Received: 9 March 2017 Revised: 31 October 2018 Accepted: 13 November 2018

DOI: 10.1002/tal.1578

RESEARCH ARTICLE

Investigating the vulnerability of nonductile reinforced

concrete columns in moderate seismic regions to gravity load
collapse
Tahir Mehmood1 | Kittipoom Rodsin2 | Pennung Warnitchai3 | Kristijan Kolozvari4

1
Department of Civil Engineering, COMSATS
University Islamabad, Wah Campus, Wah SUMMARY
Cantt, Pakistan Reinforced concrete (RC) columns are important components of the lateral load‐
2
Center of Excellence in Structural Dynamics
resisting structural system of RC buildings. In moderate seismic regions, where strin-
and Urban Management Department of Civil
and Environmental Engineering King gent seismic reinforcement detailing requirements are usually not considered, RC col-
Mongkut's University of Technology North
umns are categorized as nonductile. Postearthquake studies have shown that gravity
Bangkok, Thailand
3
Department of Structural Engineering, Asian
load collapse of RC columns can trigger the progressive collapse of RC buildings. The
Institute of Technology, Bangkok, Thailand current study presents the seismic response behavior of nonductile RC columns in
4
Department of Civil and Environmental moderate seismic regions, with a particular focus on gravity load collapse. Six RC col-
Engineering, California State University,
Fullerton, Fullerton, USA umns, three with lap splices and three without lap splices with variable aspect ratios,
Correspondence were tested under reversed cyclic loading. Experimental results show that the column
Tahir Mehmood, Department of Civil
failing in shear could tolerate the maximum drift in order of 2.7–3.5%, whereas the
Engineering, COMSATS University Islamabad,
Wah Campus , Wah Cantt, Pakistan. columns failed in flexural mode could achieve the maximum drift of 4.5%. For the col-
Email: [email protected]
umns with lap splice, the lateral strength was significantly degraded, but all spliced
columns could sustain gravity load even when their displacements were more than
4% drift. This very large drift without axial collapse, observed in the current study,
is associated with the splice slip causing the large rotation just above the splice region.

KEY W ORDS

axial load, columns, gravity load collapse, lap splices, nonductile

1 | I N T RO D U CT I O N

Reinforced concrete (RC) columns are primary part of the moment‐resistant frame buildings to resist lateral loading. RC columns may also be part
of the lateral load‐resisting structural systems where RC shear walls are used primarily as lateral load‐resisting components. RC columns require to
support gravity loads as well; therefore, they should be able to sustain the gravity load safely without compromising lateral strength under lateral
drifts due to seismic action. Post‐earthquake studies of the collapsed building showed that one of the primary cause of the collapse is the loss of
the gravity load carrying capacity of RC columns.[1] During the past few decades, our understanding of seismic hazard and seismic resistant design
of buildings has been greatly improved. Many existing buildings designed in compliance with relevant standards during their construction time may
not satisfy the requirements of the present‐day standards. RC columns in such buildings historically showed poor seismic performance due to
insufficient transverse reinforcement ratios, inadequate lap‐spliced length and poor choice of lap‐spliced regions[2,3] (Kam et al., 2011).[4] Such
RC columns deficient in seismic detailing, designated as nonductile.
Poor performance of lightly RC columns leads to the investigation of such RC columns to predict the performance of existing buildings. Sev-
eral factors were identified to control the seismic response of nonductile RC columns. It was observed that increasing axial load ratio results in the
reduction of lateral drift capacity.[5–7] Experimental studies observed that contribution of transverse reinforcement to the shear strength in such
nonductile RC column decreases as the lateral displacement increases,[7] whereas increase in transverse reinforced was found to increase lateral

Struct Design Tall Spec Build. 2018;e1578. wileyonlinelibrary.com/journal/tal © 2018 John Wiley & Sons, Ltd. 1 of 14
https://doi.org/10.1002/tal.1578
2 of 14 MEHMOOD ET AL.

drift capacity.[8,9] The decrease in the shear strength with increase in lateral displacement in widely spaced transverse reinforced columns are pri-
marily due to decrease in effective section depth due to poor bond of flexural reinforcement and concrete and presence of 90‐degree hooks fur-
ther aggravate this process. Nonseismic detailing of longitudinal and transverse reinforcement could lead to buckling of longitudinal reinforcement
and thus limiting the lateral strength and drift capacity of nonductile columns.[10,11]
A recent survey of existing RC buildings in Thailand by Suesuttajit[12] revealed that most columns are nonductile as the seismic design provi-
sions are not usually regulated in this region. Hence, detailing of most columns is generally poor. The significant detailing deficiencies typically
found in such columns are the use of widely spaced stirrups and the use of lap splices in the potential plastic hinge area. When the column is sub-
ject to seismic lateral forces, insufficient number of stirrups could not prevent concrete from crushing and the rebars from buckling, which can lead
to brittle shear failure in the plastic hinge region. Moreover, the cyclic behavior of the column may be aggravated by poor splice detailing. Lap
splices provided in these old RC buildings were usually designed as compression splices. Short splice lengths of 20–25 diameter of longitudinal
bar (db) are commonly found. In absence of proper development length, splice slip may occur leading to column failure in the splice region.
To perform satisfactory, RC columns should carry the gravity load safely under lateral deformation without gravity load collapse. Therefore, it
is important to assess the drift capacity of existing RC columns up to gravity load collapse. Lightly RC columns are prevalent in many old buildings
and common in current detailing practice in the regions of lower seismicity. The lateral drift capacity of RC column is conventionally considered
when the maximum lateral load resisting capacity is dropped to 80% of the design strength. In the regions of low to moderate seismicity such as
Thailand, most columns are nonductile and lateral strength could significantly drop below 80% soon after ultimate lateral load capacity has been
reached. These columns are deemed collapse although they can still safely carry their gravity load. In such case, the definition for the drift capacity
could be overly conservative. These types of structures are believed to have a very low lateral load and drift capacity from a conventional design
perspective. However, many post‐earthquake investigation[13,14] show that the primary cause of RC building collapse during earthquakes is the
loss of vertical‐load‐carrying capacity in critical building components leading to cascading vertical collapse, rather than loss of lateral‐load
capacity.[15]
For example, an existing building contains an overall weak lateral‐force resisting system that is susceptible to shear (point a) and a column that
is susceptible to axial failure (point b), as can be seen in Figure 1.[16] Although if the structure is upgraded by adding new shear walls and hence
stiffened and strengthened by the new wall, the building remains vulnerable to loss of vertical‐load‐carrying capacity if drifts are not controlled to
be less than the drift at axial load failure. Therefore, investigation and laboratory study of nonductile columns is needed to examine the primary
parameters that contribute to loss of column axial‐load capacity.
Analytical modeling of nonlinear behavior of RC columns is typically conducted using simplified lumped‐plasticity models (e.g., SAP[17] CSI) or
fiber‐based models with uncoupled axial/flexural and shear behavior (so‐called uncoupled models;e.g., in the study of Taucer, Spacone, and
Filippou[18]), which showed to be reliable analytical tools for predicting the flexure‐dominated responses of RC columns. However, when column
behavior is characterized with shear‐flexure interaction, shear‐dominant responses, or inadequate lap‐splice, lumped‐plasticity models, and
uncoupled fiber‐based models tend to considerably overestimate lateral force and/or deformation capacity and cannot predict reliably column

FIGURE 1 Illustration of strength and deformation demands for existing buildings

MEHMOOD ET AL. 3 of 14

hysteretic behavior.[19] Recent developments in nonlinear modeling approaches for RC components made available analytical models that can cap-
ture shear‐flexural interaction and nonlinear shear deformations providing the opportunity to simulate the behavior of columns in which shear
deformations are significant (e.g., the studies of Xu and Zhang, Jiang and Kurama, and Fischinger and Rejec[20–22]; Mehmood Hussain and
Warnitchai, 2015)[23]; the study of Kolozvari, Orakcal, and Wallace[24]). In addition, modeling approaches have been developed that can capture
bond‐slip deformations enabling simulation of splice‐related failure in RC components (e.g., study Chowdhury and Orakcal[25]). Therefore, recent
innovations in computational tools provide the opportunity for simulation of a wide range of physical and behavioral properties related to seismic
behavior of RC columns, enabling improved predictions of seismic behavior of RC buildings.
The current research study aims at investigating drift capacity and failure mechanism of nonductile columns at gravity load collapse condition.
The proposed research project involved reversed cyclic load test of RC column specimens representative of typical constructional practice in Thai-
land. Six RC columns 250 mm × 350 mm in cross section with a height of 2,050, 1,570, and 1,100 mm were tested under reversed cyclic loading.
Lap splices with a splice length of 25 bar diameter (25 db) were used in the plastic hinge of three columns. The longitudinal reinforcement ratio of
2.76% and transverse reinforcement consisting of 6‐mm diameter bars spaced at 200 mm were used for all specimens. The cantilever columns will
be subjected to 14–16% axial load level before being displaced horizontally using a displacement‐controlled actuator attached at the free end via a
pin connection. The columns will be displaced in both directions cyclically with gradual increments in displacement for every two cycles until the
column could not carry its gravity load. Finally, the experimentally measured behavior of selected RC column specimens will be simulated using
recently developed nonlinear model for RC components that captures nonlinear axial‐flexural‐shear interaction.[19,24]

2 | SIGNIFICANCE OF RESEARCH

A recent earthquake of magnitude 6.3 struck the northern areas of Thailand on May, 2014 causing significant damage to the existing structure.
This event raised serious concerns about the capital city of Thailand, Bangkok, where mid‐rise to high‐rise RC structures are in abundance.
Due to the absence of nearby seismic sources, the city has long been considered as being free from seismic risk, and most existing buildings in
Bangkok have been designed and constructed without any consideration on seismic loading. Recent seismic hazard studies,[26,27] however, indi-
cate that the city is still at risk from damaging ground motions. The first primary factor causing this risk is the presence of several distant seismic
sources capable of producing large magnitude earthquakes. These sources are active crustal faults in Western Thailand and Myanmar and the
megathrust Sunda subduction zone in the Andaman Sea. The second factor is the ability of thick soil deposits in the Bangkok basin to greatly
amplify earthquake ground motions. The third factor is the poor detailing of lateral load‐resisting components. It is only recently (2004) that
the seismic design requirements are included in the building code of Thailand. Prior to the inclusion of seismic design requirements, RC structures
in Thailand were usually designed without the consideration of seismic loading. This typical design practice leads to poorly detailed RC columns.
With all these factors combined, the seismic risk of Bangkok is comprehensible. It is thus becoming imperative to assess seismic performance of
RC columns in Bangkok.

3 | EXPERIMENTAL PROGRAM

An experimental program has been developed to investigate the reversed cyclic behavior particularly limiting drift at axial load collapse of repre-
sentative nonductile columns in Thailand. The column parameters such as axial load ratio, transverse reinforcement ratio, longitudinal reinforce-
ment ratio, and size of the column are kept constant. The variable parameters are shear span‐to‐depth ratio and splice detailing. A detailed
study was conducted by Suesuttajit[12] to document the structural indexes of the existing RC buildings in Thailand. In this study, 83 RC buildings,
ranging from the height of two floors to more than eight floors, were investigated to identify the key structural indexes of beams, walls, joints, and
columns. The axial load ratio, typical dimensions of the structural members, horizontal and vertical reinforcement of the structural members were
the key structural indexes documented in this study. The cross section dimension and axial load ratio used in this study represent the typical con-
struction practices of more than eight story existing RC buildings in Thailand. The properties of test specimens are shown in Table 1. Cross‐

TABLE 1 Properties of the tested specimens

Spec. Dimension (mm) a (mm) AR ρV ρH Ties (@mm) n f c’ (MPa)

S1L 250 × 350 2050 5.85 2.76% 0.22% R6@200 0.15 30

S1Ls 250 × 350 2050 5.85 2.76% 0.22% R6@200 0.14 32
S2M 250 × 350 1570 4.5 2.76% 0.22% R6@200 0.14 33
S2Ms 250 × 350 1570 4.5 2.76% 0.22% R6@200 0.14 32
S3S 250 × 350 1100 3.14 2.76% 0.22% R6@200 0.16 28
S3Ss 250 × 350 1100 3.14 2.76% 0.22% R6@200 0.15 30

Note. a is shear span that is a clear height of a column in this case, AR is the shear span‐to‐depth ratio defined as shear span divided by the depth, n is the
axial load ratio (ratio of the axial load to axial load‐carrying capacity or Ac f c’). ρV is the longitudinal reinforcement ratio. ρH is the lateral reinforcement (Ash/
bs). Ash = total area of transverse reinforcement; s = tie spacing; and b = column section width
4 of 14 MEHMOOD ET AL.

sectional properties of the column specimens are shown in Figure 2. The experimental set up as shown in Figure 3 was used to simulate gravity
and lateral earthquake loads. Specially designed gravity load frame was used to maintain a constant axial load throughout the test. The unique
feature of this test frame is that the built‐in rollers at the top of the frame enable the axial load to move along with the tip of the column. The
axial load that could follow the column deformation simulates the real conditions of columns under earthquake. Consequently, the point at which
the gravity load collapse occurs could be accurately monitored using this test setup. The constant axial load of 400 kN (axial load ratio about 0.14–
0.16) was applied to a specimen using a 600 kN‐jack. The axial load was monitored and kept constant throughout the test. The lateral load was
applied using a 250 kN‐hydraulic actuator. Specimens were subjected to quasi‐static loading history as suggested by Priestley and Park[28] but
using a target percent drift for convenience. A specimen was subjected to displacement‐controlled testing to a percent drift of ±0.25, ±0.5,
±0.75, ±1.0, ±1.5, ±2.0, ±2.5, ±3.0, and etc. until the column loss of its gravity load carrying capacity as shown in Figure 4. Two cycles of loading
were used with each drift ratio to ensure that the hysteretic behavior could be maintained. The test was terminated only when the column had lost
its axial load carrying capacity.
Strain gages were used to measure the strain of the longitudinal bars and stirrups. Linear variable differential transducers (LVDTs) were used
to measure flexural, shear, and yield penetration deformation. Different instruments were used to monitor the lateral load, axial load, strain of lon-
gitudinal, and transverse reinforcement, flexure, shear, and slip deformation of the specimen. The arrangement of strain gauges for specimens is
shown in Figures 5 and 6 for specimen with lap splices and without lap splices, respectively. Twenty‐four strain gauges were used for longitudinal
reinforcement, and four strain gauges were used for transverse reinforcement for the specimens with lap splices. For specimen without lap splices,
16 strain gauges were used for longitudinal reinforcement, and four strain gauges were used for transverse reinforcement. Deformations of the
specimen were measured by LVDTs of 50‐ and 100‐mm measuring capacity. The arrangement of LVDTs for measurement of deformation com-
ponents for detailed analysis is illustrated in Figure 7. Data logger equipped with a personal computer was used to record the data at each point
during application of lateral drifts. All sensors were connected to a data logger to record data from the test.

4 | EXPERIMENTAL OBSERVATIONS

4.1 | Progression of damage

Figure 8a shows the cyclic force‐deformation behavior for the specimen S1L, whereas the final stage of cracking is shown in Figure 9a. Visible
flexural crack started at 0.5% lateral drift at about 150, 450, and 550 mm above the pedestal level. With increasing the lateral drift number of
flexural cracks went on increasing. At 1.5% lateral drift, cover concrete started to spall at about 150 mm where the maximum moment occurred.
This spalling of cover concrete was found to be increased as lateral drift increased. At 2.0% lateral drift, peak lateral strength was found to be
reached. At 2% lateral drifts, cracks along the corner reinforcement started to appear, these cracks were found to be increased in length and width
as the strain in this region went on increasing with increasing lateral drift. At 3.5%, buckling of the longitudinal reinforcement started (refer to
Figure 10), and 20% drop in peak lateral strength reached. The specimen was not able to take the gravity load at 4.5% lateral drift. Half cycle
of 5% drift was applied to reach the total collapse of the column.
The specimen S2M was tested with an aspect ratio of 4.50. Hysteretic force deformation relationship and final cracked stage of the specimen
S2M is presented in Figures 8b and 9b, respectively. During testing of specimen S2M, initial flexural cracking was observed at 0.50% lateral drift
ratios, at locations of 250 and 350 mm from above the pedestal level. These cracks were found to be increased in number as the lateral drift
increased. At 2.5% lateral drift, the peak lateral strength was attained. At 2.5% lateral drift, cover concrete started to spall. At lateral drift 3.5%,
a shear crack was observed, which was later increased in width and length and caused the yielding of transverse reinforcement. Lateral strength
dropped to 20% of peak lateral strength at 3.5% drift. At 4% lateral drift, longitudinal steel was visible with significant buckling. Due to excessive
cracking at 4% lateral drift, column was not able to carry out any further gravity load, so the test was stopped at that stage.
Next, the specimen S3S was tested, which was a short RC column without lap splices having a span ratio of 3.15. Due to short shear span, this
specimen was expected to be failed in shear mode of failure. Figure 8c shows the lateral force deformation relationship, whereas Figure 9c pre-
sents the final cracked stage of the specimen S3S Initial cracking was observed at 0.50% lateral drift at 150‐ and 200‐mm height from the pedestal.

FIGURE 2 Cross‐sectional properties of the

column specimens (all dimensions in mm)
MEHMOOD ET AL. 5 of 14

FIGURE 3 Test rig and experimental setup

FIGURE 4 Loading protocol

FIGURE 5 Strain gauge locations specimen

with splices

At 1.5%, specimen reached its peak lateral strength. At this stage (1.5% lateral drift), an inclined shear crack was observed. At 2% lateral drift,
another large shear crack appears on the other side. 20% of the peak lateral strength dropped at 2% drift. With increasing drift beyond 2%, it
was observed that shear plan penetrating downward and increase in axial deformation was observed. The shear crack was found to be increased
in width as the lateral drift increased. After development of shear plan, upper part of column penetrates downward under the action of gravity
load and impose large compressive forces on longitudinal reinforcement as shown in Figure 11. At 2.5% drift, buckling of longitudinal
6 of 14 MEHMOOD ET AL.

FIGURE 6 Strain gauge locations specimen

without lap splices

FIGURE 7 Arrangement of linear variable differential transducers

reinforcement was observed at an around 500‐mm height from the pedestal as shown in photograph of Figure 11 taken after removal of debris. At
3.5% lateral drift, ultimate gravity load collapse occurred due to excessive core crushing.
Longitudinal reinforcements of the specimen S1Ls were provided with lap splices and were “cranked” at a location just above the lap splice as
shown in Figure 5; it is common practice of construction in Thailand. For specimen S1Ls height was selected same as for specimen S1L but lon-
gitudinal reinforcement provided with Lap splices. Figure 8d shows the lateral force deformation relationship, whereas Figure 9d presents the final
cracked stage of the specimen S1Ls. Initial flexural cracking was observed at 0.50% lateral drifts at 250, 350, and 400 mm. At 1.5% lateral drifts,
cover concrete start spalling. Peak lateral strength has been reached at 1.5% lateral drift without yielding the longitudinal rebar. At this stage,
cracks appeared along the lap splice, which indicated the deterioration of the bond. It was observed that the rate of lateral strength degradation
was very high after 1.5%. At 2% lateral drift, 20% of peak lateral strength was dropped. In the second cycle of 2%, lateral drift vertical cracks along
the splice length were clearly visible. The cracks along the splice length increased in width and length as the lateral drift increased. Longitudinal
reinforcement yields at 3% drift. At 4.5% lateral drift, buckling of the longitudinal rebars at just above the lap splices (i.e., 500 mm) was visible,
which ultimately lead to the collapse of the specimen at 5% lateral drift.
Next specimen to be tested was S2Ms with the same properties as S2M but provided with lap splices. Lap splice details are same as in case of
S1Ls. Figure 8e shows the lateral force deformation relationship, whereas Figure 9e presents the final cracked stage of the specimen S2Ms. Initial
MEHMOOD ET AL. 7 of 14

FIGURE 8 Hysteretic behavior of column specimens under cyclic loading

flexural cracking was observed at 0.5% lateral drift ratios, at a location of 450 mm from above the pedestal level. Longitudinal cracking along the
splice length began in the second cycle of 1% lateral drift. For 1.5% and subsequent higher drift ratios the, lateral strength started to decrease,
indicating bond deterioration along the splice length. The length and width of these cracks was found to be increasing as the drift ratio increased.
At 2.5% lateral drift, peak lateral strength of the specimen was reached. After 2.5% lateral drift substantially increase in number of cracks was
observed. At 2.5% drift, 20% drop in peak lateral strength was observed. Longitudinal bars started to buckle at 4% lateral drift at the location
of cranked portion in the longitudinal reinforcement, which ultimately leads to the collapse of the column at 5% lateral drift.
S3Ss was the last specimen to be tested provided with lap splices. Strength degradation was more prominent in testing of S3Ss than S3S due
to presence of lap splices. Figure 8f presents the lateral force deformation relationship, and Figure 9f presents the final cracked stage of the spec-
imen S3Ss. Initial cracking was observed at 0.25% lateral drift at 450 mm height from the pedestal. Transverse reinforcement yielded at 1.5% drift
with development of shear cracks. At 1.5% lateral drift, these shear cracks were found to be increased in number and size. Specimen reached its
peak lateral strength at 1.5% lateral drift without yielding of longitudinal reinforcement. At this stage (1.5% drift) cracks along the lap splices were
visible. At 2% lateral drift width of cracks along, the splice length increased and cracks became open, which accelerate the rate of lateral strength
degradation. At 2.4% lateral drift, 20% in peak lateral strength was observed. At 3% lateral drift, excessive core cracking and buckling of longitu-
dinal reinforcement was observed at around 500‐mm height from the pedestal, which ultimately resulted in the gravity load collapse at 4% lateral
drift. Again, like all other lap splice specimen buckling occurs in the cranked portion of the longitudinal reinforcement. Although, both specimen
S3S and S3Ss are short RC specimens and dominated by the brittle shear mode of failure, but the specimen S3Ss suffered more damage than
the specimen S3S due to the lap splices. The lateral force‐deformation behavior of the specimen S3Ss showed significant degradation without
any ductility. Initial cracking started earlier than the specimen S3S, and the cracks along the splice length rapidly mixed with the shear cracks which
finally lead to the gravity load collapse of the specimen. Such short RC column with lap splices should be avoided in seismic prone regions.

4.2 | Displacement and strength capacities of the specimens

In this section, performance of each specimen in terms of their displacement and strength capacity is presented. Definitions of different capacities
limit states of specimens are defined from the initial cracking to the gravity load collapse.
When the tensile strength of concrete has been exceeded, then initial cracks used to appear in a critical section of RC column. The tensile
strength of a column under axial load and bending moment can be estimated as 0.75√fc’ where fc’ is the compressive strength of concrete in
MPa. During the test, when cracks become visible that drift is recorded as initial cracking point. The first yielding displacement (δy) is yielding cor-
responding to the yielding of longitudinal reinforcement. In the current study, strain in the longitudinal reinforcement is monitored by using strain
gauges. Whereas yielding displacement (δyu) is defined based on the elastoplastic approximation of force‐deformation relationship as proposed by
Priestley and Benzoni.[29] According to elastoplastic approximation, the yielding displacement is the displacement corresponding to the 75% of the
ultimate strength. 20% drop in peak lateral strength is considered as stage of loss of lateral resistance. Deformation and strength was recorded at
that stage. When the column was no longer able to sustain the axial load that stage was considered as the gravity load collapse in this study. Axial
load was monitored during the test when the column was no longer able to carry the axial load will be considered as gravity load collapse.
This calculation procedure is explained in the Figure 12 for the specimen S1L. The ultimate strength (Fu) for the specimen S1L is 86.83 KN,
and the displacement corresponding to the 75% of the ultimate strength (65.12 KN) is 25.34 mm. The displacement corresponding to the ultimate
8 of 14 MEHMOOD ET AL.

FIGURE 9 Column specimens at the stage near collapse

strength (δu), displacement corresponding to the 20% drop in lateral strength (δ20) and the displacement corresponding to the gravity load collapse
(δcollapse) is also shown on the Figure 12.
Table 2 presents the displacement capacity of each specimen. The presence of poorly detailed lap splice (25 db) in plastic hinge regions causes
the reduction in maximum lateral strength. Soon, after the maximum strengths have been reached the lateral strengths of all specimens have sig-
nificantly degraded as shown in Figures 8d (e) 8. However, all specimens were capable of carrying full axial load although its lateral strength has
been significantly deteriorated. The maximum drift of 5, 5, and 4% of columns S1Ls, S2Ms, and S3Ss respectively could be achieved without axial
load carrying capacity being compromised.
The limiting drift at the onset of buckling in the longitudinal reinforcement, and gravity load collapse of the specimen S2Ms with lap splices
was significantly larger than the specimen S2M specimen without lap splice, although for the rest of the specimen with lap splice have at least
similar displacement capacity as compared with their counterpart specimens without lap splices. RC columns with short lap spliced length and
low level of confinement such as considered in the current study generally have limited tensile capacity at the base, which can lead to the prema-
ture failure of such RC columns. Similar observation has been made in previous studies (e.g., the studies of Samade and Samadi and Jae‐Yeol and
Moghaddam[30,31]); however, limited studies[2] have been carried out to consider the “cranked” (slight bend at the end of lap‐spliced rebar) lap
splices. This unexpected behavior is attributed to the rotation due to slippage of splices. It was observed from the experiment that after bond
deterioration, splice failure, and longitudinal bar buckle, “a concrete hinge” with very low moment capacity has formed near the cranked part of
the longitudinal reinforcement as shown in encircled parts of the Figure 9d–f. The formation of this concrete hinge allows the columns to largely
displace without gravity load collapse. Eventually, the column axial load failure was due to crushing of the “concrete hinge” located just above the
splice region. It should be noted this so called “concrete hinge” should not be confused with the conventional plastic hinge mechanism, where RC
MEHMOOD ET AL. 9 of 14

FIGURE 10 Strain variation (North side) in longitudinal reinforcement along the height of specimen S1L (a) tension strain (b) compression Strain

FIGURE 11 (a) Mechanism of shear plane formation (b) buckling longitudinal reinforcement at 500 mm due to large shear cracks in column
specimen S3S

columns could sustain large deformation without degradation in the lateral strength. Although, lap spliced columns were able to sustain large
deformation without axial load collapse but significant degradation of lateral strength was observed.
Effect of shear crack opening may play an important role in the deterioration of the concrete hinge. When the column subjected to large shear
(i.e., Column S3Ss), the crack angle is very steep and hence shear slip and crushing along the shear plane could easily occur along the open shear
crack. When the slip along the shear plane occurs, the column could no longer sustain the gravity load (Kittipoom, Nelson, John and Helen, 2006).[32]

4.3 | Axial deformation of the specimens

Axial shortening or vertical displacement of the column is very important; if the axial deformation is very high, it can produce very large secondary
moments of the beam‐column joint during an event of an earthquake. Figure 13 presents the results of axial deformation for all specimens. It
10 of 14 MEHMOOD ET AL.

FIGURE 12 Elastoplastic approximation of lateral response for specimen S1L

TABLE 2 Strength and deformation at different stages

Strength (KN‐m) Deformation (mm)/drift (%)
Col.
Mcr My Mu δcr (%) δy (%) δyu (%) δu (%) δ20 (%) δcollapse (%)

S1L 75.78 133.50 193.94 4.63 (0.25) 42 (2) 25.34 (1.2) 40 (2) 78 (4) 101.37 (5)
S1Ls 71.75 111.92 149.22 5.0 (0.25) 62.45 (3) 20 (1) 39.25 (2) 44 (2.15) 103 (5)
S2M 69.40 140.42 188.5 6.78 (0.5) 23.55 (1.5) 23.55 (1.5) 40 (2.5) 58 (3.7) 62.8 (4)
S2Ms 110.63 122.21 162.96 7.91 (0.5) 31.32 (2) 13.34 (1) 23.65 (1.5) 37.5 (2.4) 77.96 (5)
S3S 86.53 134.26 184.45 5.7 (0.5) 13.64 (1.25) 13.6 (1.24) 13.63 (1.25) 19 (1.7) 42.05 (3.8)
S3Ss 76.16 122.48 163.32 2.63 (0.25) 29.46 (2.6) 9.34 (0.85) 18.62 (1.5) 26 (2.4) 45.4 (4)

FIGURE 13 Axial deformations of specimens

should be noted that data for the specimen S2M was unfortunately lost due to breakage of the one LVDT. At initial stage, almost negligible prog-
ress in axial deformation was observed; however, a sudden increase in axial deformation was observed at one peculiar point as shown in Figure 9.
It was noted that for each specimen as the buckling of longitudinal reinforced started the axial deformation increased suddenly, which continued
to the final gravity load collapse of the specimens. Due to poor detailing and a very little confinement of the concrete core, buckling of the lon-
gitudinal reinforcement was found to be inevitable.
It was noted that the progression of axial deformation is related to the damage experienced by the specimens. When the significant degra-
dation (i.e., more than 3.5% drift in case of S1L and S1Ls) in lateral strength was observed, a sudden increase in the axial deformation was found
to be associated with it. However, in case of short specimens (S3S and S3Ss), the development of shear plane caused the increase in the axial
deformation earlier than this stage. One possible reason is the penetration of the shear cone, as explained in the Figure 11. This process is further
aggravated in case of the specimen S3Ss, splitting cracking along the lap‐splices and development of shear plan has caused the earlier increase in
the axial deformation. A sudden jump in the axial deformation was observed, when the shear plane further penetrated downward at about 3.5%
drift in case of specimen S3Ss. Different retrofitting options such as carbon fiber‐reinforced polymer wrapping can be adopted to increase the
confinement for this type of nonductile RC columns to avoid the excessive axial deformation.
MEHMOOD ET AL. 11 of 14

4.4 | Curvature distribution of the specimens

For curvature calculations, column was divided into different segments. Curvature between any two segments was calculated as the difference
between measurements of vertical transducers attached to the sides of each segment divided by the product of horizontal and vertical dimension
of each segment. Figure 14 presents the results of average curvature for all specimens. From this Figure, we can see that the curvature is concen-
trated in the region of 500 and 450 mm for columns provided with lap splices. Maximum curvature was found to be concentrated in the region just
above the lap splices due to crank portion of longitudinal reinforcement. This is also depicted that rotation at that location is higher than at the
base after bond failure. As explained earlier, due to bond slippage of the lap splice specimen with lap splices could sustain larger lateral deforma-
tion than the specimen without lap splices. For the specimen without lap splices curvature was concentrated in the region of 150 mm.

5 | ANALYTICAL MODELING

The behavior of RC column specimens S1L, S2M, and S3S is simulated using the shear‐flexure‐interaction multiple‐vertical‐line‐element model
(SFI‐MVLEM,[19,24]) available in open‐source computational platform OpenSees. This model captures interaction between axial‐flexural and shear
deformations under cyclic loading, but it does not incorporate the ability to simulate the failure mechanism associated with inadequate lap splices.
Although theoretical models that capture splice failure can be found in the literature (e.g., the study off[25]), most of them are not implemented
available in publicly available computational platforms such as OpenSees. Therefore, analytical simulation of specimens that experienced splice
failure (S1Ls, S2Ms, and S3Ss) is not considered in this study.
The SFI‐MVLEM incorporates biaxial constitutive RC panel behavior described with the fixed‐strut angle model (OpenSees nDmaterial
FSAM;[19]) into a two‐dimensional macroscopic fiber‐based model formulation of the multiple‐vertical‐line‐element model[33], as illustrated in
Figure 15a. Biaxial behavior of concrete within each RC panel element is described using a uniaxial stress–strain relationship for concrete applied

FIGURE 14 Curvature distribution along the height of specimens

12 of 14 MEHMOOD ET AL.

FIGURE 15 Modelling approach for specimen S1L: (a) shear‐flexure interaction multiple vertical line element model element, (b) model vertical
discretization, and (c) model horizontal discretization

along fixed compression struts, where mechanisms representing compression softening (Vecchio & Collins, 1993), hysteretic biaxial damage,[34]
and tension stiffening effects[35] are used. The implemented uniaxial constitutive relationship for reinforcing steel is applied along vertical and hor-
izontal reinforcement directions. The RC panel model also incorporates two shear‐resisting mechanisms to resist shear stresses along concrete
cracks including: (a) shear aggregate interlock effects[36] and (b) reinforcement dowel action.[24] Axial‐shear coupling is achieved at each macro‐
fiber (panel) level through two‐dimensional constitutive RC panel material model, which enables coupling of axial‐flexural‐shear responses under
cyclic loading at the SFI‐MVLEM element level. The element rotation occurs at the center of rotation located at the element centerline, where
constant distribution of curvature is assumed over the element height. Strains and stresses at each of the RC macrofibers are treated in the aver-
age sense. Since shear stiffness and strength of the element evolve according to computed RC panel responses and assumed material behavior,
explicit definition of shear modeling parameters is not necessary in SFI‐MVLEM, as opposed to other commonly used wall models with uncoupled
shear and axial/flexural behavior such as displacement‐based beam‐column element in OpenSees[18] or shear wall element in Perform 3D (CSI).
The behavior of unconfined and confined concrete is simulated using uniaxial material model for concrete Concrete02 (Yassin, 1994), whereas
the behavior of reinforcing steel is simulated with uniaxial constitutive model for steel SteelMPF,[19] both available in OpenSees.
The simulated column specimens were discretized with a number of model elements along the height of the column (n), and a number of RC
macro‐fibers along the column cross section (m) based on the number and locations of vertical reinforcing bars. Figure 15 illustrates typical
discretization of specimen geometry for a column specimen S1L with seven elements along the height (n = 7) and 11 macro‐fibers (m = 11) used
to represent column cross section. Discretization of models for column specimens S2M and S3S considered in this study was performed in a sim-
ilar manner. The reinforcing ratio in vertical direction for each macro‐fiber (RC panel) was obtained based on the areas of vertical reinforcing bars
and concrete within the macro‐fiber and reinforcing ratio in horizontal direction was obtained based on the area of transverse reinforcement pro-
vided within the model element. Uniaxial material models for concrete and reinforcing steel were calibrated to match as‐tested material properties
according to material calibration procedure described by Orakcal and Wallace.[37] Concrete compressive strength was increased due to effect of

FIGURE 16 Experimentally measured and analytically predicted load‐deformation responses for specimen: (a) S1L, (b) S2M, and (c) S3S
MEHMOOD ET AL. 13 of 14

confinement based on provided transverse reinforcement using the confinement model proposed by Mander, Priestely, and Park.[38] Finally, axial
load and lateral displacement history matching test conditions were applied at the top node of the analytical models.
Figure 16 compares the experimentally measured and analytically predicted lateral load versus deformation response for the test specimens
S1L, S2M, and S3S. It can be observed from the figure that the SFI‐MVLEM predicts reasonably well the lateral strength under reversed‐cyclic
loading, and the pinching characteristics of the load‐deformation response for all three specimens; lateral strength is underestimated by approx-
imately 5% at intermediate drift levels for specimen S2M and S3S. Initial stiffness of column specimens S1L and S3S is captured accurately by the
model, whereas for specimens S2M the initial stiffness is slightly overestimated. Loading and unloading stiffness and cyclic stiffness degradation is
also reasonably well‐predicted by the analytical model for all specimens. Finally, SFI‐MVLEM captures reasonably well the strength loss observed
in the experiments due to flexure, shear‐flexure, or shear failure.

6 | C O N CL U S I O N S

Existing stocks of RC buildings designed before the modern provisions of the seismic detailing have been proved to behave poorly during an earth-
quake. In moderate seismic areas like Thailand, buildings are usually designed for gravity loads. A research project was designed to assess the seis-
mic performance of these nonductile columns. Quasi‐static reversed cyclic testing of representative columns revealed valuable information.
Experiments were carried until the gravity load collapse of the specimens occurred. When the specimen was not able to carry out any further
axial load, then the test was terminated. Column specimens without lap splices could sustain the gravity load up to 4% lateral drift, whereas the
column provided with lap spliced can carry the axial load up to 5% lateral drift. It was observed that axial deformation gradually increases as spec-
imens approaching towards the collapse. Axially deformation increase substantially after the buckling of longitudinal reinforcement due to the
shortening of column length. Due to poor detailing and confinement, the buckling of longitudinal reinforcement is inevitable. Longitudinal rein-
forcement buckled after spalling of cover concrete when no restraint available on the outer side of reinforcement. In case of long and medium
columns, longitudinal reinforcement buckled at 100 mm from the bottom where the maximum moment occurred. It was noted that after buckling,
gravity load failure is warranted. For short specimen, premature buckling of longitudinal reinforcement was observed at an early stage after devel-
oping large shear cracks.
Finally, analytical investigation of the behavior of selected column specimens showed that currently available analytical models can accurately
capture the lateral load and deformation capacity of RC columns without splice failure.
Based on the findings of the current study, it is clear that the RC column in Thailand could suffer serious damage in the wake of a major seis-
mic event. Hence, it is imperative to follow the guidelines of the buildings codes of Thailand strictly in the design of new RC buildings. Meanwhile,
existing RC buildings need to be assessed and retrofitted accordingly. Suitable retrofitting techniques for such nonductile RC columns should be
adopted to increase the lateral deformation capacities; carbon fiber‐reinforced polymer (CFRP) wrapping can be a useful option to increase the
confinement of such nonductile RC columns.

ORCID

Tahir Mehmood https://orcid.org/0000-0002-0612-4691

RE FE R ENC E S
[1] J. P. Moehle, K. J. Elwood, H. Sezen, ACI Special Publications. 2001, 197, 215.
[2] A. Boys, D. K. Bull, S. Pampanin, Rotorua, New ZealandNew Zealand Concrete Industry Conference 2008, 2008.
[3] K. J. Elwood, J. P. Moehle, ACI Structural Journal 2005, 102(4), 578.
[4] W. Y. Kam, S. Pampanin, and K. Elwood. “Seismic performance of reinforced concrete buildings in the 22 February Christchurch (Lyttleton) earth-
quake.” (2011).
[5] A. C. Lynn, J. P. Moehle, S. A. Mahin, W. T. Holmes, Earthq. Spectra 1996, 12(4), 715.
[6] T. Nakamura, M. Yoshimura, J Asian Archit Build Eng 2002, 1(1), 21.
[7] H. Sezen, Department of Civil and Environmental Engineering, University of California, Berkeley 2002.
[8] I. Kogoma, T. Hayashida, Minowa, tenth world conference on earthquake engineering, Rotterdam, Holland 1992 3013.
[9] H. Ousalem, T. Kabeyasawa, A. Tasai, Thirteenth world conference on earthquake engineering, Vancouver, British Columbia, Canada 2004 Paper No. 370.
[10] T. H.‐K. Kang, S.‐S. Ha, D.‐U. Choi, ACI Structural Journal 2010, 107(1).
[11] J. Su, J. Wang, Z. Bai, W. Wang, D. Zhao, Eng. Struct. 2015, 103, 174.
[12] C. Suesuttajit, Asian Institute of Technology, Master thesis, 2007.
[13] S. Otani, Earthq. Spectra 1999, 15(3), 485.
[14] B. Kafle, M. A. Kermani, A. Wibowo, Electronic Journal of Structural Engineering 2008, 1.
[15] W. M. Ghannoum, J. P. Moehle, B. Yousef, 100th Anniversary Earthquake Conference, 2006.
[16] J. P. Moehle, Proc. Luis Garcia Symposium,2008.
[17] SAP, Computers and Structures Inc. (CSI). 2000.
14 of 14 MEHMOOD ET AL.

[18] F. F. Taucer, E. Spacone, F. C. Filippou, Report No. UCB/EERC‐91/17. Earthquake Engineering Research Center, College of Engineering, University of Cal-
ifornia Berkeley,1991.
[19] K. Kolozvari, K. Orakcal, J. W. Wallace, Computers and Structures Journal 2018, 196, 246.
[20] S. Y. Xu, J. Zhang, Earthq. Eng. Struct. Dyn. 2010, 40(3), 315.
[21] H. Jiang, Y. Kurama, ACI Structural Journal 2010, 107(4), 400.
[22] M. Fischinger, K. Rejec, T. Isaković, 15th World Conference on Earthquake Engineering, Vol. 2120, Lisbon, Portugal 2012.
[23] T. Mehmood, K. Hussain, P. Warnitchai, Magazine of Concrete Research 2015, 67(18), 1003.
[24] K. Kolozvari, K. Orakcal, J. W. Wallace, ASCE Journal of Structural Engineering 2015, 141(5), 04014135.
[25] S. R. Chowdhury, K. Orakcal, ACI Structural Journal 2013, 110(5), 735.
[26] T. Ornthammarath, P. Warnitchai, K. Worakanchana, S. Zaman, R. Sigbjörnsson, C. G. Lai, Bull Earthquake Eng 2010, 9(2), 367.
[27] N. Poovarodom, N. Plalinyot, Journal of Earthquake Engineering 2013, 17(2), 209.
[28] M. J. N. Priestley, Park, Structural Journal 1987, 84(1), 61.
[29] M. J. N. Priestley, G. Benzoni, ACI Structural Journal 1996, 93(4), 474.
[30] C. H. O. Jae‐Yeol, J. A. Pincheira, 13th World Conference on Earthquake Engineering,Vancouver, B.C., Canada august 1‐6, Paper No 2004 1506.
[31] M. Samadi, H. Moghaddam, K. Pilakoutas, Proc. of the 15th World Conference on Earthquake Engineering, Lisbon, Portugal 2012.
[32] R. Kittipoom, L. Nelson, W. John, G. Helen, Canberra 2006, 24– November.
[33] K. Orakcal, J. W. Wallace, J. P. Conte, Structural Journal 2004, 101(5), 688.
[34] M. Y. Mansour, T. C. Hsu, J. Y. Lee, ACI Structural Journal 2002, 205, 293.
[35] A. Belarbi, T. C. Hsu, ACI Structural Journal 1994, 91(4), 465.
[36] K. Orakcal, D. Ulugtekin, L. M. Massone, (2012). Proceedings, 15th World Conference on Earthquake Engineering, Lisbon, Portugal, 2012.
[37] K. Orakcal, J. W. Wallace, ACI Structural Journal 2006, 103(2), 196.
[38] J. B. Mander, M. J. N. Priestely, R. Park, ASCE Journal of Structural Engineering 1988, 114(8), 1804.

AUTHOR BIOGRAPHI ES
Tahir Mehmood is currently working as an assistant professor at Civil Engineering Department of COMSATS Institute of Information Technol-
ogy, Wah Cant, Pakistan. He obtained his Master of Science degree and Doctor of Engineering degree from the Asian Institute of Technology,
Thailand, in 2010 and 2015, respectively. His doctoral research work is focused on the seismic behavior of high‐rise structures. His areas of
interests are structural dynamics, earthquake engineering, performance‐ based seismic design, and seismic performance evaluation of existing
structures.

Kittipoon Rodsin is currently an associate professor at Department of Civil and Environmental Engineering Technology at King Mongkut's Uni-
versity of Technology (KMUTNB). He obtained his Doctor of Engineering degrees from the University of Melbourne (Australia) in 2007 and
Master of Engineering degree from Asian Institute of Technology (AIT) in 1999. His research of interest covers the area of structural dynamics,
earthquake engineering, structural health monitoring, and railway engineering. He may be contacted at [email protected].

Pennung Warnitchai is currently a professor and head of Department of Civil and Infrastructure Engineering (CIE) at the Asian Institute of
Technology (AIT). He obtained his Master and Doctor of Engineering degrees from the University of Tokyo (Japan) in 1985 and 1990, respec-
tively. His research encompasses the areas of structural dynamics, earthquake engineering, wind effects on structures, bridge engineering, and
control of structural vibration. Since 2002, he has been the leader of a long‐term research program on “mitigation of earthquake risk in Thai-
land,” funded by the Thailand Research Fund and participated by researchers from eight universities and three government departments in
Thailand. In 2009, he was leading a team of experts and professional engineers to develop the first official seismic design standard for build-
ings and structures in Thailand. He may be contacted at HYPERLINK “mailto:[email protected]” [email protected].

Kristijan Kolozvari is an assistant professor in the Department of Civil and Environmental Engineering at California State University, Fullerton.
He received his Bachelor of Science degree from the University of Belgrade, Serbia, and his MS and PhD degrees from the University of Cal-
ifornia, Los Angeles. His research interests include development and application of innovative analytical tools for nonlinear analysis of rein-
forced concrete structures, performance‐based seismic design, seismic retrofit, tall buildings behavior and design, and earthquake resiliency.

How to cite this article: Mehmood T, Rodsin K, Warnitchai P, Kolozvari K. Investigating the vulnerability of nonductile reinforced con-
crete columns in moderate seismic regions to gravity load collapse. Struct Design Tall Spec Build. 2018;e1578. https://doi.org/10.1002/
tal.1578