You are on page 1of 16

Exploratory Study on Incorporating Glass FRP Reinforcement

to Control Damage in Steel-Reinforced Concrete Bridge


Pier Walls
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

Ahmed Arafa, M.ASCE1; Nourhan Ahmed2; Ahmed Sabry Farghaly3; Omar Chaallal4; and Brahim Benmokrane5

Abstract: The need to demonstrate that a steel-reinforced concrete bridge pier wall resilient to strong earthquakes could be attained by the
incorporation of glass fiber–reinforced polymer (GFRP) reinforcement has been brought to the fore by recent experimental results on GFRP-
reinforced concrete bridge pier walls. The test results show that the GFRP bars assisted in crack recovery and the self-centering of walls
between load reversals. Hence, GFRP bars could potentially be used to control the unrecoverable damage in steel-reinforced bridge pier
walls after an earthquake. This study will use nonlinear finite element analysis (FEA) as a powerful tool to verify this expectation. A series
of analyses will be implemented on concrete bridge pier walls reinforced with either steel or GFRP bars to demonstrate that the finite element
(FE) procedure can provide quick and reliable simulation. The study is then extended to investigate the effect of using hybrid reinforcement
through a comprehensive parametric study. Different configurations of GFRP bars are examined and compared with similar configurations of
steel bars. The results show that hybrid reinforced bridge pier walls can undergo large displacements with minimal residual deformations.
Nevertheless, a sensible selection of the GFRP bars location is necessary. The findings of this study could be considered as a fundamental
step toward the development of code provisions for the use of hybrid GFRP/steel (GS) reinforcement in concrete bridge pier walls.
DOI: 10.1061/(ASCE)BE.1943-5592.0001648. © 2020 American Society of Civil Engineers.
Author keywords: Steel/GFRP bars; Concrete; Bridge pier walls; Seismic; Residual deformations; Damage control; Finite-element
analysis.

Introduction et al. 2009; Takahashi et al. 2013; Luna et al. 2015) revealed that
the behavior is dominated by inelastic shear deformations, which
Many bridges around the world depend on bridge pier walls as the then rapidly degrade the strength and stiffness and cause substan-
primary resistance component for earthquake excitations due to tial uncontrollable damage. This situation requires service disrup-
their in-plane lateral stiffness. These walls are classified as tion, lengthy repair time, or, in some cases, complete replacement.
squat walls with a height-to-length ratio (aspect ratio) typically The experimental findings are supported by previous earthquake
<2.0 [AASHTO 2018 and CSA S6-19 (CSA 2014b)]. Although reconnaissance reports. For example, many bridge pier walls
current seismic design practice promotes energy dissipation were demolished in Taiwan after the 1999 Chi-Chi earthquake
through the ductile mechanism after the yielding of flexural rein- due to the excessive damage and the subsequent high repair
forcement, bridge pier walls tend to fail in diagonal shear, or in costs (Moehle 1999).
sliding, or both due to their low aspect ratios (Fig. 1). Experimen- In the last decade, noncorrodible fiber–reinforced polymer
tal observations (Saatcioglu 1991; Sittipunt et al. 2001; Massone (FRP) reinforcing bars have increasingly been used as the main re-
inforcement for concrete structures in harsh environments (ACI
1
Lecturer, Dept. of Civil Engineering, Faculty of Engineering, Sohag 440.1R 2015). The widespread use of bridge pier walls and their
University, Sohag 82749, Egypt; Univ. of Sherbrooke, Quebec, Canada. exposure to deterioration induced by steel corrosion meant that test-
Email: ahmed_arafa@eng.sohag.edu.eg; ahmed.arafa@usherbrooke.ca ing the applicability of the use of FRP in these structural elements
2
Master’s Student, Dept. of Civil Engineering, Faculty of Engineering, was necessary. This prompted an experimental investigation at
Sohag Univ., Sohag 82749, Egypt. Email: na8072080@gmail.com
3 the Sherbrooke, Quebec, Canada (Arafa et al. 2018a, b, c, d).
Research Associate, Dept. of Civil Engineering, Univ. of Sherbrooke,
Sherbrooke, Quebec, Canada J1K 2R1. Email: ahmed.farghaly@
Nine large-scale squat walls with a height-to-length ratio of 1.33
usherbrooke.ca were constructed and tested under reversed cyclic loading to
4
Professor of Structural Engineering at ÉTS, Dept. of Construction En- simulate loads caused by earthquakes. One wall was reinforced
gineering, Montreal, Quebec, Canada H3C 1K3. Email: omar.chaallal@ with steel and served as a reference specimen, and the others
etsmtl.ca were reinforced with glass fiber–reinforced polymer (GFRP) bars
5
Professor of Civil Engineering, Tier-1 Canada Research Chair in Ad- in different configurations. The reported test results showed that
vanced Composite Materials for Civil Structures and NSERC Research the tested GFRP-reinforced concrete (RC) walls had satisfactory
Chair in Innovative FRP Reinforcement for Concrete Structures, Dept. of strength and stable cyclic behavior. Due to the elastic nature of
Civil Engineering, Univ. of Sherbrooke, Sherbrooke, Quebec, Canada GFRP bars, the cracks tended to realign and close between load re-
J1K 2R1 (corresponding author). Email: brahim.benmokrane@
versals. The walls showed recoverable, self-centering behavior up
usherbrooke.ca
Note. This manuscript was submitted on February 10, 2020; approved to the allowable drift limits. In contrast, the steel-reinforced wall
on July 23, 2020; published online on November 20, 2020. Discussion pe- exhibited excessive sliding damage at low drift ratios. These results
riod open until April 20, 2021; separate discussions must be submitted for indicated that the GFRP-reinforced walls could potentially be re-
individual papers. This paper is part of the Journal of Bridge Engineering, stored after an earthquake event, and the steel-reinforced wall
© ASCE, ISSN 1084-0702. would face technical difficulties in repairing. This supports the

© ASCE 04020116-1 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Failure along
the main
diagonal
crack
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

(a)

Concrete struts
crushing

Concrete
crushing

(b)

Sliding along
horizontal
plane

(c)

Fig. 1. Schematic details of failure modes in squat walls: (a) diagonal tension failure; (b) diagonal compression failure; and (c) sliding failure.

findings of previous reports. In addition, the results strongly sug- behavior of hybrid reinforced walls (Ghazizadeh et al. 2018).
gest that combining GFRP and steel reinforcement in bridge pier Moreover, to the best of the authors’ knowledge, this study is the
walls could be a viable method to control damage in regions first of its kind that investigates whether the characteristics of
prone to strong earthquakes. This could reduce or even eliminate GFRP bars could be harnessed to control sliding shear and the con-
the lengthy structural repair process after a strong earthquake sequent incalculable damage and provide a cost-effective solution
event. In this study, finite element analysis (FEA) is used as a tool for corrosion problems.
to verify these expectations. A comparison of the FEA results with
measured experimental data that included bridge pier walls rein-
forced with either GFRP or steel bars is first conducted to demon- Description of the Test Specimens
strate the ability of FEA to provide a reasonable simulation of the
behavior. A comprehensive parametric study is then implemented The available literature (Ibrahim et al. 2016; Bencardino et al.
to investigate the effect of using hybrid reinforcement [GFRP/steel 2016; Qin et al. 2017) establishes that the finite element model
(GS)] in bridge pier walls response. The findings of this study are ex- (FEM) developed for structural elements reinforced solely with
pected to support the work of the North American technical commit- steel or FRP bars accurately predicted the behavior of hybrid re-
tees engaged in the development of standards and design provisions inforced elements. Considering these findings and the lack of the
for the use of hybrid steel/FRP reinforcement in concrete bridge pier experimental tests on hybrid reinforced walls, the FEM in this
wall elements in the future edition of the AASHTO LRFD Bridge De- study was built and validated based on results of squat walls re-
sign Guide Specifications for GFRP Reinforced Concrete inforced with either steel or GFRP bars that were previously
(AASHTO, 2018) and the Canadian Highway Bridge Design tested by the authors (Arafa et al. 2018a). The specimens were
Code (CSA S6-19, CSA 2014b). representative of bridge pier walls, since both AASHTO (2018)
It is noted that many studies on bridge columns (Iemura et al. and CSA S6-19 (2014b) require that bridge pier walls are de-
2006; ElGawady and Sha’lan 2011; Wu et al. 2009; Saiidi et al. signed as squat shear walls in the strong direction and as columns
2009; Fahmy and Wu 2016) have investigated the concept of pro- in the out-of-plane direction. The specimens were 1,500 mm long
viding a primary steel-reinforced structural system with a more 2,000 mm heigh, and 200 mm thick. Fig. 2 shows the typical con-
flexible secondary system. The purpose of the latter was to exhibit crete dimensions, reinforcement details, test setup, and instrumenta-
elastic deformations when the main system underwent inelastic de- tion. The specimen designations are S4-80 and G4-80 for steel- and
formations. Nevertheless, there is a lack of knowledge about the GFRP-reinforced walls, respectively. Table 1 lists the mechanical

© ASCE 04020116-2 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

(c)
(d)

Fig. 2. Details of the test specimens: (a) concrete dimensions; (b) reinforcement details; (c) test setup; and (d) LVDTs instrumentations.

Table 1. Tensile properties of reinforcement


Designated bar Nominal areaa Tensile modulus of Tensile Average strain at
Bar diameter (mm) (mm2) elasticity (GPa) strengthb,c (MPa) ultimate (%)
Straight bars
#3 GFRP 9.5 71 65 1,372 2.1
#3 steel 9.5 71 200 fy = 420 ɛy = 0.2
#4 steel 12.7 129 200 fy = 420 ɛy = 0.2
Bent #3 GFRP (rectilinear spiral)
Straight 9.5 71 50 1,065 2.1
Bent — 460 —
Bent #4 GFRP (horizontal bar)
Straight 12.7 129 50 1,020 2.0
Bent — 459 —
Bent #6 GFRP (horizontal bar)
Straight 19.1 285 50 1,028 2.0
Bent — 463 —
Source: Data from Arafa et al. (2018a).
Note: GFRP = glass fiber–reinforced polymer; 1 mm = 0.0394 in.; 1 MPa = 145 psi; fy = steel yielding strength; and ɛy = steel yielding strain.
a
According to CSA S807 (CSA 2010).
b
Tensile properties were calculated using nominal cross-sectional areas.
c
Guaranteed tensile strength: average value 3 × standard deviation (ACI 440.1R-15).

Table 2. Reinforcement ratios and measured concrete strengths properties of the steel and GFRP bars. Table 2 lists the reinforcement
ratios and the measured concrete compressive strengths. The prelim-
Reinforcement ratio inary design and elaboration of details were according to CSA A23.3
′ (2014a) and CSA S806 (2012) for steel- and GFRP-reinforced walls,
Wall f (MPa)
c ρl (%) ρt (%) ρv (%) ρh (%) ρd (%) ρs (%)
respectively. Since CSA S806 (2012) contains no seismic provi-
S4-80 35 1.43 0.89 0.59 1.58 — 0.48
G4-80 40 1.43 0.89 0.59 1.58 — 0.48 sions, similar methodologies in the companion code [CSA A23.3
(2014a)] were followed, which accounted for the difference in me-
Source: Data from Arafa et al. (2018a). chanical characteristics between steel and GFRP bars. The two
Note: f ′c = concrete compressive strength; ρl = boundary longitudinal bar
reinforcement ratio; ρt = boundary tie reinforcement ratio; ρv = web
specimens were reinforced with the same reinforcement ratio and
vertical bar reinforcement ratio; ρh = horizontal web reinforcement ratio; configuration. Two boundary elements of eight #3 (steel or GFRP)
ρd = bidiagonal web reinforcement ratio; and ρs = bidiagonal sliding– bars of equal length and width (200 × 200 mm) were placed at
shear reinforcement ratio. each end of the wall length. The longitudinal reinforcement in the

© ASCE 04020116-3 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


boundaries was laterally tied against premature buckling with #3 observations indicated that the initial propagated cracks were flex-
transverse reinforcement spiral ties (steel or GFRP) spaced at ural and associated with a reduction in the lateral stiffness
80 mm along the total wall height. Two layers of horizontal and ver- [Fig. 3(a)]. Due to the lower modulus of GFRP bars compared
tical web reinforcement were used. The horizontal web reinforce- with steel bars, specimen S4-80 exhibited stiffer behavior than
ment consisted of #4 bars (steel or GFRP) spaced at 80 mm, and its reference specimen G4-80 [Fig. 3(b)]. Under increased loading,
the vertical web reinforcement consisted of #3 bars (steel or horizontal cracks continued to form up to a height of approxi-
GFRP) spaced at 120 mm. Given the lower strength and stiffness mately one-third the wall height and gained inclination in the cen-
of GFRP bars in the transverse direction (ACI 440.1R-15), there tral zone of the web due to shear stresses. As loading continued,
was some concern that the failure of G4-80 might be preceded by new shear cracks propagated near the top of the wall. Then, verti-
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

sliding. Therefore, one additional layer of bidiagonal sliding–shear cal splitting cracks quickly initiated at the boundary element under
reinforcement was added across the potential sliding plane and suf- compression, followed by progressive concrete cover spalling.
ficiently anchored onto each side of the shear plane. The specimens Following this stage, due to the yielding of the steel bars under ten-
were tested under reversed cyclic loading without axial load, since sion and compression, a major horizontal crack developed along
the aim was to investigate the sliding resistance provided by the wall length and above the sliding reinforcement [Fig. 3(a)].
GFRP bars and conservatively neglecting the effect of axial load. This crack remained open even under compressive stresses with
The lateral load was applied at 2,550 mm above the concrete base load reversals. Therefore, the load was primarily transferred
through a rigid steel beam. Lateral displacement reversal was applied along this crack by longitudinal reinforcement dowel action,
under displacement control, starting with two cycles at 2 mm of lat- since the friction–resistance forces maintained by aggregate inter-
eral displacement, then two cycles at each displacement level in in- lock deteriorated under reversed cyclic loading. Therefore, this
crements of 2 mm up to a lateral displacement level of 10 mm. This was associated with localized sliding deformation along this
was followed by increments of 5 mm up to 50 mm of lateral dis- zone, accompanied by concrete deterioration and progressive deg-
placement and then increments of 10 mm up to failure. A series of radation in lateral strength [Fig. 3(b)]. Ultimately, a sudden de-
linear variable differential transducers (LVDTs) and strain gauges crease in lateral strength occurred due to excessive buckling of
were used to measure critical response quantities. the bars [Fig. 3(b)]. In contrast, replacing steel with GFRP bars,
which have no yield point, resulted in crack realignment and clo-
sure in G4-80 between load reversals and distributed deformations
Summary of Test Results along the wall height, thereby hindering the continuous horizontal
path that was responsible for preventing sliding from appearing.
Fig. 3 shows the pattern of the observed cracks for the tested walls Therefore, the specimen continued to carry load without strength
and plots the load–drift ratio envelope curves. In general, test degradation to achieve ultimate strength and drift capacity that
deterioration above
Concrete

Longitudinal bar
Concrete
buckling
crushing

S4-80 G4-80
(a)
Lateral drift (%)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
1000
G4-80
800 S4-80 Cover spalling
Lateral load (kN)

Initial Concrete
600 yielding deterioration
Concrete
400 crushing
Cover splitting

200 First initial Excessive bars buckling


crack
0
0 10 20 30 40 50 60
(b) Lateral displacement (mm)

Fig. 3. Test results: (a) cracks pattern and failure mode; and (b) load–drift ratio hysteretic response. (Data from Arafa et al. 2018a).

© ASCE 04020116-4 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


were 71% and 50% higher, respectively than in its counterpart
steel-reinforced specimen [Fig. 3(b)]. Specimen G4-80 exhibited
flexural compression failure [Fig. 3(a)]; more information can be
found in the literature (Arafa et al. 2018a).

FEA Numerical Model and Validation

Numerical analyses were implemented in this study with VecTor2


Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

specialized finite element (FE) software (Wong and Vecchio 2002).


The program is a two-dimensional FE simulation tool based on a se-
cant–stiffness formulation using a total load, iterative procedure,
which employs constitutive relations for concrete and reinforcement
based on the modified compression field theory (MCFT) (Vecchio
and Collins 1993). The MCFT is a powerful analytical model that
predicts the load–deformation response of RC elements that are sub-
jected to in-plane shear and normal stresses. The theory is based on a
smeared crack approach in which the cracked concrete is treated as a
new material with unique stress–strain characteristics. Equilibrium,
compatibility, and material stress–strain relationships are considered
to represent average stresses and average strains, in addition to
local stresses at crack positions. More details about the program
can be found in the user manual (Wong and Vecchio 2002).

Model Geometry and Meshing


Four-node quadrilateral elements were used to model the concrete
Reinf. 1: 3#3 straight GFRP bars.
and the longitudinal and horizontal bars were simulated with truss Reinf. 2: 2#3 straight GFRP bars.
elements. For simplicity, the diagonal sliding reinforcement was Reinf. 3: 2#3 @80 GFRP ties.
modeled as smeared reinforcement. A representative FE mesh Reinf. 4: 2#4 @80 mm bent GFRP bars.
and set of truss elements was used in the analysis (Fig. 4). To elim- Conc. 1: plain concrete (specimen).
inate the localization effect, the elements’ aspect ratios were <1.5, Conc. 2: reinforced concrete with diagonal sliding reinf.
as recommended by Palermo and Vecchio (2007). This resulted in a Rigid Material
total of 32 and 48 elements in the shorter and longer directions of
Fig. 4. Typical FE meshing.
the simulated walls, respectively. The convergence criteria were
examined based on iterative simulations of different meshes with
different numbers of wall elements. The hysteretic response, failure
Gulec and Whittaker 2009; Vecchio and McQuade 2011; Mo-
mode, and strain readings were examined for the simulated squat
hamed et al. 2014; Ghazizadeh et al. 2018). The hysteric response
walls with different mesh sizes and compared with the experimen-
of concrete was modeled based on Palermo and Vecchio (2002).
tal results of the tested squat walls. The mesh size was then chosen
This model was adopted in this study, since it explicitly considers
that had no additional effect of refined mesh on the numerical re-
concrete damage caused by shear deformations. In addition, the
sults, which considered no localization effect.
model was formulated to provide substantial compatibility with
nonlinear FEA in the context of smeared rotating cracks in both
Material Modeling the compression and tension stress regimes. The concrete confine-
VecTor2 provides useful features when modeling RC elements: ment provided by the closed stirrups at the wall boundaries was
(1) a comprehensive nonlinear constitutive model of concrete, considered using Kupfer model (1969). The model was established
(2) implementation of the effect of cracks on the behavior of RC, for steel-reinforced elements. To make this model suitable for ele-
(3) inclusion of the effect of tension stiffening and compressive ments reinforced with GFRP bars and stirrups, which have lower
strength reduction due to cracks, and (4) implementation of a user- moduli of elasticity than steel reinforcement, an equivalent area
defined bond–stress relationship between reinforcement and con- of GFRP stirrups was used as recommended by Ghomi and El-
crete. The following section discusses the constitutive models for Salakawy (2018). The equivalent area was defined as the area of
the concrete and the steel and GFRP bars. a steel bar that produced the same stiffness as the actual GFRP bars.
Compression softening was simulated using the Vecchio model
(Vecchio and Collins 1993), which considers the reduction in com-
Concrete pressive strength and stiffness caused by coexisting transverse
Table 3 and Fig. 5 illustrate the constitutive models used in model- cracking and tensile straining. This model was adopted because it
ing concrete. The Hognestad parabola (Hognestad 1951) and a considers the softening of both strength and strain. The modified
modified Kent-Park formulation (Scott et al. 1982), which were in- Bentz model (Bentz 1999) was used to model the tension stiffening
tended primarily for the analysis of RC structures subjected to cy- effects. This model accounts for the tensile stresses in concrete be-
clic and dynamic loads, were used to model the pre and postpeak tween cracks induced by the bond action between the reinforce-
response of the concrete. These models were broadly used in sev- ment and concrete. The model was formulated to account for the
eral investigations to model structural walls and showed an accu- bond characteristics of the reinforcement, and therefore, the influ-
rate simulation of the behavior (Palermo and Vecchio 2007; ence extends into surrounding elements that do not contain any

© ASCE 04020116-5 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Table 3. Constitutive models used in the FE analysis
Models Parameter Models used in FE analysis
Concrete Hysteretic response Palermo and Vecchio with decay (Palermo and Vecchio 2007) [Fig. 6(a)]
Concrete compressive pre and postpeak response Hognestad model (Hognestad 1951) and Modified Park-Kent (Scott et al. 1982) [Fig. 6(b)]
Slip distortion Vecchio-Lai model (Lai 2001)
Tension stiffening Bentz model (Bentz 1999)
Confinement strength Kupfer-Richart model (Kupfer et al. 1969)
Dilation Variable-Kupfer (Kupfer et al. 1969)
Cracking criterion Mohr–Coulomb (stress)
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

Steel bars Hysteretic response Seckin (1981) [Fig. 6(c)]


Dowel action Tassios model (He and Kwan 2001)
FRP bars Hysteretic response Fig. 6(d)
Dowel action Omitted as proposed by ACI 440.1R (2015)
Note: FE = finite element; FRP = fiber–reinforced polymer.

-fc
-fc Base curve for compression
softened response fc
fp

εp -εc
p -εp
c -εc Pre-peak model
fc -fc
fcr fp
Base curve for tension stiffened
response
Edes
1
0.2fc

εcr p εc εp -εc
c
Post-peak model
(a) (b)

fs ffrp
Esh ffrp u
fy 1

Es Efrp
1 1
εs εfrp

ffrp comp.= 0.5 ffrpu

(c) (d)

Fig. 5. The used constitutive models: (a) concrete hysteretic response (data from Palermo and Vecchio 2007); (b) concrete pre and postpeak response
(data from Hognestad 1951 and Scott et al. 1982); (c) steel reinforcement hysteretic (data from Seckin 1981); and (d) FRP reinforcement hysteretic
response.

reinforcement. Tension softening was modeled with a linear descend- Reinforcement Material
ing branch after cracking. The dilation of concrete, which represents
Table 3 and Fig. 5 illustrate the constitutive models used to model
the lateral expansion of concrete due to internal microcracking and the steel and GFRP bars. The hysteretic behavior of the steel bars
increases as the compressive stresses increase, was modeled with was modeled according to Seckin (1981). This model was selected
the variable Poisson’s ratio that was proposed by Kupfer and Gerstle because it considers the Bauschinger effect beyond the yielding of
(1973). The constitutive models used to simulate the tension stiffen- steel and the associated softening of both strength and strain.
ing effects, the tension softening, and the dilation of concrete were However, the stress–strain curve was set to linear for the GFRP
adopted based on the previous numerical simulations for steel- bars with a modulus of elasticity equal to the elastic modulus of
reinforced walls (Palermo and Vecchio 2007) and GFRP-reinforced the GFRP reinforcement. The dowel action of the steel bars was
midrise shear walls (Mohamed et al. 2014). represented with the Tassios model (He and Kwan 2001) and

© ASCE 04020116-6 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


the dowel action was neglected for the GFRP bars, according to level of accuracy and could be extended to include the effect of
ACI 440.1R (2015). using hybrid GS reinforcement.
VecTor2 provides the option to define the bond–slip relationship
between reinforcing bars and the surrounding concrete. Based on
the findings of an experimental investigation by Vint (2012) on
Combining Steel and GFRP Reinforcement
the bond–slip behavior between sand-coated GFRP bars and con-
in Concrete Bridge Pier Walls
crete, the modified Bertero-Eligehausen model (Eligehausen et al.
1983) was adopted to simulate the bond between GFRP bars and
Five configurations of hybrid GS reinforcement were investigated.
concrete. A complete bond was used between steel bars and con-
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

Two main issues were considered when selecting the GS reinforce-


crete, which has been reported as being reliable when modeling
ment configurations. The first was to attain a more durable system
steel-reinforced walls (Palermo and Vecchio 2007).
in regions exposed to a harsh environment (i.e., accommodating
corrosion-induced problems). Here, the horizontal web reinforce-
Boundary Conditions ment and the confinement at the boundary elements were the clos-
est to the outer concrete surface. Consequently, they were more
Nodes at the base of the simulated walls were fixed against move- susceptible to severe environmental conditions and the associated
ment in both the horizontal and vertical directions. The lateral dis- deterioration that reduced the service life of the structure. Hence,
placement was applied at the top of the rigid steel beams (550 mm replacing the conventional steel horizontal web reinforcement
high), which was divided into 11 mesh elements in the vertical di- and confinement at boundary elements with GFRP was selected
rection and maintained the same horizontal discretization used in in all the tested configurations. The second issue was to avoid rup-
the wall. ture of the GFRP longitudinal bars before the bridge pier walls de-
veloped adequate ductile behavior and energy dissipation (i.e., the
longitudinal steel bars should exhibit strain that was much higher
Verification of the FE Model
than that at yielding before failure). This concern was raised be-
Fig. 6(a) shows the experimentally obtained cracks pattern and fail- cause the ultimate strain of GFRP bars is much lower than steel.
ure modes with their counterparts that were obtained analytically. Therefore, to reduce the probability of a wall failure before achiev-
The FEA reasonably predicted the pattern of the experimental ing a high level of energy dissipation, all the added GFRP bars were
cracks in terms of cracks initiation, intensity, and inclination. The selected in the web zone away from the extremely tensioned fiber.
results demonstrated that the FEA analyses predicted the experi- The selected reinforcement configurations are shown in Fig. 7.
mentally observed failure modes: excessive sliding in S4-80, and Configuration 1 was identical to that used in S4-80, except for the
concrete crushing at the toe under compression in G4-80. Discrep- horizontal web reinforcement, the confinement at the boundary ele-
ancies were subtle and represented in the underestimation of the ments, and the bidiagonal sliding reinforcement, which were re-
area of the crushed zone in G4-80 (Arafa et al. 2019). placed with GFRP bars. To prevent any rupture of the bidiagonal
In addition to the importance of properly predicting crack patterns GFRP bars, the reinforcement was concentrated in the web zone
and failure modes, the hysteretic response represents a major concern and maintained the same amount used in S4-80. To achieve this,
for future parametric studies. Fig. 6(b) shows the analytical and ex- the spacing was reduced to 90 mm; therefore, the reinforcement
perimental load–drift ratio hysteric response. The figure reveals the ratio changed from 0.35% to 0.4%. In configuration 2, one layer
ability of the FEA model to predict the main features of the experi- of bidiagonal web reinforcement was used with a reinforcement
mental hysteric response in terms of strength, stiffness, and deforma- ratio of 0.4% (the same as the sliding reinforcement ratio in config-
tion capacities. As evidenced experimentally, the FEA model uration 1). In the third configuration, instead of using bidiagonal
showed that S4-80 exhibited widening in the hysteretic loops beyond bars, one layer of GFRP vertical web reinforcement was used with
yielding of the flexural reinforcement. In addition, the model suc- a reinforcement ratio of 0.4%. Configurations 4 and 5 served as refer-
cessfully simulated the initial softening in the hysteretic loops, fol- ences to show whether the damage could be reduced (or eliminated)
lowed by gradual stiffening that appeared after a few cycles of the by using added steel reinforcement in arrangements similar to that
yielding point (induced by the effect of sliding–shear deformations). used in the hybrid configurations. Configurations 4 and 5 were,
The model accurately predicted the hysteresis response of G4-80, therefore, identical to configurations 2 and 3, respectively, except
which was characterized by narrow loops with minimal residual de- for the added reinforcement was steel instead of GFRP.
formation at zero loading due to the elastic nature of GFRP bars. All specimens were simulated under lateral reversed cyclic load-
Table 4 lists the experimentally and analytically predicted ultimate ing in the absence of axial loading, because the aim was to study the
strengths and drift ratios. The ratios of the experimentally obtained damage controlled solely by the proposed reinforcement configura-
ultimate strengths and drift ratios to their predicted analytical values tions while conservatively omitting the effect of axial load. Never-
generally reflect that the model yielded accurate estimations since the theless, this parameter was tested after selection of the
average difference was within 10%. reinforcement configuration that yielded the best performance.
Overall, the established model simulated the response of the The following subsections present the results of the simulated
steel-reinforced and GFRP-reinforced walls with an acceptable GS walls that consider cracks pattern, failure mode, drift ratio,

Table 4. Verification of the accuracy of the FEA results


Experimental FEA
P fpred Pupred dupred
Wall ID Pfexp (kN) Puexp (kN) duexp (%) Pfpred (kN) Pupred (kN) dupred (%) P f exp Pu exp du exp
S4-80 160 534 2.0 158 485 2.20 0.98 0.91 1.1
G4-80 164 912 3.0 153 842 2.75 0.93 0.92 0.92
Note: FEA = finite element method; Pfexp = experimental first flexural cracking load; Puexp = experimental ultimate capacity; duexp = experimental ultimate
drift ratio; Pfpred = predicted first flexural cracking load; Puexp = predicted ultimate capacity; and duexp = predicted ultimate drift ratio.

© ASCE 04020116-7 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


ultimate strength, and load–drift ratio hysteretic response. The im- specimen S4-80. For simplicity, the simulated GS specimens
pact of the reinforcement configurations on seismic performance were identified as GS followed by the configuration number.
indices in terms of residual deformations, stiffness, ductility, and
energy dissipation is introduced and discussed. To characterize
the potential effect of the proposed reinforcement configurations, Failure Progression and Hysteretic Response
a comparative study was performed between the GS bridge pier Fig. 8 shows the typical crack pattern and failure mode of the simu-
walls and the previously simulated specimen that was reinforced lated specimens alongside the reference steel-reinforced specimen
solely with steel bars (S4-80). To facilitate comparison, the con- (S4-80). GS1 exhibited a crack pattern and failure mode identical
crete used to simulate the GS bridge pier walls was identical to to that observed in S4-80 [Figs. 8(a and b)], which indicated that
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

(a)

(b)

Fig. 6. Experimental versus FEA results: (a) cracks pattern and failure mode (image by author); and (b) load–drift ratio hysteretic response.
(Data from Arafa et al. 2019.)

© ASCE 04020116-8 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


8#3 steel 8#3 steel
Boundary element Boundary element
#3 steel @ 120 #3 steel @ 120
#3 GFRP ties @80

#3 GFRP ties @80


vertical reinf. vertical reinf.
#4 GFRP @ 80 #4 GFRP @ 80
horizontal reinf. horizontal reinf.
#3 GFRP @ 90 #3 GFRP @ 90
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

Θ = 45° Θ = 45°

(a) (b)

8#3 steel 8#3 steel


Boundary element Boundary element
#3 steel @ 120 #3 steel @ 120
#3 GFRP ties @80

#3 GFRP ties @80


vertical reinf. vertical reinf.
#4 GFRP @ 80 #4 GFRP @ 80
horizontal reinf. horizontal reinf.
#3 GFRP @ 90 #3 steel @ 90
Θ = 45°

(c) (d)

8#3 steel
Boundary element
#3 steel @ 120 Note: the variable between the
#3 GFRP ties @80

vertical reinf. specimens is in red color


#4 GFRP @ 80
horizontal reinf.
#3 steel @ 90

(e)

Fig. 7. Details of the simulated specimens: (a) GS1; (b) GS2; (c) GS3; (d) GS4; and (e) GS5.

the reinforcement configuration (GS1) had no effect. Beyond a few the load was then the dowel action of the yielded longitudinal bars.
cycles of the yielding point of longitudinal reinforcement, a major Therefore, this was associated with localized sliding deformations
continuous crack formed along the wall length and above the bi- that gradually dominated the behavior, which caused degradation
diagonal sliding reinforcement. Under reversed cyclic loading, in lateral strength. Failure occurred shortly afterwards due to exces-
these cracks were exposed to abrasive rubbing between their asper- sive buckling of the longitudinal reinforcement, which resulted in
ities and deterioration of aggregate interlock. The only resistance to abrupt lateral strength loss.

© ASCE 04020116-9 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Fig. 9 shows the applied cyclic load versus the lateral drift ratio
and the points that illustrate the special events of damage during the
loading process. The behavior of all specimens was initially linear
up to the formation of the first crack. During early loading, the lon-
gitudinal reinforcement at the boundary yielded, followed by widen-
Sliding
ing in the hysteretic loops and a gradual decrease in overall stiffness.
Sliding After a few cycles, the hysteretic loops for specimens S4-80, GS1,
and GS5 widened noticeably with excessive residual deformations
(deformation at zero loading), as shown in Figs. 9(a, b, and f), re-
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

spectively. The reloading process (in the opposite loading direc-


tion) exhibited initial softening followed by gradual stiffening,
(a) (b)
which was attributed to the localized shear deformations. How-
ever, specimens GS2, GS3, and GS4 exhibited wide hysteretic
loops, but with noticeably controlled residual deformations com-
pared with their companion specimens (S4-80, GS1, and GS5)
up to a lateral drift ratio of 1.5% [Figs. 9(c–e)]. At this stage, speci-
men GS4 experienced strength degradation with excessive residual
damage due to shear distortion, followed by a decrease in lateral
Bars strength. In contrast, specimens GS2 and GS3 maintained their re-
Bars
rupture rupture sponse with controlled deformations up to failure.

Envelope Curve Response


(c) (d)
Fig. 10 shows the load–drift ratio envelope curves for the simulated
specimens. Comparable initial stiffness was seen in all specimens
up to initiation of the first flexural crack. Following this point,
and due to the lower modulus of elasticity of GFRP bars compared
with steel bars, GS bridge pier walls exhibited a slightly softer re-
sponse than S4-80 until the yielding point that occurred at a lateral
Wall drift ratio from 0.35% to 0.4%. This behavior would be preferable
distortion
in terms of reducing the seismic force demand, although it would
Sliding
slightly increase the displacement demand. After reaching the
yielding point, GS1, GS4, and GS5 exhibited envelope curves sim-
ilar to S4-80. Fig. 10 shows that GS1, GS4, and GS5 specimens
(e) (f) continued to carry load, but with much softer behavior. GS4 and
Fig. 8. Cracks pattern and failure mode: (a) S4-80; (b) GS1; (c) GS2; GS5 exhibited slightly stiffer behavior than their companions
(d) GS3; (e) GS4; and (f) GS5. (S4-80 and GS1) up to the stage at which shear deformations
started to dominate the behavior at a lateral drift of 1.3%, and
The behavior of specimens GS2 and GS3 with distributed GFRP 1.5%, respectively. After this point, the ultimate strength deterio-
bars along the wall height was completely different [Figs. 8(c and d)], rated, followed by failure at a lateral drift of 2.1%. In contrast,
specimens GS2 and GS3 continued to carry load without strength
since the bars prevented the type of horizontal crack that caused
degradation to achieve ultimate strengths and drift ratios that
sliding in other specimens. This was due to the elastic nature of
were sustainably higher than those of their companions. Specimen
GFRP bars, which helped cracks close and realign during load re-
GS2 outperformed its counterparts S4-80 and GS1 by approxi-
versals and distributed deformations along the wall height instead
mately 41% and 20% in terms of ultimate strength and drift ratio,
of mobilizing it at specific points. Hence, the walls continued to
respectively. These values were slightly lower in specimen GS3
carry load without strength degradation despite the specimens (25% and 15%, respectively), because fewer GFRP bars (relative
exhibiting much higher displacement than that at yielding to GS2) intersected the wall base which, in turn, contributed to
point. Ultimately, failure occurred due to sequential rupture of the wall’s flexural capacity. Two main reasons account for the
the GFRP bars [Figs. 8(c and d)] and was accompanied by an maintenance of stiffness and strength gain after the yielding point
abrupt loss of lateral strength. It is noted that even though the in specimens GS2 and GS3: (1) GFRP bars along the web height
wall failure was flexural rupture, the results show that the concrete maintained the integrity of shear resistance along the propagated
compressive strain was 0.0064, which was much higher than the cracks, thereby, avoiding premature failure, and (2) GFRP bars
value specified in CSA codes for unconfined sections (0.0035), have much higher tensile strength (approximately three times the
and concrete crushing was imminent as bar rupture took place. yield strength of steel) with no yielding plateau.
This indicated the effectiveness of the GFRP ties used to confine As a result of the lower ultimate strain of GFRP bars relative to
the concrete core and delay the failure. steel, a major concern in designing GS bridge pier walls is the abil-
Sliding was avoided in specimen GS4, and its behavior was no- ity of the wall to achieve a high level of ductility before the GFRP
ticeably improved. This control; however, was lost as the diagonal bars rupture. To confirm this point, the ductility factor (μ), mea-
bars yielded. At this stage, the bottom wall panel substantially dis- sured as the ratio of ultimate displacement to yield displacement,
torted [Fig. 8(e)], which was associated with strength degradation was calculated for the simulated specimens and plotted alongside
followed by an abrupt strength decrease. The result with specimen their envelope curves (Fig. 10). All specimens reached a compara-
GS5 was quite different since the specimen exhibited sliding similar ble ductility factor, regardless of the added reinforcement type; the
to GS1, except for the sliding zone being shifted down [Fig. 8(f)]. average value of μ = 6.0. This was attributed to the distribution of

© ASCE 04020116-10 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Lateral drift (%) Lateral drift (%)
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
800 800
600 600
400 400

Lateral load (kN)

Lateral load (kN)


200 200
0 0
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

-200 -200
Yielding point Yielding point
-400 Onset of sliding -400 Onset of sliding
-600 Excessive buckling -600 Excessive buckling
-800 -800
-50 -40 -30 -20 -10 0 10 20 30 40 50 -50 -40 -30 -20 -10 0 10 20 30 40 50
(a) Lateral displacement (mm) (b) Lateral displacement (mm)
Lateral drift (%) Lateral drift (%)
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
800 800
600 600
400 400

Lateral load (kN)


Lateral load (kN)

200 200
0 0
-200 Yielding point -200
Yielding point
-400 Bars rupture -400 Bars rupture
-600 -600
-800 -800
(c) -50 -40 -30 -20 -10 0 10 20 30 40 50 -50 -40 -30 -20 -10 0 10 20 30 40 50
Lateral displacement (mm) (d)
Lateral displacement (mm)
Lateral drift (%) Lateral drift (%)
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
1000 1000
800 800
600 600
Lateral load (kN)

Lateral load (kN)

400 400
200 200
0 0
-200 -200
-400 -400 Yielding point
Yielding point
-600 Diagonal bars yielded -600 Onset of sliding
-800 Excessive distortion -800 Excessive buckling
-1000 -1000
-50 -40 -30 -20 -10 0 10 20 30 40 50 -50 -40 -30 -20 -10 0 10 20 30 40 50
(e ) Lateral displacement (mm) (f) Lateral displacement (mm)

Fig. 9. Lateral load versus drift ratio hysteretic response: (a) S4-80; (b) GS1; (c) GS2; (d) GS3; (e) GS4; and (f) GS5.

GFRP bars in the web far away from the outermost wall fibers. damage. In contrast, spreading the GFRP bars along the web
Hence, it could be inferred that concern about the ductility issue height in specimens GS2 and GS3 controlled residual deforma-
could be offset by carefully selecting the location of the GFRP bars. tions. This control; however, appears to be a function of drift
ratio since the normalized residual deformations significantly
changed after the yielding point. Specifically, whereas the nor-
Residual Deformations malized residual deformations of specimens of GS2 and GS3
To facilitate comparison between the reinforcement configura- were 0.86 and 0.90, respectively, at the yielding point these values
tions studied in terms of their efficiency in controlling damage, were 0.72 and 0.81, at a lateral drift of 1.1% at which point shear
the residual displacement (displacement at zero loading) for deformations began to dominate the behavior of specimens S4-80
each specimen was determined at each drift level and normalized and GS1 and the impact of the GFRP bars became more apparent.
to that of the reference specimen (S4-80) at the same drift level. As result, the specimens achieved normalized residual deforma-
Fig. 11 shows the obtained normalized residual deformations (r) tions of 0.5 and 0.6, respectively at a lateral drift ratio of 2%,
versus drift ratio. For specimens GS1 and GS5, the value of r which corresponded to the failure of specimen S4-80.
was almost equal to unity at all drift levels, which indicated the Spreading diagonal steel bars along the wall height in specimen
ineffectiveness of the reinforcement configurations in controlling GS4 appeared to be effective in controlling damage in the initial

© ASCE 04020116-11 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Lateral drift (%)
0 0.5 1 1.5 2 2.5
800
S4-80 GS1 μ = 6.2
700 GS2 GS3
GS5 GS4 μ = 5.8
600

Lateral load (kN)


Yielding point
500
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

μ = 6.0
400
Strength degradation
300 μ = 6.1

200

100

0
0 10 20 30 40 50
Lateral displacement (mm)

Fig. 10. Envelop curves for load–drift ratio hysteretic response.

Drift ratio (%) Lateral drift (%)


0 0.5 1 1.5 2 0 0.5 1 1.5 2 2.5
Cumulative dissipated energy (kN.mm)
120000
Normalized by S4-80 residual deformations

GS1 GS2 GS3 S4-80 GS1 GS2


1.4 100000 GS3 GS4 GS5
GS5 GS4

1.2 80000

60000
1
40000
Yielding point

0.8 20000

0
0.6 0 10 20 30 40 50
Lateral displacement (mm)

0.4 Fig. 12. Drift ratio versus cumulative dissipated energy.


0 10 20 30 40
Lateral displacement (mm)
12%). The higher strength of specimens with controlled deforma-
Fig. 11. Drift ratio versus normalized residual deformations. tions (GS2 and GS3) appeared to compensate the wider loops of
specimens with higher residual deformation (S4-80, GS1, and
GS5), which led to a similar area enclosed in the hysteretic
stages of loading up to a lateral drift of 1.5%; however, it was much loops. Therefore, it was concluded that controlling the damage
lower than its companion specimen (GS2). This effect; however, did not significantly impair the walls’ ability to dissipate energy.
vanished as the diagonal bars yielded.

Discussion of Analytical Results


Energy Dissipation
Based on the previous discussions, specimens GS2, GS3, and GS4
Energy dissipation through hysteretic damping is an important could achieve the demand of ductile recoverable performance.
index when designing structures subjected to strong earthquakes, Controlling damage in specimen GS4; however, was related to
because it reduces the amplitude of the seismic response. This con- the yielding point, since it vanished as the diagonal steel bars
sequentially reduces the strength demands of the structure, which yielded at a lateral drift of 1.5%. This drift ratio was much lower
makes the design more economical. Therefore, investigating the than the limit of drift level of 2% specified in the CCBFC (2015)
ability of GS bridge pier walls to dissipate energy is necessary. and ASCE/SEI 41 (2013). In contrast, specimens GS2 and GS3
In this context, the cumulative energy dissipation was calculated safely achieved this level of drift capacity as well as a stable hys-
for each specimen by summing up the dissipated energy values teretic response with no sign of premature failure. Furthermore,
in consecutive load–displacement loops up to failure and plotting both specimens controlled the damage by 46% and 40%, respec-
against drift ratio (Fig. 12). It can be seen that specimen GS4 exhib- tively, compared with the reference specimen (S4-80), and
ited the highest energy dissipation at all loading stages with modest achieved comparable displacement ductility and cumulative energy
differences relative to the other specimens (a difference within dissipation. This performance was caused by spreading GFRP bars

© ASCE 04020116-12 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


either diagonally or vertically along the web height which, in turn, part or even all of a bridge’s gravity loads. Therefore, it is important
prevented the shear deformations that constitute a large portion of to study the impact of this parameter on the walls’ behavior. Three
residual displacement. Despite the slightly higher ability of diago- levels of constant axial load ratio (ALR) were investigated: 5%,
nal bars to control residual deformation, difficulties in their placing, 10%, and 15% of the product of concrete compressive strength
cutting, and anchoring might hinder their use. Hence, using vertical and gross cross-sectional area ( f ′c × Ac ). The selected ALR were
GFRP bars might represent an attractive alternative. Specimen GS3 previously reported as representative of axial load applied at the
was; therefore, selected to extend this study to include the effect of base of bridges (Su and Wong 2007). For simplicity, the specimens
axial loading, as discussed in the following sections. were identified as GS3 followed by ALR.
Fig. 13 shows the predicted failure modes of GS3 under different
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

ALR. The results demonstrated that axial load noticeably decreased


Effect of Axial Load the intensity of the propagated cracks. The results indicated that the
failure mode in all simulated specimens was preceded by sequential
In addition to their importance in resisting lateral loads induced by rupture of vertical GFRP bars, except for specimen GS3-15. In this
earthquakes, bridge pier walls are used as bearing elements to carry case, the failure was preceded by concrete core crushing at the

(a) (b) (c) (d)

Fig. 13. Effect of axial load on crack pattern and failure mode (wall GS3): (a) ALR = 0.0%; (b) ALR = 5%; (c) ALR = 10%; and (d) ALR = 15%.

Lateral drift (%) Lateral drift (%)


-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
1000 1000
800 800
600 600
400 400
Load (kN)

Load (kN)

200 200
0 0
-200 -200
-400 -400
-600 Yielding point -600 Yielding point
-800 Bars rupture -800 Bars rupture
-1000 -1000
-50 -40 -30 -20 -10 0 10 20 30 40 50 -50 -40 -30 -20 -10 0 10 20 30 40 50
Lateral displacement (mm) Lateral displacement (mm)
(a) (b)
Lateral drift (%) Lateral drift (%)
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
1000 1000
800 800
600 600
400 400
Load (kN)

200 200
0 0
-200 -200
-400 -400
Yielding point Yielding point
-600 -600
Bars rupture Concrete crushing
-800 -800
-1000 -1000
-50 -40 -30 -20 -10 0 10 20 30 40 50 -50 -40 -30 -20 -10 0 10 20 30 40 50
(c) Lateral displacement (mm) (d) Lateral displacement (mm)

Fig. 14. Effect of ALR on load–drift ratio hysteretic response (wall GS3): (a) ALR = 0.0%; (b) ALR = 5%; (c) ALR = 10%; and (d) ALR = 15%.

© ASCE 04020116-13 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Lateral drift (%) Lateral drift (%)
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
1000 25
μ = 5.8

Residual displacement (mm)


μ = 6.2 ALR = 0.0%
800 20
μ = 5.9
Load (kN)
μ = 5.8
600 15 ALR = 5%

400 10
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

ALR = 0.0%
ALR = 10%
ALR = 15%
Yielding point

ALR = 5%
ALR = 10%
200 Bars rupture
5
Concrete crushing
0 ALR = 15%
0
0 10 20 30 40 50
0 10 20 30 40 50
Lateral displacement (mm)
Lateral displacement (mm)
Fig. 15. Envelop curves for the investigated effect of ALR (reference
wall GS3). Fig. 16. Effect of axial load on residual displacement (reference wall
GS3).

boundary element under compression. This shift in failure mode


was due to the added compressive strain induced by axial load; Lateral drift (%)

Cumulative dissipated energy (kN.mm)


therefore, the concrete core attained its ultimate strain before the 0 1 2 3
GFRP bars. The lateral load–drift ratio hysteric curves for the simu- 140000
ALR = 0.0%
lated walls are shown in Fig. 14. It can be seen that the hysteresis 120000 ALR = 5%
loops changed to a flag shape, for instance, rounded hysteretic
100000 ALR = 10%
loops aimed approximately at the origin of the load–drift ratio rela-
tionship at zero loading. This was due to the combined effect of the ALR = 15%
80000
elastic nature of the GFRP bars and axial load. This indicated the
beneficial effect of axial load in the walls’ recovery. Fig. 15 60000
shows the envelope curves of the lateral load–drift ratio hysteretic 40000
response. It can be seen that the axial load delayed the initiation of The onset of
steel yielding
the first flexural cracks, the higher the applied axial load, the 20000
higher the first cracking load. Specifically, the first flexural crack- 0
ing load almost doubled when the ALR increased by 15% (speci- 0 20 40 60
men GS3-0.0 compared with specimen GS3-15). This might be Lateral displacement (mm)
attributed to the enhanced concrete strength due to confinement
from the surrounding concrete at a high ALR. After the first flex- Fig. 17. Effect of axial load on energy dissipation (reference wall GS3).
ural cracks, the specimens exhibited softer behavior; however,
with different trends as a function of ALR that appeared to result residual deformation in specimen GS3-15 was 30% higher compared
in stiffer behavior. This caused yielding of the longitudinal steel with specimen GS3-0.0 at the yielding point, but this value amplified
bars at lower drift ratios in specimens exposed to higher ALR. For thereafter to reach 90% at the failure point. Overall, axial loading
example, the yielding of the longitudinal steel bars occurred at a could be considered as an effective component to enhance the recov-
lateral drift ratio of 0.42% and 0.3% for specimens GS3-0.0 and ery of GS bridge pier walls after an earthquake event.
GS3-15, respectively. After this stage, all specimens continued to Fig. 17 shows the calculated cumulative dissipated energy ver-
carry load without strength degradation up to failure (Fig. 15). At sus the lateral drift ratio. Despite the lower residual displacement
this point, the specimens exposed to higher ALR attained higher with relatively tight loops in the walls under high ALR, similar
ultimate strength (e.g., GS3-15 achieved 44% gain in ultimate cumulative energy dissipation was noted for all specimens at
strength compared with specimen GS3-0.0), but at a lower drift the same drift level. This was due to the stiffer behavior of the
ratio. The reduction in ultimate drift ratio appeared to be insignif- specimens under higher axial load; hence, the area enclosed by
icant as long as the failure started by bar rupture (the difference in hysteretic loops were similar. Due to the lower ultimate drift
drift capacity between GS3-0.0 and GS3-10 was 11%), and it be- ratio of specimen GS3-15, the cumulative energy dissipation at
came substantial when the failure was preceded by concrete failure was significantly lower than that in its counterpart speci-
crushing (the difference in drift capacity between GS3-0.0 and mens. Specimen GS3-15 achieved cumulative energy dissipation
GS3-15 was 29%). However, specimen GS3-15 exhibited similar approximately 30%, 36%, and 40% of the cumulative energy of
ductility factor to that in the counterpart specimen GS3-0.0 (duc- its counterparts GS3-0.0, GS3-5, and GS3-10, respectively. This
tility factor was equal to 6.0). This occurred since the axial load- indicated the detrimental effect of high axial loading on cumula-
ing decreased the ultimate drift ratio and caused a reduction in the tive energy dissipation. This could be overridden through special
lateral drift that corresponded to yielding of the longitudinal steel confinement at the boundary elements.
bars, which offset the former reduction.
To further evaluate the effect of axial loads, the residual displace-
ment for each specimen was plotted against the lateral drift ratio Conclusions
shown in Fig. 16. This figure shows the beneficial effect of axial
load in controlling residual displacement; however, this effect was This study aimed to provide insight into the potential of the incor-
more prominent beyond the yielding point. The control of the poration of GFRP bars to develop more resilient bridge pier walls

© ASCE 04020116-14 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


that are recoverable after strong earthquakes. Based on the analysis Arafa, A., A. Farghaly, and B. Benmokrane. 2018c. “Prediction of flexural
of the numerical results, the following conclusions were drawn: and shear strength of concrete squat walls reinforced with GFRP bars.”
1. The FEA provided accurate simulations of the experimental re- J. Compos. Constr. 22 (4): 06018001. https://doi.org/10.1061/(ASCE)
sponse of the tested steel and GFRP-reinforced concrete bridge CC.1943-5614.0000854.
pier walls. Arafa, A., A. Farghaly, and B. Benmokrane. 2018d. “Evaluation of flexural
and shear stiffness of concrete squat walls reinforced with glass fiber-
2. Extending the GFRP bars, either arranged diagonally or verti-
reinforced polymer bars.” ACI Struct. J. 115 (1): 211–221. https://doi
cally, along the wall height was necessary to ensure their effec- .org/10.14359/51700987.
tiveness in controlling damage. Arafa, A., A. Farghaly, and B. Benmokrane. 2019. “Nonlinear finite-
3. Controlling damage in GS bridge pier walls did not significantly element analysis for predicting the behavior of concrete squat walls re-
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

impair the walls’ ability to dissipate energy or the ductility inforced with GFRP bars.” J. Struct. Eng. 145 (10): 04019107. https://
index, but careful selection of GFRP bars location was necessary. doi.org/10.1061/(ASCE)ST.1943-541X.0002399.
4. Within the range of the simulated GS walls, specimens GS2 and ASCE. 2013. Seismic evaluation and retrofit of existing buildings. ASCE/
GS3, which were provided with diagonal or vertical GFRP bars SEI 41-13. Reston, VA: ASCE.
along the wall height, satisfied the demand of ductile recoverable Bencardino, F., A. Condello, and L. Ombres. 2016. “Numerical and analyt-
performance up to the maximum limit of drift ratio (2%) recom- ical modeling of concrete beams with steel, FRP and hybrid FRP-steel
mended by the CCBFC (2015) and ASCE/SEI 41 (2013). Never- reinforcements.” Compos. Struct. 140: 53–65. https://doi.org/10.1016/j
.compstruct.2015.12.045.
theless, using vertical GFRP bars might be more attractive.
Bentz, E., C. 1999. “Sectional analysis of reinforced concrete structures.”
5. Axial load on GS walls enhanced their ultimate strength, but Ph.D. thesis, Civil Engineering Dept., Toronto Univ.
slightly reduced their ultimate drift ratio; however, this effect CAN/CSA (Canadian Standards Association). 2012. Design and construc-
appeared to be more prominent when the failure mode was flex- tion of building components with fiber-reinforced polymers. CSA
ural compression. S806-12. Mississauga, ON, Canada: CSA.
6. Axial load could be considered as an effective component to en- CAN/CSA (Canadian Standard Association). 2014a. Design of concrete
hance the recovery of GS walls after an earthquake event. structures. Standard CSA-A23.3-14. Mississauga, ON, Canada: CSA.
7. The achieved results might be affected by changing the wall’s CAN/CSA (Canadian Standards Association). 2014b. Canadian highway
aspect ratio. Further investigations are, therefore, needed to as- bridge design code. CSA S6-19. Mississauga, ON, Canada: CSA.
sess the effect of this parameter. CCBFC (Canadian Commission on Building and Fire Codes). 2015.
8. More studies are needed to test the durability of the proposed National Building Code of Canada (NBCC). Montreal, QC, Canada:
National Research Council of Canada.
GS configurations under different environmental conditions,
CSA (Canadian Standards Association). 2010. Specification for fibre rein-
such as chloride permeability, repeated freeze–thaw cycles, forced polymers. CSA S807. Mississauga, ON, Canada: CSA.
and various chemical environments. ElGawady, M., and A. Sha’lan. 2011. “Seismic behavior of self-centering
precast segmental bridge bents.” J. Bridge Eng. 16 (3): 328–339. https://
doi.org/10.1061/(ASCE)BE.1943-5592.0000174.
Data Availability Statement Eligehausen, R., E. Popov, and V. Bertero. 1983. Local bond stress-slip re-
lationship of deformed bars under generalized excitations, 69–80. Rep.
All data, models, and code generated or used during the study No. UCB/EERC-83/23. Berkeley, CA: Earthquake Engineering Center,
appear in the published article. Univ. of California.
Fahmy, M. F. M., and Z. Wu. 2016. “Exploratory study of seismic response
of deficient lap-splice columns retrofitted with near surface–mounted
Acknowledgments basalt FRP bars.” J. Struct. Eng. 142 (6): 04016020. https://doi.org/10
.1061/(ASCE)ST.1943-541X.0001462.
This research was conducted with funding from the Tier-1 Canada Ghazizadeh, S., C. A. Cruz-Noguez, and F. Talaei. 2018. “Analytical
Research Chair in Advanced Composite Materials for Civil Struc- model for hybrid FRP-steel reinforced shear walls.” Eng. Struct. 156:
tures, the Natural Sciences and Engineering Research Council of 556–566. https://doi.org/10.1016/j.engstruct.2017.11.060.
Canada (NSERC), and the NSERC Industrial Research Chair in Ghomi, S., and E. El-Salakawy. 2018. “Seismic behavior of exterior
FRP Reinforcement for Concrete Infrastructure. The authors are es- GFRP-RC beam–column connections: Analytical study.” J. Compos.
Constr. 22 (4): 04018022. https://doi.org/10.1061/(ASCE)CC.1943
pecially grateful to Sohag University (Egypt) for the scholarship it
-5614.0000858.
awarded to the first author. Gulec, C. K., and A. S. Whittaker. 2009. Performance-based assessment
and design of squat reinforced concrete shear walls. New York:
Multidisciplinary Center for Earthquake Engineering Research.
References He, X. G., and A. K. H. Kwan. 2001. “Modeling dowel action of rein-
forcement bars for finite element analysis of concrete structures.”
AASHTO (American Association of State Highway and Transportation Comput. Struct. 79 (6): 595–604. https://doi.org/10.1016/S0045
Officials). 2018. AASHTO LRFD bridge design guide specifications -7949(00)00158-9.
for GFRP–reinforced concrete. 2nd ed. Washington, DC: AASHTO Hognestad, E. 1951. A study on combined bending and axial load in rein-
LRFD. forced concrete members, 43–46. Monticello, IL: Univ. of Illinois
ACI (American Concrete Institute). 2015. Guide for the design and con- Engineering Experiment Station.
struction of structural concrete reinforced with fiber-reinforced poly- Ibrahim, A., M. F. M. Fahmy, and Z. Wu. 2016. “3D finite element mod-
mer (FRP) bars. ACI 440.1R-15. Farmington Hills, MI: ACI. eling of bond-controlled behavior of steel and basalt FRP-reinforced
Arafa, A., A. Farghaly, and B. Benmokrane. 2018a. “Experimental behav- concrete square bridge columns under lateral loading.” Compos.
ior of GFRP-reinforced concrete squat walls subjected to simulated Struct. 143: 33–52. https://doi.org/10.1016/j.compstruct.2016.01.014.
earthquake load.” J. Compos. Constr. 22 (2): 04018003. https://doi Iemura, H., Y. Takahashi, and N. Sogabe. 2006. “Two-level seismic design
.org/10.1061/(ASCE)CC.1943-5614.0000836. method using post-yield stiffness and its application to unbonded bar
Arafa, A., A. Farghaly, and B. Benmokrane. 2018b. “Effect of web rein- reinforced concrete piers.” Struct. Eng./Earthquake Eng. 23 (1):
forcement on the seismic response of concrete squat walls reinforced 109s–116s. https://doi.org/10.2208/jsceseee.23.109s.
with glass-FRP bars.” Eng. Struct. 174 (1): 712–723. https://doi.org Kupfer, H., H. K. Hilsdorf, and H. Rusch. 1969. “Behavior of concrete
/10.1016/j.engstruct.2018.07.092. under biaxial stress.” ACI Struct. 87 (2): 656–666.

© ASCE 04020116-15 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116


Kupfer, H. B., and K. H. Gerstle. 1973. “Behavior of concrete under biaxial Scott, B. D., R. Park, and M. Priestley. 1982. “Stress-strain behavior of
stresses.” J. Eng. Mech. Div. 99 (4): 853–866. concrete confined by overlapping hoops at low and high strain rates.”
Lai, D. 2001. “Crack shear-slip in reinforced concrete elements.” Master ACI Struct. J. 79 (1): 13–27.
thesis, Dept. of Civil Engineering, Univ. of Toronto. Seckin, M. 1981. “Hysteretic behavior of cast-in-place exterior beam-
Luna, B. N., J. P. Rivera, and A. S. Whittaker. 2015. “Seismic behavior of column-slab subassemblies.” Ph.D. thesis, Dept. of Civil Engineering,
low-aspect-ratio reinforced concrete shear walls.” ACI Struct. J. Univ. of Toronto.
112 (5): 593–604. Sittipunt, C., S. L. Wood, P. Lukkunaprasit, and P. Pattararattanakul. 2001.
Massone, L. M., K. Orakcal, and J. W. Wallace. 2009. “Modeling of squat “Cyclic behavior of reinforced concrete structural walls with diagonal
structural walls controlled by shear.” ACI Struct. J. 106 (5): 646–655. web reinforcement.” ACI Struct. J. 98 (4): 554–562.
Moehle, J. P. 1999. TAI-139: Wu-Hsi (U-Shi) bridge. photograph, NISEE Su, R. K. L., and S. M. Wong. 2007. “Seismic behaviour of slender rein-
Downloaded from ascelibrary.org by Imperial College of Science, Technology and Medicine on 11/23/20. Copyright ASCE. For personal use only; all rights reserved.

e-Library. Berkeley, CA: Univ. of California. forced concrete shear walls under high axial load ratio.” Eng. Struct.
Mohamed, N., A. Farghaly, B. Benmokrane, and K. W. Neale. 2014. 29: 1957–1965. https://doi.org/10.1016/j.engstruct.2006.10.020.
“Numerical simulation of mid-rise concrete shear walls Takahashi, S., K. Yoshida, T. Ichinose, Y. Sanada, K. Matsumoto, H.
reinforced with GFRP bars subjected to lateral displacement rever- Fukuyama, and H. Suwada. 2013. “Flexural drift capacity of rein-
sals.” Eng. Struct. 73: 62–71. https://doi.org/10.1016/j.engstruct forced concrete wall with limited confinement.” ACI Struct. J.
.2014.04.050. 110 (1): 95–104.
Palermo, D., and F. J. Vecchio. 2002. “Behavior of three-dimensional rein- Vecchio, F. J., and M. P. Collins. 1993. “Compression response of cracked
forced concrete shear walls.” ACI Struct. J. 99 (1): 81–89. reinforced concrete.” J. Struct. Eng. 119 (12): 3590–3610. https://doi
Palermo, D., and F. J. Vecchio. 2007. “Simulation of cyclically loaded .org/10.1061/(ASCE)0733-9445(1993)119:12(3590).
concrete structures based on the finite-element method.” J. Struct. Vecchio, F. J., and I. McQuade. 2011. “Towards improved modeling of
Eng. 133 (5): 728–738. https://doi.org/10.1061/(ASCE)0733-9445(2007) steel-concrete composite wall elements.” Nucl. Eng. Des. 241 (8):
133:5(728). 2629–2642. https://doi.org/10.1016/j.nucengdes.2011.04.006.
Qin, R., A. Zhou, and D. Lau. 2017. “Effect of reinforcement ratio on the Vint, L. M. 2012. “Investigation of bond properties of glass fibre reinforced
flexural performance of hybrid FRP reinforced concrete beams.” polymer bars in concrete under direct tension.” Ph.D. thesis, Dept. of
Compos. Part B: Eng. 9: 108–200. Civil Engineering, Univ. of Toronto.
Saatcioglu, M. 1991. “Hysteretic shear response of low-rise walls.” In Proc., Wong, P. S., and F. J. Vecchio. 2002. Vector 2 & formworks user’s man-
of Int. Workshop on Concrete Shear in Earthquake, 105–114. Houston, uals. Toronto: Dept. of Civil Engineering, Univ. of Toronto.
TX: National Science Foundation, Univ. of Houston. Wu, Z. S., M. F. M. Fahmy, and G. Wu. 2009. “Safety enhancement of
Saiidi, M. S., M. O’Brien, and M. S. Zadeh. 2009. “Cyclic response of con- urban structures with structural recoverability and controllability.”
crete bridge columns using super elastic nitinol and bendable concrete.” J. Earthquake Tsunami 3 (3): 143–174. https://doi.org/10.1142
ACI Struct J. 106 (1): 69–77. /S1793431109000561.

© ASCE 04020116-16 J. Bridge Eng.

J. Bridge Eng., 2021, 26(2): 04020116

You might also like