Engineering Structures 79 (2014) 32–44

Contents lists available at ScienceDirect

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

Story shear and torsional moment-based pushover procedure for

asymmetric-plan buildings using an adaptive capacity spectrum method
Karim Tarbali a,b,⇑, Kazem Shakeri b
a
International Institute of Earthquake Engineering and Seismology, Tehran, Iran
b
Faculty of Engineering, University of Mohaghegh Ardabili, Ardabil, Iran

a r t i c l e i n f o a b s t r a c t

Article history: A single-run pushover procedure is proposed to assess the seismic response of asymmetric-plan build-
Received 26 August 2013 ings, when subjected to unidirectional earthquake ground motions. Effects of the higher and torsional
Revised 2 August 2014 modes are incorporated into an invariant load pattern, which is calculated based on the height-wise
Accepted 5 August 2014
distribution of the modal story shear and torsional moment. In order to consider the effect of instanta-
neous changes in dynamic characteristics of the structure in the nonlinear phase, capacity curve of the
structure is obtained using an adaptive capacity spectrum method. Results from numerical investigations
Keywords:
indicate appropriate accuracy of the proposed procedure in capturing the relative displacement of struc-
Nonlinear analysis
Asymmetric-plan buildings
tures when compared to the results from nonlinear response history analyses. The proposed load pattern
Torsional modes can be utilized in seismic performance assessment software available for the engineering community for
Adaptive capacity spectrum seismic assessment of asymmetric-plan buildings.
Higher modes Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction the accuracy of pushover methods in estimating seismic responses

Design of engineering structures according to a specific level of
performance as well as assessing the seismic performance of exist-
ing structures have gained tremendous attention during the past
decades. One of the challenging steps in performance-based earth-
quake engineering methodology is to determine the seismic
response of structures with an appropriate accuracy, and yet by
practically efficient methods. Due to the severe effect of a seismic
loading, deformations beyond the linear behavior should be taken
into account in a seismic assessment process. Nonlinear response
history analysis, as an elaborative method, can be used to obtain
time-varying responses of structures subjected to a specific
ground-motion time series. However, implementation of the
response history analysis in seismic assessment of structures is
accompanied by some elusiveness such as the uncertainty in
selecting appropriate ground motion time series and difficulties
in processing the time-varying response of the structure. In this
regard, nonlinear static analysis method has been proposed in seis-
mic performance assessment guidelines, e.g., [1,2], as a practical
method to estimate the seismic response of 2D models of building
structures. In this regard, methods have been proposed to enhance
⇑ Corresponding author at: International Institute of Earthquake Engineering and
Seismology, Tehran, Iran. Tel.: +98 9358591447, +64 224058996.
E-mail address: k.tarbali@iiees.ac.ir (K. Tarbali).
of structures [3–10].
Since the actual response of 3D asymmetric-plan buildings can
be considerably different from the response of its individual 2D
frames, accurate seismic evaluation of building structures requires
conducting response analysis of a 3D model of the structure and
considering the effects of the higher modes and torsional behavior
of the structure. Many attempts have been previously made to
assess the seismic performance of asymmetric-plan building
frames, using nonlinear static analysis procedures (e.g., [11–28]).
A brief review of these procedures has been presented by Shakeri
et al. [28]. Contribution of the torsional and higher modes in 3D
frames can be considered through a multi-run modal pushover
procedure, which requires multiple nonlinear analyses of the
structural model (i.e., [19]). In multi-run methods, the effect of
the interaction between responses from different modes in esti-
mating the total response of the structure is neglected. On the
other hand, the effect of the higher modes, as well as the torsional
modes, can be incorporated into a single load pattern, so that, the
pushover analysis can be performed through a single-run proce-
dure (e.g., [28,7]). Since seismic evaluation of building structures
using a 3D model of the structure requires more computational
efforts than implementation of 2D models, application of
single-run procedures for pushover analysis of 3D frames is more
practical than multi-run procedures.

http://dx.doi.org/10.1016/j.engstruct.2014.08.006
0141-0296/Ó 2014 Elsevier Ltd. All rights reserved.
 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44 33

In an effort to consider the instantaneous changes in dynamic approach to derive the load pattern. Based on the spectral dynamic
characteristics of asymmetric-plan building structures during non- analysis approach [31], maximum modal forces and the torsional
linear analysis, effects of the higher and torsional modes, and also moment, corresponding to each mode of vibration, are computed
interaction between the modes in nonlinear phase, a single-run by Eqs. (1)–(3):
adaptive modal pushover procedure has been proposed by Shakeri
f xij ¼ Cyj /xij mxi Sayj ð1Þ
et al. [28], denoted as the story Shear and Torsional moment Adap-
tive (STA) pushover method. This method was an extension to the
previously proposed story shear-based pushover procedure for 2D f yij ¼ Cyj /yij myi Sayj ð2Þ
frames [7]. The load pattern in the STA method is derived based on
the instantaneous distribution of the modal story shear and tor- T hij ¼ Cyj /hij Ihi Sayj ð3Þ
sional moment in stories of a 3D structure. In the present study, UTj MIy
a practically simplified version of the STA procedure is proposed where Cyj ¼ UTj MUj
is the modal participation factor of j-th mode for
to assess the seismic response of 3D asymmetric-plan buildings. excitation in y direction; Uj ¼ hUxj Uyj Uhj iT is mode shape vector of
The proposed method still makes use of the advantage of defining j-th mode consisting of the components in three directions of motion
the load pattern based on the modal story shear and torsional for floors of the 3D frame with in-plane rigidity; Uxj ¼
moment [28], however, in the proposed procedure the load pattern hUx1j Ux2j . . . UxNj iT , Uyj ¼ h/y1j /y2j . . . /yNj iT , Uhj ¼ h/h1j /h2j . . . /hNj iT
is taken to be invariant during the pushover analysis, as opposed to
are respectively mode shape vectors in translational (x and y) and
the STA method in which the load pattern is being updated at each
rotational directions in j-th mode; /xij ; /yij ; /hij are mode shape com-
step of the analysis based on the instantaneous changes in modal
characteristics of the structure. Since the instantaneous computa- ponents in x, y, and rotational directions of i-th story in j-th mode;
tion of the load pattern is a demanding task, and also is not appli- ly = h0 1 0iT is the influence vector for seismic excitation in y direc-
cable in conventional nonlinear analysis software for an tion; mxi and myi are the translational mass of the i-th story in x
engineering practice, the proposed invariant load pattern facili- and y directions; Ihi is the rotational mass of the i-th story; M is the
tates straightforward application of the procedure. The proposed mass matrix based on a 3D model of the structure; and Sayj is the
procedure is a single-run method, therefore, responses of the struc- spectral acceleration ordinate in the seismic excitation direction
ture can be obtained by performing a single nonlinear analysis of corresponding to j-th mode of vibration.
the 3D model, and the elusiveness associated with conducting The story shear in the translational directions and the torsional
multiple analyses is averted. In addition, the effects of the higher moment of each story are calculated by Eqs. (4)–(6), for each mode.
and torsional modes are incorporated into the proposed load pat- Then, the combined modal story shear and combined modal
tern. The capacity curve of the structure in the proposed procedure torsional moment are calculated by Eqs. (7)–(9), using square-
is established based on the instantaneous deformed shape of the root-of-the-sum-of-the-squares (SRSS) rule, or other combination
structure, using the adaptive capacity spectrum method (ACSM) rules:
proposed by Casarotti and Pinho [29]. Implementation of the ACSM
X
N
is not requiring an extra computational effort during the analysis. SSxij ¼ f xkj ð4Þ
The capacity curve of the structure can be established at the end of k¼i
the nonlinear analysis, using the recorded responses of the
structure. In the following sections, first, the proposed method is X
N

developed. Then the results from numerical analyses are presented SSyij ¼ f ykj ð5Þ
k¼i
and discussed. Finally, the concluding remarks are presented.

X
N

2. Development of the proposed procedure SThij ¼ Thkj ð6Þ

k¼i

Two main aspects of the proposed method: (a) derivation of the vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX
load pattern; and (b) establishing the capacity curve of the u m
CSSxi ¼ t SS2xij ð7Þ
structure are presented separately in this section.
j¼1

2.1. Derivation of the load pattern vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

uX
u m
CSSyi ¼ t SS2yij ð8Þ
In the proposed method, the load pattern is calculated based on j¼1
the height-wise distribution of the modal story shear and torsional
moment in stories of a 3D model of the structure. This load pattern vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX
will induce the same distribution of relative displacements and u m
CSThi ¼ t ST2hij ð9Þ
torsional rotations in the structure as obtained from spectral j¼1
dynamic analysis of the linear-elastic model of the structure.
Although dynamic characteristics of a structure alter during a non- where SSxij and SSyij and are respectively the shear in i-th story in x
linear analysis and distribution of the modal story shear and tor- and y translational directions associated with mode j; SThij is the
sional moment will change accordingly [28,7], in order to torsional moment in i-th story associated with mode j about the
maintain the simplicity and practicality of the proposed procedure, perpendicular direction to the floor at the mass center; CSSxi and
the proposed load pattern is chosen to be invariant during the CSSyi are the combined story shear of story i in x and y translational
analysis. Implementation of the invariant load pattern facilitates directions, respectively; CSThi is the combined torsional moment of
straightforward application of the method in common nonlinear story i in rotational direction; m is the number of modes used to cal-
analysis software for an engineering practice (such as SAP2000 culate the combined responses; and N is the number of the stories.
[30]) and lowers the computational cost of the nonlinear analysis. In order to induce the same distribution of relative displace-
Spectral dynamic analysis of 3D building frames with in-plane ments and torsional rotations in the structure as obtained from
rigid diaphragms is discussed here in order to illustrate the spectral dynamic analysis of the linear-elastic model of the
34 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44
structure, the applied load pattern should be calculated by sub-
tracting the combined modal story shear and torsional moment
of consecutive stories, using Eqs. (10)–(12):
 F  ¼ Ska ¼
F mi ¼ CSSxi  CSSxiþ1 i<N
F xi ¼ CSSxi i¼N
 C/r
F yi ¼ CSSyi  CSSyiþ1 i<N
F yi ¼ CSSyi i¼N
 tion at k-th step of the pushover analysis; ur,k is the roof displace-
T hi ¼ CSThi  CSThiþ1 i<N
T hi ¼ CSThi i¼N
A pushover analysis based on the proposed procedure makes
use of the distribution of the lateral forces (in two translational
directions) and torsional moment (in the direction perpendicular
to the floor), presented through Eqs. (10)–(12). These lateral forces
and torsional moments will be applied at the mass center of the
stories. In order to increase the base shear of the structure in the
excitation direction (i.e., y direction) by an appropriate amount
at each step of the pushover analysis, components of the calculated
load pattern are normalized with respect to the sum of the forces
in the excitation direction, using Eqs. (13)–(15):
Fx
F xi ¼ P i
Fy
F yi
F yi ¼ P
Fy
T hi
T hi ¼ P
Fy
These normalized components of the load pattern are
multiplied by the amount of the increment in the base shear of
the structure in the excitation direction (DV by ), using Eqs. (16)–(18):
DF xi ¼ DV by  F xi
DF yi ¼ DV by  F yi
DT h i ¼ DV b y  T h i
The calculated quantities by Eqs. (16)–(18) are the components
of the load pattern that will be applied at each step of the pushover
analysis on the structure.
2.2. Establishing the capacity curve
Using an invariant load pattern to conduct pushover analyses
facilitates straightforward application of the proposed method
with an appropriate computational cost, as opposed to implemen-
tation of the adaptive load pattern [28]. However, in order to
achieve an appropriate accuracy in estimating the nonlinear
responses and consider the effect of the instantaneous changes in
dynamic characteristics of the structure in nonlinear phase, capac-
ity curve of the structure (i.e., the equivalent single-degree-of-free-
dom (SDOF) system) is established based on the instantaneous
deformed shape of the multi-degree-of-freedom (MDOF) system,
using the ACSM [29]. As mentioned before, this approach is not
requiring any computational effort during the nonlinear analysis,
since the deformed shape of the MDOF system can be recorded
during the analysis and the capacity curve can be easily established
afterward. By implementing the ACSM, instantaneous changes in
the characteristics of the structure during the nonlinear analysis
are considered in the proposed procedure when establishing the
capacity curve of system rather than being considered in the
applied load pattern.
Generally, in order to establish the capacity curve of the
structure, response of the MDOF system resulted from the push-
over analysis of the structure is converted to the characteristic
force and displacement coordinates of an equivalent nonlinear
SDOF system, by considering an assumed fundamental mode shape
using Eqs. (19) and (20):
V b;k
ð19Þ
M
ð10Þ ur;k
 k
D ¼ Sd ¼ ð20Þ
ð11Þ
where Vb,k is the base shear of the structure in the excitation direc-
ð12Þ ment of the structure in the excitation direction at the k-th step;
/r is the component of the assumed fundamental mode shape at
roof level; C = (UTM l)/(U TM U) and M* = C. L are the participation
factor and effective mass based on the assumed fundamental mode
shape, respectively, in which L = (UTM l); and l is the influence vec-
tor to take into account the excitation of masses in the direction of
the seismic loading. Maximum response of the MDOF system is esti-
mated by applying the seismic demand spectrum on the equivalent
SDOF system and calculating the maximum displacement (target
displacement) [1,2]. Maximum displacement of the equivalent
SDOF system will be treated as the performance point of the struc-
ture under the applied demand spectrum.
The assumed fundamental mode shape can be obtained based
ð13Þ on the mode shape of the structure, corresponding to the applied
distribution of the effective modal forces [e.g., 19], or the pattern
of the applied pushover load on the structure [e.g., 28]. As a novel
ð14Þ
approach, the deformed shape of the structure under the applied
load pattern can be used as the assumed fundamental mode shape
ð15Þ
in order to establish the capacity curve of the structure, based on
the ACSM [29]. Since the deformed shape of the structure continu-
ously changes during the nonlinear analysis, using the instanta-
neous deformed shape can result in an accurate representation of
the capacity curve. This idea has been proposed by Casarotti and
ð16Þ Pinho for seismic assessment of bridges, denoted as the adaptive
capacity spectrum method (ACSM) [29]. Application of the ACSM
ð17Þ on 3D building structures has been previously presented in Refs.
ð18Þ [24–27]. In the present study, the ACSM is integrated with the pro-
posed load pattern in order to estimate the seismic response of 3D
asymmetric-plan building frames using a simplified procedure
based on the proposed invariant load pattern.
The ACSM is basically the adaptive form of Eqs. (19) and (20), in
which the assumed fundamental mode shape is defined as the
deformed shape of the structure. Force coordinate of the capacity
curve is defined by the same formula as in Eq. (19), however, the
effective mass is instantaneously updated based on the instanta-
neous deformed shape of the structure at each step of the analysis
(see Eq. (21)). Since at each step of the analysis, roof component of
the fundamental mode shape (/r) is equal to roof displacement
(ur,k), displacement coordinate of the capacity curve will be simpli-
fied to inverse of the modal participation factor (see Eq. (22)). Thus,
the characteristic force and displacement coordinates of the
adaptive SDOF system for 3D building frames are calculated using
Eqs. (21) and (22), respectively:
V b;k
Ska ¼ ð21Þ
M k
ðDTK M DK Þ
Skd ¼ ð22Þ
DTK M ly
where Dk is the vector of the deformed shape of the structure (i.e.,
two orthogonal translational displacements and a torsional rotation
at the mass center of all stories) at k-th step of the pushover analysis,
ðDT M l Þ
2
and Mk ¼ ðDKT M Dy is the instantaneous effective mass based on the
K KÞ
instantaneous deformed shape of the structure (i.e., Dk) at k-th.
In order to obtain the maximum displacement of the equivalent
SDOF system under a given ground motion and determine the
 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44
response of the MDOF system, inelastic demand spectrum of the
motion can be plotted against the capacity curve of the structure
in the acceleration–displacement response spectrum format. The
intersection between these two spectra will be the performance
point (i.e., the maximum displacement) of the system. In this
study, the obtained capacity curve of the structure will be replaced
with an equivalent bilinear SDOF system and the maximum dis-
placement will be determined by performing a nonlinear response
history analysis of this equivalent bilinear SDOF system under the
corresponding ground motion. Response of the MDOF system will
be determined at the pushover step corresponding to the calcu-
lated maximum displacement.
2.3. Sequential steps of the proposed method for an automated
application
Sequential steps toward conducting a pushover analysis based
on the proposed method are presented here. First stage is to calcu-
late the load pattern and conduct the analysis, as presented in
steps 1–8:
1. Create a 3D model of the structure, considering the possible
nonlinear mechanisms.
2. Calculate the maximum lateral modal forces and the maxi-
mum torsional moment in each story of the structure, using
Eqs. (1)–(3).
3. Calculate the story shear in the translational directions and
the torsional moment of each story, using Eqs. (4)–(6).
4. Calculate the combined modal story shear and the com-
bined modal torsional moment of each story, using Eqs.
(7)–(9).
5. Calculate the components of the applied load pattern by
subtracting the combined modal story shear and the com-
bined torsional moment of consecutive stories, using Eqs.
(10)–(12).
6. Normalize the calculated load pattern with respect to the
sum of the forces in the excitation direction, using Eqs.
(13)–(15).
7. Multiply the normalized components of the load pattern by
the amount of the increment in the base shear of the
structure (in the excitation direction), using Eqs. (16)–(18).
8. Conduct a pushover analysis by applying the calculated load
pattern until the base shear of the structure in the excitation
direction reaches to an extreme value or the structure is not
stable to resist any load. Record the desired responses of the
structure at each step of the pushover analysis.
As presented in steps 1–8, the load pattern is calculated once,
and it is used to conduct the pushover analysis (i.e., invariant load
pattern concept). The second stage is to establish the capacity
curve of the structure through steps 9–12:
9. Extract the displacement vector in the mass center of stories
(along two translational directions and the torsional direc-
tion) at each step of the pushover analysis from the data-
base of the recorded responses.
10. Assume the obtained displacement vector at each step as
the instantaneous fundamental mode shape of the structure
and calculate the instantaneous effective mass of the struc-
2
ðDTK M ly Þ
ture, using Mk ¼ ðDTk M Dk Þ
.
11. Calculate the characteristic force coordinate of the
equivalent SDOF system by dividing the base shear of the
structure at each step of the analysis (in the excitation
direction) by the instantaneous effective mass of the struc-
ture, using Eq. (21).
35
12. Calculate the characteristic displacement coordinate of the
equivalent SDOF system, using Eq. (22).
As presented in steps 9–12, the capacity curve is based on the
instantaneous deformed shape of the structure during the analysis
(i.e., adaptive capacity spectrum concept). Therefore, while the
applied load pattern is invariant, the instantaneous changes in
the capacity of the structure throughout the analysis are
considered in the capacity curve. The step toward obtaining the
performance point of the structure is as follows:
13. Plot the inelastic demand spectrum against the capacity
curve of the structure in the acceleration–displacement
response spectrum format and obtain the intersection
between these two spectra. This intersection point is the
performance point of the structure.
Since the proposed load pattern is invariant throughout the
nonlinear analysis, it can be easily calculated for a given structural
system and plugged into nonlinear analysis software such as
SAP2000 [30] for seismic performance assessment purposes. After
conducting the pushover analysis, the capacity curve of the
structure can be calculated separately, based on the obtained
responses from the software.
While increasing the accuracy of the pushover methods by pro-
posing novel load patterns and procedures, it is important to main-
tain the simplicity of the pushover methods in comparison to the
nonlinear response history analysis. An easy application, lower
computational demand, and the capability to trace the hierarchy
of the occurrence of nonlinear mechanism in various components
of the structure throughout the analysis are the advantages of
the pushover analysis that should not be compromised. In this
regard, the proposed method in this study is an effort to simplify
the application of the story shear and torsional moment-based
pushover procedure considering the abovementioned points, while
keeping the accuracy of the method appropriately acceptable.
The results from the conducted numerical analyses are
presented in the following section to evaluate the accuracy of the
proposed method. Since the proposed load pattern is based on
the story shear and torsional moment distribution in stories of
the structure and the adaptive capacity spectrum method is used
to establish the capacity curve of the structure, the proposed
pushover procedure is denoted as ST-ACS.
3. Numerical study
In order to investigate the accuracy of the proposed method in
estimating the seismic response of 3D asymmetric-plan buildings,
the proposed procedure is implemented on two steel moment-
resisting building frames, known as 9- and 20-story buildings of
the SAC project [32]. These building frames have been designed
for phase II of the SAC project for Los Angeles, California, according
to 1994 UBC seismic code. The height of the first story of these
structures is 5.5 m and the height of the other stories is 4 m. 5%
damping ratio is assigned to the first mode and a higher mode with
the vibration period equal to 0.1 times the first-mode period, using
the Rayleigh damping method. More details required for modeling
of these structures are presented in Ref. [33]. The mass center in
each story of these buildings is moved as far as 10% of the length
of the plan along the direction perpendicular to the direction of
the applied seismic load to induce an asymmetry in the structures.
The nonlinear 3D model of the structures are created in OpenSees
finite element platform [34], using fiber-discretized elements.
Floors of the buildings are assumed to be rigid in plane. Nonlinear
responses of the structures are observed in the mass center, and in
36 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44
two perimeter moment-resisting frames of the buildings, denote as
stiff and flexible sides. The stiff and flexible sides of the plan are,
respectively, the farthest and the closest edges of the plan to the
mass center of the stories (Fig. 1). In order to conduct nonlinear
response history analyses (NRHA), a total of 10 ground-motion
time series have been selected from the ensemble of the SAC pro-
ject ground motions [35]. According to the seismic hazard curve of
the region, the selected ground motions are categorized into two
sets of motions with a 2% and 10% probability of exceedance in
50 years. Characteristics of the selected motions are presented in
Table 1.
Since the damage state of building structures subjected to a
seismic action is significantly governed by the relative displace-
ment in consecutive stories (referred to as drift), accuracy of the
proposed method in estimating the seismic response of the struc-
tures are investigated by comparing the drift profiles obtained
from the proposed pushover method with those from NRHA. In
addition, estimations from the modal pushover analysis (MPA)
method for asymmetric-plan buildings [19] and the pushover anal-
ysis with load pattern based on distribution of the effective seismic
forces in the first elastic mode of the structure (M1) are also
included. It is also valuable to evaluate the accuracy of the pro-
posed method (along with the other methods) based on other
important seismic responses such as rotation in the plastic hinge
zones; deformation in the columns; and internal forces in beams,
columns, and beam-column joins. Such effort is the topic of an
ongoing study by the authors. The presented results in this study
are limited only to the drift estimations.
In order to evaluate the accuracy of the proposed procedure in
estimating the drift profile of the structures and compare it with
the accuracy of the other considered pushover methods, an error
index is calculated for the estimated drift profiles. Relative differ-
ences between the estimations from the corresponding pushover
procedure and the results from NRHA in all of the stories of the
structure are taken into account in the error index, presented in
Eq. (23):
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u N 
1 u X DiNRHA  DiNSP 2
Total Error ð%Þ ¼ 100   t
N DiNRHA
i¼1
where Di-NRHA is the maximum drift of the i-th story resulted from
NRHA, Di-NSP is the drift of the i-th story from the corresponding
Fig. 1. Asymmetric plan of the 9- and 20-story building frames, illustrating mass center (MC), stiffness center (SC), and the perimeter frames at stiff and flexible sides.
nonlinear static analysis procedure, and N is the number of the sto-
ries of the structure.
Different aspects of the proposed procedure, related to the load
pattern and establishing the capacity curve of the structures will
be investigated in the following sections. Without integrating the
ACSM with the proposed load pattern, application of the proposed
load pattern for pushover analysis of 3D asymmetric-plan build-
ings will be demonstrated. Then, by performing pushover analysis
with the proposed load pattern and establishing the capacity curve
of the structure using the ACSM, accuracy of the obtained results
for the proposed pushover procedure will be investigated. After-
ward, in order to study the effect of using the adaptive form of
the proposed load pattern, a comparison will be made between
the accuracy of the estimations from pushover analyses based on
the proposed invariant load pattern and the adaptive form of it,
while the capacity curve of the structures is established by the
ACSM for both cases. At the end, a discussion will be presented
on mean responses of the structures when subjected to each set
of the selected ground motions.
3.1. Pushover analysis based on the proposed procedure
In addition to the propose procedure (i.e., ST-ACS method),
accuracy of conducting pushover analysis based only on the pro-
posed load pattern without using the ACSM is investigated. This
method is denoted as ST method. In order to establish the capacity
curve of the structure in the ST method, deformed shape of the
structure under the applied load pattern at the elastic phase is
taken as the assumed fundamental mode shape, using Eqs. (19)
and (20). Therefore, the presented results for the ST method dem-
onstrate only performance of the proposed load pattern without
including the effect of the ACSM.
Drift profiles of the 9- and 20-story buildings under Northridge
and Tabas motions with a 10% and 2% probability of exceedance in
50 years are presented in Figs. 2 and 3, respectively. As shown in
these figures, relative to the considered pushover methods, the
proposed method (i.e., ST-ACS) is successful in estimating the drift
profiles of the structures. While the drift profiles from the M1
method is always underestimated in the higher stories, and the
ð23Þ
estimations from the MPA method does not resemble the NRHA
results, trend of the obtained drift profiles from the proposed
method are similar to the trend of the NRHA results. In lower sto-
ries of the structures, estimations from the proposed method are
closer to the NRHA results compared to the estimations from the
 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44 37

Table 1
Characteristics of the selected ground motions.

Ground motion ID Earthquake Year Magnitude Closest distance to fault (km) PGA (g) PGV (cm/s)
Ground motions with a 10% probability of exceedance in 50 years
1 Imperial Valley 1940 6.9 10.0 0.68 58.7
2 Imperial Valley 1979 6.5 4.1 0.39 81.4
3 Landers 1992 7.3 36.0 0.42 64.8
4 Loma Prieta 1989 7.0 12.0 0.67 77.6
5 Northridge 1994 6.7 7.5 0.53 96.6
Ground motions with a 2% probability of exceedance in 50 years
6 Kobe 1995 6.9 3.4 1.28 140
7 Loma Prieta 1989 7.0 3.5 0.42 72.4
8 Northridge 1994 6.7 7.5 0.87 157
9 Northridge 1994 6.7 6.4 0.93 128
10 Tabas 1974 7.4 1.2 0.81 69.7

9 9 9
(a) (b) (c)
8 8 8
7 NRHA 7 NRHA 7 NRHA
6 ST-ACS 6 ST-ACS 6 ST-ACS
5 ST ST
Story

ST 5 5
4 M1 4 M1 4 M1
3 MPA MPA MPA
3 3
2 2 2
1 1 1
0 0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0 0.05 0.1 0.15
Dri (m) Dri (m) Dri (m)

20 20 20
(a) (b) (c)
18 18 18
16 16 16
NRHA NRHA NRHA
14 14 ST-ACS 14
ST-ACS ST-ACS
12 12 ST 12
ST
Story

ST
10 10 M1 10 M1
M1
8 8 MPA 8 MPA
MPA
6 6 6
4 4 4
2 2 2
0 0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Dri (m) Dri (m) Dri (m)

Fig. 2. Drift profiles of the 9- and 20-story buildings subjected to Northridge motion with a 10% probability of exceedance in 50 years resulted from NRHA and the ST-ACS, ST,
M1, and MPA pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.

other methods. However, in the higher stories, drift profiles
obtained from the proposed method are mostly underestimated.
Comparison between the estimations from the ST-ACS and ST
methods indicates that in higher stories these two methods result
in approximately equal responses. However, in lower stories, esti-
mations from the proposed procedure (i.e., ST-ACS) are closer to
the NRHA results which is more accentuated in the drift profiles
of the 20-stories building (see Figs. 2 and 3 for responses of the
20-story building). This implicates the positive effect of integrating
the ACSM with the proposed load pattern in order to obtain drift
estimations with an improved accuracy. This matter will also be
discussed based on the mean responses of the structures.
In order to investigate the accuracy of the obtained results, total
error indices calculated for the drift estimations from the ST-ACS
and ST methods, along with those for the MPA and M1 methods,
are compared through the presented results in Figs. 4 and 5 for
the 9- and 20-story buildings, respectively. As shown in Figs. 4a
and 5a, at the stiff side of the 9- and 20-story buildings,
estimations from the ST method have higher error indices for most
of the ground motions, even in some cases higher than those for
M1 method. However, at the flexible side (Figs. 4b and 5b) and
the mass center (Figs. 4c and 5c) of the structures, in comparison
to both of the MPA and M1 methods, the ST method demonstrates
a better accuracy in most of the cases. The difference between the
accuracy of the ST and M1 methods is more significant in the 20-
story building than the 9-story building. Considering the fact that
both of the ST and M1 methods are single-run procedures, imple-
mentation of the ST method (in comparison to the M1 method)
would result in estimations with a higher accuracy, especially for
tall buildings since the contribution of the higher modes are con-
sidered in the load pattern of the ST method. By comparing the
total error indices for the ST and MPA methods, at the stiff and flex-
ible sides, and the mass center in both of the buildings, no superi-
ority can be observed for one of these methods over the other.
Although the MPA method is more successful at the stiff side of
the buildings, ST method shows an appropriate performance at
38 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44

9 9 9
8
(a) 8
(b) 8
(c)
7 NRHA 7 NRHA 7 NRHA
6 ST-ACS 6 ST-ACS 6 ST-ACS
5 ST
Story

ST 5 ST 5
M1
4 M1 4 M1 4
MPA
3 MPA 3 MPA 3
2 2 2
1 1 1
0 0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Dri (m) Dri (m) Dri (m)

20 20 20
18 (a) 18
(b) 18
(c)
16 NRHA 16 NRHA 16 NRHA
14 ST-ACS 14 ST-ACS 14 ST-ACS
12 ST 12 ST 12 ST
Story

10 M1 10 M1 10 M1
8 MPA 8 MPA 8 MPA
6 6 6
4 4 4
2 2 2
0 0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Dri (m) Dri (m) Dri (m)

Fig. 3. Drift profiles of the 9- and 20-story buildings subjected to Tabas motion with a 2% probability of exceedance in 50 years resulted from NRHA and the ST-ACS, ST, M1,
and MPA pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.

the flexible side and the mass center of the structures. In addition,
implementation of the ST method is more straightforward than
MPA method, since the ST method is a single-run procedure, as
opposed to MPA method which is a multi-run procedure.
Comparison between the accuracy of the ST, M1, and MPA
methods demonstrated the competency of the proposed load pat-
tern in estimating the response of the structures. At this stage, the
ACSM is integrated with the proposed load pattern (i.e., the pro-
posed procedure) to obtain drift estimations with an improved
accuracy. As shown in Fig. 4 for the 9-story building, total error
index of the ST method is higher than that of the MPA method
for motions 1, 2, 9, and 10 at the stiff side; and motions 2 and 4
at the flexible side and mass center. By integrating the ACSM with
the ST method (i.e., ST-ACS), total error index has been lowered in
these cases, so that, estimations of the ST-ACS method are more
accurate than estimations of the MPA method. The same trend is
observed in the results of the 20-story building, under motions 2,
3, 5, and 7 at the stiff side; and motions 2, 5, and 10 at the mass
center (see Fig. 5).
Based on the error indices presented in Fig. 4 for the 9-story
building, the accuracy of the ST-ACS method is higher or equal to
the accuracy of the ST method, except for two cases at the stiff side
and the mass center (i.e., motions 7 and 8), and one case at the flex-
ible side (i.e., motion 7) of the structure. Also, based on the pre-
sented results in Fig. 5 for the 20-story building, except for two
cases at the stiff side (i.e., motions 8 and 9), four cases at the flex-
ible side (i.e., motions 5, 6, 8, and 9), and three cases at the mass
center (i.e., motions 6, 8, and 9), the ST-ACS method demonstrates
a higher accuracy than ST method.
When compared to the estimations from the MPA method, as
presented in Fig. 4, estimations from the ST-ACS method have a
higher or approximately equal accuracy. Except for motions 3, 4,
and 7 at the stiff side, motion 9 at the flexible side, and motions
7, and 9 at the mass center of the 9- story building. Also in the
20-story building (see Fig. 5), except for motions 4 and 9 at the stiff
side, and motions 6, 8, 9, and 10 at the flexible side, and motions 8
and 9 at the mass center of structure, ST-ACS method demonstrates
a higher or equal accuracy in comparison to the MPA method.
In both of the ST-ACS and MPA methods, the effect of the higher
modes is considered in the analysis, however, by different
approaches. While in the ST-ACS method this effect is included in
a single load pattern, in the MPA method it is considered by con-
ducting multiple nonlinear analyses of structures with load pat-
terns conforming to the modal distribution of the seismic forces
in the corresponding modes. Conducting multiple analyses
requires more computational effort when compared to the
straightforward implementation of the ST-ACS method. The
advantage of implementing the single-run pushover method (i.e.,
ST-ACS) to multi-run modal procedure (i.e., MPA) is more accentu-
ated in case of high-rise buildings with a large number of contrib-
uting modes.
3.2. Using an adaptive load pattern in the proposed procedure
In the proposed ST-ACS method, an invariant load pattern is uti-
lized to conduct pushover analyses. The proposed invariant load
pattern can be replaced by an adaptive load pattern [28], so that,
instantaneous changes in dynamic characteristics of the structure
and subsequently changes in the distribution of the modal story
shear and torsional moment can be directly reflected in the applied
load pattern. For a pushover analysis with an adaptive load pattern,
the applied load pattern is recalculated at each step of the analysis
 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44 39

ST- ACS ST M1 MPA ST-ACS ST M1 MPA

25 15
(a) (a)

Index on Dri (%)

Index on Dri (%)

20 12
Total Error

Total Error
15 9
10 6
5 3
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Ground Moon ID Ground Moon ID
25 15
(b) (b)
Index on Dri (%)

Index on Dri (%)

20 12
Total Error

Total Error
15 9
10 6
5 3
0 0
1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10
Ground Moon ID Ground Moon ID

25
(c) 15
Index on Dri (%)

(c)

Index on Dri (%)

20
Total Error

12
Total Error
15
9
10
6
5
3
0
1 2 3 4 5 6 7 8 9 10 0
1 2 3 4 5 6 7 8 9 10
Ground Moon ID
Ground Moon ID
Fig. 4. Total error index of the estimated drift profiles for the ST-ACS, ST, MPA, and
M1 methods at the: (a) stiff side; (b) flexible side; and (c) mass center of the 9-story Fig. 5. Total error index of the estimated drift profiles for the ST-ACS, ST, MPA, and
building. M1 methods at the: (a) stiff side; (b) flexible side; and (c) mass center of the 20-
story building.

[28]. This variant of the proposed method in which the load pattern
has an adaptive form and the capacity curve of the structure is
established based on the ASCM is denoted as STA-ACS method.
Drift profiles of the 9- and 20-story buildings subjected to
Northridge and Tabas motions with a 10% and 2% probability of
exceedance in 50 years are presented in Figs. 6 and 7, respectively.
In addition, total error indices of the STA-ACS and ST-ACS methods
under all of the ground motions are compared through the pre-
sented results in Figs. 8 and 9. As shown in Figs. 6 and 7, agreement
between the estimations of the STA-ASC and ST-ASC methods in
lower stories of the 9-story building is evident. However, in the
higher stories, estimations from the ST-ACS method are closer to
NRHA results. As seen in Fig. 8b, estimations from the ST-ACS
method are generally more accurate than the estimations from
the STA-ACS method at the flexible side of the 9-story building.
However, no general trend can be observed that one of these meth-
ods is more accurate than the other one at the stiff side and the
mass center of the 9-story building (see Fig. 8a and c).
In case of the 20-story building, estimations of the STA-ASC
method are mostly underestimated in comparison with those from
ST-ASC (see Figs. 6 and 7 for responses of the 20-story building). As
presented in Fig. 9a, none of these methods seems more accurate
than the other one at the stiff side of the 20-story building. How-
ever, at the flexible side and the mass center of the 20-story build-
ing, the ST-ACS method yields better estimations compared to the
STA-ACS method (see Fig. 9b and c). Generally, in both of the build-
ings, estimations of the ST-ACS are mostly closer to the NRHA
results than the estimations of the STA-ACS method.
Based on the presented results, it can be inferred that imple-
menting the adaptive form of the load pattern cannot guarantee
better estimations over the invariant load pattern. Considering
the fact that conducting a pushover analysis with an adaptive load
pattern (e.g., STA-ACS method) requires nonlinear analysis soft-
ware with the capability of updating the load pattern at each step
of the analysis (which is also computationally demanding), ST-ACS
method emerges as a superior method over the STA-ACS method.
3.3. Discussion based on the mean responses of the structures
Estimations from the proposed method (i.e., ST-ACS) and the
other considered pushover methods (i.e., M1 and MPA methods)
are compared with the mean responses of the structures obtained
from NRHA. Total error index is calculated for the mean drift pro-
files resulted from the corresponding pushover analyses, in which,
Di-NRHA and Di-NSP in Eq. (23) are replaced with the corresponding
mean values.
Mean drift profiles of the 9-story building subjected to the
motions with 10% and 2% probability of exceedance in 50 years
are presented in Figs. 10 and 11, respectively. Total error indices
of the mean drift profiles are also presented in Fig. 12 for the 9-
story structure. As seen in Figs. 10 and 11, at the stiff and flexible
sides, and the mass center of the 9-story building, the trend of the
drift profiles estimated by the proposed method is closer to the
trend of the drift profile from NRHA in comparison to the other
considered pushover methods. At the flexible side of the 9-story
structure under both sets of motions, drift profiles from the pro-
posed method is underestimated in the higher stories when com-
pared to the NRHA results. As presented in Fig. 12, total error
index for estimations of the proposed method is smaller than that
for the M1 and MPA methods. In addition, the proposed method
yields better estimations in comparison to the ST method, although
40 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44

9 9 9
8
(a) 8
(b) 8
(c)
7 7 7
6 NRHA NRHA NRHA
6 6
ST-ACS ST-ACS ST-ACS
5
Story STA-ACS
5 5
4 STA-ACS STA-ACS
4 4
3 3 3
2 2 2
1 1 1
0 0 0
0 0.05 0.1 0.15 0 0.05 0.1 0.15 0 0.05 0.1 0.15
Dri (m) Dri (m) Dri (m)

20 20 20
18 (a) 18
(b) 18
(c)
16 16 16
14 NRHA 14 NRHA 14 NRHA
12 ST-ACS 12 ST-ACS 12 ST-ACS
Story

10 STA-ACS 10 STA-ACS 10 STA-ACS

8 8 8
6 6 6
4 4 4
2 2 2
0 0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Dri (m) Dri (m) Dri (m)

Fig. 6. Drift profiles of the 9- and 20-story buildings subjected to Northridge motion with a 10% probability of exceedance in 50 years resulted from NRHA, and the STA-ACS
and ST-ACS pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.

9 9 9
8
(a) 8
(b) 8
(c)
7 7 7 NRHA
6 NRHA 6 NRHA 6
ST-ACS
5 ST-ACS 5 ST-ACS
Story

5
STA-ACS STA-ACS
4 STA-ACS 4 4
3 3 3
2 2 2
1 1 1
0 0 0
0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2 0 0.05 0.1 0.15 0.2
Dri (m) Dri (m) Dri (m)

20 20 20
18 (a) 18
(b) 18
(c)
16 16 16
NRHA NRHA NRHA
14 14 14
ST-ACS ST-ACS ST-ACS
12 12 12
Story

10 STA-ACS 10 STA-ACS 10 STA-ACS

8 8 8
6 6 6
4 4 4
2 2 2
0 0 0
0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08
Dri (m) Dri (m) Dri (m)

Fig. 7. Drift profiles of the 9- and 20-story buildings subjected to Tabas motion with a 2% probability of exceedance in 50 years resulted from NRHA, and the STA-ACS and ST-
ACS pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.

for motions with a 2% probability of exceedance in 50 years accu- profiles are also presented in Fig. 15. As seen in Figs. 13 and 14,
racy of the ST-ACS and ST methods are approximately equal. under both sets of motions, the proposed method underestimates
Mean drift profiles of the 20-story building are presented in the drift profile of the 20-story building in higher stories. Underes-
Figs. 13 and 14, respectively. Total error indices of the mean drift timation in drift profiles obtained from the ST-ACS method is more
 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44 41

STA-ACS ST-ACS STA-ACS ST-ACS

25
(a)

Index on Dri (%)

25
(a) 20
Index on Dri (%)

Total Error
20
15
Total Error

15
10
10
5
5
0
0 1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10 Ground Moon ID
Ground Moon ID
25
(b)

Index on Dri (%)

25
(b) 20
Index on Dri (%)

Total Error
20
15
Total Error

15
10
10
5
5
0
0 1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10 Ground Moon ID
Ground Moon ID
25
(c)

Index on Dri (%)

25
(c) 20
Index on Dri (%)

Total Error
20 15
Total Error

15 10
10 5
5 0
0 1 2 3 4 5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10 Ground Moon ID
Ground Moon ID
Fig. 9. Total error index of the estimated drift profiles for the STA-ACS and ST-ACS
Fig. 8. Total error index of the estimated drift profiles for the STA-ACS and ST-ACS methods at the: (a) stiff side; (b) flexible side; and (c) mass center of the 20-story
methods at the: (a) stiff side; (b) flexible side; and (c) mass center of the 9-story building.
building.

As it can be observed in the presented mean drift profiles of the

evident under the motions with a 2% probability of exceedance in
50 years, especially at the flexible side and the mass center.
However, unlike the other considered pushover methods, estima-
tions from the proposed method follow the trend of the NRHA
results. Comparison between the error indices calculated for the
estimations of the MPA procedure with those of the proposed
method (in Fig. 15) indicates that for the motions with a 2% prob-
ability of exceedance in 50 years estimations from the ST-ACS
method are more accurate than the estimations from the MPA
method at the stiff side of the 20-story building. For the motions
with a 10% probability of exceedance in 50 years, the proposed
method demonstrates a higher accuracy when compared to the
other considered pushover methods at the stiff and flexible sides
and the mass center of the 20-story building.
structures, in lower stories, estimations of the ST-ACS method are
closer to NRHA results than estimation of the ST method, which is
more accentuated in drift profiles of the 20-story building. How-
ever, in the higher stories, these two methods result in approxi-
mately equal responses. In general, integrating the ACSM with
the proposed invariant load pattern leads to better estimations.
It is evident from the presented mean drift profiles and the
calculated total error indices for both of the structures that the esti-
mations from the proposed method have a higher accuracy than the
estimations from the M1 method under both sets of motions. Con-
sidering the fact that both of these methods (i.e., M1 and ST-ACS)
are single-run pushover procedures, it would be more advanta-
geous to conduct a pushover analysis using the proposed method.
In addition, although accuracy of the proposed method is slightly

9 9 9
8
(a) 8
(b) 8
(c)
7 NRHA 7 NRHA 7 NRHA

6 ST-ACS 6 ST-ACS 6 ST-ACS

5 ST ST ST
Story

5 5
4 M1 M1 M1
4 4
MPA MPA MPA
3 3 3
2 2 2
1 1 1
0 0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Dri (m) Dri (m) Dri (m)

Fig. 10. Mean drift profiles of the 9-story building subjected to the motions with a 10% probability of exceedance in 50 years resulted from NRHA and the ST-ACS, ST, M1, and
MPA pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.
42 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44

9 9 9
8
(a) 8
(b) 8
(c)
7 NRHA 7 NRHA 7 NRHA
6 ST-ACS 6 ST-ACS 6 ST-ACS
ST
Story 5 ST 5 ST 5
4 M1 4 M1 4 M1
3 MPA 3 MPA 3 MPA
2 2 2
1 1 1
0 0 0
0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3
Dri (m) Dri (m) Dri (m)

Fig. 11. Mean drift profiles of the 9-story building subjected to the motions with a 2% probability of exceedance in 50 years resulted from NRHA and the ST-ACS, ST, M1, and
MPA pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.

ST-ACS ST M1 MPA
8 8
(a) (b)
Index on Mean Dri (%)

6 6
Total Error

4 4

2 2

0 0
SffSide Flexible Side Mass Center SffSide Flexible Side Mass Center

Fig. 12. Total error index of the mean drift profiles calculated for the ST-ACS, ST, M1, and MPA methods at the stiff side, flexible side, and mass center of the 9-story building
subjected to the motions with a (a) 10% and (b) 2% probability of exceedance in 50 years.

20 20
(a) 20 (b) (c)
18 18 18
16 NRHA 16 NRHA 16 NRHA
14 ST-ACS 14 ST-ACS 14 ST-ACS
12 ST 12 ST 12 ST
Story

10 M1 10 M1 10 M1
8 MPA 8 MPA 8
MPA
6 6 6
4 4 4
2 2 2
0 0 0
0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1
Dri (m) Dri (m) Dri (m)

Fig. 13. Mean drift profiles of the 20-story building subjected to the motions with a 10% probability of exceedance in 50 years resulted from NRHA and the ST-ACS, ST, M1,
and MPA pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.

20 20 20
18
(a) 18
(b) 18
(c)
16 NRHA 16 NRHA 16 NRHA
14 ST-ACS 14 ST-ACS 14 ST-ACS
12 ST 12 ST 12 ST
Story

10 M1 10 M1 10 M1
8 MPA 8 MPA 8 MPA
6 6 6
4 4 4
2 2 2
0 0 0
0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1 0 0.02 0.04 0.06 0.08 0.1
Dri (m) Dri (m) Dri (m)

Fig. 14. Mean drift profiles of the 20-story building subjected to the motions with a 2% probability of exceedance in 50 years resulted from NRHA and the ST-ACS, ST, M1, and
MPA pushover methods at the: (a) stiff side; (b) flexible side; and (c) mass center.
 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44
ST-ACS
10
(a)
Index on Mean Dri (%)
8 8
Total Error
6 6
4 4
2 2
0 0
SffSide Flexible Side Mass Center
Fig. 15. Total error index of the mean drift profiles calculated for the ST-ACS, ST, M1, and MPA methods at the stiff side, flexible side, and mass center of the 20-story building
subjected to the motions with a (a) 10% and (b) 2% probability of exceedance in 50 years.
lower than the accuracy of the MPA method at the flexible side and
the mass center of the 20-story building for the motions with a 2%
probability of exceedance in 50 years, performing multiple nonlin-
ear analyses of high-rise buildings to consider the effect of the
higher modes is a computationally demanding task. In this regard,
it can be inferred that by conducting pushover analyses using the
proposed method the effect of the higher modes can be efficiently
considered through a single-run procedure.
4. Conclusion
A pushover procedure is proposed to assess the seismic
response of 3D asymmetric-plan buildings subjected to a unidirec-
tional excitation. The Effect of the higher and torsional modes is
incorporated into a single-run invariant load pattern. The load pat-
tern is derived based on the height-wise distribution of the modal
story shear and torsional moment in stories of the structure. In
order to consider the instantaneous changes in dynamic character-
istics of the structure in the nonlinear phase, capacity curve of the
structure is obtained based on the instantaneous deformed shape
of the structure, using the adaptive capacity spectrum method
(ACSM). The proposed method (i.e., ST-ACS) is a single-run proce-
dure, which facilitates tracing the nonlinear response of the struc-
ture during the pushover analysis. Two asymmetric-plan steel
moment-resisting frames with 9 and 20 stories have been used
to evaluate the accuracy of the proposed procedure against the
results from nonlinear response history analyses (NRHA). Floors
of the buildings are assumed to be rigid in plane and the asymme-
try in the structures is induced by the eccentricity in distribution of
the floor mass. Two sets of ground motions were selected for
NRHA, conforming to the hazard level of the region with a 2%
and 10% probability of exceedance in 50 years. Different aspects
of the proposed procedure, related to the proposed load pattern
and establishing the capacity curve of the structure are investi-
gated. Without integrating the ACSM into the procedure, applica-
tion of the proposed load pattern for pushover analysis of 3D
asymmetric-plan buildings is demonstrated and the accuracy of
the obtained results from this method (i.e., ST) is discussed. By per-
forming pushover analyses using the proposed load pattern and
establishing the capacity curve of the structure using the ACSM,
accuracy of the proposed pushover procedure (i.e., ST-ACS) is also
investigated. In addition, in order to study the effect of implement-
ing adaptive form of the proposed load pattern on accuracy of the
obtained drift profiles, estimations from a variant of the proposed
procedure with an adaptive load pattern (i.e., STA-ACS method) are
considered for comparison. Estimations from the modal pushover
analysis (MPA) procedure and the pushover analysis with a load
pattern conforming to the distribution of the effective seismic
forces in the first elastic mode of the structure (i.e., M1 method)
are also considered in the evaluation study. Findings of this study
are summarized as follows:
43
ST M1 MPA
10
(b)
SffSide Flexible Side Mass Center
1. Comparison between the estimations from the ST-ACS and ST
methods indicates that in the higher stories these two methods
mostly result in approximately equal responses. However, esti-
mations from the proposed procedure (i.e., ST-ACS) are closer to
the NRHA results in lower stories. This issue is more accentu-
ated in the 20-stories building. Therefore, integrating the ACSM
with the proposed invariant load pattern leads to better
estimations.
2. Implementing the adaptive form of the proposed load pattern
cannot guarantee better estimations compared to the proposed
invariant load pattern. Considering the fact that a pushover
analysis with the adaptive load pattern (i.e., STA-ACS method)
requires nonlinear analysis software with the capability of
updating the load pattern at each step of the analysis (which
is also computationally demanding), the proposed method
emerges as a practically desirable method compared to the
STA-ACS method.
3. Estimations from the ST-ACS method have mostly a higher or
equal accuracy when compared to estimations from the MPA
method. The proposed procedure is also successful in capturing
the trend of drift profiles resulted from NRHA better than the
other considered pushover methods. Implementation of the
proposed method is practically easier than the MPA method
(especially for high-rise buildings) due to the fact that the ST-
ACS method is a single-run procedure and the contribution of
the higher modes are included in one load pattern.
In this study, drift values in stories of the structure is used to
evaluate the accuracy of the proposed method, because the dam-
age state of building structures subjected to earthquake ground
motions is significantly governed by this important seismic
response. However, it is also valuable to evaluate the accuracy of
the proposed method (along with other pushover methods) based
on the local (i.e., component-based) nonlinear responses in struc-
tures. Such efforts are the topic of an ongoing study by the authors.
References
[1] ATC. Seismic evaluation and retrofit of concrete buildings, Report ATC-40.
Redwood City, CA: Applied Technology Council; 1996.
[2] FEMA. Prestandard and commentary for the seismic rehabilitation of buildings,
FEMA 356. Washington, DC: Federal Emergency Management Agency; 2000.
[3] Chopra AK, Goel RK. A modal pushover analysis procedure for estimating
seismic demands for buildings. Earthquake Eng Struct Dynam
2002;31:561–82.
[4] Aydinoğlu MN. An incremental response spectrum analysis procedure based
on inelastic spectral displacements for multi-mode seismic performance
evaluation. Bull Earthquake Eng 2003;1:3–36.
[5] Antoniou S, Pinho R. Development and verification of a displacement-based
adaptive pushover procedure. J Earthquake Eng 2004;8(5):643–61.
[6] Kalkan E, Kunnath SK. Adaptive modal combination procedure for nonlinear
static analysis of building structures. J Struct Eng, ASCE
2006;132(11):1721–31.
44 K. Tarbali, K. Shakeri / Engineering Structures 79 (2014) 32–44
[7] Shakeri K, Shayanfar MA, Kabeyasawa T. A story shear-based adaptive
pushover for estimating seismic demands of buildings. Eng Struct
2010;32:174–83.
[8] Shakeri K, Tarbali K, Mohebbi M. Modified adaptive modal combination
procedure for nonlinear static analysis of bridges. J Earthquake Eng
2013;17(6):918–35.
[9] Jiang Y, Li G, Yang D. A modified approach of energy balance concept based
multimode pushover analysis to estimate seismic demands for buildings. Eng
Struct 2010;32:1272–83.
[10] Panyakapo P. Cyclic pushover analysis procedure to estimate seismic demands
for buildings. Eng Struct 2014;66:10–23.
[11] Tso WK, Moghadam AS. Seismic response of asymmetrical buildings using
push-over analysis. In: Proceeding of workshop on seismic design
methodologies for the next generation of codes, Bled, Slovenia
1997. Rotterdam: Balkema; 1997.
[12] Moghadam AS, Tso WK. Pushover analysis for asymmetrical multistory
buildings. In: Proceedings of the 6th U.S. national conference on earthquake
engineering, EERI, Oakland, CA; 1998.
[13] Kilar V, Fajfar P. Simple push-over analysis of asymmetric buildings.
Earthquake Eng Struct Dynam 1997;26:233–49.
[14] Faella G, Kilar V. Asymmetric multistorey R/C frame structures: push-over
versus nonlinear dynamic analysis. In: Proceedings of the 11th European
conference on earthquake engineering. Rotterdam: Balkema; 1998.
[15] Moghadam AS, Tso WK. 3-D pushover analysis for damage assessment of
buildings. J Seismol Earthquake Eng 2000;2(3).
[16] Ayala AG, Tavera EA. A new approach for the evaluation of the seismic
performance of asymmetric buildings. In: Proceeding of 7th national
conference on earthquake engineering; 2002.
[17] Penelis GR.G, Kappos A.J. 3D pushover analysis: the issue of torsion. In
proceeding of 12th European Conference on Earthquake Engineering Paper
Reference 015 2002.
[18] Penelis GR G, Kappos AJ. Inelastic torsional effects in 3D pushover analysis of
buildings. In proceeding of 4th European workshop on the seismic behavior of
irregular and complex structures. Greece, paper No. 51; 2005.
[19] Chopra AK, Goel RK. A modal pushover analysis for estimate seismic demands
for unsymmetric-plan buildings. Earthquake Eng Struct Dynam
2004;33:903–27.
[20] Reyes JC, Chopra AK. Three-dimensional modal pushover analysis of buildings
subjected to two components of ground motion, including its evaluation for
tall buildings. Earthquake Eng Struct Dynam 2011;40:789–806.
[21] Poursha M, Khoshnoudian F, Moghadam AS. A consecutive modal pushover
procedure for nonlinear static analysis of one-way unsymmetric-plan tall
building structures. Eng Struct 2011;33:2417–34.
[22] Fajfar P, Marusic D, Perus I. Torsional effects in the pushover-based seismic
analysis of buildings. J Earthquake Eng 2005;9(6):831–54.
[23] Meireles H, Pinho R, Bento R. Verification of an adaptive pushover technique
for the 3D case. In: Proceeding of First European conference on earthquake
engineering and seismology Geneva, Switzerland, Paper Number: 619; 2006.
[24] Pinho R, Bento R. Bhatt C. Assessing the 3-D irregular SPEAR building with
nonlinear static procedures. In: Proceedings of the 14th world conference on
earthquake engineering, China; 2008.
[25] Bento R, Bhatt C, Pinho R. Adaptive capacity spectrum method for seismic
assessment of 3D Irregular SPEAR buildings. Earthquakes Struct
2010;1(2):177–95.
[26] Adhikari G, Pinho R. Development and application of nonlinear static
procedures for plan-asymmetric buildings. Report No. ROSE-2010/01. Pavia,
Italy: IUSS Press; 2010.
[27] Bhatt C, Bento R. Comparison of nonlinear static methods for the seismic
assessment of plan irregular frame buildings with non seismic details. J
Earthquake Eng 2012;16:15–39.
[28] Shakeri K, Tarbali K, Mohebbi M. An adaptive modal pushover procedure for
asymmetric-plan buildings. Eng Struct 2012;36:160–72.
[29] Casarotti C, Pinho R. An adaptive capacity spectrum method for assessment of
bridges subjected to earthquake action. Bull Earthquake Eng
2007;5(3):377–90.
[30] CSI (Computers and Structures Inc.). SAP2000 v16 Analysis Reference Manual.
CSI, Berkeley; 2013.
[31] Chopra AK. Dynamic of structures: theory and applications to earthquake
engineering. 2nd edn. Englewood Cliffs, NJ: Prentice-Hall; 2003.
[32] FEMA. State of the art report on systems performance of steel moment frames
subject to earthquake ground shaking, FEMA 355C. Sacramento, CA: SAC Joint
Venture; 2000.
[33] Gupta A, Krawinkler H. Seismic demands for performance evaluation of steel
moment resisting frame structures (SAC Task 5.4.3). Report no. 132. Palo Alto,
CA: John A. Blume Earthquake Engineering Center, Stanford University; 1999.
[34] OpenSees. Open system for earthquake engineering simulation. <http://
opensees.berkeley.edu>; 2013.
[35] Somerville P, Smith N, Punyamurthula S, Sun J. Development of ground motion
time histories for Phase 2 of the FEMA/SAC Steel Project. SAC Background
Document, Report No. SAC/BD09/04, Sacramento, California; 1997.