Journal of Constructional Steel Research 79 (2012) 83–90

Contents lists available at SciVerse ScienceDirect

Journal of Constructional Steel Research

Response modification factor for steel moment-resisting frames by different pushover

analysis methods
Mohssen Izadinia a,⁎, Mohammad Ali Rahgozar b, Omid Mohammadrezaei a
a
Department of Civil Engineering, Najafabad Branch, Islamic Azad University, Najafabad, Iran
b
Department of Civil Engineering, Faculty of Engineering, University of Isfahan, Isfahan, Iran

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

Article history: The earthquake loads imposed to the structures are generally much more than what they are designed for.
Received 26 August 2011 This reduction of design loads by seismic codes is through the application of response modification factor
Accepted 23 July 2012 (R-factor). During moderate to severe earthquakes, structures usually behave inelastically, and therefore in-
Available online 30 August 2012
elastic analysis is required for design. Inelastic dynamic analysis is time consuming and interpretation of its
results demands high level of expertise. Pushover analysis, recently commonly used, is however, a simple
Keywords:
Response modification factor
way of estimating inelastic response of structures. Despite its capabilities, conventional pushover analysis
Conventional pushover analysis (CPA) does not account for higher mode effects and member stiffness changes. Adaptive pushover analysis
Adaptive pushover analysis (APA) method however, overcomes these drawbacks. This research deals with derivation and comparison
Steel moment-resisting frames of some seismic demand parameters such as ductility based reduction factor, Rμ, overstrength factor, Ω,
and in particular, response modification factor, R, from capacity curves obtained from different methods of
APA and CPA. Three steel moment-resisting frames of 3, 9 and 20 stories adopted from SAC steel project
are analyzed. In pushover analyses for each frame, eight different constant as well as adaptive lateral load
patterns are used. Among the main conclusions drawn is that the maximum relative difference for response
modification factors was about 16% obtained by the methods of conventional and adaptive pushover
analyses.
© 2012 Elsevier Ltd. All rights reserved.

1. Introduction seismic response evaluation of tall buildings. Therefore, Moghadam

and Tso [5] and later Chopra and Goel [6] introduced multi-mode
Previously elastic analysis was the main tool in seismic design of methods to overcome this problem. The most applicable method
structures. However, behavior of structures during recent earth- among them is modal pushover analysis (MPA) in which the struc-
quakes indicates that relying on just elastic analysis is not sufficient. ture under a load pattern corresponding to the elastic mode shapes
On the other hand, nonlinear dynamic analysis, although yields accu- is pushed to a certain lateral displacement. Then the results obtained
rate results, is time consuming and at times complex. Such analysis for each mode are combined using SRSS or CQC methods. Another
must be repeated for a group of acceleration time histories, not to drawback that is common in both CPA as well as MPA is the lack of ac-
mention the need for delicate interpretation of its results. Researchers counting for the change in member and/or global stiffness matrices at
have long been interested in developing fast and efficient methods to subsequent steps of analysis. In each step plastic hinges form and the
simulate nonlinear behavior of structures under earthquake loads. structure further goes in inelastic range, followed by a reduction in
The idea of inelastic static pushover analysis was first introduced in structural global stiffness. However, the load pattern is still kept
1975 by Freeman for single degree of freedom (SDOF) systems. based on the original stiffness and elastic mode shapes. In other
Then other researchers extended this method for multi-degree of words, the lateral load pattern is not in conformance with the re-
freedom systems [1–4]. Conventional pushover analysis (CPA), de- duced stiffness. Bracci et al. [7], Sasaki et al. [8], Satyarno et al. [9],
spite its strengths, has some drawbacks. For example, the shape of lat- Matsumori et al. [10], Gupta and Kunnath [11], Reqena and Ayala
eral load pattern stays the same during analysis. This shape is usually [12], and Elnashai [13] proposed different ways of conforming the
based on the first elastic mode of the structure. In other words, the loading pattern with the structural stiffness. The method of adaptive
higher mode effects or the role of more effective modes are not pushover analysis (APA) was first developed in 2004 by Antoniou
accounted for. The latter may be the source of significant errors in and Pinho [14]. Not only is this method multi-mode based, but also,
the lateral loading pattern is adapted according to the changes in stiff-
⁎ Corresponding author. Tel.: +98 3112231331; fax: +98 3112240297.
ness matrix at each step of the analysis. Following the Northridge
E-mail addresses: izadinia@iaun.ac.ir (M. Izadinia), rahgozar@eng.ui.ac.ir earthquake in 1994, the seismic design provisions of design and ma-
(M.A. Rahgozar). terial codes such as ASCE, UBC, AISC and ACI codes fundamentally

0143-974X/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jcsr.2012.07.010
84 M. Izadinia et al. / Journal of Constructional Steel Research 79 (2012) 83–90

changed. Equivalent static method in most seismic design codes is solving eigenvalue problem, the mode shapes and mode participation
based on the use of response modification factor, R (or sometimes factors are derived. Floor loads at each step, then, would be:
called force reduction factor). In fact design loads are obtained by re-
ducing/dividing the earthquake loads by the R factor. By reducing the F ij ¼ Γ j φij Mi ð2Þ
earthquake loads the structure will enter into inelastic range. There-
fore, in order to dissipate the earthquake energy, the structure will where i is the floor number, j is the mode number, Гj is the j th mode
have to experience rather large inelastic deformations. The structural participation factor, ϕij is the j th mode value at the i th floor, and Mi
capacity in withstanding the earthquake loads is related to its capac- is the mass at the i th floor. Eq. (2) provides floor load corresponding
ity in deforming in inelastic range, or its ductility capacity. For a sys- to unit response acceleration. For a given mode j, with known fre-
tem with idealized bilinear behavior (see Fig. 1), structural ductility, μ quency ωj or period Tj, spectral acceleration Sa,j would be available
is defined as the ratio of maximum displacement to the displacement and Eq. (2), changes as below:
corresponding to the yielding point. Structures with higher force re-
duction factor, R, require higher ductility capacity, μ. Therefore, R F ij ¼ Γ j φij Mi Sa;j ð3Þ
and μ factors are interrelated and play important role in energy dissi-
pation mechanism of the structures. To obtain the value of floor load, Fi, for a given floor number i, the
In order to overcome the shortcomings of the CPA, Antoniou and floor loads corresponding to different modes are combined using
Pinho proposed two different methods for adaptive pushover analy- SRSS or CQC methods. Therefore, at each step, there would be one sin-
ses; force-based adaptive pushover analysis (FAPA) [14] and displace- gle load pattern. Since the shape of load pattern is important and not
ment based adaptive pushover analysis (DAPA) [15]. its magnitude, the load values for each floor are normalized by the
total value, i.e., the sum of all the floor loads in that step:
2. Force-based adaptive pushover analysis (FAPA)
Fi
F i ¼ : ð4Þ
In any adaptive pushover analysis, in each step, the software up- X
N
Fi
dates the lateral loading. In FAPA the updating algorithm includes i¼1
four parts: 1) defining nominal lateral load vector and floor inertial
mass; 2) derivation of load factor; 3) derivation of normalized load Having known F t , λt and Δλt at any analysis step t, and P0, the
vector applicable to the structure; and 4) updating the load vector. adaptive load vector Pt can be obtained using either of the below
The first part happens only once in the beginning of analysis. The equations relating to incremental or total updating.
other three parts repeat for each step in FAPA analysis. The load pat-
tern vector is automatically obtained and updated according to the P t ¼ P t−1 þ Δλt ⋅F t ⋅P 0 ð5Þ
above algorithm. The nominal force vector P0 is defined uniformly
along the height. The distribution of load along the height at each P t ¼ λt ⋅F t ⋅P 0 ð6Þ
step is through the normalized force vector F , which is derived
based on dynamic characteristics of the structure for that step and where Pt − 1 is the adaptive load vector for the previous step. Re-
the elastic response spectra of the given earthquake. Therefore, the searchers showed that the results obtained using FAPA method may
floor inertial masses are also required. The load vector P at each be at times erroneous. This could be attributed to the use of SRSS
step is obtained by multiplying the nominal load vector P0 by the method for combining the modal floor loads. Since irrespective of
load factor λ for that step (Eq. (1)). The load factor λ depends on the sign of modal values, they always participate in the combined
the type of analysis (load control or response control) and the num- floor load with a positive sign, and this may discount the higher
ber of steps. In other words, the management of lateral load increase mode effects [15,16]. Therefore, Antoniou and Pinho introduced dis-
is by application of this factor. placement based adaptive pushover (DAPA) method [15] that is to
overcome the aforementioned issues.
P ¼ λ⋅ P 0 ð1Þ
3. Displacement based adaptive pushover analysis (DAPA)
Normalized load vector F , computed in the beginning of each step,
provides the shape of increasing load vector at each step. Any stiffness The proposed algorithm at each step contains four parts: 1) defining
changes must be reflected in this vector. Therefore, at each step, by nominal lateral load vector U0 and floor inertial mass; 2) derivation of

Base shear

Veu Actual Response

1 / Rμ

1 / Ru Vy
1/ Ω
1/ RW
Idealized Response
First Plastic
Vs Hinge μ
1/ Y

Vw Cds

Cdw
Story Drift
Δw Δs Δy Δe Δm

Fig. 1. Capacity curve for a structure along with its bilinear idealization in pursuit of seismic demand parameters [18].
 M. Izadinia et al. / Journal of Constructional Steel Research 79 (2012) 83–90 85

load factor; 3) derivation of normalized load vector applicable to the vector Ut at any step t of the DAPA analysis can be updated following
structure; and 4) updating the displacement vector. The first two one of the equations, Eq. (12) or (13). Ut − 1 is the adaptive load vector
parts are similar to FAPA method, except that the load vector is now dis- in the previous step t − 1.
placement based. The load vector U is defined as:
 ⋅U
U t ¼ U t−1 þ Δλt ⋅D ð12Þ
t 0
U ¼ λ⋅U 0 ð7Þ
 ⋅U
U t ¼ λt ⋅ D ð13Þ
The normalized D  vector, computed in the beginning of each step or t 0

at the end of last step represents the shape of the load vector (or the in-
crease in load vector). At the end of each step (after application of each
4. Response modification factor (R factor)
load increment) an eigenvalue problem is solved and depending on the
current stiffness of the system, the mode shapes and participation fac-
Researchers have so far proposed different methodologies for der-
tors are derived. Modal floor loads can be combined by SRSS or CQC
ivation of R factor. These methods in general, fall into two main
methods. Normalization of load vector in DAPA method is either
groups: the European and the American methods. In this study one
based on the story displacement or the interstory displacement. Both
of the most important American methods, so-called Uang method, is
are explained in the following.
adopted. The parameters used in Uang method, illustrated in Fig. 1,
are defined in the following [17,18].
3.1. Normalization based on story displacement
Fig. 1 depicts variation of structural base shear versus story total
drift in a typical pushover analysis. This curve is idealized as the re-
The load vector, in this method, can be obtained directly from
sponse of bilinear elasto-plastic system in pursuit of seismic demand
modal analysis (Eq. (8)). This is similar to FAPA method, except that
parameters including R factor.
instead of force components, the story modal displacements are com-
bined using SRSS method [15].
4.1. Global ductility of the structure
vffiffiffiffiffiffiffiffiffiffiffiffiffiffi vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
uX u n  2
u n 2 uX Global ductility ratio of the structure, μs is defined as ratio of max-
Di ¼ t Dij ¼ t Γ j ⋅φij ð8Þ
j¼1 j¼1 imum lateral displacement (Δmax) to lateral displacement at yield
(Δy).
Dij is the i th floor displacement due to the j th mode.
Δ max
μS ¼ ð14Þ
3.2. Normalization based on interstory displacement Δy

Since maximum interstory displacement better describes the level

of damage during an earthquake than maximum story displacement, 4.2. Ductility based force reduction factor (Rμ)
Antoniou and Pinho [14] proposed the following method of normaliz-
ing lateral load vector that is based on interstory displacement. It is Structures with ductility capacity can dissipate hysteretic energy
assumed that floor displacement Di at each floor level i is the sum of of earthquakes. Therefore, maximum elastic earthquake force (base
interstory displacements below that level. Also, it is assumed that shear, Veu) can be reduced to structural general yield strength (Vy)
interstory displacement at level i is the SRSS combination of modal at collapse occurrence. Ductility based reduction factor Rμ can then
interstory displacements [15]. be defined as
vffiffiffiffiffiffiffiffiffiffiffiffiffiffi vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
X uX u n h  i2 V eu
i u n 2 uX Rμ ¼ : ð15Þ
Di ¼ Δk with Δi ¼ t Δij ¼ t Γ j φi;j −φi−1;j ð9Þ Vy
k¼1 j¼1 j¼1

For a given earthquake and/or design response spectrum Eq. (9)

4.3. Overstrength factor (Ω)
turn into Eq. (10) where Sd,j is the j th modal displacement.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffi vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Overstrength factor represents the reserved strength in the struc-
uX uX
X
i n
iu u n h   i2
tures between the general yield point (Vy) and the formation of the
Di ¼ Δk with Δi ¼ t
Δij ¼ t
2
Γ j φi;j −φi−1;j Sd;j ð10Þ
k¼1 j¼1 j¼1 first plastic hinge (Vs).

Vy
Although results show an improvement using the latter method for Ω¼ ð16Þ
Vs
derivation of load vector, this is still an approximate method, because
of the assumed simultaneous occurrence of maximum interstory dis-
placements, while this is not usually the case. However, due to the im-
provement of the results this method is adopted as standard DAPA 4.4. Allowable stress factor (Y)
analysis method in this research [15]. The final shape of the load pat-
tern in DAPA method is taken from the shape of normalized floor dis- Depending on the definition of design stresses (allowable or ulti-
placements. Normalized floor displacement D  i is obtained by dividing mate stress) in different design codes (ASD and LRFD), the Y factors
the i th floor displacement by the maximum floor displacement. could have different values. In general, the allowable stress factor Y
is defined as the ratio of structural strength (base shear) at formation
 ¼ Di of the first plastic hinge (Vs) to the strength corresponding to allow-
D i ð11Þ
max Di able design stresses (Vw).

Having derived normalized vector D  i ; primary nominal load vector Vs

Y¼ ð17Þ
U0, load factor λt and/or incremental load factor Δλt, the adaptive load Vw
86 M. Izadinia et al. / Journal of Constructional Steel Research 79 (2012) 83–90

Table 1 the structure cannot take any more lateral load; and 2) the arrival
First three modal periods of structures (s). of an interstory drift limit usually set by the seismic codes. In the seis-
Building Mode number mic design code of Iran, Standard 2800 [23], for structure with a fun-
damental period T ≤ 0.7 s, this limit is 3.57% and for T > 0.7 s, it is
1 2 3
2.85%. For a given performance level, this lateral target displacement
SAC-3 0.379 0.125 0.080
is the maximum lateral displacement the structure likely experiences
SAC-9 1.008 0.385 0.225
SAC-20 1.656 0.588 0.343 during the design earthquake for the given hazard level. The capacity
(pushover) curve is, of course, sensitive to the degradation and
post-peak negative stiffness. According to section 3.3.3.2.5 of ASCE
41.06 [22] this displacement shall be the calculated target displace-
ment or the displacement corresponding to the maximum base
Table 2 shear whichever is least. This approach is consistent with the primary
Component yield criteria for different members in steel moment resisting frames [22]. life safety performance objective of seismic regulations of model
Rotation at yield Expected flexural strength Member building codes [26]. Several methods for determination of target dis-
θy ¼
ZF ye Lb
MCE = ZFye Beams
placement have been proposed in literature and associated with
6EI b
    structural performance levels. In the present study the allowable dis-
ZF ye Lc
θy ¼ 1− PPye M CE ¼ 1:18ZF ye 1− PPye ≤ZF ye Columns
6EI c
placement for the design earthquake of Iranian seismic code has been
Fye = expected yield strength of the material, Lb = beam length, Lc = column height, considered.
MCE = expected flexural strength, P = axial force at target displacement in pushover
analysis, Pye = expected axial yield force of the member, θy = yield rotation and Z =
7. Derivation of response modification factor, R, by different
plastic section modulus.
pushover analysis methods

Therefore, response modification factor (R-factor) in seismic codes The strategy in this paper for derivation of seismic demand pa-
allowing ASD (allowable stress design) method would be: rameters such as R factor is to first obtain the capacity and/or push-
over curves for the given structure. Then by bilinear idealization of
V eu V eu V y V s such curve as per Fig. 1, the demand parameters can be derived.
R¼ ¼ ¼ Rμ ΩY: ð18Þ
Vw Vy Vs Vw Three main types of pushover analysis are performed in this
study: 1) conventional pushover analysis (CPA), 2) force based adap-
tive pushover analysis (FAPA), and 3) displacement based adaptive
5. Structural models pushover analysis (DAPA).
In CPA the load pattern is kept constant throughout the analysis.
Following the objectives in this research, three buildings of 3, 9, Two constant load patterns considered in this study are: 1) UNIFORM
and 20 stories, previously designed and studied by SAC steel project load pattern in which for a given floor, the lateral floor load is propor-
[19,20] are used. These models hereafter are called SAC-3, SAC-9 tional to its mass/weight, and 2) MODAL load pattern, where the lateral
and SAC-20. The lateral load resisting system in these buildings is floor load is proportional to the story shear distribution calculated by
moment-resisting frames that are located in the perimeter of the SRSS combination of modal responses (Seismic design code of Iran,
buildings. The building site is in Los Angeles, CA, USA and the design Standard 2800 [23]). Response spectrum analysis of the structures
code is UBC-1994 [19]. Geometric properties of the SAC frames are il- was performed by the ETABS software [24].
lustrated in Fig. 11 and their first three modal periods are given in The Sd,j and Sa,j used in this study respectively for DAPA and FAPA, are
Table 1. The Seismostruct software is used to perform all pushover 5% damped response spectra of three major earthquakes: Northridge
analyses [21]. This software takes advantage of fiber elements that 1994, Tabas 1978 and Imperial Valley 1940. Table 3 shows the character-
are capable of accounting for material nonlinearity. The PΔ effect is istics of these ground motions. In all modal and adaptive pushover anal-
considered in the analyses. The steel properties are selected similar to yses the first 10 modes are used. For each building eight pushover
the original study, i.e., yield stress for beams Fy = 36 ksi (248 MPa) analyses are performed: two CPA (with constant load patterns of Uni-
and for columns= 50 ksi (345 MPa) and modulus of elasticity, Es = form and Modal), three FAPA (for three mentioned earthquakes), and
29,000 ksi (200,000 MPa). Nonlinear behavior of steel is assumed to three DAPA (for the three earthquakes).
be bilinear with 3% strain hardening. Load vector in each analysis step t can be obtained by total updating
or incremental updating [14]. Incremental updating is considered in all
6. Performance criteria numerical computations of the models. This is performed in accordance
with Eqs. (5) and (12). Load factor increment (Δλt ) in each step was
The yield criteria for steel beams and columns, applicable for determined according to loading or solution schemes including “load
pushover analysis, are taken from ASCE 41-06 [22]. Table 2 depicts control” and “response control”. The response control scheme was
component yield criteria for different members in steel moment employed in FAPA and CPA analyses and the load control scheme in
resisting frames [22]. DAPA analysis. The parameters P0 equal to 100 (kN) and U0 equal
The output of any pushover analysis is the variation of base shear to 0.01 m was selected and total number of loading steps was as-
with lateral displacement. The criteria as to where the ultimate capac- sumed equal to 500. Load factor increment (Δλt ) in DAPA analysis
ity of the building structure is arrived and the analysis is completed was equal to ((1 / 500) ∗ (allowable roof displacement for design
are twofold: 1) formation of collapse and/or mechanism, i.e., where earthquake adopted by Iranian seismic code in meter)) and in FAPA

Table 3
Earthquake characteristics from PEER site database.

Earthquake name, date Station Component PGA (g) PGV (cm/s) PGD (cm)

Tabas, 1978/09/16 9101TABAS TAB-LN 0.836 97.8 36.92

Northridge, 1994/01/17 24278 Castaic-Ridge Route ORR090 0.568 52.1 4.21
Imperial Valley, 1940/05/19 117 El Centro Array # 9 I-ELC180 0.313 29.8 13.32
 M. Izadinia et al. / Journal of Constructional Steel Research 79 (2012) 83–90 87

8000 14000
7000 12000
6000
BASE SHEAR (KN)

BASE SHEAR (KN)

10000
5000
8000
4000
UNIFORM UNIFORM
MODAL
6000
3000 MODAL

4000
2000

1000 2000
TOTAL DRIFT TOTAL DRIFT
0 0
0% 1% 2% 3% 0% 0.5% 1% 1.5% 2% 2.5%

Fig. 2. CPA pushover/capacity curves for SAC-3 structure for two load pattern, and drift Fig. 5. CPA pushover/capacity curves for SAC-9 structure for two load pattern, and drift
limit of 3.57%. limit of 2.85%.

7000 14000

6000 12000
BASE SHEAR (KN)

BASE SHEAR (KN)

5000 10000

4000 NORTHRIDGE 8000

TABAS NORTHRIDGE
3000 ELENTRO 6000 TABAS
ELENTRO
2000 4000

1000 2000
TOTAL DRIFT TOTAL DRIFT
0 0
0% 1% 2% 3% 0% 0.5% 1% 1.5% 2% 2.5%

Fig. 3. FAPA pushover/capacity curves for SAC-3 structure for three earthquake spectra, Fig. 6. FAPA pushover/capacity curves for SAC-9 structure for three earthquake spectra,
and drift limit of 3.57%. and drift limit of 2.85%.

and CPA analyses was calculated by software for attainment of a de- 8. Results and discussion
termined response displacement increment in a controlled node.
Figs. 2 to 10 illustrate the results for these analyses for the three Tables 4 to 6 list all seismic parameters shown in Fig. 1 from eight
structural models of SAC-3, SAC-9, and SAC-20. The locations of plas- different pushover analyses required in derivation of ductility ratio, μ,
tic hinges are depicted in Fig. 11 at the arrival of interstory drift limit overstrength factor, Ω, for SAC-3, SAC-9, and SAC-20 structures,
set by the seismic design code of Iran. The FAPA and DAPA results in respectively.
this figure are based on Northridge 1994 earthquake. In derivation of Rμ, the maximum elastic base shear, Veu is re-
The bilinear idealization of the pushover curves shown in Figs. 2 to quired. Subjected to a given acceleration time history and assuming
10 (Fig. 1) was performed following the criteria set by ASCE 41-06 elastic behavior for the structure, the maximum base shear recorded
[22] by the MATLAB software. would be Veu. Miranda and Bertero [25], having conducted a large
parametric study on nonlinear response of SDOF systems to different

7000 12000

6000 10000
BASE SHEAR (KN)

BASE SHEAR (KN)

5000
8000
4000
NORTHRIDGE 6000 NORTHRIDGE

3000 TABAS TABAS

ELENTRO ELCENTRO
4000
2000

1000 2000
TOTAL DRIFT TOTAL DRIFT
0 0
0% 1% 2% 3% 0% 0.5% 1% 1.5% 2% 2.5%

Fig. 4. DAPA pushover/capacity curves for SAC-3 structure for three earthquake spectra, Fig. 7. DAPA pushover/capacity curves for SAC-9 structure for three earthquake spectra,
and drift limit of 3.57%. and drift limit of 2.85%.
88 M. Izadinia et al. / Journal of Constructional Steel Research 79 (2012) 83–90

9000 Having obtained Rμ and assuming a design allowable stress factor

8000 of Y = 1.5 [18], the R factors for all structures and under all pushover
types are derived and listed in Tables 7 to 9.
7000
BASE SHEAR (KN)

6000
5000
4000
UNIFORM
3000 MODAL
2000
1000 TOTAL DRIFT
0
0% 0.5% 1% 1.5% 2% 2.5%

Fig. 8. CPA pushover/capacity curves for SAC-20 structure for two load patterns, and
drift limit of 2.85%.

9000
8000
7000
BASE SHEAR (KN)

6000
5000 NORTHRIDGE
TABAS
4000 ELCENTRO

3000
2000
1000 TOTAL DRIFT
0
0% 0.5% 1% 1.5% 2% 2.5%

Fig. 9. FAPA pushover/capacity curves for SAC-20 structure for three earthquake spectra,
and drift limit of 2.85%.

earthquake records on different sites, proposed the following empir-

ical relationship for Rμ.

μ−1
Rμ ¼ þ1 ð20Þ
Φ

μ is the global ductility ratio and Φ as defined below (for sediment

foundation soils) is a function of μ and the fundamental period of the
structure, T.

1 2 −2ð lnðT Þ−0:2Þ2

Φ¼1þ − e ð21Þ
12T−μT 5T

9000
8000
7000
BASE SHEAR (KN)

6000
5000 NORTHRIDGE
TABAS
4000 ELCENTRO

3000
2000
1000
TOTAL DRIFT
0
0% 0.5% 1% 1.5% 2% 2.5%
Fig. 11. Frame geometries and locations of plastic hinges at arrival of code drift limit, a.
Fig. 10. DAPA pushover/capacity curves for SAC-20 structure for three earthquake DAPA with Sd,j from Northridge earthquake, b. FAPA with Sa,j from Northridge earthquake,
spectra, and drift limit of 2.85%. c. CPA with uniform load pattern, d. CPA with modal load pattern (1 ft=0.3048 m).
 M. Izadinia et al. / Journal of Constructional Steel Research 79 (2012) 83–90 89

Fig. 11 (continued).

For each building structure, depending on the number of earth- efficient methods to simulate nonlinear behavior of structures under
quakes or the number of constant load patterns, there are more earthquake loads. Conventional pushover analysis (CPA), despite its
than one or two R factors. ASCE 41-06 recommends using the smaller strengths, has some drawbacks. For example, the shape of lateral
R factor. This is because the smaller R factor will lead to a larger de- load patterns is constant and stays the same during analysis. This
sign base shear and/or a safer design. Table 10 provides a short/final shape is usually based on the first elastic mode of the structure. In
list of the R factors for the three structures used in this study. other words, the higher mode effects or the role of more effective
modes are not accounted for. Later modal pushover analysis (MPA)
9. Summary and conclusions was introduced which accounts for higher mode effects. A common
drawback in both CPA as well as MPA is the lack of accounting for
Depending on the severity of the design earthquake, the structures the change in member and/or global stiffness matrices during push-
may undergo nonlinear behavior. Nonlinear dynamic analysis, al- over analysis. Adaptive pushover analysis (APA) was therefore devel-
though yields accurate results, is time consuming and at times com- oped in 2004 by Antoniou and Pinho, which not only is multi-mode
plex. Researchers have long been interested in developing fast and based, but also, the changes in stiffness matrix at each step of the

Table 4 Table 5
Results of different pushover analyses, ductility ratios, μ, and overstrength factors, Ω, Results of different pushover analyses, ductility ratios, μ, and overstrength factors, Ω,
for SAC-3. for SAC-9.

Load pattern Δs Δy Δ max Vs Vy Vmax Ω μ Load pattern Δs Δy Δ max Vs Vy Vmax Ω μ

(cm) (cm) (cm) (kN) (kN) (kN) (cm) (cm) (cm) (kN) (kN) (kN)

CPA, uniform 7.56 14.57 39.48 3305.1 6376 7107 1.93 2.71 CPA, uniform 18.70 38.25 60.56 5613.8 11,495 12,635 2.05 1.58
CPA, modal 8.41 14.02 32.94 2968.8 4948 5794 1.67 2.35 CPA, modal 23.37 44.98 77.56 5463.6 10,519 11,622 1.92 1.72
DAPA, Northridge 8.15 15.00 38.71 2965.4 5455 6239 1.84 2.58 DAPA, Northridge 22.95 44.62 77.77 5464.1 10,624 11,695 1.94 1.74
DAPA, Tabas 8.23 15.00 37.18 3000.8 5461 6283 1.82 2.48 DAPA, Tabas 24.20 43.91 70.12 5392.8 9775 11,330 1.81 1.60
DAPA, El Centro 8.15 15.00 37.61 2965.0 5454 6222 1.84 2.51 DAPA, El Centro 24.22 47.67 88.4 5506.6 10,788 11,947 1.97 1.85
FAPA, Northridge 8.24 14.01 34.89 3069.3 5221 6088 1.70 2.49 FAPA, Northridge 21.46 42.85 66.72 5488.1 10,967 12,028 2.00 1.56
FAPA, Tabas 7.98 14.43 37.53 3138.5 5676 6464 1.81 2.60 FAPA, Tabas 19.76 40.37 62.26 5612.3 11,477 12,505 2.04 1.54
FAPA, El Centro 8.24 14.15 34.98 3053.7 5247 6102 1.72 2.47 FAPA, El Centro 20.19 40.73 63.11 5549.7 11,207 12,313 2.02 1.55
90 M. Izadinia et al. / Journal of Constructional Steel Research 79 (2012) 83–90

Table 6 Table 10
Results of different pushover analyses, ductility ratios, μ, and overstrength factors, Ω, Response modification factor, R from different pushover analysis methods.
for SAC-20.
Structural models CPA FAPA DAPA
Load patterns Δs Δy Δ max Vs Vy Vmax Ω μ
SAC-3 5.31 5.71 6.06
(cm) (cm) (cm) (kN) (kN) (kN)
SAC-9 5.54 5.31 4.97
CPA, uniform 27.23 53.08 95.54 3235.9 6308 7250 1.95 1.80 SAC-20 5.65 5.35 5.13
CPA, modal 32.31 63.08 113.54 2978.5 5816 6570 1.95 1.80 Average 5.5 5.45 5.39
DAPA, Northridge 33.69 66.16 126.93 3057.8 6002 6871 1.96 1.92
DAPA, Tabas 36.92 73.85 118.62 3292.7 6583 7354 2.00 1.61
DAPA, El Centro 34.62 68.46 139.39 3113.2 6156 7139 1.98 2.04
FAPA, Northridge 30.00 58.46 100.20 3145.3 6130 7035 1.95 1.71 The results' confidence can be improved by more analytic models
FAPA, Tabas 28.15 54.62 97.39 3231.4 6269 7213 1.94 1.78 with other assumptions in numerical computations such as method of
FAPA, El Centro 28.62 56.16 97.84 3142.2 6167 7101 1.96 1.74 maximum lateral displacement computation and period dependent
relations for ductility based force reduction factor (Rμ).

References
Table 7
Rμ and R factors for different pushover analyses for SAC-3 structure. [1] Freeman SA, Nicoletti JP, Tyrell JV. Evaluations of existing buildings for seismic
risk — a case study of Puget Sound Naval Shipyard, Bremerton, Washington.
CPA, CPA, DAPA, DAPA, DAPA, FAPA, FAPA, FAPA, Proc. of the First U.S. National Conference on Earthquake Engineering, Oakland,
uniform modal Northridge Tabas El Centro Northridge Tabas El Centro California; 1975. p. 113-22.
[2] Shibata A, Sozen MA. Substitute structure method for seismic design in reinforced
Rμ 2.41 2.12 2.30 2.22 2.25 2.23 2.32 2.21 concrete division. J Struct Div ASCE 1976;102(ST1):1–18.
R 6.98 5.31 6.35 6.06 6.21 5.69 6.30 5.71 [3] Saiidi M, Sozen M. Simple nonlinear seismic analysis of R/C structures. J Struct Div
ASCE 1981;10(ST5):937-52.
[4] Fajfar P, Fischinger MA. N2 — a method for non-linear seismic damage analysis of
regular buildings. Proceedings of the 9th WCEE, Tokyo–Kyoto, Japan, vol. V; 1998.
p. 111-6.
[5] Moghadam AS, Tso WK. A pushover procedure for tall buildings. Proc. of the Twelfth
Table 8 European Conference on Earthquake Engineering, London, United Kingdom, Paper,
Rμ and R factors for different pushover analyses for SAC-9 structure. No. 395; 2002.
[6] Chopra AK, Goel RK. A modal pushover analysis procedure for estimating seismic
CPA, CPA, DAPA, DAPA, DAPA, FAPA, FAPA, FAPA, demands for buildings. Earthquake Eng Struct Dyn 2002;31(3):561-82.
uniform modal Northridge Tabas El Centro Northridge Tabas El Centro [7] Bracci JM, Kunnath SK, Reinhorn AM. Seismic performance and retrofit evaluation
of reinforced concrete structures. J Struct Eng 1997;123(1):3–10.
Rμ 1.80 1.99 2.02 1.83 2.17 1.77 1.74 1.76 [8] Sasaki KK, Freeman SA, Paret TF. Multi-mode pushover procedure (MMP) — a
R 5.54 5.54 5.88 4.97 6.41 5.31 5.32 5.33 method to identify the effects of higher modes in a pushover analysis. Proceed-
ings of the Sixth U.S. National Conference on Earthquake Engineering; 1998.
[9] Satyarno I, Carr AJ, Restrepo J. Refined pushover analysis for the assessment of older
reinforced concrete buildings. Proc., NZSEE Technology Conference, Wairakei, New
analysis are accounted for. Antoniou and Pinho (2004) later intro- Zealand; 1998. p. 75-82.
duced two different versions of APA; namely, force based adaptive [10] Matsumori T, Otani S, Shiohara H, Kabeyasawa T. Earthquake member deforma-
tion demands in reinforced concrete frame structures. Proc., US–Japan Workshop
pushover analysis (FAPA) and displacement based adaptive pushover PBEE Methodology for R/C Building Structures, PEER Center Report, UC Berkeley,
analysis (DAPA) methods. Using different constant and adaptive load Maui, Hawaii; 1999. p. 79-94.
patterns in pushover analysis methods, this study dealt with deriva- [11] Gupta B, Kunnath SK. Adaptive spectra-based pushover procedure for seismic
evaluation of structures. Earthquake Spectra 2000;16(2):367-91.
tion of seismic demand parameters for steel moment-resisting.
[12] Requena M, Ayala G. Evaluation of a simplified method for the determination of
Among the main conclusions drawn are: the nonlinear seismic response of RC frames. Proc., 12th World Conference on
Earthquake Engineering (WCEE), Auckland, New Zealand, Paper No. 2109; 2000.
[13] Elnashai AS. Advanced inelastic static (pushover) analysis for earthquake applica-
1) R factors obtained by the methods of conventional (CPA) and
tions. Struct Eng Mech 2001;12(1):51-69.
adaptive (FAPA or DAPA) pushover analyses tend to be different. [14] Antoniou S, Pinho R. Advantages and limitations of adaptive and non-adaptive
The maximum relative difference for response modification fac- force-based pushover procedures. J Earthquake Eng 2004;8(4):497-522.
tors was about 16% due to larger results in adaptive pushover con- [15] Antoniou S, Pinho R. Development and verification of a displacement-based adap-
tive pushover procedure. J Earthquake Eng 2004;8(5):643-61.
sidering different seismic records in Tables 7, 8 and 9. [16] Shakeri K, Shayanfar MA, Kabeyasawa T. A story shear-based adaptive pushover pro-
2) Ductility ratios (μ) obtained by the methods of conventional and cedure for estimating seismic demands of buildings. Eng Struct 2010;32:174-83.
adaptive pushover analyses tend to be different. The maximum [17] Uang CM. Establishing R (or Rw) and Cd factors for building seismic provisions.
ASCE J Struct Eng 1991;117(1):19-28.
relative difference in ductility ratios was about 17% due to larger [18] Building and Housing Research Center. Derivation of response modification fac-
results in adaptive pushover. 3) Displacement based adaptive tors for concrete moment resisting frames. Tasnimi A., Masoumi A., Publication
pushover analyses (DAPA) yield higher inelastic lateral displace- No. 4361st edition. ; 2006.
[19] Gupta A, Krawinkler H. Seismic demands for performance evaluation of steel mo-
ments and/or ductility ratios compared to the other pushover ment resisting frame structures. (SAC Task 5.4.3). Report no. 132Palo Alto, CA:
methods. 4) For high-rise and mid-rise buildings (SAC-20 and John A. Blume Earthquake Engineering Center, Stanford University; 1999.
SAC-9) the different shapes of constant load pattern in CPA result [20] SAC Joint Venture. State of the art report on systems performance of steel mo-
ment frames subject to earthquake ground shaking, FEMA-355C; 2000.
in close R factors. 5) The use of different earthquake response [21] SeismoSoft. SeismoStruct — a computer program for static and dynamic nonlinear anal-
spectra for high-rise and mid-rise buildings in FAPA method ysis of framed structures [online]. Available from URL: http://www.seismosoft.com
does not have considerable effect on the R factors and related 2004.
[22] ASCE standard ASCE/SEI 41-06. Seismic rehabilitation of existing buildings. American
parameters.
Society of Civil Engineers9780784408841; 2007.
[23] Building and Housing Research Center. Iranian code of practice for seismic resis-
tant of buildings, Standard No. 2800. Publication No. 2533rd edition. ; 2006.
Table 9 [24] Computers and Structures, Inc., “Etabs 2000 — extended 3D analysis of building
Rμ and R factors for different pushover analyses for SAC-20 structure. systems, nonlinear version 9.5.0”, Berkeley, California 94704, USA.
[25] Miranda E, Bertero. Evaluation of strength reduction factors for earthquake-
CPA, CPA, DAPA, DAPA, DAPA, FAPA, FAPA, FAPA,
resistance design. J Eng Struct 1994;2(10):357-79.
uniform modal Northridge Tabas El Centro Northridge Tabas El Centro [26] FEMA. Quantification of building seismic performance factors, FEMA P695. Washington,
Rμ 1.93 1.93 2.07 1.71 2.21 1.83 1.91 1.86 D.C., U.S.A.: Federal Emergency Management Agency; 2009
R 5.65 5.65 6.09 5.13 6.56 5.35 5.56 5.47