Model for the Lateral Behavior

of Reinforced Concrete Columns

Including Shear Deformations
Eric J. Setzlera) and Halil Sezena)

This research is focused on modeling the behavior of reinforced concrete

columns subjected to lateral loads. Deformations due to flexure, reinforcement
slip, and shear are modeled individually using existing and new models.
Columns are classified into five categories based on a comparison of their
predicted shear and flexural strengths, and rules for combining the three
deformation components are established based on the expected behavior of
columns in each category. Shear failure in columns initially dominated by
flexural response is considered through the use of a shear capacity model. The
proposed model was tested on 37 columns from various experimental studies.
In general, the model predicted the lateral deformation response envelope
reasonably well. 关DOI: 10.1193/1.2932078兴

INTRODUCTION
Numerous reinforced concrete buildings exist in the United States and around the
world that do not have sufficient detailing to ensure satisfactory performance in earth-
quakes. Past earthquakes have caused widespread damage to reinforced concrete struc-
tures that were not designed according to modern seismic design codes. Many of these
structures still exist in seismically active areas, and they may be susceptible to major
damage or collapse within their expected lifetimes. It is possible to retrofit an existing
structure to improve its shear strength and flexural deformation capacity, allowing it to
perform satisfactorily in an earthquake. This requires the ability to model the as-built
capacity of the structure, so that additional strength and/or deformation capacity require-
ments can be determined.
Columns are often the most critical components of earthquake damage-prone struc-
tures. The goal of the research reported here is to develop a model that can serve as a
response envelope for the behavior of a reinforced concrete column subjected to axial
and cyclic lateral loading. While the primary motivation and focus of this research is the
modeling of lightly reinforced columns that experience flexural yielding followed by
shear failure such as those described above, the model is general enough such that it is
also applicable for columns failing in shear (e.g., very short columns) or columns de-
veloping plastic hinges and failing in flexure (e.g., well-reinforced long columns).

a)
Department of Civil and Environmental Engineering and Geodetic Science, The Ohio State University, 470
Hitchcock Hall, 2070 Neil Ave., Columbus, OH 43210

493
Earthquake Spectra, Volume 24, No. 2, pages 493–511, May 2008; © 2008, Earthquake Engineering Research Institute
494 E. J. SETZLER AND H. SEZEN

Figure 1. Components of lateral deformation in a reinforced concrete column.

LATERAL DEFORMATION COMPONENTS

The total lateral deformation of a column subjected to lateral loads at its ends is
comprised of deformations due to three response mechanisms: flexure, reinforcement
slip at the column end(s), and shear. The interaction of these components determines the
overall behavior of a column. These deformations are depicted schematically in Figure 1.
Each deformation component can be modeled individually.

FLEXURAL DEFORMATIONS
The flexural response of a reinforced concrete section can be calculated through a
moment-curvature analysis. The moment-curvature analysis uses constitutive models for
the concrete and reinforcing steel that can capture the nonlinear behavior of each mate-
rial accurately. Moment-curvature analysis is used to calculate sectional response in this
research. A steel constitutive model with a yield plateau and nonlinear strain-hardening
region is specified to match the experimentally observed behavior of the reinforcing
steel. Concrete confinement is considered by using the constitutive model by Mander et
al. (1988) to model the core and cover concrete separately. The full strain capacity of the
confined and unconfined concrete was used, as calculated by the Mander et al. model.
The specified maximum unconfined concrete strain was 0.006.
For an applied lateral load at the column end, the moment can be determined at any
point in the column. Then, the moment-curvature relationship can be used to determine
the curvature distribution over the column height. The lateral displacement of a column
due to flexural deformations, ⌬f, can be calculated by integrating the curvature distribu-
tion over the height of the column as follows:
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 495

⌬f = 冕 ␾共x兲xdx
L

0
共1兲

where ␾共x兲 is the curvature distribution, x is measured along the axis of the column, and
L is the height of the column. In this research, Equation 1 is used to compute the flexural
deformations for lateral loads up to the load that initiates yielding in the longitudinal
steel. After yielding has occurred, the flexural deformations are calculated from a plastic
hinge model as

⌬f = ⌬f,y + 共␾ − ␾y兲Lp a −冉 冊Lp

2
共2兲

where ⌬f,y is the flexural deformation at yield, calculated using the integration method
(Equation 1), ␾ is the curvature at the column end, and ␾y is the curvature at yield. a is
the shear span, and is equal to L for a cantilever column or L / 2 for a column with fixed
supports at both ends. For columns with fixed supports at both ends, the second term on
the right side of Equation 2 must be multiplied by 2. The plastic hinge length, Lp, is
taken as one half of the total section depth per the recommendations of Moehle (1992).
Complete details of this flexural deformation model, including the constitutive models
and moment-curvature analysis, are given in Setzler (2005).

REINFORCEMENT SLIP
When a reinforcing bar embedded in concrete is subjected to a tensile force, strain
accumulates over the embedded length of the bar. This strain causes the reinforcing bar
to slip relative to the concrete in which it is embedded. Slip of a column’s reinforcing
bars in the anchoring concrete (i.e., the footing or beam-column joint) will cause rigid-
body rotation of the column, as shown in Figure 1. This rotation is not accounted for in
a flexural analysis, where the column ends are assumed to be fixed.
The bar slip model used in this study was originally developed by Sezen and Moehle
(2003), and includes further developments by Sezen and Setzler (2008). It is illustrated
in Figure 2. This model assumes a stepped function for bond stress between the concrete
and reinforcing steel over the embedment length of the bar. Based on experimental ob-
servations (Sezen 2002), the bond stress is taken as 12冑fc⬘ psi for elastic steel strains (ub
in Figure 2) and 6冑fc⬘ psi for inelastic steel strains 共u⬘b兲, where fc⬘ is the concrete com-
pressive strength. The rotation due to slip, ␪s, is calculated as

slip
␪s = 共3兲
d−c
where slip is the extension of the outermost tension bar from the column end (Figure 2),
and d and c are the distances from the extreme compression fiber to the centroid of the
tension steel and the neutral axis, respectively. The column lateral displacement is equal
to the product of the slip rotation and the column length.
496 E. J. SETZLER AND H. SEZEN

Figure 2. Proposed slip rotation model.

SHEAR DEFORMATIONS
Shear displacements are calculated in the proposed model using a combination of
two existing models. The computer program Response-2000 (Bentz 2000) uses Modified
Compression Field Theory (Vecchio and Collins 1986) to compute the monotonic shear
behavior of cracked concrete. It is used in the proposed model to compute the shear
force-deformation relationship up to the attainment of peak strength. After this point, a
shear model by Patwardhan (2005) is adopted. The proposed shear model is shown in
Figure 3. The peak strength, Vpeak, is the maximum strength from Response-2000. The

Figure 3. Proposed shear model.

MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 497

drift at the onset of shear failure, ⌬v,u, is adopted from Gerin and Adebar (2004), and is
calculated as

冉
⌬v,u = 4 − 12
vn
冊
fc⬘
⌬v,n 共4兲

where vn is the shear stress at peak strength (vn = Vpeak / bd, where b is the column width)
and ⌬v,n is the maximum drift at peak strength computed by Response-2000. The drift at
axial load failure, ⌬v,f, is calculated as

⌬v,f = ⌬ALF − ⌬f,f − ⌬s,f 艌 ⌬v,u 共5兲

where ⌬ALF is the total drift at axial load failure, and ⌬f,f and ⌬s,f are the flexural and slip
deformations, respectively, at the point of axial load failure. The total drift at axial load
failure under cyclic loading is calculated using the axial capacity model by Elwood and
Moehle (2005a):

⌬ALF 4 1 + tan2 ␪

冉 冊
= 共6兲
L 100 s
tan ␪ + P
Asvfyvdc tan ␪
where ␪ is the angle of the shear crack, P is the axial load, Asv is the area of transverse
steel with yield strength fyv at spacing s, and dc is the depth of the core concrete, mea-
sured to the centerlines of the transverse reinforcement. In the derivation, ␪ was assumed
to be 65 degrees. The rules governing the post-peak column behavior, given in the next
section, define the values for ⌬f,f and ⌬s,f. Complete details of the proposed shear model
are available in Setzler (2005).

TOTAL LATERAL RESPONSE

The total lateral response of a reinforced concrete column can be modeled using a
set of springs in series. The flexure, bar slip, and shear deformation models discussed
above are each modeled by a spring. Each spring is subjected to the same force, and the
total displacement response is the sum of the responses of each spring. Figure 4 illus-
trates the spring model schematically. Up to the peak strength of the column, the three
deformation components are simply added together to predict the total response. Rules
are established for the post-peak behavior of the springs based on a comparison of the
shear strength 共Vn兲, the yield strength 共Vy兲, and the flexural strength 共Vp兲. The yield
strength is defined as the lateral load corresponding to the first yielding of the tension
bars in the column during flexural analysis. The flexural strength is the lateral load cor-
responding to the maximum moment sustainable by the column cross section. Both val-
ues are obtained from the moment-curvature analysis. The shear strength is calculated
from the equation proposed by Sezen and Moehle (2004):
498 E. J. SETZLER AND H. SEZEN

Figure 4. Spring representation of the proposed model for a column with fixed ends.

Vn = k 冋冉 冑
6冑fc⬘
a/d
1+
P
6冑fc⬘Ag
冊 0.80Ag +
Asvfyvd
s
册 共7兲

where Ag is the gross cross sectional area and a / d is the aspect ratio. k is a factor related
to the displacement ductility, which is the ratio of the maximum displacement to the
yield displacement. k is equal to 1.0 for displacement ductilities less than 2, it is equal to
0.7 for displacement ductilities greater than 6, and it varies linearly for intermediate dis-
placement ductilities. k is taken as 1.0 in the proposed model for classification purposes,
because the classification system outlined below is based on the initial, or low-ductility
shear and flexural strengths.
Figure 5 plots the flexural response and shear strength for five fictitious columns.
These columns all have the same shear strength, but different flexural and yield
strengths. By comparing Vn, Vy, and Vp, these columns can be classified into one of the
following five categories.
Category I: Vn ⬍ Vy The shear strength is less than the lateral load causing yielding
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 499

Figure 5. Lateral load-displacement ductility relationship for columns in each category.

in the tension steel. The column fails in shear while the flexural behavior remains elastic.
Category II: Vy ⱕ Vn ⬍ 0.95Vp The shear strength is greater than the yield strength,
but less than the flexural strength of the column. The column fails in shear, but inelastic
flexural deformation occurring prior to shear failure affects the post-peak behavior.
Category III: 0.95Vp ⱕ Vn ⱕ 1.05Vp The shear and flexural strengths are essentially
identical. Due to the inherent variability in the models used to predict the strengths, it is
not possible to predict conclusively which mechanism will govern the peak response.
Shear and flexural failure are assumed to occur “simultaneously,” and both mechanisms
contribute to the post-peak behavior.
Category IV: 1.05Vp ⬍ Vn ⱕ 1.4Vp The shear strength is greater than the flexural
strength of the column. The column experiences large flexural deformations potentially
leading to a flexural failure. Inelastic shear deformations affect the post-peak behavior,
and shear failure may occur as displacements increase.
Category V: Vn ⬎ 1.4Vp The shear strength is much greater than the flexural strength
of the column. The column fails in flexure while the shear behavior remains elastic.
Category IV and V specimens are those that are expected to fail in flexure, because
their flexural strength is lower than their initial shear strength. However, the shear
strength of a column decreases as displacements increase (Sezen and Moehle 2004). If
the initial shear strength is greater than the flexural strength, but shear degradation
causes the shear strength to become less than the flexural strength, shear failure could
occur in the column after the flexural strength has been reached. Category IV specimens
are those where shear failure could occur at high displacements, while Category V speci-
mens are those whose shear strength is high enough that shear failure is not expected
500 E. J. SETZLER AND H. SEZEN

Figure 6. Behavior of (a) shear spring and (b) flexure and slip springs for each category.

even as the shear strength decreases. The factor k in the Sezen-Moehle shear strength
model (Equation 7) can take a minimum value of 0.7. Therefore, if the shear strength is
less than 1 / 0.7⬵ 1.4 times the flexural strength, shear failure is possible and the column
will be in Category IV. It is noted that these response categories are defined for the pro-
posed macro-scale model shown in Figure 4, and are not intended for finite element
analysis.
The classification system given above describes the expected column behavior based
on a comparison of the shear and flexural strengths. From these descriptions, rules gov-
erning the flexure, slip, and shear springs are defined. These rules are described below,
and are illustrated in Figure 6. As stated above, the total deformation is calculated by
summing the three deformation components for the initial response, up to the peak
strength of the column, for columns in all categories.
Category I: The peak strength of the column is the shear strength, as calculated in
the proposed shear model. After the peak strength is reached, the shear behavior domi-
nates the response. As the column strength decreases, shear deformations continue to
increase according to the shear model (solid line in Figure 6a), while the flexure and slip
springs unload along their initial responses (dashed line in Figure 6b). The post-peak
deformation at any lateral load level is the sum of the post-peak shear deformation and
the pre-peak flexural and slip deformations corresponding to that load.
Category II: The peak strength of the column is the shear strength calculated from
the proposed shear model. As the column strength decreases, shear deformations con-
tinue to increase according to the shear model (solid line in Figure 6a), but the flexure
and slip springs are locked at their values at peak strength (dot-dashed line in Figure 6b).
The post-peak deformation at any lateral load level is the sum of the flexural and slip
deformations at peak strength and the post-peak shear deformation corresponding to that
load.
Category III: The peak strength is the smaller of the shear strength and the flexural
strength. As the column strength decreases, all deformations continue to increase ac-
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 501

cording to their individual models (solid line in Figures 6a and 6b). The post-peak de-
formation at any lateral load level is the sum of the post-peak flexure, slip, and shear
deformations corresponding to that load.
Category IV: The peak strength of the column is the flexural strength, calculated in
the flexure model. As the column strength decreases, flexural and slip deformations con-
tinue to increase according to their models (solid line in Figure 6b), but the shear spring
is locked at its value at peak strength (dot-dashed line in Figure 6a). The post-peak de-
formation at any lateral load level is the sum of the post peak flexural and slip defor-
mations corresponding to that load and the shear deformation at peak strength. The col-
umn may experience a shear failure after being subjected to large deformations. Shear
failure is predicted through the use of a shear capacity model described below.
Category V: The peak strength of the column is the flexural strength calculated from
the flexure model. If the column strength decreases, flexural and slip deformations con-
tinue to increase according to their models (solid line in Figure 6b), while the shear
spring unloads with an unloading stiffness equal to its initial stiffness (dashed line in
Figure 6a). The post-peak deformation at any lateral load level is the sum of the post-
peak flexural and slip deformations and the pre-peak shear deformation corresponding
to that load.

SHEAR FAILURE DUE TO HIGH DISPLACEMENT DUCTILITY

It has been established that shear strength decreases as displacement ductility in-
creases, potentially leading to shear failure in columns initially dominated by flexure
(Sezen and Moehle 2004). However, this possibility cannot be accounted for in a simple
combination of the three component models, as outlined above.
Elwood (2004) proposed the idea of imposing a shear failure surface on the lateral
load-total displacement diagram, as shown in Figure 7. In this proposal, if the response
of the column intersects the shear failure surface, shear failure will occur. This model is
intended for the prediction of shear failure after the occurrence of flexural yielding. The
shear failure surface is defined by the empirical drift capacity model proposed by El-
wood and Moehle (2005b):

⌬SF 3 1 v 1 P 1
= + 4␳v − − 艌 共8兲
L 100 500 冑fc⬘ 40 Agfc⬘ 100

where ⌬SF is the drift at shear failure, ␳v is the transverse reinforcement ratio, and v is
the nominal shear stress. fc⬘ and v have units of psi.
The shear failure surface proposed by Elwood is implemented in the proposed model
to account for delayed shear failure following inelastic flexural response for Category IV
specimens. The drift at shear failure is calculated from Equation 8 using the peak model
strength to calculate the shear stress (i.e., v = Vp / bd). If the predicted drift in the unmodi-
fied model exceeds the calculated drift at shear failure, shear failure is assumed to have
occurred. The model is modified to decrease linearly from the point of shear failure to
zero strength at the drift at axial load failure, calculated from Equation 6.
502 E. J. SETZLER AND H. SEZEN

Figure 7. Shear failure surface model developed by Elwood (2004).

Equation 7 should not be used to predict the displacement at shear failure as it is

intended for calculation of Vn using displacement ductility, because the result is overly
sensitive to small variations in the input parameters. In other words, if there is an error
in prediction of displacement ductility in Equation 7, its effect on the predicted shear
strength will be relatively small. However, small changes in Vn may result in large varia-
tions in the corresponding displacement ductility.

COMPARISON OF MODEL AND TEST DATA

Thirty-seven column tests from eight different researchers were assembled from the
Pacific Earthquake Engineering Research Center’s Structural Performance Database
(Eberhard 2003). These were selected to cover a wide range of shear and flexural
strengths, aspect ratios, and transverse reinforcement ratios. Table 1 lists key properties
for the test columns. (In Table 1, fy is the yield strength of the longitudinal reinforcement
and ␳l is the longitudinal reinforcement ratio.) Each column was modeled using the
component and overall models proposed in this research. Table 2 lists the calculated
shear strength, yield strength, and flexural strength for each column. Each column was
assigned a category from I to V based on the rules outlined previously. The columns are
sorted in order of increasing Vn / Vp ratio, and hence from Category I to Category V, in
Table 2.
The first twelve columns in Table 1 were part of a test matrix and research program
at the University of California, Berkeley, to investigate the effect certain parameters on
the seismic response of older building columns. As shown in Table 2, these columns had
failures marked by shear effects. Similarly, the next set of six specimens, or twelve col-
umns on each side of six stubs, tested by Wight and Sozen (1975) was also selected
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 503

Table 1. Properties of test columns

L b d f⬘c fy fyv ␳l ␳v s/d P

Column Ref.a Typeb in. in. in. a/d psi ksi ksi % % in. kip P / Agf⬘c
2CLD12 1 DC 116 18 15.4 3.76 3060 63 69 2.88 0.175 0.78 150 0.151
2CHD12 1 DC 116 18 15.4 3.76 3060 63 69 2.88 0.175 0.78 600 0.605
c c
2CVD12 1 DC 116 18 15.4 3.76 3030 63 69 2.88 0.175 0.78
2CLD12M 1 DC 116 18 15.4 3.76 3160 63 69 2.88 0.175 0.78 150 0.147
2CLH18 2 DC 116 18 15.6 3.74 4800 48 58 2.23 0.068 1.15 113 0.073
2SLH18 2 DC 116 18 15.6 3.74 4800 48 58 2.23 0.068 1.15 113 0.073
3CLH18 2 DC 116 18 15.5 3.74 3710 48 58 3.63 0.068 1.16 113 0.094
3SLH18 2 DC 116 18 15.5 3.74 3710 48 58 3.63 0.068 1.16 113 0.094
2CMH18 2 DC 116 18 15.6 3.74 3730 48 58 2.23 0.068 1.15 340 0.281
3CMH18 2 DC 116 18 15.5 3.74 4010 48 58 3.63 0.068 1.16 340 0.262
3CMD12 2 DC 116 18 15.5 3.74 4010 48 58 3.63 0.175 0.77 340 0.262
3SMD12 2 DC 116 18 15.5 3.74 3730 48 58 3.63 0.175 0.77 340 0.281
25.033E 3 DE 34.5 6 10 3.45 4880 72 50 2.95 0.327 0.5 25 0.071
25.033W 3 DE 34.5 6 10 3.45 4880 72 50 2.95 0.327 0.5 25 0.071
40.033AE 3 DE 34.5 6 10 3.45 5030 72 50 2.95 0.327 0.5 42.5 0.117
40.033AW 3 DE 34.5 6 10 3.45 5030 72 50 2.95 0.327 0.5 42.5 0.117
40.033E 3 DE 34.5 6 10 3.45 4870 72 50 2.95 0.327 0.5 40 0.114
40.033W 3 DE 34.5 6 10 3.45 4870 72 50 2.95 0.327 0.5 40 0.114
40.048E 3 DE 34.5 6 10 3.45 3780 72 50 2.95 0.467 0.35 40 0.147
40.048W 3 DE 34.5 6 10 3.45 3780 72 50 2.95 0.467 0.35 40 0.147
40.067E 3 DE 34.5 6 10 3.45 4840 72 50 2.95 0.654 0.25 40 0.115
40.067W 3 DE 34.5 6 10 3.45 4840 72 50 2.95 0.654 0.25 40 0.115
40.092E 3 DE 34.5 6 10 3.45 5150 72 46 2.95 0.920 0.4 40 0.108
40.092W 3 DE 34.5 6 10 3.45 5150 72 46 2.95 0.920 0.4 40 0.108
CUS 4 DC 36 9 14.4 1.23 5060 64 60 3.34 0.324 0.24 120 0.162
CUW 4 DC 36 16 7.4 2.40 5060 64 60 3.65 0.364 0.47 120 0.162
U1 5 C 39.4 13.8 12 3.28 6320 62 68 3.80 0.272 0.49 0 0
U3 5 C 39.4 13.8 12 3.28 5050 62 68 3.80 0.544 0.25 135 0.141
U4 5 C 39.4 13.8 12 3.28 4640 64 68 3.80 0.814 0.17 135 0.153
U6 5 C 39.4 13.8 12 3.28 5410 64 62 3.80 0.835 0.22 135 0.131
A1 6 C 91.9 15 22.3 4.12 3950 65 62 2.38 0.308 0.19 138 0.098
A2 6 C 91.9 15 22.3 4.12 3950 65 62 2.38 0.308 0.19 338 0.239
C1-1 7 C 55.1 15.7 13.8 4.00 3610 72 67 2.43 0.643 0.14 101 0.113
C1-2 7 C 55.1 15.7 13.8 4.00 3870 72 67 2.43 0.643 0.14 152 0.158
C1-3 7 C 55.1 15.7 13.8 4.00 3780 72 67 2.43 0.643 0.14 202 0.216
SC3 8 C 48 36 15.6 3.07 3180 63 58 2.23 0.096 1.03 0 0
SC9 8 C 48 18 33.6 1.43 2320 63 58 2.08 0.077 0.48 0 0
a
1: Sezen (2002); 2: Lynn (2001); 3: Wight and Sozen (1975); 4: Umehara and Jirsa (1984); 5: Saatcioglu and
Ozcebe (1989); 6: Wehbe et al. (1999); 7: Mo and Wang (2000); 8: Aboutaha et al. (1999)
b
C = cantilever, DC= double curvature, DE= double-ended
c
Axial load varied from −60 kip to 600 kip
504 E. J. SETZLER AND H. SEZEN

Table 2. Shear and flexural strengths of test columns

Column Vna,b V yb V pb Vn / Vp Category SPDc Vmodelb Vtestb Vmodel / Vtest

SC3 88.6 108.7 168.2 0.53 I S 103.9 101.2 1.03
SC9 132.3 143.0 236.2 0.56 I S 109.8 144.5 0.76
3CLH18 46.4 54.7 68.6 0.68 I S 46.8 62.3 0.75
3SLH18 46.4 54.7 68.6 0.68 I S 46.8 60.7 0.77
3CMH18 62.1 73.9 80.4 0.77 I S 57.7 73.7 0.78
CUS 91.7 93.1 115.1 0.80 I S 62.7 73.3 0.86
CUW 59.4 65.6 72.2 0.82 I S 65.8 60.1 1.09
U1 52.7 42.7 65.2 0.81 II F 55.2 62.1 0.89
25.033E 19.3 18.2 20.9 0.92 II S 20.0 19.1 1.05
25.033W 19.3 18.2 20.9 0.92 II S 20.0 20.3 0.99
2CMH18 61.4 59.2 63.6 0.97 III F-S 59.2 68.8 0.86
2CVD12(T)d 56.5 39.3 58.2 0.97 III F-S 53.5 55.4 0.97
3CMD12 79.2 73.9 81.4 0.97 III S 81.3 80.0 1.02
40.033E 20.5 19.8 21.0 0.98 III F-S 21.0 20.4 1.03
40.033W 20.5 19.8 21.0 0.98 III F-S 21.0 22.8 0.92
40.033AE 20.8 20.1 21.3 0.98 III F-S 21.3 22.2 0.96
40.033AW 20.8 20.1 21.3 0.98 III F-S 21.3 22.6 0.94
2CLH18 50.5 40.9 51.7 0.98 III F-S 46.5 54.1 0.86
2SLH18 50.5 40.9 51.7 0.98 III F-S 46.5 52.4 0.89
2CLD12 69.0 58.3 70.4 0.98 III F-S 70.4 70.8 0.99
2CLD12M 69.0 58.3 70.4 0.98 III F-S 70.4 66.2 1.06
3SMD12 78.1 73.1 79.4 0.98 III F-S 79.4 82.5 0.96
40.048E 23.9 19.3 22.1 1.08 IV F-S 20.8 22.6 0.92
40.048W 23.9 19.3 22.1 1.08 IV F-S 20.8 21.3 0.98
40.067E 30.3 19.8 24.3 1.25 IV F-S 21.5 20.7 1.04
40.067W 30.3 19.8 24.3 1.25 IV F-S 21.5 20.6 1.04
2CVD12(C)d 92.3 66.7 71.7 1.29 IV F-S 71.7 67.6 1.06
2CHD12 92.3 66.7 71.7 1.29 IV F-S 71.7 80.7 0.89
U3 93.6 56.4 67.7 1.38 IV F 67.7 60.9 1.11
A1 97.0 49.1 67.8 1.43 V F 67.8 75.7 0.90
U6 118.3 57.4 81.3 1.46 V F 81.3 77.1 1.05
40.092E 36.3 19.9 24.3 1.49 V F-S 24.3 25.4 0.96
40.092W 36.3 19.9 24.3 1.49 V F-S 24.3 25.4 0.96
U4 123.2 56.4 80.5 1.53 V F 80.5 73.3 1.10
A2 128.2 62.8 74.7 1.72 V F 74.7 81.7 0.91
C1-3 124.5 51.1 59.5 2.09 V F 59.5 68.6 0.87
C1-2 121.8 47.3 57.6 2.11 V F 57.6 60.2 0.96
C1-1 117.8 43.0 55.2 2.13 V F 55.2 56.2 0.98
a
Shear strength calculated from Equation 7, assuming k = 1.0 mean: 0.95
b
All strengths are given in kips standard deviation: 0.10
c
Failure classification reported in Structural Performance Database (F = flexure, S = shear, and
F-S = flexure-shear)
d
共T兲 = axial tension portion of cycle, 共C兲 = axial compression
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 505

because the failure modes varied widely (Category II through V in Table 2). These speci-
mens were very appropriate to study the effect of shear on column behavior. Four col-
umns tested by Saatcioglu and Ozcebe (1989) also covered a wide range of response
(Categories II, IV, and V). In addition, these specimens were unique because both the
lateral load-shear displacement and lateral load-slip displacement relations were re-
ported. For this study, such cyclic experimental data were valuable in understanding the
deformation components and their contribution to the total deformations. To the authors’
knowledge, except for Reference 1, no such data are available for columns included in
the University of Washington database. The remaining nine columns had to be selected
to verify and populate certain failure categories. CUS and CUW (Umehara and Jirsa
1984) and SC3 and SC9 (Aboutaha et al. 1999) were tested by the same research group
at the University of Texas, Austin. These specimens had clear shear failure with very
limited or no flexural effects at failure. Thus, they were perfect candidates to include in
Category I. One of the factors affecting the selection of last five columns was their rela-
tively large aspect ratios (Wehbe et al. 1999; Mo and Wang 2000).
Some trends can be observed over the range of data. All columns with aspect ratios,
a / d, less than 2.5 are in Category I, and all those with aspect ratios of 4 or greater are
in Category V. All columns in the intermediate categories have aspect ratios between 2.5
and 4, although there is not an increasing trend from Category II to III to IV. Columns
U4, U6, and 40.092E/W are all in Category V despite having aspect ratios less than 4.
These columns have high transverse reinforcement ratios, which appears to prevent
shear failure and allow high levels of ductility to develop. Conversely, Columns SC3,
3CLH18, 3SLH18, and 3CMH18 have widely spaced transverse reinforcement 共s / d
ⱖ 1.0兲. This causes them to fail in shear despite aspect ratios greater than 3, and pushes
them into Category I.
The failure mode given in the Structural Performance Database (Eberhard 2003) is
listed in Table 2 for each column. There is generally good agreement between these re-
ported failure modes and the category assigned to the columns in the proposed model. It
should be noted that the placement of a column into Category V does not absolutely
preclude the possibility of shear failure under high ductility demand. The division be-
tween these categories of Vn ⬎ 1.4Vp was based on Equation 7 in this study. If the pro-
posed model is used in an engineering design application, it may be advisable to in-
crease this limit in order to achieve an acceptable factor of safety against shear failure.
Table 2 gives the maximum lateral strengths from the proposed model and the test
data. Vmodel is the peak lateral strength in the proposed model. It is the lesser of the
strength predicted by the flexure component model 共Vp兲 and the strength predicted by
the shear component model (Vpeak, as shown in Figure 3). For shear-dominant columns
or columns failing in shear without significant flexural damage, Vpeak should theoreti-
cally match Vn from Equation 7. Since Equation 7 and the shear component model use
different methods for predicting the shear strength, they do not agree exactly. The dif-
ference is generally small for the test columns; the mean of Vpeak / Vn is 1.02 with a stan-
dard deviation of 0.12. The ratio of predicted strength to observed strength is given for
each column, and the model predictions generally agree quite well with the test data.
The mean of Vmodel / Vtest is 0.95, and the standard deviation is 0.10.
506 E. J. SETZLER AND H. SEZEN

Experimental and predicted displacement capacities are given in Table 3 for the test
columns. The displacement ductilities, µtest, given in Table 3 were taken from the litera-
ture when available, and were calculated graphically for the remaining columns accord-
ing to the procedure suggested by Sezen and Moehle (2004). For the yield and ultimate
displacements and the associated displacement ductilities calculated using the graphical
procedure, the reported values are the average of the two displacements or displacement
ductilities corresponding to positive and negative load cycles. The mean ratio of pre-
dicted displacement ductility to observed displacement ductility, µmodel / µtest is 1.90, and
the standard deviation is 1.14. This indicates that the model does not predict the dis-
placements as well as strengths. A trend is noticeable in the ductility values, with higher
µ values typically corresponding to higher categories. However, columns tested under
high axial load (i.e., 2CHD12, 2CVD12, and 2CMH18) did not follow this trend. The
decrease in displacement ductility caused by high axial load has been documented in the
literature (e.g., Elwood and Moehle 2005b; Patwardhan 2005).
Lateral force-displacement relationships are shown in Figures 8 and 9 for eight of the
37 test columns modeled in this study. Comparisons for the other columns can be found
in Setzler (2005). The plots compare the response envelopes predicted by the proposed
model to cyclic test data reported in the literature for each column. At least one column
from each of the five categories is included in the figures. The model predicts reasonable
response envelopes for the columns examined in the study. For columns in Category IV
(Figure 9, plots (a) and (b)), the dashed lines show the proposed model before modifi-
cation for delayed shear failure. The solid line is the final model prediction, after con-
sideration of the Elwood shear failure surface (Equation 8). The shear failure surface
was used successfully in predicting the lateral response of Category IV columns. From
moment-curvature analysis, a sudden drop in lateral resistance was calculated for the
Category IV column, 2CHD12 which was subjected to very high axial load. Using the
shear failure surface model, a smoother response beyond shear failure was obtained in
Figure 9a. As discussed previously, Category V columns are those whose shear strengths
are high enough such that they are not expected to experience shear failure even at large
displacements. However, the Elwood shear failure surface and point of axial load failure
were computed for these columns for comparison purposes. As shown in Figure 9, plots
(c) and (d), it is not appropriate to modify the model using the Elwood shear failure
surface for Category V columns. The proposed model predicts the behavior of these col-
umns well without any shear failure modifications.

SUMMARY AND CONCLUSIONS

The focus of this research was the creation of a model that can predict the monotonic
lateral force-displacement relationship for reinforced concrete columns subjected to lat-
eral loading. The research concentrated on lightly reinforced columns that experience
flexure-shear failures. However, the model can be applied to columns with any ratio of
shear and flexural strengths. Therefore, it is applicable to columns that experience shear,
flexure, or flexure-shear failures.
The overall lateral deformation of a reinforced concrete column was modeled as
three springs in series, one for each of the deformation components. Column shear
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 507

Table 3. Experimental and predicted displacement capacities of test columns

Column Category ⌬y,testb ⌬u,testb µtest ⌬y,modelb ⌬u,modelb µmodel µmodel / µtest
SC3 I 0.27 0.96 3.50 0.42 1.64 3.90 1.11
SC9 I 0.30 0.38 1.27 0.14 1.20 8.46 6.66
3CLH18 I 0.75 1.20 1.58a 0.50 1.13 2.26 1.43
3SLH18 I 0.62 1.15 1.69a 0.50 1.13 2.26 1.34
3CMH18 I 0.89 1.20 2.14a 0.53 0.57 3.40 1.59
CUS I 0.20 0.38 1.95 0.12 0.54 4.50 2.31
CUW I 0.26 0.62 2.40 0.47 1.01 2.15 0.90
U1 II 0.67 2.09 3.12 0.25 1.37 5.43 1.74
25.033E II 0.44 1.23 2.77 0.29 1.59 5.48 1.98
25.033W II 0.42 1.18 2.82 0.29 1.59 5.48 1.94
2CMH18 III 0.65 1.20 1.94a 0.53 1.80 3.40 1.75
2CVD12(T)c III 1.13 3.41 3.01a 0.90 6.34 7.04 2.34
3CMD12 III 0.77 1.80 2.50a 0.70 1.82 2.60 1.04
40.033E III 0.48 1.27 2.65 0.30 2.13 7.10 2.68
40.033W III 0.44 1.58 3.55 0.30 2.13 7.10 2.00
40.033AE III 0.25 1.20 5.04 0.29 2.08 7.17 1.42
40.033AW III 0.24 1.23 5.46 0.29 2.08 7.17 1.31
2CLH18 III 0.59 3.00 4.17a 0.48 2.28 4.75 1.14
2SLH18 III 0.51 2.40 2.65a 0.48 2.28 4.75 1.79
2CLD12 III 1.03 2.97 2.88a 0.93 3.34 3.59 1.25
2CLD12M III 1.06 3.33 3.14a 0.93 3.34 3.59 1.14
3SMD12 III 0.89 1.80 2.73a 0.72 1.92 2.67 0.98
40.048E IV 0.47 1.68 3.55 0.33 1.34 4.06 1.14
40.048W IV 0.45 1.90 4.44 0.33 1.34 4.06 0.91
40.067E IV 0.42 2.34 5.64 0.30 1.63 5.43 0.96
40.067W IV 0.36 2.35 6.55 0.30 1.63 5.43 0.83
2CVD12(C)c IV 0.82 2.23 2.72a 0.73 1.69 2.32 0.85
2CHD12 IV 0.79 1.02 1.29a 0.62 1.53 2.47 1.91
U3 IV 0.63 1.77 2.81 0.29 1.56 5.38 1.91
A1 V 0.92 4.76 5.3a 0.53 4.36 8.24 1.55
U6 V 0.49 3.47 7.37 0.28 4.01 14.08 1.91
40.092E V 0.38 2.05 5.36 0.28 6.18 22.07 4.12
40.092W V 0.39 2.05 5.59 0.28 6.18 22.07 3.95
U4 V 0.47 3.44 7.43 0.30 4.30 14.14 1.90
A2 V 0.75 3.93 5.2a 0.53 3.22 5.67 1.09
C1-3 V 0.70 3.67 5.26a 0.45 7.03 15.63 2.97
C1-2 V 0.69 3.72 5.38a 0.44 7.33 16.66 3.10
C1-1 V 0.63 3.47 5.50a 0.44 7.56 17.18 3.12
a
Ductilities reported in literature mean: 1.90
b
All displacements are given in inches standard deviation: 1.14
c
共T兲 = axial tension portion of cycle, 共C兲 = axial compression
508 E. J. SETZLER AND H. SEZEN

Figure 8. Model predictions and test data for the lateral displacement of (a) SC3, (b) 25.033W,
(c) 3CMD12, and (d) 2CLD12.

strength was calculated using the shear strength equation proposed by Sezen and Moehle
(2004), taking the displacement ductility parameter k as 1.0. This shear strength was
compared to the yield and flexural strengths determined from the flexural analysis to
classify columns into one of five categories. Category I columns experience a pure shear
failure. Category II columns also fail in shear, but with flexural effects. Columns in Cat-
egory III fail in shear and flexure at nearly the same time. Category IV columns initially
fail in flexure or develop plastic hinges, but may experience shear failure as displace-
ment increases. Columns that experience pure flexural failures are in Category V. For
each category, expected behavior and rules for the combination of the deformation com-
ponents were presented.
Category IV specimens are those which are susceptible to shear failure after flexural
capacity is reached. A shear capacity model (Elwood 2004; Elwood and Moehle 2005b)
was used to predict the onset of delayed shear failure for these columns. If shear failure
was predicted, the model was modified and the strength is reduced linearly to the point
of axial load failure.
A database of 37 test columns was assembled, which covered a wide range of col-
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 509

Figure 9. Model predictions and test data for the lateral displacement of (a) 40.048W, (b)
2CHD12, (c) A1, and (d) U6.

umn geometries, properties, and shear to flexural strength ratios. Each of these speci-
mens was analyzed using the model proposed in this research. In general, the proposed
model did an acceptable job of predicting the response envelope for the cyclic test data.
There were several instances where the model predictions were poor, but the behavior of
most columns was represented well. The model also predicted the maximum strength of
the columns well overall. The average of the ratio of predicted strength to experimental
strength was 0.95, with a standard deviation of 0.10.
The classification system used in the combined lateral response model appeared to
represent the shear and flexural behaviors well. The shear capacity model proposed by
Elwood (2004) was used successfully to predict delayed shear failure in Category IV
columns. The proposed model was able to predict the experimental behavior best in the
flexure-shear failure range (Categories II through IV), the range for which it was in-
tended.
Predicted and experimental lateral deformation plots were given for eight of the 37
columns examined in this study. The experimental data was modeled reasonably well for
these columns.
510 E. J. SETZLER AND H. SEZEN

ACKNOWLEDGMENTS
This work was partially supported by a National Science Foundation Graduate Re-
search Fellowship.

REFERENCES
Aboutaha, R. S., Engelhardt, M. D., Jirsa, J. O., and Kreger, M. E., 1999. Rehabilitation of
shear critical concrete columns by use of rectangular steel jackets, ACI Struct. J. 96, 68–78.
Bentz, E., 2000. Response-2000, available at http://www.ecf.utoronto.ca/~bentz/home.shtml.
Eberhard, M., 2003. Structural Performance Database, Pacific Earthquake Engineering Re-
search Center, University of Washington, Seattle. Accessed 1 Nov. 2005.
Elwood, K. J., 2004. Modeling failures in existing reinforced concrete columns Can. J. Civ.
Eng. 31, 846–859.
Elwood, K. J., and Moehle, J. P., 2005a. Axial capacity model for shear-damaged columns, ACI
Struct. J. 102, 578–587.
––—, 2005b. Drift capacity of reinforced concrete columns with light transverse reinforcement,
Earthquake Spectra 21, 71–89.
Gerin, M., and Adebar, P., 2004. Accounting for shear in seismic analysis of concrete struc-
tures, in Proceedings of the 13th World Conference on Earthquake Engineering, paper 1747.
Vancouver, B.C., Canada, Aug. 2004.
Lynn, A. C., 2001. Seismic evaluation of existing reinforced concrete building columns, Ph.D.
thesis. University of California, Berkeley.
Mander, J. B., Priestley, J. N., and Park, R., 1988. Theoretical stress-strain model for confined
concrete, J. Struct. Eng. 114, 1804–1825.
Mo, Y. L., and Wang, S. J., 2000. Seismic behavior of RC columns with various tie configu-
rations, J. Struct. Eng. 126, 1122–1130.
Moehle, J. P., 1992. Displacement-based design of RC structures subjected to earthquakes
Earthquake Spectra 8, 403–428.
Patwardhan, C., 2005. Strength and deformation modeling of reinforced concrete columns,
M.S. thesis, The Ohio State University, Columbus.
Saatcioglu, M., and Ozcebe, G., 1989. Response of reinforced concrete columns to simulated
seismic loading, ACI Struct. J. 86, 3–12.
Setzler, E. J., 2005. Modeling the behavior of lightly reinforced concrete columns subjected to
lateral loads, M.S. thesis, The Ohio State University, Columbus, Ohio, 2005.
Sezen, H., 2002. Seismic behavior and modeling of reinforced concrete building columns,
Ph.D. thesis, University of California, Berkeley.
Sezen, H., and Moehle, J. P., 2003. Bond-slip behavior of reinforced concrete members, fib-
Symposium: Concrete Structures in Seismic Regions, CEB-FIP, Athens, Greece.
––—, 2004. Shear strength model for lightly reinforced concrete columns, J. Struct. Eng. 11,
1692–1703.
Sezen, H., and Setzler, E. J., 2008. Reinforcement slip in reinforced concrete columns, ACI
Struct. J. 105, 280–289.
MODEL FOR THE LATERAL BEHAVIOR OF REINFORCED CONCRETE COLUMNS 511

Umehara, H., and Jirsa, J. O., 1984. Short rectangular RC columns under bidirectional loadings,
J. Struct. Eng. 110, 605–618.
Vecchio, F. J., and Collins, M. P., 1986. The modified compression-field theory for reinforced
concrete elements subjected to shear, ACI J. 83, 219–231.
Wehbe, N. I., Saiidi, M. S., and Sanders, D. H., 1999. Seismic performance of rectangular
bridge columns with moderate confinement, ACI Struct. J. 96, 248–258.
Wight, J. K., and Sozen, M. A., 1975. Strength decay of RC columns under shear reversals, J.
Struct. Div. 101, 1053–1065.
(Received 15 July 2006; accepted 22 December 2007兲