus of concrete and is equal to
approximately 325%. The steel-reinforced specimen exhib- Ec/(2(1 + υ)); Ec is the modulus of elasticity of the concrete;
ited torsional toughness approximately 12% higher than its υ is the Poisson’s ratio; and C is the Saint-Venant torsional
counterpart with GFRP reinforcement. It can be concluded constant.
that the torsional toughness significantly increased as the The reduction factor (R) considers the stiffness loss
longitudinal reinforcement ratio increased. observed in the laboratory tests, which was almost 20 to
40% of the Kun(Theo) value. It was already reported by Leon-
THEORETICAL INVESTIGATION hardt and Schelling (1974) and adopted by Bernardo and
Theoretical prediction of torsional Lopes (2008). The torsional reinforcement can be neglected
moment-twist curve in calculations of the Kun(Theo) (Bernardo and Lopes 2008;
In RC members subjected to the pure torsional moment, Hsu 1973). For rectangular hollow sections, the Saint-
the torque is generally resisted by longitudinal rein- Venant torsional constant (C) is calculated as
forcement, transverse reinforcement, and concrete diag-
onal struts. Therefore, the torsional moment-twist curve _
4A2h
C = u (2)
comprises three distinct zones (Bernardo and Lopes 2008).
The first zone is characterized by the linear-elastic behavior where A is the area enclosed by the centerline of the wall
of materials and represents the torsional behavior until the of the hollow section; h is the wall thickness of the hollow
first diagonal crack occurs. The cracking strength in this section; and u is the perimeter of the centerline of the wall of
zone can be predicted theoretically with the skew-bending the hollow section.
theory, Bredt’s thin-tube theory, and the theory of elasticity. Cracking torsional strength—The cracking torsional
The slope of the curve in the first zone represents uncracked strength (Tcr) can be theoretically predicted with three
torsional stiffness, which can be predicted with Saint- different theories (the skew-bending theory, Bredt’s thin-
Venant stiffness. The second zone corresponds to the torsional tube theory, and the theory of elasticity), in addition to three
behavior in a cracked state with linear-elastic behavior of codes (CSA A23.3:19, CSA S6:19, and AASHTO 2018),
the materials, and the slope of the curve represents the post- which are briefly described as follows.
cracking torsional stiffness. The space-truss analogy with an Skew-bending theory—Hsu (1968) derived an equation
inclination angle of diagonal concrete struts of 45 degrees for the nominal torsional strength of solid concrete members
can theoretically predict the torsional behavior in the second based on the bending mechanism of torsional failure. This
zone. The third zone represents the nonlinear behavior of theory contends that the failure of a torsional member is
the materials with a softening effect because at least one of reached when the tensile stress induced by a 45-degree
the reinforcement types (longitudinal or transverse) reaches bending component of torque on the wider face reaches
the ultimate strength or because of the nonlinearity of the a reduced modulus of rupture of concrete. The torsional
diagonal concrete struts. In contrast, based on the experi- cracking strength is given by
mental test results from this study, the ultimate torque for
the specimens reinforced with relatively lower GFRP longi- ____
fc′ ( x )for h ≤ 4
3 _
4h _x
tudinal reinforcement ratios (1.10 and 1.66%) coincided
T
cr = 6y(
x
2
+ 10)
√ (3)
with the cracking torque. Therefore, the torsional moment-
twist curve for these specimens terminated in the first where x and y are the shorter and longer dimensions of
zone. In contrast, the specimens reinforced with relatively the cross section (in.), respectively; and fc′ is the concrete
@seismicisolation
@seismicisolation

compressive strength (psi). If h > (x/4), h should be consid- be is the minimum thickness of the exterior webs or flanges
ered as h = (x/4). comprising the closed box section (in.) and does not exceed
Bredt’s thin-tube theory—Hsu and Mo (1985) proposed a Ac/Pc. Note that fc′ is in ksi.
formula for the torsional cracking of tubular sections based
on Bredt’s thin-tube theory. This theory relates to the shear Cracked torsional behavior (second zone)
stresses due to torsion in a thin-walled tube. The shear stress Post-cracking torsional stiffness—Hsu (1973) proposed
is set equal to the tensile strength of concrete in a biaxial an equation to predict the torsional stiffness after cracking
tension-compression state. The torsional cracking strength by considering a concrete thin tube with a constant wall
is expressed as thickness. He demonstrated the validity of this theory
____
through laboratory tests in which the torsional behavior after
= 2Act(2.5√fc′ )
Tcr (4) cracking was unaffected by the concrete core of the cross
section. The post-cracking torsional stiffness equation was
where Ac is the area enclosed by the outer perimeter of the derived based on the equilibrium and compatibility equa-
hollow section (in.2); and t is the wall thickness of the hollow tions of the space-truss analogy proposed by Rausch (1953).
section (in.). The post-cracking torsional stiffness is given as
Theory of elasticity—Saint-Venant’s theory has also been
extended to calculate the torsional cracking strength. The Esx12y12Ac
_______________________

Kcr(Theo)
=   
    (9)
(x1+ y1)2(
(x1+ y1)h ρl)
torsional failure of a solid concrete section occurs when _ 2nAc _1
+
the maximum principal tensile strength σmax is equal to the
tensile strength of concrete ft′. When σmax = τmax, the elastic
cracking torque can be considered as where Es is the modulus of elasticity of the reinforcement;
x1 and y1 are the shorter and longer dimensions of the area
Tcr = 2Ahft′ (5) limited by the centerline of the reinforcement, respectively;
n = Es/Ec; and ρl is the longitudinal reinforcement ratio.
Canadian design code CSA A23.3:19—CSA A23.3:19 Ultimate torsional strength—CSA S806-12 provides an
presents the cracking torsional strength in accordance with equation to calculate the ultimate torsional strength of RC
the thin-walled tube, hollow space-truss analogy of the RC members based on the contribution of the GFRP longitudinal
cross section as follows reinforcement only. This equation is based on a thin-tube
space-truss analogy. Because the shear flow due to torsion
= 0.38ϕc√fc′ (P)
____ A2 is constant at all points around the tube perimeter, the resul-
_c

Tcr (6)
c tant of the axial tension force (N) acts through the midheight
of each wall. As a result, half of the axial tension force is
where Pc is the outer perimeter of the concrete cross section resisted by each of the top and bottom chords, as given in
(mm); and ϕc is the resistance factor for concrete and is equal Eq. (10). Longitudinal reinforcement with a tensile strength
to 0.65. For hollow sections, Ac is replaced by 1.5Ag if the AlEflεfl is required to resist the sum of the axial tension forces
wall thickness is less than 0.75(Ac/Pc); Ag is the gross area of acting in all walls of the tube, as given in Eq. (11)
the section. Note that fc′ is in MPa.
Canadian Highway Bridge Design Code CSA S6:19—The
= 2 = 2 = 1.3( 2A ) (10)
∑N _
_ AlEfl εfl 0.45poTf
_
Canadian Highway Bridge Design Code (CSA S6:19) uses
FTop
= FBottom

o
the same basic equation as CSA ____ A23.3:19 with a different
concrete tensile strength (0.33√fc′ ). This equation is identical Ao AlEfl εfl
___________
to the cracking torsional strength equation adopted in ACI
Tf = (11)
1.3(0.45po)
318-19, where the cracking torsional strength is given as
where Tf is the factored torsional resistance; Ao is the gross
fc′ (P)
Ac2 ____
_ area enclosed by the shear flow path around the tube perim-

Tcr
= 0.33ϕc√ (7)
c
eter; Al is the total area of the GFRP longitudinal bars; Po is
the length of the line joining the centers of the bars in the
AASHTO LRFD Bridge Design Guide Specifications
corners of the tube; Efl is the modulus of elasticity of the
(AASHTO 2018)—According to the “AASHTO LRFD
GFRP longitudinal bars; and εfl is the tensile strain of the
Bridge Design Guide Specifications for GFRP-Reinforced
GFRP longitudinal bars.
Concrete” (2018), the cracking torsional strength Tcr is given
Based on the test results, the observed tensile strain in the
as
GFRP longitudinal bars at specimen failure was insignificant
____ compared to its ultimate value. This is due to the specimens’
Tcr
= 0.126√fc′ 2Ao be (8)
failure mode, which was not controlled by the rupture of
the GFRP longitudinal bars. Past studies reported the same
where Ao is the area enclosed by the shear flow path,
behavior with the rectangular solid beams reinforced with
including any area of holes (in.2) (for hollow sections, this
only GFRP longitudinal bars (Mohamed and Benmokrane
can be taken as the area limited by the centerlines of the
2016; Hadhood et al. 2020). Therefore, in RC box girders
exterior webs and flanges that form the closed section); and
@seismicisolation
@seismicisolation

subjected to pure torsion, the tensile strain in the GFRP factor (R). The points corresponding to cracking and ulti-
longitudinal bars should be limited to εfl = 2000 με. mate strength are identified in the graph. In general, Fig. 11
shows that the theoretical curves predicted the experimental
Comparison of experimental and curves well, thereby validating the calculation procedure
theoretical results described previously. Figure 11 shows that the torsional
Considering the preceding discussion, the torsional strength calculated with Eq. (11) was nearly 4% lower
moment-twist curves were predicted theoretically for the than the ultimate torsional strength of specimen BGW-16
concrete box girders reinforced with only longitudinal but 9% higher than that of specimen BGW-20. In contrast,
GFRP bars. Figure 11 shows the experimental and theoret- Eq. (11) provided ultimate torsional strength for specimens
ical torsional moment-twist curves for all GFRP-reinforced BGW-8 and BGW-12 lower than their cracking torsional
box girders. The theoretical value of the cracking torsional strength. This is in good agreement with the experimental
strength in Fig. 11 was determined with Bredt’s thin-tube test results. Table 4 presents the ratios of the experimental to
theory. Furthermore, the ultimate torsional strength was esti- theoretical torsional stiffness in the first and second zones,
mated theoretically with Eq. (11). To assess the uncracked the theoretical cracking torsional moment, and the ratios of
torsional stiffness, a value of 0.6 was used for the reduction the experimental to theoretical cracking torsional moment
(Tcr(exp)/Tcr(Theo)). Different theories and code equations were
used to predict the cracking torsional moment—namely, the
skew-bending theory, Bredt’s thin-tube theory, the theory
of elasticity, CSA A23.3:19, CSA S6:19, and AASHTO
2018. The resistance factor of concrete ϕ for all codes was
taken as being equal to unity. The ____ concrete tensile strength
ft′ was taken as being equal to 5 √ fc′ in the prediction with
the theory of elasticity as reported by Hsu (1984). Table 4
indicates that the skew-bending theory and theory of elas-
ticity overestimated the cracking strength of the specimens,
as the average values of Tcr(exp)/Tcr(Theo) were 0.86 and 0.89,
respectively. In contrast, the results given by Bredt’s thin-
tube theory were in good agreement with the experimental
test results, with an average value of 0.96. Table 4 shows
that the CSA S6:19 equation underestimated the cracking
torsional strength, with an average Tcr(exp))/Tcr(Theo) value of
1.28. In contrast, the CSA A23.3:19 and AASHTO 2018
equations showed reasonable but rather conservative results,
where the average values of Tcr(exp)/Tcr(Theo) were 1.11 and
1.17, respectively. Saint-Venant’s theory overestimated the
uncracked torsional stiffness, as the average value of Kun(exp)/
Kun(Theo) was 0.84. Hsu’s equation accurately predicted
the post-cracking torsional stiffness, with an average
Fig. 11—Experimental versus theoretical torsional moment- Kcr(exp)/Kcr(Theo) value of 1.01.
twist curve. (Note: 1 kN∙m = 0.7376 kip∙ft; 1 rad/m =
0.305 rad/ft.)
Table 4—Experimental to theoretical cracking torque strength and stiffness assessment
Skew-bending Bredt’s thin-tube Theory of
theory theory elasticity CSA A23.3:19 CSA S6:19 AASHTO 2018
Specimen Kun(exp)/ Kcr(exp)/ Tcr(Theo), Tcr(exp)/ Tcr(Theo), Tcr(exp)/ Tcr(Theo), Tcr(exp)/ Tcr(Theo), Tcr(exp)/ Tcr(Theo), Tcr(exp)/ Tcr(Theo), Tcr(exp)/
ID Kun(Theo) Kcr(Theo) kN∙m Tcr(Theo) kN∙m Tcr(Theo) kN∙m Tcr(Theo) kN∙m Tcr(Theo) kN∙m Tcr(Theo) kN∙m Tcr(Theo)
BSW-12 0.91 0.99 42.54 0.87 37.88 0.97 41.11 0.90 32.96 1.12 28.63 1.29 31.15 1.18
BGW-8 0.75 — 42.54 0.85 37.88 0.95 41.11 0.88 32.96 1.10 28.63 1.26 31.15 1.16
BGW-12 0.81 — 42.54 0.86 37.88 0.96 41.11 0.89 32.96 1.11 28.63 1.28 31.15 1.17
BGW-16 0.82 0.95 42.54 0.87 37.88 0.97 41.11 0.90 32.96 1.12 28.63 1.29 31.15 1.18
BGW-20 0.93 1.10 42.54 0.85 37.88 0.96 41.11 0.88 32.96 1.10 28.63 1.26 31.15 1.16
Average 0.84 1.013 — 0.86 — 0.96 — 0.89 — 1.11 — 1.28 — 1.17
Standard
0.075 0.078 — 0.009 — 0.010 — 0.009 — 0.011 — 0.013 — 0.012
deviation
COV, % 8.85 7.67 — 1.047 — 1.042 — 1.011 — 0.99 — 1.02 — 0.103
Note: 1 kN∙m = 0.7376 kip∙ft. @seismicisolation

@seismicisolation

CONCLUSIONS AUTHOR BIOS
This study investigated the contribution of the longitudinal Ibrahim T. Mostafa is a Doctoral Candidate in the Department of Civil
and Building Engineering at the University of Sherbrooke, Sherbrooke, QC,
reinforcement to the torsional strength of reinforced concrete Canada. He received his BSc and MSc in civil engineering from the Faculty
(RC) box girders reinforced with longitudinal glass fiber- of Engineering, Helwan University, Helwan, Cairo, Egypt. His research
reinforced polymer (GFRP) bars but no transverse reinforce- interests include the use of fiber-reinforced polymers (FRPs) in reinforced
concrete structures.
ment. Five full-scale concrete box girders reinforced with
GFRP or steel bars were constructed and tested under pure Salaheldin Mousa is a Postdoctoral Fellow in the Department of Civil and
torsional loading up to failure. The test parameters included Building Engineering at the University of Sherbrooke, where he received
his PhD. He is also a Lecturer at the Faculty of Engineering at Shoubra,
the longitudinal reinforcement ratio and the type of longitu- Benha University, Benha, Egypt, where he received his BSc and MSc. His
dinal reinforcement (GFRP or steel). The torsional moment- research interests include the use of FRPs in reinforced concrete structures.
twist curves were predicted theoretically. In addition, the
Hamdy M. Mohamed is a Research Associate and Lecturer in the Depart-
experimental cracking strength was compared with the three ment of Civil and Building Engineering at the University of Sherbrooke. He
different theories and the current codes. The following are received his BSc and MSc from the Faculty of Engineering, Helwan Univer-
the main conclusions drawn from this investigation: sity, and his PhD from the University of Sherbrooke. His research inter-
ests include the use and field applications of FRPs in reinforced concrete
1. The torsional failure of all specimens was controlled structures.
by concrete splitting attributed to the absence of transverse
torsional reinforcement, and a major diagonal spiral crack Brahim Benmokrane, FACI, is a Professor in the Department of Civil
and Building Engineering at the University of Sherbrooke, Tier-1 Canada
developed with a crushed surface formed at the midheight of Research Chair in Advanced Composite Materials Used for Civil Engi-
the test region’s front face. neering Structures, Senior Industrial Research Chair in Innovative FRP
2. The test results pointed out that the increase in the GFRP Composite Materials for Concrete Infrastructure, and Director of the
University of Sherbrooke Research Center on Structural FRP Composite
longitudinal reinforcement ratio of specimens from 1.10% Materials for Concrete Structures (CRUSMAC). He is a member and past
(BGW-8) to 1.66% (BGW-12) had no significant effect on Co-Chair of ACI Subcommittee 440-K, FRP-Material Characteristics,
the post-cracking torsional stiffness and strength. Moreover, and a member of ACI Committees 435, Deflection of Concrete Building
Structures, and 440, Fiber-Reinforced Polymer Reinforcement; and ACI
the ultimate torsional capacity of these specimens coincided Subcommittees ACI 440-E, FRP-Professional Education; ACI 440-F,
with the corresponding cracking torque. FRP-Repair-Strengthening; 440-H, FRP-Reinforced Concrete; ACI 440-I,
3. The specimens with relatively higher GFRP longitu- FRP-Prestressed Concrete; and ACI 440-L, FRP-Durability. He received
the ACI Foundation Arthur J. Boase Award in 2022. His research interests
dinal reinforcement ratios (BGW-16 and BGW-20) exhib- include the development of FRP reinforcement for concrete structures and
ited an increase in their torsional strength of approximately their durability, structural performance, and field applications.
8% and 14%, respectively, after the appearance of the first
diagonal crack. ACKNOWLEDGMENTS
This research was conducted with funding from the Natural Sciences and
4. The torsional strength and stiffness of the RC box Engineering Research Council of Canada (NSERC), the NSERC-Alliance
girders without web reinforcement after cracking were grant program, the Fonds de recherche du Québec – Nature et technologies
highly dependent on the axial stiffness of the longitudinal (FRQNT), and the Tier-1 Canada Research Chair in Advanced Composite
Materials Used for Civil Engineering Structures. The authors are grateful
reinforcing bars. to Pultrall Inc. (Thetford Mines, QC, Canada) for the donation of GFRP
5. The ultimate torsional strength of RC box girders rein- materials, and the technical staff of the Canada Foundation for Innovation
forced with only GFRP longitudinal bars was predicted with (CFI) structural lab in the Department of Civil and Building Engineering at
the University of Sherbrooke.
the CSA S806-12 (2017) design equation with a modifica-
tion related to the GFRP tensile strain.
REFERENCES
6. The skew-bending theory and theory of elasticity over- AASHTO, 2018, “AASHTO LRFD Bridge Design Guide Specifications
estimated the predictions of the cracking torsional strength for GFRP-Reinforced Concrete,” second edition, American Association of
of the tested specimens, whereas Bredt’s thin-tube theory State Highway and Transportation Officials, Washington, DC.
ACI Committee 318, 2019, “Building Code Requirements for Structural
was in good agreement with the experimental test results. Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved
7. The torsional cracking strength equation in CSA S6:19 2022),” American Concrete Institute, Farmington Hills, MI, 624 pp.
(2019) provided conservative predictions for all specimens ACI Committee 440, 2015, “Guide for the Design and Construction of
Structural Concrete Reinforced with Fiber-Reinforced Polymer (FRP) Bars
tested. In contrast, the CSA A23.3:19 (2019) and AASHTO (ACI 440.1R-15),” American Concrete Institute, Farmington Hills, MI,
2018 (2018) equations yielded reasonable but rather conser- 88 pp.
vative results. ASTM C39/C39M-21, 2021, “Standard Test Method for Compressive
Strength of Cylindrical Concrete Specimens,” ASTM International, West
Based on the results found, this study considerably Conshohocken, PA, 8 pp.
contributes to an understanding of the torsional strength ASTM C496/C496M-17, 2017, “Standard Test Method for Splitting
and behavior of RC box girders reinforced with longitudinal Tensile Strength of Cylindrical Concrete Specimens,” ASTM International,
West Conshohocken, PA, 5 pp.
GFRP bars under a pure torsional moment. Future work ASTM D7205/D7205M-21, 2021, “Standard Test Method for Tensile
should focus on the structural performance of concrete box Properties of Fiber Reinforced Polymer Matrix Composite Bars,” ASTM
girders reinforced internally with longitudinal GFRP bars International, West Conshohocken, PA, 13 pp.
ASTM D7914/D7914M-21, 2021, “Standard Test Method for Strength
and with GFRP web reinforcement to study the contribu- of Fiber Reinforced Polymer (FRP) Bent Bars in Bend Locations,” ASTM
tion of the web reinforcement to the torsional behavior and International, West Conshohocken, PA, 6 pp.
strength. Bernardo, L. F. A., and Lopes, S. M. R., 2008, “Behaviour of Concrete
Beams under Torsion: NSC Plain and Hollow Beams,” Materials and Struc-
tures, V. 41, No. 6, July, pp. 1143-1167. doi: 10.1617/s11527-007-9315-0
CSA A23.3:19, 2019, “Design of Concrete Structures,” CSA Group,
Toronto, ON, Canada, 301 pp.
@seismicisolation
@seismicisolation

CSA S6:19, 2019, “Canadian Highway Bridge Design Code,” CSA Leonhardt, F., and Schelling, G., 1974, Torsionsversuche an Stahlbeton-
Group, Toronto, ON, Canada, 1182 pp. balken, Deutscher Ausschuss für Stahlbeton: Schriftenreih, V. 239, Ernst &
CSA S806-12(R2017), 2017, “Design and Construction of Building Sohn, Berlin, Germany, 122 pp. (in German)
Components with Fibre-Reinforced Polymers,” CSA Group, Toronto, ON, Mitchell, D., and Collins, M. P., 1974, “Diagonal Compression Field
Canada. Theory — A Rational Model for Structural Concrete in Pure Torsion,” ACI
CSA S807:19, 2019, “Specification for Fibre-Reinforced Polymers,” Journal Proceedings, V. 71, No. 8, Aug., pp. 396-408.
CSA Group, Toronto, ON, Canada, 67 pp. Mohamed, H. M., and Benmokrane, B., 2014, “Design and Perfor-
Deifalla, A.; Hamed, M.; Saleh, A.; and Ali, T., 2014, “Exploring GFRP mance of Reinforced Concrete Water Chlorination Tank Totally Rein-
Bars as Reinforcement for Rectangular and L-Shaped Beams Subjected forced with GFRP Bars: Case Study,” Journal of Composites for Construc-
to Significant Torsion: An Experimental Study,” Engineering Structures, tion, ASCE, V. 18, No. 1, Feb., p. 05013001. doi: 10.1061/(ASCE)
V. 59, Feb., pp. 776-786. doi: 10.1016/j.engstruct.2013.11.027 CC.1943-5614.0000429
El-Salakawy, E.; Masmoudi, R.; Benmokrane, B.; Brière, F.; and Mohamed, H. M., and Benmokrane, B., 2015, “Torsion Behavior of
Desgagné, G., 2004, “Pendulum Impacts into Concrete Bridge Barriers Concrete Beams Reinforced with Glass Fiber-Reinforced Polymer Bars and
Reinforced with Glass Fibre Reinforced Polymer Composite Bars,” Cana- Stirrups,” ACI Structural Journal, V. 112, No. 5, Sept.-Oct., pp. 543-552.
dian Journal of Civil Engineering, V. 31, No. 4, Aug., pp. 539-552. doi: doi: 10.14359/51687824
10.1139/l04-006 Mohamed, H. M., and Benmokrane, B., 2016, “Reinforced Concrete
Eladawy, M.; Hassan, M.; and Benmokrane, B., 2019, “Experimental Beams with and without FRP Web Reinforcement under Pure Torsion,”
Study of Interior Glass Fiber-Reinforced Polymer-Reinforced Concrete Journal of Bridge Engineering, ASCE, V. 21, No. 3, Mar., p. 04015070.
Slab-Column Connections under Lateral Cyclic Load,” ACI Structural doi: 10.1061/(ASCE)BE.1943-5592.0000839
Journal, V. 116, No. 6, Nov., pp. 165-180. doi: 10.14359/51716803 Mohamed, H. M.; Ali, A. H.; Hadhood, A.; Mousa, S.; Abdelazim, W.;
Hadhood, A.; Gouda, M. G.; Agamy, M. H.; Mohamed, H. M.; and and Benmokrane, B., 2020, “Testing, Design, and Field Implementation
Sherif, A., 2020, “Torsion in Concrete Beams Reinforced with GFRP of GFRP RC Soft-Eyes for Tunnel Construction,” Tunnelling and Under-
Spirals,” Engineering Structures, V. 206, Mar., Article No. 110174. doi: ground Space Technology, V. 106, Article No. 103626. doi: 10.1016/j.
10.1016/j.engstruct.2020.110174 tust.2020.103626
Hsu, T. T. C., 1968, “Torsion of Structural Concrete—A Summary on Mohamed, H. M.; Chaallal, O.; and Benmokrane, B., 2015, “Torsional
Pure Torsion,” Torsion of Structural Concrete, SP-18, American Concrete Moment Capacity and Failure Mode Mechanisms of Concrete Beams Rein-
Institute, Farmington Hills, MI, pp. 165-178. forced with Carbon FRP Bars and Stirrups,” Journal of Composites for
Hsu, T. T. C., 1973, “Post-Cracking Torsional Rigidity of Rein- Construction, ASCE, V. 19, No. 2, Apr., p. 04014049. doi: 10.1061/(ASCE)
forced Concrete Sections,” ACI Journal Proceedings, V. 70, No. 5, May, CC.1943-5614.0000515
pp. 352-360. Mousa, S.; Mohamed, H. M.; and Benmokrane, B., 2018, “Flexural
Hsu, T. T. C., 1984, Torsion of Reinforced Concrete, Van Nostrand Rein- Strength and Design Analysis of Circular Reinforced Concrete Members
hold, New York, NY. with Glass Fiber-Reinforced Polymer Bars and Spirals,” ACI Structural
Hsu, T. T. C., and Mo, Y. L., 1985, “Softening of Concrete in Torsional Journal, V. 115, No. 5, Sept., pp. 1353-1364. doi: 10.14359/51702282
Members—Theory and Tests,” ACI Journal Proceedings, V. 82, No. 3, Park, S.-K.; Ko, W.-J.; and Kim, H.-Y., 2001, “Estimation of Torsional
May-June, pp. 290-303. doi: 10.14359/10335 Crack Width for Concrete Structural Members,” Magazine of Concrete
Jeng, C.-H.; Chiu, H.-J.; and Peng, S.-F., 2013, “Design Formulas for Research, V. 53, No. 5, Oct., pp. 337-345. doi: 10.1680/macr.2001.53.5.337
Cracking Torque and Twist in Hollow Reinforced Concrete Members,” Pultrall Inc., 2019, Composite Reinforcing Rods Technical Data Sheet,
ACI Structural Journal, V. 110, No. 3, May-June, pp. 457-467. doi: Thetford Mines, QC, Canada.
10.14359/51685603 Rahal, K. N., and Collins, M. P., 1995, “Analysis of Sections Subjected
JSCE, 1997, “Recommendation for Design and Construction of Concrete to Combined Shear and Torsion—A Theoretical Model,” ACI Structural
Structures Using Continuous Fiber Reinforcing Materials,” Concrete Engi- Journal, V. 92, No. 4, July-Aug., pp. 459-469. doi: 10.14359/995
neering Series 23, A. Machida, ed., Japan Society of Civil Engineers, Rausch, E., 1953, Drillung, Schub und Scheren im Stahlbetonbau
Tokyo, Japan, 325 pp. (Torsion, Diagonal Tension, and Shear in Reinforced Concrete), Deutscher
Khagehhosseini, A. H.; Porhosseini, R.; Morshed, R.; and Eslami, A., Ingenieur-Verlag GmbH, Düsseldorf, Germany, 168 pp. (in German)
2013, “An Experimental and Numerical Investigation on the Effect of Shehab, H. K.; El-Awady, M. E.; and Husain, M., Sayed Mandour, 2009,
Longitudinal Reinforcements in Torsional Resistance of RC Beams,” “Behavior of Concrete Beams Reinforced by FRP Bars under Torsion,”
Structural Engineering and Mechanics, V. 47, No. 2, July, pp. 247-263. doi: Proceedings of the Thirteenth International Conference on Structural and
10.12989/sem.2013.47.2.247 Geotechnical Engineering (13th ICSGE), Cairo, Egypt, pp. 931-942.
Koutchoukali, N.-E., and Belarbi, A., 2001, “Torsion of High-Strength Zhou, J.; Shen, W.; and Wang, S., 2017, “Experimental Study on
Reinforced Concrete Beams and Minimum Reinforcement Require- Torsional Behavior of FRC and ECC Beams Reinforced with GFRP
ment,” ACI Structural Journal, V. 98, No. 4, July-Aug., pp. 462-469. doi: Bars,” Construction and Building Materials, V. 152, Oct., pp. 74-81. doi:
10.14359/10289 10.1016/j.conbuildmat.2017.06.131
@seismicisolation
@seismicisolation

Title No. 120-S83
Experimental Study on Drilled Shaft Reinforcing Bar

Anchorage in Footings
by Yousun Yi, Hyunsu Kim, Jongkwon Choi, Juan Murcia-Delso, and Oguzhan Bayrak
This paper presents an experimental study on the anchorage To ensure full tensile yield capacity of the tie elements
behavior of drilled shaft reinforcement subjected to tension in comprising a strut-and-tie model, sufficient development
drilled shaft footings loaded under combined axial force and length (anchorage length) should be provided beyond the
uniaxial bending moment. Four large-scale tests were conducted point at which the tie meets the strut anchoring it. Current
on drilled shaft footing specimens employing an equivalent loading
STM provisions1,2 establish the critical section at which
condition introducing tension in the drilled shaft reinforcement.
the bar should be developed based on the geometry of the
Three different anchorage details were tested: straight bars,
hooked bars, and headed bars. The drilled shaft reinforcement extended nodal zone bounded by the edge of the strut in a
was capable of developing its full yield strength in tension in all singular node (for example, a node with a well-bounded
the tests, regardless of the anchorage detail. The tensile stresses bearing face). Specifically, the critical section is defined at
in drilled shaft bars were primarily developed in the region of the the point where the centroid of the tie element intersects the
embedment length closest to the interface between the drilled shaft extended nodal zone, as shown in Fig. 2(a). For smeared
and the footing, while negligible stress and slip were measured in nodes, it is not possible to identify the critical section for
the vicinity of the unloaded end of the bars for all three anchorage tension development based on the extended nodal zone
details. Finally, a critical section is also proposed in this study to owing to the absence of a well-defined nodal geometry. An
perform the anchorage check for the drilled shaft reinforcement alternative method was proposed by Yi et al.3 to determine
in drilled shaft footings designed with the strut-and-tie method.
the critical section of the column ties in Fig. 2(b), corre-
The definition of the critical section provides a safe estimate of the
sponding to a drilled shaft footing subjected to combined
available development length of the drilled shaft bars according to
the findings of the experimental program. axial force and a moderate uniaxial bending moment. This
method assumes a large compression field bounded by the
Keywords: bar anchorage; bond; development length; drilled shaft footing; idealized diagonal struts near the column tie element of
drilled shaft reinforcement; large-scale tests; strut-and-tie method (STM). the 3-D strut-and-tie model, and the stress field is used to
define the critical section for the column tie element at the
INTRODUCTION smeared node. The proposed compression field is considered
A drilled shaft footing is a component of a pier founda- to perform a similar role to the extended nodal zone in deter-
tion transferring forces from the pier to a group of drilled mining the critical section, as shown in Fig. 2(b). The safety
shafts. Drilled shaft footings are generally considered to of this critical section criterion was verified experimentally.3
behave as D-regions due to their reduced shear span-depth However, the same criterion cannot be used for the case
ratios and are therefore recommended to be designed using represented by the 3-D strut-and-tie model in Fig. 1(b) due
the strut-and-tie method (STM). Three-dimensional (3-D) to its different configuration of struts and ties. Therefore, it
strut-and-tie models are required to represent internal forces remained unclear how to perform the anchorage check for
in drilled shaft footings supported on a grid of drilled shafts. the column and drilled shaft tie elements in the 3-D strut-
The configuration of the 3-D strut-and-tie model depends on and-tie model for drilled shaft footings under combined
the loading combination applied through the pier. Figure 1 axial force and a large uniaxial bending moment.
presents two strut-and-tie models for a drilled shaft footing Most previous experimental studies4-10 on drilled shaft
subjected to uniaxial flexure and vertical compression footings were conducted for the simplest loading condition:
loading. The model in Fig. 1(a) corresponds to a column a pier subjected to uniaxial compression. To the authors’
subjected to combined axial force and a moderate axial knowledge, a limited number of experimental studies3,11
bending moment, resulting in tension at one face of the have been conducted on drilled shaft footings subjected to
column and nonuniform compression in drilled shafts, while uniaxial compression combined with the moderate flexural
the model in Fig. 1(b) corresponds to a case of combined moment, inducing tension at the column reinforcement
axial force and a large uniaxial bending moment, resulting and nonuniform compression in drilled shafts. However,
in tension not only at one face of the column but also at no experimental studies have been reported to investigate
two of four drilled shafts. Both column and drilled shaft tie
elements in the model are connected with smeared nodes, in
which the nodal geometry cannot be clearly established due ACI Structural Journal, V. 120, No. 5, September 2023.
MS No. S-2022-257.R1, doi: 10.14359/51738837, received January 21, 2023, and
to the absence of well-defined geometrical constraints (for reviewed under Institute publication policies. Copyright © 2023, American Concrete
Institute. All rights reserved, including the making of copies unless permission is
example, bearing area or strut width limitation). obtained from the copyright proprietors. Pertinent discussion including author’s
@seismicisolation
@seismicisolation

Fig. 1—Three-dimensional strut-and-tie model for drilled shaft footing on four drilled shafts subjected to axial force and
uniaxial bending moments.
Fig. 2—Critical section for anchorage verification of column tie elements in drilled shaft footings subjected to combined
axial force and moderate uniaxial bending moment.
the response of drilled shaft footings subjected to flexural drilled shaft footings subjected to combined axial force and
moments large enough to induce tension in the drilled shafts. a large uniaxial bending moment, resulting in tension in two
Therefore, due to this limited experimental database, of four drilled shafts. They proposed anchorage details for
studies12-20 on STM-based analytical models for drilled drilled shaft reinforcement under tension based on conser-
shaft footings that correspond only to uniaxial compres- vative assumptions and best construction practices given
sion have been conducted. A few studies21,22 proposed the lack of in-depth experimental or computational studies
STM-based design recommendations and examples for on this problem. The design example provided by Widianto
@seismicisolation
@seismicisolation

Fig. 3—Detailing of column reinforcement anchorage proposed in literature.
and Bayrak21 proposed using headed ends to develop the EXPERIMENTAL PROGRAM
tension capacity of both the drilled shaft reinforcing bars Equivalent loading condition
and column bars extending beyond the mat reinforcement The experimental program presented in this paper is part
on the opposite face of the footing, as shown in Fig. 3(a). of a comprehensive research project on drilled shaft foot-
This was a conservative approach because it assumed that ings23 conducted in the Ferguson Structural Engineering
the bars had to be fully developed at the end point of the Laboratory at The University of Texas at Austin. The exper-
ties, without consideration of an extended nodal zone. On iments presented here are specifically intended to study the
the other hand, 180-degree hooked drilled shaft reinforcing effects of a high bending moment applied at the interface
bars were proposed in a similar design example by Williams of the column, resulting in tensile reactions at two of four
et al.,22 as shown in Fig. 3(b). The hooked bars were recom- drilled shafts. To investigate the response of a footing spec-
mended to extend up to right before the mat reinforcement imen subjected to such a high moment demand, a large load
in the opposite face of the footing, considering successful eccentricity or lateral load would need to be applied to the
long-term construction practice. The development length column. Furthermore, yielding of the column reinforcement
of the drilled shaft reinforcement was not verified against would need to be prevented to investigate the anchorage of
code requirements in any of these design examples, and the drilled shaft reinforcement developing their yield strength,
recommended anchorage details were not experimentally which would require a very large amount of column rein-
verified. forcement. Hence, it is difficult to directly reproduce this
There are currently a number of unknowns and ambigu- boundary condition in large-scale laboratory testing. To
ities related to the development length requirements and overcome the difficulty, this study adopted a simpler, equiv-
detailing of drilled shaft reinforcement subjected to tension alent loading condition for the experimental program by
in drilled shaft footings. Experimental data are needed to directly applying a tensile load to the vertical reinforcement
resolve open questions related to the development of this of two drilled shafts, representing the expected tensile reac-
type of reinforcement in the context of the 3-D STM. The tions under a high moment demand. To further simplify the
present study conducted large-scale structural tests on drilled test setup, the column tie elements shown in Fig. 1(b) were
shaft footing specimens where the drilled shaft reinforce- substituted by post-tensioning forces applied on the top
ment was loaded under tension. Three different anchorage surface of the footing to prevent the failure induced by the
details were investigated: straight bars, 90-degree hooked column tie elements, as illustrated in Fig. 4.
bars, and headed bars. This paper provides a comprehensive The strut-and-tie model obtained from the equivalent
description and discussion of the experimental program and loading is simpler than that resulting from combined axial
test results. Design recommendations for the development force and a large uniaxial bending moment, as depicted in
length calculation of the drilled shaft reinforcement are also the comparison between Fig. 4 and Fig. 1(b). The discrep-
provided and discussed using experimental results. ancies between the models are caused by the existence or
not of a compressive reaction at the other two drilled shafts
RESEARCH SIGNIFICANCE of the footing. Although the horizontal ties and struts placed
The large-scale experimental tests conducted in this study on the plane of the bottom mat reinforcement do not exist in
provide valuable data on the behavior of drilled shaft foot- the equivalent model, they do not affect the anchorage of the
ings governed by yielding of the drilled shaft reinforce- vertical tie element for drilled shaft reinforcement because
ment. The test results characterize the anchorage response these elements are self-equilibrated at the node. Similarly,
of drilled shaft reinforcement with three different anchorage the confinement effect owing to the horizontal struts is also
details. The development length verification recommenda- negligible. However, those two models have identical strut-
tions can contribute to overcoming the lack of guidance and and-tie model configurations near the tip of the drilled shaft
uncertainties related to the use of the 3-D STM in the design ties, which represents the force-transfer mechanism of the
of drilled shaft footings subjected to large flexural moments. drilled shaft ties. Therefore, the equivalent loading condition
was employed for the experimental program of this study to
@seismicisolation
@seismicisolation

Fig. 4—Three-dimensional strut-and-tie model for drilled shaft footing under equivalent loading condition.
Fig. 5—Drilled shaft reinforcement anchorage details installed in specimens.

investigate the anchorage behavior of the drilled shaft rein- Table 1—Test matrix
forcement subjected to tension in drilled shaft footings. Drilled shaft reinforcement
Specimen
Test ID ID Number and size Anchorage detail
Test variables
By employing the equivalent loading condition, the drilled 1 VII–TD-ST Straight
VII-TD
shaft reinforcement behavior in each pair of drilled shafts 2 VII–TD-HD Five No. 6 Headed
can be investigated without influencing the pair of drilled (reinforcement
3 VII–TK-ST ratio: 1.09%*) Straight
shafts on the opposite side. Therefore, two tests can be VII-TK
4 VII–TK-HK 90-degree hooked
conducted per each footing specimen. Two footing spec-
imens were fabricated, and each specimen contained two *
For 16 in. (406 mm) diameter drilled shaft.
different anchorage types of drilled shaft reinforcement. To
compare the behavior of different anchorage types, the first Specimen design
specimen had drilled shaft reinforcement with straight bars The geometry of the footing specimens was determined
and headed anchorages, and the second one had reinforce- from a database comprising 35 drilled shaft footings
ment with straight bars and hooked anchorages, as depicted constructed between 1994 and 2004 for 16 bridge projects in
in Fig. 5. The headed bar has a disk-shaped mechanical the state of Texas. 23
The test specimens were approximately
anchor at the end of the reinforcing bar, which has an one-third scale of the average footing within the database.
effective bearing area of four times the nominal area of The footing specimens were 132 in. (3353 mm) long, 96 in.
the reinforcing bar used. The hooked bars employed stan- (2438 mm) wide, and 40 in. (1016 mm) deep. The drilled
dard 90-degree hooks employing permissible bend radii in shaft reinforcement embedded in the footing corresponded
accordance with AASHTO LRFD Bridge Design Specifi- to four 16 in. (406 mm) diameter drilled shafts. To repli-
cations (AASHTO LRFD) and ACI 318-19. As illustrated
1 2 cate the reinforcement details of the in-service drilled shaft
in Fig. 5(c), the hooked drilled shaft reinforcing bars were footing, reinforcing bars were placed at all faces of foot-
oriented radially toward the outside of the reinforcement ings (that is, bottom mat, top mat, and side reinforcement).
cage, which is the typical detail employed for in-practice The footing specimens were constructed and tested upside
footings. The test matrix is given in Table 1. down
@seismicisolation to allow the direct application of an upward (tension)
@seismicisolation

Fig. 6—Geometry of test specimens.
force to the drilled shaft reinforcement and to develop stable the footing. Therefore, the test specimens in this study were
support capable of resisting the large overturning moment designed with side reinforcement ratios larger than 0.30%
due to the equivalent loading condition. For clarity, the rein- in both the transverse (0.31%) and longitudinal (0.37%)
forcing bars comprising the specimen will be conventionally directions. The minimum reinforcement ratio for shrinkage
referred to with respect to their locations in an actual drilled and temperature reinforcement (>0.18%) was also provided
shaft footing (for example, top mat for the longitudinal rein- to the bottom mat reinforcement of the footing in accor-
forcement on the drilled shafts side and bottom mat for the dance with AASHTO LRFD.1 Additionally, large-diameter
longitudinal reinforcement on the column side). Figure 6 threaded rods (1.625 in. [40 mm] diameter) were inserted
illustrates the geometry of the test specimens. through polyvinyl chloride (PVC) pipes embedded in the
The amount of the top mat reinforcement within the footing to apply the post-tensioning forces that replace the
footing and the column reinforcement within the footing vertical column tie elements for the strut-and-tie model of
specimens was determined to prevent its yielding during the equivalent loading. Those rods were post-tensioned and
testing, based on the predicted tie forces in the 3-D strut- bolted to a supporting frame to prevent a slack between the
and-tie model at ultimate state governed by fracture of the specimen and the frame during the testing. To avoid any
drilled shaft reinforcement. The nodal positions in the strut- damage related to the bursting and bearing actions intro-
and-tie model and the effective width of the ties for the duced by post-tensioning forces, anti-bursting and spiral
top mat reinforcement were determined based on criteria reinforcement were provided at the midheight of the footing
proposed by Williams et al.22 Specifically, the CCC nodes and bearing faces, respectively (refer to Fig. 7).
(that is, the nodes subjected to triaxial compression) beneath The size and embedment length of drilled shaft rein-
the bearing pad for post-tensioning force are positioned forcement was determined based on the compiled data-
at a depth of 10% of the footing height, and only the top base of drilled shaft footings constructed in the state of
mat reinforcement placed within the drilled shaft diameter Texas to design the drilled shaft reinforcement with the
is considered to resist the horizontal tie forces. To preclude size proportioned accordingly to the specimen size. The
any premature failures in the footing, side-face reinforce- drilled shaft reinforcement of some footings in the data-
ment was provided orthogonally on the side surfaces of base was extended up to the level of the top mat reinforce-
the test specimens in an amount exceeding the minimum ment—the same detail as that proposed by Williams et al.22
crack control reinforcement ratio (0.30%) of the STM spec- Nevertheless, most of the drilled shaft reinforcement in the
ifications in AASHTO LRFD.1 Even though crack control database usually did not extend to the top mat reinforce-
reinforcement would be required in the same plane as the ment, and the provided embedment lengths are presumed
struts to control the width of the splitting cracks induced to satisfy the anchorage requirement in accordance with the
by these struts, a previous experimental study10 on drilled sectional design approach. The collected footings might not
shaft footings under uniform compression showed that side- have been designed for load cases inducing tension in the
face reinforcement effectively controls the surface crack and drilled shafts or lacked the STM-based design recommenda-
ensures the triaxial confinement at the internal struts within tions regarding drilled shaft reinforcement anchorage at the
@seismicisolation
@seismicisolation

Fig. 7—Reinforcement details of test specimens.
Table 2—Footing reinforcement details
Bottom mat reinforcement Side-face reinforcement Top mat reinforcement
N-S direction W-E direction Longitudinal direction Transverse direction N-S direction W-E direction
12-2xNo. 9 @ 6.00 No. 6 @ 5.50 in. 12-No. 7 @ 6.00 and
21-No. 9 @ 5.50 in. No. 5 @ 5.00 in. 21-No. 5 @ 5.50 in.
and 7.50 in. No. 8 @ 6.00 and 7.50 in. 7.50 in.
Note: N-S is north-south; W-E is west-east; 1 in. = 25.4 mm.
time of design. Therefore, this study defined the available drilled shaft of the specimens. The reinforcement details are
embedment length (lp,a), an extended embedment length of summarized in Table 2 and illustrated in Fig. 7.
drilled shaft reinforcement that extends up to the elevation
of the top mat reinforcement, to determine an adequate size Material properties
for the drilled shaft reinforcement. Given the geometry of Two batches of concrete were used to fabricate each
the footing specimens, the size of the drilled shaft bars was footing specimen. The mixture proportions were identical for
determined to provide a realistic ratio between the available all batches of concrete with a design compressive strength
embedment length (lp,a) and bar diameter (db,s) according to of 3.6 ksi (24.8 MPa). ASTM A706/A706M-1624 Grade 60
the footing database. The number of bars was then deter- reinforcing bars were used for the drilled shaft reinforce-
mined to have a similar drilled shaft reinforcement ratio ment. ASTM A706/A706M24 bars were chosen to be able to
(ρs) as that observed in the footing database. To reproduce weld them to a steel plate needed to apply the tensile load,
a similar available embedment length ratio (lp,a/db,s = 43.1) as will be discussed later. The rest of the reinforcement in
and drilled shaft reinforcement ratio (ρs = 1.05%) averaged the test specimens were ASTM A615/A615M-2025 Grade 60
from the footing database, five No. 6 reinforcing bars (ρs = reinforcing bars. The mechanical properties of the concrete
1.09%) were provided and extended up to the top mat rein- and reinforcement were determined from the average results
forcement (lp,a/db,s = 46.3) for each 16 in. (406 mm) diameter of materials testing on a minimum of three samples of each
@seismicisolation
@seismicisolation

Table 3—Mechanical properties of materials
Test ID VII-TK-ST VII-TK-HK VII-TD-ST VII-TD-HD
Anchorage detail of drilled shaft reinforcement Straight Hooked Straight Headed
Compressive strength (fc′),
Concrete 5.18 5.44 4.44 4.56
ksi (day of test)
Yield strength (fy,b), ksi 62.9
Bottom mat
Tensile strength (fu,b), ksi 107.9
Reinforcement
Yield strength (fy,s), ksi 68.2
Drilled shaft
Tensile strength (fu,s), ksi 102.8
Note: 1 ksi = 6.89 MPa.
Fig. 8—Test setup for planned equivalent loading conditions.

batch. Furthermore, the compressive concrete strengths of applied in the normal direction with respect to the plane of
the two batches used in a single test specimen were averaged the footing, and the proper load-transfer mechanisms from
to represent the compressive concrete strength of the footing. the drilled shaft footing to the strong floor. The test setup
The materials testing for concrete and reinforcement was configuration is illustrated in Fig. 8.
conducted in accordance with ASTM C39/C39M-2126 and The eccentrically applied upward force (48 in. [1.2 m]
ASTM A370-21,27 respectively. The average material prop- eccentricity with respect to the centroid of the specimen)
erties are summarized in Table 3. results in a large overturning moment at the center reac-
tion region on the bottom surface of the specimen, which
Test setup represents the interface between the column and the footing.
The equivalent loading condition consisted of applying A support frame was designed to redistribute the moment to
an upward (tensile) force to the drilled shaft reinforcement the strong floor with tensile and compressive reactions. The
extending from the footing. A test setup was meticulously supporting frame consists of a large pedestal supporting the
designed and fabricated to ensure the tensile load was specimen, and staggered box beams are placed under it to
@seismicisolation
@seismicisolation

Fig. 9—Connection detail between drilled shaft reinforcement and drilled shaft plate.
transfer the moment to six support pedestals on the strong To further examine the anchorage behavior of the drilled
floor. shaft reinforcement, bar slip was measured at the top and
Each support pedestal was connected to the box beam with bottom ends of the drilled shaft reinforcement. On the top
four threaded rods, and the rods were post-tensioned with surface of the footing, linear potentiometers were installed
50 kip (0.2 MN) to compensate for tensile reactions caused near the center of the four side surfaces of the drilled shaft
by the overturning moment. The test specimen was anchored plates. While fabricating the specimens, small PVC pipes
to the support frame with two rows of five large-diameter were installed right under the tips of two out of the five drilled
(1.625 in. [40 mm] diameter) post-tensioning threaded rods shaft bars for each drilled shaft. The PVC pipes ensured the
passing through the PVC pipes embedded in the footing. creation of deep and small voids in the concrete. Linear
For each test, the rods in a row located at the axis of the potentiometers were installed through the holes to measure
vertical tie elements on the column side were post-tensioned the reinforcement slip at the bottom end of the drilled shaft
with 150 kip (0.7 MN) per rod to prevent a slack between reinforcement during testing. Whereas the straight and the
the specimen and the supporting frame before yielding the headed anchorages have a flat surface at the bottom tip, the
drilled shaft reinforcement. hooked anchorage does not due to its bend radius. Therefore,
Figure 9 illustrates the detail of the connection between a small steel rod was welded at the bend radius of the hooked
the drilled shaft reinforcement extending out from the anchorage to make a flat surface at its tip to monitor the slip
footing and the steel plates used to apply the vertical load. of the hooked reinforcement.
The drilled shaft bars were welded to these plates inside Test specimens were loaded under displacement control
pockets 0.75 in. (19 mm) deep and 2.5 in. (64 mm) in diam- using the actuators at a rate of 0.025 in. (0.6 mm) per minute.
eter. The two drilled shaft plates were connected to a cross- During testing, loading was stopped at 50 kip (0.2 MN)
head box beam through squat steel pedestals. Two 330 kip increments, up to the load when at least one drilled shaft
(1.5 MN) capacity servo-controlled actuators were placed at reinforcement yielded, to inspect and document the condi-
both sides of the specimen under the crosshead box beam to tion of the specimens. After all drilled shaft reinforcements
apply a tensile force to the drilled shaft reinforcement. Each exceeded their yield strain, the specimens were continuously
actuator had a swivel head at both ends, permitting rotation loaded under displacement control until approximately 90%
and translation in the loading point of the specimen. of the expected ultimate load corresponding to the full
tensile strength of the drilled shaft bars (400 kip [1.8 MN])
Instrumentation and loading protocol was attained. The tests were stopped at 400 kip (1.8 MN)
The applied load was monitored in two ways: by means of to prevent fracture of the drilled shaft reinforcement, which
load cells embedded in the actuators, and load cells installed can potentially cause impact damage on the test setup or
in the support pedestals. The support frame was placed on actuators.
top of four corner pedestals and two center-located pedes-
tals, as shown in Fig. 8. A total of three 500 kip (2.2 MN) EXPERIMENTAL RESULTS
capacity load cells were provided at the base of each corner Overall response
pedestal, and one 500 kip (2.2 MN) capacity load cell was The overall response of the specimens is compared in
provided at the base of each center-located pedestal. Table 4 in terms of the loads at yielding of drilled shaft
To investigate the strain development along the embed- bars and maximum bar stresses. The applied loading could
ment length of the drilled shaft reinforcement during testing, not be evenly distributed among all drilled shaft reinforce-
the drilled shaft bars were instrumented with strain gauges. ment during the tests due to the deflection of the specimen.
The measured bar strains were used to develop tensile stress For example, the maximum stress difference between bars
profiles to compare the behavior of different anchorage corresponding to the same drilled shaft were between
details. The strain gauges were installed on longitudinal ribs 12 and 15 ksi (83 and 103 MPa) (approximately 15%) at
of the drilled shaft bars to minimize an adverse effect on the a load of 400 kip (1.8 MN). Furthermore, the load corre-
bond properties. In addition, the bottom mat reinforcement sponding to the first yielding of the bars in each test presents
was also instrumented with strain gauges to monitor their a wide range of values (209 to 290 kip [0.9 to 1.3 MN]).
strains during the tests. However, there was no significant difference between the
@seismicisolation
@seismicisolation

Table 4—Summary of test results
Test ID VII-TK-ST VII-TK-HK VII-TD-HD VII-TD-ST

Anchorage detail of drilled shaft reinforcement Straight Hooked Headed Straight
Measured load at the first yielding of a drilled shaft bar
*
209 (E04) 249 (W10) 290 (E02) 277 (E06)
Pfy,s, kip (reinforcing bar position†)
Measured load corresponding to yielding of all drilled
344 329 374 342
shaft bars* Py,s, kip
Maximum bar stress at 400 kip loading* fs,max, ksi
94.8 (E04) 90.5 (W07) 92.4 (W10) 90.3 (E06)
(reinforcing bar position†)
Py,s/PSTM 1.15 1.10 1.25 1.14
Strain data analyzed to find stresses of reinforcement.

*
Refer to figure in table.

†
Note: 1 kip = 4.4 kN; 1 ksi = 6.9 MPa.
loads corresponding to the yielding of all the drilled shaft shaft plate on the specimen measured unstable data after
reinforcing bars. As shown in Table 4, these load values are drilled shaft reinforcement yielding due to the conical cracks
within 10% of the average value of 347 kip (1.5 MN) of the formed around the drilled shaft plate. Therefore, the west-
four tests. The maximum stresses of the drilled shaft rein- side plate was selected for VII-TK-ST (straight drilled shaft
forcement were computed based on strain gauge data at the reinforcement), and the east-side plate was selected for the
maximum applied load of 400 kip (1.8 MN). For all spec- other specimens to represent the behavior of the drilled shaft
imens, the maximum stresses were commonly developed reinforcement embedded in the footing. The load in Fig. 10
at the measurement location closest to the top surface of corresponds to the load measured by the load cell in the same
the footing (2 in. [51 mm] below the top surface), and they side’s actuator. Figure 10 also presents the results in terms
exceeded 90% of the tensile strength of the bars in all cases. of the average stress level on the drilled shaft reinforcement
The theoretical capacity of the specimens according to the by dividing the applied load by the total area of reinforce-
3-D STM would correspond to the yielding of the drilled ment connected to one drilled shaft plate. As shown, all four
shaft reinforcement at a load of 300 kip (1.3 MN). The differ- tests presented a consistent bar stress-elongation response.
ence between the theoretical yield load and the experimental The average drilled shaft reinforcement stress exceeded the
values of the load for first yielding and complete yielding of yield stress at a load of approximately 300 kip (1.3 MN) and
the bars, indicated in Table 4, can be explained by the fact reached a tensile stress of approximately 90 ksi (621 MPa)
that the distribution of tensile forces among the bars was not at a load of 400 kip (1.8 MN).
perfectly uniform, as will be discussed later based on the The reinforcement slip measured at the bottom tip of the
analysis of bar stress data. drilled shaft reinforcement showed a trend to increase as
The overall test response is also analyzed in terms of the load increased; however, the values were very small (of the
load applied to one drilled shaft versus the average elon- order of 10–5 in. [10–4 mm]) for all anchorage types. Hence,
gation of its drilled shaft bars in Fig. 10. To this end, the the slip was negligible.
measurements obtained from the linear potentiometers
attached on the drilled shaft plates were averaged to calcu- Visual observations
late the average displacement of the plate with respect to During each test, all faces of the tested span of the footing
the footing, which represents the average elongation (or slip) were visually inspected. During testing of VII-TK-ST, a
of the drilled shaft reinforcement at the top surface of the series of horizontal cracks formed at midheight of the north
footing. Some linear potentiometers installed at one drilled face of the specimen as the load increased. Those cracks are
@seismicisolation
@seismicisolation

stress-strain relationships obtained from tension tests on
reinforcing bars were used to convert measured strains to
stresses. The drilled shaft reinforcing bars in a drilled shaft
plate were subjected to different loads because the position
of the plate was not perfectly perpendicular to the direction
of the applied load during testing owing to the deformation
of the specimen. However, the stress profiles of all drilled
shaft reinforcing bars showed a similar tendency regard-
less of their position and anchorage type. The drilled shaft
reinforcing bar in the east-side drilled shaft reinforcement
group positioned closest to the central axis of the footing
(E06 for VII-TK-ST and VII-TK-HK; E12 for VII-TD-ST
and VII-TD-HD) was selected to investigate the behavior
of the drilled shaft reinforcement and establish a compar-
ison between the specimens, as this was the only bar in all
Fig. 10—Load versus bar elongation. specimens with all strain gauges functional until the end of
testing. Figure 13 presents the stress profiles corresponding
presumed to be microcracks that already existed between
to the different load levels ranging from 50 to 400 kip (0.4
two concrete batches comprising the specimen. Most of
to 1.8 MN).
the cracks were first observed at the load levels of 100 and
All tested drilled shaft reinforcement anchorages were
150 kip (0.4 and 0.7 MN), and no additional crack propa-
able to develop the yield stress within approximately 16 in.
gation was observed afterward. Furthermore, those cracks
(406 mm) of their embedment length measured from the
were not observed in the other three tests. Otherwise, all
topmost strain gauge, installed at 2 in. (50 mm) below the
specimens showed a similar crack pattern. Only a few hair-
top surface of the footing. This is inferred from the stress
line cracks occurred on the side surfaces of the footing in all
curves corresponding to the bar yielding (at a load of 300 kip
specimens during the tests.
[1.3 MN]), which practically go to zero at a depth of 16 in.
The post-tensioning forces applied for anchoring the test
(406 mm). Beyond this load level, the stress level near the top
specimen to the support frame increased during the testing
of the footing increased to approximately 90 ksi (621 MPa),
as the load increased. It indicates that the post-tensioning
which is 90% of the tensile strength (99 ksi [683 MPa]) of
rods were elongated during the testing; therefore, the desired
the reinforcement. In contrast, the stresses near the bottom
fixed support condition could not be perfectly provided to
tip of the drilled shaft reinforcement were negligible in all
the test specimen. Because of the actual boundary condi-
cases. This indicates that the headed and hooked drilled shaft
tion, the test specimens showed a rocking behavior during
reinforcement did not activate the bearing action of the head
the testing, which hindered the formation of flexural cracks
and hook to resist the tensile force in the reinforcing bar, and
in the footings. This observation is in line with the bottom
that tension was developed solely by bond stresses on the
mat reinforcement strains measured during the tests. All the
surface of the bar.
bottom mat reinforcing bars experienced a low strain level
To examine the bond behavior of the drilled shaft rein-
(corresponding to a tensile stress less than 1.5 ksi [10 MPa])
forcement, average bond stresses (τb) between two consec-
at the installed strain gauge locations at any time of the tests.
utive gauges were calculated based on equilibrium using
In spite of this rocking motion, all tested drilled shaft
Eq. (1)
reinforcement were successfully loaded beyond their yield
capacity and up to a maximum stress of over 90% of their
τb = ( )
_ Δfs
As _
tensile strength. The test specimens were dissected diago- (1)
db π Δl
nally after the testing to inspect internal cracks, as shown in
Fig. 11. Regardless of the anchorage types, concrete spalling where As is the cross-sectional area of reinforcement to be
around the drilled shaft plates was observed after the tests anchored; db is the diameter of reinforcement to be anchored;
due to the formation of shallow conical failure planes near the Δfs is the change in stress between two consecutive gauges;
loaded end of the bar above the bottom footing mat, caused by and Δl is the center-to-center distance between two consec-
the bond of the drilled shaft reinforcement, whereas almost utive gauges. Figure 14 presents the average bond stress
no cracking or damage was observed from the middepth to profiles of the drilled shaft reinforcement obtained from
the bottom end of the drilled shaft reinforcement. Figure 12 each test. For reference, Fig. 14 also includes the theoret-
presents the inspected crack maps after testing the span of ical value of the local bond strength (τbu,split) corresponding
each drilled shaft reinforcement anchorage type. to a splitting bond failure calculated in accordance with fib
Model Code (MC) 201028 using Eq. (2)
Tensile stress and bond stress profiles of
column reinforcement
τbu,split = η26.5(  25 )  ( ) [( ) (cmin) ] (2)
0.25
fc′
____ _
25
0.2
cmin 0.33 _
_ cmax 0.1
The anchorage response of drilled shaft reinforcement
during the testing was investigated through the data obtained db db
from strain gauges installed at 7 in. (178 mm) spacing. The where η 2 is a reinforcing bar position factor (equal to 1 for
vertical
@seismicisolation
@seismicisolation bars); fc ′ is the concrete strength; and cmin and cmax

Fig. 11—Diagonal cut section of VII-TK-ST.
are parameters that depend on the concrete cover and bar specimens, the impact of such variation on bond resistance
spacing as defined in Eq. (3a) and (3b) is very limited, as evidenced by the 5% maximum difference
in theoretical bond strength, as shown in Fig. 14.
cmin = min{cs/2, cx, c y}
(3a)
DISCUSSION
cmax = max{cs/2, cx} (3b) Effects of anchorage type
The experimental study by Yi et al.3 on drilled shaft
where cs is the clear spacing of the reinforcing bars; cx is footings subjected to combined axial force and a moderate
the clear side cover to the bar in the direction parallel to uniaxial bending moment, represented by the 3-D strut-
the splitting crack; and cy is the clear cover to the bar in and-tie model in Fig. 1(a), showed that an assumed large
the direction perpendicular to the splitting crack. The compression field bounded by the idealized diagonal struts
confinement term for transverse reinforcement of the orig- induced bar stresses in the vicinity of the bottom tip of the
inal fib MC 2010 equation was not included in Eq. (2) by column reinforcement. Hence, different column reinforce-
neglecting the confinement effect achieved from the side- ment force-transfer actions were found depending on the
face reinforcement. anchorage details provided at the ends of the column rein-
The magnitude of the bond stresses increases with the level forcing bars.
of loading, and relatively large bond stresses concentrate In contrast, the overall behavior of the drilled shaft rein-
near the top surface of the specimen, as shown in Fig. 14. forcement observed in all four tests of this study was compa-
The maximum bond stress values at and after bar yielding rable regardless of the anchorage type. The development of
vary between 1.0 and 1.5 ksi (6.9 and 10.3 MPa), and are tensile capacity of the bars was concentrated in the upper
developed in the upper 7 in. (178 mm) segment in most cases half of the embedment length owing to the bond forces in this
or in the segment right below it. As shown, these maximum region (Fig. 13 and 14), and the bearing action of the heads
values are smaller than the theoretical local bond resistance or hooks was not activated. Almost no tensile reinforcing
τbu,split, but this is reasonable because the local peak of the bar stresses were developed near the end of drilled shaft
actual bond stress distribution is expected to be larger than reinforcement, even though the strut-and-tie model in Fig. 6
the average value obtained from a 7 in. (178 mm) segment. considers a diagonal strut flowing to the end of the vertical
The bond stress diagrams also confirm that bond stresses drilled shaft tie element. A possible explanation is that the
are negligible at depths below 16 in. (406 mm) at yielding compression field stemming from the post-tensioning force
(load of 300 kip [1.3 MN]) and below 23 in. (584 mm) at the was combined with a tensile stress field in the concrete in
maximum load (400 kip [1.8 MN]). All bond stress distribu- the upper half of the embedment length. This observation
tions were very similar regardless of the anchorage details is in line with observations from the inspected cut section
and concrete strength of the specimens. While the values (Fig. 11), which does not reveal cracking of concrete along
of fc′ presented a maximum difference of 20% between the embedment length of the bar, except for the shallow cone
failure at the very top.
@seismicisolation
@seismicisolation

Fig. 12—Crack map of test specimens after testing.
Critical section for anchorage length verification for the purpose of developing the tensile forces of the tie.
Although the diagonal strut in Fig. 6 did not seem to influ- The point at which the drilled shaft reinforcing bar intersects
ence reinforcing bar stress developments in the vicinity of the boundary of the fan-shaped strut would correspond to
the drilled shaft reinforcement during the tests, a critical the critical section at which the bar starts to be developed,
section of the drilled shaft tie element needs to be estab- between 12.1 and 17.2 in. (307 and 437 mm) from the top
lished to ensure the development of its full yield strength surface depending on the bar. Figure 15 also indicates the
consistent with the existing stress flows in the context of the section at which the yield strength of the bar was actually
3-D STM-based design procedure. Therefore, the internal developed at the end of the test, which lies somewhere
force flow of the test specimens was reviewed to propose the between 2 and 9 in. (51 and 229 mm) from the top surface.
critical section beyond which the drilled shaft reinforcement This indicates that the assumed critical section for the devel-
needs to be developed. opment of drilled shaft reinforcement is conservative with
The diagonal strut acting at the end of the drilled shaft respect to the test results. To simplify, the critical section can
reinforcement embedded is classified as a fan-shaped strut be assumed to be at the point where the theoretical compres-
because this end corresponds to a smeared node, as shown sion field meets the vertical tie of the drilled shaft regardless
in Fig. 15. Therefore, the strut boundary spreads out from of the bar position in the drilled shaft.
the edge of the bearing pad placed on the test specimen for The available development lengths measured from the
applying the post-tensioning force. The minimum strut angle proposed critical section at all drilled shaft reinforcing bar
specified in AASHTO LRFD,1 25 degrees, is employed for positions in the footing are compared with the minimum
defining the upper boundary of the fan-shaped strut. The development lengths required for the different anchorage
resulting compression field, shaded in blue in Fig. 15, is types in accordance with both AASHTO LRFD1 and ACI
318.2 The available development length is measured from
assumed to perform the same role as an extended nodal@seismicisolation
zone
@seismicisolation

Fig. 13—Stress profiles of drilled shaft reinforcement.
the end of the drilled shaft reinforcement to the proposed to represent the effect of column reinforcement using post-
critical section. The comparison results are summarized in tensioning forces. In an actual footing subjected to combined
Table 5. As shown, all drilled shaft reinforcing bars in the axial force and a large uniaxial bending moment, an analo-
specimens satisfy the anchorage requirement, except for the gous critical section can be defined based on the compression
straight bars. Even though the available length measured for field that forms between the drilled shaft reinforcement and
the straight bars is shorter than the minimum development the column reinforcement, as shown in Fig. 16. The assumed
lengths computed with the provisions, the straight column compression field (shaded in blue in Fig. 16) represents a
reinforcement could develop its yield strength during the noncontact lap splice transfer mechanism between the column
testing. This confirms the conservativeness of the proposed and drilled shaft reinforcement. Similar to the compression
critical section. field conservatively defined for the equivalent loading condi-
The significant level of conservatism of the critical section tion, a minimum strut angle of 25 degrees is also employed
estimation for these tests could be explained in part by the for the compression stress field enabling the noncontact lap
capacity of concrete to carry tensile stresses in the anchorage splice, as depicted in Fig. 16. Therefore, the critical section of
zone of the bars, as mentioned earlier. This type of behavior the drilled shaft reinforcement can be defined at the position
is not guaranteed in general as larger amounts of drilled where the drilled shaft tie element intersects the boundary of
shaft reinforcement could result in cracking of the anchorage the assumed compression field. Likewise, the critical section
zone, leading to more widely distributed bond stresses in of the column reinforcement under this loading condition is
agreement with the proposed inclined compression field. defined as the intersection of the column tie and the boundary
Regardless, the proposed definition of the critical section of the compression field.
is consistent with the general principles of the STM and
current anchorage checks for ties based on extended nodal CONCLUSIONS
zones, and as such is intended to provide safe solutions. This experimental study investigated the anchorage
The critical section proposed in Fig. 15 is specific to test response of drilled shaft reinforcement subjected to tension
specimens where an equivalent loading condition was used in drilled shaft footings subjected to combined axial force
@seismicisolation
@seismicisolation

Fig. 14—Bond stress profiles of drilled shaft reinforcement.
and a large uniaxial bending moment by means of an equiv- bearing action of the hook or head was not able to be acti-
alent boundary condition. Two drilled shaft footing speci- vated because no splitting crack was caused by the diagonal
mens containing various types of drilled shaft reinforcement fan-shaped strut flowing down to the end of the drilled shaft
(straight, headed, and 90-degree hooked) were fabricated, tie element, as evidenced by post-testing section cuts. There-
and the anchorage behavior of the drilled shaft reinforce- fore, the development of these bars was solely provided by
ment embedded in each half span of the footing was exam- bond stresses along the top and central portion of the embed-
ined in four independent tests. All the bar types were loaded ment lengths.
in tension beyond their yield capacity. The main conclusions 4. A compression field formed by the boundary of the
and findings of the study are the following: fan-shaped strut was used for theoretical anchorage verifi-
1. All drilled shaft reinforcing bars were able to develop cations using a three-dimensional (3-D) strut-and-tie model
their full yield strength and sustained large inelastic defor- of the test specimens. The proposed compression field is
mations regardless of the anchorage type. In addition, the regarded as an extended nodal zone to determine the critical
load-bar elongation responses of the four test specimens section of the drilled shaft tie element. This simple criterion
were practically identical. provides conservative results about the available length to
2. The experimentally derived stress profiles of bars with develop the bars in tension. For actual drilled shaft foot-
different anchorage types of bars were comparable to each ings subjected to combined axial force and a large uniaxial
other. The tensile stress increment of the drilled shaft rein- bending moment, similar critical section definitions for the
forcement was observed in the upper half of their embed- column and drilled shaft reinforcement are proposed based
ment length (near their loaded end). The largest bond on a compression field representing the noncontact lap splice
stresses typically occurred near the top surface of the spec- behavior between the column and drilled shaft tie elements.
imen, which represented the interface between the footing Although the proposed critical section of the drilled shaft
and drilled shaft. reinforcement provided conservative estimations of the
3. The slip and tensile bar stresses at the unloaded end available development length in the test specimens, the vali-
of the drilled shaft reinforcement were negligible.@seismicisolation
The dation of this approach is limited to four tests studying a
@seismicisolation

Fig. 15—Assumed compression field and proposed critical section of drilled shaft reinforcement in test specimens.
Fig. 16—Proposed critical sections for column and drilled shaft reinforcement in drilled shaft footings under combined axial
force and large uniaxial bending moment.
Table 5—Anchorage check based on proposed critical section

Test ID
VII-TK-ST VII-TK-HK VII-TD-ST VII-TD-HD
Anchorage types Straight Hooked Straight Headed
Available length (la), in. 14.9
AASHTO Minimum development length (ld), in. 17.1 10.5 18.4

N/A*
LRFD (2020) la/ld 0.87 1.42 0.81
Minimum development length (ld), in. 21.6 11.1 23.3 7.9
ACI 318-19 (2019)
la/ld 0.69 1.62 0.64 1.88
*
Minimum development length for headed bars is not specified in AASHTO LRFD.1
Note: 1 in. = 25.4 mm.
@seismicisolation
@seismicisolation

single design parameter (anchorage type) and an idealized 7. Suzuki, K.; Otsuki, K.; and Tsuchiya, T., “Influence of Edge Distance
on Failure Mechanism of Pile Caps,” Transactions of the Japan Concrete
loading scheme. Therefore, further studies are recommended Institute, V. 22, 2000, pp. 361-367.
to study the response of noncontact lap splices between the 8. Suzuki, K., and Otsuki, K., “Experimental Study on Corner Shear
column and drilled shaft reinforcement, including the effects Failure of Pile Caps,” Transactions of the Japan Concrete Institute, V. 23,
2002, pp. 303-310.
of other design parameters (for example, footing geometry 9. Clarke, J. L., “Behaviour and Design of Pile Caps with Four Piles,”
or drilled shaft reinforcement size and area). Technical Report No. 42.489, Cement and Concrete Association, Wexham
Springs, UK, 1973, 19 pp.
10. Kim, H.; Boehm, R. A.; Yi, Y.; Mühlberg, S.; Webb, Z. D.; Choi, J.;
AUTHOR BIOS Murcia-Delso, J.; Hrynyk, T. D.; and Bayrak, O., “Effects of Reinforcement
Yousun Yi is a Postdoctoral Fellow in the Ferguson Structural Engi-
Details on Behavior of Drilled Shaft Footings,” ACI Structural Journal,
neering Laboratory at The University of Texas at Austin, Austin, TX, where
V. 120, No. 1, Jan. 2023, pp. 285-301.
he received his PhD in civil engineering in 2022. His research interests
11. Miguel-Tortola, L.; Miguel, P. F.; and Pallarés, L., “Strength of Pile
include the strut-and-tie method and nonlinear analysis of reinforced
Caps under Eccentric Loads: Experimental Study and Review of Code
Provisions,” Engineering Structures, V. 182, 2019, pp. 251-267. doi:
10.1016/j.engstruct.2018.12.064
Hyunsu Kim is a Structural Specialist at COWI North America, Inc.
12. Dey, S., and Karthik, M. M., “Modelling Four-Pile Cap Behaviour
He received his BS and MS from Seoul National University, Seoul, South
Using Three-Dimensional Compatibility Strut-and-Tie Method,” Engi-
Korea, in 2008 and 2010, respectively, and his PhD from The University
neering Structures, V. 198, 2019, Article No. 109499. doi: 10.1016/j.
of Texas at Austin in 2022. His research interests include the refinement of
engstruct.2019.109499
the strut-and-tie method for drilled shaft footings and numerical analysis of
13. Mogili, S., and Hwang, S.-J., “Softened Strut-and-Tie Model for
Shear and Flexural Strengths of Reinforced Concrete Pile Caps,” Journal
of Structural Engineering, ASCE, V. 147, No. 11, 2021, p. 04021169. doi:
ACI member Jongkwon Choi is an Assistant Professor in the Department
10.1061/(ASCE)ST.1943-541X.0003141
of Civil and Environmental Engineering at Hongik University, Seoul, South
14. Meléndez, C.; Sagaseta, J.; Sosa, P. F. M.; and Rubio, L. P., “Refined
Korea. He received his BS and MS from Seoul National University and his
Three-Dimensional Strut-and-Tie Model for Analysis and Design of Four-
PhD from The University of Texas at Austin. His research interests include
Pile Caps,” ACI Structural Journal, V. 116, No. 4, July 2019, pp. 15-29. doi:
the mechanical behavior, experiment, and analysis of reinforced concrete
10.14359/51714485
and prestressed concrete structures, and the structural assessment of aging
15. Broms, C. E., “Strut-and-Tie Model for Punching Failure of Column
Footings and Pile Caps,” ACI Structural Journal, V. 115, No. 3, May 2018,
pp. 689-698. doi: 10.14359/51702042
ACI member Juan Murcia-Delso is an Assistant Professor at Univer-
16. Chetchotisak, P.; Yindeesuk, S.; and Teerawong, J., “Interactive
sitat Politècnica de Catalunya - BarcelonaTech (UPC), Barcelona, Cata-
Strut-and-Tie-Model for Shear Strength Prediction of RC Pile Caps,”
lonia, Spain. He received his BS and MS from UPC and his PhD from the
Computers and Concrete, V. 20, No. 3, 2017, pp. 329-338.
University of California, San Diego, La Jolla, CA. He is a member of Joint
17. Mathern, A.; Chantelot, G.; Svahn, P.-O.; Kettil, P.; Rempling, R.;
ACI-ASCE Committee 408, Bond and Development of Steel Reinforcement.
and Engström, B., “Enhanced Strut-and-Tie Model for Reinforced Concrete
His research interests include the nonlinear behavior, design, and assess-
Pile Caps,” International Association for Bridge and Structural Engi-
ment of concrete structures.
neering, 39th IABSE Symposium – Engineering the Future, Sept. 21-23,
2017, Vancouver, BC, Canada, pp. 607-614.
Oguzhan Bayrak, FACI, is a University Distinguished Teaching Professor
18. de Araújo, J. M., “Design of Rigid Pile Caps through an Iterative
and Phil M. Ferguson Professor in the Department of Civil, Architectural
Strut-and-Tie Model,” Journal of Advanced Concrete Technology, V. 14,
and Environmental Engineering at The University of Texas at Austin. He
No. 8, 2016, pp. 397-407. doi: 10.3151/jact.14.397
is a member of ACI Committees S803, Faculty Network Coordinating
19. Park, J.; Kuchma, D.; and Souza, R., “Strength Predictions of Pile
Committee, and 341, Performance-Based Seismic Design of Concrete
Caps by a Strut-and-Tie Model Approach,” Canadian Journal of Civil Engi-
Bridges; and Joint ACI-ASCE Committees 441, Reinforced Concrete
neering, V. 35, No. 12, 2008, pp. 1399-1413. doi: 10.1139/L08-062
Columns, and 445, Shear and Torsion.
20. Yun, Y. M.; Chae, H. S.; Kim, B.; and Ramirez, J. A., “Verification
of Three-Dimensional Grid Strut-and-Tie Model Approach in Structural
ACKNOWLEDGMENTS Concrete,” ACI Structural Journal, V. 115, No. 1, Jan. 2018, pp. 27-40. doi:
The authors wish to express their gratitude and sincere appreciation to 10.14359/51700946
the Texas Department of Transportation (TxDOT) for funding this study 21. Widianto, and Bayrak, O., “Example 11: Deep Pile Cap with Tension
through Project 0-6953. The findings and suggestions reported in this paper Piles,” Further Examples for the Design of Structural Concrete with Strut-
are those of the authors and do not necessarily reflect the perspectives of and-Tie Models, SP-273, K.-H. Reineck and L. C. Novak, eds., American
TxDOT. Concrete Institute, Farmington Hills, MI, 2011, pp. 169-190.
22. Williams, C.; Deschenes, D.; and Bayrak, O., “Strut-and-Tie Model
Design Examples for Bridges,” Report No. FHWA/TX-12/5-5253-01-1,
REFERENCES Center for Transportation Research, The University of Texas at Austin,
1. AASHTO, “AASHTO LRFD Bridge Design Specifications,” ninth Austin, TX, 2012, 276 pp.
edition, American Association of State Highway and Transportation Offi- 23. Yi, Y.; Kim, H.; Boehm, R. A.; Webb, Z. D.; Choi, J.; Wang, H.-C.;
cials, Washington, DC, 2020, 1912 pp. Murcia-Delso, J.; Hrynyk, T. D.; and Bayrak, O., “3D Strut-and-Tie
2. ACI Committee 318, “Building Code Requirements for Structural Modeling for Design of Drilled Shaft Footings,” Report No. FHWA/
Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved TX-21/0-6953-R1, Center for Transportation Research, The University of
2022),” American Concrete Institute, Farmington Hills, MI, 2019, 624 pp. Texas at Austin, Austin, TX, 2021, 466 pp.
3. Yi, Y.; Kim, H.; Bohem, R. A.; Webb, Z. D.; Choi, J.; Murcia-Delso, J.; 24. ASTM A706/A706M-16, “Standard Specification for Deformed and
Hrynyk, T. D.; and Bayrak, O., “Experimental Study on Column Rein- Plain Low-Alloy Steel Bars for Concrete Reinforcement,” ASTM Interna-
forcing Bar Anchorage in Drilled Shaft Footings,” ACI Structural Journal, tional, West Conshohocken, PA, 2016, 7 pp.
V. 120, No. 4, July 2023, pp. 191-206. doi: 10.14359/51738721 25. ASTM A615/A615M-20, “Standard Specification for Deformed and
4. Sabnis, G. M., and Gogate, A. B., “Investigation of Thick Slab (Pile Plain Carbon-Steel Bars for Concrete Reinforcement,” ASTM Interna-
Cap) Behavior,” ACI Journal Proceedings, V. 81, No. 1, Jan.-Feb. 1984, tional, West Conshohocken, PA, 2020, 8 pp.
pp. 35-39. 26. ASTM C39/C39M-21, “Standard Test Method for Compressive
5. Suzuki, K.; Otsuki, K.; and Tsubata, T., “Influence of Bar Arrange- Strength of Cylindrical Concrete Specimens,” ASTM International, West
ment on Ultimate Strength of Four-Pile Caps,” Transactions of the Japan Conshohocken, PA, 2021, 8 pp.
Concrete Institute, V. 20, 1998, pp. 195-202. 27. ASTM A370-21, “Standard Test Methods and Definitions for
6. Suzuki, K.; Otsuki, K.; and Tsubata, T., “Experimental Study on Four- Mechanical Testing of Steel Products,” ASTM International, West Consho-
Pile Caps with Taper,” Transactions of the Japan Concrete Institute, V. 21, hocken, PA, 2021, 50 pp.
1999, pp. 327-334. 28. fib, “fib Model Code for Concrete Structures 2010,” International
Federation for Structural Concrete, Lausanne, Switzerland, 2013, 434 pp.
@seismicisolation
@seismicisolation

Title No. 120-S84
Design of Glass Fiber-Reinforced Polymer-Reinforced

Concrete Columns per ACI CODE-440.11-22
by Zahid Hussain and Antonio Nanni
This paper is an attempt at a better understanding of design limitations to non-seismic zones and structures not requiring
provisions of ACI CODE-440.11-22, building code for the design fire resistance. The addition of provisions for compressive
of glass fiber-reinforced polymer (GFRP)-reinforced concrete members is a critical development for practitioners inter-
(RC) columns. Sway and a non-sway column examples origi- ested in nonmetallic reinforcement as it allows designing
nally designed with steel reinforcement were redesigned using
and construction of a building entirely with GFRP-RC.
GFRP longitudinal bars and ties as per provisions of ACI CODE-
Though ACI CODE-440.11-22 permits the design of
440.11-22 to analyze the effect of changing reinforcement type.
Columns were designed with both low-modulus (Ef = 6500 ksi), columns using GFRP bars, due to their lower reliability, the
and high-modulus (Ef = 8700 ksi) GFRP bars. A parametric study minimum compressive strength properties of GFRP bars
was carried out by varying the concrete compressive strength, are not specified in ASTM D7957.12 As stipulated in ACI
the cross-section aspect ratio, and the resultant load eccentricity. CODE-440.11-22, in pure compression, their presence can
GFRP-RC columns require larger cross-section dimensions and be treated as having the same stiffness and strength as those
more reinforcement area than steel-RC columns irrespective of of the surrounding concrete. However, in the presence of
the GFRP elastic modulus when subjected to the same demand. moment, GFRP reinforcement may effectively contribute
The concrete strength has a significant effect on the dimensions to the column capacity. Therefore, this study is carried out
of GFRP-RC columns, and rectangular sections were found to be to show the implications of current Code provisions on the
more efficient than square sections with the same gross concrete
design of GFRP-RC compressive members.
area in the presence of moment. GFRP-RC columns subject to high
eccentricity loads take advantage of GFRP tensile properties and,
thus, are more efficient. RESEARCH SIGNIFICANCE
The recently published ACI CODE-440.11-2211 allows
Keywords: building code; concrete columns; eccentricity; glass fiber-rein- the design of columns with GFRP reinforcement. Due to
forced polymer (GFRP) reinforcement. remaining knowledge gaps in the behavior of GFRP-RC
columns, some Code provisions were only analytically
INTRODUCTION developed and verified by incorporating differences in material
Fiber-reinforced polymer (FRP) bars, being a competitive properties with steel-RC. This study shows the implications of
option for reinforced concrete (RC) members in aggressive Code provisions and highlights the areas for further research.
environments, were not allowed in compression members in
the previous editions of the ACI 440 Guide.1 The primary METHODOLOGY
reason for this exclusion was a lack of information regarding In this study, a column part of a sway frame from the ACI
the behavior of FRP-RC members subjected to compressive Reinforced Concrete Design Handbook,13 a Companion to
loads. However, researchers have been actively investigating ACI 318-19,14 is selected and redesigned using GFRP rein-
the behavior of glass FRP (GFRP)-RC columns during the forcement. This column is part of an interior, continuous,
last decade and have found GFRP-RC columns to be permis- six-bay frame, and built integrally with a 7 in. (178 mm)
sible structural elements. In fact, several experimental deep slab, as shown in Fig. 1(a). The constituent mate-
studies investigated the effect of the compressive behavior rials selected for column design are listed in Table 1.
of longitudinal GFRP bars by testing RC columns2-5 with The concrete strength fc′ is 5000 psi (35 MPa), while the
an overall positive assessment of their feasibility. Jawaheri GFRP type is compliant with material specification based
Zadeh and Nanni6 provided information on flexural stiff- on ASTM D7957.12 For the non-sway case, a column from
ness in frame analysis for GFRP-RC that resulted in close a frame part of an industrial building was taken from the
correspondence to limits proposed by Bischoff.7 Similarly, textbook by Wight and Macgregor,15 as shown in Fig. 1(b).
Hadhood et al., among other researchers, proposed a 1%
8 9
This is a laterally braced column with a beam on one side, as
minimum reinforcement necessary to maintain section shown in Fig. 1(b). The concrete strength fc′ for this column
integrity to achieve a nominal capacity of columns. Khor- was 4000 psi (28 MPa) (as given in the textbook), while
ramian and Sadeghian10 performed structural tests validating
the performance of GFRP-RC columns with reinforcement ACI Structural Journal, V. 120, No. 5, September 2023.
ratios as high as 5.3%. Given, these significant advances in reviewed under Institute publication policies. Copyright © 2023, American Concrete
research over the past decade, the new ACI CODE-440.11-22 Institute. All rights reserved, including the making of copies unless permission is
Building Code permits the use of GFRP-RC columns with
11
@seismicisolation
@seismicisolation

Table 1—Properties of GFRP reinforcement and concrete
Concrete strength, psi
Nominal Elastic Guaranteed tensile Ultimate
Designation diameter, in. Nominal area, in. 2
modulus, ksi strength, ksi strain, % Sway column Non-sway column Clear cover, in.
GFRP-4 0.5 0.2 108 0.016
GFRP-8 1.0 0.79 6500 84.5 0.013
GFRP-9 1.128 1.0 82 0.013
5000 4000 1.5
GFRP-4* 0.5 0.2 139.5 0.016
GFRP-8* 1.0 0.79 8700 120 0.013
GFRP-9 *
1.128 1.0 115 0.013
*
New-generation GFRP bars with higher modulus of elasticity and guaranteed strength as proposed in an ASTM material spec under development.
Note: GFRP-4 = M13; GFRP-8 = M25; GFRP-9 = M29; 1 in. = 25.4 mm; 1 in.2 = 645 mm2; 1 ksi = 6.89 MPa.
Table 2—Strength reduction factor Φ for moment,

axial force, or combined moment and axial force
(ACI CODE-440.11-22, Section 21.2.2)
Net tensile strain at failure
in outermost layer of GFRP
reinforcement εf Classification Φ
εf = εfu Tension-controlled 0.55
εfu > εf > 0.8εfu Transition 1.05 to 0.5εf/εfu
Compression-
εf ≤ 0.8εfu 0.65
controlled
considered. This study uses No. 8 and 9 (M25 and M29)

nominal bar sizes for longitudinal reinforcement and No. 4
(M13) for stirrups/ties in all columns. The mechanical prop-
erties of GFRP bars in tension affecting design are listed in
Table 1 and include guaranteed ultimate tensile strength ffu,
corresponding ultimate strain εfu, and modulus of elasticity
Ef. GFRP compressive properties (that is, strength and stiff-
ness) are not provided because, in design, the area of longi-
tudinal GFRP bars in compression is considered equivalent
to concrete.
COLUMN PROVISIONS IN ACI CODE-440.11-22

For applicable factored load combinations, design strength
at all sections shall satisfy the requirements of ACI CODE-
440.11-22, Section 10.5.1.1, given as follows
ΦSn ≥ U (1)
where Sn is nominal moment, shear, axial, or torsional

strength; U is strength of a member or cross section required
to resist factored loads; and Φ is strength reduction factor as
per ACI 440.11-22 and given in Table 2.
Because GFRP compression reinforcement will not
contribute to the compression capacity of the cross section,
the strength of a column subject to pure axial load is calcu-
Fig. 1—Geometrical dimensions for: (a) sway column; and lated using the gross concrete area and fc′, while treating
(b) non-sway column. GFRP as if it were concrete, as given in the Code Section
22.4.2.2
the GFRP has the same properties as for the sway column.
Given that a new ASTM material specification is under Po = 0.85fc′Ag (2)
development for a class of GFRP bars with higher modulus
of elasticity and strength, this class of GFRP bars was also where Po is nominal axial strength at zero eccentricity.
@seismicisolation
@seismicisolation

The design tensile strain and strength of GFRP bar, in this ACI CODE-440.11-22 provides three conditions to deter-
study, were used as provided in Code Section 10.3.2.1, given mine if the frame can be considered non-sway—namely,
as follows Code Sections 6.2.5, 6.6.4.3(a), and 6.6.4.3(b), listed as
follows:
If Pu > 0.10fc′Ag 1. 6.2.5 states that, if the stiffness of bracing elements
exceeds 12 times the gross lateral stiffness of the columns
Then, the limit on tensile strain is in the direction considered, a column in that story can be
considered as non-sway.
εf = 0.01 2. 6.6.4.3 implies analyzing the columns as non-sway if
condition (a) or (b) is satisfied:
Also, (a) The increase in column end moments due to second-
order effects does not exceed 5% of the first-order end
(0.01Ef)
ffu
moments.
Design strength = m
in (b) The stability index does not exceed 0.05. The stability

index for a given story, Q, shall be calculated as shown
Code Section 10.6.1 specifies a minimum reinforcement
of 1% of the gross concrete area (Ag) to provide resistance to ∑ Pu δo
_

Q = (3)
bending and possibly concrete creep. Similarly, maximum Vus
lc
reinforcement of 8% is specified to avoid congestion of
reinforcing bars and to ensure that concrete can be properly where ∑Pu is total factored vertical load; Vus is horizontal
consolidated. story shear; δo is first-order relative lateral displacement
The minimum number of bars is indicated by the Code between the top and bottom of that story; and lc is height of
Section 10.7.3, given as the column from the center to center of the joints.
GFRP-RC columns are more susceptible to the slender-
Minimum number of longitudinal bars = ness effects than steel-RC due to the lower stiffness of GFRP
⎧4 Rectangular or circular ties reinforcement compared to steel bars; therefore, more strict
limits are imposed when checking slenderness effects for
⎨
⎪
3
  
    Triangular ties
GFRP-RC columns. Slenderness effects can be neglected in
⎩6
⎪
Enclosed by spirals both sway and non-sway frames if the following conditions
of Code Sections 6.2.5.1(a) or 6.2.5.1(b) are satisfied, given
For longitudinal reinforcement, the minimum clear herein as Eq. (4a), (4b), and (4c)
spacing between bars is specified in the Code Section 25.2.3 (a) For columns not braced against sidesway
as follows
k lu
_
1.5 in.(38 mm)
r ≤ 17 (4a)
( 4 / 3dagg
)
Minimum spacing between bars = max 1.5db
(b) For columns braced against sidesway
where db is diameter of the longitudinal bar; and dagg is diam- k lu

_ _ M1

r ≤ 29 + 12 M (4b)
eter of the aggregate. 2
Code Section 25.7.2.3 states that every corner or alternate

bar shall have lateral support by the corner of a tie with an k lu
_

r ≤ 35 (4c)
included angle of not more than 135 degrees. Also, every bar
shall have less than 6 in. (152 mm) clear on each side along where M1/M2 is negative if the column is bent in single
the tie from a laterally supported bar. curvature and positive for double curvature; lu is unsup-
The column size may be found from Eq. (2) by introducing ported length of column; and k is effective length factor for
a strength reduction factor; however, the values obtained in compression members. The effective length factor reflects
this study were significantly lower than those required by column-end restraint conditions, which depend on the rela-
GFRP-RC columns. The stricter limits on slenderness for tive stiffness of the columns to the floor members at the top
GFRP-RC columns will usually require bigger size columns. and bottom of joints given by
Authors, by trial and error, found following relations to give
good approximation for an initial estimate for the size of a
( ) + ( )
_
EI _
EI
square column

l
______________________
c col,above
lc col,below

ω =    
    (5)
( )
l beam,left ( l ) beam,right
_
EI _
EI
Pu
_________ +
Sway columns: Ag =    
0.25 fc′
The values obtained by Eq. (5) are used to calculate k,
Pu
_________
Non-sway columns: Ag =     using Fig. R6.2.5.1a and Fig. R6.2.5.1b for non-sway and
0.15 fc′
sway frames, respectively, shown as Fig. 2, which is then
@seismicisolation
@seismicisolation

Fig. 2—Jackson and Moreland alignment charts (as given in ACI CODE-440.11-22).
used in Eq. 6.2.5.1a (4a), 6.2.5.1b (4b), and 6.2.5.1c (4c) to Table 3—Moment of inertia and cross-sectional
determine if the slenderness of columns could be neglected. area for elastic analysis
r is the radius of gyration. Its value can be calculated as Cross-sectional Cross-sectional
given in Code Section 6.2.5.2 Moment of area for axial area for shear
Member end condition inertia deformations deformations
_

√Ig / Ag Columns 0.4Ig
Uncracked 0.4Ig
Walls 1.0Ag bw x h
which is (a) 0.30 times the dimension in the direction Cracked 0.15Ig
stability being considered for rectangular columns; or (b)
Beams 0.15Ig
0.25 times the diameter of circular columns.
The moment of inertia and cross-sectional areas for elastic
sidesway, calculated using a first-order frame analysis; M2s
analysis at factored load level may be calculated by Code
is the factored end moment on a compression member at the
Section 6.6.3.1.1 and is shown in Table 3. It should be noted
end at which M2 acts, due to loads that cause appreciable
that moment of inertia values in ACI CODE-440.11-22 are
sidesway, calculated using a first order elastic frame anal-
lower than those provided in ACI 318-19 for steel-RC due to
ysis; and δs is the moment magnification factor for sway
lower stiffness of GFRP reinforcement.
frames. Code Section 6.6.4.6.2 provides two ways to calcu-
Code Section 6.6.4.6.4 requires magnifying the first-order
late its value, given as Eq. (7a) and (7b)
moment to consider the second-order effects produced by
slenderness in sway frames, given as Eq. (6a) and (6b)
_ 1

δs = 1 − Q ≥ 1 (7a)
M1 = M1ns + δsM1s (6a)
___________
1
M2 = M2ns + δsM2s (6b)
δs = ≥ 1 (7b)
_ ∑ Pu
1−
0.75∑ Pc
where M1 is the lesser factored end moment on a compres-
sion member; M1ns is the factored end moment on a compres- The critical buckling load Pc is calculated by Code Section
sion member at the end at which M1 acts, due to loads that 6.6.4.4.2, given as
cause no appreciable sidesway, calculated using a first-order
elastic frame analysis; M1s is the factored end moment on π2( EI)eff
_
a compression member at the end at which M1 acts, due Pc = (8)
(k lu )2
to loads that cause appreciable sidesway, calculated using
a first-order elastic frame analysis; and M2 is the greater where Pc is critical buckling load; klu is effective length (the
factored end moment (always positive) on a compression
member. If transverse loading occurs between supports, M2 length of a pin-ended column having same buckling load as
is taken as the largest moment occurring on a member; M2ns original column); and (EI)eff is effective moment of inertia,
is the factored end moment on a compression member at the which can be calculated by Code Section 6.6.4.4.4a and
end at which M2 acts, due to loads that cause no appreciable 6.6.4.4.4b, given herein as Eq. (9a) and (9b)
@seismicisolation
@seismicisolation

0.24 EcIg
_

(EI)eff = 1 + B (9a) Table 4—Factored axial, shear, and moment values
dns
Ultimate loads Sway column Non-sway column
0.2EcIg
( EI)eff = _
1 + B + 0.75 EfIf (9b) Vu, kip 22 —
dns
Pu, kip 789 134
where Bdns is the ratio of maximum factored sustained axial (Mu)top, kip-ft 145 38
load to maximum factored axial load; and If is the moment (Mu)bott, kip-ft 197 94.4
of inertia of the GFRP bars about the centroid of the cross
Note: 1 kip = 4.44 kN; 1 kip-ft = 1.35 kN-m.
section.
Code Section 6.6.4.5 implies amplifying M2 for the effects laterally unbraced; therefore, slenderness effects were
of member curvature in a non-sway frame given as checked as per ACI CODE-440.11-22, Eq. (6.2.5.1a) (that
is, klu/r < 17). The unsupported length of the column, as
Mc = δM2 (10) shown in Fig. 1(a), is 15.5 ft (4.72 m) and its cross-section
dimensions were calculated using relations provided in this
where Mc is factored moment amplified for the effects paper equal to 26 x 26 in. (660 x 660 mm). The effective
of member curvature; and δ is magnification factor for length factor was calculated using alignment charts given
non-sway frames as given in Code Section 6.6.4.5.2 in Fig. R6.2.5.1 in ACI CODE-440.11-22 (that is, Fig. 2),
which depend on the relative stiffnesses of columns to the
_
C m floor members at column top and bottom joints. In this
δ = Pu ≥ 1.0 (11)
_ design example, the column frames into beams at the top
1 − 0.75P
c joint, whereas it frames at bottom in a two-way slab. It was
assumed that the columns in the stories above and below
where Cm (factor relating actual moment diagram to an equiv- had the same cross-section dimensions. The gross moment
alent uniform moment diagram) shall be in accordance with of inertia of the column was equal to 38,080 in.4 (15.8 × 109
6.6.4.5.3a and 6.6.4.5.3b, given herein as Eq. (12a) and (12b): mm4), and the effective moment of inertia calculated as per
(a) For columns without transverse loads applied between Table 3 was equal to 15,232 in.4 (6.3 × 109 mm4).
supports As stated in ACI CODE-440.11-22, Section R6.6.3.1.1, it
is sufficiently accurate to take the gross moment of inertia of
_M1 a T-beam equal to twice that of its web. Using this approach,
Cm = 0.6 − 0.4 M (12a)
2
the moment of inertia of T-beams framing into the column
at the top joint was calculated equal to 81,000 in.4 (33.7 ×
(b) For columns with transverse loads applied between
109 mm4), and the effective moment of inertia as per Table 3
supports
was equal to 12,150 in.4 (5 × 109 mm4). Similarly, the
moment of inertia at the lower joint was calculated for the
Cm = 1.0 (12b)
slab framing into the column. The width of the slab in the
transverse direction was considered equal to 14 ft (4.3 m)
EXAMPLES OF COLUMN DESIGN AND
and its thickness equal to 7 in. (178 mm), as given in the ACI
DISCUSSION
318-19 Design Handbook.13 Its gross moment of inertia was
The required strength for the two columns subjected to
calculated equal to 4802 in.4 (2 × 109 mm4), which reduced to
lateral and gravity loads was checked using the factored
720 in.4 (0.3 × 109 mm4) when calculating effective moment
load combinations in Chapter 5 and analysis procedures in
of inertia as per Table 3.
Chapter 6 of the ACI CODE-440.11. The calculated values
The relative stiffness at the top and bottom joints was calcu-
of axial load, moment, and shear demands used in this study
lated as per Eq. (5) of this paper (as given in ACI CODE-
are given in Table 4 as originally available from the sources of
440.11-22), which were found equal to 3 for top and 30 for
the steel-RC cases.12,15 It should be noted that for simplicity, a
bottom joints, respectively. Using relative stiffness factors in
single combination of ultimate axial load and moment (that is,
alignment charts given in Fig. 2, the effective length factor
Pu and Mu) for each of the two columns was adopted, whereas
was calculated equal to 2.8 and radius of gyration equal to
in practice, the demand of several combinations of loads and
7.5 from ACI CODE-440.11-22, Section 6.2.5.2. The values
moments must be satisfied. Also, the design was carried out
of effective length factor, unsupported length of column,
by keeping the reinforcement amount as close as possible to
and radius of gyration were used in Eq. (6.2.5.1a). It was
the minimum requirements of the Code (that is, 1%Ag).
observed that slenderness effects cannot be neglected; hence
the column should be designed by considering the second-
Sway column using low- and high-modulus GFRP
order effects.
bars
The second step after slenderness is to investigate if
The column was designed using the Code-referenced
the column should be analyzed as sway or non-sway. ACI
low-modulus (Ef = 6500 ksi [44,815 MPa]) GFRP bars, with
CODE-440.11-22, Section 6.6.4.3 states that a column can
a concrete strength of 5000 psi (35 MPa). The columns were
be analyzed as part of a non-sway frame if: (a) column end
designed as per procedure provided in ACI CODE-440.11-
moments due to second-order effects do not exceed 5%
22, Fig. R6.2.5.3. The column in this design example is
of the first-order end moments; and (b) the stability index
@seismicisolation
@seismicisolation

Fig. 3—Interaction diagram for GFRP-RC sway columns.
calculated as per 6.6.4.4.1 does not exceed 0.05. The sum to 2.5. The magnified moments calculated as per Section
of all factored column and wall gravity loads were consid- 6.6.4.6.1 were the lesser moment (M1) equal to 368 kip-ft
ered as given in ACI 318-19 Design Handbook13 equal to (500 kN-m) and the larger end moment (M2) equal to 493
25,700 kip (114,320 kN) and horizontal story shear equal to kip-ft (670 kN-m).
775 kip (3450 kN). Because the first story of a building is ACI CODE-440.11-22, Section 10.6.1.1 states that area
often assumed hinged at 0.67lu, the following equation was of longitudinal reinforcement shall be at least 0.01Ag. Once
used to calculate deflection at a distance l to the hinge the concrete cross-section dimensions and reinforcement
amount are selected, the strength interaction diagram for
_ l3
Vus that case is constructed by means of the spreadsheet specif-
δo = (13) ically developed for GFRP-RC. The spreadsheet allows
3∑ EI
placing the reinforcement in the first (d1) and last (dn) layers
The deflection was found equal to 1.16 in. (29 mm), and as close to the outer column face as permitted by the Code.
the stability index (calculated as per Eq. (3)) was equal to Bars could also be inserted along the two lateral sides of the
0.176, which is greater than 0.05; hence, the column was cross section. The spreadsheet recomputes capacity resulting
analyzed and designed as part of a sway frame. For a sway from changes in sectional strain to create a smooth plot of a
frame, the secondary moments at the end of the column nominal strength interaction diagram (Pn-Mn). The values of
due to differential movement of the ends of columns were nominal axial force and moment are multiplied by strength
calculated as per ACI CODE-440.11-22 Section 6.6.4.6.4 as reduction factors, and limits on axial strength are applied to
given by Eq. (6a) and (6b) in this paper. The sway magnifi- create the design strength interaction diagram (ϕPn-ϕMn). As
cation factor was calculated as per Section 6.6.4.6.2, given an example of the output of the spreadsheet, Fig. 3 shows
as Eq. (7a) and (7b) in this paper. ACI CODE-440.11-22 the design interaction diagram developed for the 26 x 26 in.
allows three approaches for calculating moment magnifier, (660 x 660 mm) cross section of the sway column with rein-
including: the Q method, the sum of P method, and second- forcement consisting of eight No. 9 (M29) bars for both low-
order elastic analysis. Because the example is based on hand and high-modulus GFRP. The horizontal dotted line in the
calculations, the sum of P method was used to calculate interaction diagrams shows the limit on the nominal axial
sway magnification factor, as given by Eq. (7b). The critical compressive strength to account for accidental eccentricities
buckling load can be calculated as per Section 6.6.4.4.2 in as per ACI CODE-440.11-22, Table 22.4.2.1 (that is, 0.8 for
ACI CODE-440.11-22, given by Eq. (8) in this paper. The a column with ties).
effective stiffness was calculated as per Section 6.6.4.4.4a, The interaction diagram shown in Fig. 3 was constructed
given by Eq. (9a), where factor Bdns was considered equal to by locating critical points according to x (that is, the location
zero for short-term lateral loads as allowed in ACI CODE- of neutral axis) based on selected Pn-Mn pairs. The two P-M
440.11-22, Section R6.6.4.6.2. The effective stiffness was curves show nominal strength (that is, points shown with
calculated equal to 36.8 × 106 kip-in.2 (105.6 × 109 kN-mm2) plain letters A, B, C, D, and E) and design strength (that is,
and critical load equal to 1340 kip (5960 kN). The sway points shown with dashed letters A′, B′, C′, D′, and E′). Specif-
magnification factor calculated as per Eq. (7b) was equal ically, point A in Fig. 3 represents the maximum nominal
@seismicisolation
@seismicisolation

Table 5—Steel-RC and GFRP-RC sway column analysis and design
Column analysis Column design
Effective Moment magni-
Ultimate loads length factor fication factor Steel-RC column GFRP-RC column
Applied Moment, Magnified Size, Reinforcement Magnified Size, Reinforcement
load, kip kip-ft Steel GFRP Steel GFRP moment, kip-ft in. area, in.2 (ratio, %) moment, kip-ft in. area, in.2 (ratio, %)
789 197 2.2 2.8 1.12 2.5 221 24 x 24 8 No. 8 (1.0) 493 26 x 26 8 No. 9 (1.1)
Note: 1 kip = 4.44 kN; 1 kip-ft = 1.35 kN-m; 1 in. = 645 mm .

2 2
compressive force corresponding to zero eccentricity (Mn = the capacity (9%) for the column subjected to same demand
0 and x = +¥), which is the uppermost point in the interaction (that is, as a low-modulus GFRP-RC column given in
diagram. Point B′ on the design domain limit represents the Table 5) was noticed. The column cross section, reinforce-
case of maximum compressive force usable in design. The ment details, and interaction diagrams developed for both
two modes of failure (tension and compression-controlled low- and high-modulus GFRP-RC columns are shown in
modes) are separated by the “balanced failure” shown by Fig. 3.
point C′, representing FRP rupture (note: the guaranteed
strength of GFRP bars is replaced by 0.01Ef, as specified in Non-sway column using low- and high-modulus
the Code) and concrete crushing simultaneously (x = xb). If GFRP bars
the neutral axis shifts beyond xb, the failure mode shifts from The column example taken from the textbook by Wight
compression to tension. The lowermost point in the interac- and Macgregor15 was redesigned with GFRP reinforcement,
tion diagram (E′) corresponds to maximum tensile force (Mn considering it as part of a non-sway frame as the stability
= 0 and x = –¥), and maximum strain in the reinforcement. index (Q = 0.04) and magnified moment were within the
Any combination of ultimate axial load and moment (that limits stated in Section 6.6.4.3 for a non-sway frame. The
is, Pu-Mu, shown by a black dot) laying within the interac- magnification factor calculated by analysis was 0.7; hence,
tion curve represents safe (and outside, an unsafe) column a minimum magnification factor of 1.0 was used to calcu-
design. late the magnified moment. The P-M diagram developed for
The GFRP-RC column subjected to same ultimate loads non-sway columns showing the nominal and design capacity
required larger cross section compared to steel-RC as the curves is shown in Fig. 4. The (Pn, Mn) curve shows the
axial strength of GFRP reinforcement is not considered capacity before and (ΦPn, ΦMn) after the application of the
in resistance calculations and is replaced with equal area strength reduction factors. The column failed with dimen-
of concrete. Also, the higher magnification factor resulted sions of the steel-RC section (14 x 14 in. [356 x 356 mm])
in a very large, magnified moment, as given in Table 5, and reinforcement consisting of four No. 8 (M25) bars, as
together with the limits on the maximum strength of GFRP shown in Fig. 4(a), where the demand (Pu-Mu) shown by the
bars, the GFRP-RC column required bigger cross-sectional black dot lies outside the design capacity curve. Therefore,
dimensions than a steel-RC column. For example, a column the cross-section dimensions were increased to 18 x 18 in.
designed with GFRP reinforcement failed with dimensions (460 x 460 mm) and reinforcement consisting of four No. 9
similar to that of the steel-RC (24 x 24 in. [610 x 610 mm]); (M29) bars to satisfy the demand. As shown in Fig. 4(b),
therefore, the column size was increased to 26 x 26 in. (660 x the demand is within the design capacity curve, representing
660 mm) to augment its load-carrying capacity to exceed a safe column. It was observed that GFRP-RC columns
the demand. The minimum reinforcement depends on the require bigger cross sections and reinforcement areas than
gross area of the cross section (Afmin = 0.01Ag); therefore, steel-RC. The effective length factor, moment magnification
a GFRP-RC column required more reinforcement area than factor, cross-sectional dimensions, and required reinforce-
steel-RC. For example, a column designed with GFRP-RC ment are shown in Table 6.
required eight No. 9 (M29) bars, whereas that with steel-RC Similar to the case of sway frames, an effort was made
required eight No. 8 (M25) bars. The values of effective to investigate the impact of using high-modulus bars
length factor, moment magnification factor, cross-sectional on non-sway column design. The high-modulus bars
area, and longitudinal reinforcement for a column in sway showed a 12% increase in the column capacity compared to
frame are provided in Table 5, which includes values for low-modulus GFRP-RC column subjected to same demand
steel-RC taken from ACI 318-19 Design Handbook.13 (that is, the low-modulus GFRP-RC column given in Table 6).
This study also investigated the effect of high-modulus The column cross section, reinforcement details, and interac-
(Ef = 8700 ksi [60,000 MPa]) GFRP reinforcement on column tion diagrams developed for non-sway column are shown in
design. The compressive strength of GFRP reinforcement is Fig. 4.
not considered in resistance calculations when the GFRP
is in compression (that is, the area of GFRP replaced with PARAMETRIC STUDY
concrete); hence, the column dimensions remained same as A parametric study was carried out by varying concrete
with low-modulus GFRP bars (26 x 26 in. [660 x 660 mm]). compressive strength fc′, aspect ratio, and applied load eccen-
Further, due to a limit on maximum GFRP tensile strain (that tricity to evaluate implications on the design of GFRP-RC
is, 0.01) by Code Section 10.3.2.1, only a slight increase in column cross sections. For comparison, steel-RC sections
@seismicisolation
@seismicisolation

Fig. 4—Interaction diagram of GFRP-RC non-sway columns.
Table 6—Steel-RC and GFRP-RC non-sway column analysis and design

Column analysis Column design
Effective Moment magni-
Ultimate loads length factor fication factor Steel-RC column GFRP-RC column
Applied Moment, Magnified Size, Reinforcement Magnified Size, Reinforcement
load, kip kip-ft Steel GFRP Steel GFRP moment, kip-ft in. area, in.2 (ratio, %) moment, kip-ft in. area, in.2 (ratio, %)
134 94.4 0.77 0.80 0.53 0.7 94.4 14 x 14 4 No. 7 (1.2) 94.4 18 x 18 4 No. 9 (1.2)
Note: 1 kip = 4.44 kN; 1 kip-ft = 1.35 kN-m; 1 in.2 = 645 mm2.
were also designed by changing the parameters stated 22 x 22 in. (560 x 560 mm), which decreased to 18 x 18 in.
previously. The yielding strength of steel used was 60 ksi (460 x 460 mm) and 16 x 16 in. (406 x 406 mm) as concrete
(414 MPa) and modulus of elasticity 29,000 ksi (200 GPa), compressive strength increased to 7500 and 10,000 psi (52
whereas the GFRP reinforcement used was compliant with and 70 MPa) respectively. Similarly, the required reinforce-
ASTM D7957 as referenced by ACI CODE-440.11-22. ment decreased from six No. 9 (M29) bars at fc′ = 5000 psi
To compare results with steel-RC, both sections (steel-RC (35 MPa) to four No. 9 (M29) and four No. 8 (M25) at
and GFRP-RC) were subjected to same demand (that is, fc′ = 7500 and 10,000 psi (52 and 70 MPa), respectively
no magnification factors were applied). Therefore, a cross (note: reinforcement used at all three concrete strengths
section of 20 x 20 in. (508 x 508 mm) was used and varied is 1.2%Ag). It was further noticed that, at higher concrete
as required. The reinforcement ratio was kept as close to strength, GFRP-RC sections performed similar to steel-RC.
minimum required 1% as possible throughout the parametric For example, as shown in Table 7, at fc′ = 5000 psi (35 MPa)
study. the required dimensions for GFRP-RC section are 22 x 22 in.
(560 x 560 mm), whereas those for steel-RC are 20 x 20 in.
Design with different fc′ values (508 x 508 mm). However, when concrete strength increased
Four different values of concrete strength (fc′ = 2500, to 7500 psi (52 MPa) and above, the required dimensions for
5000, 7500, and 10,000 psi [18, 35, 52, 70 MPa]) were used. both RC sections are the same.
The cross sections (steel-RC and GFRP-RC) were subjected In contrast, when concrete strength was decreased to
to ultimate axial compressive load of 789 kip (3510 kN) and 2500 psi (18 MPa), the required cross sections signifi-
ultimate moment of 2367 kip-in. (267 kN-m) (no magnifi- cantly increased to satisfy the demand. Similar effects were
cation factors applied). The reinforcement area was kept as observed in the case of steel-RC; however, unlike steel-RC,
close to 1% of gross concrete area as possible. As expected, GFRP-RC dimensions and reinforcement area increased
RC cross-section dimensions significantly decreased more rapidly. For example, the steel-RC section satisfied
with increasing concrete strength. For example, as shown the demand with cross-sectional dimensions equal to 26 x
in Table 7, at concrete strength of 5000 psi (3 MPa), the 26 in. (660 x 660 mm) with eight No. 9 (M29) bars (ρ =
required GFRP cross section to satisfy the demand (Pu = 0.011Ag), whereas GFRP-RC required 28 x 28 in. (710 x
789 kip [3510 kN] and Mu = 2367 kip-in. [267 kN-m]) is 710 mm) with eight No. 9 (M29) bars (ρ = 0.01Ag). It has
@seismicisolation
@seismicisolation

Table 7—Cross sections at different fc′ values
Demand Steel-RC GFRP-RC
Applied Ultimate moment, Eccentricity, Concrete Reinforcement area, Reinforcement area,
load, kip kip-in. in. strength, psi h/6 Size, in. in.2 (ratio, %) h/6 Size, in. in.2 (ratio, %)
fc′ = 2500 4.3 26 x 26 8 No. 9 (1.1) 4.6 28 x 28 8 No. 9 (1.0)
fc′ = 5000 3.3 20 x 20 4 No. 9 (1.0) 3.6 22 x 22 6 No. 9 (1.2)
789 2367 3.0
fc′ = 7500 3.0 18 x 18 4 No. 9 (1.2) 3.0 18 x 18 4 No. 9 (1.2)
fc′ = 10,000 2.7 16 x 16 4 No. 8 (1.2) 2.7 16 x 16 4 No. 8 (1.2)
Note: 1 kip = 4.44 kN; 1 kip-in. = 113 kN-mm; 1 in. = 25.4 mm; 1 in. = 645 mm .
2 2
Fig. 5—Cross sections at different fc′ values.

been noted that higher concrete strengths have profound Design with different cross-section aspect ratio
effects on cross-sectional dimensions and reinforcement The cross-section aspect ratio was changed from 1.0 to 1.5
area of GFRP-RC. The effect of changing concrete strength and 2.0. The ultimate axial load and moment were kept same
on cross-sectional dimensions and reinforcement area can be for both RC cross sections (steel-RC and GFRP-RC) with
visualized in Table 7 and Fig. 5. no magnification factors applied (Pu = 789 kip [3510 kN],
Also, it is worth noting here that kern distance signifi- Mu = 2367 kip-in. [267 kN-m]). The reinforcement ratio was
cantly decreases with increasing concrete strength. For kept as close to 1% of the gross concrete area as possible and
example, for GFRP-RC, at fc′ = 5000 psi (35 MPa), the concrete strength was 5000 psi (35 MPa).
applied axial load acts at e = 3.0 in. (76 mm), which is within As expected, when changing the cross-section aspect ratio
kern distance (h/6 = 3.6 in. [92 mm]), implying that the from 1.0 to 1.5 and 2.0, the required dimensions for both RC
axial load does not cause tension in the section. However, sections decreased. For example, as shown in Table 8, the
at a concrete strength of 7500 psi (52 MPa), the kern GFRP-RC cross-sectional area decreased to 17 x 26 in. (432
distance significantly decreased and ultimately, the eccen- x 660 mm) from 22 x 22 in. (560 x 560 mm) and steel-RC to
tricity falls outside the kern (e = 3.0 in. [76 mm] and h/6 16 x 24 in. (406 x 610 mm) from 20 x 20 in. (508 x 508 mm)
= 2.7 in. [69 mm]) when fc′ = 10,000 psi (70 MPa). There- when changing aspect ratio from 1.0 to 1.5. It further
fore, as observed by calculations, higher concrete strength decreased to 14 x 28 in. (356 x 710 mm) and 13 x 26 in.
help decreasing cross-sectional dimensions and axial load (330 x 660 mm) for GFRP-RC and steel-RC, respectively,
causes tension in the section. Subsequently, GFRP-RC cross when increasing the aspect ratio to 2.0. It can be observed
sections take advantage of tensile properties of GFRP rein- in Table 8 that the required dimensions and reinforcement
forcement and require dimensions similar to steel-RC. For area for GFRP-RC are larger than steel-RC for the three
example, as shown in Table 7, at fc′ = 7500 and 10,000 (52 aspect ratios investigated. It should be noted that, in all three
and 70 MPa), GFRP-RC required cross-sectional areas of cross sections the axial load is applied at e = 3.0 in. (76 mm),
18 x 18 in. (460 x 460 mm) and 16 x 16 in. (406 x 406 mm) which falls within the kern distance (h/6). This implies that
and reinforcement of four No. 9 (M29) and four No. 8 (M25) the applied load does not cause tension in the cross section.
bars, respectively—the same as steel-RC. The effect of changing the aspect ratio may be more promi-
nent for GFRP-RC cross sections subjected to highly eccen-
tric loads (that is, eccentricity exceeding kern distance h/6).
Therefore, an attempt was made to investigate the cross
@seismicisolation
@seismicisolation

Table 8—Cross sections at different aspect ratios
Demand Steel-RC GFRP-RC
Applied Ultimate moment, Aspect Reinforcement area, Reinforcement area,
load, kip kip-in. Eccentricity, in. ratio h/6 Size, in. in.2 (ratio, %) h/6 Size, in. in.2 (ratio, %)
1.0 3.3 20 x 20 4 No. 9 (1.0) 3.7 22 x 22 6 No. 9 (1.2)
789 2367 3.0 1.5 4.0 16 x 24 4 No. 9 (1.0) 4.3 17 x 26 6 No. 8 (1.0)
2.0 4.3 13 x 26 4 No. 9 (1.1) 4.6 14 x 28 4 No. 9 (1.0)
1.0 3.0 18 x 18 4 No. 9 (1.2) 3.0 20 x 20 4 No. 9 (1.0)
395 2367 6.0 1.5 3.7 14 x 22 4 No. 8 (1.0) 3.7 14 x 22 4 No. 8 (1.0)
2.0 4.0 12 x 24 4 No. 8 (1.0) 4.0 12 x 24 4 No. 8 (1.0)
Note: 1 kip = 4.44 kN; 1 kip-in. = 113 kN-mm; 1 in. = 25.4 mm; 1 in.2 = 645 mm2.
Fig. 6—Cross sections at different aspect ratios.

sections by increasing the eccentricity to 6.0 in. (152 mm). while keeping the moment constant (Mu = 2367 kip-in.
The axial load was decreased to half (395 kip [1760 kN]) [267 kN-m]), with no magnification factors applied, to eval-
while moment was kept the same (2367 kip-in. [267 kip-ft]). uate the effect on cross-sectional dimensions and reinforce-
For the GFRP-RC, it was observed that at an aspect ratio ment area. The eccentricities were gradually enhanced at
of 1.0, the required dimensions decreased to satisfy the increments of 0.5 in. (13 mm), except for the points inside
demand (20 x 20 in. [508 x 508 mm] from 22 x 22 in. [560 (0.1h), exactly on (h/6), and outside (0.2h) the kern (note:
x 560 mm]), as the axial load decreased to half of original the kern distances are based on GFRP-RC cross-section
load). It was further observed that for GFRP-RC the required dimensions). For square cross sections, the eccentricity
dimensions and reinforcement area significantly decreased values used in the calculations are 1.0., 1.5, 2.0, 2.4, 2.5, 3.0,
with increasing aspect ratio and were similar to steel-RC at 3.6, and 4.0. The reinforcement ratio was kept as close to a
aspect ratios equal to 1.5 and 2.0. For example, as shown in minimum 1% of the gross concrete area as possible and the
Table 8, at aspect ratios of 1.5 and 2.0, the required dimen- concrete strength used was 5000 psi (35 MPa). To compare
sions for GFRP-RC are 14 x 22 in. (356 x 560 mm) and the efficacy of GFRP-RC, steel-RC cross sections were
12 x 24 in. (305 x 610 mm), which are 23% and 28% less also designed by changing the same parameters as stated
from a cross section with an aspect ratio of 1.0 (that is, 20 x previously. It should be noted that both RC sections were
20 in. [508 x 508 mm]) and are the same as steel-RC at same subjected to same demand (that is, no magnification factors
aspect ratios. For all three aspect ratios, the eccentricity is were applied).
outside the kern distance (h/6), implying that there is tension The cross-section design started with minimum cross-
in the cross sections. Figure 6 shows the trend of changing sectional area sufficient to satisfy the demand at e = 1.0 in.
the cross-sectional area when increasing the aspect ratio at (25.4 mm), as shown in Table 9. It can be observed in Table 9
eccentricities of 3.0 and 6.0 in. (76 and 152 mm). that when axial loads are high, the required cross-sectional
dimensions for GFRP-RC increase more rapidly compared
Design with different load eccentricities to steel-RC. For example, at an eccentricity value equal to
The analysis was carried out with the intent of 1.0 in. (25.4 mm) (Pu = 2367 kip [10,530 kN]), the required
changing eccentricities by altering the values of axial@seismicisolation
load dimensions for GFRP-RC are 34 x 34 in. (860 x 860 mm),
@seismicisolation

Table 9—Cross sections with different eccentricities at constant moment
Steel-RC GFRP-RC
Ultimate load, Reinforcement area, Reinforcement area, in.2
Eccentricity, in. kip e/h Cross-section area, in. in.2 (ratio, %) e/h Cross-section area, in. (ratio, %)
1.0 2367 0.03 32 x 32 12 No. 9 (1.1) 0.03 34 x 34 12 No. 9(1.0)
1.5 1578 0.05 26 x 26 8 No. 9 (1.1) 0.05 28 x 28 8 No. 9(1.0)
2.0 1184 0.09 0.08
24 x 24 6 No. 9(1.0)
2.4 (0.1h) 1075 0.10 22 x 22 6 No. 9 (1.2) 0.10
2.5 947 0.11 0.11
22 x 22 6 No. 9(1.2)
3.0 789 0.15 0.13
3.6 (h/6) 710 0.18 20 x 20 4 No. 9 (1.0) 0.18
20 x 20 4 No. 9(1.0)
4.0 (0.2h) 592 0.2 0.2
Note: 1 in. = 25.4 mm; 1 kip = 4.44 kN; 1 in.2 = 645 mm2.
Fig. 7—Cross sections with different eccentricities at constant moment.

and those for steel-RC are 32 x 32 in. (812 x 812 mm). Vn = Vc + Vf (14)
However, when the eccentricity value increases to 4.0 in.
(100 mm), the required dimensions for two RC sections are where Vn is nominal shear strength; Vc is nominal shear
the same. Similarly, the required reinforcement area also strength provided by the concrete; and Vf is nominal shear
decreases and is the same for GFRP-RC and steel-RC (four strength provided by GFRP shear reinforcement.
No. 9 [M29] bars). Despite stricter design limits compared The shear strength provided by concrete is calculated
to steel-RC, the GFRP was found effective to resist loads as the greater of two expressions, given by Code Sections
with high eccentricities, especially for values exceeding the 22.5.5.1a and 22.5.5.1b as follows
kern distance (h/6). For example, in Table 9, at eccentricity _
e = 4.0 in. (100 mm), the axial load acts outside the kern c = 5λ kcr
V √fcʹbd (US units)
distance (h/6 = 3.6), causing tension in the cross section and
    _ (15a)
Vc = 0 . 42λ kcr √fcʹ bd ( SI units)
requires that cross-section dimensions for two RC sections
_
are same. The required reinforcement and cross-sectional
c = 0 . 8λ √fcʹbd
V (US units)
area with changing eccentricity values can be seen in Table 9
   
_ (15b)
and increase in reinforcement requirements and cross- Vc = 0 . 066λ √ fcʹbd ( SI units)
sectional area with increasing eccentricities are visualized
in Fig. 7. where kcr is ratio of the depth of elastic cracked section neutral
axis to the effective depth given by the Code Commentary
COLUMN PROVISIONS FOR SHEAR IN
Section R22.5.5.1, shown as follows
CODE-440.11-22
ACI CODE-440.11-22 Section 10.5.3.1 references Section ____________
22.5 for the calculation of nominal shear strength of column, = √2  
kcr,rect ρfnf+ (ρfnf)2 − ρfnf
which can be calculated as given in Section 22.5.1.1
@seismicisolation
@seismicisolation

where ρf = (Af/bd) is the reinforcement ratio; Af is the area of EXAMPLE OF SHEAR DESIGN AND DISCUSSION
GFRP reinforcement; nf = (Ef/Ec) is the modular ratio; and Ec The sway GFRP-RC column was designed for gravity
is the concrete elastic modulus calculated as given by Code load and magnified moment (Pu = 789 kip [3510 kN], Mu =
Section 19.2.2.1 493 kip-ft [670 kN-m]) that required a cross-sectional area
_ equal to 26 x 26 in. (660 x 660 mm). This section discusses
Ec = 57, 000 √ fc US units ( ) the shear design of the aforementioned cross section
    _ (17)
Ec = 4700 √fc (SI units) subjected shear force (Vu) of magnitude 22 kip (98 kN), as
given in Table 4. The shear strength provided by the concrete
____________ _____________ cross section (ΦVc) was calculated with Eq. (22.5.5.1a) and
and λ = √2   /(1 + [d/10]) (√   2 /(1 + 0 . 004d) is the size effect (22.5.5.1b), as given in column provisions for shear. The
factor as given in ACI CODE-440.11-22, Section 22.5.5.1.3. GFRP bars used as transverse reinforcement were compliant
The size effect factor was considered for these examples with ASTM D7957,12 which states that the guaranteed ulti-
because h exceeded 10 in. (254 mm). mate tensile force of the bent portion of a bar shall be greater
The shear strength provided by the GFRP reinforcement is than or equal to 60% of the values of guaranteed ultimate
as given in Code Section 22.5.8.5.3 tensile force provided in ASTM D7957, Table 3.12 Also,
for transverse reinforcement, No. 4 (M13) bars were used,
Vf = Afvfft(d/s) (18) having a minimum inside diameter of the bend equal to 3 in.
(76 mm) as given in ASTM D7957, Table 4.12
where Afv is the area of shear reinforcement as given in Code It was observed from the calculations that shear strength
Commentary Eq. (R22.5.8.5) provided by the concrete cross section alone is not suffi-
cient to resist the shear force. Hence the shear capacity must
_Afv _ Vu − Φ Vc increase by means of shear reinforcement to satisfy the shear
s = (19)
Φ fft d demand. It should be noted that limits on the shear strength
provided by concrete resulted in lower Vc together with a
and fft is the permissible stress in the GFRP shear reinforce- 40% reduction in the strength at the bend of GFRP trans-
ment. The design tensile strength of GFRP transverse rein- verse reinforcement12 significantly affect shear design. The
forcement is controlled by the strength of the bent portion limits on the maximum strength of GFRP transverse rein-
of the bar and by a strain limit of 0.005, as given by Code forcement given by Code Section 20.2.2.6 only allowed the
Section 20.2.2.6 maximum design tensile strength of reinforcement equal to
32.5 ksi (224 MPa), which is 70% less than actual strength
fft ≤ (ffb, 0.005Ef) (20) value of No. 4 (M13) bars.
In this column example, to augment the shear strength
where ffb is the guaranteed ultimate tensile strength of the of the cross section, No. 4 (M13) bars were used at 9 in.
bent portion of the bar. Its minimum value be taken as spec- (228 mm) center-to-center. Whereas the same column when
ified in ASTM D7957; and s is center-to-center spacing of designed with steel-RC only required minimum shear rein-
transverse reinforcement. forcement (that is, No. 4 [M29] at 16 in. [406 mm] center-to-
Maximum spacing smax between legs of shear reinforce- center) to hold the longitudinal reinforcement, as the shear
ment is calculated as the least of maximum spacing limita- strength provided by the concrete cross section for steel-RC
tions given by Code in 10.6.2.2 and its Commentary in (24 x 24 in. [610 x 610 mm]) was sufficient to satisfy the
R22.5.8.5 shear demand.
The Code Sections 25.7.2.3a, 25.7.2.3b, and 25.7.2.3c
_ Afv Φ fft d require that lateral support from ties be provided for bars at
smax = V − Φ V (21)
u c every corner, and to bars with greater than 6 in. (152 mm)
clear on each side. Therefore, in this column example, two
_ Afv fft_ C-shaped tie bars forming a diamond shape for middle
smax = (US units)
0 . 75 √fcb longitudinal bars, in addition to two overlapping C-shaped
   
(22a)
_
A
f
fv ft _ tie bars for corner longitudinal reinforcement, were used,
smax = (SI units) as shown in Fig. 8. The tie size, its dimensions, and their
0 . 062 √fcb
distribution along column height are schematically shown in
Afv fft Fig. 8 for a sway column.
_
smax = (US units) The non-sway GFRP-RC column was subjected to an
50b
   
(22b) axial compressive load of 134 kip (596 kN), magnified
_ Afv fft
smax = (SI units) moment of 94.4 kip-ft (128 kN-m), and no shear force value
0 . 35b
is specified in reference source.15 To resist the gravity load
The maximum tie spacing requirement is also provided in and applied moment, the column required a cross-sectional
the Code Section 22.7.2.1. The maximum tie spacing shall area equal to 18 x 18 in. (460 x 460 mm) and four No. 9
not exceed 12db of longitudinal bar, 24db of tie bar, h, or b. (M29) longitudinal bars. Though no shear force is given in
the column example, minimum reinforcement was still used
to hold the longitudinal bars in position and avoid buckling.
@seismicisolation
@seismicisolation

Fig. 8—Schematic of reinforcement details in sway GFRP-RC column.
The transverse reinforcement was used at maximum speci- diameter; ffr is the GFRP tensile stress required to develop
fied spacings as given in Code Sections 10.6.2.2 and 25.7.2.1 the full nominal section capacity; ω is the bar location modi-
and Code Commentary Section R22.5.8.5. fication factor, taken equal to 1.5, if more than 12 in. (300
In this column, for transverse reinforcement, No. 4 (M13) mm) of fresh concrete is placed below the reinforcement
bars having an inside bend diameter equal to 3 in. (76 mm) being developed and 1.0 for all other cases.
as stated in ASTM D7957,12 Table 4 was used. As mentioned
in previous sections, the maximum spacing of ties cannot 20db (23b)
exceed 12db, 24 diameter of tie bar, and the smallest dimen-
sions of the member. Therefore, with the given cross-section 12 in. (300 mm) (23c)
dimensions and longitudinal reinforcement information in
this column example, No. 4 (M13) ties were used at 12 in. The lap splice lengths in columns shall be calculated in
(300 mm) center-to-center. accordance with 10.7.5 and 25.5. Code Section 10.7.5.2
states that in a column subjected to moment and axial force,
COLUMN PROVISIONS FOR DETAILING IN CODE- tensile stresses may occur on one face of the column for
440.11-22 moderate or large eccentricities. If such stresses occur, Code
Code Section 25.4.1.1 requires that tension or compres- Section 10.7.5.2.2 requires tension splices to be used, which
sion reinforcement at each section of a member shall be can be classified as Class A or Class B lap splices and calcu-
developed on each side of that section by embedment lated in accordance with Code Section 25.5.2.1, as given in
length, hook, mechanical device, or a combination thereof. Table 10.
Development length ld for bars in tension shall be greater Code Section 10.7.5.2.1 states that if the bar is compres-
of the values calculated by Code Sections 25.4.2.1(a), sive due to factored loads, compression lap splices shall
25.4.2.1(b), and 25.4.2.1(c) be permitted. Given no experimental data on development
length for GFRP bars in compression (ldc), Code Section
(√ )
_ ffr 25.4.9.1 states that the development length in compression
d b _ − 340 shall be conservatively taken the same as that for tension as
fc
_____________
ld =    cb ω ( US units) in Code Section 25.4.2.1.
_
13 . 6 + The minimum overlap of tie bar ends shall be greater of
db

     (23a) 20db or 6 in. (152 mm), as in 25.7.2.3.1.
( 0 . 083 √fc )
_ ffr _
db − 340 Code Section 10.7.6.2 states that the bottom tie shall be
__________________ located not more than one-half the tie spacing above the top
ld =       _ cb ω ( SI units)
13 . 6 + of footing or slab; similarly, the top tie shall be located not
db
more than one-half the tie spacing below the lowest hori-
in which cb/db shall not be taken greater than 3.5, and where zontal reinforcement in the slab, drop panel, or shear cap.
cb is the lesser of: (a) the distance from center of a bar to If beams frame into all sides of column, the top tie shall be
nearest concrete surface, and (b) one-half the center-center located not more than 3 in. (76 mm) below the lowest hori-
spacing of bars being developed; db is the nominal bar zontal reinforcement in the shallowest beam.
@seismicisolation
@seismicisolation

Table 10—Lap splice length of GFRP bars in tension (ACI CODE-440.11-22, Section 25.5.2.1)
Maximum percent of Af spliced within
Af,provided/Af,required* over length of splice required lap length Splice type lst
50 Class A Greater of: 1.0ld, 20db, and 12 in.
≥2.0
100 Class B
Greater of: 1.3ld, 20db, and 12 in.
<2.0 All classes Class B
*
Ratio of area of reinforcement provided to area of reinforcement required by analysis at splice location.
Note: 1 in. = 25.4 mm.
DETAILING OF REINFORCEMENT FOR COLUMN The development length equation for GFRP bars in tension
EXAMPLES (Eq. (25.4.2a)) was conservatively adopted for this case as
The Code permits the use of the same equation for devel- well. Also, the bar location modification factor (ω = 1.5)
opment lengths of GFRP bars in compression and tension was used to calculate the development length. The term cb/
(Eq. (25.4.2)). Therefore, the development length equation db in this column resulted in 1.83, which is well below the
for GFRP bars was conservatively adopted as given in Code permissible limit of 3.5. The development length calculated
Section 25.4.2. Also, there is more than 12 in. (300 mm) for non-sway column as per Code Section 25.4.2 resulted
of fresh concrete to be placed below the longitudinal bar in 75 in. (1900 mm). Because the tensile bar stress reaches
being developed in a sway column; hence, the bar location its full capacity at a point (limited by maximum strain 0.01)
modification factor (ω = 1.5) was also used. The Code spec- which is greater than 0.5ffu, and the ratio of area of reinforce-
ifies a maximum limit for the term cb/db as 3.5, which in ment provided to area of reinforcement required is less than
this column resulted as 1.83, well below the permitted limit. 2.0, therefore, Class B lap splices were selected. The splice
The development length was calculated using three equa- length was calculated as given in Code Section 25.5.2.1,
tions mentioned in Code Section 25.4.2 and Eq. (25.4.2.1(a)) which resulted a value equal to 98 in. (2490 mm).
governed, which resulted in 64 in. (1625 mm). Code Section 10.7.6.2 explains the distribution of ties in
The tensile bar stress at a point reaches its maximum a beam-column joint. The first tie was conservatively placed
value (limited by maximum strain 0.01), which is greater at 3 in. (76 mm), as required by Code Section 10.7.6.2,
than 0.5ffu; also, the ratio of area of reinforcement provided followed by the required spacing. C-shaped ties were used
to the area of reinforcement required in this example is less in the non-sway column, with an overlap as stated in Code
than 2.0 (that is, 1.18); therefore, Class B lap splices were Section 25.7.2.3.1, which resulted in 10 in. (254 mm) when
used. The splice length was calculated as given in Code using No. 4 (M13) ties.
Section 25.5.2.1 (that is, the greater of 1.3ld, 20db, and 12 in.
[300 mm]), which resulted 84 in. (3134 mm). CONCLUSIONS AND RECOMMENDATIONS
The lap splice calculated for the steel-RC column was In this study, a sway column example originally
33 in. (840 mm), which shows that GFRP-RC columns designed with steel reinforcement was taken from the ACI
require very large splice lengths (2.5 times greater than Design Handbook,13 a companion to ACI 318-19,14 and a
steel-RC). Unlike steel, GFRP bars cannot be bent on site non-sway column example from the textbook by Wight and
and together with more GFRP reinforcement required, they Macgregor.15 These two columns were redesigned with glass
increase the complexity in cage preparation. It is the sole fiber-reinforced polymer (GFRP) reinforcement to show the
responsibility of the contractor to splice column reinforce- implications of ACI CODE-440.11-2211 with both low- (Ef =
ment cages. As shown in Fig. 7, the bottom cage bars are 6500 ksi [44,815 MPa]) and high-modulus (Ef = 8700 ksi
shown slightly tilted just after the start of the beam to differ- [60,000 MPa]) GFRP bars. A limited parametric study
entiate from top bars. was carried out to evaluate the effects of changing values
Code Section 10.7.6.2 explains the distribution of ties in of fc′, cross-section aspect ratio, and eccentricity. Based on
a beam-column joint. It states that the bottom tie shall be the outcomes of this study, the following conclusions were
located not more than one-half the tie spacing above the top drawn:
of footing or slab; similarly, the top tie shall be located not • The stiffness values for GFRP reinforcement result in
more than one-half the tie spacing below the lowest hori- higher moment magnification factors for GFRP-rein-
zontal reinforcement in the slab, drop panel, or shear cap. In forced concrete (RC) compared to steel-RC columns.
this example, the first tie was placed at 3 in. (76 mm) from • The advantage of high modulus/strength of new-gen-
the floor top, followed by others at the required spacing. eration GFRP bars can be beneficial to resist condi-
Two pieces of C-shaped stirrups with minimum overlap as tions of large eccentricities. However, due to limits on
the greater of 20db and 6 in. (152 mm) were used as per maximum tensile strain (0.01 in./in. [0.01 mm/mm]) to
Code Section 25.7.2.3.1. In current column design, No. 4 control column curvature, these benefits are not fully
(M13) bars are used; hence, an overlap of 10 in. (254 mm) used.
is provided. • The compressive strength and stiffness of GFRP rein-
The development length for non-sway GFRP-RC forcement is replaced with an equal area of concrete;
column (18 x 18 in. [460 x 460 mm]) was calculated using hence, bigger cross sections are typically required for
Code-specified Eq. 25.4.2.1(a), 25.4.2.1(b), and 25.4.2.1(c). GFRP-RC columns when compared to steel-RC.
@seismicisolation
@seismicisolation

• Minimum reinforcement depends on the concrete gross REFERENCES
area; hence, because of larger cross-section dimen- 1. ACI Committee 440, “Guide for the Design and Construction of Struc-
tural Concrete Reinforced with Fiber-Reinforced Polymer (FRP) Bars (ACI
sions, GFRP-RC will require more reinforcement than 440.1R-15),” American Concrete Institute, Farmington Hills, MI, 88 pp.
steel-RC. 2. Zhang, P.; Lv, X.; Zhang, H.; Liu, Y.; Chen, B.; Gao, D.; and Shaikh,
• It is obvious that increasing concrete strength helps S. A., “Experimental Investigations of GFRP Reinforced Columns with
Composite Spiral Stirrups under Concentric Compression,” Journal
decrease dimensions of RC sections. However, the of Building Engineering, V. 46, No. 4, 2022, pp. 1-21. doi: 10.1016/j.
concrete strength has an additional effect on the perfor- jobe.2021.103768
mance GFRP-RC cross sections. For the case consid- 3. Hasan, H. A.; Sheikh, M. N.; and Hadi, M. N. S., “Performance Eval-
uation of High Strength Concrete and Steel Fiber High Strength Concrete
ered and at concrete strength of 7500 psi (52 MPa) Columns Reinforced with GFRP Bars and Helices,” Construction and
and above, the required dimensions for steel-RC and Building Materials, V. 134, No. 2, 2017, pp. 297-310. doi: 10.1016/j.
GFRP-RC were the same, as opposite at lower concrete conbuildmat.2016.12.124
4. De Luca, A.; Matta, F.; and Nanni, A., “Behavior of Full Scale GFRP
strengths. Reinforced Concrete Columns under Axial Load,” ACI Structural Journal,
• As expected, the rectangular sections performed better V. 107, No. 2, Mar.-Apr. 2010, pp. 589-596.
than square sections, and in most cases, GFRP-RC and 5. De Luca, A.; Matta, F.; and Nanni, A., “Structural Response of
Full Scale Reinforced Concrete Columns with Internal FRP Reinforce-
steel-RC required the same cross sections when axial ment under Compressive Load,” 9th International Symposium on Fiber
load acted outside the kern (that is, large eccentricities). Reinforced Polymer Reinforcement for Concrete Structures, D. Oehlers,
• The current development length equation in the Code M. Griffith, and R. Seracino, eds., Sydney, Australia, July 13-15, 2010.
6. Jawaheri Zadeh, H., and Nanni, A., “Flexural Stiffness and Second
result in very large values compared to steel-RC Order Effects in FRP-RC Frames,” ACI Structural Journal, V. 114, No. 2,
because there is no distinction in the requirements for Mar. 2017, pp. 533-544.
compression and tension splices. Research should be 7. Bischoff, P. H., “Member Stiffness for Frame Analysis for GFRP Rein-
forced Concrete Structures,” IABSE Symposium—Engineering the Future,
undertaken to reassess provision parameters by incor- Sept. 21-23, 2017.
porating improvements in material properties. 8. Hadhood, A.; Mohamed, H. M.; Benmokrane, B.; Nanni, A.; and
• It is observed that replacing the contribution of GFRP Shield, C. K., “Assessment of Design Guidelines of Concrete Columns
Reinforced with Glass Fiber-Reinforced Polymer Bars,” ACI Structural
reinforcement in compression with an equal area of Journal, V. 116, No. 4, July 2019, pp. 193-207. doi: 10.14359/51715663
concrete significantly affected the design. With the 9. Guérin, M.; Mohamed, H. M.; Benmokrane, B.; Shield, C. K.; and
recent advancements in material properties and manu- Nanni, A., “Effect of Glass Fiber-Reinforced Polymer Reinforcement
Ratio on Axial-Flexural Strength of Reinforced Concrete Columns,”
facturing techniques, there is need to re-investigate the ACI Structural Journal, V. 115, No. 4, July 2018, pp. 1049-1061. doi:
contribution of GFRP bars in the axial compressive 10.14359/51701279
capacity of GFRP-RC columns using high-modulus 10. Khorramian, K., and Sadeghian, P., “Experimental and Analytical
Behavior of Short Concrete Columns Reinforced with GFRP Bars under
GFRP reinforcement. Eccentric Loading,” Engineering Structures, V. 151, 2017, pp. 761-773.
doi: 10.1016/j.engstruct.2017.08.064
AUTHOR BIOS 11. ACI Committee 440, “Building Code Requirements for Structural
Zahid Hussain is a PhD Student in Civil and Architectural Engineering Concrete Reinforced with Glass Fiber-Reinforced Polymer (GFRP) Bars
Department at the University of Miami, Coral Gables, FL. He is a member (ACI CODE-440.11-22) and Commentary,” American Concrete Institute,
of ACI Committee 440, Fiber Reinforced Polymer Reinforcement. His Farmington Hills, MI, 266 pp.
research interests include sustainable materials, computational methods, 12. ASTM D7957/D7957M-22, “Standard Specifications for Solid
design, and behavior of fiber-reinforced polymer (FRP)-reinforced Round Glass Fiber Reinforced Polymer Bars for Concrete Reinforcement,”
structures. ASTM International, West Conshohocken, PA, 2022, 5 pp.
13. MNL-17(21), ACI Reinforced Concrete Design Handbook, A
Antonio Nanni, FACI, is an Inaugural Senior Scholar, Professor, and Chair Companion to ACI 318-19, American Concrete Institute, Farmington Hills,
of the Department of Civil and Architectural Engineering at the Univer- MI, 2019, 568 pp.
sity of Miami. He is a member of ACI Committees 440, Fiber Reinforced 14. ACI Committee 318, “Building Code Requirements for Structural
Polymer Reinforcement, and 549, Thin Reinforced Cementitious Products Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved
and Ferrocement. 2022),” American Concrete Institute, Farmington Hills MI, 2019, 624 pp.
15. Wight, J. K., and Macgregor, J. G., Reinforced Concrete Mechanics
and Design, sixth edition, Pearson Education, Inc., Upper Saddle River,
ACKNOWLEDGMENTS NJ, 2019.
The authors would like to thank the National Science Foundation, Grant
No. 1916342, and the Higher Education Commission of Pakistan for their
financial support of the lead author.
@seismicisolation
@seismicisolation

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
• Access to ACI Structural Journal and ACI Concrete Codes, Specifications, and Prac-
Materials Journal – free to all ACI student tices – in conjunction with ACI’s chapters,
members students 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@seismicisolation
in-class study material
@seismicisolation
https://www.concrete.org/educatorsandresearchers/aciinyourclassroom.aspx
Title No. 120-S85
Cover Spalling in Reinforced Concrete Beams Subjected to

Pure Torsion
by Allan Kuan, Evan C. Bentz, and Michael P. Collins
Cover concrete plays an important role in the torsional behavior of showed that cracked concrete exhibits a weaker compres-
reinforced concrete members because the resulting shear stresses sive response than what would be observed during a
are concentrated in these areas. Modeling its behavior is difficult cylinder test, a phenomenon known as compression soft-
due to: 1) compression softening; and 2) the possibility of spalling ening.10 Its discovery led to the issue of spalling in torsion
at high loads. Traditional approaches, which only consider one
being revisited by Hsu and Mo11 in 1985. They suggested
effect or the other, are limited in their ability to model the ultimate
that the tendency of the DCFT to overestimate the torsional
strength and torque-twist response of members over a wide range
of cover thicknesses. This paper presents a mechanics-based model strength of a member when the unspalled dimensions were
which can predict when torsional spalling occurs and quantify its used was not due to cover spalling. Instead, they argued that
effect on a member’s strength and stiffness. Its application within it was because the compression-softening effect—which
a nonlinear analysis framework and a design procedure based on would weaken the response of the concrete under torsional
ACI 318-19 is shown. Using the proposed model together with stresses, but had not yet been discovered in 1974—was not
existing compression-softening models results in improved strength considered in the model. They proposed a new model which
predictions of 187 pure torsion tests found in the literature. was similar to the DCFT but, like analytical tools for shear
modeling, neglected spalling and accounted for compression
Keywords: design; reinforced concrete; softening; space truss; spalling;
torsion.
softening. Their model showed good agreement with exper-
iments whose cover thickness were within a narrow range
INTRODUCTION of values.
Cover spalling has commonly been observed in exper- Despite further improvements in understanding how
iments of reinforced concrete beams subjected to pure reinforced concrete resists shear and torsion, there is
torsion, like those shown in Fig. 1, or torsion in combina- still a lack of consensus on the role of spalling in torsion
tion with other actions.1,2 Spalling, which usually affects the behavior and how to address it. Most analytical models
corners of the cross section but can propagate into the side for torsion developed since the 1980s have followed Hsu
cover as well, has important implications for design.3 For and Mo’s “softened approach” by neglecting spalling and
example, design codes such as ACI 318-19,4 CSA A23.3:19,5 considering compression softening,12-23 while most codi-
and AASHTO LRFD6 have special detailing requirements to fied design provisions4-6 adopt a “spalled approach” that
avoid premature failures caused by torsional spalling, and considers spalling but neglects compression softening like
neglect the cover concrete when determining a member’s the original DCFT. There are weaknesses to both methods.
torsional strength. The strength predictions made by “spalled approaches” can
Early analysis tools for torsion, such as the Diag- be very conservative for members that have thick covers or
onal Compression Field Theory7 (DCFT) developed by contain small quantities of reinforcement, and the stiffness is
Mitchell and Collins in 1974, assumed that the entirety of a usually underestimated. Spalled approaches are also incon-
member’s cover would spall to the depth of the hoop rein- sistent with modeling practices for flexural or shear behavior
forcement at failure. This is because the authors found that because: 1) the assumed section geometry neglects the
if the unspalled section geometry and concrete stress-strain cover; and 2) compression softening is neglected when the
response obtained from a cylinder test were used together, concrete is subjected to biaxial or triaxial stress states. On
the torsional strength would be systematically overesti- the other hand, softened approaches can seriously overesti-
mated. To address this issue, they recommended using the mate the torsional strength of members with thick covers.24
fully spalled dimensions because it improved the quality of A weakness shared by both methods is that spalling is treated
the strength predictions and appeared to be consistent with as a binary issue, with the cover being entirely considered
their experimental observations. The DCFT still forms the or entirely neglected, when experimental evidence—such as
basis for the torsion provisions in the previously mentioned the beams shown in Fig. 1—suggests that the reality is some-
design codes, which share its assumption that the cover fully where in between. The shortcomings of these approaches
spalls at failure.8,9
In the 1980s, research investigating the shear behavior of MS No. S-2022-311.R1, doi: 10.14359/51738839, received April 21, 2023, and
reinforced concrete led to further improvements in under- Institute. All rights reserved, including the making of copies unless permission is
standing how reinforced concrete members resist torsion. obtained from the copyright proprietors. Pertinent discussion including author’s
Experiments performed on panels subjected to pure@seismicisolation
shear is received within four months of the paper’s print publication.
@seismicisolation

by others previously. The application of the general
methodology is demonstrated in two ways: within a nonlinear
analysis framework based on the DCFT, and within a design
procedure based on the ACI 318-19 torsion provisions. Both
applications are then validated using pure torsion experi-
ments drawn from the literature.
Traditional methods of accounting for cover spalling in
reinforced concrete members can be inadequate for members
with very thick or very shallow covers. This paper presents
a simple model which can determine when torsional spalling
occurs and quantify its effect on a member’s strength and
stiffness. Implementing this model into existing analysis
tools and design procedures improves their ability to predict
the torsional strength and torque-twist response of members
over a wide range of cover thicknesses. The improved accu-
racy can be particularly beneficial for the evaluation of
torsion-critical members in existing structures.
PROPOSED MODEL
Consider the cracked concrete member subjected to pure
torsion shown in Fig. 2(a). If the torsion is assumed to be
primarily resisted by circulating shear stresses and not
warping torsion, then the member can be represented by a thin
tube which carries a uniform shear flow around its perimeter.
If tensile stresses in the cracked concrete are neglected, the
shear flow will be the result of diagonal compressive stresses
Fig. 1—Pure torsion experiments by Mitchell and Collins
in the concrete, which are equilibrated by tensile stresses in
showing effect of cover thickness on degree of spalling.1
the longitudinal and hoop reinforcement. These compressive
require engineers to use their judgement when addressing stresses, which vary through the thickness of the tube, can be
issues of torsion because a more general method is not yet represented using an equivalent rectangular stress block7,11
available. with an average stress of f2,av = α1βcsfc′ that acts over a width
Accounting for the effects of both compression softening ao, where fc′ is the concrete strength obtained from a cylinder
and spalling on torsion behavior can address the weaknesses test, βcs is a factor accounting for compression softening, and
of these traditional approaches. However, it is shown later in α1 is a stress block factor which is a function of the prin-
this paper that simply considering both at the same time is cipal compressive strain on the surface, ε2s. If the member
too conservative. The main challenge is that spalling, which is unspalled, the line of action of the shear flow will be at
can but does not always influence the torsional response, is a depth of 0.5ao beneath the outside surface of the cross
difficult to account for because the underlying mechanism section. This is shown in Fig. 2(b).
is not fully understood. One attempt at solving this problem Case 1 in Fig. 2(c) shows a corner detail of a member with
was made by Rahal and Collins, who formulated a model
25 a very thin cover as it resists an applied torque. A simple
to predict when spalling due to torsion would occur and representation of the tensile forces in the hoop reinforce-
implemented it into an analytical tool which also considered ment, T, and the compressive forces which make up the
compression softening. Although their model showed good shear flow, C, is also shown. To satisfy equilibrium at the
agreement with a selection of torsion experiments, their corner where these forces change direction, an additional
spalling check was empirically derived using a small set diagonal force must be present. This force will be compres-
of torsion tests and is relatively complex to use for design. sive because the tensile forces in the hoops will be closer to
Furthermore, it is unable to predict how much concrete is the surface than the compressive forces from the shear flow.
lost after spalling took place, which, as seen in Fig. 1, is not In this situation, the cover is not expected to spall because
always the full cover. there are no tensile stresses in the concrete.
This paper attempts to improve design and analysis prac- Case 2 in Fig. 2(c) again shows the corner detail of a
tices for torsion by presenting a general methodology which member subjected to torsion. Here, the cover is thicker than
considers both compression softening and spalling. A central in Case 1 and hence, the compressive forces which make up
part of the methodology is a new model which derives the the shear flow will be closer to the outside surface than the
mechanism of torsional spalling from first principles. In tensile forces in the hoops. The diagonal force in the corner
addition to having the ability to predict when spalling occurs, will now be tensile, and if the resulting tensile stress exceeds
the model also considers and quantifies the resulting loss of the cracking stress of the concrete, diagonal cracking causing
concrete, which is something that has not been investigated spalling will occur. For a 90-degree corner with a uniform
@seismicisolation
@seismicisolation

Fig. 2—Model conditions (top and middle) and corner details if shear flow path is within (bottom left) or outside (bottom right)
hoop reinforcement.
cover thickness, the diagonal tensile force per unit length plane will be equal to scr × 0.5√2. If it is then assumed that
along the beam will be the tensile force in Eq. (1) acts uniformly over this plane,
the average tensile stress in the concrete at the moment of
_ A f cracking will be equal to

Ft = √ _
2 s (1)
h h
_ A f Ah fh

_ h h _ 1_ _
where fh is the stress in the hoop; and Ah and s are the ft = √
2 s ⋅ = 2 s crs (2)
0 . 5 √2 ⋅ s cr
area and spacing of the hoop reinforcement, respectively.
When spalling occurs, a diagonal failure plane will form
The average crack spacing can be calculated using the
at a 45-degree angle that intersects with the horizontal and
1978 CEB-FIP equation,26 though this value should not
vertical faces of the member, causing cracks to appear on the
exceed the smaller of the side lengths which meet at the
surface. If the distance between these cracks, measured along
corner, bi
the perimeter around the corner, is equal to the transverse
crack spacing, scr, then the length of the diagonal failure
db
scr = 2(c + 0 . 1s)+ k1 k2 _
ρh ≤ min{bi} (3)
@seismicisolation
@seismicisolation

In Eq. (3), c is the cover, measured from the outside
surface to the centerline of the hoops; k1 and k2 are factors
which account for bond and strain conditions (taken as 0.4
and 0.25, respectively)27; db is the bar diameter of the hoops;
and ρh is the quantity of hoop reinforcement. If pcp and Acp
are the perimeter and area defined by the unspalled dimen-
sions of the cross section, respectively, then a nominal value
of ρh can be taken as
Ah pcp
_

ρh = (4)
Acp
s
Having defined the tensile stress in the concrete due to the

applied loads, the cracking stress must now be determined
to predict when spalling takes place. The concrete, which
is triaxially stressed, will crack at a lower stress than under
uniaxial stress conditions because of the coexisting diagonal
compressive stresses which circulate around the member.
This phenomenon was noted by Kupfer et al.28 when
performing tests on biaxially loaded specimens and Foster
et al.29 when studying cover spalling in columns subjected to
axial compression. A simple equation that accounts for this
effect is a modified version of the model proposed by Kupfer
and Gerstle,30 which is shown as follows
= ft′(1 − 0.8
)
_ f2
fcr (5) Fig. 3—Simplified model for section loss following cover
f2,max
spalling.
Ah fh _____
In Eq. (5), fcr is the cracking strength; ft′ is the uniaxial 2 _ scrs ≥ 4 √ fc′ (1 − 0.8 α1); fc′ in psi (8b)
tensile strength; f2 is the accompanying principal compres-
sive stress; and f2,max is the peak compressive stress after
If Eq. (8) is satisfied and the centroid of the shear flow
considering compression softening, equal to βcsfc′. Note
falls outside of the hoop reinforcement, the corner of the
that unlike Kupfer and Gerstle’s original model, the soft-
cross section will spall off and no longer contribute to the
ened compressive strength is used instead of the uniaxial
member’s torsional resistance. An idealization of the spalled
compressive strength. This is because the concrete at the
member is shown in Fig. 3. The distance between the failure
corner is in a triaxial state of stress instead of being biaxially
planes that define the spalled concrete is the crack spacing,
loaded as in Kupfer et al.’s experiments.28
which matches the assumption used to derive Eq. (8), and
For a member subjected to torsion, f2—and hence, fcr—
within the cover, the failure plane is assumed to extend
will not be constant though the thickness of the cover.
along the centerline of the hoop reinforcement due to micro-
However, the average cracking stress, fcr,av, can be obtained
cracking. These L-shaped failure patterns at the corners of
by using the average compressive stress in the rectangular
the section lead to the following reductions to the overall
stress block, f2,av. Substituting the definitions of f2,av and f2,max
cross-section geometry
into Eq. (5) gives
− ∑(8 scr,i+ ci2) ≥ Aoh

n _1 2
= ft′(1 − 0.8  
βcsfc′ )
α1βcsfc′
_______ Asp
= Acp (9)
fcr,av   = ft′(1 − 0.8 α1) (6) i=1
1 _
psp = pcp− ∑scr,i(1 − 2 √ 2 ) ≥ ph
n _
The uniaxial tensile strength can be taken as (10)
i=1
_____
ft′ = 0.33 √fc′ ; fc′ in MPa (7a)
where Asp and psp are the reduced area and perimeter after
_____ spalling, respectively; and n is the number of corners that
ft′ = 4 √fc′ ; fc′ in psi (7b) have spalled off, which can be conservatively taken as four
for rectangular sections. Asp is generally larger than the area
Spalling will then occur when the applied stress exceeds enclosed by the hoop reinforcement, Aoh, but will approach
the cracking stress, which results in the following condition Aoh for members that have widely spaced cracks relative
when Eq. (2), (6), and (7) are combined to the size of the cross section. Note that Eq. (9) and (10)
assume that the failure plane forms at a 45-degree angle,
Ah fh _____ which is a reasonable assumption for members subjected to
2 _ scrs ≥ 0.33√fc′ (1 − 0.8 α1) ; fc′ in MPa (8a) pure torsion. Instances where the failure plane is unlikely
@seismicisolation
@seismicisolation

Fig. 4—Model summary.
to assume this shape include cases where the cover thick- expected to spall at high torsional stresses, while internal or
ness varies around the member’s perimeter, or if the shear re-entrant corners are expected to remain unspalled.
stresses do not remain constant as they turn the corner, such
as in the case of combined torsion and shear. IMPLEMENTATION IN NONLINEAR ANALYSIS
The proposed model, whose key equations are summa- To study the application of the proposed model within
rized in Fig. 4, suggests that spalling is strongly affected a nonlinear analysis framework, a modified version of the
by the crack control provided by the hoop reinforcement. DCFT, which can predict a member’s complete torque-twist
According to Eq. (8), spalling is less likely to occur in behavior, was developed that accounts for both compres-
lightly reinforced members because the cracks in such sion softening and corner spalling. Details of the original
members will be widely spaced as a result of their poor crack DCFT, which assumes full spalling and neglects both tensile
control. However, if spalling occurs, then the impact on the stresses in the cracked concrete and compression softening,
member’s torsional response will be significant because can be found elsewhere.7
more concrete is expected to be lost, as shown in Eq. (9) In addition to employing the proposed spalling model
and (10). Conversely, heavily reinforced members with good instead of assuming full spalling, the model differed from
crack control, and hence a small crack spacing, are more the DCFT in its choice of constitutive relationships for the
likely to spall. The impact of spalling on such members is concrete stress-strain behavior. To be applicable to both
expected to be relatively minor however, since less concrete normal-strength and high-strength concrete, a modified
will be lost. In the limiting case where the amount of rein- version of Popovics model27,32-34 was used as a base curve for
forcement provided results in a failure caused by crushing the compressive response of the concrete. The compression
before yielding, good predictions of strength can be obtained softening relationship suggested by the Modified Compres-
by neglecting spalling completely.31 These predictions are sion Field Theory10 (MCFT) was chosen, though alternative
consistent with experimental evidence, like the beams shown relationships by others could also have been used.35-39 The
in Fig. 1, and previous remarks about spalling. It is inter-
1,2
MCFT equation for βcs is
esting to note that although the CEB-FIP equation was not
originally intended to be used for torsional spalling, it gives _____________
1
βcs(t) =   
0.8 + 170 ε1(t) ≤ 1.0 (11)
reasonable predictions of how much concrete is lost after it
occurs. Consider the PT series of beams tested by Mitchell
and Collins,1 which are shown in Fig. 1. The predicted crack In Eq. (11), βcs is a compression-softening coefficient
spacings for PT4 and PT6, which had unspalled dimen- used to reduce the strength of the concrete, so f2,max = βcsfc′;
sions of 381 × 381 mm and 432 × 432 mm (15 × 15 in. and ε1 is the principal tensile strain; and t is the depth from the
17 ×17 in.), are 179 and 258 mm (7.0 and 10.2 in.), respec- outside surface of the cross section. ε1 can be calculated
tively. These values represent losses of approximately 47% using the following equation if positive values of the prin-
and 60% of the cover along each face, which has reasonable cipal compressive strain, ε2, indicate compression
agreement with the photos shown in Fig. 1.
It should be noted that—based on equilibrium of the ε1(t) = εl(t) + εh(t) + ε2(t) (12)
forces drawn in Fig. 2—spalling is only expected to occur
if the centerline of the shear flow acts within the clear cover. The degree of softening through the tube walls is not
However, the forces drawn in the figure are a simplification constant because ε1 varies through the thickness. Deter-
of the actual state of stress at the corners of the cross section. mining the variation of βcs therefore requires ε1—and by
In reality, the compressive stresses that make up the shear extension, εl, εh, and ε2—to be known at each point in the
flow are distributed over the depth of compression instead depth of compression, td. This can be done by revisiting
of only being concentrated at the centerline, so some tensile some of the assumptions employed by the DCFT in its orig-
stresses in the concrete will always be needed to turn the inal formulation. First, the longitudinal reinforcement is
corner, even if the resultant shear flow does not appear to assumed to be uniformly distributed around the perimeter of
require any. Therefore, all external or salient corners are the member, so it will remain straight as it twists and hence,
@seismicisolation
@seismicisolation

the longitudinal strain εl will not be a function of t. ε2 is Figure 5 shows the strength predictions obtained from
assumed to vary linearly from a maximum on the surface each modeling approach plotted against the normalized
to a value of zero at the depth of compression, as shown cover. The spalled-only approach made reasonable strength
in Fig. 2(b), and this variation can be determined from the predictions over the full range of normalized cover values.
twist of the member.7 Finally, the transverse strains, εh, can Most of the predictions were conservative and had test-to-
be found using the assumption that the angle of the diag- predicted ratios which fell between 1.0 and 2.0. The strength
onal compression field, θ, remains constant throughout the predictions made using the softened-only approach were
depth of compression td. This is done using the following on average more accurate than those obtained from the
relationship between θ and the longitudinal, transverse, and spalled-only method. However, the number of specimens
principal compressive strains whose strength was overestimated increased from 26.2
to 69%, and the predictions were increasingly unconser-
εl(t)+ ε2(t)
_ vative as the normalized cover increased. The test-to-pre-
εh(t) = − ε2(t); 0 ≤ t ≤ td (13)
tan2θ dicted ratios of the eight members with thick covers—that
is, those exceeding a normalized cover of 50%—were all
With the principal tensile and principal compressive less than 0.65, which is less than the reduction factor used
strains known at all locations in the tube wall, the principal by ACI 318-19 for torsion, ϕ = 0.75, and therefore unsafe.
compressive stresses in the concrete f2 can be obtained using Accounting for both compression softening and full cover
the Popovics model and Eq. (11). The stress block factors, α1 spalling gave less-accurate but more conservative predic-
and β1, and the average value of the compression softening tions, with an average test-to-predicted ratio of 1.41
coefficient, βcs, can then be determined for any given strain compared to 1.12 and 0.93 made by the first two approaches,
state by integrating the stress distribution so that the equiv- respectively. The spread of the predictions was much larger
alent rectangular stress block shares the same net force and than the first two approaches. The proposed model gave the
line of action of the actual stress distribution.27 most accurate predictions of the four methods studied, and
In determining a member’s torque-twist behaviour, the unlike the softened-only approach, maintained its accuracy
DCFT calculates the stress block factors, the stresses in the as the normalized cover increased.
hoop reinforcement, fh, and the depth of the shear flow path, The trends observed in Fig. 5 can also be seen in the test-
ao, at each load step. Using this information, spalling was to-predicted summary statistics shown in Table 1. Based on
considered using Eq. (8) and checking if the centerline of the statistics obtained using the full dataset, the proposed
the shear flow path was inside of the clear cover after each method made the most accurate predictions of strength
converged load step. If both conditions were satisfied, then (average test-to-predicted ratio of 1.07) and had the smallest
the area and perimeter of the member were reduced using coefficient of variation (COV) and spread. Table 1 also
Eq. (9) and (10) for the remainder of the analysis. contains statistical parameters which were generated by
fitting a normal distribution to the lower half of the test-
Experimental validation to-predicted data, which are the tests of interest from the
The numerical model was validated using 187 tests of perspective of safety. For methods one and three, the lower-
nonprestressed beams subjected to pure torsion.1,40-52 Each half analyses yielded better average test-to-predicted values
of the beams considered in the study had a rectangular and coefficients of variation than the corresponding values
cross section which varied in size from 160 × 275 to 600 obtained from the full dataset. This indicates that strength
× 600 mm (6.3 × 10.8 to 23.6 × 23.6 in.). The normalized predictions made by the fully spalled methods were not
cover, defined as the ratio of the cover concrete to the total symmetrically distributed, but were instead biased towards
cross-sectional area or c×pcp/Acp, varied from 7 to 83%, the conservative side due to several test-to-predicted ratios
and the concrete cylinder compressive strength varied from exceeding 2.0. The opposite was true for the softened-only
2080 to 16,000 psi (14.3 to 110 MPa). The specimens in the method, which saw an increase in the COV from 17.4 to
dataset covered a range of failure modes, including yielding 20.3%. The predictions made using the proposed approach
of the reinforcement, crushing of the concrete before any had comparable average test-to-predicted values and coef-
yielding, or spalling following initial cracking. ficients of variation for both statistical analyses, indicating
Four modeling approaches were taken to study how that the test-to-predicted values were distributed with low
different assumptions about compression softening and cover skew. Examining the first percentile values, which approxi-
spalling would affect the strength predictions of these beams. mate the required strength reduction factor to offset the 1%
The first was a “spalled-only approach,” which neglected likelihood of failure,53 shows that the proposed method with
compression softening and assumed that the full cover was corner spalling and compression softening is as safe as the
spalled during the entire analysis. The second was a “soft- spalled-only approach despite being 5% less conservative on
ened-only approach,” which employed the MCFT compres- average.
sion-softening relationship but ignored cover spalling. The The role of the proposed corner spalling model on the
third approach considered both full cover spalling and predicted torque-twist response can be seen in Fig. 6,
compression softening. Finally, the fourth approach used the which shows the observed and predicted behavior of two
proposed framework by considering compression softening companion beams tested at the University of Toronto. These
and the suggested corner spalling model. beams, whose appearances after failure are shown in Fig. 1,
contained identical reinforcement cages but had different
@seismicisolation
@seismicisolation

Fig. 5—Summary of strength predictions obtained from nonlinear analyses.
Table 1—Summary of analysis results: nonlinear models
Full dataset
Test/Pred Spalled only Softened only Softened and spalled Proposed
Count 187 187 187 187
Mean 1.12 0.93 1.41 1.07
COV 18.3% 17.4% 19.9% 14.6%
Maximum 1.80 1.30 2.56 1.48
Minimum 0.69 0.40 0.83 0.74
% Tests < 1.0 26.2% 69.0% 2.7% 34.8%
Lower half of dataset only
Mean 1.10 0.96 1.35 1.05
COV 15.6% 20.3% 13.5% 13.7%
First percentile 0.70 0.51 0.92 0.72
cover thicknesses. Despite the thicker cover, PT6 failed especially well. It should be noted the initial stiffness of both
at the same torque as PT5 and exhibited severe spalling at beams was underestimated because tensile stresses in the
failure. The proposed model correctly predicted that PT5 cracked concrete (that is, tension stiffening) were neglected
did not experience spalling. This is because the shear flow in the analysis.
was contained within the hoop reinforcement for the entire The results of a more detailed study comparing the four
analysis, so diagonal tension did not develop at the corners. modeling approaches against a broader variety of variables
The ultimate strength and ductility were somewhat under- (such as concrete strength, reinforcement strength, aspect
estimated, though this may be because the confinement ratio, and so on) can be found in the first author’s doctorate
provided by the hoop reinforcement was not considered thesis.54
when modeling the concrete’s compressive response. The
ultimate strength and torque-twist response of PT6 were SIMPLIFICATION FOR DESIGN
predicted with excellent accuracy. Like the experiment, the Although the proposed spalling model improves the
cover was predicted to spall, and the calculated post-spalling predictive capabilities of the DCFT, its full formula-
response followed the experimentally observed behavior tion is too complex to use in design. In this section, a
@seismicisolation
@seismicisolation

relationship, which is shown in Fig. 7. The proposed expres-
sion for fcr then becomes
fcr = 0.02fc′ (17)
It should be emphasized that Eq. (17) represents the

cracking stress under biaxial conditions when the concrete
is also subjected to significant coexisting compressive
stresses. As shown in Fig. 7, the cracking stress predicted by
Eq. (17) is typically smaller than the uniaxial tensile strength
predicted by Eq. (7).
At ultimate limit states, spalling must be considered when
the applied torque exceeds a threshold value, Tspall. Tspall can
be determined by equating Eq. (15) and (17), approximating
Ao as 0.85Acp, and taking θ as 45 degrees for nonprestressed
members
Tspall = 0.017Acpscrfc′ (18)
Equation (18) can be simplified further by noting that

Fig. 6—Modeling PT5 and PT6 beams. Eq. (7) can underestimate the tensile strength of concrete,
especially when the tensile stresses are concentrated in a
simplified version, which retains the essential characteristics
small volume of concrete,55 which is the case when dealing
of the general model, is derived, and its application within
with cover concrete in the corner of a cross section. There-
a modified version of the ACI 318-19 torsion procedures is
fore, the suggested expression for Tspall becomes
demonstrated.
The condition for spalling given in Eq. (8) can be rewritten
Tspall = 0.02Acpscrfc′ (19)
in terms of the applied torsion, Tu, if it is assumed that there
are no tensile stresses in the cracked concrete. This allows
For design purposes, using Eq. (3) to determine scr is
the torque to be written in terms of the stresses carried by
inconvenient because it requires the arrangement of the
the hoop reinforcement fh, the area enclosed by the shear
transverse reinforcement to be known in advance. An
flow path Ao, and the angle of inclination of the diagonal
approximate value for design can be obtained by considering
compression θ
that the maximum spacing of hoop reinforcement allowed by
ACI 318-19 is 305 mm (12 in.), and the maximum spacing
Ah fh
Tu = 2 Ao _
s cot θ (14) of longitudinal torsional reinforcement around the perimeter
is also 305 mm (12 in.). The diagonal crack spacing, sθ, can
Substituting Eq. (14) into Eq. (2) and rearranging terms then be taken as half the maximum reinforcement spacing
allows the tensile stress which initiates spalling, ft, to be as done in the CSA A23.3:19 shear provisions,5 giving a
expressed in terms of the applied torque value of sθ = 152 mm (6 in.). The diagonal crack spacing
can be related to the longitudinal crack spacing, sx, and the
_ Tu transverse crack spacing, sy, using the following equation

ft = A s tan θ (15)
o cr suggested by Vecchio and Collins10
As before, spalling will occur when this tensile stress is _ 1

sθ =
equal to the cracking stress of the concrete, fcr. The cracking _
sin θ _ cos θ
sx + sy (20)
stress can be determined using Kupfer et al.’s model and a
suitable value of α1. Kuan et al.31 derived an expression for
α1 corresponding to a principal compressive strain of 0.003 If it is assumed that θ = 45 degrees, sθ = 152 mm (6 in.), and
on the surface of the cross section, which is appropriate for both sx and sy are equal to scr, then scr will equal to 216 mm
ultimate strength calculations. Their expression is (8.5 in.). Modifying this base value to include the influence
of the cover thickness on scr in the same way as the CEB-FIP
α1 = 1.0 – 0.005fc′; fc′ in MPa (16a) equation gives the following simple expression for scr
_ scr = 200 + 2c ≤ min{bi}; c in mm (21a)

α1 = 1.0 − 0.035 fcʹ ; fc′ in psi
(16b)
1000
scr = 8 + 2c ≤ min{bi}; c in in. (21b)
Substituting Eq. (16) into Eq. (8) allows the cracking stress
to be expressed solely in terms of fc′. fcr can be expressed When the applied torque exceeds Tspall, the spalled geom-
in an even simpler manner by linearizing the resulting etry should be obtained using Eq. (9) and (10). Although
@seismicisolation
@seismicisolation

fails by yielding, the longitudinal strain and hoop strain can
be taken as 0.002 for 400 MPa (60 ksi) steel. If the principal
compressive strain in the concrete on the surface equals
to 0.003 at failure, then the principal tensile strain can be
conservatively taken as 0.007, which results in a value of
βcs = 0.50 if the MCFT compression softening relation-
ship given in Eq. (11) is used. Substituting these terms into
Eq. (23) gives
___________
( √ 2 )
_Asp
_ Tu psp

ao = psp 1 − 1 − 4 (24)
α1fcʹ Asp
In Eq. (24), the stress block factor α1 is calculated using

Eq. (16), and if the terms under the square root sign result in
a negative number, ao should be taken as ao = Asp/psp. Once
ao has been found, then Ao and po can be calculated as
_1

Ao = Asp
− 2 pspao (25)
po = psp – 4ao (26)

Fig. 7—Derivation of simplified cracking stress equation.
ACI 318-19 also includes a third equation for torsional
the general model also requires the shear flow to act within strength to account for crushing or excessive crack widths.
the cover to cause spalling, a reasonable simplification for In the absence of an applied shear force, this equation is
design is to neglect this check and use the partially spalled
geometry whenever the applied torsion exceeds Tspall. Tu ph _
_

2 ≤ 0.83 √ fcʹ ; fc′ in MPa (27a)
1.7Aoh
Implementation in design procedures
The basic equations for torsional strength given by ACI Tu ph _
_
318-19 which correspond to yielding of the hoop reinforce-
2 ≤ 10 √ fc′; fc′ in psi (27b)
1.7Aoh
ment and longitudinal reinforcement, respectively, are
When using this equation within the proposed framework,
_ ⎧
Ah fyh

⎪
2 Ao s cot θ Aoh and ph can be replaced with Asp and psp if Tu exceeds Tspall.
Tn = min⎨  
Alfyl (22) If Tu is less than Tspall, then each instance of Asp and psp in
⎪ _
⎩ 2 Ao po tan θ Eq. (23) to (27) can be replaced with the unspalled geometry,
Acp and pcp.
where po is the perimeter of the shear flow path; and fyh and To determine the torsional strength of a member using the
fyl are the yield strengths of the hoop and longitudinal rein- proposed design procedure, the following steps should be
forcement, respectively. ACI 318-19 defines po = ph, Ao = followed:
0.85Aoh, and suggests a value of θ = 45 degrees for nonpre- 1. Determine Tspall using Eq. (19) and (21). If the applied
stressed members. torque Tu exceeds this value, determine Asp and psp using
Using the model presented in this paper, Ao and po can Eq. (9) and (10), respectively.
be determined by considering corner spalling, equilibrium 2. Obtain the depth of the stress block, ao, using Eq. (24)
conditions, and compression softening instead of simply and calculate Ao and po using Eq. (25) and (26).
assuming Ao = 0.85Aoh and po = ph, which is currently done 3. The torsional strength, Tn, is the smallest value obtained
in ACI 318-19. To determine Ao and po from Asp and psp, the from Eq. (22) and (27). When using Eq. (27) to check
width of the equivalent rectangular stress block, ao, must be concrete crushing or excessive cracking, the relevant modi-
known. Its value can be obtained if the equation previously fications described in the previous paragraph shall be used.
derived by Collins and Mitchell9,27 is modified to account for
corner spalling and compression softening Experimental validation
The predictive ability of the proposed design method was
_______________________ evaluated using the same dataset of beams that was used to
( √ )
_Asp
_ Tu psp
ao = psp 1 −    
1− (tan θ + cot θ) (23) validate the spalling model within the nonlinear analysis
α1βcsfcʹ Asp
2 framework. Figure 8 shows the strength predictions obtained
using ACI 318-19, the proposed method where scr is calcu-
Equation (23) can be simplified by introducing a series of lated using Eq. (3), and the proposed method if scr is calcu-
assumptions about the strain state of the member at failure. lated using Eq. (21). Again, the predictions are plotted with
Like before, θ can be taken as 45 degrees, and if the member respect to the normalized cover. The ACI predictions are
@seismicisolation
@seismicisolation

Fig. 8—Summary of strength predictions obtained from design procedures.
generally conservative across the full range of cover thick- Table 2—Summary of analysis results: design
nesses, and 23 tests have a test-to-predicted ratio of 2.0 or procedures
higher. The predictions obtained using the proposed design Full dataset
methods are generally more accurate, but can be unconser-
Proposed— Proposed—
vative for very thick covers if the CEB-FIP equation is used
Test/Pred ACI 318-19 CEB-FIP scr Simplified scr
to calculate scr. Using the simpler equation for scr generally
gives more conservative predictions than the CEB-FIP equa- Count 187 187 187
tion, with fewer test-to-predict ratios below 1.0, but is still Mean 1.54 1.24 1.43
more accurate than using ACI 318-19. COV 27.9% 19.2% 18.6%
Table 2 shows the summary statistics obtained using the
Maximum 3.00 1.99 2.20
three design methods represented in Fig. 8. Like before,
two sets of statistics, one corresponding to the full dataset Minimum 0.85 0.74 0.84
and another corresponding to a normal distribution fitted to % Tests < 1.0 2.7% 12.3% 2.7%
the lower half of the test-to-predicted data, are shown. The Lower half of dataset only
proposed method with the CEB-FIP equation to calculate
Mean 1.42 1.18 1.40
scr made significantly better predictions than ACI 318-19,
with the average test-to-predicted ratio and COV improving COV 16.6% 13.7% 16.4%
from 1.54 to 1.24 and 27.9 to 19.2%, respectively, when First percentile 0.87 0.81 0.86
considering the full dataset. Using the simpler equation to
calculate scr gave comparable statistics to the more complex increasingly unconservative as the cover thickness increases,
equation, with the mean test-to-predicted value being higher and are unsafe when applied to members whose normalized
at 1.43, but the COV being 0.6% smaller. Despite the cover exceeds 50%.
proposed methods being less conservative than ACI 318-19 2. Considering both full cover spalling and compression
on average, the lower-half analyses resulted in comparable softening is conservative but gives less accurate strength
first percentile values for all three methods. This indicates predictions than traditional spalled and softened methods.
that the proposed methods offer a similar margin of safety This suggests that quantifying the loss of the cover concrete
against failure than the current code, and because they give at the corner, rather than assuming that the full cover is lost
first-percentile values exceeding the reduction factor used following spalling, is more appropriate when compression
by ACI 318-19 for torsion, ϕ = 0.75, are also compatible softening is also considered.
with the current ACI design framework. 3. The proposed spalling model predicts that the crack
control provided by the transverse reinforcement plays an
CONCLUSIONS important role in determining when spalling occurs and how
In this paper, a model was presented that can predict much concrete is lost. Based on equilibrium and experi-
when torsional spalling occurs and quantify its effect on mental evidence, spalling is assumed to be concentrated at
a member’s strength and stiffness. The model was imple- the corners of the cross section.
mented into a nonlinear analysis framework and a design 4. Although it was originally derived for other applica-
procedure and then validated using a large database of pure tions, the 1978 CEP-FIP crack spacing equation makes good
torsion experiments. predictions when used in the proposed spalling model for
The main findings from this study are: torsional spalling.
1. Modeling approaches that consider compression soft- 5. Nonlinear analysis tools and design procedures that use
ening but not spalling are generally more accurate those that the proposed spalling model and the Modified Compression
consider spalling but not softening. However, they become
@seismicisolation
@seismicisolation

Field Theory compression-softening relationship make T = tension force in hoop reinforcement as result of torsion; torsion
Tn = nominal torsional strength
better strength predictions than existing alternatives. Tspall = threshold torsion above which corner spalling must be
Future work should investigate how spalling propagates considered
into the side cover, a phenomenon which appears to be Tu = factored torsional demand at section
t = distance through thickness of equivalent tube
caused by a different mechanism than the corner spalling td = depth of compression within equivalent tube
discussed in this paper. Experiments examining the influ- α1 = ratio of average unsoftened stress in rectangular compression
ence of member size on spalling should also be investigated, block to concrete cylinder strength
β1 = ratio of depth of rectangular compression block to thickness of
as the proposed model suggests that the impact of spalling equivalent tube
becomes smaller as the overall size gets larger. Finally, βcs = factor accounting for compression-softening effects
spalling due to combined moment, shear, and torsion should εc′ = concrete strain corresponding to peak compressive stress
εh = strain in transverse direction
also be investigated. εl = strain in longitudinal direction
ε1 = principal tensile strain
AUTHOR BIOS ε2 = principal compressive strain
Allan Kuan is a Structural Designer at RJC Engineers, Toronto, ON, ϕ = ACI 318-19 reduction factor for torsion; ϕ = 0.75.
Canada. He received his BASc in engineering science in 2015 and his PhD θ = angle of inclination of diagonal compressive stresses to longitu-
in civil engineering in 2022 from the University of Toronto, Toronto, ON, dinal axis of member
Canada. His research interests include shear and torsion in reinforced and ρh = nominal quantity of transverse torsional reinforcement.
prestressed concrete structures.
Evan C. Bentz, FACI, is Professor and Associate Chair (undergrad) of

REFERENCES
1. Mitchell, D., “The Behaviour of Structural Concrete Beams in Pure
civil engineering at the University of Toronto. He received his BASc from
Torsion,” PhD thesis, Department of Civil Engineering, University of
the University of Waterloo, Waterloo, ON, Canada, in 1994, and his PhD
Toronto, Toronto, ON, Canada, 1974, 140 pp.
from the University of Toronto in 2000. He is a member of Joint ACI-ASCE
2. Rahal, K. N., and Collins, M. P., “Effect of Thickness of Concrete
Committee 445, Shear and Torsion, and is past Chair of ACI Committee
Cover on the Shear-Torsion Interaction—An Experimental Investigation,”
365, Service Life Modeling. He received the ACI Chester Paul Seiss Award
ACI Structural Journal, V. 92, No. 3, May-June 1995, pp. 334-342.
in 2018 and the ACI Design Award twice.
3. Mitchell, D., and Collins, M. P., “Detailing for Torsion,” ACI Journal
Proceedings, V. 73, No. 9, Sept. 1976, pp. 506-511.
ACI Honorary Member Michael P. Collins is a University Professor Emer-
4. ACI Committee 318, “Building Code Requirements for Structural
itus of structural engineering at the University of Toronto, where he has led
Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved
a long-term research project aimed at developing rational but simple shear
2022),” American Concrete Institute, Farmington Hills, MI, 2019, 623 pp.
and torsion design procedures for both reinforced and prestressed concrete
5. CSA A23.3:19, “Design of Concrete Structures,” Canadian Standards
members. The results of this work have influenced design provisions for
Association, Mississauga, ON, Canada, 2019, 295 pp.
buildings, bridges, nuclear containment structures, and offshore concrete
6. AASHTO, “AASHTO LRFD Bridge Design Specifications and
platforms. He has six awards from ACI.
Commentary,” ninth edition, American Association of State Highway and
Transportation Officials, Washington, DC, 2020, 1912 pp.
NOTATION 7. Mitchell, D., and Collins, M. P., “Diagonal Compression Field Theo-
Acp = area enclosed by outside perimeter of concrete cross section ry—A Rational Model for Structural Concrete in Pure Torsion,” ACI
Ah = area of one leg of transverse torsional (hoop) reinforcement Journal Proceedings, V. 71, No. 8, Aug. 1974, pp. 396-408.
Al = total area of longitudinal torsional reinforcement 8. MacGregor, J. G., and Ghoneim, M. G., “Design for Torsion,” ACI
Ao = area enclosed by shear flow path Structural Journal, V. 92, No. 2, Mar.-Apr. 1995, pp. 211-218.
Aoh = area enclosed by centerline of closed transverse reinforcement 9. Collins, M. P., and Mitchell, D., “Shear and Torsion Design of
Asp = area enclosed by outside perimeter of concrete cross section Prestressed and Non-Prestressed Concrete Beams,” PCI Journal, V. 25,
after considering corner spalling No. 5, 1980, pp. 32-100. doi: 10.15554/pcij.09011980.32.100
ao = depth of equivalent rectangular stress block/equivalent tube 10. Vecchio, F. J., and Collins, M. P., “The Modified Compression Field
thickness Theory for Reinforced Concrete Elements Subjected to Shear,” ACI Journal
bi = shorter of two side lengths which meet at i-th corner of cross Proceedings, V. 83, No. 2, Mar.-Apr. 1986, pp. 219-231.
section 11. Hsu, T. T. C., and Mo, Y. L., “Softening of Concrete in Torsional
C = compression force in concrete making up shear flow due to Members – Theory and Tests,” ACI Journal Proceedings, V. 82, No. 3,
tension May-June 1985, pp. 290-303.
c = cover, measured from outside surface to centerline of hoop 12. PCI Design Handbook—Precast and Prestressed Concrete, eighth
reinforcement edition, Precast/Prestressed Concrete Institute, Chicago, IL, 2017.
db = bar diameter of hoop reinforcement 13. Zia, P., and Hsu, T. C., “Design for Torsion and Shear in Prestressed
fc′ = specified compressive strength of concrete Concrete Flexural Members,” PCI Journal, V. 49, No. 3, 2004, pp. 34-42.
fcr = cracking strength of concrete doi: 10.15554/pcij.05012004.34.42
fcr,av = average concrete cracking strength within tube thickness 14. Lucier, G.; Walter, C.; Rizkalla, S.; Zia, P.; and Klein, G., “Devel-
fh = stress in transverse (hoop) reinforcement opment of a Rational Design Methodology for Precast Concrete Slender
ft = tensile stress in concrete Spandrel Beams: Part 2, Analysis and Design Guidelines,” PCI Journal,
ft′ = uniaxial tensile strength of concrete V. 56, No. 4, 2011, pp. 106-133. doi: 10.15554/pcij.09012011.106.133
fyh = specified yield strength of transverse reinforcement 15. Rahal, K. N., “Torsional Strength of Normal and High Strength Rein-
fyl = specified yield strength of longitudinal reinforcement forced Concrete Beams,” Engineering Structures, V. 56, 2013, pp. 2206-
f2 = principal compressive stress in concrete 2216. doi: 10.1016/j.engstruct.2013.09.005
f2,av = Average concrete principal compressive stress within tube 16. Rahal, K. N., and Collins, M. P., “Simple Model for Predicting
thickness Torsional Strength of Reinforced and Prestressed Concrete Sections,” ACI
n = number of spalled corners Structural Journal, V. 93, No. 6, Nov.-Dec. 1996, pp. 658-666.
pcp = outside perimeter of concrete cross section 17. Rahal, K. N., “A Unified Approach to Shear and Torsion in Rein-
ph = perimeter of centerline of closed transverse torsion reinforcement forced Concrete,” Structural Engineering and Mechanics, V. 77, No. 5,
po = perimeter of shear flow path Mar. 2021, pp. 691-703.
psp = outside perimeter of concrete cross section after considering 18. Jeng, C. H., and Hsu, T. T. C., “A Softened Membrane Model for
corner spalling Torsion in Reinforced Concrete Members,” Engineering Structures, V. 31,
q = shear flow as result of applied torque No. 9, 2009, pp. 1944-1954. doi: 10.1016/j.engstruct.2009.02.038
s = spacing of transverse torsional reinforcement 19. Kuan, A.; Bruun, E. P. G.; Bentz, E. C.; and Collins, M. P., “Nonlinear
scr = average crack spacing in transverse direction Sectional Analysis of Reinforced Concrete Beams and Shells Subjected to
sx = average crack spacing in longitudinal direction Pure Torsion,” Computers & Structures, V. 222, Oct. 2019, pp. 118-132.
sy = average crack spacing in transverse direction doi: 10.1016/j.compstruc.2019.07.001
@seismicisolation
@seismicisolation

20. Vecchio, F. J., and Selby, R. G., “Towards Compression-Field 38. Miyahara, T.; Kawakami, T.; and Maekawa, K., “Nonlinear Behavior
Analysis of Reinforced Concrete Solids,” Journal of Structural Engi- of Cracked Reinforced Concrete Plate Element Under Uniaxial Compres-
neering, ASCE, V. 117, No. 6, 1991, pp. 1740-1758. doi: 10.1061/ sion,” Concrete Library International, V. 11, 1988, pp. 306-319.
(ASCE)0733-9445(1991)117:6(1740) 39. Kollegger, J., and Mehlhor, G., “Experimentelle Untersuchungen zur
21. Valipour, H. R., and Foster, S. J., “Nonlinear Reinforced Concrete Bestimmung der Druckfestigkeit des gerissenen Stahlbetons bei einer Quer-
Frame Element with Torsion,” Engineering Structures, V. 32, No. 4, 2010, zugbeanspruchung,” Report 413, Deutscher Ausschuss Für Stahlbeton,
pp. 988-1002. doi: 10.1016/j.engstruct.2009.12.026 Berlin, Germany, 1990.
22. Nguyen, T.-A.; Nguyen, Q.-H.; and Somja, H., “Nonlinear Analysis 40. Bernardo, L. F. A., and Lopes, S. M. R., “Torsion in High-Strength
of RC Members Subjected to Combined Bending–Shear–Torsion Stresses: Concrete Hollow Beams: Strength and Ductility Analysis,” ACI Structural
A Numerical Multi-Fiber Displacement-Based Finite Element Model with Journal, V. 106, No. 1, Jan.-Feb. 2009, pp. 39-48.
Warping,” Acta Mechanica, V. 232, No. 7, 2021, pp. 2635-2658. doi: 41. Fang, I. K., and Shiau, J. K., “Torsional Behavior of Normal- and
10.1007/s00707-021-02966-x High-Strength Concrete Beams,” ACI Structural Journal, V. 101, No. 3,
23. Bairan Garcia, J. M., and Mari, A. R., “Coupled Model for the Nonlinear May-June 2004, pp. 304-313.
Analysis of Anisotropic Sections Subjected to General 3D Loading. Part 2: 42. Hsu, T. T. C., “Torsion of Structural Concrete—Behavior of Rein-
Implementation and Validation,” Computers & Structures, V. 84, No. 31-32, forced Concrete Rectangular Members,” Torsion of Structural Concrete,
2006, pp. 2264-2276. doi: 10.1016/j.compstruc.2006.08.035 SP-18, American Concrete Institute, Farmington Hills, MI, 1968,
24. Ibrahim, M. S., and Gebreyouhannes, E., “Spalling of Concrete pp. 261-306.
Cover due to Torsion: Examination of Advanced Theoretical Models,” 43. Jeng, C. H.; Peng, S. F.; Chiu, H. J.; and Hsiao, C. K., “New Torsion
Proceedings of the fib Symposium 2020, Nov. 2020, pp. 694-701. Experiment on large-Sized Hollow Reinforced Concrete Beams,” ACI
25. Rahal, K. N., and Collins, M. P., “Analysis of Sections Subjected Structural Journal, V. 111, No. 6, Nov.-Dec. 2014, pp. 1469-1480. doi:
to Combined Shear and Torsion – A Theoretical Model,” ACI Structural 10.14359/51687166
Journal, V. 92, No. 44, July-Aug. 1995, pp. 459-469. 44. Koutchoukali, N. E., and Belarbi, A., “Torsion of High-Strength
26. CEB-FIP, 1978, “Model Code for Concrete Structures,” CEB-FIP Reinforced Concrete Beams and Minimum Reinforcement Requirement,”
International Recommendations, third edition, Comité Euro-International ACI Structural Journal, V. 98, No. 4, July-Aug. 2001, pp. 462-469.
du Béton, Paris, France, 348 pp. 45. Lampert, P., and Thürlimann, B., “Torsionsversuche an Stahlbeton-
27. Collins, M. P., and Mitchell, D., Prestressed Concrete Structures, balken,” Institut fur Baustatik, ETH Zurich, Germany, January 1968.
Prentice Hall, Englewood Cliffs, NJ, 1991, 766 pp. 46. McMullen, A. E., and Rangan, B. V., “Pure Torsion in Rectangular
28. Kupfer, H. B.; Hilsdorf, H. K.; and Rusch, H., “Behavior of Concrete Sections—A Re-Examination,” ACI Journal Proceedings, V. 75, No. 10,
Under Biaxial Stresses,” ACI Journal Proceedings, V. 66, No. 8, Aug. 1969, Oct. 1978, pp. 511-519.
pp. 656-666. 47. Rasmussen, L. J., and Baker, G., “Torsion in Reinforced Normal and
29. Foster, S. J.; Liu, J.; and Sheikh, S. A., “Cover Spalling in HSC High-Strength Concrete Beams – Part 1: Experimental Test Series,” ACI
Columns Loaded in Concentric Compression,” Journal of Structural Structural Journal, V. 92, No. 1, Jan.-Feb. 1995, pp. 56-62.
Engineering, ASCE, V. 124, No. 12, 1998, pp. 1431-1437. doi: 10.1061/ 48. Chiu, H. J.; Fang, I. K.; Young, W. T.; and Shiau, J. K., “Behavior
(ASCE)0733-9445(1998)124:12(1431) of Reinforced Concrete Beams with Minimum Torsional Reinforcement,”
30. Kupfer, H. B., and Gerstle, K. H., “Behaviour of Concrete under Engineering Structures, V. 29, No. 9, 2007, pp. 2193-2205. doi: 10.1016/j.
Biaxial Stresses,” Journal of Engineering Mechanics, V. 99, No. 4, 1973, engstruct.2006.11.004
pp. 853-866. 49. Peng, X. N., and Wong, Y. L., “Behavior of Reinforced Concrete
31. Kuan, A.; Bentz, E. C.; and Collins, M. P., “Torsional Strength of Walls Subjected to Monotonic Pure Torsion—An Experimental Study,”
Reinforced Concrete Members Failing by Crushing,” ACI Structural Engineering Structures, V. 33, No. 9, 2011, pp. 2495-2508. doi: 10.1016/j.
Journal, V. 119, No. 6, Nov. 2022, pp. 233-244. doi: 10.14359/51736115 engstruct.2011.04.022
32. Popovics, S., “A Numerical Approach to the Complete Stress-Strain 50. Joh, C.; Kwahk, I.; Lee, J.; Yang, I. H.; and Kim, B. S., “Torsional
Curve of Concrete,” Cement and Concrete Research, V. 3, No. 5, 1973, Behavior of High-Strength Concrete Beams with Minimum Reinforcement
pp. 583-599. doi: 10.1016/0008-8846(73)90096-3 Ratio,” Advances in Civil Engineering, V. 2019, Jan, 2019, p. 1432697. doi:
33. Thorenfeldt, E.; Tomaszewics, A.; and Jensen, J. J., “Mechanical 10.1155/2019/1432697
Properties of High-Strength Concrete and Application in Design,” Proceed- 51. Ibrahim, M. S.; Gebreyouhannes, E.; Muhdin, A.; and Gebre, A.,
ings of the Symposium Utilization of High Strength Concrete, Tapir, Trond- “Effect of Concrete Cover on the Pure Torsional Behavior of Reinforced
heim, Norway, 1987, pp. 149-159. Concrete Beams,” Engineering Structures, V. 216, Aug, 2020, p. 110790.
34. Collins, M. P., and Porasz, A., “Shear Design for High Strength doi: 10.1016/j.engstruct.2020.110790
Concrete,” CEB Bulletin d’Information, V. 193, Dec. 1989, pp. 77-83. 52. Nagataki, S.; Okamoto, T.; and Lee, S. H., “A Study on Mechanism
35. Belarbi, A., and Tsu, T. T. C., “Constitutive Laws of Reinforced of Torsional Resistance of Reinforced Concrete Members,” JSCE Transac-
Concrete in Biaxial Tension-Compression,” Research Report UHCEE 91-2, tions, V. 1988, No. 390, 1988, pp. 179-188. doi: 10.2208/jscej.1988.390_179
University of Houston, Houston, TX, 1991. 53. MacGregor, J. G., “Safety and Limit States Design for Reinforced
36. Vecchio, F. J., and Collins, M. P., “Compression Response Concrete,” Canadian Journal of Civil Engineering, V. 3, No. 4, 1976,
of Cracked Reinforced Concrete,” Journal of Structural Engi- pp. 484-513. doi: 10.1139/l76-055
neering, ASCE, V. 119, No. 12, 1993, pp. 3590-3610. doi: 10.1061/ 54. Kuan, A., “The Behaviour and Design of Reinforced and Prestressed
(ASCE)0733-9445(1993)119:12(3590) Concrete Members Subjected to Torsion,” PhD thesis, Department of Civil
37. Mikame, A.; Uchida, K.; and Noguchi, H., “A Study of Compressive & Mineral Engineering, University of Toronto, Toronto, ON, Canada, 2022,
Deterioration of Cracked Concrete,” Proceedings from the International 624 pp.
Workshop on Finite Element Analysis of Reinforced Concrete, Columbia 55. Bentz, E. C., “Empirical Modeling of Cracking in Reinforced
University, New York, NY, 1991. Concrete,” ACI Structural Journal, V. 116, No. 3, May 2019, pp. 233-242.
doi: 10.14359/51714476
@seismicisolation
@seismicisolation

Title No. 120-S86
Behavior of Concrete Bridge-Deck Slabs Reinforced with

Basalt Fiber-Reinforced Polymer and Steel Bars
by Yahia M. S. Ali, Xin Wang, Shui Liu, and Zhishen Wu
Recently, hybrid reinforcement by combining steel with fiber- govern the design of FRP-reinforced concrete (RC) members
reinforced polymer (FRP) bars has emerged as a new system in in most instances. ACI 440.1R-15 (ACI Committee 440
reinforced concrete (RC) constructions. This reinforcement system 2015) and CSA S806-12 (2017) permit using glass FRP
can effectively overcome the ductility and serviceability challenges (GFRP), carbon FRP (CFRP), and aramid FRP (AFRP) bars
of FRP-RC structures. A total of 11 full-scale bridge-deck slabs
in concrete constructions.
were constructed and tested. The test parameters were reinforce-
Basalt fibers have been introduced as a promising addition
ment type, ratio, arrangement, and slab thickness. Moreover, a
comparison between the experimental and predicted deflections to the current types of FRPs. Basalt FRP (BFRP) bars have
from design provisions was carried out to verify the efficiency of relatively greater strength and modulus, comparable costs,
the models for hybrid RC sections. Based on test results, hybrid and higher chemical resistance than GFRP bars (Wu et al.
RC slabs exhibited ductility leading to an ample warning before 2015). Thus, using BFRP bars with a relatively high elastic
failure rather than brittle shear failure observed for FRP-RC slabs. modulus, compared to GFRP bars, would significantly
In addition, hybrid RC slabs displayed good stiffness, service- decrease the amount of reinforcement required and reduce
ability, and load-carrying capacity. Furthermore, test results give the crack width (Elgabbas et al. 2016). Moreover, BFRP-RC
an average bond-dependent coefficient, kb, of 1.27, close to the beams exhibited acceptable deformability when investigated
1.2 recommended by ACI CODE-440.11-22. In addition, some in flexure and shear (Duic et al. 2018).
modifications were proposed to shear equations available in
different design codes to be valid for hybrid RC members without
LITERATURE REVIEW
shear reinforcement.
In FRP-RC members, deeper cracks reduce the contri-
Keywords: basalt fiber-reinforced polymer (BFRP) bar; concrete bridge; bution of uncracked concrete to the shear stress due to the
hybrid reinforcement; shear behavior. lower concrete depth in compression. Moreover, in the trans-
verse direction, FRP bars have lower strength and stiffness,
INTRODUCTION which led to wider cracks and lower aggregate interlock
Bridge-deck slabs are the most critical infrastructure and dowel action supplement to the tensile reinforcement
exposed to harsh environments (deicing salts, humidity, compared to that of an equivalent steel area (El-Sayed et al.
freezing-and-thawing cycles, and chlorides) that make these 2006a). Finally, the total shear strength of FRP-RC members
structures very susceptible to corrosion of steel reinforce- is lower than that of steel-RC members. However, traditional
ment. The associated deterioration can accelerate such failure stirrups are not feasible for constructing slab bridges; conse-
or reduce the expected life span of the structure. Within the quently, the mode of failure may be dominated by shear
past two decades, the most effective way to diminish mainte- (Abdul-Salam et al. 2016). Slabs reinforced with GFRP
nance costs and extend the life span of structures has been to or CFRP bars failed in diagonal tension failure, while the
use fiber-reinforced polymer (FRP) composites as an alter- steel-reinforced slabs failed in ductile flexure mode by steel
native to traditional steel reinforcement in structural compo- yielding followed by concrete crushing (El-Salakawy and
nents, especially where steel corrosion is a major concern. Benmokrane 2004). In addition, increasing the reinforce-
In addition to corrosion resistance, FRP composites have ment ratio significantly improved the shear strength and the
many characteristics over steel reinforcement, such as a post-cracking stiffness (El-Sayed et al. 2005; Matta et al.
high strength-to-weight ratio, excellent fatigue resistance, 2013). Thus, the reinforcement type and its axial stiffness
and nonmagnetic and nonconductive nature, which can be can be confirmed to have a pronounced effect on the shear
used in harsh environments for civil structures. Unfortu- strength of the RC sections.
nately, FRP composites have some drawbacks: a low elastic Gradual failure can be attained by using both FRP and steel
modulus compared to steel (Ef/Es = approximately 0.25) and reinforcements. Therefore, steel reinforcement improves the
linear-elastic behavior up to failure without presenting any ductility by the yielding of steel reinforcement and enhances
yielding plateau, which resulted in the brittle collapse of the serviceability by decreasing the deflection and crack width,
member (Goldston et al. 2016). Many investigations have while FRP reinforcement maintains the load-carrying
been performed to study the overall performance of concrete
members reinforced with FRP bars. FRP-reinforced slabs MS No. S-2022-357.R1, doi: 10.14359/51738840, received April 27, 2023, and
had larger deflections and wider crack widths and depths Institute. All rights reserved, including the making of copies unless permission is
compared to steel-reinforced slabs (Michaluk et al. 1998; obtained from the copyright proprietors. Pertinent discussion including author’s
Ferrier et al. 2015). Therefore, serviceability criteria often is received within four months of the paper’s print publication.
@seismicisolation
@seismicisolation

capacity even after the yielding of steel bars (Qin et al. SFCBs and BFRP bars as the main reinforcement exhibited
2017). In addition, hybrid reinforcement resulted in lower better serviceability and ductility compared to conventional
crack spacing and smaller widths (Aiello and Ombres 2002). hybrid RC beams (Yang et al. 2020).
From a durability approach, the FRP bars were placed on To the authors’ best knowledge, no research has been
the outer layer to attack the harsh conditions, and the steel carried out studying the performance of concrete bridge-deck
bars were placed on the inner layer with an adequate cover slabs reinforced with hybrid bars. Therefore, it is necessary
to keep them away from corrosion. However, placing steel to comprehend how these types of structures behave and
and FRP bars in one layer presented better flexural strength later to allow and incorporate this concept into bridge design
than placing FRP bars in the outer layer and steel bars in codes and specifications. Based on the authors’ previous
the inner layer (Yinghao and Yong 2013). It is found that studies of this bridge deck (Ali et al. 2023), the structural
the effective reinforcement ratio has a major impact on the behavior of the hybrid RC bridge-deck slabs without shear
flexural capacity compared with axial stiffness between reinforcement was studied considering the most effective
GFRP and steel bars (Rf) (Qu et al. 2009). GFRP-steel RC parameters that affect the deck slab performance.
beams showed slightly lower flexural capacity than steel-RC
beams. However, the deflection and maximum crack width RESEARCH SIGNIFICANCE
were large in beams reinforced with steel reinforcement at The novelty of this paper is to shed light on the struc-
service load (Ruan et al. 2020). Furthermore, the ductility tural behavior of the BFRP/steel-RC deck slabs without
of hybrid reinforced beams can satisfy the specifications of shear reinforcement. The slabs’ performance was evaluated
serviceability limits by adequately regulating the reinforce- in terms of cracks propagation and failure modes, load-
ment ratio and the Af/As value (Ge et al. 2015). In addition, the deflection response, reinforcement and concrete strains, stiff-
maximum load and moment for serviceability increased with ness, ductility taking into consideration the effects of rein-
the GFRP-to-steel ratio. At the ultimate stage, the deflection forcement type and ratio, Af/As ratio, and slab thickness. The
obviously increases and provides a good pre-failure warning experimental test results were used to verify the accuracy of
(Xingyu et al. 2020). Applying the principle of equal stiff- existing models to predict the load-deflection response and
ness, the overall performance of the hybrid RC beams was to evaluate the bond-dependent coefficients, kb, of the ribbed
superior when As/Af ≤ 1.0; however, it declined intensely BFRP bars. In addition, some modifications were proposed
when As/Af > 1.0 (Wang et al. 2022). to different code equations to predict the shear strength of
An alternative method was proposed by Nanni et al. hybrid RC slabs with reasonable accuracy.
(1994) to protect the steel bar from corrosion by covering
the steel core with a braided and epoxy-coated aramid or EXPERIMENTAL INVESTIGATION
vinylon fiber FRP skin. Wu et al. (2012) developed a new Design concept
kind of hybrid bar, a steel-FRP composite bar (SFCB) that The sum of the FRP reinforcement ratio, ρf, and steel rein-
combines a ribbed steel bar inside and a longitudinal FRP forcement ratio, ρs, cannot be applied to directly express the
outside in a pultrusion process. The SFCB was created by reinforcement ratio of hybrid sections owing to the varia-
modifying FRP pultrusion technology. The benefits of an tions in mechanical properties between steel and FRP bars.
SFCB are: 1) the pultrusion process and the ribbed inner Two types of reinforcement ratios, ρsf,s and ρsf,f, were defined
steel bar ensure good interface properties and optimal use to account for combinations of the elastic modulus and
of each material; 2) excellent durability with the outer FRP; strength, respectively (Pang et al. 2016). The corresponding
3) a high elastic modulus of the SFCB due to the contri- balanced reinforcement ratios can be calculated as: ρs,b is
bution of steel at the initial stage; 4) noticeable post-yield the reinforcement ratio when concrete crushing and steel
modulus in the stress-strain relationship after the inner steel yielding happen synchronously; and ρf,b is the reinforcement
bar yields; and 5) high strength, good ductility, anti-erosion ratio when concrete crushing and FRP bar rupturing occur
properties, and low cost (Wu et al. 2009). simultaneously as follows
Many studies investigated the behavior of concrete
members reinforced with SFCBs: beams (Sun et al. 2012; _ Ef
Ge et al. 2020; Yang et al. 2020), slabs (Ali et al. 2023), and
ρsf,s = ρs+ Eρf (1)
s
columns (Sun et al. 2011; Ibrahim et al. 2017). It is found
that beams reinforced with SFCBs displayed stable post-
yield stiffness after the inner steel bar yielded. Although _fy
yielding the same ratio between steel and FRP, the ultimate
ρsf,f = ρs+ ρf (2)
ffu

capacity of the beam reinforced with BFRP and steel bars
showed only 72% of that reinforced with SFCBs. This is
attributed to the high bond stress in the hybrid beam, which fc′_
____ εcu
led to the early slip of BFRP bars (Sun et al. 2012). More-
ρs,b = α1β1   ε + ε (3)
fy cu y
over, SFCB-RC beams showed enhanced stiffness, reduced
crack width, and higher moment capacity than their coun-
terparts reinforced with FRP bars (Ge et al. 2020). Using fc′_
____ εcu
ρf,b = α1β1    ε + ε (4)
ffu
cu f
@seismicisolation
@seismicisolation

Hybrid FRP-steel slabs were designed in which ρsf,s ≤ ρs,b Fig. 1. BFRP bars are made of basalt fibers and epoxy resin,
and ρsf,f ≥ ρf,b. Therefore, the designed failure mode can be and basalt fiber content by weight is 70% according to the
defined as follows: firstly, the steel bars yield; subsequently, manufacturer. For SFCB production, 10 mm diameter inner
the concrete is crushed in compression; and lastly, the slab ribbed steel bars were used and wrapped with basalt fibers
fails. This failure would provide sufficient warning before that consisted of BFRP combined with a vinyl ester resin.
slab failure can be attained. Axial tensile tests were conducted to evaluate the mechan-
ical properties of the different reinforcing bars according to
Material properties ASTM D7205/D7205M (2016), as applicable. The results of
The deck slab specimens are fabricated using normal- the average three specimens of reinforcing bars are reported
weight ready mixed concrete to cast all the slabs on the in Table 1 and shown in Fig. 2.
same day. The concrete mixture design proportions per
cubic meter are 169 L of water, 452 kg of cement, 639 kg of Test specimens
sand, 1088 kg of aggregate, and an air content of 5 to 6% to Eleven large-scale RC slabs with a total length (Lt)
achieve a slump of 150 ± 30 mm. The average compressive of 2900 mm and a width (b) of 1000 mm were used. The
strength was evaluated by testing three standard 150 mm boundaries of the slabs were delineated considering the
concrete cubes after 28 days of curing. The concrete cubes contraflexural lines. Moreover, these dimensions are the
yielded an average compressive strength of 52 ± 1.3 MPa. It most popular size of the bridge-deck slabs for girder-type in
is worth mentioning that the cylinder compressive strength North America (El-Salakawy et al. 2003; El-Salakawy and
of concrete, fc′, was calculated based on ACI codes, whereas Benmokrane 2004). Nine slabs had a depth (h) of 200 mm
fc′ = 0.8fcu. The reinforcing bars used in this paper included (according to the requirements of CSA S6 [2019]), one
ribbed steel bars, BFRP bars, and SFCBs, as shown in slab had a depth of 250 mm, and the last one had a depth
of 120 mm. The investigated parameters were the reinforce-
ment type, effective reinforcement stiffness ρsf,s, the ratio
between the area of FRP and steel bars Af/As, reinforcement
arrangement, and slab thickness. All slabs have the same
steel reinforcement in all directions, 12 mm with a spacing
of 225 mm, except the bottom longitudinal reinforcement.
The clear concrete covers were 50 mm and 25 mm for the
top and bottom reinforcement, respectively. The slab spec-
imens were identified according to the amount and type of
longitudinal reinforcement (XB or XBXS), where the letters
X, B, and S indicate the number of bars, BFRP reinforcing
bar, and steel reinforcing bar, respectively.
The test program was categorized into four groups based
on the test parameters studied. Group A (reference group)
consists of two slabs (8B and 5B3S) studying the effect of
reinforcement type. In Group B, slabs (5B7S, 6B7S, and
10B6S) investigate the influence of increasing the effective
reinforcement stiffness, ρsf,f, compared to Group A. It is worth
noting that the slabs in this group were designed with similar
ρsf,s to study the impact of the Af/As ratio. Group C includes
slabs (5B5S, 5S5B, and 5B5S-S) that vary in reinforce-
ment arrangement, where S (the last letter in Slab 5B5S-S)
represents that the reinforcement is arranged in a single layer.
Regarding Slab 5S5B, which placed steel bars at the outer
layer, it is not efficient from the durability point of view;
Fig. 1—Configuration of BFRP and SFCB reinforcing
this system is examined only to evaluate the influence of the
bars: (a) geometry of BFRP bars and SFCBs; and (b) cross
reinforcement arrangement on the structural performance
sections.
Table 1—Properties of reinforcing bars

Bar type d, mm EI, GPa fy, MPa EII, GPa fu, MPa εu, %
10 200 420 — 632 3.30
Steel 12 200 452 — 667 3.16
16 200 494 — 688 3.50
BFRP 15.6 55 ± 0.60 — — 1163 ± 27.3 2.12 ± 0.12
SFCB 15.5 105 ± 2.4 263 ± 4.7 32 ± 3.2 756 ± 16.2 3.71 ± 0.16
Note: 1 mm = 0.0394 in.; 1 MPa = 145 psi; 1 GPa = 0.145 ksi. @seismicisolation
@seismicisolation

Table 2—Details of tested slabs
Group No. Slab Af/As ρsf,s % ρsf,f % ρsf,s/ρs,b ρsf,f/ρf,b
8B — 0.25 0.89 0.09 3.14
A
5B3S 1.58 0.59 0.75 0.20 2.65
5B7S 0.68 1.18 1.00 0.40 3.53
B 6B7S 0.81 1.19 1.12 0.41 3.95
10B6S 1.58 1.18 1.48 0.40 5.22
5B5S 0.88 0.93 0.30 3.28
C 5S5B 0.95 0.79 1.00 0.27 3.53
5B5S-S 0.76 0.88 0.26 3.11
9SFCB-200 0.59 0.82 0.20 2.05
Fig. 2—Stress-strain relationships of reinforcements. D 9SCFB-250 1.40 0.46 0.64 0.15 1.60
9SCFB-120 1.14 1.59 0.38 3.98
of the hybrid deck slabs. The last group comprised three
slabs (9SFCB-200, 9SFCB-250, and 9SFCB-120) with three cracking load, as listed in Table 3. As the load increased,
different thicknesses (200, 250, and 120 mm) to study the more flexural cracks began to develop below or between the
effect of different slab thicknesses. Table 2 summarizes the point loads. With further loading, flexural-shear and shear
details of the slabs. Figure 3 shows the cross sections of all cracks appeared and spread in the shear spans, demonstrating
deck slabs. the shear stresses in the shear span. Once the steel yielded,
the number of cracks increased and propagated, accompa-
Test program and instrumentation nied by the widening of the existing cracks.
The slabs were tested under four-point bending loading Slab 8B showed a larger number of shear cracks, and
up to failure. Figure 4 provides the test setup and schematic the crack length propagated quickly due to the low stiff-
diagram of the slabs. A steel spreader beam was used to ness modulus of the BFRP bar. In addition, Slab 8B yielded
transform the two concentrated loads 900 mm apart, yielding higher crack spacing, more severe cracking, and a lower
a shear span (a) of 900 mm on both sides with a clear span number of cracks than Slab 5B3S, thus tending to suggest
length (L) of 2700 mm. Two half-cylinders with 100 mm substituting BFRP bars with steel bars to improve the axial
diameters were used for loading; however, two full cylinders stiffness of the slab. Increasing the reinforcement ratio
were used to support the slab specimens, as shown in Fig. 4. increases the number of major and minor cracks, improves
The load was applied with displacement control at a constant the crack distribution length, and reduces the average minor
rate of 0.6 mm/min by the hydraulic jack of 500 kN capacity and major crack spacing. This result is clarified by better
to measure the applied loads, with a displacement sensor to bond strength as the number of longitudinal bars increases
measure the corresponding deflection. The load gradient was (Nguyen et al. 2020). Moreover, Slab 5B7S displayed low
established at 5 kN up to an applied load of 100 kN to detect average crack spacing at the same load level compared to
crack widths. After the applied load reached 100 kN, the load Slab 10B6S. The same phenomenon was reported by Ge
gradient was increased to 10 kN until the failure of the slab. et al. (2015); the average crack spacing diminishes with the
Three linear variable differential transformers (LVDTs) were reduction of Af/As.
positioned at the bottom midspan and the two loading points
of each slab, and two LVDTs were installed at the supports Modes of failure
to offset their settlements. Five electrical strain gauges were The combined effect of high shear force and bending
attached to the slab surface to measure the concrete strain moment leads to a spatially high-stress area; hence, failure
along the depth of the slab, and four strain gauges were also occurs in the shear span. Four different failure modes were
attached to the surface of the tensile reinforcement. The observed in the experimental tests and are listed in Table 3.
applied loads, deflections, and strain readings at a frequency Mode I: Diagonal tension failure (DTF)—This mode was
of 10 Hz were automatically recorded using a data acquisi- observed only for Slab 8B reinforced with BFRP bars owing
tion system. The crack width was measured using a hand- to the low stiffness of BFRP bars, as shown in Fig. 5. At
held readout microscope with a magnification factor of 40× a high load level, the shear cracks continued to widen due
with an accuracy of 0.01 mm. to the absence of shear reinforcement until failure occurred.
The diagonal shear crack occurred at 100 mm far from the
EXPERIMENTAL RESULTS AND DISCUSSION support, making it approximately 42 degrees with the hori-
Crack patterns and propagation zontal, then extended toward the loading point and widened
Figure 5 shows the crack distribution of the tested deck and propagated, leading to slab collapse. Moreover, the
slabs upon failure. The first crack was initiated in the pure failure was accompanied by local bending of steel bars in
bending moment zone. Generally, the cracking load was the compression zone keeping the slab intact as one part,
recorded at a similar load level for slabs of the same thick- as shown in Fig. 6. This phenomenon led to improved
ness, while increasing the slab thickness increased the ductility and integrity of the slab. In contrast, a previous
@seismicisolation
@seismicisolation

Fig. 3—Reinforcement details of tested deck slabs: (a) Group A; (b) Group B; (c) Group C; and (d) Group D. (Note: Dimen-
sions in mm; 1 mm = 0.0394 in.)
study observed that using FRP bars in the compression zone Fig. 5. The main crack that makes the failure for these slabs
resulted in the shearing off of FRP bars, and the slabs were has an inclined degree of 55 to 68 degrees. Additionally,
divided into two parts (Abdul-Salam et al. 2016). the failure was accompanied by concrete-cover spalling
Mode II: Flexural-shear failure (FSF)—This mode was on the tension side without FRP bars shearing off, and the
observed for Slabs 5B3S, 5B5S, 5S5B, and 9SFCB-200, slab maintained its integrity even after failure, which can
which initiated as a flexural crack closer to the loading point be attributed to the high reinforcement ratio and Af/As in
and then propagated inclining upward to the loading point. these slabs. Accordingly, a minimum amount of steel rein-
As shown in Fig. 5, the failure was gradual, and the slabs forcement with a reasonable value of Af/As is recommended
displayed some ductility before reaching the ultimate load, to ensure ductility and prevent the devastating failure of
rather than the brittle shear failure observed in Slab 8B. This concrete bridge-deck slabs.
may be attributed to the higher load levels reached, which Regarding the arrangement of reinforcement, Slab 5B5S-S,
confirms the enhancement of the shear resistance by the with reinforcement arranged in a single layer, exhibited
improved dowel action of the double-layer reinforcement severe failure compared with its counterparts arranged in a
(Yoo et al. 2016). double layer owing to deteriorated bond performance caused
Mode III: Shear-compression failure (SCF)—In this by the small spacing of the reinforcement (Yang et al. 2020).
mode, the compression zone in the slab was reduced by Minor differences were noted in the distribution and shape
the inclined flexure-shear cracks; thus, the concrete was of the flexural cracks for all tested slabs. However, the slab
crushed in Slabs 6B7S, 10B6S, and 5B5S-S, as shown in failure mechanism changes between the BFRP-RC slab and
@seismicisolation
@seismicisolation

Fig. 4—Test setup: (a) schematic drawing; and (b) slab specimen ready for testing. (Note: Dimensions in mm; 1 mm = 0.0394 in.)
hybrid RC slabs could be due to the difference in the shear- failure; thus, this type of section is admissible in the design
crack location. The shear crack in Slab 8B initiated closer to of hybrid RC members without shear reinforcement.
the support due to the diagonal tension stresses, resulting in a
crack closer to the support (100 mm). In contrast, Slab 5B3S Strain distribution
experienced no significant shear cracks, which moved the Figure 7 shows the strain distribution of concrete along the
crack location closer to the loading point (300 mm). It can be cross-section height at various loading stages. The average
related to the contribution of arch action, which was depen- concrete strain along the cross-section height is almost
dent on the critical shear-crack location. The crack was far linear and proportional to the distance from the neutral axis,
from the support in hybrid slabs, allowing the arch action to reflecting that the assumption of the plane cross section is
contribute more. valid for hybrid RC slabs, as shown in Fig. 7. The depth of
A discussion of flexural analysis was presented (refer to the the neutral axis of FRP-reinforced deck slab 8B was smaller
Appendix*), demonstrating that the hybrid RC slabs reached than the depth of the neutral axis of hybrid reinforced deck
their flexural capacities before failure. Therefore, this failure slab 5B3S, indicating the influence of the modulus of the
is considered a combination of flexural and shear failure. steel bars when added to the slab reinforcement.
Mode IV: Flexural failure (FF)—This mode was observed
for Slabs 5B7S, 9SFCB-250, and 9SFCB-120 by steel Ultimate capacity
yielding where the main crack was approximately under the The ultimate capacity for all tested slabs is listed in Table 3.
point load. These slabs did not display any bond-splitting The test results revealed that hybrid slab 5B3S showed a
cracks, which shifted the crack location closer to the loading slight increase of 6% in the ultimate capacity compared with
point. Generally, using hybrid reinforcement to reinforce the Slab 8B reinforced with pure BFRP bars. This is attributed
deck slabs, either separate reinforcing bars or SFCBs, could to the amount of BFRP bars in Slab 5B3S being less than
achieve some plastic deformations of concrete before total that of Slab 8B, whereas BFRP bars are responsible for
carrying the additional load after the steel yielded. In addi-
tion, the ultimate capacity increased, as did the reinforce-
ment ratio (Tureyen and Frosch 2002). For example, Slab
*
The Appendix is available at www.concrete.org/publications in PDF format,
appended to the online version of the published paper. It is also available in hard copy 6B7S showed an increase in the ultimate capacity of 39%
from ACI headquarters for a fee equal to the cost of reproduction plus handling at the compared to control slab 5B3S. This is due to the function
time of the request. @seismicisolation
@seismicisolation

Fig. 5—Crack patterns and failure modes of tested slabs: (a) Group A; (b) Group B; (c) Group C; and (d) Group D.
of the longitudinal reinforcement in decreasing the opening layer is more effective than placing the BFRP bars at the outer
and propagation of cracks resulting in smaller crack width, layer. Experimental results from Slab 9SFCB-250 showed
which increases the uncracked concrete depth and improves that shear capacity could be enhanced with increasing
the aggregate interlock across the cracks. Moreover, Slab slab stiffness compared to Slab 9SFCB-200. The ultimate
10B6S contains the highest FRP-to-steel content (Af/As = shear capacity of Slab 9SFCB-200 was 292 kN, while Slab
1.58) and showed an improvement in ultimate capacity by 9SFCB-250 failed at 359 kN, that is, a 23% improvement.
13% compared with Slab 5B7S (Af/As = 0.68). Increasing the slab thickness increases the surface area that
The reinforcement arrangement affects the ultimate resists the shear stresses, yielding higher shear capacity.
capacity because Slab 5B5S-S, reinforced in a single layer,
exhibited higher ultimate capacity than slabs arranged in Load-deflection relationship
a double layer. The ultimate capacity of Slab 5B5S-S is Figure 8 shows the load-midspan deflection curves for the
approximately 1.22 and 1.18 times the ultimate capacity of tested deck slabs. All slabs showed a steep linear behavior
Slabs 5B5S and 5S5B, respectively. From the mechanical at the pre-cracking stage with low deflection values. In the
point of view, placing the BFRP and steel bars at the outer post-cracking stage, slab stiffness was significantly degraded
@seismicisolation
@seismicisolation

Table 3—Test results of tested slabs
Group Slab Δcr, mm Δy, mm Δu, mm Pcr, kN Py, kN Pu, kN Mexp/Mpred Failure mode
8B 2.3 — 63 36 — 295 0.80 DTF
A
5B3S 1.7 19.1 78 41 179 314 1.43 FSF
5B7S 1.2 22.3 67 42 260 414 1.56 FF
B 6B7S 2.0 20.1 49 46 251 437 1.59 SCF
10B6S 1.9 20.3 45 47 294 470 1.62 SCF
5B5S 2.3 20.6 68 43 190 325 1.42 FSF
C 5S5B 2.3 18.4 73 38 190 335 1.34 FSF
5B5S-S 3.7 16.5 58 49 211 396 1.51 SCF
9SFCB-200 1.5 13.3 77 34 146 292 1.23 FSF
D 9SFCB-250 3.1 13.2 65 99 212 359 1.22 FF
9SFCB-120 8.8 21.9 144 26 69 147 2.18 FF
Average *
— — — — — 1.51 —
Standard deviation* — — — — — 0.26 —
Coefficient of variation , %
*
— — — — — 17 —
*
These characteristics were calculated only for hybrid RC deck slabs; DTF is diagonal tension failure; FSF is flexural-shear failure; FF is flexural failure; SCF is shear-compression
failure.
Note: 1 mm = 0.0394 in.; 1 kN = 0.225 kip.
Fig. 6—Bending of steel bars in compression zone: (a) 8B; and (b) 10B6S.
due to flexural cracks, which reduced the moment of inertia. yielding, a secondary stiffness was detected only for hybrid
Slab 8B exhibited a bilinear load-midspan deflection curve deck slabs. This observation proves the concept of the
and degraded faster than hybrid RC slabs due to the lower hybrid section that the significant role of FRP bars brightens
elastic modulus of the BFRP bars, as shown in Fig. 8(a). after steel yielding. For Group B, increasing the reinforce-
Conversely, hybrid RC slabs exhibited trilinear load-deflec- ment ratio increased the post-cracking stiffness, and hence
tion curves owing to the presence of steel bars. As shown decreased the deflection at similar load levels (El-Sayed
in Fig. 8, after steel yielded, a pronounced reduction in the et al. 2006b).
slope of the load-deflection curves as the load increased to Regarding the Af/As, there was a slight influence of the
high levels means that steel reinforcement cannot resist any ratio of Af/As on the stiffness after cracking. However, a
additional load, and only the BFRP reinforcement carried the significant enhancement in stiffness was noticed after steel
load upon failure. In this stage, the deflection of hybrid deck yielding for Slab 5B7S in comparison to Slab 10B6S; thus,
slabs was lower than the deflection of Slab 8B, attributed as Af/As increases, the deflection decreases (Safan 2013). For
to the efficiency of the BFRP bars restricting the excessive example, at the load of 294 kN, the deflection of Slab 10B6S
deflection even after steel yielding. Hence, using hybrid bars was 30% lower than the deflection of Slab 5B7S because the
to reinforce the concrete bridge-deck slabs keeps the load former had a high Af/As, which enhanced the slab rigidity
growing with a reasonable deflection value. by restricting the excessive deflection after the steel yielded.
In Group A, Slab 8B suffered a larger deflection than The foregoing results proved the significant influence of the
Slab 5B3S under the same load level, as shown in Fig. 8(a). Af/As on the post-elastic strength of bridge-deck slabs with
This is attributed to the high elastic stiffness of steel bars sufficient deformability and stiffness.
compared to BFRP bars, which increased the rigidity of the Considering the reinforcement arrangement in Group C,
hybrid slabs. At the yielding load of Slab 5B3S, the deflec- both slabs reinforced in a double layer developed larger
tion decreased by 43% compared to Slab 8B. After @seismicisolation
steel deflection than the slabs arranged in a single layer, as shown
@seismicisolation

Fig. 7—Strain distribution of tested slabs: (a) 8B; (b) 5B3S; and (c) 10B6S. (Note: 1 mm = 0.039 in.)
Fig. 8—Load-deflection relationships of tested slabs: (a) Groups A and B; (b) Group C; and (c) Group D. (Note: 1 kN =
0.225 kip; 1 mm = 0.0394 in.)
@seismicisolation
@seismicisolation

in Fig. 8(b). At a load of 200 kN (approximately the yielding the secondary stiffness depended on BFRP reinforcement
load of this group), the deflections of Slabs 5B5S, 5S5B, due to steel yielding, which could not bear any additional
and 5B5S-S were 23.9, 20.2, and 15.7 mm, respectively. load. The post-yielding stiffness of hybrid RC slabs can lead
This means that by increasing the depth of the steel bars to a smaller residual deformation during unloading.
layer, the stiffness degradation of the double-layer hybrid Consequently, the equal stiffness design principle is
deck slab was faster than that of the single-layer hybrid recommended to realize a damage-controllable structure
deck slab (Yinghao and Yong 2013). The slab thickness had under earthquakes (Wu et al. 2009; Wang et al. 2022). For
a clear effect on the stiffness of SFCB slabs, as shown in Group C, the secondary stiffness, KII, did not show a big
Fig. 8(c). An apparent reduction in the deflection was found difference between the two slabs arranged in double layers.
when increasing the slab thickness from 120 mm to 200 and However, KII for Slab 5B5S-S was higher than both Slabs
250 mm while maintaining a constant reinforcement amount. 5B5S and 5S5B, which means that the stiffness degradation
This is due to increasing the moment of inertia of the slabs. of the double-layer hybrid deck slab was faster than that of
the single-layer hybrid deck slab (Yinghao and Yong 2013).
Stiffness Among all the tested slabs, Slab 9SFCB-250 showed the
Based on the experimental test results, two stiffness highest initial stiffness, 16.15 kN/mm, due to increasing
factors were calculated to measure the stiffness of the tested the moment of inertia, hence improving the slab stiffness;
deck slabs before and after yielding: the initial equivalent on the other hand, Slab 9SFCB-120 showed the lowest one,
stiffness, KI, and secondary stiffness, KII, respectively (Sun 3.13 kN/mm.
et al. 2019).
Ductility
_ y
P Ductility refers to the amount of inelastic deformation

KI = Δ (5) that can be undergone without losing load-carrying capacity
y
before complete failure. This paper adopted the deforma-
tion-based approach to evaluate the ductility of hybrid rein-
_ Pu − Py forced deck slabs. The overall performance factor, J, was

KII = Δ − Δ (6) adopted by CSA S6:19, which combines the strength and
u y
deformability provided by Eq. (7).
The stiffness factors are listed in Table 4. The results
showed that the slab stiffness tended to improve with ΨuMu
_

J = Ψ M (7)
increasing the effective reinforcement ratio, ρsf,f, due to the s s
improvement in the axial stiffness of the deck slab. Thus,

Slab 5B3S designed with a low ρsf,f suffered from stiffness The moment and curvature at the serviceability limit state
degradation after cracking. Increasing the reinforcement are taken as the point when the maximum concrete compres-
ratio from 0.59 to 1.19% in Slabs 5B3S and 10B6S increased sive strain reaches a value of 0.001. CSA S6:19 states that
the initial and secondary stiffness KI and KII by 54% and the overall performance factor, J, should be at least 4.0 for
218%, respectively. A slight influence of Af/As on the initial rectangular sections. Equation (7) uses service moments,
stiffness, KI, and a significant enhancement of 83% in the Ms, at a concrete strain of 0.001, as recommended by CSA
secondary stiffness, KII, were observed for Slab 6B7S in S6:19, neglecting the yielding of steel bars. Therefore, a
comparison to Slab 5B7S. This indicated that the initial modified overall performance factor, Jmod, was suggested for
equivalent stiffness was affected by steel reinforcement hybrid reinforced beams taken as the ratio of the product
rather than BFRP reinforcement due to the significant varia- of moment and curvature at the ultimate to the product of
tion in the elastic modulus (Nguyen et al. 2020). In contrast, moment and curvature at the yielding of steel reinforce-
ment, as given in Eq. (8) (El Refai et al. 2015). It should be
Table 4—Stiffness and ductility of tested slabs mentioned that Jmod was used only for slabs reinforced with
Group Slab KI KII Jmod
hybrid reinforcement.
8B — — 4.2
A _ΨuMu
5B3S 9.44 2.28 7.5
Jmod
= Ψ M (8)
y y
5B7S 11.64 3.48 5.3
B 6B7S 12.52 6.38 5.1
Table 4 shows the values of the modified overall perfor-
mance factor, Jmod, of all tested deck slabs. The inclusion of
10B6S 14.50 7.24 4.8
steel bars in Slab 5B3S improved the Jmod by 79% compared
5B5S 9.23 2.86 5.7 to Slab 8B. Slabs with lower reinforcement ratios showed
C 5S5B 10.32 2.63 6.2 a higher performance due to the proportional decrease in
5B5S-S 12.80 4.52 6.5
the stiffness of the slabs with the decrease in the reinforce-
ment ratio (Liu et al. 2022). Increasing the reinforcement
9SFCB-200 10.96 2.31 7.5
ratio from 0.59 to 1.18% (Slabs 5B3S and 10B6S) reduced
D 9SFCB-250 16.15 2.80 6.3 the Jmod from 7.5 to 4.8. This is attributed to the enhanced
9SFCB-120 3.13 1.13 6.0 reinforcement axial stiffness; increasing the reinforcement
@seismicisolation
@seismicisolation

Fig. 9—Load-maximum crack width relationships of tested slabs: (a) Groups A and B; (b) Group C; and (c) Group D. (Note:
1 kN = 0.225 kip; 1 mm = 0.0394 in.)
axial stiffness restricts the propagation of cracks, leading outer layer recorded a wider crack width for Slab 5B5S,
to narrower and shorter cracks. While the overall perfor- followed by Slabs 5S5B and 5B5S-S. Figure 9(c) shows that
mance factor retracted slowly with the increase in Af/As, it higher crack width values were recorded for Slab 9SFCB-
can record its maximum value at a low value of Af/As. This 120, indicating that decreasing the slab thickness to 120 mm,
is because the ultimate deflection and the ultimate load show which is less than the CSA S6:19 minimum allowable thick-
variations, changing trends with the increase in Af/As. With ness of 175 mm, resulted in wider cracks.
respect to the reinforcement arrangement, it can be found Different codes and design guidelines proposed a limit
that the overall performance factor of the single-layer hybrid for the crack width of bridge-deck slabs. The American
RC slab 5B5S-S is 6.5, which is larger than that of the Association of State Highway and Transportation Offi-
double-layer hybrid RC slab (5.7 for 5B5S). The values of cials (AASHTO) LRFD Bridge Design Specifications
the modified overall performance factor satisfied the CSA (AASHTO 2018) recommended a limit value of 0.5 mm for
S6:19 requirements, which ranged from 4.2 to 7.5. The maximum crack width when using GFRP reinforcement.
higher values of the overall performance factor imply more Furthermore, CSA S6:19, Clause 16.8.2.3, states that crack
ample warning by exhibiting a significant deformation at the width should not exceed 0.5 mm and 0.7 mm for exterior
ultimate state of the hybrid RC deck slabs before failure. and interior exposure, respectively. As shown in Fig. 9, the
Therefore, the hybrid RC slabs can meet the requirements of maximum crack widths for all test slabs reached 0.4 mm
deformability by using an adequate value of Af/As. after nearly 54 to 79% of the ultimate load. These loads were
much higher than the service loads; thus, the authors recom-
Crack width mended that for hybrid RC deck slabs, the maximum crack
Figure 9 shows the applied load-crack width curves. The width limit should be 0.4 mm for exterior exposure.
load-crack width relationship was almost linear for the pure
BFRP-RC slab; however, all hybrid deck slabs exhibited a Reinforcement and concrete strains
bilinear curve owing to the yielding of steel bars. In Group A, The applied load versus measured strains at the midspan
replacing the BFRP bars with steel bars could restrain crack of both concretes at the top fiber, and the tensile BFRP bars
depth and fast propagation due to the higher elastic modulus or SFCBs, are shown in Fig. 10. Before cracking, all the
of the steel bars. As shown in Fig. 9(a), the crack widths were tested deck slabs showed approximately similar concrete
inversely proportional to the reinforcement ratio. Thus, the and reinforcement strains. After cracking, the BFRP strain
measured crack width was lower in the case of Slab 6B7S of Slab 8B showed a rapid linear increase, increasing the
than in Slab 5B3S. Moreover, a higher ratio of FRP-to-steel load up to failure. Using steel bars instead of BFRP bars,
amount Af/As led to better secondary stiffness and hence Slab 5B3S yielded a significant reduction in BFRP strain
narrower crack widths. Placing BFRP reinforcement@seismicisolation
in the compared to Slab 8B, as shown in Fig. 10(a). The influence
@seismicisolation

Fig. 10—Load-BFRP reinforcement and concrete-strain relationships: (a) Groups A and B; (b) Group C; and (c) Group D.
(Note: 1 kN = 0.225 kip.)
of the reinforcement ratio can be realized in Slabs 5B3S COMPARISON OF TEST RESULTS WITH
and 10B6S; high reinforcement and lower strains were DIFFERENT DESIGN CODE PREDICTIONS
recorded at the same load levels. Increasing the steel rein- Prediction of midspan deflection
forcement controlled the BFRP bar strain development in Equations from different design codes and available
the hybrid deck slabs, as demonstrated by the lower tensile models in the literature (refer to the Appendix) were used
strain readings in the BFRP bars of Slab 5B3S compared to predict the midspan deflections of the tested slabs at
to Slab 5B7S. These results tend to indicate the effect of two load levels, 30 and 60% of the ultimate load capacity,
increasing the effective reinforcement ratio to improve the Pu, of each slab. Figure 11 compares the experimental
serviceability performance of the hybrid slabs. Moreover, as and predicted deflections up to 60% of the ultimate load.
the Af/As increased, the BFRP strains decreased, as shown Figure 11(a) shows clearly that both the Bischoff model
in Fig. 10(a). The BFRP strain in Slab 10B6S (Af/As = 1.58) (Bischoff 2007) and ACI 440.1R-15 yielded highly under-
recorded 2422 με, which was considerably lower than the estimated deflections for Slab 8B; however, CSA S806-12
BFRP strain of Slab 6B7S (Af/As = 0.81) of 4273 με at the showed good results with experimental deflections. As
yielding load. The secondary stiffness (provided by either shown in Fig. 11(b), the midspan deflections of Slab 5B3S
BFRP bars or the outer FRP of SFCBs) can restrict the reinforced with a low reinforcement ratio (ρeff = 0.59%)
strain development of tensile reinforcements after the steel were underestimated by all the equations; however, the CSA
is yielded (Yang et al. 2020). As shown in Fig. 10(b), there S806-12 equation showed good results with experimental
was no clear effect of the reinforcement arrangement on the deflections. However, better prediction of deflections was
yielding load, while Slab 5B5S read a high value of BFRP noticed when increasing the reinforcement ratio in Group B.
strain of 1932 με, followed by values of 1518 με and 1198 με Moreover, a similar trend is also valid for the hybrid slabs
for Slabs 5S5B and 5B5S-S, respectively. Therefore, the of Group C. An excellent agreement between the predicted
BFRP reinforcement undergoes more tensile stress when and the experimental deflections was noticed for Slabs
placed in the outer layer. 6B7S and 5B5S, especially for CSA S806-12, illustrated
in Fig. 11(e). From a design point of view, CSA S806-12
showed the most accurate method of predicting the deflec-
tion of hybrid RC members among all current models at both
0.30Pu and 0.60Pu, with an average of Vc,exp/Vc,pred of 1.02 ±
@seismicisolation
@seismicisolation

Fig. 11—Comparisons of midspan deflection versus experimental and calculated results of tested deck slabs: (a) 8B; (b) 5B3S;
(c) 5B7S; (d) 6B7S; (e) 10B6S; (f) 5B5S; (g) 5S5B; (h) 5B5S-S; (i) 9SFCB-200; (j) 9SFCB-250; and (k) 9SFCB-120. (Note:
1 mm = 0.0394 in.)
0.26 and 1.12 ± 0.10. This might be caused by the cracked Bond-dependent coefficient prediction
moment of inertia in the closed-form equation suggested by Both design codes ACI CODE-440.11-22 (ACI Committee
CSA S806-12. 440 2022) and CSA S6:19 provided the same expression for
@seismicisolation
@seismicisolation

predicting the maximum crack width of FRP-RC members, kb determined at the different load levels. The average kb
as shown in Eq. (9). The bond-dependent coefficient, kb, value was 1.27 ± 0.14 with a coefficient of variation (COV)
considered the variety of concrete sections and different of 11% for BFRP bars with ribbed surfaces. It was found
FRP bars. In addition, the bond degree of FRP bars within that this value is close to that recommended by Shield et al.
the concrete, which is always indicated by the bond coeffi- (2019) and ACI CODE-440.11-22 of 1.2 for GFRP bars.
cient in existing design codes, should be determined before However, CSA S6:19 proposed using a kb of 1.0 for ribbed
computing the crack width. The determination of the bond FRP bars; thus, CSA S6:19 is nonconservative.
coefficient, kb, is suggested to be realized by substituting
the experimentally obtained crack widths and tensile strains Prediction of shear strength
of the longitudinal reinforcements into Eq. (9). It should be The shear capacities of hybrid reinforced slabs were
noted that w in Eq. (9) denotes the maximum crack width predicted with the shear models provided in different design
at the bottom of the tension face, while the measured crack codes. To date, no design code or guidelines have been
widths of the present study were at the level of BFRP rein- published for designing concrete structures with hybrid rein-
forcement. Therefore, the amplification factor β in Eq. (9), forcement. Therefore, the authors suggested a modification
which transfers the crack width at the level of reinforcements for some equations that consider the effect of hybrid rein-
to that of the tension face, was not considered. forcement. For example, for the AASHTO specifications
_ (AASHTO 2018), it is recommended to account for the ratio
between the stiffness of longitudinal steel reinforcement, Es,

_s 2
√
w = 2εfβkb dc2+ (2 ) (9a)
to that of FRP reinforcement, Ef. For CSA S806-12, which
considers the effect of FRP reinforcement, ρf, the authors
proposed replacing the FRP reinforcement with the mechan-
Ma ical reinforcing index, ρf,eq, as shown in Eq. (10). However,
_____________________

εf =   
   (9b) CSA S6:19 considers FRP axial stiffness in the shear equa-
( EfAf+ EsAs) [1 − (3 )]d
_k
tions; thus, the authors suggested accounting for the axial
stiffness of both steel and FRP reinforcement rather than
only FRP reinforcement, as shown in Eq. (11). Moreover,
_
h − kd ACI CODE-440.11-22 indirectly incorporates both steel and

β = (9c)
d − kd FRP reinforcement through the coefficient k. Table 6 (refer
to the Appendix) summarizes all shear equations after intro-
Herein, the slab results were used to assess the value of ducing the modifications.
the bond coefficient of BFRP bars, kb, with Eq. (9). The kb
was calculated at two load levels: at 0.3Pu, which is consid- _fy
ered the service load level (Mota et al. 2006; Bischoff et al.
ρf,eq = ρf+ ρs (10)
ffu

2009; El-Nemr et al. 2013), and at 0.67Pu, in which the
crack pattern reached a stabilized state and no new cracks
appeared. Table 5 reports the average kb measured for each EA = EsAs + EfAf (11)
slab specimen. The results showed some variations between
The accuracy of the predicted results was evaluated and
Table 5—Bond-dependent coefficient, kb discussed by comparing their predictions with the values
kb that were experimentally determined from the tested slabs.
Slab 0.30Pu 0.67Pu Average The experimental-to-predicted shear ratios, Vc,exp/Vc,pred,
are presented in Table 7. From the results of the prediction,
8B 1.11 1.07 1.09
AASHTO yielded very good agreement with the experi-
5B3S 1.38 1.37 1.38 mental results, with an average Vc,exp/Vc,pred of 1.23 ± 0.36.
5B7S 1.22 1.42 1.32 After introducing the modifications, ACI CODE-440.11-22
6B7S 0.92 1.45 1.19 and CSA S806-12 yielded good yet conservative predic-
tions with average Vc,exp/Vc,pred of 1.65 ± 0.21 and 1.59 ±
10B6S 1.32 0.97 1.15
0.33 and a COV of 13 and 21%, respectively. However,
5B5S 1.42 1.44 1.43 CSA S6:19 showed very conservative predictions with an
5S5B 1.43 1.41 1.42 average Vc,exp/Vc,pred of 2.01 ± 0.30 compared to AASHTO.
5B5S-S 0.97 1.25 1.11
This might be attributed to CSA S6:19 accounting for the
effect of the longitudinal FRP reinforcement ratio and the
9SFCB-200 1.28 1.42 1.35
elastic modulus in one term (axial stiffness, EA) rather than
9SFCB-250 1.03 1.12 1.08 the other codes calculating them separately. Generally, the
9SFCB-120 1.46 1.38 1.42 modifications in the shear equations proposed in this study
Average 1.27
are suitable for the predictions of the shear capacities of
hybrid RC members reinforced with steel and BFRP bars,
Standard deviation 0.14
while extensive investigations are needed in the future to
Coefficient of variation, % 11 confirm these modifications.
@seismicisolation
@seismicisolation

Table 6—Modified shear equations for hybrid RC members
Reference Equation
_____
ACI CODE-440.11-22 Vc = 0.4√ fc′ bw kd
_____
AASHTO (2018) Vc = 0.16√ fc′ bw c(Es/Ef); c = kd
_____ ____
Vc,mod
= 0.11ϕc√fc′ bw d_v ≤ 0.05λϕckm (fc′)  1/3bw dv ≤ 0.22ϕc√fc′ bw dv
kr,mod
CSA S806-12
√
Vu d
_
km = M ≤ 1.0; kr,mod = 1 + (Efρsf,f)1/3
u
Vc = 2.5βϕcfcrbwdv
βmod =[   1 + 1500εx,mod][1000 + sze]
____________
0.4 _ 1300
CSA S6:19
Mu
_
+ Vu
dv
_____________ _ 35sz
εx,mod =    ≤ 0.003; s ze = 15 + a ≤ 0.85sz
2(EfAf+ EsAs) g
Table 7—Experimental-to-predicted shear capacity

Vu/Vu,pred
Slab Vu ACI CODE-440.11-22 AASHTO (2018) CSA S806-12 CSA S6:19
8B 147.5 2.06 2.06 1.24 1.83
5B3S 157 1.49 1.48 1.31 1.77
6B7S 218.5 1.70 1.07 2.01 2.24
10B6S 235 1.86 1.17 2.19 2.46
5B5S 162.5 1.45 0.91 1.51 1.71
5S5B 167.5 1.55 0.98 1.55 1.79
5B5S-S 198 1.71 1.17 1.67 2.44
9SFCB-200 146 1.41 0.97 1.24 1.83
Average 1.65 1.23 1.59 2.01
Standard deviation 0.21 0.36 0.33 0.30
Coefficient of variation, % 13 29 21 15
Note: Vc,exp is factored shear (kN); 1 kN = 0.225 kip.
DESIGN RECOMMENDATIONS FOR HYBRID RC SUMMARY AND CONCLUSIONS

MEMBERS WITHOUT SHEAR REINFORCEMENT Eleven deck slabs were constructed and tested up to failure
Based on the experimental results in this study, under- to study the mechanical behavior of hybrid deck slabs. The
reinforced hybrid RC slabs exhibit higher stiffness and test parameters are reinforcement type, ratio, and arrange-
strength with ductile behavior before failure compared to ment; the fiber-reinforced polymer (FRP)-to-steel ratio Af/
FRP-RC slabs, which is still adequate to achieve the safety As; and slab thickness. Based on the results and discussion
requirement. Hence, it is recommended that designers should presented herein, the following conclusions are obtained:
follow the under-reinforced section procedures when using 1. The observed failure mode of the basalt FRP (BFRP)-
hybrid (FRP and steel) bars as tensile reinforcement. More- reinforced concrete (RC) slab was an undesirable brittle
over, it is necessary that engineers should be aware of the shear failure, while hybrid RC slabs showed gradual shear-
FRP-to-steel ratio, Af/As, when designing such sections. It is compression failure, flexural-shear failure, and flexural
suggested to use a high Af/As, which means a large amount failure. The failure of the hybrid RC slabs is significantly
of FRP bars to provide high strength and avoid excessive affected by the effective reinforcement ratio and Af/As; it
deflection and rupture of FRP bars after the steel yields. In becomes more severe when increasing the reinforcement
contrast, using a low Af/As can ensure the ductile perfor- ratio with a high value of Af/As.
mance guaranteed by the yielding of steel bars. In addition, 2. The effective reinforcement ratio and Af/As had consid-
using the proposed shear equations presented in this study erable influence on the post-cracking stiffness of the hybrid
can assure the ductile failure of RC flexural elements by deck slabs. In addition, increasing the reinforcement ratio
making the shear capacity at all sections equal to or greater by 100% increased the ultimate capacity by 50%. Increasing
than the flexural one. Recently, some design recommenda- Af/As from 0.68 to 1.58 significantly reduced the midspan
tions were proposed through an experimental and numerical deflections and crack widths and increased the ultimate
study made by the authors (Ali et al. 2023). However, further capacity by 14%.
research on the concrete contribution to the shear strength of 3. The existing models for predicting the midspan deflec-
hybrid RC members is needed. tions are conservative in predicting the deflections of hybrid
RC slabs. However, the models give good predictions for
slabs that have high effective reinforcement ratios. CSA
@seismicisolation
@seismicisolation

S806-12 provided accurate predictions with the experi- Icr = cracked moment of inertia
Ie = effective moment of inertia
mental results. Ig = gross moment of inertia
4. The test results were used to assess the bond-depen- J = overall performance factor
dent coefficient of BFRP bars. The calculations yielded an Jmod = modified overall performance factor
kb = bond coefficient
average bond-dependent coefficient, kb, of 1.27, close to the km = moment-shear interaction factor
1.2 recommended by ACI CODE-440.11-22 and larger than kr,mod = reinforcement stiffness factor
the 1.0 proposed by CSA S6:19. L = slab length
Lg = distance from support to point where M = Mcr
5. The proposed modifications to the design code equa- Ma = applied moment at critical section
tions were verified to accurately predict the shear capacity of Mcr = cracking moment
hybrid RC slabs without shear reinforcement. The AASHTO Mexp = experimental moment capacity
Mpred = predicted moment capacity
equation provided the most accurate shear capacity predic- Ms = service moment
tion for hybrid RC slabs, with an average ratio of Vc,exp/Vc,pred Mu = ultimate moment
of 1.23. My = yielding moment
nf = ratio of modulus of elasticity of BFRP bars to modulus of elas-
ticity of concrete
AUTHOR BIOS ns = ratio of modulus of elasticity of steel bars to modulus of elas-
ACI member Yahia M. S. Ali is a PhD Candidate in the School of Civil ticity of concrete
Engineering at Southeast University, Nanjing, Jiangsu, China, and a Pcr = cracking load
Teaching Assistant in the Department of Civil Engineering at Assiut Univer- Pth = theoretical flexural capacity
sity, Assiut, Egypt. He received his BS and MSc in structural engineering Pu = ultimate load
from Assiut University. His research interests include structural design Py = yielding load
and the analysis and testing of concrete structures reinforced with fiber- s = spacing of reinforcing bars
reinforced polymers (FRPs). sze = effective crack spacing for members without stirrups
Vc,mod = modified one-way shear strength provided by concrete and flex-
Xin Wang is a Professor in the Key Laboratory of Concrete and Prestressed ural reinforcement
Concrete Structures of Ministry of Education at Southeast University, and a Vu = factored shear
Professor in the Department of Civil Engineering at Southeast University. w = maximum crack width
He received his BS and MSc from Southeast University in 2002 and 2006, α1 = ratio of average stress of equivalent rectangular stress block to
respectively, and his PhD from Ibaraki University, Mito, Ibaraki, Japan, in cylinder compressive strength of concrete
2010. His research interests include FRP composites and concrete struc- β = ratio of distance between neutral axis and tension face to
tures with FRP composites. distance between neutral axis and centroid of reinforcement
β1 = ratio of depth of equivalent rectangular stress block to depth of
Shui Liu is a PhD Candidate in the Key Laboratory of Concrete and neutral axis
Prestressed Concrete Structures of Ministry of Education at Southeast Δcr = cracking deflection at midspan of slab
University and in the Department of Civil Engineering at Southeast Univer- Δm = maximum deflection at midspan of slab
sity. His research interests include the analysis and design of concrete Δu = ultimate deflection at midspan of slab
structures reinforced with a combination of FRP and steel bars. Δy = yielding deflection at midspan of slab
εcu = maximum concrete compressive strain (0.003 for ACI 318-19
Zhishen Wu is a Professor in the School of Civil Engineering at Southeast [ACI Committee 318 2019] provisions)
University. He received his BS and MS from Southeast University, and his εf = tensile strain in BFRP bars
PhD from Nagoya University, Nagoya, Aichi, Japan. His research interests εs = yield strain in steel bars
include FRP composites and technologies, maintenance and life-cycle engi- εx,mod = modified longitudinal strain at middepth of cross section
neering, failure mechanics of composite materials, and structural health ϕc = material reduction factor for concrete (taken as unity in this
monitoring infrastructure systems. paper)
η = reduction coefficient
ACKNOWLEDGMENTS λ = concrete density factor
The authors gratefully acknowledge the financial support provided ρf = BFRP reinforcement ratio
by the National Key Research and Development Program of China ρf,b = FRP balanced reinforcement ratio
(No. 2022YFB3706503) and the National Natural Science Foundation of ρf,eq = mechanical reinforcing index
China (Grant 52278244). ρs = steel reinforcement ratio
ρs,b = steel balanced reinforcement ratio
ρsf,f = effective reinforcement ratio in hybrid sections with respect to
NOTATION FRP
Af = area of BFRP reinforcement ρsf,s = effective reinforcement ratio in hybrid sections with respect to
As = area of steel reinforcement steel
a = distance between support and point load (shear span) Ψs = curvature at service moment
ag = nominal maximum size of coarse aggregate Ψu = curvature at ultimate moment
bw = width of cross section Ψy = curvature at yield moment
c = distance from extreme fiber in compression to neutral axis
d = distance from extreme fiber in compression to center of tensile
reinforcement REFERENCES
df = distance from extreme compression fiber to center of FRP bars AASHTO, 2018, “AASHTO LRFD Bridge Design Guide Specifications
ds = distance from extreme compression fiber to center of steel bars for GFRP-Reinforced Concrete,” second edition, American Association of
dv = effective shear depth State Highway and Transportation Officials, Washington, DC.
Ec = modulus of elasticity of concrete Abdul-Salam, B.; Farghaly, A. S.; and Benmokrane, B., 2016, “Mecha-
Ef = modulus of elasticity of BFRP bars nisms of Shear Resistance of One-Way Concrete Slabs Reinforced with FRP
EI = elastic modulus Bars,” Construction and Building Materials, V. 127, Nov., pp. 959-970. doi:
EII = post-elastic modulus 10.1016/j.conbuildmat.2016.10.015
Es = modulus of elasticity of steel bars ACI Committee 318, 2019, “Building Code Requirements for Structural
fc′ = concrete compressive strength Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved
fcr = cracking strength 2022),” American Concrete Institute, Farmington Hills, MI, 624 pp.
fcu = cube compressive strength ACI Committee 440, 2015, “Guide for the Design and Construction of
ffu = ultimate tensile stress in BFRP bars Structural Concrete Reinforced with Fiber-Reinforced Polymer (FRP) Bars
fy = yield stress in steel reinforcement (ACI 440.1R-15),” American Concrete Institute, Farmington Hills, MI,
h = overall member thickness 88 pp.
@seismicisolation
@seismicisolation

ACI Committee 440, 2022, “Building Code Requirements for Structural Construction and Building Materials, V. 87, July, pp. 28-37. doi: 10.1016/j.
Concrete Reinforced with Glass Fiber-Reinforced Polymer (GFRP) Bars— conbuildmat.2015.03.113
Code and Commentary (ACI CODE-440.11-22),” American Concrete Insti- Goldston, M.; Remennikov, A.; and Sheikh, M. N., 2016, “Experimental
tute, Farmington Hills, MI, 260 pp. Investigation of the Behaviour of Concrete Beams Reinforced with GFRP
Aiello, M. A., and Ombres, L., 2002, “Structural Performances of Bars under Static and Impact Loading,” Engineering Structures, V. 113,
Concrete Beams with Hybrid (Fiber-Reinforced Polymer-Steel) Reinforce- Apr., pp. 220-232. doi: 10.1016/j.engstruct.2016.01.044
ments,” Journal of Composites for Construction, ASCE, V. 6, No. 2, May, Ibrahim, A. I.; Wu, G.; and Sun, Z.-Y., 2017, “Experimental Study of
pp. 133-140. doi: 10.1061/(ASCE)1090-0268(2002)6:2(133) Cyclic Behavior of Concrete Bridge Columns Reinforced by Steel Basalt-
Ali, Y. M. S.; Wang, X.; Ding, L.; Liu, S.; and Wu, Z., 2023, “Exper- Fiber Composite Bars and Hybrid Stirrups,” Journal of Composites for
imental and Numerical Investigations of the Effects of Various Tensile Construction, ASCE, V. 21, No. 2, Apr., p. 04016091. doi: 10.1061/(ASCE)
Reinforcement Types on the Structural Behavior of Concrete Bridge Deck CC.1943-5614.0000742
Slabs,” Engineering Structures, V. 285, June, Article No. 116036. doi: Liu, S.; Wang, X.; Ali, Y. M. S.; Su, C.; and Wu, Z., 2022, “Flexural
10.1016/j.engstruct.2023.116036 Behavior and Design of Under-Reinforced Concrete Beams with BFRP and
ASTM D7205/D7205M-06(2016), 2016, “Standard Test Method for Steel Bars,” Engineering Structures, V. 263, July, Article No. 114386. doi:
Tensile Properties of Fiber Reinforced Polymer Matrix Composite Bars,” 10.1016/j.engstruct.2022.114386
ASTM International, West Conshohocken, PA, 13 pp. Matta, F.; El-Sayed, A. K.; Nanni, A.; and Benmokrane, B., 2013, “Size
Bischoff, P. H., 2007, “Deflection Calculation of FRP Reinforced Effect on Concrete Shear Strength in Beams Reinforced with Fiber-Rein-
Concrete Beams Based on Modifications to the Existing Branson Equa- forced Polymer Bars,” ACI Structural Journal, V. 110, No. 4, July-Aug.,
tion,” Journal of Composites for Construction, ASCE, V. 11, No. 1, Feb., pp. 617-628.
pp. 4-14. doi: 10.1061/(ASCE)1090-0268(2007)11:1(4) Michaluk, C. R.; Rizkalla, S. H.; Tadros, G.; and Benmokrane, B.,
Bischoff, P. H.; Gross, S.; and Ospina, C. E., 2009, “The Story Behind 1998, “Flexural Behavior of One-Way Concrete Slabs Reinforced by Fiber
Proposed Changes to ACI 440 Deflection Requirements for FRP-Rein- Reinforced Plastic Reinforcements,” ACI Structural Journal, V. 95, No. 3,
forced Concrete,” Serviceability of Concrete Members Reinforced with May-June, pp. 353-364.
Internal/External FRP Reinforcement, SP-264, C. Ospina, P. Bischoff, Mota, C.; Alminar, S.; and Svecova, D., 2006, “Critical Review of
and T. Alkhrdaji, eds., American Concrete Institute, Farmington Hills, MI, Deflection Formulas for FRP-RC Members,” Journal of Composites
pp. 53-76. for Construction, ASCE, V. 10, No. 3, June, pp. 183-194. doi: 10.1061/
CSA S6:19, 2019, “Canadian Highway Bridge Design Code,” CSA (ASCE)1090-0268(2006)10:3(183)
Group, Toronto, ON, Canada, 1182 pp. Nanni, A.; Henneke, M. J.; and Okamoto, T., 1994, “Tensile Properties
CSA S806-12 (R2017), 2017, “Design and Construction of Building of Hybrid Rods for Concrete Reinforcement,” Construction and Building
Structures with Fibre-Reinforced Polymers,” CSA Group, Toronto, ON, Materials, V. 8, No. 1, pp. 27-34. doi: 10.1016/0950-0618(94)90005-1
Canada. Nguyen, P. D.; Dang, V. H.; and Vu, N. A., 2020, “Performance of
Duic, J.; Kenno, S.; and Das, S., 2018, “Performance of Concrete Concrete Beams Reinforced with Various Ratios of Hybrid GFRP/
Beams Reinforced with Basalt Fibre Composite Rebar,” Construc- Steel Bars,” Civil Engineering Journal, V. 6, No. 9, pp. 1652-1669. doi:
tion and Building Materials, V. 176, July, pp. 470-481. doi: 10.1016/j. 10.28991/cej-2020-03091572
conbuildmat.2018.04.208 Pang, L.; Qu, W.; Zhu, P.; and Xu, J., 2016, “Design Propositions for
El-Nemr, A.; Ahmed, E. A.; and Benmokrane, B., 2013, “Flexural Hybrid FRP-Steel Reinforced Concrete Beams,” Journal of Composites
Behavior and Serviceability of Normal- and High-Strength Concrete Beams for Construction, ASCE, V. 20, No. 4, Aug., p. 04015086. doi: 10.1061/
Reinforced with Glass Fiber-Reinforced Polymer Bars,” ACI Structural (ASCE)CC.1943-5614.0000654
Journal, V. 110, No. 6, Nov.-Dec., pp. 1077-1088. Qin, R.; Zhou, A.; and Lau, D., 2017, “Effect of Reinforcement Ratio
El-Salakawy, E., and Benmokrane, B., 2004, “Serviceability of Concrete on the Flexural Performance of Hybrid FRP Reinforced Concrete Beams,”
Bridge Deck Slabs Reinforced with Fiber-Reinforced Polymer Composite Composites Part B: Engineering, V. 108, Jan., pp. 200-209. doi: 10.1016/j.
Bars,” ACI Structural Journal, V. 101, No. 5, Sept.-Oct., pp. 727-736. compositesb.2016.09.054
El-Salakawy, E. F.; Kassem, C.; and Benmokrane, B., 2003, “Flex- Qu, W.; Zhang, X.; and Huang, H., 2009, “Flexural Behavior of Concrete
ural Behaviour of Bridge Deck Slabs Reinforced with FRP Composite Beams Reinforced with Hybrid (GFRP and Steel) Bars,” Journal of
Bars,” Fibre-Reinforced Polymer Reinforcement for Concrete Structures: Composites for Construction, ASCE, V. 13, No. 5, Oct., pp. 350-359. doi:
Proceedings of the Sixth International Symposium on FRP Reinforcement 10.1061/(ASCE)CC.1943-5614.0000035
for Concrete Structures (FRPRCS-6), K. H. Tan, ed., World Scientific Ruan, X.; Lu, C.; Xu, K.; Xuan, G.; and Ni, M., 2020, “Flexural Behavior
Publishing, Singapore, pp. 1291-1300. and Serviceability of Concrete Beams Hybrid-Reinforced with GFRP Bars
El-Sayed, A.; El-Salakawy, E.; and Benmokrane, B., 2005, “Shear Strength and Steel Bars,” Composite Structures, V. 235, Mar., Article No. 111772.
of One-Way Concrete Slabs Reinforced with Fiber-Reinforced Polymer doi: 10.1016/j.compstruct.2019.111772
Composite Bars,” Journal of Composites for Construction, ASCE, V. 9, Safan, M. A., 2013, “Flexural Behavior and Design of Steel-GFRP Rein-
No. 2, Apr., pp. 147-157. doi: 10.1061/(ASCE)1090-0268(2005)9:2(147) forced Concrete Beams,” ACI Materials Journal, V. 110, No. 6, Nov.-Dec.,
El-Sayed, A. K.; El-Salakawy, E. F.; and Benmokrane, B., 2006a, “Shear pp. 677-686.
Strength of FRP-Reinforced Concrete Beams without Transverse Rein- Shield, C.; Brown, V.; Bakis, C. E.; and Gross, S., 2019, “A Recalibra-
forcement,” ACI Structural Journal, V. 103, No. 2, Mar.-Apr., pp. 235-243. tion of the Crack Width Bond-Dependent Coefficient for GFRP-Reinforced
El-Sayed, A. K.; El-Salakawy, E. F.; and Benmokrane, B., 2006b, “Shear Concrete,” Journal of Composites for Construction, ASCE, V. 23, No. 4,
Capacity of High-Strength Concrete Beams Reinforced with FRP Bars,” Aug., p. 04019020. doi: 10.1061/(ASCE)CC.1943-5614.0000978
ACI Structural Journal, V. 103, No. 3, May-June, pp. 383-389. Sun, Z.; Fu, L.; Feng, D.-C.; Vatuloka, A. R.; Wei, Y.; and Wu, G., 2019,
El Refai, A.; Abed, F.; and Al-Rahmani, A., 2015, “Structural Perfor- “Experimental Study on the Flexural Behavior of Concrete Beams Rein-
mance and Serviceability of Concrete Beams Reinforced with Hybrid forced with Bundled Hybrid Steel/FRP Bars,” Engineering Structures,
(GFRP and Steel) Bars,” Construction and Building Materials, V. 96, Oct., V. 197, Oct., Article No. 109443. doi: 10.1016/j.engstruct.2019.109443
pp. 518-529. doi: 10.1016/j.conbuildmat.2015.08.063 Sun, Z.-Y.; Wu, G.; Wu, Z.-S.; and Zhang, M., 2011, “Seismic Behavior
Elgabbas, F.; Ahmed, E. A.; and Benmokrane, B., 2016, “Experimental of Concrete Columns Reinforced by Steel-FRP Composite Bars,” Journal
Testing of Concrete Bridge-Deck Slabs Reinforced with Basalt-FRP of Composites for Construction, ASCE, V. 15, No. 5, Oct., pp. 696-706. doi:
Reinforcing Bars under Concentrated Loads,” Journal of Bridge Engi- 10.1061/(ASCE)CC.1943-5614.0000199
neering, ASCE, V. 21, No. 7, July, p. 04016029. doi: 10.1061/(ASCE) Sun, Z. Y.; Yang, Y.; Qin, W. H.; Ren, S. T.; and Wu, G., 2012, “Exper-
BE.1943-5592.0000892 imental Study on Flexural Behavior of Concrete Beams Reinforced by
Ferrier, E.; Michel, L.; Zuber, B.; and Chanvillard, G., 2015, “Mechan- Steel-Fiber Reinforced Polymer Composite Bars,” Journal of Rein-
ical Behaviour of Ultra-High-Performance Short-Fibre-Reinforced forced Plastics and Composites, V. 31, No. 24, Dec., pp. 1737-1745. doi:
Concrete Beams with Internal Fibre Reinforced Polymer Bars,” Compos- 10.1177/0731684412456446
ites Part B: Engineering, V. 68, Jan., pp. 246-258. doi: 10.1016/j. Tureyen, A. K., and Frosch, R. J., 2002, “Shear Tests of FRP-Reinforced
compositesb.2014.08.001 Concrete Beams without Stirrups,” ACI Structural Journal, V. 99, No. 4,
Ge, W.; Wang, Y.; Ashour, A.; Lu, W.; and Cao, D., 2020, “Flexural July-Aug., pp. 427-434.
Performance of Concrete Beams Reinforced with Steel–FRP Composite Wang, X.; Liu, S.; Shi, Y.; Wu, Z.; and He, W., 2022, “Integrated
Bars,” Archives of Civil and Mechanical Engineering, V. 20, No. 2, June, High-Performance Concrete Beams Reinforced with Hybrid BFRP and
Article No. 56. doi: 10.1007/s43452-020-00058-6 Steel Bars,” Journal of Structural Engineering, ASCE, V. 148, No. 1, Jan.,
Ge, W.; Zhang, J.; Cao, D.; and Tu, Y., 2015, “Flexural Behaviors of p. 04021235. doi: 10.1061/(ASCE)ST.1943-541X.0003207
Hybrid Concrete Beams Reinforced with BFRP Bars and Steel Bars,” Wu, G.; Dong, Z.-Q.; Wang, X.; Zhu, Y.; and Wu, Z.-S., 2015, “Predic-
tion of Long-Term Performance and Durability of BFRP Bars under the
@seismicisolation
@seismicisolation

Combined Effect of Sustained Load and Corrosive Solutions,” Journal Experimental and Numerical Methods,” Engineering Structures, V. 206,
of Composites for Construction, ASCE, V. 19, No. 3, June, pp. 1-9. doi: Mar., Article No. 110117. doi: 10.1016/j.engstruct.2019.110117
10.1061/(ASCE)CC.1943-5614.0000517 Yang, Y.; Sun, Z.-Y.; Wu, G.; Cao, D.-F.; and Pan, D., 2020, “Experi-
Wu, G.; Sun, Z. Y.; Wu, Z. S.; and Luo, Y. B., 2012, “Mechanical Prop- mental Study of Concrete Beams Reinforced with Hybrid Bars (SFCBs and
erties of Steel-FRP Composite Bars (SFCBs) and Performance of SFCB BFRP Bars),” Materials and Structures, V. 53, No. 4, Aug., Article No. 77.
Reinforced Concrete Structures,” Advances in Structural Engineering, doi: 10.1617/s11527-020-01514-8
V. 15, No. 4, Apr., pp. 625-635. doi: 10.1260/1369-4332.15.4.625 Yinghao, L., and Yong, Y., 2013, “Arrangement of Hybrid Rebars on
Wu, Z.; Fahmy, M. F. M.; and Wu, G., 2009, “Safety Enhancement Flexural Behavior of HSC Beams,” Composites Part B: Engineering, V. 45,
of Urban Structures with Structural Recoverability and Controllability,” No. 1, Feb., pp. 22-31. doi: 10.1016/j.compositesb.2012.08.023
Journal of Earthquake and Tsunami, V. 3, No. 3, Sept., pp. 143-174. doi: Yoo, D.-Y.; Banthia, N.; and Yoon, Y.-S., 2016, “Flexural Behavior of
10.1142/S1793431109000561 Ultra-High-Performance Fiber-Reinforced Concrete Beams Reinforced
Xingyu, G.; Yiqing, D.; and Jiwang, J., 2020, “Flexural Behavior Inves- with GFRP and Steel Rebars,” Engineering Structures, V. 111, Mar.,
tigation of Steel-GFRP Hybrid-Reinforced Concrete Beams Based on pp. 246-262. doi: 10.1016/j.engstruct.2015.12.003
@seismicisolation
@seismicisolation

Title No. 120-S87
Failure Mode-Dependent Behavior of Carbon Fiber-

Reinforced Polymer Prestressed Concrete Girders
by Yail J. Kim, Jun Wang, Woo-Tai Jung, Jae-Yoon Kang, and Jong-Sup Park
This paper presents the characteristics of concrete girders in aggressive service environments. Furthermore, owing to
prestressed with carbon fiber-reinforced polymer (CFRP) tendons restricted budgets and resources, the demand for sustainable
when subjected to compression- and tension-controlled failure materials is commonplace in federal, state, and municipal
modes. To understand the full-range behavior of these girders agencies.
beyond the boundary of design specifications, progressive failure
Carbon fiber-reinforced polymer (CFRP) tendons may
is simulated using an advanced computational approach, agent-
be employed for prestressed concrete application due to
based modeling. Five bulb-tee girder sections are adopted with a
variable amount of CFRP tendons and their flexural responses are several benefits: corrosion resistance, high strength and light
examined until the intended failure modes are accomplished. The weight, nonconducting and nonmagnetic characteristics, low
rate of capacity reductions in the compression-controlled sections is relaxation, and tailorable formation.6 Findings from labora-
governed by the degree of concrete crushing in the upper flange and tory research were integrated into design specifications6,7
the depth of the girder, whereas the rate in the tension-controlled and numerous highway bridges have been erected with
girders is dominated by the sequential rupture of CFRP without CFRP-prestressed concrete girders around the world.8-10 The
demonstrating size dependency. When the girder concrete cracks, failure mode of CFRP-prestressed concrete girders plays a
locally unstable responses are observed in the compression-con- crucial role in the context of safety. Because both tension-
trolled sections, which are not noticed in the other sections. As far and compression-controlled sections are allowed,6,7 the
as deformability is concerned, both girder configurations are satis-
selection of a certain failure mode is largely dependent on
factory. Upon initiation of the progressive failure processes, the
practitioners’ discretion. Technically speaking, if concrete-
level of safety varies differently depending upon the girder type and
the arrangement of the tendons. The tension-controlled sections crushing governs, the girder suffers progressive failure in
require more activation energy, representing a transition rate from conjunction with sequential crushing across its section until
the initial to damaged states, than their compression counterparts. force equilibrium is not achieved; on the other hand, if CFRP
rupture dominates, the girder collapses as soon as tensile
Keywords: carbon fiber-reinforced polymer (CFRP); failure mode; full- strains exceed the ultimate strain of the tendons. The pros
range behavior; prestressed concrete; safety. and cons of those sections are currently inconclusive, and
the research community has been debating the effectiveness
INTRODUCTION of the design approaches over decades.11 From a cursory
Civil infrastructure constitutes the backbone of a nation’s point of view, compression-controlled sections appear to be
economy and is instrumental in the operation of modern safer than tension-controlled sections; contrarily, contem-
society. Aligning with the practical needs for controlled plating the high tensile strength of CFRP, expected safety in
quality, low maintenance, and efficient geometry, prestressed the tension-controlled sections may be compatible with that
concrete bridges are prevalent and account for a significant of the compression-controlled sections. These unapparent
portion of highway systems. For example, the National aspects are the sources of the foregoing arguments on the
Bridge Inventory of the Federal Highway Administration implications of the failure modes.
reports that 66.5% and 57.9% of constructed bridges in Assessments on the flexural response of CFRP-prestressed
Florida and New York were built with prestressed concrete, concrete members, particularly for compression-controlled
respectively.1 Despite such favorable advantages and popu- sections, are circumscribed by articles that stipulate the
larity, durability remains one of the most critical problems maximum usable strains of concrete (εcu = 0.003 and 0.0035
when managing prestressed concrete members, epitomized in ACI 440.4R-046 and SIMTReC,7 respectively), which
by recent statistics stating that an annual budget of $84,500 were empirically assumed for design purposes12 and do not
to $111,800 was spent per bridge for the last 10 years in the mean the actual failure of the members. To expand the scope
United States.2 Because hot-rolled steel outperforms cold- of investigations over the physical collapse of CFRP-pre-
formed steel from a material standpoint,3 prestressing strands stressed concrete members, an in-depth understanding of
comprising multiple wires are more susceptible to failure in full-range behavior is indispensable beyond the prescribed
comparison with reinforcing bars.4 The collapse of a precast
bridge in Lakeview Drive, Washington, PA, was attributed ACI Structural Journal, V. 120, No. 5, September 2023.
to a combination of excessive spalling and corrosion damage reviewed under Institute publication policies. Copyright © 2023, American Concrete
in the bottom flange of a box girder. Attention should thus
5 Institute. All rights reserved, including the making of copies unless permission is
be paid to the deterioration of prestressed concrete girders closure, if any, will be published ten months from this journal’s date if the discussion
@seismicisolation
@seismicisolation

Fig. 1—Design of bulb-tee girders: (a) compression-controlled section; (b) tension-controlled section; and (c) variable girder
size. (Note: Units in mm.)
Table 1—Summary of default sections
Compression-controlled section Tension-controlled section
Two layers of CFRP Three layers of CFRP Four layers of CFRP
ID H, mm df, mm Mn, kN·m df, mm Mn, kN·m df, mm Mn, kN·m df, mm Mn, kN·m
BT42 1067 843 7943 991 4720 982 4689 969 4367
BT54 1372 1148 11,535 1295 6427 1287 6389 1274 6231
BT63 1600 1376 14,271 1524 7571 1516 7533 1503 7394
BT72 1829 1605 17,072 1753 8790 1744 8752 1731 8694
BT84 2134 1910 20,786 2057 10,364 2049 10,326 2036 10,267
Note: H is girder depth; df is effective depth of tendons; Mn is nominal flexural capacity; compression-controlled section failure mode is crushing of concrete; tension-controlled
section failure mode is rupture of CFRP; 1 mm = 0.0394 in.; 1 kN·m = 0.738 kip∙ft.
limits. In this paper, an advanced modeling approach is used MODELING

to elucidate the detailed failure mechanisms of compression- Outlined in the following are the description of a theoret-
and tension-controlled CFRP-prestressed concrete girders ical framework to predict the behavior of archetypal bridge
and ensuing outcomes that can quantify the performance and girders used in the field, the background and implementation
vulnerability associated with each failure mode. of a computational platform, and the numerical representa-
tion of failure modes.
Unlike conventional steel-prestressed concrete girders, Benchmark girders
CFRP-prestressed members are designed with an intended Pursuant to published guidelines and manuals,6,13,14
failure mode of either concrete crushing or CFRP rupture. bulb-tee girders were designed (Fig. 1). The height of the
During the course of pursuing a project, bridge engineers girders varied from H = 1067 to 2134 mm (42 to 84 in.),
choose one of these two possible options; however, there is which were designated BT42 to BT84 (Table 1), and each
a concern that their selection is largely based on previous one involved 54 and 24 CFRP tendons for compression-
experiences without deliberating adverse impacts when and tension-controlled sections, respectively. The place-
the members encounter catastrophic situations, which are ment of the tendons was constant in the bottom flange when
outside the coverage of specifications. Provided that the concrete crushing was the primary failure mode (Fig. 1(a)),
process of a structural collapse entails a substantial redis- whereas two to four layers of CFRP were arrayed to generate
tribution of stresses from the origination of damage to the progressive rupture of the tendons before the concrete
adjacent regions, the progression of sequential failure and crushed (Fig. 1(b)). In line with ordinary practices in precast
corresponding consequences ought to be documented prop- plants, the 28-day compressive strength of the girder concrete
erly. Nonetheless, knowledge in the subject discipline is was set to fc′ = 65 MPa (9427 psi). The diameter of the CFRP
incomplete at present and clarifications are essential. By tendons was db = 12.7 mm (0.5 in.) and the manufacturer-re-
comprehending the repercussions of the complex failure ported tensile strength and modulus were ffu = 1724 MPa
mechanisms, effective design strategies can be established (250 ksi) and Ef = 124 GPa (18,000 ksi), respectively, with a
alongside procedural improvements and a paradigm shift in rupture strain of εfu = 0.0139. Tendon stresses at jacking and
bridge engineering with CFRP-prestressed concrete girders. after transfer were ffj = 0.65ffu and ffi = 0.6ffu, respectively,
and the effective stress was ffe = 0.52ffu.6 Given that the
@seismicisolation
@seismicisolation

Fig. 2—Agent-based modeling: (a) visualized concept; and (b) overview of modeling.
objective of this study was to examine the failure-dependent agents shared compatible strains and transferred element-
behavior of CFRP-prestressed concrete girders, a deck slab level stresses and forces so as to reproduce the development
was not included. Table 1 enumerates the flexural capacities of a girder curvature. The stress propagation of the model
of the girders (Mn) and, as expected, the compression-con- continued, and resulting forces in the modularized agents
trolled sections demonstrated higher resistance by 88.4% in were summed to monitor the status of equilibrium and to
comparison with the tension-controlled sections, on average. characterize the flexural behavior of the girders, contingent
upon failure mode, as illustrated in Fig. 2(b). The interde-
Agent-based modeling pendence of the agents was tied with those simple rules, and
Concept—Agent-based modeling is a contemporary all collective patterns at the global level were translated into
simulation technique, which is prevalent in the discipline the macroscopic behavior of the prestressed girders.
of social science, and its general objective is to emulate
complex interactions among human beings sharing common Validation
backgrounds, interests, physical activities, and spaces.15,16 The developed approach was validated using test data and
Concise descriptions are provided herein, while further existing equations. Figure 3(a) compares the flexural capac-
details on the principles and implementation methodologies ities of experimental beams21-24 and their modeling counter-
are available elsewhere. The modeling platform is intrinsi-
17
parts. The tensile strength and modulus of CFRP tendons
cally a bottom-up approach that comprises a cohort of auton- embedded in the laboratory beams (fc′ = 37 to 63 MPa
omous entities called agents. The individual components [5366 to 9137 psi]) were ffu = 1882 to 2275 MPa (273 to
engage one another without a centralized decision-making 330 ksi) and Ef = 131 to 170 GPa (19,000 to 24,656 ksi),
process18—that is, the heterogeneous behavior of each agent respectively. The prestressing level of CFRP was 0.5 to
is controlled by self-organization in local territory. Accord- 0.65ffu. The predicted capacities agreed with the measured
ingly, an agent’s response is dominated by the reactions of ones at an average margin of 9.1%. Shown in Fig. 3(b) are
neighboring agents. The most notable difference between the predicted flexural capacities of the benchmark girders
agent-based and traditional analytical models can be found (Table 1) in relation to the capacities calculated by the
in their focus levels19: the former is built upon individual analytical equations proposed by Peng and Xue,11 which can
interactions; by contrast, the latter intends to directly attain cover both compression- and tension-controlled sections.
the solution of a homogeneous system. The coefficient of determination of R2 = 0.9813 corroborates
Formulation—NetLogo, an open-source programing the adequacy of the modeling method.
language,20 was employed to execute the theory of agent-
based modeling. On the two-dimensional graphical user Sequential failure
interface linked with command tabs, a grid space was created After validation, the default sections depicted in Fig. 1
to computationally reconstruct the benchmark bridge girders were subjected to sequential failure that was required to
(Fig. 2(a)). The size of the agents was 6.35 mm (0.25 in.), simulate the full-range behavior of the girders. For the
and these discrete entities, positioned at the orthogonal compression-controlled sections (Fig. 4(a)), the strains of the
lines of the grid, were uniformly distributed in the spatial extreme compression fiber were calculated at every iteration
domain of the girders. The respective agents were mutually (Fig. 2(b)), and the values were compared with the crushing
adaptive in conformity with a reactive protocol: adjoining strain of the concrete (εcu = 0.003). If the maximum strain
@seismicisolation
@seismicisolation

Fig. 3—Validation of modeling approach: (a) experimental (A. and R. is Abdelrahman and Rizkalla21; S. is Saeed22; D. and S.
is Dolan and Swanson23; M. et al. is Mertol et al.24); and (b) analytical.
Fig. 4—Schematic of sequential failure modes: (a) crushing of concrete; and (b) rupture of CFRP.
in a layer of the agents was greater than the threshold limit, on average (Fig. 5(b)). For the replication of possible live
the layer was deemed to fail and the next layer was reset to load intensities during the course of girder failure, four live
be the ultimate compression fiber with a reduced height of load factors (αL) were adopted from published literature,26,27
the girder. This procedure was repeated until the compres- which were then coupled with the performance levels elabo-
sion zone reached the haunch depth of the girder. For the rated in NCHRP 44028: Fully Operational (FO) at αL = 1.75,
tension-controlled sections (Fig. 4(b)), the same procedure Operational (OP) at αL = 1.00, Life Safety (LS) at αL = 0.50,
was applied, except when a tensile strain of CFRP went over and Near Collapse (NC) and αL = 0.25. The physical inter-
the ultimate strain (εf ≥ εfu): the CFRP layer was regarded pretation of these performance indicators is as follows: FO
to fail without concrete crushing (εc < εcu, where εc is the is a full design load, OP is the typical service state, LS is
compressive strain of concrete), which proceeded until all an extreme event, and NC is a critical situation. Figure 5(c)
CFRP tendons ruptured. provides the distributions of the factored load effects (ME)
versus the factored resistance (MR) of the girders, where
Characterization of load effects the dead load factors of αD = 1.25 (structural components)
To evaluate the performance safety of the benchmark and αW = 1.50 (wearing surface) and the strength resistance
girders under the aforementioned failure scenarios, a rela- factor27 of ϕ = 1.0 were associated with the predefined αL
tionship was characterized between variable load effects and factors. The average ratios of the applied loads and resis-
girder resistance. Taken from actual bridge design projects in tance are charted in Fig. 5(d): Case I with the factored dead
Colorado, Illinois, Michigan, Pennsylvania, and Texas,25 the loads is intended to focus on repair design, whereas Case II
unfactored dead and live load moments (Fig. 5(a)) and the with the unfactored dead loads can be used to infer the load
nominal resistance (Fig. 5(b)) of prestressed concrete girders effects (fractions of girder resistance) representing specific
were plotted. The span length of the sampled girders ranged performance levels for safety assessment.
from L = 12 to 42 m (39 to 138 ft). The dead load of the
girders (all structural components plus wearing surface: MDL RESULTS
and MDW, respectively) generated a 327% higher moment The ramifications of compression- and tension-con-
than the live load (MLL) involving impact factors (IM) and trolled sections are delineated for the full-range behavior of
dynamic load allowance (DLA), on average (Fig. 5(a)). CFRP-prestressed concrete girders. Emphasis is placed on
The nominal resistance of the girders (Mn) was greater than failure particulars, flexural responses, safety appraisals, and
the unfactored load effects (MDL + MDW + MLL) by 53.6%, activation energy with regard to the transition of a girder state.
@seismicisolation
@seismicisolation

Fig. 5—Characterized relationship between load effects and resistance based on actual bridge design projects in Colorado,
Illinois, Michigan, Pennsylvania, and Texas: (a) dead and live moments; (b) load effects versus resistance; (c) variable live
load intensities; and (d) ratio of load effects to resistance.
Failure characteristics the compression-controlled case, the capacity variation was
Figure 6(a) shows capacity reductions in the compres- independent of the girder size (Fig. 7(a), inset). The reason
sion-controlled girders with progressive failure. As the is explained by the fact that the entire upper flange, primarily
crushing depth of the concrete increased, the capacity resisting the compressive stress, was not impaired until
steadily declined in all instances. The descending rate was the complete failure of the tendons took place and that the
rapid up to approximately 50 mm (2 in.) when the damage of progressive rupture of the closely spaced CFRP tendons in
the upper flange initiated and proceeded, beyond which the the vertical direction was an insignificant determinant for
response tended to stabilize. The flange played an important changing the lever arm. Shown in Fig. 7(b) is the impact of
role in resisting compressive stress that generated a resul- the sequential failure of CFRP. While the normalized girder
tant force counteracting the tensile component for main- capacity with two CFRP layers plummeted in a linear manner,
taining equilibrium. Although the dimension of the flange the downtrend lines of the girders with three and four layers
was identical across the board (Fig. 1(a)), the evolution of were mitigated because the remaining tendons carried stresses
the capacity decrease was size-dependent (Fig. 6(b)): the transferred from the previously ruptured tendons. It is thus
capacities of BT42 and BT84 at haunch-level crushing with stated that the vertical distribution of CFRP is desirable for
a depth of 189 mm (7.4 in.) were 58.4% and 71.3% of the tension-controlled sections and should have as many layers
initial capacities, respectively. This observation is ascribed as possible to avoid a catastrophic collapse of the prestressed
to the reliance of the lever arm between the compression and girders. The neutral axis depth of the BT42 to BT84 girders
tension forces on the girder depth; in other words, the neutral gradually descended as the rupture depth went up (Fig. 7(c));
axis location of the stocky girders was altered susceptibly by mechanically saying, less concrete was needed for responding
the concrete crushing, relative to that of the deeper girders to the lowered tensile components. Owing to the equilibrium
(Fig. 6(c)). The amount of the girder concrete was lessened requirements discussed earlier, the tendency of the neutral
with the elevated crushing depth (Fig. 6(d), inset); however, axis depth normalized by the girder depth appreciably differed
the loss of the concrete area above the haunch level did between the compression- and tension-controlled sections
not affect the normalized capacity of the girders (Fig. 6(d)) (Fig. 7(c), inset). Figure 7(d) illustrates the significance of
because the repositioned neutral axis necessitated more CFRP-area reductions, which reaffirms the betterment of the
concrete in the web that brought to a balance with the CFRP vertically distributed tendon layers in terms of preserving the
tendons. load-bearing ability of the girders.
The moment capacity of the tension-controlled sections
also dwindled due to the rupture of CFRP (Fig. 7(a)).@seismicisolation
Unlike
@seismicisolation

Fig. 6—Failure of compression-controlled sections: (a) flexural capacity; (b) normalized capacity; (c) normalized neutral axis
depth; and (d) concrete-area reduction.
Fig. 7—Failure of tension-controlled sections: (a) flexural capacity with four CFRP layers; (b) normalized capacity for BT84;
(c) neutral axis depth with four CFRP layers; and (d) CFRP-area reduction.
@seismicisolation
@seismicisolation

Fig. 8—Moment-strain relationship: (a) compression-controlled sections; (b) tension-controlled sections with two CFRP
layers; (c) normalized moment development of compression-controlled sections; and (d) normalized moment development of
tension-controlled sections with four CFRP layers.
Strain variation Moment-curvature
The moment-strain relationship of the girders is graphed Figure 9(a) demonstrates the moment-curvature rela-
in Fig. 8. For the compression- and tension-controlled tionship of the compression-controlled sections. The
sections, the girder size influenced concrete and CFRP aforementioned momentary diminution was conspicuous
strains (Fig. 8(a) and (b)). That is, as the depth of the girders in the ordinate, at which a bifurcation occurred when the
decreased, more sectional rotations were allowed on account concrete cracked, and the slope of the curves escalated with
of the low moment of inertia; consequently, the strain devel- the increased girder depth. As far as the tension-controlled
opment became pronounced. The sensitive computational sections are concerned (Fig. 9(b)), the pre-cracking stiffness
model captured the instantaneous drop of the moment in the was akin to that of the preceding sections; on the contrary,
compression-controlled girders (ρ = 2.4 to 5.4%, where ρ is the post-cracking stiffness was much lower because of the
the reinforcement ratio based on the web thickness) when relatively short neutral axis depth (Fig. 7(c)) concomitant
the concrete cracked (Fig. 8(a)), which is not available in with the noticeable cracking of the sections. The three-stage
conventional sectional analysis. These locally unstable behavior of the tension-controlled girder is epitomized in
responses were attributed to a sudden decrease in the stiffness Fig. 9(c), where BT42 is used as a representative sample:
of the girders incorporating the low-modulus CFRP mate- 1) the first stage with the uncracked concrete subjected to
rial.29 Regarding the tension-controlled girders (Fig. 8(b)), service loading (0 ≤ Mnor ≤ 0.67); 2) the second stage with
the abrupt local indentation was not obvious because the the cracked concrete up to the ultimate moment (0.67 <
transition of the moment of inertia before and after cracking Mnor ≤ 1.0); and 3) the third stage with the successive rupture
was marginal in the lightly reinforced sections (for example, of the tendons in the post-peak region. It is worth noting
ρ = 0.98% for BT84). Analogous to the normalized capaci- that the increment of the normalized curvature was minimal
ties given in Fig. 6(b) and 7(a), the propensity for the strain until the normalized moment plunged to Mnor = 0.06;
growth was reliant upon the amount of the CFRP tendons. namely, the contribution of the multilayered CFRP tendons
Under the same moment level normalized by the maximum to the modification of the girder’s curvature was negligible
moment (Mnor), BT84 exhibited more strains than BT42 for (supplementary discussions to follow). The area under the
the compression-controlled sections (Fig. 8(c)); conversely, moment-curvature curves of all girders was numerically
there was no distinguishable facet for the tension-controlled integrated, which was designated as the “characteristic
sections (Fig. 8(d)). area” in Fig. 9(d). Irrespective of girder size, the compres-
sion-controlled sections revealed higher characteristic areas
@seismicisolation
@seismicisolation

Fig. 9—Moment-curvature relationship: (a) compression-controlled sections; (b) tension-controlled sections with two CFRP
layers; (c) stepwise failure of tension-controlled BT42 section with four CFRP layers; and (d) characteristic area.
than the tension-controlled ones, leading to a physical inter- and εfe is the effective CFRP strain (εfe = ffe/Ef). Figure 10
pretation that the amount of the former’s internal energy charts the deformability indexes of the prestressed girders.
was greater and conveyed better to the constituent materials A general trend is that the indexes of the compression-con-
until a dissipation process was completed at the time of the trolled sections were less than those of the tension-controlled
girder failure. The almost constant characteristic areas of the sections; however, both of them in Fig. 10(b) exceeded the
tension-controlled sections with two to four CFRP layers suggested limits (DI = 1.5 and 1.8 for the compression- and
verify the ignorable impact of the progressive rupture of the tension-controlled sections, respectively30). This points out
tendons on the sectional deformation of the girders. that all sections were technically adequate from a design
perspective (it is noted that there is no limit for Eq. (1); thus, it
Deformability cannot be used for practice, as criticized in Kim and Nickle30).
Because the CFRP tendons and concrete are brittle in
nature, the traditional concept of ductility is not applicable.6 Sectional response
As an alternative to quantify the flexural performance of Figure 11 displays the sectional response of the compres-
CFRP-prestressed members, a deformability index (DI) may sion- and tension-controlled girders. For comparison, selected
be calculated6,30 attributes were normalized by their maximum values, except
for neutral axis depths, which were normalized by the girder
_ (1 − k) _ εfu depths (Table 1). With an increase in the moment, the neutral
DI = (1)
1 − α /(dfβ1)εfs axis depth of the entire girder series steadily declined until
a normalized moment of approximately Mnor = 0.85 was
⎧ k εcu
_ reached (Fig. 11(a)), outside of which distinctions were prom-
⎪1 + ε for compression-controlled section inent due to the abrupt failure of the tendons (Fig. 11(a),
DI = ⎨
fe
    
ε

(2) inset). The curvature response of the two girder types was
⎪ _
⎩ εfe
cu
for tension-controlled section
similar up to Mnor = 0.81 (Fig. 11(b)); then, the cracked
section of the compression-controlled girder showed tran-
where k is the ratio of the neutral axis depth to the effective- sient instability. The steeply rising curvature of the girders
ness depth (k = c/df); α is the constant (α = ρdfffu/(0.85fc′); ρ again diverged immediately after the normalized moment
is based on the flange width6; β1 is the concrete stress block reached Mnor = 1.0 by virtue of the CFRP rupture. As a result
factor specified in ACI 318-1912; εfs is the service strain of of restrictive concrete deformations at the top of the girders
CFRP, which can be approximated to be a prestressing strain6; within the maximum usable boundary of εcu = 0.003, the path
@seismicisolation
@seismicisolation

Fig. 10—Deformability index: (a) ACI 440.4R-046; and (b) Kim and Nickle.30
Fig. 11—Normalized sectional response of compression- and tension-controlled girders with moment development: (a) neutral
axis; (b) curvature; (c) concrete strain; and (d) CFRP strain.
of the normalized concrete strains was virtually unrelated to Safety assessment
the failure modes (Fig. 11(c)). Likewise, the strains of the Mensuration—The safety of the girders was estimated
most-tensioned CFRP near the bottom of the girders were using Eq. (3), indirectly expressing the probability of
alike from Mnor = 0 to 0.6 (Fig. 11(d)), implying that there failure,31 together with the extent of damage (Eq. (4))
should be no concerns about the premature failure of these
girder configurations under service loading. In excess of Mnor LN(MN /ME )
__________________
_________________
= 0.6, the CFRP strain of the compression-controlled girder
β =   
   (3)
√ (  
COVR )2+ (COVE )2
forged ahead (εf < εfu) relative to that of the tension-controlled
girder suffering the rupture of the tendons (εf = εfu).
Mn0
_ − MnD

Ω = M (4)
n0
@seismicisolation
@seismicisolation

Fig. 12—Safety index with progressive damage in compression-controlled sections: (a) damage level; (b) size effect; (c) perfor-
mance category at girder-top crushing; and (d) performance category at haunch crushing.
where β is the safety index appertaining to the actual distri- the design-level failure. As the crushing continued to the
bution of bridge loads and responses ; MN is the flexural
31 haunch location, the margin of safety remarkably decreased
resistance of the girder with and without damage (Fig. 6(a) from the limit (Fig. 12(d)). Therefore, strict traffic control
and 7(a)); ME is the load effect representing the characterized will be imperative under the NC category (αL = 0.25) if
four performance categories (Case II in Fig. 5(d)); COVR and concrete crushing penetrates through the upper flange of
COVE are the coefficients of variation for the MN and ME such CFRP-prestressed girders.
terms, respectively (referring to the literature ; COVR =
31,32 Rupture of CFRP—Even if the safety indexes of the
0.075 and COVE = 0.18 were assumed); Ω is the damage tension-controlled sections were the same as those of the
index; and Mn0 and MnD are the initial moment capacity and compression-controlled sections at Ω = 0 (no damage),
the damaged moment resistance, respectively. the brittle rupture of CFRP resulted in swift reductions
Crushing of concrete—The safety indexes of the compres- (Fig. 13(a)). As before, the indexes of the tension-controlled
sion-controlled sections consistently diminished with the sections at the initial stage were unrelated to the girder
accumulated damage (Fig. 12(a)). The intervals between the depth (Fig. 13(b)), and this trend was maintained when the
performance levels from NC to FO were retained without bottommost layer of the tendons failed (the damaged state
being affected by the damage index and, in addition, the indicated in Fig. 13(b)), which was different from the rising
safety responses were not engaged with the depth of the pattern of the compression-controlled sections (Fig. 12(b)).
girders. Figure 12(b) explicates the reliance of the girder The indexes linked with the succeeding tendon ruptures
safety on its depth. Before the occurrence of damage (the were negative and had no practical significance (not shown
initial state), the indexes were constant; per contra, the in Fig. 13(b)). The number of the vertically distributed
indexes with concrete crushing (the damaged state) ascended CFRP layers was a factor that adjusted the level of safety
as the depth increased. The use of a deeper section was thus (Fig. 13(a) and (b)). Specifically, at the moment of the first
beneficial in the sense of safety when subjected to compres- tendon rupture, the safety indexes of all performance cate-
sion failure. The influence of the progressive crushing is gories in the two-tendon girders were below the AASHTO
visible in Fig. 12(c) and (d). With the exception of the FO limit of 2.5 (Fig. 13(c)); however, the average index of the
category (full design load), the safety indexes of the girders NC category was 2.7 in the four-tendon girders (Fig. 13(d)).
undergoing concrete crushing at the top were higher than the Considering these safety features, CFRP tendons should
AASHTO limit of 2.5 for constructed bridges (Fig. 12(c)), 33 be arranged carefully if the intended failure mode of a
denoting that the girders were still functional in spite of prestressed girder is tension-controlled, which is vulnerable
in comparison with the compression-controlled case.
@seismicisolation
@seismicisolation

Fig. 13—Safety index with progressive damage in tension-controlled sections: (a) damage level with four CFRP layers; (b) size
effect with four CFRP layers; (c) performance category with two CFRP layers; and (d) performance category with four CFRP
layers.
Activation energy for the tension-controlled girders (Fig. 14(b)). Figure 14(c)
To definitize the dependency of the flexural resistance on contrasts the activation energy belonging to all girder series.
the girder types, the concept of Arrhenius theory, which is The shift of the intended failure modes from compression to
prevalent in physical chemistry, may be taken34 tension brought about the soaring of the activation energy,
signifying the susceptible reactivity of the tension-con-
Rr = e−Q/(kB T)
(5) trolled girders when the state of damage evolved. The degree
of changes in the activation energy of the girders was deter-
where Rr is the response rate; Q is the activation energy; kB mined by the secant slope of the resistance-energy curves,
is the Boltzmann constant (kB = 1.381 × 10–23 J/K); and T is connecting the initial state with the failure state, and is
the temperature in Kelvin (T = 298 K was used, equivalent rendered in Fig. 14(d). The slopes clearly demonstrate the
to 25°C [77°F]). Equation (5) was rearranged to solve for size-dependency of the activation energy: the reaction of
the activation energy in tandem with replacing the Rr term the smallest girder, BT42, was conspicuous, during which a
by the moment ratio of MnD/Mn0 transition was made from the undamaged to damaged states,
regardless of the failure modes.
Q = –ln(MnD/Mn0)(kBT) (6)
SUMMARY AND CONCLUSIONS
The physical meaning of the activation energy is that it This paper has investigated the full-range behavior
measures the transition rate of reactions between consecu- of carbon fiber-reinforced polymer (CFRP)-prestressed
tive states at a specific temperature35; scilicet, Eq. (6) can concrete girders that failed in compression and tension.
figure out a minimum amount of energy that is needed from Simulations were conducted using an advanced computa-
the initial state to certain damage levels of the girders. As tional platform, agent-based modeling, which handled decen-
described in Fig. 14(a) and (b), the downward flexural resis- tralized interactions between multiple entities. Employing
tance related to the increased damage raised the activation five bulb-tee sections (BT42 to 84) with a variable amount
energy. The progression of the activation energy was restric- of CFRP tendons, the consequences of progressive failure
tive for the compression-controlled girders up to the haunch- were expounded with the aim of quantifying the funda-
level crushing (Fig. 14(a)), whereas more activation energy mental hazard of those girders beyond design-level flexural
was necessary while the sequential ruptures were in progress responses. Technical interests lied in load-resisting abili-
ties, performance levels, strain development, deformability,
@seismicisolation
@seismicisolation

Fig. 14—Activation energy: (a) compression-controlled sections; (b) tension-controlled sections with four CFRP layers;
(c) comparison of activation energy at impending failure; and (d) comparison of secant slope at impending failure.
structural safety, and state transitions, depending upon the distributed CFRP tendons was a crucial factor that
failure modes. The following conclusions are drawn: governed the level of safety. Under the evolution of
• The capacity-reduction rate of the compression-con- damage, the tension-controlled girders demanded more
trolled girders was prompt when half of the upper-flange activation energy than the compression-controlled ones.
concrete crushed, whereas the rate became stable as the Without reference to the failure modes, the develop-
crushing further progressed. Owing to the relocation of ment of the activation energy was size-dependent.
neutral axis depth, the capacities of the damaged girder
normalized by the initial capacity were not impinged by AUTHOR BIOS
the loss of concrete above the haunch level. Different Yail J. Kim, FACI, is President of the Bridge Engineering Institute, An
International Technical Society, and a Professor in the Department of Civil
from the compression-controlled girders, the capacity Engineering at the University of Colorado Denver, Denver, CO. He is Chair
decrease of the tension-controlled girders was scale-in- of ACI Subcommittee 440-I, FRP-Prestressed Concrete; past Chair of ACI
variant. The sequential rupture of CFRP transferred Committee 345, Bridge Construction and Preservation; and a member of
ACI Committees 342, Evaluation of Concrete Bridges and Bridge Elements;
tensile stresses from one layer to another. 377, Performance-Based Structural Integrity & Resilience of Concrete
• The degree of sectional rotations was a function of the Structures; 440, Fiber-Reinforced Polymer Reinforcement; and Joint
girder depth and controlled the magnitude of CFRP ACI-ASCE Committee 343, Concrete Bridge Design. He received the ACI
Chester Paul Siess Award for Excellence in Structural Research in 2019.
strains. A change in the section stiffness with the His research interests include advanced composite materials for rehabilita-
low-modulus tendons induced local instability in the tion; structural informatics; complex systems; and science-based structural
compression-controlled girders at the time of cracking, engineering including statistical, interfacial, and quantum physics.
while this trend was not apparent in the tension-con- ACI member Jun Wang is a Postdoctoral Fellow in the Department of Civil
trolled girders by virtue of the dissimilar reinforcement Engineering at the University of Colorado Denver. She received her BS and
ratios. Both girder configurations satisfied deforma- MS from Northeast Forestry University, Harbin, China, and the Univer-
sity of Colorado Denver, respectively. She is a member of ACI Committee
bility requirements, and the premature failure of CFRP 345, Bridge Construction and Preservation. Her research interests include
was not expected under service loading. multi-object interaction, advanced modeling, and concrete structures.
• After the initiation of the sequential failure process, the
Woo-Tai Jung is a Senior Researcher in the Structural Engineering
safety indexes of the compression-controlled sections Research Division of the SOC Research Institute at the Korea Institute
rose with the increased girder depth; in contrast, the of Civil Engineering and Building Technology. He received his PhD from
indexes of the tension-controlled sections were not Myongji University, Seoul, South, Korea. His research interests include
strengthening deteriorated concrete structures with fiber-reinforced
concerned with the depth. The number of vertically polymer (FRP) composites.
@seismicisolation
@seismicisolation

Jae-Yoon Kang is a Research Fellow in the Structural Engineering 13. CDOT, “Bridge Design Manual,” Colorado Department of Transpor-
Research Division of the SOC Research Institute at the Korea Institute of tation, Denver, CO, 2012.
Civil Engineering and Building Technology. He received his MS and PhD 14. PCI, PCI Bridge Design Manual, third edition, Precast/Prestressed
from Dongguk University, Seoul, South, Korea. Concrete Institute, Chicago, IL, 2014.
15. Duffy, V. G., Digital Human Modeling and Applications in Health,
Jong-Sup Park is a Research Fellow in the Structural Engineering Safety, Ergonomics and Risk Management, Springer Nature, Cham, Swit-
Research Division of the SOC Research Institute at the Korea Institute of zerland, 2021.
Civil Engineering and Building Technology. He received his MS and PhD 16. Engel, U.; Quan-Haase, A.; Liu, S. X.; and Lyberg, L., Handbook of
from Myongji University. Computational Social Science, Routledge, Oxfordshire, UK, 2021.
17. Railsback, S. F., and Grimm, V., Agent-Based and Individual-Based
Modeling: A Practical Introduction, second edition, Princeton University
ACKNOWLEDGMENTS Press, Princeton, NJ, 2019.
The research was conducted as part of the Strategic Research Project of 18. DeAngelis, D. L., and Diaz, S. G., “Decision-Making in Agent-Based
the Korea Institute of Civil Engineering and Building Technology (KICT). Modeling: A Current Review and Future Prospectus,” Frontiers in Ecology
The authors thank the members of ACI Subcommittee 440-I, FRP-Pre- and Evolution, V. 6, 2019, p. 237. doi: 10.3389/fevo.2018.00237
stressed Concrete, for their discussions and valuable suggestions. To avoid 19. Gilbert, N., Agent-Based Models, second edition, SAGE Publica-
commercialism, all proprietary information, such as product names and tions, Inc., Thousand Oaks, CA, 2020.
manufacturers, was not included. The technical contents presented in this 20. Wilensky, U., and Rand, W., An Introduction to Agent-Based
paper are based on the opinion of the authors and do not necessarily repre- Modeling: Modeling Natural, Social, and Engineered Complex Systems
sent that of others. with NetLogo, MIT Press, Cambridge, MA, 2015.
21. Abdelrahman, A. A., and Rizkalla, S. H., “Serviceability of Concrete
REFERENCES Beams Prestressed by Carbon-Fiber-Reinforced-Plastic Bars,” ACI Struc-
1. FHWA, “National Bridge Inventory (NBI),” Federal Highway Admin- tural Journal, V. 94, No. 4, July-Aug. 1997, pp. 447-457.
istration, Washington, DC, 2022, https://www.fhwa.dot.gov/bridge/nbi/ 22. Saeed, Y. M., “Behavior of Prestressed Concrete Beams with CFRP
ascii.cfm. (last accessed Aug. 16, 2023) Strands,” MS thesis, Portland State University, Portland, OR, 2016.
2. Fiorillo, G., and Ghosn, M., “Risk-Based Life-Cycle Analysis of 23. Dolan, C. W., and Swanson, D., “Development of Flexural Capacity
Highway Bridge Networks under Budget Constraints,” Structure and of a FRP Prestressed Beam with Vertically Distributed Tendons,” Compos-
Infrastructure Engineering, V. 18, No. 10-11, 2022, pp. 1457-1471. doi: ites Part B: Engineering, V. 33, No. 1, 2002, pp. 1-6. doi: 10.1016/
10.1080/15732479.2022.2059525 S1359-8368(01)00053-1
3. Bamonte, P.; Kalaba, N.; and Felicetti, R., “Computational Study 24. Mertol, H. C.; Rizkalla, S.; Scott, P.; Lees, J. M.; and El-Hacha, R.,
on Prestressed Concrete Members Exposed to Natural Fires,” Fire Safety “Durability of Concrete Beams Prestressed with CFRP Bars,” Case Histo-
Journal, V. 97, 2018, pp. 54-65. doi: 10.1016/j.firesaf.2018.02.006 ries and Use of FRP for Prestressing Applications, SP-245, R. El-Hacha
4. Kotsovinos, P.; Judge, R.; Walker, G.; and Woodburn, P., “Fire and S. H. Rizkalla, eds., American Concrete Institute, Farmington Hills,
Performance of Structural Cables: Current Understanding, Knowledge MI, 2007, pp. 1-20.
Gaps, and Proposed Research Agenda,” Journal of Structural Engi- 25. Nowak, A. S., “Calibration of LRFD Bridge Design Code,” NCHRP
neering, ASCE, V. 146, No. 8, 2020, p. 03120002. doi: 10.1061/(ASCE) Report 368, National Academy of Sciences, Washington, DC, 1993.
ST.1943-541X.0002703 26. Ghosn, M.; Moses, F.; and Wang, J., “Design of Highway Bridges
5. Naito, C.; Sause, R.; Hodgson, I.; Pessiki, S.; and Desai, C., “Forensic for Extreme Events,” NCHRP Report 489, National Academy of Sciences,
Evaluation of Prestressed Box Beams from the Lake View Drive over I-70 Washington, DC, 2003.
Bridge,” Report No. FHWA-PA-2006-017-EMG002, Center for Advanced 27. AASHTO, “LRFD Bridge Design Specifications,” ninth edition,
Technology for Large Structural Systems, Lehigh University, Bethlehem, American Association of State Highway and Transportation Officials,
PA, 2006. Washington, DC, 2020.
6. ACI Committee 440, “Prestressing Concrete Structures with FRP 28. Marsh, M. L., and Stringer, S. J., “Performance-Based Seismic
Tendons (ACI 440.4R-04) (Reapproved 2011),” American Concrete Insti- Bridge Design (NCHRP Synthesis 440),” Transportation Research Board,
tute, Farmington Hills, MI, 2004, 35 pp. Washington, DC, 2013.
7. Campbell, I., “Prestressing Concrete Structures with Fibre-Reinforced 29. Karayannis, C. G.; Kosmidou, P.-M. K.; and Chalioris, C. E.,
Polymers (Design Manual No. 5),” Structural Innovation and Monitoring “Reinforced Concrete Beams with Carbon-Fiber-Reinforced Polymer
Technologies Resource Centre (SIMTReC), Winnipeg, MB, Canada, 2008. Bars—Experimental Study,” Fibers, V. 6, No. 4, 2018, p. 99. doi: 10.3390/
8. Frankhauser, W.; Elkaissi, J.; Khal, S.; McMilan, S.; Potter, W.; fib6040099
Rister, D.; Wilson, D.; and O’Connor, J., “Advances in Fiber-Reinforced 30. Kim, Y. J., and Nickle, R. W., “Long-Term Multipliers and Deform-
Polymer (FRP) Composites in Transportation Infrastructure,” NCHRP ability of Fiber-Reinforced Polymer-Prestressed Concrete,” ACI Structural
20-68A, Transportation Research Board, Washington, DC, 2013. Journal, V. 115, No. 1, Jan. 2018, pp. 223-234. doi: 10.14359/51700988
9. Rizkalla, S., “Innovative Use of FRP for Sustainable Precast Concrete 31. Barker, R. M., and Puckett, J. A., Design of Highway Bridges: An
Bridges and Structures,” Proceedings of the Transport Research Arena, LRFD Approach, fourth edition, John Wiley & Sons, Inc., Hoboken, NJ,
Paris, France, 2014. 2021.
10. Ozyildirim, H. C., and Sharp, S. R., “Concrete Beams Prestressed 32. Nowak, A. S., and El-Hor, H. H., “Serviceability Criteria for
Using Carbon Fiber Reinforced Polymer,” Report No. VTRC 19-R29, Prestressed Concrete Bridge Girders,” 4th International Bridge Engineering
Virginia Transportation Research Council, Charlottesville, VA, 2019. Conference, Transportation Research Board, National Research Council,
11. Peng, F., and Xue, W., “Analytical Approach for Flexural Capacity Washington, DC, 1995, pp. 181-187.
of FRP Prestressed Concrete T-Beams with Non-Prestressed Steel Bars,” 33. AASHTO, The Manual for Bridge Evaluation, third edition, Amer-
Journal of Composites for Construction, ASCE, V. 22, No. 6, 2018, ican Association of State Highway and Transportation Officials, Wash-
p. 04018063. doi: 10.1061/(ASCE)CC.1943-5614.0000903 ington, DC, 2018.
12. ACI Committee 318, “Building Code Requirements for Structural 34. Atkins, P.; de Paula, J.; and Keeler, J., Physical Chemistry: Molec-
Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved ular Thermodynamics and Kinetics, Oxford University Press, Oxford, UK,
2022),” American Concrete Institute, Farmington Hills, MI, 2019, 624 pp. 2019.
35. McMurry, J. E., Organic Chemistry, eighth edition, Cengage
Learning, Boston, MA, 2012.
@seismicisolation
@seismicisolation

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.
@seismicisolation
@seismicisolation
Title No. 120-S88
Deflection Control Methodologies for Curvilinear

Concrete Members Reinforced with Glass Fiber-Reinforced
Polymer Bars
by Seyed Mohammad Hosseini, Salaheldin Mousa, Hamdy M. Mohamed, and Brahim Benmokrane
This paper reports the results of a comprehensive analytical Therefore, employing effective methodologies to predict the
study implemented to develop deflection prediction methodol- deflection of GFRP-RC members with high accuracy is of
ogies for curvilinear reinforced concrete (RC) members with great importance.
glass fiber-reinforced polymer (GFRP) reinforcement, focusing Two common approaches can be employed to calculate the
on precast concrete tunnel lining (PCTL) segments. The first step
immediate deflection of flexural RC elements: 1) using the
involved modifying the procedures for estimating elastic deflection,
general assumptions of elastic deflection calculation along
cracking moment, and cracked moment of inertia, which were then
introduced for use with curvilinear members. In the next step, three with the effective moment of inertia (Ie); and 2) integration
methodologies of effective moment of inertia, integration of curva- of curvature along the length of the member. Ie considers the
ture, and integration of curvature considering tension stiffening were effective transition between the gross moment of inertia (Ig)
developed for curvilinear members. Then, the analytical results in uncracked regions of a member to the cracked moment
were compared to the experimental database, and a novel method of inertia (Icr) in the cracked part considering the effect of
was developed for predicting deflection in curvilinear GFRP-RC tension stiffening. Branson8 originally assumed the rigidi-
members. In the third and final step, a procedure was developed to ties of the cracked and uncracked parts of a RC element as
adapt the presented methodologies for use with a tunnel segment springs in parallel. His assumption can be written in the form
under real load and boundary conditions. The results indicate that of Eq. (1) as a general model to predict Ie in RC members
the proposed method could predict the deflection of curvilinear
GFRP-RC members with high accuracy.
Ie = k1Ig + k2Icr ≤ Ig (1)
Keywords: curvilinear reinforced concrete (RC) members; deflection;
effective moment of inertia; glass fiber-reinforced polymer (GFRP); precast where k1 and k2 are functions of the ratio of cracking to
concrete tunnel lining (PCTL) segments; reinforced concrete (RC). applied moment (Mcr/Ma), which has been empirically
proposed. Equation (1), based on Branson’s recommenda-
INTRODUCTION tions, can predict the deflection of simply supported straight
Curvilinear reinforced concrete (RC) elements are widely rectangular concrete beams reinforced with typical amounts
used in many types of structures, such as tunnels, bridges, of steel reinforcement with reasonable accuracy. Such
water tanks, and culverts.1 Precast concrete tunnel lining models, however, underestimate the deflection of FRP-RC
(PCTL) segments number among the most frequently used elements, as it was correlated for beams with Ig/Icr smaller
curvilinear RC elements. When a tunnel is bored with a than approximately 4.0, while Ig/Icr in FRP-RC members
tunnel boring machine (TBM), such segments are placed generally ranges between 5 and 25.9 Besides, supposing
sequentially as the boring advances.2 Corrosion is one of the parallel springs for the rigidities of uncracked and cracked
major problems associated with RC structures reinforced sections in Eq. (1) is an incorrect assumption because they
with conventional steel reinforcement. Such issues are exac- are series springs.10,11 By neglecting such wrong assump-
erbated in the corrosive environment of tunnels.3,4 Replacing tions, various researchers tried to modify the values of k1 and
steel reinforcement with glass fiber-reinforced polymer k2 based on the experimental results of FRP-RC beams.12-21
(GFRP) reinforcement is recognized as a viable solution for Bischoff9 developed a new form of equation (Eq. (2)) for
dealing with corrosion issues.5 Serviceability often governs Ie based on the true assumption of series springs for the
the design of flexural members reinforced with GFRP bars cracked and uncracked rigidities in a flexural member
either through cracking, deflection, or stress verification.6
In general, curvature, loading distribution, span length,
and boundary conditions do not play a considerable role in
cracking control and stress verification procedures as they
are mainly related to sectional properties. These issues must
be considered in predicting deflection. Due to commercial ACI Structural Journal, V. 120, No. 5, September 2023.
MS No. S-2022-397.R1, doi: 10.14359/51738842, received May 1, 2023, and
GFRP bars having lower moduli of elasticity than steel rein- reviewed under Institute publication policies. Copyright © 2023, American Concrete
forcing bars, deflection in GFRP-RC flexural members at Institute. All rights reserved, including the making of copies unless permission is
service load is generally greater than in steel-RC members.7 closure, if any, will be published ten months from this journal’s date if the discussion
@seismicisolation
@seismicisolation

Icr

______________ account for serviceability control in straight GFRP-RC

Ie =    ≤ Ig (2)
1 − ηβ( Mcr
/Ma ) members. Moreover, there is a lack of methodologies
available in the literature for predicting the deflection of
where η = 1 – Icr/Ig; and β is the tension-stiffening factor, curvilinear GFRP-RC elements. This study fills this gap
varying between 0 and 1 for the case of no tension stiff- and presents novel deflection prediction methodologies
ening and full tension stiffening, respectively, suggested to specifically developed for curvilinear GFRP-RC elements,
be taken as Mcr/Ma. Employing Eq. (2) yielded reasonably with a particular focus on PCTL segments. The proposed
conservative estimations for deflection of simply supported methodologies were validated through testing 11 full-scale
FRP-RC members. GFRP-reinforced PCTL segments. The findings of this study
Integration of curvature along the member is another will significantly benefit design engineers and contribute to
approach to general deflection calculation proposed in the improving design standards. Notably, this study introduces
literature by various researchers and in standards.22-25 The novel equations and procedures that cannot be found else-
general concept of this method is to obtain curvature in where in the literature.
each section and integrate it along the length of the member.
Curvature in uncracked and cracked parts is calculated SUMMARY OF EXPERIMENTAL PROGRAM
using the gross and cracked moment of inertia, respectively, Eleven full-scale curvilinear GFRP-reinforced PCTL
neglecting the effect of tension stiffening. The effect of segments were constructed and tested under bending load.
tension stiffening can be considered by obtaining the curva- An overview of the specimens’ details, reinforcement char-
ture of the cracked part with the effective moment of inertia acteristics, test setup, and instrumentation is reported in
in each section22 or by linear interpolation of the section the following. Detailed information can be found in papers
curvature.24 Bischoff and Gross22 employed an approach previously published by the authors.27,28 The specimens had
integrating curvature, assuming the gross moment of inertia internal and external radii of 3250 and 3500 mm, respec-
in the uncracked regions and the effective moment of inertia tively, and measured 3100 mm in length, 1500 mm in width,
derived from Eq. (2) for the cracked regions. They proposed and 250 mm in thickness (Fig. 1). The test parameters include
the following equation as the equivalent moment of inertia reinforcement ratio (0.48, 0.69, 0.90, and 1.28%), concrete
by simplifying the integrals based on different load and type (normal-strength concrete [NSC] and fiber-reinforced
boundary conditions concrete [FRC]), concrete strength (NSC and high-strength
concrete [HSC]), and tie configuration (closed ties and
Icr

_______________ U-shaped ties). The reinforcement ratio of 0.48% was chosen
Ie =   2 ≤ Ig (3)
1 − γη(Mcr /Ma ) to ensure that the specimen exhibits a favorable compression-
controlled failure, as per the requirements outlined in the
where γ is a factor considering the effect of load and boundary provisions of ACI 440.1R-156 and CSA S806-12.23 Subse-
conditions. Equation (3) seems to be the most theoretically quently, higher reinforcement ratios (0.69, 0.90, and 1.28%)
correct method in the literature; it has also been adopted by were employed to accommodate larger bar sizes and closer
ACI 440.1R-15.6 In general, neglecting tension stiffening bar spacing while still considering construction feasibility.
led to deflection being overestimated, while considering it Sand-coated No. 5 and No. 6 GFRP bars were used as
equal to what was proposed by Bischoff and Gross22 under- longitudinal reinforcement and end-anchorage U-shaped
estimated deflection. The accuracy, however, depends on the bars. The transverse ties were sand-coated No. 4 GFRP
reinforcement’s axial stiffness, estimated cracking moment, bars, either U-shaped or closed ties. It should be noted
estimated concrete modulus of elasticity, moment level on that the tie configuration is a critical parameter that signifi-
which deflecting is calculated, load and boundary condi- cantly impacts the construction process of GFRP cages in
tions, and so on.9,10,22,26 GFRP-RC curvilinear members such as tunnel segments.
The deflection prediction methodologies in the literature Therefore, it is imperative to compare the serviceability
for FRP-RC members generally deal with straight members behavior of specimens reinforced with two common types
without axial load. In addition, there are no recommenda- of transverse ties—namely, closed ties and U-shaped ties.
tions for adapting the available methodologies for complex Tables 1 and 2 report the properties of reinforcement and
load and boundary conditions in real applications. This paper concrete, respectively. The specimens were tested under
presents an analytical study performed to propose deflection three-point bending load until failure (Fig. 1). The selected
prediction methodologies to estimate the deflection of curvi- test setup provides a determined system for establishing the
linear GFRP-RC members under service load conditions. relationship between internal forces, external loads, and
The focus of this study was to adapt the methodologies for deflection required for analytical procedures. This allows for
PCTL segments. The methodologies, however, are general reliable evaluation of deflection-calculation procedures and
and can be used for any type of curvilinear member. the development of general prediction models for any load
or boundary condition. Three linear potentiometers (LPOTs)
RESEARCH SIGNIFICANCE recorded midspan deflection; the quarterspan deflection was
Predicting the deflection of GFRP-RC curvilinear recorded with two LPOTs installed at the quarterspan. Table
members at the service load stage is crucial for their design. 2 presents the test matrix and the key experimental results for
However, the current deflection prediction methodologies, the tested specimens. To investigate the deflection behavior
as proposed by ACI 440.1R-156 and CSA S806-12,23 only at the service stage and compare the experimental results
@seismicisolation
@seismicisolation

Fig. 1—Specimen geometry, reinforcement details, test setup, and instrumentations. (Note: Dimensions in mm; 1 mm =
0.0394 in.)
Table 1—Mechanical properties of GFRP reinforcement
Nominal cross- Tensile modulus of Ultimate strength,
Reinforcement type Bar size Bar diameter, mm sectional area, mm2 elasticity, GPa MPa Ultimate strain, %
Curvilinear longitudinal No. 5 15.0 199 55.1 ± 1.25 1115 ± 60 2.0 ± 0.1
GFRP bars No. 6 20.0 284 52.9 ± 0.6 1068 ± 49 2.0 ± 0.1
No. 5 15.0 199 53.5 ± 1.1 1283 ± 42 2.4 ± 0.1
U-shaped GFRP bars*
No. 6 20.0 284 53.2 ± 2.9 1131 ± 35 2.1 ± 0.0
U-shaped and closed
No. 4 3.0 129 55.6 ± 1.6 1248 ± 74 2.2 ± 0.1
GFRP ties*
*
Reported values are based on applying tension to straight bars manufactured with same process as bent bars.
Note: 1 mm = 0.0394 in.; 1 mm2 = 0.00155 in.2; 1 GPa = 145 ksi; 1 MPa = 145 psi.
Table 2—Test matrix and test results

Deflection at service
moment, mm
Concrete Longitudinal fc′, fp, f150
D , Mcr, Mn,
Specimen ID type reinforcement Tie configuration MPa MPa MPa kN∙m kN∙m 0.3Mp 2000 µε 1.1Mcr
7G No. 5 NC 7 No. 5 bars Closed ties 48 — — 38 213 11.7 2.5 3.0
7G No. 5U NC 7 No. 5 bars U-shaped ties 44 — — 37 177 9.5 2.6 3.3
7G No. 5H HSC 7 No. 5 bars Closed ties 86 — — 49 247 10.4 2.7 3.4
13G No. 5H HSC 13 No. 5 bars Closed ties 90 — — 44 257 6.5 3.7 1.9
7G No. 5HU HSC 7 No. 5 bars U-shaped ties 87 — — 41 227 9.8 2.1 2.9
7G No. 5F FRC *
7 No. 5 bars Closed ties 50 4.5 0.8 33 210 9.8 2.5 2.5
13G No. 5F FRC 13 No. 5 bars Closed ties 44 4.0 1.3 30 273 8.2 3.8 1.9
7G No. 5FU FRC 7 No. 5 bars U-shaped ties 46 4.3 1.0 31 230 9.6 3.1 1.9
*
Polypropylene fibers, 12 mm in length, were used to fabricate FRC.
Note: 1 mm = 0.0394 in.; 1 kN·m = 0.7376 kip∙ft; 1 MPa = 145 psi.
@seismicisolation
@seismicisolation

Fig. 2—Geometrical parameters, deformation parameters, external loads, and internal forces in curvilinear member with load
and boundary conditions similar to test specimens (deformations and curvatures are exaggerated).
with the analytical procedure, some reference points must the service load deflection of different types of curvilinear
be specified. As the design of GFRP-RC flexural elements is GFRP-RC members.
generally governed based on serviceability requirements, the
bending moment under service load conditions is typically Calculating elastic deflection in curvilinear
much lower than their nominal bending moment capacity. In members
the following, two reference points for service load condi- When the ratio of radius to the sectional height in a curved
tions of GFRP-reinforced flexural elements were defined member is greater than 2, the fundamental concepts related
according to the literature26,29: 1) moment corresponding to to the relationship between curvature and deflection, as well
30% of the peak moment; and 2) moment corresponding to as the strain energy due to the bending, can be approximated
a strain of 2000 µε in the tensile reinforcement. The latter by that of straight members.30 As follows, the two methods
sometimes leads to defining a service moment that is lower commonly used to calculate elastic deflection in straight
than the cracking moment. This might lead to unrealistic members were modified for use with curvilinear members.
predictions for deflection at the service load. Therefore, a Figure 2 shows the centerline of a curvilinear member
moment corresponding to 1.1 times the cracking moment before and after deformation induced by an external force.
was introduced as an alternative to the moment corre- The radial deflection at each point is rθ. Consider a small
sponding to a strain of 2000 µε when the obtained service element of CD with arc length ds. The exaggerated shape
moment is smaller than the cracking moment. Mota et al.7 shows that the centerline of the segment is specified as
applied the same approach. CD and C′D′ after deformation. Radial deformation at the
point C is rθ, while such deformation is rθ + dr at point D.
ANALYTICAL INVESTIGATIONS Therefore, the relationship between the radial deflection and
This section presents the deflection prediction methodol- curvature in the element is
ogies developed for use with GFRP-reinforced curvilinear
members. Initially, two methods were proposed to calcu- _
d2r _Mθ
2 = − ϕθ = E I (4)
late elastic deflection in curvilinear structural elements. ds θ θ
The cracking moment to be used in calculating deflection

is discussed and proposed. After that, the procedures and Obtaining the previous equation assumed small deforma-
models developed to obtain the cracked moment of inertia tions. Moreover, the influence of curvilinearity on the funda-
in GFRP-reinforced NSC, HSC, and FRC PCTL segments mental assumptions of the distribution of internal stresses
are described. Subsequently, three procedures for calcu- and the curvature of the cross section was not considered.
lating deflection are presented and adapted for use in curvi- This is attributed to the high ratio of curvature to thick-
linear RC members. Thereafter, the results obtained from the ness within the considered curvilinear members. In such
presented procedures are compared with the experimental scenarios, the impact of curvilinearity on stress distribution
results. Subsequently, a model capable of predicting deflec- and the moment-curvature correlation in the cross section
tion in curvilinear GFRP-RC members with high accuracy is considered insignificant, as supported by Boresi et al.30
is proposed. Lastly, a procedure was developed to employ Considering ds = Rdθ, rotation can be obtained with the
the deflection prediction methodologies presented for tunnel following equation
segments under actual loading and boundary conditions.
While the methodologies presented focus mainly on tunnel Θθ = ∫ −ϕθRdθ + C1 (5)
segments, they can also be used effectively to estimate @seismicisolation
@seismicisolation

where C1 can be obtained according to the boundary condi- experimental results. However, considering the various
tions. Using a similar approach, radial deflection in each factors that can influence the cracking moment value in
section can be obtained as follows different conditions, it is reasonable to use more conser-
vative values for the cracking moment in practical design
rθ = ∫(∫ − ϕθds + C1)Rdθ + C2 (6) applications.
To obtain the vertical deflection, dr, each point should be Cracked moment of inertia
multiplied by sin(θ) in the integral. Therefore, the following Calculating the cracked moment of inertia is essential
equation is obtained to calculate the vertical deflection in a in all the deflection-calculation methods presented. In the
polar coordinate system in a curvilinear member following, the procedures to calculate the cracked moment
of inertia for GFRP-reinforced NSC, HSC, and FRC tunnel
Δθ = ∫(∫ − ϕθRdθ + C1)Rsin(θ)dθ + C2 (7) segments were presented. The contribution of reinforcing
top bars is neglected in the presented procedures as a simpli-
where C2 can be obtained using boundary conditions. When fying and conservative assumption.
it is aimed to calculate the deflection at a certain location, GFRP-reinforced NSC and HSC curvilinear members—
deflection can be obtained using a virtual work method with When there is no axial load, the cracked moment of inertia
this equation does not depend on the applied bending moment on a section
in which a linear stress-strain relationship is considered for
mθMθ concrete in compression. When axial load is present, both
Δ = ∑∫_
E I Rdθ (8)
θ θ the axial load and bending moment in the section affect the
cracked moment of inertia. The cracked moment of inertia
where mθ is the moment induced because of a unit dummy in such conditions can be calculated with the following
load applied at the point where deflection is being calcu- equation when the contribution of the top reinforcement is
lated. It should be noted that the deflections resulting from neglected.
shear and axial forces are neglected as they are generally
insignificant when the span length-to-depth ratio is large.30 Icr = (bd3/3)ka3 + nfAf d2(1 – ka)2 (9)
According to the presented methods, an equation for calcu-
lating the elastic deflection of the test specimens was where Eq. (10) can be used to calculate ka
developed, as reported in Appendix A.*
__________________________
√   
N2− 2Nωnρf+ ω2nρf(nρf+ 2) − nρfω + N
_____________________________________
Cracking moment ka =        
ω (10)
Cracking moment (Mcr) is one of the most influential
parameters in estimating the deflection in an RC member. where ω = Ecεcbd. As can be inferred from Eq. (10), the
Therefore, predicting the exact cracking moment is of great uncracked depth depends on the level of axial load and the
importance in accurately estimating deflection. Equalizing maximum concrete compressive strain. Equation (11) pres-
the maximum tensile stress in the uncracked section______ to the ents the relationship between the bending moment, axial
maximum tensile capacity of concrete (0.62√fc′   according load, ka, and εc in a section.
to ACI 318-1931) yields the theoretical cracking moment
(Mcr,theo). Shrink restraint in an RC member might lead to
pre-existing tensile stresses in the member, which reduce the

_
(
3 − ka _h _
) (
ka d
Ma = nρfωd(1 − ka ) 9 + N 2 − 3 ) (11)
cracking moment.32 Bischoff and Gross22 reported a range of
0.48 to 1.44 with a median of 0.85 for the ratio of theoret- Inserting ka from Eq. (10) into (11) yields an equa-
ical to experimental cracking moment based on the exten- tion with εc as its unknown variable. Due to complexity,
sive data from the literature for FRP-RC flexural members. however, there is no closed-form solution for that equation.
ACI 318-1931 recommends multiplying the theoretical In such situations, the value of εc can be found by trial and
cracking moment by 0.67 in the deflection-calculation proce- adjustment. Subsequently, the values of ka and Icr can be
dure. The ratio of theoretical to experimental cracking moment calculated. The presented equations are based on the linear
in the tested specimens was 0.7 ± 0.02, 0.58 ± 0.04, and 0.61 concrete stress-strain assumption in compression, which is
± 0.02 for NSC, HSC, and FRC specimens, respectively. valid until approximately 0.7fc′.33 According to a prelim-
There is a need for an extensive study to determine the value inary comparison conducted by the authors for the tested
of cracking moment in different concrete types for FRP-RC specimens, neglecting the effect of axial load led to an error
members. Given the lack of such study, the author’s study of approximately 10% in the value of ka for a given value of
recommends taking Mcr equal to 0.7Mcr,theo for NSC GFRP- Ma when the axial load was below ±0.0045fc′Ag. When the
reinforced tunnel segments and 0.6Mcr,theo for HSC and axial load was increased, neglecting such contributions led
FRC GFRP-reinforced tunnel segments according to the to considerable errors.
Curvilinear GFRP-reinforced FRC members—Finding the
cracked moment of inertia in an FRC section considering
*
appended to the online version of the published paper. It is also available in hard copy the contribution of fibers requires assuming a stress-strain
from ACI headquarters for a fee equal to the cost of reproduction plus handling at the model for FRC in compression and tension. The authentic
time of the request. @seismicisolation
@seismicisolation

accuracy is expected for simply supported beams with point
load or distributed load.22 In addition, the axial-load level
and its variation along the member might greatly affect accu-
racy. In the case of FRC, the contribution of the fibers can be
considered using the effect of fibers on the cracked moment
of inertia as well as their effect on the tension-stiffening
behavior of the concrete. Bischoff36 proposed modifications
to Eq. (2) to consider the contribution of fibers for both
cracking moment of inertia and tension stiffening. This study
relies only on the effect of fibers on the cracked moment
of inertia to consider the contribution of fibers in the calcu-
Fig. 3—Assumed model for distribution of stress and strain lation of GFRP-reinforced FRC PCTL segments because
in cross section of FRC specimens. considering the effect of FRC on the tension stiffening leads
stress-strain properties of FRC under compression and tension to impractical and complex procedures.
can be accurately determined through the corresponding tests
and used in the analysis. Nevertheless, in the design of FRC Deflection prediction using integration
components, simple models are commonly used.34 Figure 3 of curvature
presents the assumed stress and strain distribution in the For predicting deflection using integration of curvature,
section based on the stress-strain model for FRC. This model the cracked and uncracked parts of the member should be
was adopted and simplified for the service stage based on specified based on the bending-moment diagram. Thereafter,
the provisions in ACI 544.4R-18.34 The parameters used in the deflection-calculation equations can be used by substi-
the stress-strain response of FRC can be obtained with beam tuting Iθ with gross and cracked moment of inertia in the
testing as well as stress and strain compatibility. σcr and uncracked and cracked sections, respectively. This method is
referred to as the Integ. method herein. Equation (15) can be
σp can be taken equal to fp and 0.37f 1D50 , respectively.2,34,35
used to calculate the deflection through integration of curva-
Equations (12) and (13) form a system of equations with
ture in the virtual work method.
two unknown variables—εc and ka—which can be obtained
through trial and adjustment. mθMθ mθMθ
Δ = ∑ ∫_

_
Rdθ + ∑ ∫E Rdθ (15)
Uncracked EcIg
Cracked cIcr,θ
Ff + Fcr + Fp + N – Fc = 0 (12)
Equation (15) could be simplified with this approach
(
Ma = Ffd + Fcr
_
3 )
ka d + 2 dcr

+ Fp ( 2 )+ N _
dcr
_ + h
()
h
( )ka d
− Fc _
2 3
Δ = Δg + δΔcr (16)
(13)
where Δg and δΔcr can be calculated with these equations
(refer to Appendix B)
After obtaining εc and ka, the cracked moment of inertia
considering the contribution of fibers can be calculated with _ 1
the following equation
Δg = EI ∫mθMθ Rdθ (17)
c g
Ma ka d
_

Icr
= Eε (14) n η m M
δ Δcr = ∑ _ i ∫_
c c
θ θ Rdθ (18)
i =1 Ec Icr,i

Deflection prediction using effective moment of
inertia These equations are valid for all types of loading and
The effective moment of inertia proposed by Bischoff9 boundary conditions, provided that no settlement or move-
(Eq. (2)) can be replaced with the value of Iθ in the methods ment has occurred in the supports and that the relationship
presented to calculate deflection when the relationship between the applied load and internal forces can be deter-
between the applied loads and internal forces can be spec- mined. Because variations in the axial-load level changes
ified. This method is referred to herein as Ie. This procedure the value of Icr, using a constant value for Icr might be a
does not consider the effect of load and boundary conditions source of errors in the calculation procedure. To account for
or variations in the axial load in the member. It does consider the effect of axial load on deflection calculation, the cracked
the effect of tension stiffening by using the tension-stiff- sections along the member can be divided into a reasonable
ening factor β. This factor theoretically varies between 0 number of parts. The values of ηi and Icr,i for each part can
and 1 depending on the level of bending moment. Bischoff9 be calculated by obtaining δΔcr and summing δΔcr,i along
recommended using Mcr/Ma as the tension-stiffening factor. the cracked section. When the level of axial load and its
As this method supposes a constant effective moment of variation along the member are not significant, however, the
inertia along the member, it is simple to use, but its accu- minimum value of Icr along the member expected in a section
racy depends on the types of loading, boundary conditions, with the greatest bending moment and the lowest axial load
reinforcement ratio, and level of bending moment. The@seismicisolation
best
@seismicisolation

can be used for all the cracked sections as a conservative used, employing the same equation as Eq. (17) for Δg. In this
simplification. case, δΔcr should be calculated as
Simplifying the presented equations for load and
boundary conditions similar to the tested specimens leads to η
δ Δcr = ∑ _
n
i ∫ mθMθ (1 − ηiβi(Mcr,i/Mθ )) Rdθ (23)
the following equation to calculate the deflection at midspan i=1 cr,iEc

I
(refer to Appendix B)
Note that Eq. (23) requires greater computational effort
than Eq. (18), especially when the effect of axial load is to
Δ = λ/EcIcr (19)
be considered. For load and boundary conditions similar to
those of the test specimens when the axial load and its vari-
where λ is calculated using Eq. (20)
ation are not significant, it yields
[0 ]
⎢ ⎥
_π _
π
⎡ ⎤
2 −α 2 −α
η
λ = 2(1 − η) ∫ Cθ Rdθ+ (1 − η )∫ Cθ Rdθ (20)
_ _π
2 −α

θg (1 − η) ∫ mθMθ Rdθ+η
_ 2 0

Δ = EI   
π
_
−α
(24)
c cr 2
where Cθ = m(π/2)–αMθ. Equation (20) is valid for those types ∫ mθMθ (1 − β(Mcr
/Mθ ))Rdθ
of load and boundary conditions where the distribution of ⎣θg ⎦
bending moment is symmetrical, and the uncracked section
starts at the supports of an angle θg, followed by a cracked Equation (19) can be used to calculate the midspan deflec-
section from θg to midspan. In addition, Icr of the critical tion once the value of λ has been determined with this equa-
section (often located at midspan) was used as the cracked tion (refer to Appendix B)
moment of inertia for all the cracked sections in Eq. (15),
[0 ]
because the level of axial load was not significant. The angle _π
2 −α
_π
2 −α
θg corresponds to the angle from the support to the point at
λ = 2(1 − η) ∫ Cθ Rdθ+ η ∫ ( Cθ − Cθ g) Rdθ (25)
which Mθ = Mcr. For the load and boundary conditions of the θg
tested specimens, θg can be obtained with this equation

where Cθg = (Mcr)2/P.

θ____________________
g = Evaluation of presented methods with
( )
experimental data
cot (α ) (2 Cb − Ca ) − √(  
2
2PR) + 4 Ca Cb − ( Ca )
______________________________________
2
sin−1          − α Table 3 presents the ratio of the theoretical to the exper-
2PR( cot2(α ) + 1) imental midspan deflection of the tested specimens. In
(21)
addition, Fig. 4 and 5 compare the experimental and
where Ca = 2Pfcot(α) + Pl – 4Mcr; and Cb = PRcot(α). It analytical moment-deflection curves of the specimens. The
should be noted that Mcr is dependent on the level of axial moment-deflection curves were drawn up to 50% of the
load. When the variation in axial load is not significant, experimental bending-moment capacity of the specimens.
however, Mcr obtained from the minimum axial load along Table 3 provides a comparison of the midspan deflection
the member can be used as a conservative assumption for at the loads corresponding to the three reference points of
simplicity. For the tested specimens, the integrals in Eq. (20) 2000 µε, 1.1Mcr, and 0.3Mn, when applicable. The deflection
were obtained and are reported in Appendix B. in FRC specimens was obtained according to two scenarios
of considering or neglecting the contribution of the fibers.
Deflection prediction using integration of The average and standard deviation are presented __
sepa-
curvature considering tension stiffening rately for each concrete type. In this study, 3 320√ fc′   + 6900
Due to the effect of tension stiffening, the stiffness in was used as the concrete modulus of elasticity according to
the cracked parts of an element is greater than the cracked ACI 363R-10.37
moment of inertia.22 By supposing Iθ = Icr,θ/[1 – ηθβθ According to Table 3, applying the Ie and Integ. methods
(Mcr,θ/Mθ)] in the calculations related to the cracked parts of overestimated the deflection of NSC tunnel segment speci-
the section, the effect of tension stiffening can be considered. mens by 20% and 50%, respectively, on average, for different
In such situations, deflection can be obtained with reference points. In contrast, considering tension stiffening
based on the Integ. TS method underestimated deflection by
mθMθ 33%, on average. In the HSC specimens, the overestima-
Δ = ∑ ∫_ EI Rdθ
Uncrackd c g tion yielded by the Ie and Integ. methods was 3% and 33%,
mθMθ (1 − ηθβθ( Mcr,θ
/ Mθ ) ) respectively. In contrast, considering the tension stiffening
+ ∑ ∫_______________________
       Rdθ (22) with the Integ. TS method underestimated the deflection by
Cracked EcIcr,θ

54%. Neglecting the contribution of the fibers in FRC tunnel
This method is referred as Integ. TS herein when β is segment specimens resulted in a significant overestimation
considered as Mcr/Mθ according to the recommendation of of the deflection (42% for Ie and 81% for Integ. methods).
Bischoff and Gross. To simplify Eq. (22), Eq. (16) can be
22 Considering the contribution of the fibers by considering
their effect on the cracked moment of inertia reduced the
@seismicisolation
@seismicisolation

Table 3—Comparison of experimental and analytical results
Deflection-calculation method
Ie Integ. Integ. TS Mod. model
Δtheo./Δexp.
ID 2000 µε 1.1Mcr 0.3Mn 2000 µε 1.1Mcr 0.3Mn 2000 µε 1.1Mcr 0.3Mn 2000 µε 1.1Mcr 0.3Mn
NSC specimens
7G No. 5 — 0.93 0.96 — 1.05 1.13 — 0.29 0.51 — 1.14 1.00
7G No. 6 1.02 0.91 1.13 1.38 1.23 1.37 0.35 0.32 0.66 0.98 0.91 1.07
13G No. 5 1.44 1.68 1.35 1.74 2.24 1.61 0.85 0.64 0.85 1.14 1.15 1.09
13G No. 6 1.17 1.53 1.13 1.39 2.00 1.33 0.81 0.63 0.82 0.91 0.80 0.91
7G No. 5U — 0.84 0.80 — 1.15 1.04 — 0.27 0.35 — 1.10 0.90
Average 1.21 1.18 1.07 1.50 1.53 1.30 0.67 0.43 0.64 1.01 1.02 0.99
STD 0.21 0.39 0.21 0.21 0.55 0.22 0.28 0.19 0.21 0.12 0.16 0.09
HSC specimens
7G No. 5H — 0.63 1.10 — 0.86 1.40 — 0.20 0.53 — 1.01 0.98
13G No. 5H 1.15 1.00 1.23 1.51 1.34 1.50 0.48 0.36 0.73 0.94 1.05 1.01
7G No. 5HU — 0.92 1.17 — 1.28 1.45 — 0.26 0.63 — 1.15 1.04
Average 1.15 0.85 1.17 1.51 1.16 1.45 0.48 0.27 0.63 0.94 1.07 1.01
STD — 0.19 0.07 — 0.26 0.05 — 0.08 0.10 — 0.07 0.03
FRC specimens neglecting contribution of fibers
7G No. 5F 1.09 1.09 1.25 1.50 1.50 1.51 0.33 0.33 0.76 1.34* 1.34 1.32
13G No. 5F 1.83 2.53 1.35 2.23 3.27 1.52 1.07 1.14 1.01 1.68 2.29 1.28
7G No. 5FU 1.32 1.04 1.32 1.78 1.43 1.55 0.47 0.34 0.89 1.57 1.41 1.40
Average 1.41 1.55 1.31 1.84 2.07 1.53 0.62 0.60 0.89 1.63 1.68 1.33
STD 0.38 0.85 0.05 0.37 1.04 0.02 0.39 0.46 0.13 0.08 0.53 0.06
FRC specimens considering contribution of fibers
7G No. 5F 0.97 0.97 1.18 1.33 1.33 1.43 0.31 0.31 0.72 0.98 0.98 1.11
13G No. 5F 1.39 1.83 1.13 1.69 2.33 1.28 0.83 0.90 0.85 1.06 1.22 0.98
7G No. 5FU 0.90 0.66 1.06 1.27 0.97 1.27 0.37 0.30 0.72 1.00 0.92 1.06
Average 1.09 1.15 1.12 1.43 1.54 1.33 0.50 0.50 0.76 1.01 1.04 1.05
STD 0.27 0.61 0.06 0.23 0.70 0.09 0.28 0.34 0.08 0.04 0.16 0.07
*
Values were obtained with same modification parameters as NSC.
Note: Integ. refers to integration-of-curvature method; Integ. TS refers to integration-of curvature method considering tension stiffening; STD refers to standard deviation.
overestimation of the Ie and Integ. methods by 12% and to Table 3. Therefore, although using effective moment
43%, respectively. The Integ. TS method underestimated the of inertia is simpler than the other methods, it yielded the
deflection by 30% and 41%, respectively, on average, when most accurate results with acceptable conservativeness.
neglecting or considering the contribution of the fibers. As mentioned previously, however, the method’s accuracy
Table 3 and Fig. 4 and 5 reveal that the accuracy of these depends on the load and boundary conditions.22
deflection prediction methods depends on the reinforcement
ratio and concrete type. For instance, Ie underestimated the Proposed model
deflection of 7G No. 5 by 5%, while it overestimated the The method based on integration of curvature considering
deflection of 7G No. 6 and 13G No. 5 by 2% and 49%, the tension-stiffening factor equal to Mcr/Mθ seems to be
respectively. Generally, Integ. TS significantly underesti- theoretically correct. As reported in the preceding section, it
mated the midspan deflection, which was more pronounced considerably underestimated the midspan deflection. Under-
at the lower reinforcement ratio. The approach to calculating estimation with such methods is consistent with some studies
deflection yielded relatively more reasonable results for FRC in the literature.22,38 The underestimation was greater at the
specimens (Fig. 5). In general, according to the average for reference points 2000 µε and 1.1Mcr, especially in the NSC
all the specimens and reference points, the ratio of theoret- specimens with lower reinforcement ratios. To help demon-
ical to experimental deflection was 1.11, 1.41, and 0.55 for strate the reason for such underestimation, Fig. 6 presents
the Ie, Integ., and Integ. TS methods, respectively, according the theoretical midspan moment-curvature diagrams of 7G
@seismicisolation
@seismicisolation

Fig. 4—Comparison of experimental and analytical moment-deflection diagrams of NSC specimens. (Note: 1 mm = 0.0394 in.;
1 kN·m = 0.7376 kip·ft.)
No. 5 and 13G No. 6 specimens obtained based on the Ie, tension-stiffening factor was overestimated. Increasing the
Integ., and Integ. TS methods. In addition, the experimental reinforcement ratio in 13G No. 6 eliminated the initial curva-
moment-curvature diagram was drawn for comparison using ture increase, and the tension-stiffening factor was predicted
the strain values recorded during the test. The Ie and Integ. with good accuracy. Comparing the moment-curvature
TS methods yielded similar curvatures in a section because diagrams of all tested specimens revealed that the rapid
they use the same equation for the moment of inertia in a increase in the curvature at cracking and the accuracy of Mcr/
section. Figure 6 shows that the curvature of specimen 7G Mθ as the tension-stiffening factor depended primarily on the
No. 5 increased rapidly after initiation of the first crack reinforcement ratio, concrete strength, and concrete type.
and thereafter approached the curvature obtained from the This study developed the following equations to modify the
cracked moment of inertia by increasing the applied bending values of η and β for use in the presented method based on
moment. In contrast, considering tension stiffening by setting Integ. TS
the tension-stiffening factor equal to Mcr/Mθ did not follow
the experimental trend. In that case, the initial increase p p
of curvature upon cracking was not considered and the βm = m ( 
fc′ ρf _
______
fc,n′ρfb ( )
 )  M where m
Mcr

θ
( 
fc,n′ρfb)
fc′ ρf
______
    ≤ 1.0 (26)
@seismicisolation
@seismicisolation

Fig. 5—Comparison of experimental and analytical moment-deflection diagrams of HSC and FRC specimens. (Note: 1 mm =
0.0394 in.; 1 kN·m =q0.7376 kip·ft.) q
ηn = 1 − n( 
_____
()
fc′ρf ) Ig
    where n ( 
fc,n′ ρfb _ Icr

fc′ρf )
fc,n′ ρfb
_____
    ≥ 1.0 (27) Table 4—Proposed coefficients to be used in
proposed model
where fc,n′ is 80 MPa (11.6 ksi) for HSC and 40 MPa (5.8 ksi) Concrete type m n p q
for NSC and FRC. The modification constants m, n, p, and
NSC 0.14 4.0 1.12 1.0
q were obtained using regression analysis according to
the results for different types of concrete and are reported HSC 0.65 5.0 0.26 0.8
in Table 4. The effect of the reinforcement ratio on the FRC 0.30 7.7 0.70 2.3
tension-stiffening characteristic was considered using the
ratio of ρf to ρfb. Yost et al.20 and Mousavi and Esfahani17 with closed ties and U-shaped ties, identical constants were
used a similar approach to consider the effect of the rein- proposed for both tie configurations. The modified deflection
forcement ratio on the deflection of GFRP-RC beams. In values can be obtained by replacing the values of η and β in
addition, the ratio of fc,n′/fc′ was added to the proposed model the equations presented to calculate the deflection consid-
to consider the effect of variations in concrete strength in ering the tension-stiffening factor. Equation (28) presents the
the specimens. Furthermore, as the difference was minimal modified value of curvature in each section in the proposed
between the stiffness behavior of the specimens reinforced modified model.
@seismicisolation
@seismicisolation

Fig. 6—Comparison of experimental and analytical moment-curvature diagrams for specimens 7G No. 5 and 13G No. 6. (Note:
1 mm = 0.0394 in.; 1 kN·m = 0.7376 kip·ft.)
Fig. 7—Comparison of curvature response of different deflection-calculation methods: (a) moment-curvature response; and
(b) curvature along specimen. (Note: 1 mm = 0.0394 in.; 1 kN·m = 0.7376 kip·ft.)

_
(
Mθ
) Mcr
_
ϕθ = EI 1 − ηnβmM
c cr θ
(28) which reduces the value of Mcr/Mθ according to the rein-
forcement ratio, concrete strength, and the modification
factors. Figure 7(b) presents the curvature value along the
Appendix C describes the procedure used to obtain the
tunnel segment specimen obtained using the Ie, Integ., Integ.
modified equations. Figure 7(a) shows the main concepts
TS, and Mod. methods (note that the specimens’ centerline
of the assumed moment-curvature response in the modi-
was selected as the x-axis for a better view). As can be seen,
fied (referred to as Mod. herein) and Integ. TS methods (the
the methods based on the integration of curvature yielded
graphs are for specimen 7G No. 5). Tension stiffening at a
minimal curvature in the uncracked sections. In the cracked
certain bending moment refers to a change in curvature (ΔΦ)
sections, using Icr in the Integ. method eventually increased
relative to the curvature of the cracked member obtained
the curvature right after passing θg; the curvature increased
using Icr. ΔΦmax is the maximum possible tension stiffening
linearly up to the midspan. In contrast, using the effective
at cracking. This is the tension-stiffening value considered
moment of inertia according to Eq. (2) led to a gradual
in Integ. TS right after the formation of the first crack. This
increase in curvature. In the Mod. method, an increase in
can be the main reason for the significant underestimation
curvature after passing from the uncracked region is consid-
of the integration-of-curvature method considering tension
ered, and the curvature is modeled to gradually increase up
stiffening in calculating deflection for the bending moments
to midspan.
near the cracking load. An indirect method was used to
Figure 6 presents a comparison of the moment-
consider the curvature increase when the first crack appeared
curvature obtained using the proposed equations and the
by modifying the value of η by ηn (refer to Appendix C). By
experimental results for 7G No. 5 and 13G No. 6. As can
using ηn, the maximum change in curvature will be limited to
be seen, the modified model fitted well with the exper-
ΔΦm. Increasing the bending moment decreases the tension-
imental moment-curvature of the sections. The moment-
stiffening effect. The ratio of change in curvature at a certain
deflection relationships obtained with the modified model
bending moment (ΔΦ) to the maximum change in curva-
for different specimens appear in Fig. 4 and 5. As shown, the
ture at cracking is known as the tension-stiffening factor.
modified model was quite consistent with the experimental
Because using Mcr/Mθ overestimated the tension-stiffening
results for all the specimens. In addition, Table 3 gives the
effect, the modified tension-stiffening factor βm is proposed,
@seismicisolation
@seismicisolation

ratio of theoretical to experimental midspan deflection for segments or other curvilinear members under real load and
the modified model. Considering an average for different boundary conditions. The equations presented require the
reference points, the developed modified model predicted relationship between the applied loads and the internal forces
the midspan deflection with conservativeness of 1%, 2%, in the member. This is because the complexity of the load
and 3% in the NSC, HSC, and FRC tunnel segment spec- and boundary conditions in tunnel segments and some curvi-
imens, respectively. Therefore, the modified model accu- linear members make it generally impossible to determine
rately predicted the midspan deflection for the tested spec- such relationships. In addition, the axial load on the member
imen with acceptable conservativeness of 2% on average. relates the cracked moment of inertia in each section to the
combination of axial load and bending moment. Therefore,
DISCUSSION some main parameters in the equations vary from section
As mentioned, the coefficients presented for the modified to section. The following procedure adapts the deflection-
model were obtained based on the experimental data. As calculation procedures presented for designing GFRP-
the number of specimens was limited, the accuracy of the reinforced PCTLs under real loading and boundary
coefficients needs to be further verified with an extensive conditions.
database. In particular, the coefficients for the FRC speci- Step I: Finding forces and initial deflection at joints using
mens are valid when the properties of the FRC are similar first-order analysis—Generally, there is interaction between
or superior to that in the current study. Should the mechan- the applied loads and deflection in tunnel segments and other
ical properties of the FRC be lower than the FRC in this RC structures. In such situations, the internal forces and esti-
study, the coefficients for NSC herein should be used. In mated approximate deformation must be found with a first-
addition, the average compressive strengths of NSC and order analysis, which requires an estimate of the moment
HSC in this study were 40 and 88 MPa (5.8 and 12.8 ksi), of inertia to be used in the analysis procedure. This must be
respectively. Therefore, caution should be exercised when accomplished despite the moment of inertia varying section
using the proposed coefficients in Table 4 for NSC or HSC by section in RC structures. The common design practice
with compressive strengths that differ significantly from in such cases is to use an initial estimation of the moment
those used in this study. The validity of the coefficients in of inertia in the members. Zadeh and Nanni40 proposed the
the presence of axial load also needs to be validated. In fact, following equations for the first-order estimation of the
the diagrams in Fig. 6 are valid when then the axial load moment of inertia of GFRP-RC slab members and columns.
is not significant. It is expected, however, that compressive
axial load would improve the tension-stiffening behavior.39 Islab = [0.10 + 0.15(Ef/Es)]Ig ≤ 0.25Ig (29)
Furthermore, variations in the surface configuration of the
reinforcement can affect the tension-stiffening characteris-
tics and, consequently, influence the accuracy of deflection Icolumn = [0.40 + 0.15(Ef/Es)]Ig ≤ 0.55Ig (30)
prediction models9,11,26 Furthermore, further verification is
required to confirm the accuracy of the proposed model for When a designer expects the axial load of a member to
different span-depth ratios. be greater than 0.1fc′Ag, it can be assumed to be a column.
In general, the Integ. TS method should not be used to Otherwise, the initial moment of inertia proposed for slabs
calculate deflection in curvilinear GFRP-RC members with can be used in the analysis. The reduced flexural rigidity (Ir)
low reinforcement ratios as it significantly underestimates for use in the related calculations can be obtained with the
deflection. A designer may, however, opt for the Ie, Integ., or following equation (where relevant)3
Mod. method based on design considerations. In fact, each
of these methods has advantages and disadvantages. Using Ir = Ij + (4/ns)2 × I (31)
Ie simplifies the deflection-calculation procedure but does
not guarantee that the effect of loading and boundary condi- where Ij is the moment of inertia at the joint, which is taken
tions is considered. The Mod. method could provide more as zero in the design; ns is the number of segments in a ring,
accurate results than the other methods, but it requires more excluding the key segment, which should be considered
computational effort. Lastly, the Integ. method could be a greater than four; and I denotes the lining moment of inertia,
suitable conservative option when the designer is unsure which can be calculated with Eq. (29) or (30), according to
about the member’s tension-stiffening characteristics. In the axial-load level. The internal forces and joint deforma-
addition, this method requires less computational effort than tions can be estimated with the first-order analysis using one
the Mod. method. The following section provides the proce- of the analysis methods such as elastic equation method,
dure developed for adopting the Ie, Integ., or Mod. method beam-spring method, finite element modeling (FEM), and
for GFRP-reinforced PCTL segments under real loading and discrete element method (DEM).3
boundary conditions. The procedure is general and can be Step II: Calculating rotation and deflection in selected
used for other types of curvilinear members. segments—The critical segments for the deflection control
can be determined based on the results from Step I. Figure 8
Development of methods for use under real provides a schematic view of a segment considered for the
loading and boundary conditions deflection control procedure. The effect of the other segments
The integration-based methods presented previously and the joints on the boundary conditions of the segment
include several assumptions that might not be met in tunnel is modeled by vertical, horizontal, and rotational springs.
@seismicisolation
@seismicisolation

Fig. 8—Schematic view of tunnel segment considered for deflection calculation.
Fig. 9—Schematic view of proposed deflection-calculation procedure for GFRP-reinforced PCTL segments under real load and
boundary conditions.
A schematic arbitrary external load is shown, which varies 4. Calculate the cracking moment (Mcr,i) for each point
according to the loading conditions. Note that the displayed according to the axial-load level.
springs and external load do not play a role in the calcula- 5. For the points where M ≥ Mcr,i, calculate the cracked
tion procedure, which will be described in the following and moment of inertia (Icr,i).
presented solely to show a semi-real condition of a segment. 6. For deflection calculation based on the effective moment
The analysis in Step 1 is supposed to yield the forces and of inertia, consider the moment of inertia for each point (Ii)
moments as well as the deformations and rotation of the according to Eq. (32). For the other methods, where M ≥
joints. The following procedure is proposed to obtain the Mcr,i, calculate the moment of inertia based on the deflection-
rotation and deflection diagrams of GFRP-reinforced PCTL calculation method to be used. Equations (33) and (34)
segments. In addition, Fig. 9 shows a hypothetical schematic represent the moment of inertia based on the Integ. and Mod.
view of the proposed procedure. methods, respectively. When M ≤ Mcr,i, consider Ii equal to
1. Divide the tunnel segment into certain number of Ig.
elements (even number) referred to as nseg. In such situa-
tions, Δθ = 2θmax/nseg. Icr,i

______________

Ii =    (32)
2. Designate each element as ni, i = 1, 2, 3,…, nseg. The
angle between starting point of the element ni and the starting
point of the segment is iΔθ.
Mcr,i
_
1 − ηiβi M ( )
θ,i
3. Specify the axial load (Ni) and bending moment (Mi) at Ii = Icr (33)
the points corresponding to α + iΔθ, where i = 1, 2, 3,…, nseg
based on the results in Step I (refer to Fig. 5 for the defini- Icr,i

________________
Ii =    (34)
( )
tions of α and θmax). Mcr,i

_
1 − ηn,iβm,i M
@seismicisolation θ,i
@seismicisolation

7. Calculate the curvature value at each point using the 4. Employing the effective moment of inertia is a simple
following equation method with acceptable conservativeness (11% on average
in the tested specimens). Its accuracy, however, depends on
ϕi = Mi/EcIi (35) the load and boundary conditions. The proposed method
can provide more accurate results than the other methods,
8. Use Eq. (5) to find the rotation at each point (Θi) using although it requires more computational effort. Lastly, the
numerical methods to calculate the integral. C1 is equal to the integration of curvature while neglecting tension stiffening
rotation at the starting point of the segment (ΘA), determined is a conservative option (41% on average) when a designer
in Step I. The trapezoidal rule or Simpson’s rule might be is not sure about the tension-stiffening characteristics of the
used to find the rotation based on the curvature at each point member.
(refer to Appendix D for the detailed equations). AUTHOR BIOS
9. Use Eq. (36) to find the deflection (Δ) at each point
using numerical methods (ΔA is determined in Step I). Again, Seyed Mohammad Hosseini is a Doctoral Candidate in the Department
of Civil and Building Engineering at the University of Sherbrooke, Sher-
the trapezoidal rule or Simpson’s rule might be used to find brooke, QC, Canada. He received his BSc and MSc from Isfahan Univer-
deflection (refer to Appendix D). sity of Technology (IUT), Isfahan, Iran. His research interests include the
internal and external use of fiber-reinforced polymers (FRPs) in reinforced
Δθ = ∫ΘθRsin(θ)dθ + ΔA (36)
Salaheldin Mousa is an FRQNT Postdoctoral Fellow in the Department of
Appendix D presents a design example according to the Civil and Building Engineering at the University of Sherbrooke, where he
also received his PhD. He is also a Lecturer at the Faculty of Engineering
presented procedure for one of the specimens tested. In the at Shoubra, Benha University, Cairo, Egypt, where he received his BSc and
example, the procedure overestimated the deflection by 8% MSc. His research interests include the use of FRPs in reinforced concrete
when following the Mod. method. The procedure entails structures.
two main sources of error: 1) using a numerical model to Hamdy M. Mohamed is a Lecturer and Research Associate in the Depart-
solve the integral; and 2) using an initial moment of inertia ment of Civil and Building Engineering at the University of Sherbrooke,
to obtain the initial internal forces, rotations, and deforma- where he received his PhD. He received his BSc and MSc from Helwan
University, Cairo, Egypt. His research interests include the use and field
tions at the supports in Step I. The former can be minimized applications of FRPs in reinforced concrete structures.
by increasing the number of elements in the analysis. In
addition, the latter might be improved by performing further Brahim Benmokrane, FACI, is a Professor in the Department of Civil
and Building Engineering at the University of Sherbrooke, Tier-1 Canada
rounds of analysis with the data obtained from the preceding Research Chair in Advanced Composite Materials Used for Civil Engi-
rounds. In addition to these sources of error, estimating the neering Structures, Senior Industrial Research Chair in Innovative FRP
exact cracking moment is of great importance in minimizing Composite Materials for Concrete Infrastructure, and Director of the
University of Sherbrooke Research Center on Structural FRP Composite
errors in calculating deflection. Materials for Concrete Structures (CRUSMAC). He is a member and past
Co-Chair of ACI Subcommittee 440-K, FRP-Material Characteristics,
CONCLUSIONS and a member of ACI Committees 435, Deflection of Concrete Building
Structures, and 440, Fiber-Reinforced Polymer Reinforcement; and ACI
The following conclusions can be drawn from the results Subcommittees ACI 440-E, FRP-Professional Education; ACI 440-F,
of this study: FRP-Repair-Strengthening; 440-H, FRP-Reinforced Concrete; ACI 440-I,
1. Applying the effective moment of inertia developed FRP-Prestressed Concrete; and ACI 440-L, FRP-Durability. He received
the ACI Foundation Arthur J. Boase Award in 2022. His research interests
for curvilinear members based on Bischoff’s model9 over- include the development of FRP reinforcement for concrete structures and
estimated deflection by 20%, 3%, and 12% in the normal- their durability, structural performance, and field applications.
strength concrete (NSC), high-strength concrete (HSC), and
fiber-reinforced concrete (FRC) specimens, respectively, on ACKNOWLEDGMENTS
This research was conducted with funding from the Natural Sciences and
average, for different reference points. Engineering Research Council of Canada (NSERC), the Pole de recherche
2. Neglecting tension stiffening in the integration-of- et d՚innovation en materiaux avancés au Quebec (PRIMA Quebec), Math-
curvature method overestimated deflection by 50%, 33%, ematics of Information Technology and Complex Systems (MITACS), the
Fonds de recherche du Québec en nature et technologies (FRQNT), and the
and 43% in the NSC, HSC, and FRC curvilinear specimens, Tier-1 Canada Research Chair in Advanced Composite Materials for Civil
respectively. Considering the contribution of tension stiff- Structures. The authors are grateful to the precast company (Sym-Tech
ening in curvilinear glass fiber-reinforced polymer-rein- Béton Préfabriqué, Sainte-Hyacinthe, QC, Canada) and to the GFRP bar
manufacturer (Pultrall Inc., Thetford Mines, QC, Canada) for their effective
forced concrete (GFRP-RC) members, however, underesti- involvement in this project. The authors also acknowledge the contribution
mated deflection by 33%, 54%, and 41% in the NSC, HSC, of the technical staff of the structural lab in the Department of Civil Engi-
and FRC specimens, respectively. neering at the University of Sherbrooke.
3. A comparison of the experimental and analytical results
revealed that the accuracy of the methodologies presented REFERENCES
1. Khaloo, A.; Moradi, H.; Kazemian, A.; and Shekarchi, M., “Exper-
depends on the reinforcement ratio, concrete strength, and imental Investigation on the Behavior of RC Arches Strengthened by
concrete type. A new model was proposed and validated GFRP Composites,” Construction and Building Materials, V. 235, 2020,
based on the experimental data to consider the effect of these p. 117519. doi: 10.1016/j.conbuildmat.2019.117519
2. ACI Committee 544, “Report on Design and Construction of Fiber-Re-
parameters on a deflection prediction method. The proposed inforced Precast Concrete Tunnel Segments (ACI 544.7R-16),” American
model could predict deflection at the service load with 2% Concrete Institute, Farmington Hills, MI, 2016, 41 pp.
conservativeness. 3. ACI Committee 533, “Guide for Precast Concrete Tunnel Segments
(ACI 533.5R-20),” American Concrete Institute, Farmington Hills, MI,
2020, 85 pp.
@seismicisolation
@seismicisolation

4. Caratelli, A.; Meda, A.; Rinaldi, Z.; and Spagnuolo, S., “Precast Tunnel Materials and Structures, V. 48, No. 10, 2015, pp. 3279-3293. doi: 10.1617/
Segments with GFRP Reinforcement,” Tunnelling and Underground Space s11527-014-0398-0
Technology, V. 60, 2016, pp. 10-20. doi: 10.1016/j.tust.2016.07.011 22. Bischoff, P. H., and Gross, S. P., “Equivalent Moment of Inertia
5. Robert, M., and Benmokrane, B., “Combined Effects of Saline Solu- Based on Integration of Curvature,” Journal of Composites for Construc-
tion and Moist Concrete on Long-Term Durability of GFRP Reinforcing tion, ASCE, V. 15, No. 3, 2011, pp. 263-273. doi: 10.1061/(ASCE)
Bars,” Construction and Building Materials, V. 38, 2013, pp. 274-284. doi: CC.1943-5614.0000164
10.1016/j.conbuildmat.2012.08.021 23. CSA S806-12, “Design and Construction of Building Components
6. ACI Committee 440, “Guide for the Design and Construction of with Fibre Reinforced Polymers,” CSA Group, Toronto, ON, Canada, 2012,
Structural Concrete Reinforced with Fiber-Reinforced Polymer Bars (ACI 187 pp.
440.1R-15),” American Concrete Institute, Farmington Hills, MI, 2015, 24. Rasheed, H. A.; Nayal, R.; and Melhem, H., “Response Prediction of
88 pp. Concrete Beams Reinforced with FRP Bars,” Composite Structures, V. 65,
7. Mota, C.; Alminar, S.; and Svecova, D., “Critical Review of No. 2, 2004, pp. 193-204. doi: 10.1016/j.compstruct.2003.10.016
Deflection Formulas for FRP-RC Members,” Journal of Composites 25. Razaqpur, A.; Svecova, D.; and Cheung, M. S., “Rational Method for
for Construction, ASCE, V. 10, No. 3, 2006, pp. 183-194. doi: 10.1061/ Calculating Deflection of Fiber-Reinforced Polymer Reinforced Beams,”
(ASCE)1090-0268(2006)10:3(183) ACI Structural Journal, V. 97, No. 1, Jan.-Feb. 2000, pp. 175-184.
8. Branson, D. E., “Instantaneous and Time-Dependent Deflections of 26. Bischoff, P. H., and Gross, S. P., “Design Approach for Calculating
Simple and Continuous Reinforced Concrete Beams,” HPR Report No. Deflection of FRP-Reinforced Concrete,” Journal of Composites for
7, Part I, Alabama Highway Department, Bureau of Public Roads, Mont- Construction, ASCE, V. 15, No. 4, 2011, pp. 490-499. doi: 10.1061/(ASCE)
gomery, AL, 1963, pp. 1-78. CC.1943-5614.0000195
9. Bischoff, P. H., “Reevaluation of Deflection Prediction for Concrete 27. Hosseini, S. M.; Mousa, S.; Mohamed, H. M.; and Benmokrane, B.,
Beams Reinforced with Steel and Fiber Reinforced Polymer Bars,” Journal “Structural Behavior of Precast Reinforced Concrete Tunnel Segments with
of Structural Engineering, ASCE, V. 131, No. 5, 2005, pp. 752-767. doi: Glass Fiber-Reinforced Polymer Bars and Ties under Bending Load,” ACI
10.1061/(ASCE)0733-9445(2005)131:5(752) Structural Journal, V. 119, No. 1, Jan. 2022, pp. 307-319.
10. Bischoff, P. H., “Comparison of Existing Approaches for Computing 28. Hosseini, S. M.; Mousa, S.; Mohamed, H. M.; Eslami, A.; and
Deflection of Reinforced Concrete,” ACI Structural Journal, V. 117, No. 1, Benmokrane, B., “Experimental and Analytical Study on Precast High-
Jan. 2020, pp. 231-240. doi: 10.14359/51718072 Strength Concrete Tunnel Lining Segments Reinforced with GFRP Bars,”
11. Bischoff, P. H., and Scanlon, A., “Effective Moment of Inertia for Journal of Composites for Construction, ASCE, V. 26, No. 5, 2022,
Calculating Deflections of Concrete Members Containing Steel Rein- p. 04022062. doi: 10.1061/(ASCE)CC.1943-5614.0001257
forcement and Fiber-Reinforced Polymer Reinforcement,” ACI Structural 29. ISIS Canada Research Network, “Reinforcing Concrete Structures
Journal, V. 104, No. 1, Jan.-Feb. 2007, pp. 68-75. with Fibre Reinforced Polymers,” Design Manual No. 3, University of
12. Adam, M. A.; Said, M.; Mahmoud, A. A.; and Shanour, A. S., Manitoba, Winnipeg, MB, Canada, 2007.
“Analytical and Experimental Flexural Behavior of Concrete Beams 30. Boresi, A. P.; Sidebottom, O. M.; and Saunders, H., Advanced
Reinforced with Glass Fiber Reinforced Polymers Bars,” Construc- Mechanics of Materials, fourth edition, John Wiley & Sons Inc., New York,
tion and Building Materials, V. 84, 2015, pp. 354-366. doi: 10.1016/j. 1985, 763 pp.
conbuildmat.2015.03.057 31. ACI Committee 318, “Building Code Requirements for Structural
13. Alsayed, S.; Al-Salloum, Y.; and Almusallam, T., “Performance of Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved
Glass Fiber Reinforced Plastic Bars as a Reinforcing Material for Concrete 2022),” American Concrete Institute, Farmington Hills, MI, 2019, 624 pp.
Structures,” Composites Part B: Engineering, V. 31, No. 6-7, 2000, 32. Scanlon, A., and Bischoff, P. H., “Shrinkage Restraint and Loading
pp. 555-567. doi: 10.1016/S1359-8368(99)00049-9 History Effects on Deflections of Flexural Members,” ACI Structural
14. Arabshahi, A.; Tavakol, M.; Sabzi, J.; and Gharaei-Moghaddam, N., Journal, V. 105, No. 4, July-Aug. 2008, pp. 498-506.
“Prediction of the Effective Moment of Inertia for Concrete Beams Rein- 33. Park, R., and Paulay, T., Reinforced Concrete Structures, John Wiley
forced with FRP Bars Using an Evolutionary Algorithm,” Structures, V. 35, & Sons, Inc., New York, 1991.
2022, pp. 684-705. 34. ACI Committee 544, “Guide to Design with Fiber-Reinforced
15. Benmokrane, B.; Chaallal, O.; and Masmoudi, R., “Flexural Concrete (ACI 544.4R-18),” American Concrete Institute, Farmington
Response of Concrete Beams Reinforced with FRP Reinforcing Bars,” ACI Hills, MI, 2018, 44 pp.
Structural Journal, V. 93, No. 1, Jan.-Feb. 1996, pp. 46-55. 35. ASTM C1609-19, “Standard Test Method for Flexural Performance
16. Gao, D.; Benmokrane, B.; and Masmoudi, R., “A Calculating Method of Fiber-Reinforced Concrete (Using Beam with Third-Point Loading), ”
of Flexural Properties of FRP-Reinforced Concrete Beam: Part 1: Crack ASTM International, West Conshohocken, PA, 2019, 9 pp.
Width and Deflection,” technical report, Department of Civil Engineering, 36. Bischoff, P., “Deflection Calculation Using an Effective Moment
University of Sherbrooke, Sherbrooke, QC, Canada, 1998. Inertia for FRC,” Deflection and Stiffness Issues in FRC and Thin Structural
17. Mousavi, S. R., and Esfahani, M. R., “Effective Moment of Inertia Elements, SP-248, P. H. Bischoff and F. Malhas, eds., American Concrete
Prediction of FRP-Reinforced Concrete Beams Based on Experimental Institute, Farmington Hills, MI, 2007, pp. 17-30.
Results,” Journal of Composites for Construction, ASCE, V. 16, No. 5, 37. ACI Committee 363, “Report on High-Strength Concrete (ACI 363R-
2012, pp. 490-498. doi: 10.1061/(ASCE)CC.1943-5614.0000284 10),” American Concrete Institute, Farmington Hills, MI, 2010, 65 pp.
18. Thériault, M., and Benmokrane, B., “Effects of FRP Reinforcement 38. Mousa, S.; Mohamed, H. M.; and Benmokrane, B., “Deflection
Ratio and Concrete Strength on Flexural Behavior of Concrete Beams,” Prediction Methodology for Circular Concrete Members Reinforced with
Journal of Composites for Construction, ASCE, V. 2, No. 1, 1998, pp. 7-16. Fiber-Reinforced Polymer Bars,” ACI Structural Journal, V. 116, No. 2,
doi: 10.1061/(ASCE)1090-0268(1998)2:1(7) Mar. 2019, pp. 279-293. doi: 10.14359/51713293
19. Toutanji, H. A., and Saafi, M., “Flexural Behavior of Concrete Beams 39. Ng, P. L.; Gribniak, V.; Jakubovskis, R.; and Rimkus, A., “Tension
Reinforced with Glass Fiber-Reinforced Polymer (GFRP) Bars,” ACI Stiffening Approach for Deformation Assessment of Flexural Reinforced
Structural Journal, V. 97, No. 5, Sept.-Oct. 2000, pp. 712-719. Concrete Members under Compressive Axial Load,” Structural Concrete,
20. Yost, J. R.; Gross, S. P.; and Dinehart, D. W., “Effective Moment of V. 20, No. 6, 2019, pp. 2056-2068. doi: 10.1002/suco.201800286
Inertia for Glass Fiber-Reinforced Polymer-Reinforced Concrete Beams,” 40. Zadeh, H. J., and Nanni, A., “Flexural Stiffness and Second-Order
ACI Structural Journal, V. 100, No. 6, Nov.-Dec. 2003, pp. 732-739. Effects in Fiber-Reinforced Polymer-Reinforced Concrete Frames,” ACI
21. Zhang, L.; Sun, Y.; and Xiong, W., “Experimental Study on the Structural Journal, V. 114, No. 2, Mar.-Apr. 2017, pp. 533-543. doi:
Flexural Deflections of Concrete Beam Reinforced with Basalt FRP Bars,” 10.14359/51689257
@seismicisolation
@seismicisolation

ARE YOU A RESEARCHER?
SIGN UP FOR TODAY!
ORCID provides a persistent digital identifier 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
@seismicisolation
@seismicisolation
Title No. 120-S89
Seismic-Fire-Combined Loadings Applied to Carbon

Fiber-Reinforced Polymer-Confined Reinforced Concrete
Columns
by Ju-Hyung Kim, Yail J. Kim, and Jun Wang
This paper presents analytical investigations into the behavior configurations are frequently adopted for experimental and
of a reinforced concrete column with and without carbon fiber- theoretical investigations.4,5
reinforced polymer (CFRP) confinement when subjected to Numerous buildings built prior to the enforcement of
earthquake and fire loadings. A data set of 100 ground motions seismic codes, particularly before the 1970s, are consid-
covering short and long durations is collected and integrated
ered deficient and may not safely accommodate lateral sway
with 0 to 3 hours of fire exposure. Two strengthening categories
precipitated by earthquakes. Non-seismically designed
are implemented: 1) one to six CFRP layers; and 2) six layers of
CFRP with a 40 mm (1.6 in.) thick insulation. A computational members thus encounter increased vulnerability to failure
platform incorporating autonomous discrete entities is used for and, in lieu of costly reconstruction, rehabilitation is
the simulation of heat transfer, while static pushover and nonlinear preferred to address assorted issues induced by the inade-
dynamic analyses predict the seismic response of the unconfined quate capacity of those nonconforming ones.3 Because the
and confined columns. Thermal gradients are generated across the primary objective of seismic retrofitting is to ameliorate the
column section to identify the physical and mechanical properties strength and ductility of structural elements against exces-
of constituents at elevated temperatures, which are linked with sive drift ratios,6 relevant rehabilitation strategies would
the static and dynamic models. The CFRP-confined column with mitigate the risk of physical impairment. Among prevalent
insulation outperforms its unconfined counterpart from a behav- strengthening techniques that enhance the capacity of rein-
ioral standpoint, specifically for axial capacities, flexural failure,
forced concrete columns, such as enlarging cross sections
energy dissipation, and deformability. The implications of the
and adding steel plates,7 confinement with carbon fiber-
seismic-fire-combined loadings are remarkable in terms of
degrading the load-resisting ability of the columns compared with reinforced polymer (CFRP) sheets is regarded as a convenient
those of the uncoupled actions. The duration of the ground motions and efficient method8 and is widely implemented around the
dominates the development of a relationship between the spectral world.9 Specifically speaking, the efficacy of CFRP strength-
acceleration and drift ratio of the columns. Design recommenda- ening is remarkable for seismic upgrading associated with
tions are rendered to address the limitations of current practice. the ductile failure, inelastic rotations, and energy dissipa-
tion of substandard columns.10 Another notable benefit of
Keywords: carbon fiber-reinforced polymer (CFRP); column; earthquake the CFRP application is that it raises the strength of existing
duration; fire endurance; modeling; seismic performance.
columns without changing stiffness, which is desirable for
preserving the magnitude of seismic forces.11
INTRODUCTION
Seismic events may ravage energy lines and electricity
The seismic failure of structural elements is contingent
networks in built environments and can prompt fires. A clas-
upon the degree of resistance to earthquake intensities and
sical instance is found in the 1906 San Francisco earthquake,
wave directions. Sequential earthquakes (also known as
which was accompanied by multiple fires: the disaster devas-
aftershocks) aggravate the deterioration of load-bearing
tated the city, and 498 people died and more than 225,000
components, thereby reducing the overall performance of
people became homeless.12 Accordingly, extensive research
building structures. Previous studies enunciated that the
has been conducted for the last several decades to figure
amplitude, frequency, and duration of oscillating ground
out the implications of earthquakes, fires, and a combina-
motions are crucial factors controlling the detrimental impact
tion thereof.13,14 Most cases, however, focused on their own
of seismic waves.1,2 Technical regulations allow a certain
individual consequences,8,15 and insufficient efforts were
extent of seismic damage within the boundary of preventing
expended to understand interactions between these hazards,
the collapse of buildings by developing plastic hinges that
which can bring about a significantly high level of damage
alleviate externally generated excitations.3 As far as the
in constructed facilities.16 It is worth noting that the seismic
stability of a structural system is concerned, columns play
resistance of fire-damaged members is not comparable to that
an important role in maintaining force equilibrium. When
of intact members, and the likelihood of collapse increases
seismic loading is applied to a column, the adequacy of axial
capacity needs to be combined with appropriate ductility ACI Structural Journal, V. 120, No. 5, September 2023.
so that the physical failure of the member is retarded until MS No. S-2023-014.R1, doi: 10.14359/51738843, received April 26, 2023, and
substantial energy is dissipated. The behavior of columns Institute. All rights reserved, including the making of copies unless permission is
near the base where connections are made with floors obtained from the copyright proprietors. Pertinent discussion including author’s
or footings is of interest and, that being so, cantilevered is received within four months of the paper’s print publication.
@seismicisolation
@seismicisolation

Fig. 1—Benchmark column: (a) dimensions; and (b) strengthening.
under the multi-hazard environment.17 Contemplating a each, where As is the cross-sectional area of the reinforcing
dearth of provisions in published specifications related to the bar), and No. 4 closed ties (As = 129 mm2 [0.2 in.2] each)
effects of thermomechanical loadings,3,18 practitioners may were placed at spacings of 200 mm (8 in.), as depicted in
not properly prepare for such an extreme event at the design Fig. 1(a). The compressive strength of the concrete was fc′ =
stage of new and rehabilitated structures. 30 MPa (4350 psi) and the yield strength of the reinforcing
This paper explores the repercussions of seismic-fire- steel was fy = 420 MPa (60 ksi). Given that the column was
combined loadings on the behavior of a non-slender rein- built on a mat foundation,20 its geometric configuration was
forced concrete column confined with CFRP sheets. A treated as a cantilever: conforming to previous studies,4,5
two-fold analytical program, comprising static pushover this simplified boundary condition was deemed suitable for
and dynamic analyses, is carried out to deal with short- and evaluating hysteretic responses at the connection level under
long-duration earthquakes alongside the ASTM E119-20 seismic loadings. To upgrade the capacity and ductility of
standard fire,19 including a heat transfer model. After the column, one to six layers of CFRP sheets were applied
validating the predictive approaches, an extensive para- (Fig. 1(b)). The tensile strength, elastic modulus, and ulti-
metric study is executed with the aim of proposing perfor- mate strain of CFRP were ffu = 3800 MPa (550 ksi), Ef =
mance-based design guidelines. 227 GPa (33,000 ksi), and εfu = 0.167, respectively, based on
an equivalent fiber thickness of tf = 0.165 mm (0.0065 in.).
RESEARCH SIGNIFICANCE The column was assumed to carry an axial load of P =
The current state of knowledge is incomplete to address 0.2fc′Ag, where Ag is the gross-sectional area of the column,
concerns arising from synergistic distress consisting of representing a combination of service design loads.21 The
fire and seismic loadings, especially under variable earth- influence of insulation was also examined using the six-layer
quake durations. In addition, there is a lack of information CFRP case (designated C6). The insulation type was a spray-
on the behavior of CFRP-strengthened reinforced concrete able cementitious material,22 which is commonplace for fire-
columns exposed to those adverse circumstances that are proofing,23 and possessed a thickness of 40 mm (1.6 in.),
presumable in nonconforming buildings. To accomplish an average density of 256 kg/m3 (15.9 lb/ft3), a thermal
resilient structural systems, a scientific understanding of conductivity of 0.0815 W/(m∙°C) (0.047 BTU/(h∙ft∙°F)),
the interrelations between the seismic and fire loadings is and a specific heat of 1047 J/(kg∙°C) (0.25 BTU/(lb∙°F)).
a prerequisite, whereas prescriptive specifications do not Because a fire rating of 3 hours is the well-accepted norm in
offer provisions to effectively handle this important aspect.18 the building community,24 the thermomechanical behavior
The outcomes from the present study are intended to clarify of the column was studied from 0 (a prefire state for refer-
ambiguous design schemes with and without CFRP for ence) to 3 hours.
columns subjected to seismic-fire-combined loadings.
DURATION OF EARTHQUAKES
BENCHMARK COLUMN Earthquake ground motions incorporating a variety of
A typical reinforced concrete column was taken from durations were collected to simulate the seismic behavior
the first floor of a 20-year-old building,20 and minor adjust- of the benchmark column. A characteristic duration was
ments were made for a diameter of D = 860 mm (34 in.) defined as a time interval between 5 and 75% of cumulative
and a height of h = 4200 mm (13.8 ft). The circular column Arias intensity measures (DS5-75), which is useful to appraise
was reinforced with 24 No. 9 bars (As = 645 mm2 [1.0 in.2] the seismic response of a structure25,26
@seismicisolation
@seismicisolation

Fig. 2—Description of characteristic earthquake duration (DS5-75): (a) time history; (b) Arias intensity; and (c) spectrally
equivalent short- and long-duration motions.
Table 1—Fifty sets of spectrally equivalent earthquake ground motions
Short duration (DS5-75 < 25 seconds) Long duration (DS5-75 ≥ 25 seconds) Short duration (DS5-75 < 25 seconds) Long duration (DS5-75 ≥ 25 seconds)
Title PGA, DS5-75, Title PGA DS5-75 Scale Title PGA, DS5-75, Title PGA, DS5-75, Scale
No. (station) g seconds (station) (g) (seconds) factor No. (station) g seconds (station) g seconds factor
Chalfant
Valparaiso Petrolia Tohoku
1 Valley 0.444 2.20 0.437 27.6 1.31 26 0.498 2.30 1.044 67.5 1.50
(Llol.) (Petrolia) (Koho.)
(CZR)
Big Bear Landers Loma Prieta Tohoku

2 0.111 11.8 0.102 25.1 1.04 27 0.442 1.40 0.414 75.8 0.71
(SB) (ICC) (G#1) (Niho.)
Mt. Lewis Landers Whittier Tohoku

3 0.147 1.00 0.271 25.8 0.92 28 0.537 2.60 0.414 73.5 0.81
(HVGP) (IJR) (TCHN) (Niho.)
Landers Landers Coalinga Tohoku

4 0.135 12.8 0.099 25.5 0.97 29 0.288 3.10 0.418 56.7 1.63
(BVH) (TPPC) (CCS) (Aidu.)
Alum Rock Kocaeli Loma Prieta Tohoku

5 0.160 1.40 0.192 27.8 1.31 30 0.351 1.90 0.418 69.4 1.67
area (CL) (Fatih) (G#2) (Aidu.)
IA = (π / 2g)∫tda( t)2dt

determined. In accordance with the methodology proposed
(1)
0
by Chandramohan et al.,27 ground motions were classified
where IA is the Arias intensity; g is the gravitational accelera- into two categories (short [DS5-75 < 25 seconds] and long
tion (1g = 9.81 m/s = 32.2 ft/s ); and td and a(t) are the time
2 2 [DS5-75 ≥ 25 seconds] durations), and these were paired to
span and history of the recorded acceleration, respectively. isolate the effects of earthquake durations with a focus on
As exemplified in Fig. 2(a) and (b), accelerations recorded magnitude and frequency. For implementation, 300 accel-
at a seismograph station (Fig. 2(a)) were normalized to erograms of short-duration earthquakes with a magni-
identify a range of the cumulative intensities from 5 to 75% tude greater than 6.0 were gleaned from various sources
(Fig. 2(b)); then, the characteristic duration (DS5-75 ) was (Table 1, footnote) and compared against 100 long-duration
@seismicisolation
@seismicisolation

Table 1 (cont.)—Fifty sets of spectrally equivalent earthquake ground motions
Chalfant
Hokkaido Northridge Tohoku
6 Valley 0.125 7.70 0.099 26.3 1.10 31 0.239 8.30 0.318 77.3 1.05
(Oiwa.) (LA-B) (Fukush.)
(BSS)
Chino Hills Hokkaido Anza Tohoku

7 0.177 0.50 0.132 28.0 1.07 32 0.397 0.90 0.299 80.8 0.96
(GGCG) (Haya.) (LQBD) (Kawa.)
Coyote
Hokkaido La Habra Tohoku
8 Lake 0.249 1.40 0.132 25.3 1.05 33 0.361 0.60 0.299 85.4 1.01
(Haya.) (Walnut) (Kawa.)
(CLD)
Chi-Chi Hokkaido Northridge Tohoku

9 0.138 6.70 0.109 40.8 1.47 34 0.316 5.90 0.503 78.7 1.97
(HWA003) (Kuri.) (LA-C) (Miharu)
Loma
Maule Whittier Tohoku
10 Prieta 0.542 1.70 0.684 30.2 1.29 35 0.392 1.50 0.900 71.3 1.68
(Angol) (LA116) (Miyak.)
(GSF)
Calexico Maule Northridge Tohoku

11 0.383 14.9 0.527 31.9 1.62 36 0.434 2.00 0.407 75.5 1.39
(El Centro) (Const.) (SPD) (Sakun.)
Northridge Maule Big Bear Tohoku

12 0.389 5.90 0.465 38.2 1.16 37 0.545 5.50 0.407 66.7 0.85
(LA-H) (Curico) (BBL) (Sakun.)
Supersti- Maule Landers Tohoku

13 0.341 7.10 0.375 33.7 1.53 38 0.284 21.7 0.352 69.3 1.31
tion Hills (Hualane) (JTFS) (Kakuda)
Chi-Chi Maule Tottori Tohoku

14 0.139 6.30 0.182 27.6 1.21 39 0.391 4.60 0.352 70.8 0.78
(CHY047) (Santiago) (TTR008) (Kakuda)
Loma
Maule Landers Tohoku
15 Prieta 0.356 1.60 0.462 51.4 1.34 40 0.171 21.4 0.254 70.4 1.25
(Talca) (DHS) (Iwanu.)
(GGCS)
Coyote Maule Petrolia Tohoku

16 0.254 0.90 0.462 51.7 1.61 41 0.193 4.20 0.199 73.3 1.23
Lake (GH) (Talca) (Fortuna) (Higash.)
San
Kocaeli El Mayor Tohoku
17 0.253 9.70 0.244 23.6 0.82 42 Fernando 1.171 5.40 0.199 69.1 0.20
(Ambarli) (Chih.) (Higash.)
(SP)
El
El Mayor Coalinga Tohoku
18 Mayor-Cu- 0.183 15.4 0.244 26.8 1.29 43 0.109 6.10 0.120 81.0 1.42
(Chih.) (2W) (Kamin.)
capah (3)
Kocaeli El Mayor Coalinga Tohoku

19 0.230 6.40 0.201 28.3 0.83 44 0.176 5.60 0.207 78.1 1.39
(İzmit) (Tama.) (2E) (Yonez.)
Tottori Tohoku Niigata Tohoku

20 0.179 5.70 0.283 77.1 1.25 45 0.132 16.2 0.174 71.2 1.21
(SMN002) (Yana.) (NIG013) (Tendou)
Parkfield Tohoku Landers Tohoku

21 0.229 3.20 0.283 75.8 1.15 46 0.118 22.4 0.174 64.0 1.84
(Zone 15b) (Yana.) (Mecca) (Tendou)
Parkfield Tohoku Northridge Tohoku

22 0.546 1.60 0.577 74.0 1.15 47 0.178 7.70 0.204 81.7 1.14
(Zone 8) (Fukush.) (PM) (Takaha.)
La Habra Tohoku Big Bear Tohoku

23 0.356 1.30 0.577 71.2 1.14 48 0.225 5.70 0.204 79.7 0.84
(Fullerton) (Fukush.) (DHS) (Takaha.)
La Habra Tohoku Yountville Tohoku

24 0.330 1.60 0.506 76.3 1.32 49 0.340 1.60 0.208 65.3 0.67
(La Habra) (Iitate) (Napa) (Yonez.)
La Habra Tohoku Tohoku

25 0.703 0.60 0.506 77.9 1.73 50 Landers (FI) 0.122 7.90 0.208 70.0 1.50
(Brea) (Iitate) (Yonez.)
Note: Web sources: Center for Engineering Strong Motion Data (https://www.strongmotioncenter.org/), NGA-West2 database (https://peer.berkeley.edu/ngawest2/databases/), and
National Research Institute for Earth Science and Disaster Prevention (https://www.kyoshin.bosa.go.jp/); PGA is peak ground acceleration; DS5-75 is earthquake duration corre-
sponding to 5 to 75% of cumulative Arias intensity; 1 g = 9.81 m/s2 = 32.2 ft/s2.
earthquakes excerpted from Chandramohan et al.27; after- complete set of the manipulated ground motions is visible
ward, their spectrally equivalent motions were identified in Fig. 3.
in tandem with 5%-damped pseudo-acceleration spectra.3
Each of the 50 pairs between the 100 short- and long-dura- HEAT TRANSFER
tion earthquakes revealed minimum squared errors and opti- Conduction and thermal properties
mized scale factors for the individual sets listed in Table 1. The governing equation of heat transfer in the benchmark
By using these spectrally equivalent ground motions, unnec- column may be expressed by Fourier’s law
essary distractors arising from the morphological dissimi-
larity of the paired durations are eliminated (Fig. 2(c)). A ∂ T(x, t)
_ ∂ T2(x, t)
_

@seismicisolation
@seismicisolation = α (2)
∂t ∂ x2
Fig. 3—Fifty spectrally equivalent earthquake motion sets.
where T(x,t) is the temperature at location x and time t; others was invariant (Fig. 4(h) and (i)), except for the thermal
and α is the thermal diffusivity (α = λ/(Cpρ), in which λ is decomposition of CFRP at 400°C (752°C). Figures 4(j) to
the thermal conductivity; Cp is the specific heat; and ρ is (l) summarize the thermal diffusivity of the concrete, CFRP,
the density). In compliance with the recommendations of and insulation. Equation (3) reproduces the ASTM E119
preceding research,28,29 the thermal contribution of rein- standard fire,36 as graphed in Fig. 5(a)
forcement in the concrete column was ignored. Figure 4 plots _ _
the thermal properties of the constituent materials predicted T = 750(1 − exp(− 3.79553√th))+ 170.41√th + T0 (3)
by previously reported equations.30-32 While the conduc-
tivity of the concrete and CFRP descended with tempera- where T is the applied temperature in Celsius; th is the
ture (Fig. 4(a) and (b)), that of the insulation was constant heating time in hours; and T0 is the reference temperature
(Fig. 4(c)). The abrupt drop in the CFRP’s conductivity at (T0 = 20°C [68°F]).
400°C (752°F) was attributed to the breaking of chemical
bonds in the resin, accompanied by a chain scission process Formulation and validation
that lowered the residual mass of effective molecules in the A heat transfer model was developed using a computa-
polymeric composition and precipitated interfacial failure tional platform built with discrete entities, which is called
between the fibers and the matrix.33,34 For modeling conve- agent-based modeling. This nontraditional approach is
nience,35 the mechanical resistance of CFRP was disregarded often employed in social science to study a reciprocal
beyond the thermal decomposition temperature of 400°C relationship between autonomous individuals,40 and the
(752°F) that caused the malfunctioning of the composite. concept was useful to predict the interactive behavior of
Unlike the stable case of the concrete (Fig. 4(d)), the specific the column components during a fire. An open-source code,
heat of the CFRP and insulation was temperature-dependent NetLogo, formed the basis of heat transfer in conjunction
(Fig. 4(e) and (f), respectively). The endothermic reactions with the aforementioned material properties. Further expla-
of these materials involving the conversion of molecular nations on the background and implementation of the code
kinetic energy to chemical energy were responsible for such are available elsewhere.41 Figures 5(b) and (c) compare
erratic behavior. 36-38
The variation in the concrete’s density the theoretical temperature of concrete members with and
was marginal up to 1063°C (1945°F) on account of changes without CFRP strengthening against experimental data
in its mineralogical composition and free water contents at obtained from the literature.22,41 The compressive strength
elevated temperatures (Fig. 4(g) ); however, the density of
39
of the concrete spanned between 28 and 39 MPa (4061
@seismicisolation
@seismicisolation

Fig. 4—Temperature-dependent properties of constituent materials: (a) thermal conductivity of concrete; (b) thermal conduc-
tivity of CFRP; (c) thermal conductivity of insulation; (d) specific heat of concrete; (e) specific heat of CFRP; (f) specific heat of
insulation; (g) density of concrete; (h) density of CFRP; (i) density of insulation; (j) thermal diffusivity of concrete; (k) thermal
diffusivity of CFRP; and (l) thermal diffusivity of insulation.
Fig. 5—Validation of heat transfer model: (a) ASTM E119-2019; (b) plain concrete (test: Weerasinghe et al.42); and (c) confined
concrete with insulation (test: Williams et al.22).
and 5656 psi). The thickness of the CFRP and insulation Appendix A* contains detailed information. As shown in
layers was 1 mm (0.0394 in.) and 38 mm (1.5 in.), respec- Fig. 6(a), both the strength and stiffness of the concrete
tively, and their thermal properties were the same as those declined with the increased thermal load. The yield strength
in the present study (Fig. 4). Temperatures were recorded of the reinforcing bars was sustained up to 400°C (752°F),
at 45 mm (1.8 in.) from the concrete surface42 and in the after which noticeable degradation was rendered (Fig. 6(b)).
vicinity of concrete and insulation levels.22 The computed CFRP sheet—For the temperature-dependent mechanical
and measured values agreed across the board. properties of CFRP, Eq. (4) and (5) may be used43
PREDICTION OF CAPACITY DEGRADATION Ef(T)/Ef = 0.475tanh{–8.68 × 10–3(T – 367.41)} + 0.525

Material modeling (4)
Unconfined concrete and reinforcing bar—The full consti-
tutive relationships of plain concrete and reinforcing bars *
at elevated temperatures were attained from Eurocode 231; appended to the online version of the published paper. It is also available in hard copy
from ACI headquarters for a fee equal to the cost of reproduction plus handling at the
time of the request.
@seismicisolation
@seismicisolation

Fig. 6—Material properties at elevated temperatures: (a) compressive stress-strain of unconfined concrete; (b) yield strength of
reinforcing bar; (c) elastic modulus of CFRP sheet; (d) tensile strength of CFRP sheet; (e) confining stress; and (f) compressive
stress-strain of confined concrete.
fcc′− fc′
_________
ffu(T)/ffu = 0.45tanh{–5.83 × 10–3(T – 339.54)} + 0.55 (5) E2 =   εccu 
(9)
where Ef(T) and ffu(T) are the elastic modulus and tensile
strength of CFRP at temperature T in Celsius. The hyper- 2fc′
______
bolic responses of CFRP dwindled until thermal decomposi- εt′ =   E− E  
(10)
c 2
tion occurred (Fig. 6(c) and (d)). According to experimental
observations,44 the linearity of stress-strain in CFRP can be where fc and εc are the stress and strain of the confined
preserved in fire
concrete, respectively; and Ec is the elastic modulus of the
core concrete. The compressive strength of the confined
ff(T) = Ef(T)εf (6)
concrete (fcc′) and its maximum strain (εc,max) are shown in
Eq. (11) and (12), respectively10
where ff(T) is the stress of CFRP at temperature T; and εf is
the CFRP strain.
fcc′ = fc′ + ψf3.3κa fl (11)
Confined concrete—Pursuant to ACI 440.2R-17,10 the
confining pressure of core concrete (fl(T)) in the column is
calculated by
εc,max ≤ εccu ≤ 0.01 (12)
fl(T) = 2Ef(T)ntfεfe/D (7)
εccu = εc′(1.50 + 12κb(fl/fc′)(εfe/εc′)0.45) (13)
where n is the number of the confining layers; and εfe is the
effective strain of CFRP (εfe = 0.55εfu). When more layers
where ψf is a reduction factor (ψf = 0.95); κa and κb are the
were applied, the degree of confinement was raised and its
geometry and efficiency factors (κa = κb = 1.0 for circular
sensitivity to elevated temperatures increased (Fig. 6(e)).
columns); and εc′ is the compressive strain of the core
This fact points out that the ramifications of fire are signifi-
concrete (εc′ = 0.002). Substituting Eq. (7) and (11) into
cant, justifying the demand for insulation to retain the effec-
Eq. (8) with the Eurocode model for the unconfined concrete
tiveness of CFRP strengthening until an intended fire rating
yields the temperature-dependent constitutive relationship
is achieved. The stress-strain curve of the confined concrete
of the confined concrete (Fig. 6(f)).
is written as10
Static pushover analysis
{ for εt′ ≤ εc ≤ εc,max}
Ecεc− ( Ec− E2 )2/ (4fc′) for0 ≤ εc ≤ εt′

fc =         A plastic hinge model was formulated with the following
fc′+ E2 εc
assumptions3,21,45 to establish a moment-curvature relation-
(8)ship (Fig. 7(a)):
@seismicisolation
@seismicisolation

and db is the diameter of the longitudinal reinforcing bar.
Compared with the plastic hinge equation of ACI 440.2R-17
that merely considers a gap between adjacent CFRP wraps,10
Eq. (17) contains the confining term (fl(T)) to interconnect
thermally degraded material properties with the hinge length.
Hysteretic simulation
Conceptual development—The benchmark column was
modeled as a single-degree-of-freedom (SDOF) system. This
archetypal approximation for dynamic analysis is justified
by the fact that the plastic hinge region (Fig. 7(a)) dominates
the lateral displacement of the column mass.48 A simpli-
fied trilinear load-displacement curve was then constructed
(Fig. 7(b)). In line with ACI 374.2R-13,49 the initial stiffness
of the curve (Ke) was characterized using the point at which
a lateral load equaled 75% of the maximum load (0.75F0),
and the fundamental period of the column was expressed
to be T1 = 2π(P/(gKe))0.5, in which P is the axial load. The
load and displacement of the column at yielding (Fy and uy,
respectively, and Fy = Keuy) were calculated as instructed in
ASCE/SEI 41-17.3 When the maximum displacement (um)
coincided with the value at a post-peak load of 0.75F0, tanta-
mount to a capacity loss of 25%, the column failed.49
Framework—Because the hysteretic approach proposed
by Ibarra et al.50 adequately models in-cycle deterioration
along with seismic energy dissipation that is essential for
elucidating structural damage,7 it was adopted to simu-
Fig. 7—Hysteretic model: (a) idealized behavior of canti- late the behavior of the benchmark column. The extent of
levered column; (b) trilinear envelope curve; and (c) cyclic cyclic degradation can be quantified through a change in the
deterioration. amount of energy dissipation50
• The shear deformation of the benchmark column is
βi = (Ei/(Et− ∑ ij=1Ej))
c
negligible. (18)
• Plane sections remain plane before and after bending.
• Strain compatibility is valid between the constituents of where βi is the in-cycle degradation parameter at the i-th
the column. alternation (one full load reversal comprises the i-th and
• The interfacial slip of CFRP is negligible at the i+1-th alternations [i ≥ 1], as retraced in Fig. 7(c)); Ei and
concrete-surface level. Ej are the dissipated hysteretic energy values (∫F(u)du, 1 ≤
At fire-exposure time th, the lateral capacity and displace- j ≤ i) at the current and previous alternations (that is, ΣEj
ment of the column (F(th) and u(th), respectively) are deter- is the cumulative hysteretic energy from the first to the i-th
mined by alternations); Et is the reference energy (Et = γFyuy, in which
γ is an empirical constant: γ = 120 for UC columns51 and a
F(th) = Mn(th)/h (14) value should be found for confined ones); and c is the rate
of deterioration (c = 1.0 is used for most reinforced concrete
columns51). With the progression of cyclic loadings, the
u(th) = ue(th) + up(th) = (15) degradation parameter (βi) is updated every load reversal
1/3ϕy(th)h2 + (ϕu(th) – ϕy(th))lp(h – 0.5lp) so that the in-cycle deterioration of the column is computed
from an envelope generated from the pushover analysis
where Mn(th) is the ultimate moment; ue(th) and up(th) are the (Fig. 7(c))
elastic and plastic deformations, respectively; ϕy(th) and ϕu(th)
are the yield and maximum curvatures, respectively; and lp is Fi = (1 – βi)Fi–1 (19)
the plastic hinge length, which can be estimated by46,47
where Fi is the load of the post-peak envelope at the i-th
lp,unconf(T) = 0.08h + 0.022db fy(T) (16) alternation commencing from the initial load of the envelope
without cyclic degradation (F0).
lp,conf(T) = 0.8(fl(T)/fc′(T))h + 0.022db fy(T) (17) Validation

The validation of the predictive methods is provided
where lp,unconf(T) and lp,conf(T) are the hinge length of the in Fig. 8. The dimensions of the test columns52 were D =
unconfined (UC) and CFRP-confined columns, respectively;@seismicisolation
@seismicisolation

Fig. 8—Validation of predictive methods: (a) pushover; (b) calibration of energy constant γ; and (c) hysteretic behavior.
Fig. 9—Thermal conduction across benchmark column: (a) temperature field of unconfined column exposed to fire for 3 hours;
and (b) development of temperature at level of concrete surface.
305 mm (12 in.) and h = 2000 mm (6.5 ft), which were Incremental dynamic analysis
confined with CFRP sheets (tf = 1 mm [0.0394 in.], Ef = For the evaluation of seismic performance through the
70.6 GPa [10,240 ksi], and ffu = 849 MPa [123 ksi]). Irre- foregoing time-history model, incremental dynamic analysis
spective of axial load, the static moment envelopes created (IDA) was carried out.55 The intensity of ground motions was
by the pushover model were in agreement with experi- increased in a piecewise manner until the columns collapsed.
mental responses (Fig. 8(a)). For the hysteretic simulation A spectral acceleration at the fundamental frequency of
of the confined column, the energy constant γ was calibrated the columns was labeled as Sa(T1), which was equivalent
against the measured data (Fig. 8(b)). The converged value to the magnitude of an input motion. This numerical tech-
of γ = 240 was then employed to generate full cyclic curves nique repeatedly updated input motions using the recorded
(Fig. 8(c)). Because others reported that the γ constant was data (all short- and long-duration motions in Table 1), and
an invariable property with respect to elevated tempera- ensuing drift ratios were figured out (the execution algo-
tures,17 γ = 240 was taken for all thermally loaded columns. rithm is delineated in Vamvatsikos and Cornell55). When
the maximum drift ratios of the columns reached the preset
NONLINEAR DYNAMIC ANALYSIS limits from the pushover analysis (Fig. 7(b)), incremental
Time-history analysis iterations were terminated and IDA curves (Sa(T1) versus
The SDOF hysteretic model was expanded to conduct a drift ratio) were drawn.
time-history analysis with the elastic stiffness, fundamental
period, damping ratio, and cyclic degradation detailed previ- RESULTS
ously. Nonlinear solutions were sought using the constant Thermal gradient
average acceleration method, also known as the Newmark Figure 9(a) exhibits a temperature field in the UC column
method,53 and the modified Newton-Raphson iteration.54 exposed to the ASTM E119 standard fire for 3 hours. The
A convergence criterion was chosen at a tolerance limit of distribution of temperature was uneven across the section:
10–5 mm (3.9 × 10–7 in.). Although complete methodological the cover concrete outside the closed tie was thermally loaded
procedures are omitted due to the page limit, one can refer to over 1000°C (1832°F), the longitudinal reinforcing bars were
dynamics texts.48 Predictions included time histories for the subjected to a range of 400 to 550°C (752 to 1022°F), and the
displacement, velocity, and acceleration of the unconfined inner side of the column underwent below 400°C (752°F).
and confined columns. The rise in temperature at the level of the concrete surface
is demonstrated in Fig. 9(b). The incipient stage of the UC
@seismicisolation
@seismicisolation

Fig. 10—Capacity of columns exposed to fire: (a) interaction diagram of UC column; (b) interaction diagram of C6 column;
and (c) capacity degradation.
Table 2—Lateral resistance of unconfined (UC) The load and moment capacities of the UC column less-
and confined (C6) columns ened owing to the fire loadings (Fig. 10(a)). The conspic-
Identification UC C6
uous diminution of the load plateaus, denoting nominal
maximum compression, was ascribed to the decreased
CFRP confinement No Yes
pure axial capacity of the column caused by the thermally
Number of CFRP layers 0 6 degraded constituent properties (Fig. 6). With an increase
Insulation thickness 0 mm 40 mm in the exposure time, cracked-section responses below the
plateaus deviated from the control response at 0 hours; as a
Prefire: 0 hours 647 kN 766 kN
Peak strength Fire duration: 1 hour 574 kN 756 kN consequence, the flexural failure of the column was accel-
(Mn/h) Fire duration: 2 hours 532 kN 743 kN erated when subjected to the same magnitude of the axial
Fire duration: 3 hours 503 kN 720 kN load. Regarding the interaction diagrams of the C6 column
Prefire: 0 hours 2.3% 9.0% (Fig. 10(b)), a modest divergence was noticed in all graphs;
Drift ratio
Fire duration: 1 hour 3.5% 8.6% for example, the 3-hour moment at a transition point from
at failure
(um/h)
Fire duration: 2 hours 4.2% 7.9% the plateau to the bulged portion of the curve was 14.3%
Fire duration: 3 hours 4.4% 7.0% lower than its 0-hour counterpart. The moments of the C6
Fundamental Prefire: 0 hours 0.88 seconds 0.79 seconds column were consistent under the service design load of
period Fire duration: 1 hour 1.07 seconds 0.84 seconds 0.2fc′Ag, regardless of the exposure time; nonetheless, the
(T1 = 2π(P/ Fire duration: 2 hours 1.29 seconds 0.90 seconds
circumferentially wrapped CFRP sheets were not beneficial
(gKe))0.5) Fire duration: 3 hours 1.48 seconds 1.01 seconds
in terms of altering the pure moment capacity of the column
Note: T1 is fundamental period; P is applied axial load; g is gravitational acceleration;
Ke is initial stiffness; Mn is ultimate moment; h is column height; um is maximum
(2172 kN∙m [1602 kip∙ft]) relative to the capacity of the UC
displacement; 1 kN = 0.225 kip; 1 mm = 0.0394 in.; 1 MPa = 145 psi. column (2027 kN∙m [1495 kip∙ft]) without thermal distress.
Figure 10(c) displays the deteriorated moment capacity of
column was rapid (233°C [451°F] at 1 minute), followed by the columns: Mn,uc(th) and Mn,c(th) are the capacities of the
a gradual development to 1063°C (1945°F) at 180 minutes. unconfined and confined columns at the service design load
With the presence of CFRP, the evolutionary trend of the level, respectively. Contrary to the sustained capacity of the
surface temperature was retarded until the decomposition C6 column showing a loss of 6.0% from 0 to 180 minutes,
temperature of 400°C (752°F) was reached. The increased the capacity of the UC column diminished by over 22.3%.
CFRP layers delayed the transfer of the heat, whereas the
extended time to the decomposition was minimal, scilicet, 4 Lateral resistance
and 6 minutes with two and six layers, respectively. When Table 2 enumerates the lateral resistance of the confined
the confined column was insulated, the surface temperature and unconfined columns as a function of the fire-exposure
was maintained below 34°C (93°F) for up to 30 minutes and time. Despite the fact that the progression of high tempera-
the highest temperature was 351°C (664°F) at 180 minutes. tures exacerbated structural deterioration, the C6 column
This observation substantiates the practical significance of outperformed the UC column from the perspective of static
proper insulation for CFRP-strengthened columns in the and dynamic properties. The drops in the peak strength were
event of a fire. 22.3% and 6.0% for the UC and C6 columns, respectively,
from 0 to 3 hours of thermal loadings. On their drift ratios
Capacity degradation in fire at failure, a marked improvement was recorded as high as
Interactions between the axial load and moment of 391.3% (9.0% versus 2.3% at the prefire condition). The
the UC and confined (C6) columns at failure are given in fundamental periods of the UC and C6 columns increased
Fig. 10(a) and (b), respectively, depending upon fire-expo- by 68.2% and 27.8%, respectively, from 0 to 3 hours, which
sure time. Because the contribution of CFRP to the strength is indicative of the superior resistance of the C6 column to
of the uninsulated column was nullified within 6 minutes lateral loading; in other words, the stiffness reduction (ΔKe)
in a fire (Fig. 9(b)), interaction diagrams without insulation of C6 was lower than that of UC.
that do not furnish meaningful acquaintance are not shown.
@seismicisolation
@seismicisolation

Fig. 11—Seismic-fire-combined load effects: (a) column responses with and without thermal loading; (b) IDA curves of UC
column; and (c) IDA curves of C6 column.
Performance under seismic-fire-combined loadings short-duration hysteretic curves (Fig. 12(a)) were inclined
The thermomechanical behavior of the columns, predicted to move along the path of the pushover envelope until the
by the pushover model, is shown in Fig. 11(a). For clarity, seismic excitation attenuated (Fig. 2(c)); by contrast, the
the least favorable situation at 3 hours of fire exposure was long-duration curves suffered a sudden escalation of the
instantiated with prefire responses at 0 hours. The ultimate drift at 2.7% (the third quadrant, Fig. 12(b)) because of
strength and stiffness of the UC column dropped by 22.3% an upsurge in the ground acceleration near 100 seconds in
and 64.6%, respectively, before and after the exposure; in Fig. 2(c) that was linked with the aforementioned hysteretic
contrast, those of the C6 column decreased by 6.0% and damage. Upon initiation of the inelastic behavior, the energy
38.6%, respectively. Likewise, the maximum tolerable dissipation of the column steadily increased (Fig. 12(c)).
drift ratios of the confined column were improved consid- Even if there was a gap between the short- and long-dura-
erably. The outcomes of the IDA model for the unconfined tion scenarios, their growth rates were almost identical up to
and confined columns are compiled in Fig. 11(b) and (c), 255 kN∙m (188 kip∙ft) when the short-duration earthquake
respectively. The exposure-dependent median graphs of was active; contrarily, the amount of energy dissipation
these columns were acquired from the individual responses appertaining to the long-duration earthquake continuously
pertaining to the 50 pairs of the ground motions (Table 1). rose up to 154 seconds and stabilized owing to the decay of
As the exposure time elapsed, the acceleration response of the ground acceleration (Fig. 2(c)). The strength reduction
the UC column abated (Fig. 11(b)): the softened slope of the of the column stemming from the cyclic loading is plotted in
IDA curve at 3 hours signifies the enlarged vulnerability to Fig. 12(d), where a ratio of the peak of each hysteretic curve
lateral loadings, which would elevate the risk of structural (F0,deg) to the ultimate strength of the pushover envelop (F0)
collapse by allowing inordinate sway. The performance of is defined (Fig. 12(d), inset). The short-duration earthquake
the C6 column surpassed that of the UC column (Fig. 11(c)): led to a degradation ratio of 0.90 at 60 seconds. On the other
the initial stiffness (Ke) of the unconfined and confined hand, the long-duration earthquake resulted in a plunge and
columns decreased by 64.2% and 47.4% from 0 to 3 hours, entailed a stable ratio beyond 154 seconds when the input
respectively. The spectral acceleration and drift relationship acceleration ebbed (Fig. 2(c)). It is thus argued that current
of the C6 column tended to be bilinear: the deviation of design practices, without explicitly taking into account earth-
the secondary slope from the initial slope was attributed to quake durations,56 cannot fully embrace the implications of
the accumulated hysteretic damage in the column together seismic hazards, and that an alternative approach should be
with enhanced deformability that extended the breadth of adduced for both confined and unconfined columns.
usable drifts. The secondary slope of the IDA curves was Assessment of combined load effect—Described in Fig. 13
an indication of structural impairment, insinuating that are the median IDA curves of the columns under the vari-
accumulated thermal damage progressively weakened the able earthquake durations coupled with the thermal distress.
confinement system. While the development of spectral acceleration in the UC
column was controlled by the fire exposure, its susceptibility
Dependency of seismic capacity on to the duration of the seismic excitation was virtually none
earthquake duration (Fig. 13(a)). The reason is explained by the low maximum
Dynamic behavior—Figure 12 reveals the response of drifts of the UC column that failed before the occurrence of
the confined column subjected to short- and long-duration excessive yield deformations tied with the cyclic deteriora-
earthquakes at 3 hours of fire exposure (only the C6 column tion (Fig. 12(d)). As for the confined column (Fig. 13(b)),
is covered under the selected acceleration samples given in the influence of the 3-hour fire loading was also prominent
Fig. 2(c) for brevity, and comparative assessments in rela- and the duration effect was inappreciable until the bifurca-
tion to the UC column will follow). The maximum drifts of tion of the IDA curves emerged due to the increased funda-
the column loaded with the short and long durations were mental period associated with the dissipation of hysteretic
3.1% and 6.1%, as shown in Fig. 12(a) and (b), respectively. energy in the C6 column (Table 2). The maximum difference
It should be noted that these drifts are part of the 7.0% between the short- and long-duration-induced spectral accel-
median of the integrated values discussed in Fig. 11(c). The erations at 0 hours of exposure was 18%, whereas the degree
@seismicisolation
@seismicisolation

Fig. 12—Dynamic response of C6 subjected to spectrally equivalent earthquake ground motions at 3-hour fire exposure (EQ.
is earthquake): (a) short duration; (b) long duration; (c) hysteretic energy dissipation; and (d) cyclic strength degradation.
Fig. 13—Median IDA curves as to variable earthquake durations combined with thermal loading (EQ. is earthquake): (a) UC
column; and (b) C6 column.
of disparity was reduced to 11% at 3 hours. This fact corrob- when subjected to the 50 pairs of the spectrally equivalent
orates that the elevated temperatures mitigated the discrep- short and long ground motions under the exposure periods
ancy between the short- and long-duration earthquakes. of 0 and 3 hours. The strength of the columns waned as
Reliance on characteristic duration—Figures 14(a) to (d) the characteristic durations lengthened, which implies that
demonstrate the distribution of cyclically degraded strengths persistent earthquake loadings need to be a consideration for
with DS5-75 for the unconfined and confined columns at failure seismic design. The capacities of the UC and C6 columns,
@seismicisolation
@seismicisolation

Fig. 14—Influence of earthquake duration on strength degradation (log-scale plots): (a) cyclic strength degradation of UC
column at 0 hours; (b) cyclic strength degradation of UC column at 3 hours; (c) cyclic strength degradation of C6 column at
0 hours; (d) cyclic strength degradation of C6 column at 3 hours; (e) seismic capacity of UC column at 0 hours; (f) seismic
capacity of UC column at 3 hours; (g) seismic capacity of C6 column at 0 hours; and (h) seismic capacity of C6 column at
0 hours.
Fig. 15—Seismic capacity degradation with earthquake duration in fire (log-scale plots): (a) UC column; and (b) C6 column.
represented by the peak spectral accelerations, are appraised the earthquake duration (Fig. 15(a)). An opposite trend
in Fig. 14(e) to (h). At the threshold duration of 25 seconds, was, however, noticed for the confined column by virtue of
there was a stepwise decrement in the median capacities of the insulation layer (Fig. 15(b)): the less susceptible vari-
the columns (Fig. 14(e) to (h)) and the importance of the ations of the 1- to 3-hour lines in comparison with that of
durations was once again emphasized. the 0-hour line are ascribed to the hysteretic energy dissipa-
tion explained earlier. For a practical appraisal under seis-
DESIGN RECOMMENDATIONS mic-fire-combined loadings at the preliminary design phase,
Shown in Fig. 15 are the sets of strength degradation the quantity of the strength degradation was equated with a
data belonging to the short- and long-duration earthquakes capacity reduction factor (namely, the nominal capacity of a
consolidated with the fire-exposure time (100 ground column is multiplied by the factor to estimate its deteriorated
motions collated in Table 1 were used for each temporal capacity without going through rigorous dynamic inves-
category). Regression lines were added to definitize a rela- tigations). Table 3 arranges those factors in the context of
tionship between the degraded strength and external attri- column types, earthquake durations, and fire ratings, which
butes (dependent and independent variables, respectively). were rounded for the simplification of the exact degrada-
Aligning with Fig. 13(a), the exposure time was more influ- tion amounts. The applicable boundary of the proposal is an
ential in degenerating the strength of the UC column than
@seismicisolation
@seismicisolation

Table 3—Proposed capacity reduction factors for seismic-fire-combined loading
Fire rating
Column type Earthquake duration *
0 hours 1 hour 2 hours 3 hours
Unconfined without insulation DS5-75 < 25 seconds 0.95 0.85 0.80 0.75
(before strengthening) DS5-75 ≥ 25 seconds 0.90 0.80 0.75 0.70
CFRP-confined with insulation DS5-75 < 25 seconds 0.95 0.95 0.90 0.90
(after strengthening) DS5-75 ≥ 25 seconds 0.90 0.90 0.85 0.85
*
Applicable range: 0 seconds ≤ duration ≤ 85 seconds.
earthquake duration of 0 to 85 seconds in agreement with the established on cyclically accumulated damage plus the
range of the sampled seismic events (Table 1). widened usable drifts that enabled high-level deform-
ability, were a metric to gauge the functionality of the
SUMMARY AND CONCLUSIONS CFRP system.
This paper has discussed analytical procedures to compre- • The intensity of ground motions and earthquake dura-
hend the behavior of reinforced concrete columns with and tions dominated the pattern of conformational loops in
without carbon fiber-reinforced polymer (CFRP) confine- the hysteretic curves. Besides, the seismic-fire-com-
ment in a multi-hazard scenario, comprising earthquake bined loading was a critical factor for dissipating the
and fire loadings. A data set of 100 short- and long-dura- energy of the UC and C6 columns. Upon examining
tion earthquakes was gleaned and partitioned at 25 seconds, the features of the characteristic duration (DS5-75), the
which was amalgamated with a fire-exposure period of 0 to prominence of the persistent earthquake loadings
3 hours. The strengthening scheme involved two categories: was recognized.
1) one to six layers of CFRP sheets; and 2) six CFRP layers • The proposed reduction factors addressed the limita-
with a 40 mm (1.6 in.) thick insulation. Heat transfer was tions of current practice,56 which is reliant on risk-
modeled on a discrete entity platform, and the mechanical targeted maximum considered earthquake responses
response of the columns was predicted by static pushover and (MCER) without allowing for the duration of ground
nonlinear dynamic methods (time-history analysis and incre- motions. The factors can facilitate the design of UC and
mental dynamic analysis [IDA]). After linking the seismic CFRP-confined columns under the combined loadings
approaches with the distribution of temperature from the with an earthquake duration of 0 to 85 seconds and a fire
conduction model, the performance of the unconfined (UC) rating of up to 3 hours.
and confined columns was studied. Technical interests lay
in capacity degradation, load-moment interactions, lateral AUTHOR BIOS
resistance, a relationship between spectral acceleration and Ju-Hyung Kim is a Postdoctoral Fellow in the Department of Civil Engi-
neering at the University of Colorado Denver, Denver, CO. He received his
drift ratio, and hysteretic energy dissipation. Findings were BE, MS, and PhD from the Department of Architecture and Architectural
integrated to suggest performance-based design guidelines. Engineering at Seoul National University, Seoul, South Korea. His research
The following conclusions are drawn: interests include seismic design, performance evaluation, and statistical
analysis of reinforced concrete structures.
• The initial development of temperature was rapid at
the surface of the column concrete (233°C [451°F] Yail J. Kim, FACI, is President of the Bridge Engineering Institute, An
at 1 minute) and, as uneven thermal distributions International Technical Society, and a Professor in the Department of
Civil Engineering at the University of Colorado Denver. He is Chair of
progressed across the section, the growth rate became ACI Subcommittee 440-I, FRP-Prestressed Concrete, past Chair of ACI
stable up to 1063°C (1945°F) at 180 minutes. Whereas Committee 345, Bridge Construction and Preservation, and a member of
the degree of heat transfer was delayed in part by ACI Committees 342, Evaluation of Concrete Bridges and Bridge Elements;
377, Performance-Based Structural Integrity & Resilience of Concrete
wrapping the column with CFRP, insulation was Structures; and 440, Fiber-Reinforced Polymer Reinforcement; and Joint
required to reach a 3-hour fire rating (351°C [664°F] at ACI-ASCE Committee 343, Concrete Bridge Design. He received the
180 minutes). Chester Paul Siess Award for Excellence in Structural Research in 2019.
His research interests include advanced composite materials for rehabilita-
• The thermally degraded properties of the constituent tion; structural informatics; complex systems; and science-based structural
materials were responsible for lowering the pure axial engineering, including statistical, interfacial, and quantum physics.
capacity of the UC column and accelerating its flexural
ACI member Jun Wang is a Postdoctoral Fellow in the Department of
failure. The influence of the fire exposure was marginal Civil Engineering at the University of Colorado Denver. She received her
on the behavior of the confined column with the insu- BS and MS from Northeast Forestry University, Harbin, Heilongjiang,
lation layer (C6), leading to reasonable retention of the China, and her PhD from the University of Colorado Denver. She is a
member of ACI Committee 345, Bridge Construction and Preservation. Her
capacity with a 6.0% loss at the design service load of research interests include multi-object interaction, concrete structures, and
0.2fc′Ag. advanced modeling.
• Through the pushover model, it was construed that the
seismic-fire-combined distress was more detrimental ACKNOWLEDGMENTS
This research is in part supported by the U.S. Department of Transpor-
than the uncoupled actions. The hysteretic analysis tation through the Mountain-Plains Consortium, for which the authors are
confirmed the increased vulnerability of the UC column grateful. Proprietary information is not included to avoid commercialism.
to stability failure when the exposure period was Technical contents presented herein are based on the opinion of the writers
and do not necessarily represent that of others.
enlarged. The bilinear IDA curves of the C6 column,@seismicisolation
@seismicisolation

REFERENCES Concrete Moment-Frame Building,” PEER Report 2007/12, Pacific Earth-
1. Li, Q., and Ellingwood, B. R., “Performance Evaluation and Damage quake Engineering Research Center, Berkeley, CA, 2008, 382 pp.
Assessment of Steel Frame Buildings under Main Shock–Aftershock Earth- 21. Moehle, J., Seismic Design of Reinforced Concrete Buildings,
quake Sequences,” Earthquake Engineering & Structural Dynamics, V. 36, McGraw-Hill Professional, New York, 2014, 760 pp.
No. 3, Mar. 2007, pp. 405-427. doi: 10.1002/eqe.667 22. Williams, B.; Bisby, L.; Kodur, V.; Green, M.; and Chowdhury, E.,
2. Bessason, B.; Rupakhety, R.; and Bjarnason, J. Ö., “Comparison and “Fire Insulation Schemes for FRP-Strengthened Concrete Slabs,” Compos-
Modelling of Building Losses in South Iceland Caused by Different Size ites Part A: Applied Science and Manufacturing, V. 37, No. 8, Aug. 2006,
Earthquakes,” Journal of Building Engineering, V. 46, Apr. 2022, Article pp. 1151-1160. doi: 10.1016/j.compositesa.2005.05.028
No. 103806. doi: 10.1016/j.jobe.2021.103806 23. Lahoti, M.; Tan, K. H.; and Yang, E.-H., “A Critical Review of
3. ASCE/SEI 41-17, “Seismic Evaluation and Retrofit of Existing Build- Geopolymer Properties for Structural Fire-Resistance Applications,”
ings,” American Society of Civil Engineers, Reston, VA, 2017. Construction and Building Materials, V. 221, Oct. 2019, pp. 514-526. doi:
4. Lavorato, D., and Nuti, C., “Pseudo-Dynamic Tests on Reinforced 10.1016/j.conbuildmat.2019.06.076
Concrete Bridges Repaired and Retrofitted after Seismic Damage,” 24. SDI, “Basic Fire Door, Fire Door Frame, Transom/Sidelight Frame,
Engineering Structures, V. 94, July 2015, pp. 96-112. doi: 10.1016/j. and Window Frame Requirements (SDI 118-19),” Steel Door Institute,
engstruct.2015.01.012 Cleveland, OH, 2019, 12 pp.
5. Xu, J. J.; Demartino, C.; Shan, B.; Heo, Y. A.; and Xiao, Y., “Exper- 25. Somerville, P. G.; Smith, N. F.; Graves, R. W.; and Abrahamson,
imental Investigation on Performance of Cantilever CFRP-Wrapped N. A., “Modification of Empirical Strong Ground Motion Attenuation Rela-
Circular RC Columns under Lateral Low-Velocity Impact,” Composite tions to Include the Amplitude and Duration Effects of Rupture Directivity,”
Structures, V. 242, June 2020, Article No. 112143. doi: 10.1016/j. Seismological Research Letters, V. 68, No. 1, Jan.-Feb. 1997, pp. 199-222.
compstruct.2020.112143 doi: 10.1785/gssrl.68.1.199
6. Lignos, D. G.; Moreno, D. M.; and Billington, S. L., “Seismic 26. Foschaar, J. C.; Baker, J. W.; and Deierlein, G. G., “Preliminary
Retrofit of Steel Moment-Resisting Frames with High-Performance Fiber- Assessment of Ground Motion Duration Effects on Structural Collapse,”
Reinforced Concrete Infill Panels: Large-Scale Hybrid Simulation Exper- 15th World Conference on Earthquake Engineering (15WCEE), Lisbon,
iments,” Journal of Structural Engineering, ASCE, V. 140, No. 3, Mar. Portugal, 2012, 10 pp.
2014, p. 04013072. doi: 10.1061/(ASCE)ST.1943-541X.0000877 27. Chandramohan, R.; Baker, J. W.; and Deierlein, G. G., “Quantifying
7. FEMA 547, “Techniques for the Seismic Rehabilitation of Existing the Influence of Ground Motion Duration on Structural Collapse Capacity
Buildings,” Federal Emergency Management Agency, Washington, DC, Using Spectrally Equivalent Records,” Earthquake Spectra, V. 32, No. 2,
2006, 571 pp. May 2016, pp. 927-950. doi: 10.1193/122813eqs298mr2
8. Gkournelos, P. D.; Triantafillou, T. C.; and Bournas, D. A., “Seismic 28. O’Meagher, A. J., and Bennetts, I. D., “Modelling of Concrete
Upgrading of Existing Reinforced Concrete Buildings: A State-of-the-Art Walls in Fire,” Fire Safety Journal, V. 17, No. 4, 1991, pp. 315-335. doi:
Review,” Engineering Structures, V. 240, Aug. 2021, Article No. 112273. 10.1016/0379-7112(91)90026-U
doi: 10.1016/j.engstruct.2021.112273 29. Hajiloo, H.; Green, M. F.; Noël, M.; Bénichou, N.; and
9. Naser, M. Z.; Hawileh, R. A.; and Abdalla, J. A., “Fiber-Reinforced Sultan, M., “Fire Tests on Full-Scale FRP Reinforced Concrete Slabs,”
Polymer Composites in Strengthening Reinforced Concrete Structures: Composite Structures, V. 179, Nov. 2017, pp. 705-719. doi: 10.1016/j.
A Critical Review,” Engineering Structures, V. 198, Nov. 2019, Article compstruct.2017.07.060
No. 109542. doi: 10.1016/j.engstruct.2019.109542 30. Griffis, C. A.; Masumura, R. A.; and Chang, C. I., “Thermal
10. ACI Committee 440, “Guide for the Design and Construction of Response of Graphite Epoxy Composite Subjected to Rapid Heating,”
Externally Bonded FRP Systems for Strengthening Concrete Structures Journal of Composite Materials, V. 15, No. 5, Sept. 1981, pp. 427-442. doi:
(ACI 440.2R-17),” American Concrete Institute, Farmington Hills, MI, 10.1177/002199838101500503
2017, 112 pp. 31. EN 1992-1-2:2004, “Eurocode 2: Design of Concrete Structures -
11. Zou, X. K.; Teng, J. G.; De Lorenzis, L.; and Xia, S. H., “Optimal Part 1-2: General Rules - Structural Fire Design,” European Committee for
Performance-Based Design of FRP Jackets for Seismic Retrofit of Rein- Standardization, Brussels, Belgium, 2004, 99 pp.
forced Concrete Frames,” Composites Part B: Engineering, V. 38, No. 5-6, 32. Chowdhury, E. U.; Bisby, L. A.; Green, M. F.; and Kodur, V. K. R.,
July-Sept. 2007, pp. 584-597. doi: 10.1016/j.compositesb.2006.07.016 “Investigation of Insulated FRP-Wrapped Reinforced Concrete Columns in
12. Ager, P.; Eriksson, K.; Hansen, C. W.; and Lønstrup, L., “How the Fire,” Fire Safety Journal, V. 42, No. 6-7, Sept.-Oct. 2007, pp. 452-460.
1906 San Francisco Earthquake Shaped Economic Activity in the Amer- doi: 10.1016/j.firesaf.2006.10.007
ican West,” Explorations in Economic History, V. 77, July 2020, Article 33. Chapiro, A., “Radiation Effects in Polymers,” Encyclopedia of Mate-
No. 101342. doi: 10.1016/j.eeh.2020.101342 rials: Science and Technology, second edition, K. H. Jürgen Buschow,
13. Wang, Y.; Kodur, V. K. R.; Fu, C.; Liu, C.; Zhou, H.; and Naser, M. Z., M. C. Flemings, E. J. Kramer, P. Veyssière, R. W. Cahn, B. Ilschner, and
“Seismic Performance of Reinforced Concrete Frame Joints after Exposure S. Mahajan, eds., Elsevier, Amsterdam, the Netherlands, 2004, pp. 1-8.
to Fire,” ACI Structural Journal, V. 118, No. 3, May 2021, pp. 3-14. 34. Zhou, G.; Mikinka, E.; Golding, J.; Bao, X.; Sun, W.; and Ashby, A.,
14. Chian, S. C., and Kolathayar, S., “Recent Advances in Earthquake “Investigation of Thermal Degradation and Decomposition of Both Pristine
Engineering—An Introduction,” Recent Advances in Earthquake Engi- and Damaged Carbon/Epoxy Samples with Thermal History,” Composites
neering: Select Proceedings of VCDRR 2021, S. Kolathayar and S. C. Part B: Engineering, V. 201, Nov. 2020, Article No. 108382. doi: 10.1016/j.
Chian, eds., Lecture Notes in Civil Engineering, V. 175, Springer Nature, compositesb.2020.108382
Singapore, 2022, pp. 1-8. 35. Zhou, F.; Zhang, J.; Song, S.; Yang, D.; and Wang, C., “Effect of
15. Mohan, A. T.; Van Coile, R.; Hopkin, D.; Jomaas, G.; and Temperature on Material Properties of Carbon Fiber Reinforced Polymer
Caspeele, R., “Risk Tolerability Limits for Fire Engineering Design: (CFRP) Tendons: Experiments and Model Assessment,” Materials (Basel),
Methodology and Reference Case Study,” Fire Technology, V. 57, No. 5, V. 12, No. 7, Apr. 2019, Article No. 1025. doi: 10.3390/ma12071025
Sept. 2021, pp. 2235-2267. doi: 10.1007/s10694-021-01118-w 36. Lie, T. T., Structural Fire Protection, Manuals and Reports on Engi-
16. Uchiyama, T.; Kawaguchi, K.; and Wakabayashi, T., “Effect of neering Practice No. 78, American Society of Civil Engineers, Reston, VA,
Simultaneous Consideration for Seismically Induced Events on Core 1992.
Damage Frequency,” Journal of Power and Energy Systems, V. 5, No. 3, 37. Bai, Y., and Keller, T., “Time Dependence of Material Properties of
2011, pp. 360-375. doi: 10.1299/jpes.5.360 FRP Composites in Fire,” Journal of Composite Materials, V. 43, No. 21,
17. Wang, J. H.; Zhang, X.; Kunnath, S.; He, J.; and Xiao, Y., “Post- Oct. 2009, pp. 2469-2485. doi: 10.1177/0021998309344641
Earthquake Fire Resistance and Residual Seismic Capacity of Reinforced 38. Luo, Y.; Klima, K. M.; Brouwers, H. J. H.; and Yu, Q., “Effects of
Concrete Columns,” ACI Structural Journal, V. 118, No. 4, July 2021, Ladle Slag on Class F Fly Ash Geopolymer: Reaction Mechanism and High
pp. 123-136. Temperature Behavior,” Cement and Concrete Composites, V. 129, May
18. ACI Committee 318, “Building Code Requirements for Structural 2022, Article No. 104468. doi: 10.1016/j.cemconcomp.2022.104468
Concrete (ACI 318-19) and Commentary (ACI 318R-19) (Reapproved 39. Xing, Z.; Beaucour, A.-L.; Hebert, R.; Noumowe, A.; and
2022),” American Concrete Institute, Farmington Hills, MI, 2019, 624 pp. Ledesert, B., “Aggregate’s Influence on Thermophysical Concrete Proper-
19. ASTM E119-20, “Standard Test Methods for Fire Tests of Building ties at Elevated Temperature,” Construction and Building Materials, V. 95,
Construction and Materials,” ASTM International, West Conshohocken, Oct. 2015, pp. 18-28. doi: 10.1016/j.conbuildmat.2015.07.060
PA, 2020, 36 pp. 40. Rand, W., and Stummer, C., “Agent‐Based Modeling of New Product
20. Haselton, C. B.; Goulet, C. A.; Mitrani-Reiser, J.; Beck, J. L.; Deier- Market Diffusion: An Overview of Strengths and Criticisms,” Annals of
lein, G. G.; Porter, K. A.; Stewart, J. P.; and Taciroglu, E., “An Assessment Operations Research, V. 305, No. 1-2, Oct. 2021, pp. 425-447. doi: 10.1007/
to Benchmark the Seismic Performance of a Code-Conforming Reinforced s10479-021-03944-1
@seismicisolation
@seismicisolation

41. Wilensky, U., and Rand, W., An Introduction to Agent-Based 374.2R-13),” American Concrete Institute, Farmington Hills, MI, 2013,
Modeling: Modeling Natural, Social, and Engineered Complex Systems 18 pp.
with NetLogo, MIT Press, Cambridge, MA, 2015, 504 pp. 50. Ibarra, L. F.; Medina, R. A.; and Krawinkler, H., “Hysteretic Models
42. Weerasinghe, P.; Nguyen, K.; Mendis, P.; and Guerrieri, M., “Large- That Incorporate Strength and Stiffness Deterioration,” Earthquake Engi-
Scale Experiment on the Behaviour of Concrete Flat Slabs Subjected to neering & Structural Dynamics, V. 34, No. 12, Oct. 2005, pp. 1489-1511.
Standard Fire,” Journal of Building Engineering, V. 30, July 2020, Article doi: 10.1002/eqe.495
No. 101255. doi: 10.1016/j.jobe.2020.101255 51. Haselton, C. B.; Liel, A. B.; Lange, S. T.; and Deierlein, G. G.,
43. Bisby, L. A., “Fire Behaviour of Fibre-Reinforced Polymer (FRP) “Beam-Column Element Model Calibrated for Predicting Flexural
Reinforced or Confined Concrete,” PhD thesis, Queen’s University, Response Leading to Global Collapse of RC Frame Buildings,” PEER
Kingston, ON, Canada, 2003, 397 pp. Report 2007/03, Pacific Earthquake Engineering Research Center, Berkeley,
44. Zhou, F.; Pang, R.; Zhang, P.; and Cui, J., “Experimental Investi- CA, 2008, 152 pp.
gation of the Mechanical Properties of Carbon Fiber-Reinforced Polymer 52. Paultre, P.; Boucher-Trudeau, M.; Eid, R.; and Roy, N., “Behavior
(CFRP) Tendons during and after Exposure to Elevated Temperatures,” of Circular Reinforced-Concrete Columns Confined with Carbon Fiber–
Materials and Structures, V. 55, No. 2, Mar. 2022, Article No. 82. doi: Reinforced Polymers under Cyclic Flexure and Constant Axial Load,”
10.1617/s11527-022-01923-x Journal of Composites for Construction, ASCE, V. 20, No. 3, June 2016,
45. Nawy, E. G., Reinforced Concrete: A Fundamental Approach, sixth p. 04015065. doi: 10.1061/(ASCE)CC.1943-5614.0000624
edition, Pearson, Hoboken, NJ, 2009. 53. Werkle, H., Finite Elements in Structural Analysis: Theoretical
46. Paulay, T., and Priestley, M. J. N., Seismic Design of Reinforced Concepts and Modeling Procedures in Statics and Dynamics of Structures,
Concrete and Masonry Buildings, John Wiley & Sons, Inc., New York, Springer Nature, Cham, Switzerland, 2021.
1992, 768 pp. 54. Leondes, C. T., Structural Dynamic Systems Computational Tech-
47. Youssf, O.; ElGawady, M. A.; and Mills, J. E., “Displacement and niques and Optimization: Finite Element Analysis (FEA) Techniques, CRC
Plastic Hinge Length of FRP-Confined Circular Reinforced Concrete Press, Boca Raton, FL, 2021.
Columns,” Engineering Structures, V. 101, Oct. 2015, pp. 465-476. doi: 55. Vamvatsikos, D., and Cornell, C. A., “Incremental Dynamic Analysis,”
10.1016/j.engstruct.2015.07.026 Earthquake Engineering & Structural Dynamics, V. 31, No. 3, Mar. 2002,
48. Chopra, A. K., Dynamics of Structures: Theory and Applications to pp. 491-514. doi: 10.1002/eqe.141
Earthquake Engineering, sixth edition, Pearson, London, UK, 2016. 56. ASCE/SEI 7-22, “Minimum Design Loads and Associated Criteria
49. ACI Committee 374, “Guide for Testing Reinforced Concrete Struc- for Buildings and Other Structures,” American Society of Civil Engineers,
tural Elements under Slowly Applied Simulated Seismic Loads (ACI Reston, VA, 2022.
@seismicisolation
@seismicisolation

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.
@seismicisolation
@seismicisolation

ACI Structural Journal Sept. - 2023 V. 120 No. 5

Uploaded by

Copyright:

Available Formats

ACI Structural Journal Sept. - 2023 V. 120 No. 5

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

ACI Structural Journal Sept. - 2023 V. 120 No. 5

Uploaded by

Copyright:

Available Formats

V. 120, NO.

A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE

Staff 121 Behavior of Concrete Bridge-Deck Slabs Reinforced with Basalt

Editors Contents continued on next page

ACI Structural Journal/September 20231

26-29—Greenbuild 2023: Building for the Built Environment, Washington,

2 ACI Structural Journal/September 2023

Ultra-High-Toughness Concrete Retrofitted Boundary

ACI Structural Journal/September 2023 3

4 ACI Structural Journal/September 2023

Fig. 3—Tensile and compressive properties of plain concrete and UHTCC.

ACI Structural Journal/September 2023 5

Fig. 4—Half-grouted steel sleeve connection. @seismicisolation

6 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 7

Fig. 7—Hysteresis curves and evolving crack pattern.

8 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 9

Fig. 9—Fractal-index development.

10 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 11

Fig. 12—Curvature and plastic hinge length, compared with model.

12 ACI Structural Journal/September 2023

lp = 1.25(0.2h + 0.044H) (4) ​

ACI Structural Journal/September 2023 13

14 ACI Structural Journal/September 2023

Table 7—Calculated flexural resistance and lateral bearing force

ACI Structural Journal/September 2023 15

16 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 17

18 ACI Structural Journal/September 2023

Hysteretic Model of Coupler Box Assembly for Seismic

ACI Structural Journal/September 2023 19

20 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 21

22 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 23

Fig. 5—Details and experimental test setup for RC columns.

24 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 25

26 ACI Structural Journal/September 2023

Fig. 8—Reinforcement detailing in tested RC model frames.

ACI Structural Journal/September 2023 27

28 ACI Structural Journal/September 2023

Fig. 12—Schematic diagram depicting procedure for reestablishing damaged column.

ACI Structural Journal/September 2023 29

​δ​m​− ​δ​y​ _ ​∫E=​ ​ ​dE​

30 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 31

Fig. 17—Energy-based parametric comparison in tested frames.

Fig. 18—Comparative plots of strength and stiffness degra-

32 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 33

34 ACI Structural Journal/September 2023

ACI Structural Journal/September 2023 35

Seismic Performance of Five-Spiral Concrete Columns

ACI Structural Journal/September 2023 37

Note: 1 MPa = 145.04 psi; 1 mm = 0.0394 in.

Murcia-Delso et al.18 tested two full-scale columns RESEARCH SIGNIFICANCE

Staff 121 Behavior of Concrete Bridge-Deck Slabs Reinforced with Basalt

ACI Structural Journal/September 20231

2 ACI Structural Journal/September 2023

lp = 1.25(0.2h + 0.044H) (4)

δm− δy _ ∫E= dE