Gokhale - Effect of Loading History On Axial Cyclic Behaviour of RC Prisms Representing Wall Boundary Zones PDF

Engineering Structures 233 (2021) 111871
Contents lists available at ScienceDirect
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Effect of loading history on axial cyclic behaviour of RC prisms representing

wall boundary zones
R.A. Gokhale a, *, M. Tripathi b, F. Dashti b, R.P. Dhakal a
a
Dept. of Civil and Natural Resources Engineering, University of Canterbury, Christchurch, New Zealand
b
Quake Centre, University of Canterbury, Christchurch, New Zealand
A R T I C L E I N F O A B S T R A C T
Keywords: Susceptibility of flexure-dominated reinforced concrete walls to premature compression failure modes (global
Reinforced concrete wall lateral instability, local rebar buckling, etc.) was exposed during the recent past earthquake events in Chile and
Boundary zone New Zealand. Numerous experimental studies have been conducted since then on isolated columns representing
Loading history
wall boundary zones to study critical parameters influencing these compression failure modes. However, a
Rebar buckling
Out-of-plane buckling
specific study including the effect of different loading histories, particularly in terms of cycle content (strain
range and corresponding cycle count), is missing in the literature. In this study, uniaxial cyclic tests were per
formed on six doubly reinforced prisms idealised as the boundary elements of rectangular flexure-dominated
walls to evaluate the effect of various loading histories on different compression failure modes. The uniaxial
cyclic loading histories applied to these specimens included a conventional loading protocol (obtained directly
from lateral cyclic loading test of a prototype wall) as well as two earthquake loading histories comprising cycle
content representative of the cyclic demand expected from near-fault and far-fault earthquakes. In some of these
tests, the level of compressive strains in the adopted loading protocol was reduced to study their impact on the
compression failure modes. Although results re-affirmed the influence of imposed tensile strain on the global
lateral instability and rebar buckling phenomena, the effect of variation in the number and amplitude of the
strain cycles in the loading history was also displayed. Furthermore, results not only highlighted the impact of
compressive strain variation on the global out-of-plane buckling response but also revealed the detrimental ef
fects of rebar buckling on the low-cycle fatigue life of the specimens.
1. Introduction The review of previous experimental studies on rectangular RC

prisms representing wall boundary elements suggests that extensive
Structural walls provide relatively high in-plane stiffness and are research has been performed on understanding the parameters that
therefore considered as the preferred lateral load resisting system in induce premature compression failure modes under (i) monotonic
regions of medium to high seismicity. Performance of a flexure- compression, (ii) pre-strain (tensile strain) prior to reloading mono
dominated reinforced concrete (RC) wall (shear-span ratio ≥ 3.0) tonically into compression, and (iii) reversed cyclic tension-compression
against in-plane lateral cyclic actions depends on the response of its loading. The research reported in literature has shown the effect of
confined end regions known as the boundary zones. These confined end different loading histories on the response of isolated boundary elements
regions (or boundary zones) are subjected to tensile and compressive and reaffirmed the importance of cyclic loading over the monotonic
strain reversals during seismic events. A more desirable deformation compression loading. Chai and Elayer [4] conducted an experimental
capacity of a structural wall can be achieved by suppressing the pre study to estimate maximum tensile strain that could be imposed on the
mature compression failure modes in its boundary regions. In order to boundary elements of ductile RC walls. For this purpose, two different
assess the performance of flexure-dominated walls, a simplified loading protocols keeping the tensile to compressive strain ratio con
approach of experimentally testing RC columns (or prisms) representing stant at 5 and 7 were adopted. Hilson et al. [17] performed reversed
wall boundary zones under uniaxial loading is commonly adopted by cyclic axial loading tests on isolated rectangular boundary elements to
researchers [4,21,17;35,32,42,16,39,12]. assess the maximum tensile strain required prior to reloading into
* Corresponding author.
E-mail address: [email protected] (R.A. Gokhale).
https://doi.org/10.1016/j.engstruct.2021.111871
Received 27 June 2020; Received in revised form 4 January 2021; Accepted 5 January 2021
Available online 9 February 2021
0141-0296/© 2021 Elsevier Ltd. All rights reserved.
R.A. Gokhale et al. Engineering Structures 233 (2021) 111871
compression to initiate longitudinal bar buckling failure and the role of experimental campaign, at the Structural Engineering Laboratory of
cover concrete in delaying bar buckling. Two loading protocols were University of Canterbury to investigate the effect of different loading
employed: (i) the compressive strains were limited to 0.2% while the histories on the axial cyclic behaviour of RC prisms. While acknowl
tensile strains were gradually increased up to 4.5%, and (ii) the tensile edging that size effect could make the performance of the specimen
strains were held constant at 2% while the compressive strains were different from that of full-scale walls, these half-scaled prisms repre
gradually increased up to failure. Taleb et al. [35] tested rectangular sented boundary zones in the bottom storey of a 4 m long multi-storey
prisms simulating wall boundaries under axial cyclic loading to study high slender RC wall with a shear span ratio of 3 and an inter-storey
the impact of transverse reinforcement detailing, prism slenderness, and height of 4 m. The height of the idealised boundary element was
loading type (monotonic compression vs reversed cyclic) on the failure considered equal to the length of plastic hinge region of the wall.
patterns and compression load carrying capacity of the confined prism Experimental tests conducted by Tripathi et al. [37,38] on ductile wall
specimens. For the cyclic loading protocol, a tensile strain to compres units represented the bottom storey of the same prototype wall, where it
sive strain ratio of 5, similar to Chai and Elayer [4], was maintained was observed that the plastic tensile strains extended up to 60% of the
throughout the test. Haro et al. [16] investigated the influence of unsupported wall height. Therefore, for the current experimental pro
different loading protocols including bi-directional loading and longi gram, a prism height of 1.2 m was adopted corresponding to the scaled
tudinal reinforcement ratio on the onset of OOP buckling by conducting wall height of 2 m. However, the current experimental program
experimental tests on idealised RC columns. Uniaxial cyclic loading included two different prism thicknesses to facilitate study of both
protocols consisted of three different load histories: cyclic loads with global buckling as well as local rebar buckling failure modes. The six
tensile strains of increasing amplitudes and compressive strains held specimens in the test matrix were divided into two groups, with Group 1
constant equivalent to the specimen compressive capacity; megathrust – consisting of three prisms of 120 mm thickness, and Group 2 consisting
type earthquake; and strike-slip type earthquake. Tripathi et al. [39] of three prisms of 200 mm thickness. All six specimens had identical
performed axial tests on rectangular RC prisms to study the critical pa reinforcement detailing with the length and height of 330 mm and 1200
rameters influencing localized compression failure mode. The key pa mm, respectively. Specimens were designed in accordance with New
rameters experimentally investigated in this study included loading Zealand Concrete Standard [24] requirements for the ductile plastic
history, transverse reinforcement detailing, longitudinal reinforcement hinge regions in the wall boundary elements. Table 1 and Fig. 1 provide
detailing and cover concrete. The loading protocol adopted for this study the geometry and reinforcement details of the specimens. To enforce
was directly derived using the strain measurements from their experi fixed-fixed boundary conditions at the top and bottom of the prisms, the
mental tests on the corresponding wall units. More recently, Gokhale specimens were cast dumb-bell shaped. Two additional D12 bars that
et al. [12] conducted prism tests to investigate the influence of prism extend 200 mm into the top and bottom concrete end blocks facilitate
slenderness on their axial response. All specimens were subjected to damage to be concentrated within the prism (area of interest), rather
same loading protocol that had a constant tensile to compressive strain than at the prism-end block interface (top and bottom). Detailed spec
ratio of about 7.3. imen design for interested readers could be found in Tripathi [36].
In order to assess the performance of structural components ratio Material properties of the specimens are provided in Table 2 and
nally, it is important to consider loading protocols that have charac Table 3.
teristics representative of the various types of earthquakes. Not only the
relationship between the tensile and compressive strain peaks are 2.2. Test setup and Instrumentation
important but also the corresponding cycle count at that strain ampli
tude (or range) needs to be considered. The effect of these parameters on The test was set up on a high capacity (10 MN) Dartec Universal
different compression failure modes, particularly the global OOP buck Testing Machine. Loading was directly applied to the test prism in the
ling and local rebar buckling failure modes, have not been thoroughly form of vertical (i.e., axial) displacements at a slow rate through the
explored previously. Conventional loading protocols consider strain vertical actuator located below the bottom Dartec head. Fig. 2 below
cycles of increasing amplitudes with equal number of cycles at each illustrates the test setup.
amplitude. However, in an actual earthquake scenario, the cycle content Twenty vertical linear potentiometers were attached on the West and
is somewhat different from the one considered in the laboratory, where East faces (i.e., the longer faces) of the boundary element, as shown in
it is generally dominated by large number of small amplitude cycles [2]. Fig. 3 (a) and (b) respectively. The vertical linear potentiometers were
Moreover, a more specific study exploring the influence of compressive evenly spaced over the middle portion with a gauge length of 250 mm,
strains on the two compression failure modes is missing, even though its whereas the end ones that were close to the top and bottom end blocks
effect has been observed in the past experiments. had a gauge length of 225 mm. As can be seen in Fig. 3 (b), in addition to
In this study, the effects of different uniaxial cyclic loading protocols the vertical linear potentiometers, four horizontal linear potentiometers
on the response of RC prisms representing wall boundary zones are were also placed on the East face to measure any lateral displacement
examined through experimental investigation. The various loading along the length of the boundary elements. OOP deformations were
protocols that were employed comprised of a conventional loading measured using draw-wire potentiometers that were attached to the East
protocol as well as near-fault and far-fault earthquake loading histories. face of the specimens. Fig. 3 (c) shows positioning of the draw-wire
The conventional loading history was derived based on average axial potentiometers along the height of the prisms.
strains measured in the boundary zone of a prototype wall in a quasi-
static lateral cyclic loading test [39] and the earthquake loading his 2.3. Loading protocols
tories were derived following a numerical procedure previously devel
oped by the authors utilizing the earthquake ground motion records of The loading histories applied to each specimen typically consisted of
distinct characteristics [11]. Moreover, the impact of compressive reversed cyclic tension-compression cycles. Because of irreversible
strains, which were varied within the loading program, on the global cracking and permanent elongation of RC members [27,28], the strain
OOP buckling and local rebar buckling failure modes are also studied. history experienced by both reinforcement and concrete is characterised
by large tensile strains and small (in some cases no) compressive strains
2. Experimental program despite being subjected to comparable positive and negative drifts.
Hence, strain cycles with equal or comparable tensile and compressive
2.1. Test matrix strains do not represent typical cyclic actions induced in RC members
and should be avoided in experimental investigations. In this study,
A series of six uniaxial cyclic tests were performed, as a part of bigger while retaining the larger tensile to compressive strain ratio, three
2
Table 1
Specimen Matrix.
Specimen Thickness “b” Length “l” Height “Hprism” Longitudinal Reinforcement Transverse Reinforcement Spacing-to- Slenderness Ratio
ID (mm) (mm) (mm) Ratio “ρl” (%) Longitudinal Bar Diameter Ratio “s/db” “Hprism/b”
12BZ-1 120 330 1200 2.86 4.6 10.0

12BZ-2 120 330 1200 2.86 4.6 10.0
12BZ-3 120 330 1200 2.86 4.6 10.0
20BZ-1 200 330 1200 1.71 4.6 6.0
20BZ-2 200 330 1200 1.71 4.6 6.0
20BZ-3 200 330 1200 1.71 4.6 6.0
Dartec Top
Head (fixed)
Test
Specimen
Dartec
Bottom Head
Vertical
Actuator
Fig. 2. Test setup.
compressive peaks.
The first uniaxial cyclic loading protocol (LP#1) used in these tests
Fig. 1. Cross-sectional details of the tested specimens. was derived by Tripathi et al. [39] based on the strain histories
measured at the base of the wall boundaries during the quasi-static cy
clic tests of the prototype wall conducted at the University of Canter
Table 2
bury. Details of this experimental program on wall units can be found in
Concrete Compressive Strength.
Tripathi et al. [37]. The wall specimens were subjected to in-plane cyclic
Specimen ID f ‘c (MPa) lateral displacements with three symmetric cycles at each drift level.
12BZ-1 35.0 Based on the strain measurements of these experiments, an equation
12BZ-2 36.5 relating the maximum tensile (εmax) and compressive (εmin) strains
12BZ-3 39.7
generated at the base of wall boundaries was developed, as shown in
20BZ-1 37.3
20BZ-2 40.2
Equation (1).
20BZ-3 31.7 εmin
= 0.04 (εmax )− 0.48
(1)
εmax
However, Tripathi et al. [39] also hypothesized that reduced
Table 3
compressive strains may be used for the uniaxial cyclic protocol to ac
Reinforcement properties.
count for the strain gradient along the height of the wall boundary zone.
Bar D12 For the ductile slender walls that were tested, Tripathi et al. [37,38] had
fy (MPa) 307.5 observed that the inelastic tensile strains tended to spread over a
fu (MPa) 444.1 considerable portion of the wall height with increasing ductility de
E (MPa) 197,250
mand, whereas, compressive strains appeared to concentrate over a
εy 0.0015
εsh 0.03172 short height of about 300 mm. This observation is also in line with the
εu 0.228 postulation formulated by Segura et al. [33] that the compressive strains
tend to localize over a small height, approximately 2.5 times the wall
εsh – Hardening strain.
thickness, especially after spalling of cover concrete. Therefore, for the
εu – Ultimate tensile strain.
current experimental program, the compressive strains obtained from
Eq. (1) were reduced by a factor of 4 (=1200 mm/300 mm). Fig. 4 (a)
different loading histories were considered; namely: conventional, near-
shows this loading protocol in graphical form.
fault type and far-fault type. The loading protocol differed in each case
Rational assessment of a structural component requires a loading
with respect to the relationship between the tension and compression
protocol containing realistic cycle content in terms of relationship be
peaks and the corresponding cycle count at each strain range, where
tween tensile and compressive strains and corresponding number of
strain range is defined as the difference between the tensile and
3
Fig. 3. Instrumentation: (a) linear potentiometers – West face; (b) linear potentiometers – East face; and (c) string potentiometers – East face.
cycles at each strain range. For this purpose, a new loading protocol was account for strain gradient like in case of experimentally derived LP#1.
developed to represent the cumulative inelastic strain demands at the Following the numerical investigation, three relationships were ob
wall boundaries when the corresponding wall is displaced to a tained between the strain range (Δε), the corresponding cumulative
performance-based drift limit due to moderate-to-large magnitude number of inelastic cycles (N), and the maximum and minimum strains
earthquakes of a particular characteristic. Non-linear time history ana (εmax and εmin) as presented in Eqs. (2), (3) and (4). In these equations,
lyses were performed on a validated macro-model of the prototype wall the coefficients a, b and c depend on the ground motion characteristic
Specimen SWD-1 tested by Tripathi et al. [37] to generate these inelastic (say near-fault, far-fault, etc.). Fig. 5 portrays these relationships in
strain demands at the wall boundaries. The strain histories thus obtained graphical form leading to the proposed loading protocol. These earth
were further statistically evaluated for the strain ranges above a certain quake loading protocols are developed based on strain ranges as illus
threshold and corresponding cycle count. For this study, strain ranges trated in Fig. 5. For a chosen strain range, Eq. (2) gives the cumulative
below yield strain were ignored so that majority of the tensile strains number of cycles with strain ranges equal to or greater than that strain
obtained were inelastic. The methodology adopted in the development range. For instance, as such strain ranges below yield strain (εy) are
of this loading protocol is explained in detail in Gokhale et al. [11]. The ignored in this study, if Δε1 equal to εy is used in Eq. (2) then that would
wall modelling strategy adopted for the numerical analysis somewhat provide the total number of inelastic cycles (N = N1). Similarly, the
suited the prism testing concept as it facilitated both nearly uniform difference between the number of cycles (N2 − N1) corresponding to two
strain measurements along the wall boundary element height and strain chosen strain ranges (Δε1, Δε2) would provide the number of cycles with
values that were averaged over the entire cross-sectional area of the the strain ranges between Δε1 and Δε2. Likewise, Eq. (3) provides
boundary zone. Therefore, resulting earthquake loading protocols do maximum strain for a given strain range. With the strain range and the
not necessarily require reduction in their compressive strain values to maximum strain known, corresponding minimum strain is computed
4
Fig. 4. (a) Loading protocol based on wall experiments with reduced compressive strains (LP#1); (b) Loading protocol based on near-fault earthquake characteristics
with reduced compressive strains (LP#2); (c) Loading protocol based on near-fault earthquake characteristics (LP#3); and (d) Loading protocol based on far-fault
earthquake characteristics (LP#4).
5
Fig. 5. Development of loading protocol.
using Eq. (4). 3.1. Group 1: 120 mm thick specimens - failure in global OOP buckling
mode
N = a × (Δε) b
(2)
Even though the failure mode experienced by the 120 mm thick
εmax = c × (Δε) (3)
specimens was OOP instability, the failure stage differed with the
loading history, especially in the higher compressive strain loading
εmin = εmax − Δε (4)
history. In order to understand the effect of different loading histories,
A near-fault loading protocol was derived considering a suite of the mechanics behind OOP buckling is briefly described below.
seven ground motions having pulse-type characteristics. The numerical
investigation provided the constants a, b, and c as 0.024, − 1.0, and 0.88, 3.1.1. OOP buckling phenomenon
respectively [11]. As depicted in Fig. 4 (b) and (c), both loading pro OOP buckling in thin reinforced concrete walls was first observed by
tocols, LP#2 and LP#3, represent the same near-fault type earthquake Oesterle et al. [25], however, the development of this failure mode was
loading regime, except that the compressive strains in LP#2 were first described by Goodsir [14]. Later, Paulay and Priestley [26] and
reduced to study their influence on the specimen response. These Chai and Elayer [4] developed phenomenological analytical models
compressive strains were reduced by a factor of 4 like in the first (i.e., based on the experimental tests on idealised RC prisms subjected to
experiment-based) loading protocol LP#1 to maintain consistency. uniaxial cyclic loads. More recently, Rosso et al. [31] and Dashti et al.
A similar exercise for the far-fault type earthquake ground motion [5,6,7,8] have dissected this phenomenon in great detail with respect to
suite provided the constants a, b, and c as 0.04, − 1.1 and 0.85 respec singly and doubly reinforced slender walls, respectively.
tively and the resulting loading protocol LP#4 is depicted in Fig. 4 (d). Fig. 7 describes the development and recovery of OOP deformation
Table 4 presents the details of the loading protocols adopted for the as well as the progression towards OOP instability in an idealised RC
experimental program and Fig. 6 depicts them graphically. column representing buckled wall boundary zone. Fig. 7 (a) and (b)
It can be seen in Fig. 6 (b) that the number of lower magnitude in present the axial force versus average axial strain and OOP deformation
elastic cycles (εmax ≤ 10εy) are much greater than the higher magnitude versus average axial strain plots. Fig. 7 (c) illustrates the various OOP
cycles (εmax > 10εy) for the earthquake loading histories, taking about deformations associated with different loading stages during the
80–90% of the total inelastic cycles. This is somewhat in line with the development of global buckling phenomenon. Besides, the axial force
distribution of ground acceleration amplitude observed during a major versus average axial strain responses shown in Fig. 7 (d) and (e) for each
earthquake event. The effect of this variation in the cycle content along face across the prism thickness further aid in understanding the role of
with the variation in compressive strains on the compression failure asynchronous behaviour of the two rebar layers across the boundary
modes are evaluated and discussed in the following sections. element thickness in this phenomenon.
When a doubly reinforced isolated boundary element is subjected to
3. Experimental results and observations inelastic tensile strain (path o-a), wide horizontal cracks develop along
its height. Upon unloading (path a-b), elastic strain is recovered; how
Two distinct failure modes were observed during this experimental ever, residual inelastic strains in the rebars leave the cracks wide open.
program; namely the global OOP instability and local rebar buckling Thereafter, compressive stresses are solely carried by the longitudinal
followed by the core concrete crushing. The slender 120 mm thick reinforcement until complete or partial crack closure. Inherent eccen
specimens failed due to instability caused by global OOP deformations, tricities that exist across the cross-section due to unequal crack closures
whereas localised failure mode was observed in the stockier 200 mm in prior cycles and/or non-uniform load application will invariably lead
thick specimens. Therefore, in the following sections, both experimental to development of different tensile strains in the two rebar layers,
observations and discussion on the effects of different loading protocols thereby, yielding one layer in compression prior to the other. As a result,
are grouped based on specimen thicknesses. out-of-plane stiffness of the section reduces which is accompanied by an
increase in OOP deformation (path b-c). Typically, rebar layer (Face 1)
Table 4 that stretches more at the peak tensile loading tends to yield in
Loading protocol matrix. compression prior to the opposite rebar layer. Further reloading into
compression yields the second rebar layer (Face 2) expediting OOP
Loading Loading Constants Ratio Specimen
Protocol Protocol Type (εmin/ IDs deformation (ΔOOP). This rapid increase in OOP deformation (path c-d)
a b c is also evident from the hysteresis plots for the two prism faces shown in
ID εmax)
LP#1 Conventional – – – varies (1,2) 12BZ-1;
Fig. 7 (d) and (e), which show an increase in tensile strain on Face 2
20BZ-1 while the tensile strain on Face 1 reduces. At this stage, depending on the
LP#2 Near-fault 0.024 − 1.0 0.88 0.0341(1) 12BZ-2 magnitude of the applied tensile strain and associated crack width,
LP#3 Near-fault 0.024 − 1.0 0.88 0.136 12BZ-3; further compressive loading could either lead to crack closure (on Face
20BZ-2
1) facilitating concrete contribution towards load carrying capacity of
LP#4 Far-fault 0.04 − 1.1 0.85 0.176 20BZ-3
the section and thereby reducing the OOP deformation (path d-e) or
Note: (1) Compressive strains are reduced by 4. result in instability with abrupt loss in compression carrying capacity
(2) (εmin/εmax) = 0.01 (εmax)− 0.48
6
0.25 100% > 10εy ≤ 10εy

4
90% 4
0.2 80%
12
LP#4
% Distribution of Cycles
70%
0.15
Ratio (εmin/εmax)
60%
LP#3
50%
53
0.1
40% 15
30%
12
0.05 LP#1 20%
LP#2
10%
0
0%
0 0.01 0.02 0.03 0.04 0.05
Conventional Near-fault Far-fault
εmax
(a) (b)
Fig. 6. Cycle content for each loading protocol (a) tensile-compressive strain relationship and (b) cycle count.
(path d-f). At point e, the applied compressive strain of high magnitude Chai and Elayer [4] later developed an expression for the maximum
could possibly lead to concrete crushing limit state accompanied by permissible tensile strain comprising of three strain components as can
building up of OOP deformation (Δc) due to P-δ effects. be seen in Eqs. (7a) and (7b), where, εe is the elastic strain recovered
At small magnitude inelastic tensile strains close to yield strain of during unloading, εr is the reloading strain corresponding to the yielding
rebar, the difference between the tensile strains developed across the of both vertical bars, and ε*a is the axial strain corresponding to the first
boundary element thickness is usually small. However, as the loading crack closure. Both εe and εr are considered as a function of the yielding
advances into subsequent higher magnitude inelastic tensile strain cy strain of the rebar and can be quantified as (η1εy) and (η2εy), respec
cles, this difference also increases, thereby, accelerating the evolution of tively. Although Chai and Elayer [4] utilized the same stability criterion
OOP deformation when reloaded in compression. Depending on the (ξ) proposed by Paulay and Priestley [26] for computing the strain
magnitude of the difference between the tensile strains developed across component ε*a, they considered a sinusoidal curvature distribution
the prism thickness and also lateral stiffness of the cross-section, an OOP along the buckling length (Lo) instead of the circular curvature
deformation (Δt) may be induced in the opposite direction, as depicted distribution.
in Fig. 7, where rebar layer near Face 1 stretches more than the rebar
εsm = εe + εr + ε*a (7a)
layer near Face 2.
( )2
3.1.2. Analytical models predicting maximum permissible tensile strain π2 b
εsm = η1 εy + η2 εy + ξ (7b)
Although asymmetric formation of tensile strains across the prism 2 Lo
thickness is a factor that eventually leads to the development of OOP More recently, Haro et al. [15] proposed improvements to the Chai
deformation upon load reversal, it is the tensile strain excursion and and Elayer [4] model by providing alternative expressions for the three
associated crack width that determines the possibility of a timely crack strain components shown in Eq. (7a). The resulting computational
closure under compressive loading, thereafter, recovering the already- model is expressed in Eq. (8).
incurred OOP deformation. [ ]
( )2 ( )2
Paulay and Priestley [26] proposed a stability criterion (ξ) corre 24 b 5 b
εsm = ξ+ − 2.4 + 160 − 1.9ln(ρ) εy (8)
sponding to a crushing limit state derived using an equivalent rectan 5 Lo 9 Lo
gular stress block for the compressed concrete region at mid-height of
the buckled wall strip of length (Lo) and assuming yielding of both rebar Nevertheless, it is important to note that the two basic phenome
layers. Using this stability criterion, they formulated an expression for nological models by Paulay and Priestley [26] and Chai and Elayer [4]
maximum permissible tensile strain (εsm) that can be imposed on a wall were derived from the experimental tests conducted on RC columns with
boundary zone prior to its lateral instability failure. Eqs. (5) and (6) refer pinned end conditions, essentially, column height being equal to the
to the expressions for the stability criterion and the maximum tensile buckled length (Lo). Therefore, the maximum tensile strains obtained
strain, respectively. from these expressions are the average tensile strains over the buckling
( √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ) length and special consideration is required while utilizing these
ξ = 0.5 1 + 2.35m − 5.53m2 + 4.70m (5) analytical models for idealized prisms with different end conditions.
( )2 3.1.3. Experimental observations

b
εsm = 8β ξ (6) Fig. 8 (a-c) present the hysteretic responses of each specimen in
Lo
terms of the applied axial displacements vs. axial force. The average
where m = ρfy/f‘c is the mechanical reinforcement ratio; ρ is boundary axial strains are worked out in a global sense by dividing the applied
zone longitudinal reinforcement ratio; and β is the position of vertical axial displacements by the height of the specimen. The design
reinforcement with respect to the wall thickness (b) (for two layers of compressive strength (Po) of the prism is shown on these plots to
vertical bars, it is assumed as 0.8). compare with the maximum compressive force attained during the test.
7
Fig. 7. Development of OOP buckling (a) Axial force vs applied average axial strain response; (b) OOP displacement vs applied average axial strain response; (c) OOP
deformations at various loading stages; (d) Axial force vs local average axial strain response of Face 1; and (e) Axial force vs local average axial strain response of
Face 2.
Fig. 8 (d-f) depicts the development of OOP displacements (normalized deformation was first observed while unloading from 10εy (1.5%) ten
with respect to specimen thickness) at the mid-height of the prisms in sile strain cycle in all the tested specimens. With every subsequent cycle,
every strain cycle. Stability criterion and upper bound stability limit larger OOP displacements were noted. Buckling failure was defined
proposed by Paulay and Priestley [26] are also shown in these plots. when the compression carrying capacity of the prism dropped by about
During the initial cycles with average strains close to yield strain of 20%. At the point of instability, the OOP displacements for all the
reinforcing bars, horizontal cracks developed around the mid-height specimens approached 60 mm, which is in close to the upper bound
region. Subsequent cycles saw development of cracks distributed along stability limit proposed by Paulay and Priestley [26] as 50% of wall
the height of the prism roughly at the stirrup locations. As the loading thickness.
progressed, the existing cracks became wider while some new inclined Table 5 summarizes different milestones observed for all specimens
hairline cracks emerged between these existing cracks. OOP during their last two loading cycles, whereas, Fig. 9 depicts them
8
0.7
600
300 Asfy 0.6
Normalized OOP Displacement

0 Upper Bound Stability Limit
0.5
Axial Force (kN)
(Paulay & Priestley, 1993)

-300 Asfy 0.4 Initiation of OOP
12BZ-1
displacement
-600
Initiation of OOP displacement 0.3 Global instability
-900
Global instability 0.2 Stability Criterion
-1200 (Paulay & Priestley,
0.1 1993)
-1500
Po = -1700 kN 0
-1800
-2100 -0.1
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
Global Average Axial Strain (mm/mm) Global Average Axial Strain (mm/mm)
(a) (d)
0.7
600
300 Asfy 0.6

Upper Bound Stability Limit
0 0.5
Axial Force (kN)

12BZ-2
displacement
-600
Initiation of OOP displacement 0.3 Global instability
-900
Global instability 0.2 Stability Criterion
-1200 (Paulay & Priestley,
1993)
0.1
-1500
Po = -1756 kN 0
-1800
-2100 -0.1
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
Global Average Axial Strain (mm/mm) Global Average Axial Strain (mm)
(b) (e)
0.7
600
300 Asfy 0.6

0 Upper Bound Stability Limit

0.5
Axial Force (kN)

12BZ-3
displacement
-600
0.3 Global instability
-900 Initiation of OOP displacement
0.2 Stability Criterion
Global instability (Paulay & Priestley,
-1200
1993)
0.1
-1500
0
-1800 Po = -1878 kN
-2100 -0.1
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
(c) (f)
Fig. 8. Hysteresis responses for the 120 mm thick specimens.
recorded. Fig. 10 displays a typical progression of OOP deformation and

Table 5
its almost complete recovery in the penultimate cycle (Points a1 to e),
Average strain levels corresponding to key stages of specimen response during
followed by OOP instability of one of the tested specimens (12BZ-3) in
the last two cycles.
its final cycle (Points a2 to f). The local average strain distribution along
Specimen Loading Penultimate cycle Failure cycle
the height of the specimen at the corresponding loading stages is shown
ID Protocol ID
εmax εOOP- εmin εmax εfail in Fig. 11. Similar observations were noticed for specimens 12BZ-1 and
max 12BZ-2.
12BZ-1 LP#1 0.03 0.0138 − 0.0016 0.035 0.0122 Fig. 10 (a) and Fig. 11 (a) show the distribution of wide horizontal
12BZ-2 LP#2 0.03 0.0131 − 0.001 0.04 0.0167 cracks and local strains respectively along the height of the Specimen
12BZ-3 LP#3 0.024 0.01 − 0.00324 0.03 0.008
12BZ-3 when subjected to an average tensile strain of 2.5%. Upon
unloading, the build-up of OOP deformation led to the development of
graphically. In Table 5, penultimate cycle corresponds to the cycle prior different local strains across the thickness of the specimen. Compressive
to failure until when the specimen had maintained stability as per the strains were developed not only at the mid-height on the West face but
criteria defined by Paulay and Priestley [26], and εOOP-max refers to the also at the top and bottom of East face due to the shape of the OOP
strain when the maximum OOP displacement (stability criterion) was deformation profile imposed by fixed end conditions (as seen in Fig. 10
9
Average Axial Strain (mm /mm) Average Axial Strain (mm /mm)
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
600 60
a1 a2 f
300 30 d
e
OOP Displacement (mm)

0 0
a1 a2
Axial Force (kN)
f
-300 -30
-600 d -60
-900 -90
-1200 -120
-1500 -150
e
-1800 -180
-6 0 6 12 18 24 30 36 42 -6 0 6 12 18 24 30 36 42
Axial Displacements (mm) Axial Displacements (mm)
Fig. 9. Graphical progression of OOP deformation leading to instability.
(a) Point (a1) (b) Point (d) (c) Point (e) (d) Point (a2) (e) Point (f)
WEST EAST
Fig. 10. Visual progression of OOP deformation leading to instability.
1200 1200 1200 1200 1200

East East East East East
Face Face Face Face Face
West West West West West
Face Face Face Face Face
1000 1000 1000 1000 1000
Strain
Strain
Strain
Strain
Strain
800 800 800 800 800

Prism Height (mm)
Prism Height (mm)
Prism Height (mm)

Prism Height (mm)
OOP
Prism Height (mm)
OOP
OOP
OOP
OOP
Strain Strain Strain Strain

Strain
600 600 600 600 600
400 400 400 400 400
200 200 200 200 200
0 0 0 0 0
-0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 -0.005 -0.003 -0.001 0.001 0.003 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Local Axial Strain (mm/mm) Local Axial Strain (mm/mm) Local Axial Strain (mm/mm) Local Axial Strain (mm/mm) Local Axial Strain (mm/mm)
(a) Point (a1) (b) Point (d) (c) Point (e) (d) Point (a2) (e) Point (f)
Fig. 11. Local average strain distribution along the specimen height during the last two cycles.
10
(b) and Fig. 11 (b)). Fig. 10 (c) and Fig. 11 (c) show the recovered state 0.08
At onset of force
of the specimen at the application of peak compressive strain.
drop
In the final cycle, further widening of cracks with similar distribution
Local Compressive Strain (mm/mm)

of local strains along the height was observed at an average tensile strain
of 3.0% (Fig. 10 (d) and Fig. 11 (d)). Upon reversal from this peak tensile
0.06
strain, OOP displacements gradually escalated heading towards lateral
instability. No spalling of cover concrete was spotted in the compressed
regions; instead, core concrete crushing failure was noted at the top
point of contraflexure evidenced by the formation of a diagonal failure
plane as seen in Fig. 10 (e). 0.04
Interestingly, at this stage when the force capacity dropped by 20%
(Point f), although the average unloading strain was +0.8% (still in
Mander et
tensile range), the local compressive strain on the concave side reached al. (1988)
− 4.0% as noted in Fig. 11 (e). The ultimate compressive strain capacity 0.02
of this confined concrete section was about − 2.0% based on the Modified
Approach
analytical model proposed by Mander et al. [20], however, with a slight
variation in the assumption of the effective confined concrete area.
Considering arching action to act in the form of second-degree parabolas
0
between the restrained longitudinal bars only instead of every longitu 12BZ-1 12BZ-2 12BZ-3
dinal bar, as shown in Fig. 12 (a) and (b), resulted in a better prediction.
This observation is in line with the recent findings by Welt et al. [41]. Fig. 13. Ultimate strain capacity predictions for confined concrete in com
Fig. 13 presents the local compressive strains measured for all the parison with experimental observations.
specimens at the onset of force drop as well as at 20% drop in
compression force carrying capacity. Also, the ultimate strain capacities
of confined concrete estimated based on the original Mander et al. [20]
model and the modified approach are plotted in this figure. The ultimate
strain capacity based on the modified approach and the compressive
strains measured at the onset of force drop match fairly well. The dif
ference between the ultimate strain capacities based on the two ap
proaches may seem small for the tested specimens because of their well
confined and code-compliant detailing. However, care would be
required while working with sparsely confined sections.
As mentioned earlier, since the analytical models proposed by Paulay
and Priestley [26] and Chai and Elayer [4] provide theoretical maximum
tensile strains developed over the buckling length, it seems prudent to
consider local tensile strains averaged over the buckled region of the
prism. Hence considering the deformed profile of the prism as depicted
in Fig. 14, buckling length (Lo) was measured between the two points of
contra-flexures, which was about 60% of the prism height (i.e., Lo ≈ 720
mm) based on visual observations of the crack pattern for the three
specimens. This observed buckling length agrees well with the buckling
length expression proposed by Haro et al. [16] as the prism height
divided by the square root of 3 based on their experimental tests on
prisms with fixed end conditions. Note the total gauge length for the
Fig. 14. Buckling length imposed by the fixed end conditions.
linear potentiometers in the central region was 750 mm, and the
consequent small discrepancy in the average axial strain calculation was
ignored.
Fig. 15 not only presents the hysteresis responses for average tensile
strains developed over the buckling length (Lo), but also indicates the
analytical predictions using the three models (Eqs. (6), (7b), and (8)).
Recommended values of 1.0 and 2.0 for design purposes were adopted
for the constants η1 and η2, respectively, in the Chai and Elayer model
(Eq. (7b)). Moreover, to understand the role of unequal strain distribu
tion across the prism thickness in the development of OOP buckling
mechanism, these hysteretic responses are separately plotted for the two
prism faces. It should be noted that the side to which the OOP
displacement occurs is beyond control. Therefore, for the purpose of
consistency, OOP displacement is considered to occur towards East (i.e.,
concavity in the central region occurs towards West).
Fig. 12. Effective confined area as per (a) original Mander et al. [20] and (b)
modified approach.
11
600 600
300 Asfy 300 Asfy
0 0
Axial Force (kN)
Axial Force (kN)

-300 Asfy -300 Asfy
12BZ-1
-600 -600
Chai & Elayer (1999)

Paulay & Priestley
Paulay & Priestley

Haro et al. (2019)
Haro et al. (2019)
-900 -900
(1993)
(1993)
-1200 -1200
-1500 West Face -1500 East Face

(a) (d)
-1800 -1800
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
Average Axial Strain over Lo (mm/mm) Average Axial Strain over Lo (mm/mm)
600 600
Asfy Asfy
300 300
0 0
Axial Force (kN)

Axial Force (kN)
-300 Asfy -300 Asfy
12BZ-2
-600 -600

Paulay & Priestley
Haro et al. (2019)

Paulay & Priestley
Haro et al. (2019)
-900 -900
(1993)
(1993)
-1200 -1200
West Face East Face

-1500 (b) -1500 (e)
-1800 -1800
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
600 600
Asfy Asfy
300 300
0 0
Axial Force (kN)
Axial Force (kN)
-300 Asfy -300 Asfy
12BZ-3
-600 -600

Paulay & Priestley

Paulay & Priestley
Haro et al. (2019)

Haro et al. (2019)
-900 -900
(1993)
(1993)
-1200 -1200
-1500 West Face -1500 East Face

(c) (f)
-1800 -1800
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
Fig. 15. Local hysteresis responses across the prism thickness.
3.1.4. Comparison between the responses of specimens 12BZ-1 and 12BZ-2 case, when the lateral instability occurred, the average tensile strains
Specimen 12BZ-1 was subjected to a conventional loading protocol developed over the buckling length had exceeded the theoretical
LP#1, whereas Specimen 12BZ-2 was tested under a different loading maximum tensile strains predicted by the analytical models.
history LP#2 which was derived from the near fault-type earthquakes. Nonetheless, the comparison of these two specimens provided some
LP#1 had an equal distribution of cycles up to and beyond tensile strain insight into the effect of cycle count along with the impact of slightly
of 10εy as can be seen in Fig. 6 (b). On the other hand, LP#2 had more higher compressive strains on the development of OOP buckling. A
contribution from small magnitude cycles until the tensile strain reached closer look at Fig. 8 (d) suggests that with every subsequent cycle at the
10εy; thereafter, the loading protocol had steeper jumps in its tensile same magnitude slightly higher OOP displacements were observed.
strain magnitudes along with a solitary cycle at each tensile strain level. Moreover, Fig. 15 (a and b) shows that, particularly at the identical
Corresponding compressive strains in the LP#2 loading history were applied average tensile strain cycle of 3.0%, Specimen 12BZ-1 devel
smaller than those in LP#1 as shown in Fig. 6 (a). The difference be oped slightly higher tensile strains as compared to Specimen 12BZ-2.
tween these applied compressive strains reduced as the tensile strain This could also be attributed to the higher compressive strain levels in
magnitude increased. the earlier cycles for the Specimen 12BZ-1. The role of this higher
Despite these differences in the loading protocols, both specimens compressive strains on buckling of idealised boundary elements is
reached the same last stable cycle before buckling instability occurred. further explored in the following section.
Also, the OOP displacements measured were comparable at every stage
of the loading. The failure cycle for the two prisms may seem different 3.1.5. Comparison between the responses of specimens 12BZ-2 and 12BZ-3
since LP#2 lacked the intermediate 3.5% tensile strain cycle. In either Specimens 12BZ-2 and 12BZ-3 were subjected to loading protocols
12
LP#2 and LP#3, respectively. Both loading protocols were derived from experience escalated tensile strains in the subsequent cycles.
the same near-fault ground motion suite, however, the compressive Fig. 17 depicts the OOP deformation profile along the prism height
strains were reduced by a factor of 4 for the LP#2. The effect of this for Specimens 12BZ-2 and 12BZ-3 at various stages of the loading. It
difference in the compressive strains is studied in this section. should be noted that the following observations are limited to the cen
A clear impact of higher compressive strain can be observed in tral deformed region only (i.e., buckling length). It is evident that OOP
Fig. 16 in terms of magnitude of OOP deformation as well as early displacements experienced by the Specimen 12BZ-3 at peak tension
occurrence of OOP instability for Specimen 12BZ-3. Even though OOP loading were much more pronounced compared to the Specimen 12BZ-
displacements were initially visible after unloading from 1.5% average 2. Moreover, concavity on the East side suggests that the West face
tensile strain cycle in both prisms, its magnitude was higher in Specimen stretched more than the East face. However, unloading from this in
12BZ-3 in every subsequent cycle. Subsequently, Specimen 12BZ-3 elastic tension and reloading into compression resulted in concavity on
experienced OOP instability after unloading from 3.0% average tensile the West side because the rebar layer near the West face yielded in
strain cycle, whereas Specimen 12BZ-2 failed in the following cycle. compression prior to the other rebar layer. Similarly, under peak
Reasons behind this early occurrence of instability for Specimen 12BZ-3 compressive strain loading, the West face appeared to carry more
will be revealed when local strains within the buckled region are compression compared to the East face, thereby creating concavity on
considered later. the West side again. Such cumulative compressive stresses experienced
Fig. 15 (b, c, e, f) also displayed the relationship between the theo on the West face caused localised concrete crushing, consequently
retical maximum tensile strains predicted using the analytical models escalating the development of tensile strains in the corresponding rebar
(Eqs. (6)–(8)) and the average tensile strains developed over the buck layer in the subsequent tension cycles.
ling length for the two specimens. A good agreement can be observed Fig. 18 further supports the above explanation by comparing the
between the analytical predictions and the measured strains, regardless local tensile strains averaged over the buckling length for the specimens
of the extent of the applied compressive strain. As the local average 12BZ-2 and 12BZ-3 at the application of various identical global average
tensile strains exceeded the theoretical limits, lateral instability was tensile strains. It is evident from the figure that Specimen 12BZ-3
witnessed upon reloading in compression in both specimens. This not developed higher tensile strains compared to Specimen 12BZ-2 in
only re-affirms the influence of tensile strain on the OOP buckling every loading cycle. This is particularly important as tensile strain
phenomenon but also highlights the indirect impact of higher magnitude and the associated crack width control the possibility of OOP
compressive strain through accelerated development of tensile strains instability. Besides, Fig. 18 also depicts the asymmetric development of
over the buckling length. tensile strains across the boundary element thickness which is another
This escalation in the development of tensile strains under higher important aspect of the OOP buckling mechanism that facilitates OOP
compressive strain loading could be attributed to the path-dependent deformation due to yielding of one rebar layer prior to the other upon
hysteretic properties of steel reinforcement. An analogy could be reloading in compression. This asymmetry in the development of tensile
drawn from the experimental investigations conducted by Moyer and strains across the prism thickness is more apparent for Specimen 12BZ-3
Kowalsky [22] on laterally loaded cantilever RC columns. They than Specimen 12BZ-2. Notably, the West face of Specimen 12BZ-3
observed that the rebar layer near the column face which was com expectedly incurred slightly larger tensile strains in comparison to the
pressed first developed relatively higher tensile strain during the cyclic East face.
lateral loading of equal drifts than the rebar layer placed near the Thus, it could be concluded that higher compressive strains not only
opposite face. They attributed such behaviour to the accumulated widened the gap between the tensile strains across the prism thickness
growth strain following the load reversals, which were primarily caused but also appear to have influenced their magnitudes in general. Both
by residual deformations. Similar logic could be applied to a column factors provide conducive circumstances for accelerating the progress
experiencing OOP deformations under uniaxial cyclic loading, whereby, towards OOP instability.
compressed regions along the deformed profile would potentially
600 0.7
12BZ-2
0.6 12BZ-3
0 0.5
Axial Force (kN)
0.4 Instability
-600 0.3
Instability
0.2 Stability Criterion

12BZ-2 (Paulay &
Priestley,1993)
-1200 12BZ-3 0.1
-1800 -0.1
-0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 -0.005 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
(a) (b)
Fig. 16. Hysteresis responses of 12BZ-2 and 12BZ-3: (a) Axial force vs Global average axial strain; (b) OOP displacements at mid-height vs Global average
axial strain.
13
Peak Tension Partial Compression Peak Compression

1200 1200 1200
Strain
Strain
Strain
1000 1000 1000
OOP
OOP
OOP
Strain
Strain Strain
800 800 800
Prism Height (mm)

Prism Height (mm)
Prism Height (mm)

600 600 600
400 400 400
12BZ-2

12BZ-2

12BZ-2
T = 1.5% (1) cycle T = 1.5% (1) cycle
(Paulay & Priestley,

T = 1.5% (1) cycle
Stability Criterion
200 200 200
T = 2.5% (1) cycle T = 2.5% (1) cycle
T = 2.5% (1) cycle
1993)
T = 3.0% (1) cycle T = 3.0% (1) cycle
T = 3.0% (1) cycle
T = 4.0% (1) cycle
T = 4.0% (1) cycle
0 0 0
-0.03 -0.02 -0.01 0 0.01 0 0.25 0.5 0.75 1 0 0.01 0.02 0.03
Normalized OOP Displacement (ξ=δ/b) Normalized OOP Displacement (ξ=δ/b) Normalized OOP Displacement (ξ=δ/b)
1200 1200 1200
Strain
Strain
Strain
1000 1000 1000
OOP
OOP
OOP
Strain
Strain
Strain
800 800 800
Prism Height (mm)
Prism Height (mm)
Prism Height (mm)
600 600 600
400 400 400

12BZ-3
12BZ-3 12BZ-3
(Paulay & Priestley,
Stability Criterion
T = 1.5% (1) cycle T = 1.5% (1) cycle

200 200 T = 1.5% (1) cycle 200
1993)
T = 2.5% (1) cycle T = 2.5% (1) cycle

T = 2.5% (1) cycle
T = 3.0% (1) cycle T = 3.0% (1) cycle
0 0 0
-0.03 -0.02 -0.01 0 0.01 0 0.25 0.5 0.75 1 0 0.01 0.02 0.03
Normalized OOP Displacement (ξ=δ/b) Normalized OOP Displacement (ξ=δ/b) Normalized OOP Displacement (ξ=δ/b)
WEST EAST
Fig. 17. OOP deformation profile along the prism height at various loading stages.
3.2. Group 2: 200 mm thick specimens - failure due to rebar buckling concrete crushing was observed in these columns.
followed by concrete crushing Inelastic tension loading induces near-horizontal cracks distributed
along the prism height; the average spacing of these cracks was
The cyclic response of Group 2 specimens is presented in Fig. 19 approximately equal to the stirrup spacing. As the tensile strain
using the axial force versus global average axial strain, and the OOP magnitude increased, these cracks got wider. Following unloading from
displacement versus global average axial strain plots. Due to the pres these inelastic tensile strains and reloading into compression, the axial
ence of intrinsic eccentricities in the specimens, development of differ strength was solely provided by the longitudinal reinforcement prior to
ential tensile strains across the prism thickness and thereby initiation of crack closure. Horizontal reinforcement in the form of stirrups per
OOP deformation under compression loading is inevitable. However, formed the function of providing not only confinement to the core
smaller slenderness ratio facilitated early crack closure, thus, controlling concrete but also lateral support to the longitudinal reinforcement.
the OOP deformation well below the stability criterion [7]. Instead, a Degree of lateral support offered by the hoop legs depends on their axial
localised failure mode in the form of rebar buckling followed by core stiffness. Longitudinal rebar buckling need not necessarily occur
14
0.04 between the peak tension and compression strains. It should be noted
12BZ-2 12BZ-3 that the above model does not account for bar buckling over multiple
transverse reinforcement spacing. However, this could be considered
Average Tensile Strain over Lo (mm/mm)
0.035
implicitly by adopting appropriate buckling length corresponding to the
buckling mode. In general, the analytical predictions using Rodriguez
0.03
et al. [30] model appear to be in line with the experimental observa
tions. Moreover, these observations are also somewhat in accordance
0.025 with the postulation made by Kato et al. [19], that a well-detailed cross-
section could potentially experience rebar buckling between two adja
0.02 cent stirrups (i.e. mode-1 buckling).
The specimens in this group were also subjected to three different
uniaxial cyclic loading histories like the thinner specimens. However,
0.015
instead of LP#2 loading protocol, another loading protocol, LP#4,
derived from the far-fault ground motion suite was utilized to evaluate
0.01 the effects of large number of small magnitude inelastic cycles. Although
all three specimens lost their strength after unloading from the 4.0%
0.005 tension cycle regardless of the loading history, the experimental obser
vations provide valuable insights into the effects of variation in cycle
0 content on the axial cyclic response of these specimens.
East West East West East West
3.2.1. Effect of compressive strain on axial cyclic response of 200 mm thick
T = 1.5% T = 2.4% T = 3.0% specimens
The three loading histories adopted for this group differed both in
Fig. 18. Average tensile strains developed over buckling length (Lo) across
terms of cycle count at each strain range and compressive strain
prism thickness at the application of various global average tensile strains (T).
magnitude. The ratio of the peak compressive strain to the peak tensile
strain varied in LP#1 with the magnitude of maximum tensile strain
between the adjacent hoops (i.e., buckling mode-1) and could possibly
applied in each cycle, whereas, this ratio was kept constant throughout
span over multiple hoop spacing (buckling mode-2 or higher) depending
the experiment for LP#3 and LP#4 loading histories, as illustrated in
on the axial stiffness of the horizontal reinforcement and their
Fig. 6 (a). The compressive strain magnitudes for LP#1 were initially
arrangement [10,45]. Also, influence of prior tensile strain on the
comparable to that in LP#3 and LP#4, however, for higher tensile strain
occurrence of longitudinal rebar-buckling is well documented in the
cycles (>10εy), the corresponding compressive strains in LP#1 were
literature. Suda et al. [34] had postulated that cyclic loading instigates
considerably smaller.
the rebar-buckling phenomenon during unloading from tension, while
Specimen 20BZ-1, which was subjected to comparatively lesser
the applied strain is still on tension side whereas the stress is
compressive strains, showed more gradual bar buckling phenomenon
compressive.
than the other two specimens. This is evident from the hysteresis plot in
Formation of vertical cracks along the column edges indicated the
Fig. 19 (a) that shows a gradual change in the target reloading slope
initiation of rebar-buckling. In case of Specimens 20BZ-2 and 20BZ-3,
after the occurrence of rebar buckling. Specimens 20BZ-2 and 20BZ-3
vertical cracks under compressive stress were first observed when
suffered simultaneous buckling of the longitudinal reinforcing bars
unloading from 3.0% average tensile strain, whereas it was observed
along with extensive cover concrete spalling under compressive loading.
during the 3.5% tensile strain cycle in Specimen 20BZ-1. Subsequently,
Although the initiation of rebar buckling was delayed in Specimen 20BZ-
spalling of cover concrete revealed the buckled longitudinal rebar(s)
1, the ultimate failure cycle for all three specimens was the same. This
with varying buckling lengths. The middle unrestrained longitudinal
particularly highlights the dependence of rebar buckling phenomenon
rebars initially buckled over two stirrups spacing (buckling mode-2),
on the tensile strain magnitude under reversed cyclic loading.
whereas rebars at other locations bowed outwards between the adja
Fig. 21 shows the distribution of local tensile strains along the prism
cent hoops (buckling mode-1). Further loading aggravated the condition
height when subjected to a uniform average tensile strain of 4.0%,
of previously buckled longitudinal bars and surrounding concrete.
except for Specimen 20BZ-1, which was uniformly stretched only up to
Finally, the prisms failed due to crushing of core concrete under
3.5% average strain. It could be seen that despite being subjected to
compressive loading due to loss of confinement near the buckled lon
uniform tension, there is localization of strains at certain locations
gitudinal bars. Again, failure here is defined when the compression ca
where spalling of cover concrete and rebar buckling was later observed.
pacity is dropped by about 20%.
Fig. 22 depicts the corresponding states of the specimens after unloading
Table 6 summarizes the different milestones (which are also marked
from average tensile strains of either 3.5% (for 20BZ-1) or 4.0% (for
in Fig. 19) for the three prisms during the experiments with reference to
other specimens), while approaching zero axial displacement and at the
the peak average tensile strain and average strain at the onset of the
peak compressive loading, respectively. While the pre-existing vertical
corresponding milestone. Since strain gauges were not employed in
cracks widened in specimens 20BZ-2 and 20BZ-3 during this loading
these experiments, visual observations and global average strain read
cycle, new vertical cracks formed at the critical locations in Specimen
ings were relied upon to indicate these milestones.
20BZ-1 as the axial displacement approached compressive strain domain
Fig. 20 shows the critical strain values at the onset of rebar buckling
and the condition aggravated at the peak compressive strain leading to
for the three specimens in comparison to the values predicted using an
complete spalling of cover concrete.
analytical model proposed by Rodriguez et al. [29,30]. Rodriguez et al.
It is also interesting to note that for Specimen 20BZ-1 when the cover
[29] investigated the onset of bar buckling in reversed cyclic axial tests
concrete spalling initiated the strain was about 0.0%, which is signifi
performed on bare reinforcing bar coupons and established a relation
cantly less than the commonly adopted spalling strain value of − 0.3%.
ship between bar buckling length and critical strain which was defined
This suggests that cover concrete spalling could be initiated by bar
as the difference between the axial strain at zero stress and the peak
buckling phenomenon, thereby, potentially making the cover concrete
compression strain. More recently, Rodriguez et al. [30] proposed a
ineffective much before the commonly used spalling strain value [9].
modification to the above definition of available critical strain, essen
Nevertheless, considering the complexities involved around the buck
tially considering it equal to the strain range of that cycle i.e. difference
ling behaviour of rebar inside a concrete mass, further detailed
15
600 0.5
Asfy

0 0.4
Asfy
Axial Force (kN)
-600 0.3
Stability Criterion (Paulay & Priestley, 1993)
20BZ-1
-1200 0.2
Vertical crack formation
Spalling of cover concrete
Concrete Crushing failure
-1800 0.1
-2400 0
Po = -2772 kN
-3000 -0.1
-0.01 0 0.01 0.02 0.03 0.04 0.05 -0.01 0 0.01 0.02 0.03 0.04 0.05
(a) (d)
600 0.5
Asfy

0 0.4
Asfy
Axial Force (kN)
-600 0.3
20BZ-2
-1200 Vertical cracks formation 0.2
Concrete crushing failure
-1800 0.1
-2400 0
Po = -2959 kN
-3000 -0.1
-0.01 0 0.01 0.02 0.03 0.04 0.05 -0.01 0 0.01 0.02 0.03 0.04 0.05
(b) (e)
600 0.5
Asfy
0 0.4
Asfy
Axial Force (kN)
-600 0.3
20BZ-3

-1200 0.2
Vertical crack formation
Concrete crushing failure
-1800 0.1
-2400 Po = -2402 kN 0
-3000 -0.1
-0.01 0 0.01 0.02 0.03 0.04 0.05 -0.01 0 0.01 0.02 0.03 0.04 0.05
(c) (f)
Fig. 19. Hysteresis response for the 200 mm thick specimens.
16
Table 6
Average strain levels corresponding to key stages for the 200 mm thick specimens.
Specimen ID Loading Protocol ID Vertical Crack Initiation Bar Buckling First Observed (Spalling) Crushing Failure
εmax ε εmax ε εmax ε

20BZ-1 LP#1 0.035 0.0015 0.035 0.0 0.04 − 0.0018
20BZ-2 LP#3 0.03 − 0.004 0.04 0.01 0.04 − 0.0047
20BZ-3 LP#4 0.03 − 0.003 0.04 0.0 0.04 − 0.0051
Therefore, it is crucial to utilize a low-cycle fatigue model that in

0.06
corporates the effects of rebar buckling to evaluate the damage potential
of each loading history. Hence, in this study, the model proposed by
0.05 Tripathi et al. [40] has been employed that not only incorporates the
Mode 1 Bar buckling (s/db=4.6) effect of buckling but also considers the effect of yield strength of
reinforcing bar on their low-cycle fatigue life.
Critical Strain (εp* )
0.04
The damage potential of a loading history is evaluated using a metric
0.03
called cumulative damage index (CD) which is worked out by accumu
lating the damage caused by each strain cycle, i.e. accumulating the
Mode 2 Bar buckling (s/db=9.2) ratio between the applied number of half cycles (N*f) of a given strain
0.02
amplitude and the required number of half-cycles (Nf) to induce low-
Rodriguez & Iñiguez (2019) cycle fatigue failure at that strain amplitude, as shown in Equation
0.01 (9). The required number of half-cycles (Nf), for a given strain amplitude
(εa) and rebar buckling length, is calculated using Equations (10)–(13)
0.00 formulated by Tripathi et al. [40]. According to this model, the low-
20BZ-1 20BZ-2 20BZ-3 cycle fatigue life coefficients (a) and (β) are a function of the buckling
parameter (λ), which relates the extent of buckling to the slenderness
Fig. 20. Analytical prediction of rebar buckling.
ratio (L/D) of the longitudinal reinforcing bar and the square root of its
yield strength [43,44].
investigation is required to ascertain the contribution of cover concrete
n ( *)
in restricting the outward movement of the longitudinal reinforcement. ∑n ∑ Nf
CD = Di = (9)
i=1 i=1
Nf i
3.2.2. Effect of strain amplitude history on axial cyclic response of 200 mm
( )a
thick specimens εa = β 2Nf (10)
The three loading protocols adopted for this test group had different
combinations of number and amplitudes of the axial displacement (or − λ
β= + 0.2 (11)
average strain) cycles applied, which facilitates investigating the impact 350
of cyclic loading characteristics on the axial cyclic response of idealised
( λ )
boundary zones. This impact is investigated by comparing the low-cycle a= − + 0.441 (12)
fatigue damage caused by each loading history on the respective spec 1200
imen. Various researchers Brown and Kunnath [3], Kashani et al. [18], √̅̅̅̅̅̅̅̅
Tripathi et al. [40] have postulated the unfavourable effects of longi λ=
L fy
(14)
tudinal bar buckling phenomenon on its low-cycle fatigue life. D 100
1200 1200 1200

20BZ-1 20BZ-2 20BZ-3
T = 3.5 % T = 4.0%
1000 1000 T = 4.0 % 1000
800 800 800

Prism Height (mm)
Prism Height (mm)
Prism Height (mm)
600 600 600
400 400 400
200 200 200
0 0 0
0 0.02 0.04 0.06 0 0.02 0.04 0.06 0 0.02 0.04 0.06
Local Axial Strain (mm/mm) Local Axial Strain (mm/mm) Local Axial Strain (mm/mm)
Fig. 21. Local strain distribution along the height of the specimens at the application of uniform global average tension (T).
17
20BZ-1 20BZ-2 20BZ-3
Axial displacement approaching zero

At peak compression
Fig. 22. State of each specimen upon reversal from peak tensile strain of either 3.5% (20BZ-1) or 4.0% (20BZ-2 and 20BZ-3).
(a) Higher mode bar buckling (b) Bar rupture
Fig. 23. Detrimental effects of bar buckling on low-cycle fatigue life of Specimen 20BZ-1.
18
Even though all the tested specimens lost their compression carrying 1.0
capacity after reversal from 4.0% average tensile strain, out of interest, Tripathi et al.(2018)
these specimens were stretched further in the subsequent cycle to induce
5.0% tensile strain. In the process, fracturing of the middle unsupported L/D = 9.2
0.8 λ = 16.19
Cumulative Damage Index (CD)

rebar on the west face of Specimen 20BZ-1 was noted, whereas, speci β = 0.15
mens 20BZ-2 and 20BZ-3 did not show any signs of bar fracture. This a = -0.45
could be evidenced from Fig. 23 (b) as well as from the hysteresis plot in
Fig. 19 (a) which shows a sudden drop in the force capacity while 0.6
extending to 5.0% tensile strain. The fractured rebar had initially
buckled over two stirrup spacings (i.e. mode-2 buckling), however, later
buckled over three stirrup spacings (i.e. mode-3 buckling) at the end of
4.0% tension cycle as revealed in Fig. 23 (a). Similar observations, with 0.4
regards to higher mode buckling, were noted for the other two speci
mens, except for the rebar rupture.
Table 7 demonstrates the estimation of the gradual development of
0.2
cumulative fatigue damage for each of the three loading protocols and
Fig. 24 summarizes them graphically. Linear potentiometer readings in
the vicinity of the buckled rebar were utilized to account for any
localization of tensile strains. Also, based on the observations noted 0.0
above, a slight adjustment was made in the analytical calculations to 20BZ-1 20BZ-2 20BZ-3
account for the higher mode of buckling towards the end of each loading
history, as can be seen in the last two rows of Table 7. As shown in Fig. 24. Cumulative damage index indicating low-cycle fatigue attained by
Fig. 24, the cumulative fatigue damage caused by the conventional each specimen.
loading protocol (LP#1) is about 90%, whereas, the earthquake loading
protocols (LP#3 and LP#4) imposed about 46% and 48% fatigue-related this loading history (LP#1), unlike LP#3 and LP#4. Had full compres
damage, respectively. Consequently, bar rupture due to low-cycle fa sive strains been employed, (according to the analytical model predic
tigue was only observed in Specimen 20BZ-1. This bar fracture could be tion) the cumulative damage index (CD) could have easily exceeded 1.0,
attributed to the presence of more number of large amplitude cycles in thereby, potentially inducing bar rupture in earlier loading cycles.
LP#1, triggering higher cumulative fatigue damage. Moreover, it is Besides highlighting greater damage potential of conventional
important to note that full compressive strains were not incorporated for loading protocol, Fig. 24 also indicates that low-cycle fatigue damage
Table 7
Cumulative low-cycle fatigue damage induced by each loading protocol.
Loading Protocol (Specimen) εt εc Δε = εt − εc εa = Δε/2 N*f (Applied) Nf Required) Di = (N*f/ Nf)i CD = Σ (Di)
LP#1 (20BZ-1) 0.0010 − 0.0003 0.0013 0.0006 4 87161.2 0.00005 0.91

0.0015 − 0.0003 0.0018 0.0009 4 38912.3 0.0001
0.0030 − 0.0005 0.0035 0.0017 4 9529.7 0.0004
0.0055 − 0.0006 0.0061 0.0031 4 2749.4 0.0015
0.0110 − 0.0009 0.0119 0.0060 4 638.8 0.0063
0.0193 − 0.0011 0.0204 0.0102 4 194.8 0.0205
0.0266 − 0.0013 0.0279 0.0139 4 98.4 0.0406
0.0346 − 0.0015 0.0360 0.0180 4 55.9 0.0716
0.0428 − 0.0016 0.0444 0.0222 4 35.3 0.1134
0.0517 − 0.0018 0.0534 0.0267 4 23.5 0.1700
0.0600 − 0.0019 0.0619 0.0309 4 11.4** 0.3524
0.0750 0.0 0.0750 0.0375 1 7.5** 0.1337
LP#3 (20BZ-2) 0.0013 − 0.0002 0.0015 0.0007 6 61044.9 0.0001 0.46
0.0020 − 0.0003 0.0023 0.0011 6 25015.2 0.0002
0.0026 − 0.0004 0.0030 0.0015 6 13283.3 0.0005
0.0040 − 0.0005 0.0045 0.0022 4 5443.3 0.0007
0.0058 − 0.0007 0.0065 0.0033 2 2400.8 0.0008
0.0073 − 0.0009 0.0082 0.0041 2 1469.3 0.0014
0.0123 − 0.0011 0.0134 0.0067 2 494.9 0.0040
0.0180 − 0.0022 0.0202 0.0101 2 200.8 0.0100
0.0384 − 0.0032 0.0416 0.0208 2 40.7 0.0491
0.0446 − 0.0041 0.0487 0.0307 2 28.9 0.0693
0.0560 − 0.0055 0.0614 0.0400 2 11.5** 0.1735
0.0800 0.0 0.0800 0.0400 1 6.5** 0.1538
LP#4 (20BZ-3) 0.0013 − 0.0002 0.0015 0.0008 38 63115.3 0.0006 0.48
0.0019 − 0.0003 0.0022 0.0011 18 25019.6 0.0007
0.0026 − 0.0005 0.0030 0.0015 18 13285.7 0.0014
0.0038 − 0.0007 0.0045 0.0023 10 5444.2 0.0018
0.0062 − 0.0009 0.0071 0.0035 6 1995.8 0.0030
0.0072 − 0.0011 0.0083 0.0042 4 1406.2 0.0028
0.0092 − 0.0014 0.0105 0.0053 8 838.5 0.0095
0.0174 − 0.0027 0.0201 0.0100 4 202.2 0.0198
0.0317 − 0.0041 0.0357 0.0179 2 56.9 0.0351
0.0441 − 0.0053 0.0494 0.0247 2 28.0 0.0715
0.0578 − 0.0070 0.0648 0.0324 2 10.3** 0.1950
0.0750 0.0 0.075 0.0375 1 7.5** 0.1337
**
Half-cycles calculated corresponding to buckling mode 3.
19
caused by far-fault earthquake loading history is only slightly more than fatigue life. However, they substantially affected the lateral
near-fault earthquake loading history. As shown in Fig. 6 (b), these stiffness of the cross-section inducing rapid development of out-
earthquake loading histories had identical number of higher magnitude of-plane deformation. This may become crucial for boundary el
inelastic cycles (>10εy), whereas, the number of small magnitude in ements susceptible to out-of-plane buckling. Further research
elastic cycles were substantially different. Thus, it could be inferred that may be required to further explore this aspect.
there is minimal impact of large number of small amplitude inelastic
cycles on the cumulative damage index and it is dominated by inelastic CRediT authorship contribution statement
tensile strain cycles greater than 1.5% or 10εy. These observations are
fairly consistent with other experimental results on RC bridges columns R.A. Gokhale: Conceptualization, Methodology, Validation, Formal
subjected to lateral cyclic loading reported in literature [1,13,23]. analysis, Investigation, Writing - original draft, Visualization. M. Tri
Nevertheless, Fig. 19 (e-f) highlight the more pronounced rate of pathi: Conceptualization, Methodology, Writing - review & editing,
development of OOP deformation in Specimen 20BZ-3 compared to Supervision, Funding acquisition, Project administration. F. Dashti:
Specimen 20BZ-2, despite having similar low-cycle fatigue damage Conceptualization, Writing - review & editing, Supervision. R.P. Dha
index. The large number of small magnitude inelastic cycles seem to kal: Conceptualization, Methodology, Writing - review & editing, Su
have affected the lateral stiffness of the prism, thereby, almost doubling pervision, Funding acquisition, Project administration.
the OOP displacements in Specimen 20BZ-3. Further research may be
required to explore the effect of such large number of small magnitude
inelastic cycles on boundary elements susceptible to OOP buckling. Declaration of Competing Interest
4. Conclusions The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
A series of six uniaxial cyclic tests were performed on code-compliant the work reported in this paper.
RC prisms simulating boundary zones of flexure-dominated ductile
walls. The main objective of the experimental study was to investigate Acknowledgements
the effect of different strain histories on both global OOP buckling as
well as local bar buckling phenomena. The three different loading his The authors would like to acknowledge the financial support pro
tories adopted in this study were: conventional (measured from quasi- vided by the Ministry of Business, Innovation and Employment (MBIE)
static cyclic test of a prototype wall), near-fault type and far-fault type and the Quake Centre at the University of Canterbury for carrying out
axial strain histories, differing in terms of cycle count and the rela the research. The technical support provided by John Maley in Struc
tionship between their peak tensile and compressive strains. The tural Engineering Laboratory at the University of Canterbury is greatly
earthquake loading histories had larger contribution from small ampli appreciated.
tude cycles compared to the conventional loading protocol. Further
more, major difference between the two earthquake loading protocols References
was in the total number of cycles. While the number of high magnitude
cycles were same in both loading histories, the far-fault loading history [1] Acun B, Sucuoglu H. Performance of reinforced concrete columns designed for
flexure under severe displacement cycles. Struct J 2010;107(3).
contained substantially large number of small magnitude cycles. Be [2] Bazaez R, Dusicka P. Cyclic loading for RC bridge columns considering subduction
sides, specimens were also subjected to reduced compressive loading to megathrust earthquakes. J Bridge Eng 2016;21(5):04016009.
study the impact of compressive strain magnitude on the two failure [3] Brown J, Kunnath SK. Low-cycle fatigue failure of reinforcing steel bars. ACI Mater
J 2004;101(6):457–66.
modes. The key findings from the experimental study are as below: [4] Chai YH, Elayer DT. Lateral stability of reinforced concrete columns under axial
reversed cyclic tension and compression. Struct J 1999;96(5).
(1) Compressive strains caused localised concrete crushing, thereby, [5] Dashti F, Dhakal RP, Pampanin S. Tests on slender ductile structural walls designed
according to New Zealand standard. Bull New Zealand Soc Earthq Eng 2017;50(4).
influencing the development of tensile strains in the subsequent [6] Dashti F, Dhakal RP, Pampanin S. Blind prediction of in-plane and out-of-plane
cycles. Higher compressive strains not only widened the gap be responses for a thin singly reinforced concrete flanged wall specimen. Bull Earthq
tween the tensile strains developed in the two rebar layers across Eng 2018;16(1):427–58.
[7] Dashti F, Dhakal RP, Pampanin S. Validation of a numerical model for prediction of
the prism thickness but also appeared to have influenced the
out-of-plane instability in ductile structural walls under concentric in-plane cyclic
strain magnitudes in general. Both effects provide conducive loading. J Struct Eng 2018;144(6):04018039.
environment for accelerating the progress towards out-of-plane [8] Dashti F, Dhakal RP, Pampanin S. Evolution of out-of-plane deformation and
buckling. subsequent instability in rectangular RC walls under in-plane cyclic loading:
Experimental observation. Earthq Eng Struct Dyn 2018;47(15):2944–64.
(2) When subjected to inelastic cyclic actions, bar buckling can take [9] Dhakal RP. Post-peak response analysis of SFRC columns including spalling and
place regardless of the level of compressive strain reached. Cyclic buckling. Struct Eng Mech 2006;22:311–30.
loading with smaller (or no) compressive strains facilitate [10] Dhakal RP, Maekawa K. Reinforcement stability and fracture of cover concrete in
reinforced concrete members. J Struct Eng 2002;128(10):1253–62.
gradual evolution of bar buckling whereas higher compressive [11] Gokhale RA, Dhakal RP, Tripathi M, Dashti F. Quasi-static uniaxial cyclic loading
strains are likely to cause abrupt development of bar buckling protocol for RC columns representing wall boundary zones. Pac Conf Earthquake
accompanied by substantial cover concrete spalling. Engineering Auckland 2019.
[12] Gokhale RA, Tripathi M, Dashti F, Dhakal RP. Lateral stability limits for RC wall
(3) The low-cycle fatigue damage caused by conventional loading boundary zones based on axial response of idealized prisms. J Earthq Eng 2021.
protocol is much more severe than that caused by the earthquake https://doi.org/10.1080/13632469.2021.1884148 [in press].
loading histories, thus, asserting greater dominance of high [13] Goodnight JC, Kowalsky MJ, Nau JM. Effect of load history on performance limit
states of circular bridge columns. J Bridge Eng 2013;18(12):1383–96.
magnitude inelastic cycles on the low-cycle fatigue life over the [14] Goodsir WJ. The design of coupled frame-wall structures for seismic actions.
large number of small magnitude inelastic cycles. For instance, in University of Canterbury; 1985, http://hdl.handle.net/10092/7751.
the tested prism specimens, the cumulative damage was [15] Haro AG, Kowalsky M, Chai YH. Assessment of the boundary region stability of
special RC walls. Springer Singapore, 223–242.
approximately 90%, 46% and 48% in prisms subjected to con
[16] Haro AG, Kowalsky M, Chai YH, Lucier GW. Boundary elements of special
ventional, near-fault and far-fault earthquake loading, reinforced concrete walls tested under different loading paths. Earthq Spectra
respectively. 2018;34(3):1267–88.
(4) Similar cumulative damage indices for the near-fault and far-fault [17] Hilson CW, Segura CL, Wallace JW. Experimental studies of longitudinal
reinforcement buckling in reinforced concrete structural wall boundary elements.
earthquake loading histories suggests minimal impact of large In: Proc., 10th National Conference in Earthquake Engineering, Earthquake
number of small magnitude inelastic cycles on the low-cycle Engineering Research Institute.
20
[18] Kashani MM, Salami MR, Goda K, Alexander NA. Non-linear flexural behaviour of [33] Segura CL, Wallace JW, Arteta CA, Moehle JP. Deformation capacity of thin
RC columns including bar buckling and fatigue degradation. Mag Concr Res 2018; reinforced concrete shear walls. In: New Zealand Society for Earthquake
70(5):231–47. Engineering (NZSEE) Annual Technical Conference, Christchurch; 2016.
[19] Kato D, Kanaya J, Wakatsuki K. Buckling strains of main bars in reinforced [34] Suda K, Murayama Y, Ichinomiya T, Shimbo H. Buckling behaviour of longitudinal
concrete members. In: Proc., 5th East Asia and Pacific Conf. in Structural reinforcing bars in concrete column subjected to reverse lateral loading. In:
Engineering and Construction EASEC-5, 699-704. Eleventh World Conference on Earthquake Engineering, Elsevier Science Ltd; 1996.
[20] Mander JB, Priestley MJN, Park R. Theoretical Stress-Strain Model for Confined [35] Taleb R, Tani M, Kono S. Performance of Confined Boundary Regions of RC Walls
Concrete. J Struct Eng 1988;114(8):1804–26. under Cyclic Reversal Loadings. J Adv Concr Technol 2016;14(4):108–24.
[21] Massone L, Polanco P, Herrera P. Experimental and analytical response of RC wall [36] Tripathi M. Bar buckling in ductile reinforced concrete walls: causes, consequences
boundary elements. In: 10th U.S. National Conference on Earthquake Engineering: and control. Doctor of Philosophy, University of Canterbury; 2020.
Frontiers of Earthquake Engineering, Anchorage, Alaska; 2014. [37] Tripathi M, Dhakal RP, Dashti F. Bar buckling in ductile RC walls with different
[22] Moyer MJ, Kowalsky MJ. Influence of Tension Strain on Buckling of Reinforcement boundary zone detailing: Experimental investigation. Eng Struct 2019;198:109544.
in Concrete Columns. Struct J 2003;100(1). [38] Tripathi M, Dhakal RP, Dashti F. Nonlinear cyclic behaviour of high-strength
[23] Nojavan, A. Performance of full-scale reinforced concrete columns subjected to ductile RC walls: Experimental and numerical investigations. Eng Struct 2020;222:
extreme earthquake loading. PhD, Civil Engineering, University of Minnesota; 111116.
2015. [39] Tripathi M, Dhakal RP, Dashti F, Gokhale R. Axial response of rectangular RC
[24] NZS3101:2006 “Concrete Structures Standard.” Standards New Zealand. prisms representing the boundary elements of ductile concrete walls. Bull Earthq
[25] Oesterle RG, Fiorato AE, Johal LS, Carpenter JE, Russell HG, Corley WG. Eng 2020;18(2020):4387–420.
Earthquake resistant structural walls-tests of isolated walls. Research and [40] Tripathi M, Dhakal RP, Dashti F, Massone LM. Low-cycle fatigue behaviour of
Development Construction Technology Laboratories, Portland Cement Association, reinforcing bars including the effect of inelastic buckling. Constr Build Mater 2018;
Skokie, Illinois; 1976. 190(2018):1226–35.
[26] Paulay T, Priestley MJN. Stability of ductile structural walls. Struct J 1993;90(4). [41] Welt T, Lehman D, Lowes L, LaFave J. A constitutive model for confined concrete in
[27] Peng BH, Dhakal RP, Fenwick RC, Carr AJ, Bull DK. Elongation of plastic hinges in slender rectangular RC sections incorporating compressive energy. Constr Build
ductile RC members: Model development. J Adv Concr Technol 2011;9(3):315–26. Mater 2018;193:344–62.
[28] Peng BH, Dhakal RP, Fenwick RC, Carr AJ, Bull DK. Elongation of plastic hinges in [42] Welt TS, Massone LM, LaFave JM, Lehman DE, McCabe SL, Polanco P. Confinement
ductile RC members: Model verification. J Adv Concr Technol 2011;9(3):327–38. behavior of rectangular reinforced concrete prisms simulating wall boundary
[29] Rodriguez ME, Botero JC, Villa J. Cyclic stress-strain behavior of reinforcing steel elements. J Struct Eng 2017;143(4):04016204.
including effect of buckling. J Struct Eng 1999;125(6):605–12. [43] Dhakal RP, Maekawa K. Modeling for Postyield Buckling of Reinforcement. J Struct
[30] Rodriguez ME, Iñiguez M. Drift capacity at onset of bar buckling in RC members Eng 2002;128(9):1139–47. https://doi.org/10.1061/(ASCE)0733-9445(2002)128:
subjected to earthquakes. Springer Singapore; 2017, 185–200. 9(1139).
[31] Rosso A, Almeida JP, Beyer K. Stability of thin reinforced concrete walls under [44] Dhakal RP, Maekawa K. Path-dependent cyclic stress–strain relationship of
cyclic loads. Bull Earthq Eng 2015;14(2):455–84. reinforcing bar including buckling. Eng Struct 2002;24:1383–96. https://doi.org/
[32] Rosso A, Jiménez-Roa LA, Almeida JPd, Zuniga APG, Blandón CA, Bonett RL, et al. 10.1016/S0141-0296(02)00080-9.
Cyclic tensile-compressive tests on thin concrete boundary elements with a single [45] Tripathi M, Dhakal RP. Designing and detailing transverse reinforcement to control
layer of reinforcement prone to out-of-plane instability. Bull Earthq Eng 2017; 16 buckling-induced failure in rectangular RC walls. Bull of New Zealand Soc for
(2): 859–887. Earthq Eng 2021.
21

Gokhale - Effect of Loading History On Axial Cyclic Behaviour of RC Prisms Representing Wall Boundary Zones PDF

Uploaded by

Copyright:

Available Formats

Gokhale - Effect of Loading History On Axial Cyclic Behaviour of RC Prisms Representing Wall Boundary Zones PDF

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

Gokhale - Effect of Loading History On Axial Cyclic Behaviour of RC Prisms Representing Wall Boundary Zones PDF

Uploaded by

Copyright:

Available Formats

Engineering Structures 233 (2021) 111871

Contents lists available at ScienceDirect

Effect of loading history on axial cyclic behaviour of RC prisms representing

1. Introduction The review of previous experimental studies on rectangular RC

12BZ-1 120 330 1200 2.86 4.6 10.0

Fig. 2. Test setup.

Fig. 5. Development of loading protocol.

0.25 100% > 10εy ≤ 10εy

( )2 3.1.3. Experimental observations

300 Asfy 0.6

Normalized OOP Displacement

(Paulay & Priestley, 1993)

300 Asfy 0.6

Normalized OOP Displacement

(Paulay & Priestley, 1993)

300 Asfy 0.6

0 Upper Bound Stability Limit

(Paulay & Priestley, 1993)

recorded. Fig. 10 displays a typical progression of OOP deformation and

OOP Displacement (mm)

Fig. 9. Graphical progression of OOP deformation leading to instability.

Fig. 10. Visual progression of OOP deformation leading to instability.

1200 1200 1200 1200 1200

800 800 800 800 800

Prism Height (mm)

Prism Height (mm)

Strain Strain Strain Strain

600 600 600 600 600

400 400 400 400 400

200 200 200 200 200

Local Compressive Strain (mm/mm)

Fig. 14. Buckling length imposed by the fixed end conditions.

300 Asfy 300 Asfy

Axial Force (kN)

Chai & Elayer (1999)

Paulay & Priestley

Paulay & Priestley

-1500 West Face -1500 East Face

Axial Force (kN)

-300 Asfy -300 Asfy

Chai & Elayer (1999)

Haro et al. (2019)

West Face East Face

Axial Force (kN)

-300 Asfy -300 Asfy

Chai & Elayer (1999)

Paulay & Priestley

Haro et al. (2019)

-1500 West Face -1500 East Face

Fig. 15. Local hysteresis responses across the prism thickness.

0.2 Stability Criterion

Peak Tension Partial Compression Peak Compression

Prism Height (mm)

Prism Height (mm)

400 400 400

Upper Bound Stability Limit

(Paulay & Priestley, 1993)

(Paulay & Priestley,

1200 1200 1200

Therefore, it is crucial to utilize a low-cycle fatigue model that in