You are on page 1of 10

Engineering Structures 247 (2021) 113211

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Cyclic shear displacement model for reinforced concrete columns


Erkan Bicici a, b, *, Halil Sezen a
a
Department of Civil, Environmental and Geodetic Engineering, The Ohio State University, Columbus, OH 43210, USA
b
Department of Civil Engineering, Faculty of Engineering, Artvin Coruh University, Artvin 08100, Turkey

A R T I C L E I N F O A B S T R A C T

Keywords: Lateral response of reinforced concrete (RC) columns can be modeled and simulated considering three
Shear displacement displacement components: flexural displacement, displacement due to slippage of longitudinal bars at column
Column model ends, and shear displacement. The incorporation of shear displacement in a computational model is essential for
Concrete shear failure
accurate calculation of lateral load–displacement relationship. The contribution of shear displacement can be
Cyclic model
significant for RC columns with low displacement capacity or ductility. An easy-to-apply shear displacement
model is proposed in this study, and implemented in the open source software, OpenSees. The proposed model is
based on a pre-defined envelope or monotonic model and does not require updating of modeling parameters
during each cycle. The model is validated using experimental data. New rules are developed for the cyclic model
to capture the effects of cracking and change in stiffness and strength during the loading and unloading cycles.
Comparison of the calculated and measured shear displacements indicates that the cyclic shear response can be
captured by the proposed model.

1. Introduction structures that are vulnerable to shear failure and collapse during
earthquakes in many parts of the world.
Reinforced concrete (RC) columns experience brittle shear failure if The lateral displacement of an RC column can be represented as the
the design and detailing of the columns do not meet the requirements of summation of three displacement components: flexural displacement,
the modern seismic codes, e.g., ACI 318 [1]. Existing RC buildings displacements due to slippage of the longitudinal bars at the ends of the
constructed before mid-1970s and new buildings designed without column, and shear displacement (Fig. 1). Displacement component
following the seismic code requirements typically include columns with model (DCM) enables modeling of three displacement components
widely spaced and poorly detailed transverse reinforcement. At failure, separately and then combining them together to simulate total lateral
these columns typically have low total lateral displacement capacity and displacement of the column. The flexural displacement is caused by the
larger contribution from shear displacement component, which is bending moment over the height of the column. The slip displacement is
associated with large diagonal cracks prior to column axial failure [18]. concentrated at the ends of the column due to longitudinal bars pulling
The contribution of shear deformations to total lateral displacement is out of anchorage.
much larger in these columns than in well-designed flexure-dominated The displacements due to flexure and slippage can be modeled and
columns. Such columns are typically defined as shear-critical, low- calculated relatively accurately using the currently available models
ductility, or nonductile columns. Shear cracking and associated damage [26,9]. For instance, fiber-section models with non-linear uniaxial ma­
may cause early degradation of lateral load resistance and subsequent terials can be used to accurately calculate the moment–curvature rela­
loss of axial load carrying capacity [5,4,24,31]. The early and sudden tionship. Flexural displacement can then be obtained by integrating the
loss of column load carrying capacity may lead to partial or complete curvatures over the length of the member. The calculated stress–strain
collapse of the structure and loss of life. Accurate modeling and analysis relationship of longitudinal bars at the ends of the columns can be used
of such columns are critical for the evaluation of seismic performance to determine the column lateral displacement caused by bar slip. The
and risk assessment of these structures [29,32]. Thus, it is critical to contribution of shear displacement to total lateral response of RC col­
understand and accurately model the behavior of RC columns with large umns with seismic details is relatively small and can be less than 10%
shear displacement component for the evaluation and analysis of RC [12]. However, the contribution of shear displacement may be

* Corresponding author.
E-mail addresses: erkanbicici@artvin.edu.tr (E. Bicici), sezen.1@osu.edu (H. Sezen).

https://doi.org/10.1016/j.engstruct.2021.113211
Received 1 September 2020; Received in revised form 28 March 2021; Accepted 12 September 2021
Available online 24 September 2021
0141-0296/© 2021 Elsevier Ltd. All rights reserved.
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

Fig. 1. Three lateral displacement components of an RC column.

significantly large in poorly designed shear-critical columns [24,21]. A and system failure mode. However, research is limited and there is no
recent study showed that the shear displacement can be larger than 40% consensus about the calculation of column shear displacement [22,26].
of the total lateral displacement [13]. Therefore, the effect of shear In DCM, it is convenient to use zero-length springs at the ends of the
deformations must be included in computer models of such columns to column to simulate shear and slip behavior and the flexural behavior can
accurately determine lateral strength, displacement and axial capacity, be modeled with fiber section distributed plasticity (Fig. 1). In this

Fig. 2. Available cyclic shear models; a) Ozcebe and Saatcioglu [21], b) Pincheria et al. [20], c) Sezen and Chowdhury [25], and d) LeBorgne and Ghannoum [11].

2
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

study, an easy-to-apply and relatively accurate cyclic model is proposed to degrade. Then, the reloading path goes through the symmetry of the
to predict the shear displacement of an RC column subjected to axial and unloading point with respect to origin. In this model, axial failure of the
cyclic lateral loading. The proposed cyclic shear displacement model is column was not considered.
built upon previously developed monotonic [22] and cyclic [25] models. Sezen and Chowdhury [25] simplified the shear response calculated
The proposed model is implemented in an open-source structural anal­ from the MCFT model by considering the displacements at the onset of
ysis software, OpenSees, [15]. A uniaxial material is created and added shear strength degradation and axial failure (Fig. 2c). The pre-peak
to the OpenSees library to model the proposed shear spring at the end of lateral load-shear displacement relationship in the model is based on
the column (Fig. 1). The proposed model is validated using pure shear MCFT. The authors modified the cyclic rules proposed by Ozcebe and
response data, separated from other displacement components, obtained Saatcioglu [21] to eliminate the details and limitations of the MCFT
from six full-scale columns tested by Sezen and Moehle [24] and Saat­ model. Elwood [6] developed a shear response model, which was also
cioglu and Ozcebe [21]. used by LeBorgne and Ghannoum [11] (Fig. 2d). The main characteristic
of the model is definition of a failure curve. According to the model, the
shear displacement response is linear elastic with a predefined initial
1.1. Background information
stiffness. Once the pre-defined curve is reached, the shear strength
degradation starts. The response is represented by a negative stiffness
Better understanding of behavior of shear-critical columns is needed
until the residual capacity is reached. The cyclic characteristics of the
to model and capture the seismic behavior of RC buildings including model are defined with three pinching factors for unloading and
such columns. To accurately model and analyze these structures, the
reloading based on the rules in Takeda et al. [27].
cyclic shear behavior must be considered and included in the analysis. The experimental data used in this paper are from rectangular col­
Unfortunately, shear response has been typically ignored and not
umns. The model can be used for circular columns using the parameters
modeled in most available software or research studies. The initial at­ allowed in ACI 318 [1], e.g., the effective depth d can be set equal to
tempts to develop computational models were very limited due to the
0.80 times the column diameter. Kakavand et al. [10] found that the
lack of available experimental data and the complexity of the problem, effects of critical parameters on shear behavior are similar based on
which involves interaction of flexure, shear and axial forces [21].
detailed assessment of 325 rectangular columns and 172 circular
Modified Compression Field Theory (MCFT) proposed by Vecchio columns.
and Collins [28] mechanically explains the overall shear behavior of RC
members. Some of the current models that are used to calculate lateral
load-shear displacement relationship of RC columns are based on MCFT. 1.2. Research significance
However, the MCFT model includes limitations, i.e., it works well under
monotonic lateral load and is typically accurate until the peak strength is Most of the currently available numerical models are developed to
reached. A mathematical model for cyclic shear behavior, separate from predict the total lateral response of columns by lumping all three
the flexure and slip displacement components, dates to study of Ozcebe deformation components in a single model and without knowing the
and Saatcioglu [21] (Fig. 2a). They proposed a set of rules to calculate interaction and contributions of flexural or shear response on the total
cyclic shear response of RC columns based on a well-defined envelope response. Some of these models may seem to predict the total response
model. Ozcebe and Saatcioglu [21] used MCFT to obtain an envelope for numerically while the underlying mechanics and individual contribu­
the shear response of a column and considered yield strength of longi­ tion of flexure, bar slip, and shear contributions can be inaccurate. Such
tudinal steel as one of the critical points for shear displacement. How­ models lead to incorrect estimates of lateral drift, load redistribution
ever, recent studies have shown that the flexural yield strength is not as after failure of a member, and collapse during dynamic analyses, seismic
critical for shear displacement response [22]. Additionally, the shear assessment or design of structures.
strength degradation and axial load failure were not considered in There are only few research studies and models focusing on pure
earlier models, although the model proposed by Ozcebe and Saatcioglu shear behavior of RC members (Fig. 2) and its influence on the total
[21] includes important hysteretic features, such as strength decay, lateral response. This is mainly because it is difficult to isolate and
stiffness degradation and pinching. The application of that model is measure the shear deformations during the experiments in the labora­
limited to RC members which experienced flexural yielding prior to tory. Even when the shear displacements are measured, most re­
shear failure. searchers stop the experiments once the strength degradation starts
A tri-linear cyclic shear displacement model was proposed by Pin­ immediately after large shear cracks develop. Therefore, experimental
cheria et al. [20] (Fig. 2b). The envelope of the model consists of three studies on columns including both shear failure and axial load failure are
mechanical points: cracking, and maximum and residual strengths. For rare. To the knowledge of the authors, few researchers (e.g., [21]; and
cyclic response, a set of rules were developed. Initially, there is no cyclic [24]) report all three displacement components for column specimens.
behavior in pre-peak loading stages, i.e., the column is assumed elastic This very limited experimental evidence is used in this research.
in shear. Once the peak strength is reached the unloading stiffness starts The main objective of this study is to propose an easy-to-apply model

Fig. 3. Lateral force-shear displacement envelope used in the proposed model.

3
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

Fig. 4. Comparison of monotonic shear displacement models.

to calculate the shear displacement of RC columns. Detailed and more leads to strength degradation up to axial failure of the column.
complicated models may predict more accurate results for shear In the shear response envelope developed by Sezen [22], the effect of
displacement, however; these models will be time-consuming and will a large number of parameters are investigated on the shear, shear-
not be preferred for the analysis of multi-story buildings or complicated flexure and flexure response of RC columns. Then, using MCFT, the
structures. In this study, simple cyclic rules are proposed to model an RC lateral strengths are calculated when diagonal shear cracks develop.
column while keeping the accuracy at acceptable level. The simplicity of Based on a regression analysis, Eq. (1) was proposed to calculate the
the model can reduce the modeling and make the model applicable for cracking shear strength, Vcr .
easy modification in the future, e.g., using the open-source software ( )
P GAg
OpenSees. Besides simplicity, no user interference is needed during cy­ Vcr = ′ + 0.10 (1)
2fc Ag L
clic shear analysis because all modeling parameters, e.g., pinching
points, are pre-defined or pre-calculated in the proposed model.
where,fc is compressive strength of concrete, Ag is cross sectional area, P

is axial load of column, L is length of column, and G is shear modulus of


2. Proposed model
the column. P/fc Ag is unitless and G‧Ag /L is in MPa units. Due to being

The proposed model includes an envelope of the lateral force-shear derived from simulations with a limited range of parameters [22], Eq.
displacement relationship, which can be considered as the boundary (1) has limitations for applicability. To avoid unrealistic results, the
for the shear behavior. Rules are defined for cyclic response based on the range of parameters used in the dataset should be carefully evaluated.
defined envelope. The proposed model is based a previously developed Until the first shear cracks develop (region I in Fig. 3), shear response of
envelope model as the boundary, and a new set of rules are proposed to the column is assumed as elastic and the principles of elastic mechanics
simulate cyclic shear behavior of RC columns. are used. It is assumed that the shear strain is uniformly distributed over
the cross-section of the column and shear stiffness is equal to G‧Ag/L. The
cracking shear displacement is calculated using Eq. (2).
2.1. Shear response envelope ( )
P
Δsh,cr = + 0.10 (2)
A previously developed model by Sezen [22] for primary curve of

2fc Ag
lateral load-shear displacement relationship is used in this study. The
The next critical point of the model is peak shear strength, Vn and the
multilinear envelope model is defined with four critical points at: shear
corresponding displacement, Δsh,n (end of region II in Fig. 3). Eq. (3)
cracking, peak lateral strength, onset of strength degradation, and axial
from Sezen and Moehle [23] and Eq. (4) from Sezen [22] are adopted to
failure (Fig. 3). Based on these four critical points, in the proposed
calculate the peak strength (kN) and the corresponding shear displace­
model, four unique shear regions are defined as: elastic (I), stiffness
ment (mm), respectively.
degradation (II), peak (III), and strength degradation (IV). In RC col­
⎡⎛ √̅̅̅̅ ⎞ ⎤
umns, initial flexural cracks occur in early stages of loading. The initial ′ √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅

0.5 f
flexural cracking and subsequent softening does not have major effect on ⎢⎜
Vn = ⎣⎝
c √
√1 + √̅̅̅̅′
P ⎟
⎠0.8 Ag +
Av fyv d ⎥
⎦ (3)
a
shear behavior. Thus, the shear response during the early loading stage d 0.5 f A
c g
s
stays elastic (region I in Fig. 3). In other words, in this stage initial
flexural cracks do not appear to reduce elastic shear stiffness until di­ ⎡ ⎤
agonal shear cracks start to form. In the second stage, diagonal shear ⎢ ⎥
⎢ 1 fy ρl ⎥
cracks start to degrade shear stiffness. Thus, region II in Fig. 3 is labeled Δsh,n =⎢ √ ̅̅̅̅̅̅̅̅ − 0.0004⎥L (4)
⎢5000 a ⎥
as stiffness degradation region. Although the stiffness of shear response
P
⎣ d A ⋅f
′ ⎦
g
decrease, the lateral force keeps increasing until the peak strength is
c

reached (region III). In this model, it is assumed that, the peak shear
strength stays constant until the onset of strength degradation (at Δsh,u in where, Av is area of transverse reinforcement, fyv is yield strength of
Fig. 3). In the strength degradation region, the shear cracks become transverse reinforcement, d is the effective depth of cross section, s is
wider and start to dominate the overall lateral response. The reduction vertical spacing of transverse reinforcing steel, and ρl is longitudinal
or loss of stress transfer across the widened diagonal or shear cracks

4
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

Fig. 5. Proposed cyclic shear displacement rules if unloading starts in regions I or II.

reinforcement ratio. a is the shear span length of the column, which is negative stiffness or degrading response. The cyclic behavior of shear-
equal to L/2 for double curvature columns and L for cantilever columns. critical or nonductile RC columns may be dominated by shear and
The diagonal shear cracks become more apparent and lead to brittle failure [22,26]. To avoid overestimation of both strength and
strength degradation starting at shear displacement Δsh,u at the end of displacement capacity of the column, the cyclic shear response should be
region III in Fig. 3. Eq. (5) is used to calculate the displacement corre­ included in dynamic and cyclic analysis. Cyclic behavior of a nonductile
sponding to onset of shear strength degradation [8,19]; and [22]. column may show stiffness deterioration, strength degradation, pinch­
( ) ing effect, and early axial load failure. In the proposed model, these
υn
Δsh,u = 4 − 12 ′ Δsh,n (5) phenomena and axial load-shear-flexure interaction are considered and
fc
included in the shear model. Total lateral displacement models,
where, υn (= Vn /bd) is the shear stress corresponding to the maximum including cyclic bar slip and flexure models, are presented elsewhere
shear strength.The widening of diagonal shear cracks leads to a drastic [14,3].
decrease in the shear strength to a point that these cracks result in total The cyclic rules in previous studies were generally defined by
loss of lateral and axial load capacity. To calculate lateral displacement equations at each turning (unloading and reloading) point of cycles
at axial failure, ΔALF , Eq. (6) is used [5,4]. [21,16,30] or a ratio of peak strength and displacement [11,6]. Different
) from the previously developed models, in the proposed model, defining
ΔALF 4 1 + tan2 θ the envelope or primary curve is sufficient to simulate the entire cyclic
= ( (6)
L 100
s
shear behavior. The proposed model calculates the cyclic slopes without
tanθ + P Av ∙fyv ∙ d∙tanθ additional user input. Thus, no extra strength or deformation calculation
is needed. This approach simplifies the cyclic shear displacement cal­
where θ is the average angle of shear crack, which is assumed to be 65◦ culations; however, the user influence over the definition of cyclic
in this study. The displacement calculated by Eq. (6) includes dis­ response is reduced.
placements due to flexure and slippage of reinforcing bars at the ends of
the column. The shear displacement component at the axial failure, Δsh,a 2.2.1. Rules for cyclic model
can be calculated from Eq. (7) by subtracting the maximum flexural and The cyclic rules developed in this research are based on limited
slip displacement from the total lateral displacement at axial failure, available experimental data and consistent with the previously devel­
ΔALF . oped models. In the proposed model, the unloading/reloading slopes
depend on the maximum shear displacement for each cycle. The shear
Δsh,a = ΔALF − Δfl − Δsl ≥ Δsh,u (7) behavior at any stage is assumed to correspond to one of the four me­
chanical regions described in Section 2.1.
where, Δfl and Δsl are the maximum flexure and slip displacement, Before the analysis, two additional slopes need to be calculated in the
respectively. The maximum flexural and slip displacement can be ob­ proposed model (k1 and k2 in Fig. 3). The first slope, k1 is defined as the
tained from moment-curvature analysis of the cross section using slope of the line between the peak point in one loading direction and the
constitutive relations for steel and concrete materials [17,14,7]. cracking point in the other loading direction. k2 is defined as the slope of
Fig. 4 compares the experimental data and previously described two the line connecting the onset of shear strength degradation point in one
envelope models for shear displacement of Specimen-1 tested by Sezen loading direction to the cracking point in the other loading direction as
and Moehle [24]. Using MCFT, shear displacement can be calculated up shown in Fig. 3. The slopes are calculated from the envelope of shear
to the peak strength [19] and Sezen [22]. The MCFT model over­ displacement of the column. The response on the positive and negative
estimates the measured shear displacement. The model proposed by loading directions may be different due to the different axial loads
LeBorgne and Ghannoum [11] predicts nearly rigid elastic shear applied on the column. In this case, the unloading and reloading slopes
response initially with a large stiffness. will not be the same in both loading directions.
In the initial elastic region (region I in Fig. 3), the shear response of
2.2. Cyclic shear model the column is linear whether it is loaded or unloaded. Thus, there is no
hysteretic energy dissipation due to shear in the elastic region under
Nonlinear dynamic analysis of a building under seismic loading re­ cyclic loading. In this region, there may be initial flexural cracks in the
quires nonlinear cyclic modeling of its structural elements. To accurately column, but they do not affect the shear response because the column
model the behavior of the entire structure, the model should accurately does not have shear cracks yet.
capture the nonlinear cyclic response at the element level, including the

5
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

Fig. 6. Proposed cyclic shear displacement rules if unloading starts in regions III or IV.

Once the shear displacement exceeds the shear cracking displace­ more apparent and larger, and the unloading slopes are reduced. If the
ment, Δsh,cr in any direction, the rules for region II govern the shear unloading starts in the peak region between the peak displacement, Δsh,n
response (Fig. 3). The extension and enlargement of shear cracks reduce and onset of shear strength degradation point, Δsh,u (region III in Fig. 3),
the shear force transfer along the crack surface that reduce the shear the unloading rules are defined as a function of maximum shear
stiffness. The elastic shear mechanism and related parameters are no displacement in the other loading direction. The three possible paths are
longer valid. When unloading occurs between the cracking displace­ separately presented in Fig. 6.
ment, Δsh,cr and the shear displacement corresponding to peak strength, i) If the target response is in region I or II in the other loading di­
Δsh,n (within region II), there are three possible paths for unloading/ rection and if the unloading starts in region III, the unloading slope is k1
reloading depending on the magnitude of the maximum displacement (Fig. 6) until the unloading reaches shear cracking force, Vcr (line EF in
the column reached in the other loading direction in the previous cycles. Fig. 6). Then, the unloading path goes through the shear cracking point
The three possible paths for unloading/reloading are separately pre­ in the other loading direction (line FC in Fig. 6). Larger shear
sented in Fig. 5. displacement and larger shear cracks in this region lead to further
i) If the previous maximum displacement in the other loading di­ reduction in unloading slopes. After reaching the cracking point, the
rection was in region I or II, the unloading/reloading behavior follows loading follows the primary curve (line CD in Fig. 6).
the slope between the point where unloading started and the cracking ii) If the column is in region III in both unloading and loading di­
point in the other loading direction (path BC in Fig. 5 when the rections based on previous cycles, three discrete slopes are defined to
unloading starts at point B). Furthermore, once the unloading/reloading present the unloading/reloading behavior. The unloading slope is k1
path reached the cracking point (C in Fig. 5) on the other side, the until the unloading reaches shear cracking force, Vcr (line EF in Fig. 6).
loading path follows the primary curve (line CD in Fig. 5). Then, the unloading path reaches the intersection point, F′ . The location
In this case, the unloading stiffness (slope of line BC) is always less of F’ is calculated as intersection of two imaginary lines. The first line
than elastic stiffness (slope of line AC) since point B has a larger shear represents cracking strength, Vcr (dashed line in Fig. 6). The second line
strength and both points A and B (or any other point in region II) are starts from the previously reached peak point (G in Fig. 6) and goes to
connected to point C. In this region, the initiation of shear cracks leads to cracking strength, Vcr with the slope of k2 (FF’ in Fig. 6). Thus, the slope
reduction in unloading slope. of FF’ decreases with increasing shear displacement. After reaching F’,
ii) If the maximum displacement was in region III in the other the reloading goes through the previously reached peak strength and the
loading direction in the previous cycles, the unloading slope is equal to corresponding displacement in the loading direction (F’G in Fig. 6). The
the elastic stiffness k1 (slope shown in Fig. 5) until the unloading force loading follows the primary curve after reaching the point G.
reaches the shear cracking force, Vcr, i.e., the slope of line BQ is k1 in The opening and closing of shear cracks during multiple loading
Fig. 5. Then, the unloading path goes through the maximum displace­ cycles may lead to different stiffnesses in the unloading and reloading
ment on the other loading side and the corresponding strength (line QG directions. In the proposed approach, three different slopes are intro­
in Fig. 5). duced. Between cracking forces, due to the opening of cracks, the stiff­
The opening of previously developed cracks and appearance of new ness becomes lower than the stiffness corresponding to the loads larger
shear cracks in the other loading direction reduces the unloading slope. than the cracking force. As the shear cracks close in both loading di­
Once the loading reaches the envelope curve (at point G), the loading rections, shear stress transfer and the resistance provided by the crack
follows the envelope curve. surfaces improve. This approach also enables modeling of potential
iii) If the previous maximum displacement in the other loading di­ pinching in a given cycle depending on the magnitude of the loading and
rection is in region IV, the unloading slope is defined as k1 (slope of line unloading shear displacements and strengths.
BZ in Fig. 5). Then, the reloading path goes through the point of previous iii) If the target column response is in region IV after reloading, the
maximum displacement in the reloading direction and the correspond­ unloading slope is defined as k2 (EZ in Fig. 6). If the target column
ing strength (line ZR in Fig. 5). Then, if the loading continues, the path response is in region IV when the reloading starts, the reloading path
follows the primary curve after reaching the primary curve (at point R in goes through the previous maximum displacement in the reloading di­
Fig. 5). rection and the corresponding strength (line ZR in Fig. 6).
Once the maximum shear displacement exceeds the displacement Finally, if the unloading starts in the strength degradation region
corresponding to the peak shear strength, the rules for region III are (region IV), the slope of the unloading path is k2 (line WX in Fig. 6)
triggered to govern the response. In this region, the shear cracks become regardless of the target region in the other loading direction. However,

6
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

Fig. 7. Proposed rules for the cyclic shear model.

Table 1 Table 3
Cyclic paths based on the mechanical region. Calculated shear displacements and strengths to define envelope response of test
Target region
columns.
Test Vcr Vn Vp Δsh,cr Δsh,n Δsh,u Δsh,alf
I II III IV
column (kN) (kN) (kN) (mm) (mm) (mm) (mm)
Unloading I elastic
Specimen 104 282 311 0.176 3.32 10.33 154.4
region II line BC in line BCD in line BQG in line BZR in
1
Fig. 5 Fig. 5 Fig. 5 Fig. 5
Specimen 240 407 315 0.403 1.10 2.88 61.8
III line EFC in line EFCD in line EFF’G in line EZR in
2
Fig. 6 Fig. 6 Fig. 6 Fig. 6
Specimen 87 249 253 0.148 6.15 19.67 94.1
IV line EXZ in
3T
Fig. 6
Specimen 187 386 313 0.316 1.06 3.17 80.6
3C
Specimen 114 289 314 0.175 3.34 10.82 154.4
for reloading there are two possibilities to follow. If strength degrada­ 4
tion is not triggered on the reloading side, the reloading goes through U6 225 372 370 0.166 2.40 6.91 78.1
the previous maximum displacement in the reloading direction and the U1 147 217 293 0.125 4.06 13.96 104.4
corresponding envelope strength. However, once region IV is triggered
on both loading sides, the model compares the previous envelope
Fig. 7).
strengths corresponding to the previous maximum displacements for
both loading sides. Then, the reloading path goes through the smaller
3. Application and validation of the model
strength point. For example, in Fig. 7, once the unloading starts at point
R, the model compares the strengths at M and R. In the figure, the
In this study, a model is proposed to simulate the cyclic shear force-
strength at R is less than that at M. Thus, the reloading goes through the
shear displacement response. The shear response calculated using the
symmetry of R or T in the other loading direction, i.e., the strength and
proposed numerical model can be incorporated into total column
displacement at T and R are the same. However, if the strength at M was
response (considering the effects of two other displacement compo­
less than the strength at R, the reloading would go through the point of
nents) using the horizontal shear spring shown in Fig. 1 [6,14]. Thus, to
symmetry of M (not T in Fig. 7). In this approach, the continuum on the
simulate the shear response, a new uniaxial material object is created
primary curve is eliminated, i.e., there are gaps between points M and T,
(named Shear01) based on the proposed model and added to the uniaxial
and Y and R. The proposed cyclic rules are summarized in Table 1 based
material library of OpenSees (2020).
on the unloading and target region.
The material model for the shear spring was established by consid­
In addition to these rules for complete cycles, if the reloading hap­
ering the behavior in four unique regions of the envelope shown in
pens during an incomplete cycle (without reaching the previous
Fig. 3. The details of Shear01 material object may be found in Bicici [3].
maximum displacement), the reloading goes back to the point where
The proposed shear-spring element monitors the shear displacement of
unloading started. In these incomplete cycles, the unloading path fol­
the RC column at each stage of the loading. Exceeding each critical shear
lows the same path with complete cycles (e.g., cycles JKL and OP in
displacement limit triggers the behavior to be governed by the loading,

Table 2
Geometric and material properties of column specimens.
Test Column a (mm) b (mm) a/d fc (MPa)

fyl (MPa) fyv (MPa) ρl % ρv % s (mm) P (kN)

Specimen 1 1473 457 3.76 21.1 447 469 2.5 0.17 305 667
Specimen 2 1473 457 3.76 21.1 447 469 2.5 0.17 305 2669
Specimen 3 1473 457 3.76 20.9 447 469 2.5 0.17 305 P*
Specimen 4 1473 457 3.76 25.6 447 469 2.5 0.17 305 667
U6 1000 350 3.28 37.3 437 425 3.8 0.84 65 600
U1 1000 350 3.28 43.6 430 470 3.8 0.27 150 0

P*: axial load is varying from 249 kN tension to 2669 kN compression.

7
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

Fig. 8. Comparison of the proposed model with experimental shear response of selected columns.

unloading, and reloading rules defined for the next mechanical region as Mp are calculated from moment-curvature analysis in OpenSees. In the
described in the previous section. analysis, the core and cover concrete are modeled, separately. The peak
In typical RC column experiments, the total lateral displacement and moment generally corresponds to the analysis step that core concrete
reaction forces are recorded and presented. Measurement of shear dis­ reaches its maximum strength. The maximum flexural strength, Vp is
placements isolated from the flexural and slip displacements requires calculated by dividing Mp by shear span (Table 3). In the proposed
additional effort. Thus, it is rare to find RC column specimens with model, the peak strength for the proposed shear-spring element is the
recorded shear displacement histories presented separately. In this smaller of the calculated shear strength, Vn and flexural strength, Vp
research, six test columns with reported pure shear response are (Fig. 3). By using the lower of these two strengths, the proposed model
modeled and analyzed to validate the proposed cyclic model [21,24]. accounts for the interaction between flexural and shear behavior[3].
These six column specimens represent flexure-shear, and flexure domi­
nated failures. The geometric and material properties of the columns are
shown in Table 2. The shear response envelope is calculated for each 3.1. Comparison of calculated and measured response
column using Equations (1) through (6) and critical calculated param­
eters are presented in Table 3. Specimen U1 was tested under zero axial The calculated cyclic shear responses of test columns are compared
load [21]. Since Equations (1) through (6) were developed for columns with the experimental data in Fig. 8. The peak strength is accurately
with axial load, the axial load ratio P/fc Ag in Eq. (1), 2 and 4 is set equal

calculated with the proposed model for Specimen-1, which was tested by
to 0.05 for U1. The peak shear strength of the RC test columns, Vn are Sezen and Moehle [24]. The calculated pre-peak unloading and
calculated from Eq. (3). Peak flexural moment or plastic hinge moment reloading slopes and the post-peak response degradation compare
reasonably well with the experimental data. The proposed model

8
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

Fig. 9. Comparison of measured and calculated lateral force-total lateral displacement for Specimen-1, a) without shear spring, and b) with shear spring.

slightly underestimates the peak strength of Specimen-2 in the positive concluded that the proposed model can accurately predict the cyclic
loading direction. The proposed model simulates unloading and shear response of RC columns. Proposed model does not sufficiently
reloading cycles relatively accurately until the final shear failure is capture the shear response for U1. One reason for this may be the zero
triggered. axial load applied on this column. The proposed cyclic shear model was
Specimen-3 was tested under varying axial load. Different axial loads developed using data from columns with axial loads Sezen [22].
were used in each direction of the loading when the response envelopes
are calculated according to the proposed model. The comparison of
3.2. Lateral cyclic response of columns
experimental and calculated response confirms the accuracy and effi­
ciency of the proposed model. The peak shear strength is accurately
The lateral load-total lateral displacement relationship of Specimen-
calculated by the proposed model. The overall cyclic response is ob­
1 is calculated with and without shear springs. The results of both an­
tained reasonably well although the calculated cyclic response cannot
alyses are shown in Fig. 9. Three displacement components of the col­
accurately capture all unloading and reloading cycles. Pre-peak and
umn are separately calculated as shown in Fig. 1, and combined as
post-peak shear behavior of Specimen-4 is simulated by the proposed
described in Bicici and Sezen [2]. The model proposed in this paper was
model relatively accurately.
used to calculate the cyclic shear response shown in Fig. 8 using the
The calculated peak strength of U1 is 19% smaller than the measured
horizontal spring at the end of the column. The shear spring is repre­
strength. The pinching point of the cyclic behavior is overestimated by
sented in OpenSees by the new uniaxial material model proposed in this
the proposed model and the overall cyclic response is not captured well.
paper. The column flexural response is simulated using distributed
However, initial cycles are accurately captured by the proposed model
plasticity. Bar slip displacement was calculated using rotational springs
for U1. During the tests of U1 and U6, the columns reached the
at the ends of the column.
maximum total lateral displacement at 84 and 89 mm, respectively [21].
Fig. 9a shows the calculated and measured lateral load-total lateral
The corresponding measured shear displacement components of U1 and
displacement relationship of Specimen-1 with flexural element and
U6 are 13 and 6 mm, respectively. Thus, it should be noted that the
rotational springs. The total lateral displacements presented in Fig. 9b
contribution of shear to the overall response was less compared to the
include the shear displacements modeled using the shear spring at the
flexural response in columns U1 and U6. The peak strength of U6 is
end of the column. Total lateral displacement at shear failure is calcu­
slightly overestimated by the proposed model. The unloading and
lated as 2.78 in. (70.6 mm) using Eq. (8), which captures the actual shear
reloading cycles are closely captured by the proposed model.
failure reasonably well in the degrading response region. This model
By comparing the calculated and experimental data, it can be
was proposed by Elwood and Moehle [5,4], in which ρ˝ is transverse

Fig. 10. Comparison of measured and calculated lateral force-total lateral displacement for Specimen-2 (left) and Specimen-3 (right).

9
E. Bicici and H. Sezen Engineering Structures 247 (2021) 113211

reinforcement ratio, υ is maximum nominal shear stress (Vtest /bd) (in psi [2] Bicici E, Sezen H. Modeling of lateral response of reinforced concrete columns in
existing buildings. In: 16th World Conference On Earthquake Engineering,
units), and P is the axial load.
Santiago, Chile, Paper No. 284. 2017.
Δs 3 1 υ 1 P 1 [3] Bicici E. Development of Computational Model for Cyclic Response of Reinforced
= + 4ρ˝ − √̅̅̅̅ − ≥ (8) Concrete Columns. Ph.D. Dissertation. Columbus, Ohio: The Ohio State University;
L 100 500 f ′ 40 Ag fc′ 100 2018. p. 292.
c
[4] Elwood JK, Moehle JP. Axial capacity model for shear-damaged columns. ACI
Struct J 2005;102(4):578.
The comparison of total lateral responses in Fig. 9 shows the
[5] Elwood KJ, Moehle JP. Drift capacity of reinforced concrete columns with light
importance of considering the effect of shear displacement in modeling transverse reinforcement. Earthq Spectra 2005;21(1):71–89.
the lateral behavior of an RC column. As demonstrated in the figure, the [6] Elwood KJ. Modeling failure in existing reinforced concrete columns. Can J Civ Eng
2004;31:846–59.
lateral strength degradation is triggered and the cyclic behavior is
[7] Feng D-C, Xu J. An efficient fiber beam-column element considering flexure–shear
dominated by the shear spring. Additionally, Fig. 9b shows that the peak interaction and anchorage bond-slip effect for cyclic analysis of RC structures. Bull
lateral strength and the post-peak degradation of Specimen-1 are Earthq Eng 2018;16(11):5425–52.
calculated relatively accurately. However, the pre-peak stiffness is [8] Gerin M, Adebar P. Accounting for shear in seismic analysis of concrete structures.
In: Proceedings, 13th World Conf. on Earthquake Engineering, Vancouver, Paper
slightly overestimated. Furthermore, Fig. 10 compares the measured No. 1747, 2004.
and calculated lateral response of Specimen-2 and -3. For both columns, [9] Gu X, Hua J, Cai M. Seismic responses of reinforced concrete intermediate short
the peak lateral strength is accurately calculated. However, in the columns failed in different modes. Eng Struct 2020;206:110–73.
[10] Azadi Kakavand MR, Sezen H, Taciroglu E. Data-driven models for predicting the
negative loading side of Specimen-2, the peak strength is under­ shear strength of rectangular and circular reinforced concrete columns. ASCE J
estimated. Besides, the calculated cyclic slopes for both columns show Struct Eng 2021;147(1):04020301.
agreement with experiment results. The lateral displacement at axial [11] LeBorgne MR, Ghannoum WM. Analytical element for simulating lateral-strength
degradation in reinforced concrete columns and other frame members. ASCE J
failure is underestimated for Specimen-3. Struct Eng 2014;140(7):1–12.
[12] Lehman DE, Moehle JP. Seismic performance of well-confined concrete bridge
4. Conclusions columns. Report No. PEER-1998/01, Pacific Earthquake Engineering Research
Center, University of California, Berkeley, 2000.
[13] Li Y, Weng P, Hwang S. Seismic performance of reinforced concrete intermediate
The lateral displacement of an RC column is a combination of three short columns failed in shear. ACI Struct J 2019;116(3):195–206.
displacement components: flexure, longitudinal bar slip, and shear dis­ [14] Lodhi MS, Sezen H. Estimation of monotonic behavior of reinforced concrete
columns considering shear-flexure-axial load interaction. Earthquake Eng Struct
placements. For the columns designed to meet the requirements of
Dyn 2012;41:2159–75.
modern seismic codes, the flexural behavior is dominant and shear [15] McKenna F, Fenves GL, Scott MH. Open system for earthquake engineering
displacement can be small. However, the shear displacements can be simulation. Pacific Earthquake Engineering Research Center, University of
significantly large in older RC columns or the new columns designed California, Berkeley, 2004.
[16] Mergos PE, Kappos AJ. A distributed shear and flexural flexibility model with
without following the seismic design codes. Thus, the shear behavior shear-flexure interaction for R/C members subjected to seismic loading.
must be accurately modeled and included in analysis of columns with Earthquake Eng Struct Dyn 2008;37(12):1349–70.
low ductility. However, there is still no consensus about the mechanism [17] Mergos PE, Kappos AJ. A gradual spread inelasticity model for R/C beam–columns,
accounting for flexure, shear and anchorage slip. Eng Struct 2012;44:94–106.
and modeling of pure shear behavior of RC columns mainly due to very [18] Pan Z, Guner S, Vecchio FJ. Modeling of interior beam-column joints for nonlinear
limited experimental evidence. analysis of reinforced concrete frames. Eng Struct 2017;142:182–91.
In this study, an easy-to-use cyclic shear displacement model is [19] Patwardhan C. Shear Strength and Deformation modeling of Reinforced Concrete
Columns. M.S. Thesis. Columbus: The Ohio State University; 2005. p. 176.
proposed for RC columns regardless of the failure mode. After defining [20] Pincheria JA, Dotiwala FS, D’Souza JT. Seismic analysis of older reinforced
an envelope curve for the shear response, rules are developed for the concrete columns. Earthq Spectra 1999;15(2):245–72.
unloading and reloading portions of the cyclic response based on the [21] Saatcioglu M, Ozcebe G. Response of reinforced concrete columns to simulated
seismic loading. ACI Struct J 1989;86:3–12.
envelope curve. The novelty of this cyclic model is that the analysis and
[22] Sezen H. Shear deformation model for reinforced concrete columns. Struct Eng
modeling parameters need not be re-calculated or updated during Mech 2008;28(1):39–52.
analysis once the monotonic shear response or envelope is defined using [23] Sezen H, Moehle JP. Shear strength model for lightly reinforced concrete columns.
ASCE J Struct Eng 2004;103(1):1692–703.
a set of equations at the beginning. A new uniaxial material model is
[24] Sezen H, Moehle JP. Seismic tests of concrete columns with light transverse
created in open source software, OpenSees to represent the shear reinforcement. ACI Struct J 2006;103(6):842–9.
response with a horizontal spring in the column model. Then, the cyclic [25] Sezen H, Chowdhury T. Hysteretic model for reinforced concrete columns
shear displacements of previously tested columns are calculated to including the effect of shear and axial load failure. J Struct Eng 2009;135(2):
139–46.
validate the proposed model. The comparison of measured and calcu­ [26] Baradaran Shoraka M, Elwood KJ. Mechanical model for nonductile reinforced
lated shear response shows the efficiency and relatively good accuracy concrete columns. J Earthquake Eng 2013;17(7):937–57.
of the proposed cyclic shear displacement model. Cyclic lateral response [27] Takeda T, Sozen MA, Nielsen NN. Reinforced concrete response to simulated
earthquakes. J Struct Divis 1970;96(12):2557–73.
of one sample column (Specimen-1) is analyzed with and without shear [28] Vecchio FJ, Collins MP. The modified compression-field theory for reinforced
displacements to illustrate the importance of incorporation of shear concrete elements subjected to shear. ACI J 1986;83(2):219–31.
behavior in the model to accurately capture the shear failure of columns. [29] Wang Z, Wang J, Zhu J, Zhang J. A simplified method to assess seismic behavior of
reinforced concrete columns. Struct Concr 2020;21(1):151–68.
[30] Xu S-Y, Zhang J. Hysteretic shear-flexure interaction model of reinforced concrete
Declaration of Competing Interest columns for seismic response assessment of bridges. Earthquake Eng Struct Dyn
2011;40(3):315–37.
[31] Yu B, Ning C, Li B. Hysteretic model for shear-critical reinforced concrete columns.
The authors declare that they have no known competing financial
J Struct Eng 2016;142(9):04016056.
interests or personal relationships that could have appeared to influence [32] Xu J-G, Feng D-C, Wu G, Cotsovos DM, Lu Y. Analytical modeling of corroded RC
the work reported in this paper. columns considering flexure-shear interaction for seismic performance assessment.
Bull Earthq Eng 2020;18(5):2165–90.

References

[1] ACI 318. Building Code Requirements for Structural Concrete (ACI 318-19) and
Commentary (ACI 318R-19), Farmington Hills, Michigan, 2019.

10

You might also like