Eerc 03 03

REPORT NO.
EERC 2003-03 EARTHQUAKE ENGINEERING RESEARCH CENTER
EVALUATION OF THE MODAL PUSHOVER

ANALYSIS PROCEDURE USING VERTICALLY
REGULAR AND IRREGULAR GENERIC
FRAMES
By
Chatpan Chintanapakdee
Anil K. Chopra
A Report on Research Conducted Under

Grant No. CMS-9812531 from the
National Science Foundation
COLLEGE OF ENGINEERING
UNIVERSITY OF CALIFORNIA, BERKELEY

Evaluation of The Modal Pushover Analysis
Procedure Using Vertically Regular and
Irregular Generic Frames
By
Chatpan Chintanapakdee
And
Anil K. Chopra
A Report on Research Conducted Under Grant No. CMS-9812531

from the National Science Foundation
Earthquake Engineering Research Center

University of California, Berkeley
December, 2002
UCB/EERC 2003-03
ABSTRACT
The recently developed modal pushover analysis (MPA) for estimating seismic demands
on buildings has been shown to be a significant improvement over the pushover analysis
procedures currently used in structural engineering practice. None of the current invariant
force distributions accounts for the contribution of higher modeshigher than the
fundamental modeto the response or for redistribution of inertial forces because of
structural yielding. By including the contributions of a sufficient number of modes of
vibration (generally two to three), the height-wise distribution of responses estimated by
MPA is generally similar to the exact results from nonlinear response history analysis
(RHA). Although the results of the previous research were extremely promising, only a
few buildings were evaluated. The results presented in this thesis evaluate the accuracy of
MPA for a wide range of frame buildings with strong-columns and weak-beams and a
ground motion ensemble. The selected structures are (1) sixty vertically regular frames
of six different heights: 3, 6, 9, 12, 15, and 18 stories and five strength levels
corresponding to SDF-system ductility factor of 1, 1.5, 2, 4, and 6; and (2) forty-eight
irregular frames, all 12 stories high, designed with three types of irregularitystiffness,
strength, and combined stiffness and strengthintroduced in eight different locations
along the height using two modification factors. Each frame was analyzed for an
ensemble of twenty ground motions.
Investigated first is the basic premise that the roof displacement of a multistory
building can be determined from the deformation of an SDF system. The responses of
both systems are determined rigorously by nonlinear RHA without introducing any of the
approximations underlying the simplified methods for estimating the deformation of an
SDF system (see, e.g., FEMA-273 or ATC-40 guidelines). Data obtained indicate that the
first-mode SDF system overestimates the median roof displacement for systems
subjected to large ductility demand , but underestimates for small . The bias and
dispersion tend to increase for longer period system for every value of . The modal
pushover analysis procedure has the advantage of reducing the dispersion in the roof
displacement and the underestimation of the median roof displacement for elastic or
ii
nearly elastic cases at the expense of increasing slightly the overestimate of roof
displacement of buildings responding far into the inelastic range.
Comparing the median values of story-drift demands determined by MPA to those

obtained from nonlinear RHA shows that the MPA predicts reasonably well the changing
height-wise variation of demand with building height and SDF-system ductility factor.
Median and dispersion values of the ratios of story-drift demands determined by MPA
and nonlinear-RHA procedures were computed to measure the bias and dispersion of
MPA estimates leading to the following results: (1) the bias and dispersion in the MPA
procedure tend to increase for longer-period frames and larger SDF-system ductility
factors (although these trends are not perfect); (2) the bias and dispersion in MPA
estimates of seismic demands for inelastic frames are usually larger than for elastic
systems; (3) the well-known response spectrum analysis (RSA), which is equivalent to
the MPA for elastic systems, consistently underestimates the response of elastic
structures, e.g., up to 18% in the upper-story drifts of 18-story frames; (4) accuracy of
MPA does not deteriorate, in spite of irregularity in stiffness, strength, or stiffness and
strength provided the irregularity is in the middle or upper story; (5) the MPA procedure
is less accurate relative to the regular frame in estimating the seismic demands of
frames with strong or stiff-and-strong first story; soft, weak, or soft-and-weak lower half;
stiff, strong, or stiff-and-strong lower half; (6) in spite of the larger bias in estimating
drift demands for some of the stories in particular cases, the MPA procedure identifies
stories with largest drift demands and estimates them to a sufficient degree of accuracy,
detecting critical stories in such frames; and (7) the bias in the MPA procedure for frames
with soft, weak or soft-and-weak first story is about the same as for the regular frame.
This study also compares the seismic demands for vertically irregular and
regular frames determined by rigorous nonlinear RHA. The effects of vertical
irregularity on the median values of story drifts and floor displacements are documented
with the following results: (1) the three types of irregularities similarly influence the
heightwise variation of story drifts, with the effects of strength irregularity being larger
than stiffness irregularity, and the effects of combined-stiffness-and-strength irregularity
being the largest among the three; (2) introducing a soft and/or weak story increases the
story drift demands in the modified and neighboring stories and decreases the drift
iii
demands in other stories; (3) on the other hand, a stiff and/or strong story decreases the
drift demand in the modified and neighboring stories and increases the drift demands in
other stories; (4) drift demands in the upper stories are much more sensitive to
irregularities in the lower stories than the response of lower stories is affected by
irregularities in the upper stories; (5) while the roof displacement is usually insensitive to
vertical irregularity, it is significantly different for frames that are stiffness-and-strength
irregular in their lower half; and (6) irregularity in the base story or lower stories has
significant influence on the heightwise distribution of floor displacements.
Comparative evaluation of the accuracy of MPA and pushover analysis using

force distribution specified in FEMA-356 [BSSC, 2000] demonstrates that MPA is
almost always more accurate in estimating drifts in all stories of all frames than all of
FEMA distributions. The envelope of the four estimates corresponding to the four FEMA
force distributions is less biased than the individual estimates but is almost always more
biased than MPA for both regular and irregular frames.
Finally, the MPA procedure is simplified to facilitate its implementation in

engineering practicewhere the earthquake hazard is usually defined in terms of a
median (or some other percentile) design spectrum for elastic systemsand the accuracy
of this simplified procedure is documented. In the simplified procedure, the peak
deformation of an inelastic SDF system can be determined from the elastic deformation
spectrum multiplied by the ratio of peak deformations of inelastic and corresponding
linear SDF systems.
With the goal of developing empirical equations for the inelastic deformation
ratio, the median of this ratio is presented for 170 ground motions organized into nine
ensembles of ground motions, representing large or small earthquake magnitude and
distance, and NEHRP site classes B, C, and D; near-fault ground motions are also
included. Two sets of results are presented for bilinear non-degrading systems over the
complete range of elastic vibration period, Tn : C for systems with known ductility
factor, ; and CR for systems with known yield-strength reduction factor, Ry . The
influence of post-yield stiffness on the inelastic deformation ratios C and CR is
investigated comprehensively. All data is interpreted in the context of acceleration-

iv
sensitive, velocity-sensitive, and displacement-sensitive regions of the spectrum for broad
applications. The median C versus Tn and CR versus Tn plots are demonstrated to be
essentially independent of the earthquake magnitude and distance (over their ranges
considered), and of site class. In the acceleration-sensitive spectral region, the median
inelastic deformation ratio for near-fault ground motions is systematically different when
plotted against Tn ; however, when plotted against normalized period Tn / Tc (where Tc is
the period separating the acceleration- and velocity-sensitive regions) they become very
similar in all spectral regions. Determined by regression analysis of the data, two
equationsone for C and the other for CR have been developed as a function of
Tn / Tc , and or Ry , respectively, that are valid for all ground motion ensembles
considered. These equations for C and CR should be useful in estimating the inelastic
deformation of new or rehabilitated structureswhere the global ductility capacity can

be estimatedand existing structures with known lateral strength, respectively.
v
ACKNOWLEDGMENTS
This research investigation is funded by the National Science Foundation under

Grant CMS-9812531, a part of the U.S. Japan Cooperative Research in Urban Earthquake
Disaster Mitigation. The first author would also like to acknowledge the scholarship from
the Royal Thai Government. This financial support is gratefully acknowledged.
Our research has benefited from discussions with Prof. Rakesh Goel of California
Polytechnic State University at San Luis Obispo; Profs. Helmut Krawinkler, C. Allin
Cornell, and Eduardo Miranda of Stanford University; and Chris Poland, Jon Heintz, and
Kent Yu of Degenkolb Engineers. Claire Johnsons editing of parts of this report is
appreciated.
vi
TABLE OF CONTENTS
Page
Abstract...............................................................................................................................ii
Acknowledgments............................................................................................................. vi
Table of Contents.............................................................................................................vii
1. INTRODUCTION........................................................................................................1
1.1 Background and Motivations................................................................................ 1
1.2 Objectives............................................................................................................. 2
1.3 Report Organization.............................................................................................. 3
2. METHODS TO ESTIMATE SEISMIC DEMANDS OF MULTISTORY

BUILDINGS................................................................................................................. 5
2.1 Introduction........................................................................................................... 5
2.2 Elastic Multistory Buildings................................................................................. 5
2.2.1 Modal Response History Analysis (RHA) ............................................... 5
2.2.2 Modal Response Spectrum Analysis (RSA) ............................................ 8
2.2.3 Modal Pushover Analysis (MPA) ............................................................ 8
2.3 Inelastic Multistory Buildings...............................................................................9
2.3.1 Nonlinear Response History Analysis (NL-RHA)....................................9
2.3.2 Uncoupled Modal Response History Analysis (UMRHA).....................10
2.3.3 Modal Pushover Analysis (MPA)........................................................... 16
Chapter 2 Figures......................................................................................................... 18
3. STRUCTURAL SYSTEMS, GROUND MOTIONS, AND RESPONSE

STATISTICS.............................................................................................................. 23
3.1 Introduction......................................................................................................... 23
3.2 Structural Systems...............................................................................................23
3.2.1 Regular Frames....................................................................................23
3.2.2 Vertically Irregular Frames..................................................................... 25
3.3 Ground Motions.................................................................................................. 27
3.4 Response Statistics.............................................................................................. 27
Chapter 3 Tables.......................................................................................................... 30
Chapter 3 Figures......................................................................................................... 33
vii
4. ESTIMATING TARGET ROOF DISPLACEMENT............................................ 47
4.1 Introduction......................................................................................................... 47
4.2 Elastic Systems................................................................................................... 47
4.3 Inelastic Systems................................................................................................. 49
Chapter 4 Figures......................................................................................................... 53
5. EVALUATION OF MPA PROCEDURE FOR VERTICALLY REGULAR

FRAMES..................................................................................................................... 68
5.1 Introduction......................................................................................................... 68
5.2 Nonlinear Response History Analysis Results....................................................68
5.3 Comparison of MPA and Nonlinear RHA Results............................................. 69
5.4 Bias and Dispersion of MPA Procedure..............................................................69
5.5 Comparative Evaluation of MPA and RSA Procedures......................................73
Chapter 5 Figures......................................................................................................... 75
6. EVALUATION OF MPA PROCEDURE FOR VERTICALLY IRREGULAR

FRAMES..................................................................................................................... 96
6.1 Introduction......................................................................................................... 96
6.2 Effect of Irregularity on Story-Drift Demands................................................... 97
6.3 Effect of Irregularity on Floor Displacements.................................................... 99
6.4 Bias and Dispersion of MPA Procedure............................................................. 99
Chapter 6 Figures....................................................................................................... 104
7. COMPARATIVE EVALUATION OF MPA AND PUSHOVER ANALYSIS

USING FEMA FORCE DISTRIBUTIONS.......................................................... 121
7.1 Introduction....................................................................................................... 121
7.2 FEMA force distributions..................................................................................121
7.3 Regular Frames..............................................................................................122
7.4 Irregular Frames................................................................................................ 124
Chapter 7 Figures....................................................................................................... 126
8. PRACTICAL IMPLEMENTATION OF MPA.................................................... 133

8.1 Introduction....................................................................................................... 133
8.2 MPA Procedure B............................................................................................. 133
8.3 MPA Procedure C............................................................................................. 135
viii
8.4 Comparative Evaluation of Three Versions of the MPA Procedure.................136
8.5 Estimation of SDF-system Deformation...........................................................137
Chapter 8 Figures....................................................................................................... 139
9. INELASTIC DEFORMATION RATIOS FOR SDF BILINEAR SYSTEMS... 146

9.1 Introduction....................................................................................................... 146
9.2 Inelastic Deformation Ratio: Theory................................................................ 148
9.2.1 Bilinear Systems................................................................................... 148
9.2.2 Computing Inelastic Deformation Ratio............................................... 149
9.2.3 Limiting Values of Inelastic Deformation Ratio.................................. 150
9.3 Ground Motions and Elastic Response Spectra................................................ 152
9.4 Deformation of Systems with Known Ductility............................................... 153
9.5 Deformation of Systems with Known Ry .........................................................155
9.6 Influence of Earthquake Magnitude and Distance............................................ 157

9.7 Influence of Firm Site Classes.......................................................................... 157
9.8 Near-fault Ground Motions...............................................................................157
9.9 Estimating Deformation of Inelastic Systems...................................................158
9.9.1 Equation for C .................................................................................... 158
9.9.2 Equation for CR .................................................................................... 161
Chapter 9 Tables........................................................................................................ 164
Chapter 9 Figures....................................................................................................... 173
10. CONCLUSIONS...................................................................................................... 194

References....................................................................................................................... 202
Appendix A Bilinear Idealization............................................................................207
Appendix B Difficulty in MPA................................................................................ 209
Appendix C Regression Equations for C and CR ............................................... 211
ix
1. INTRODUCTION
1.1 BACKGROUND AND MOTIVATIONS
Estimating seismic demands at low performance levels, such as life safety and collapse
prevention, requires explicit consideration of inelastic behavior of the structure. While
nonlinear response history analysis (RHA) is the most rigorous procedure to compute
seismic demands, current structural engineering practice uses nonlinear static procedure
(NSP), or pushover analysis, described in FEMA-273 [Building Seismic Safety Council,
1997], or ATC-40 [1996], and FEMA-356 [BSSC, 2000]. The seismic demands are
computed by nonlinear static analysis of the structure subjected to monotonically
increasing lateral forces with an invariant height-wise distribution until a predetermined
target displacement is reached. The target roof displacement is determined from the
deformation of an equivalent single-degree-of-freedom (SDF) system. A formulation of
the pushover analysis can be found in [Krawinkler and Seneviratna, 1998].
This pushover analysis procedure is obviously based on two major assumptions:

(1) the response of the multi-degree-of-freedom (MDF) structure can be related to the
response of an equivalent SDF system, implying that the response is controlled by a
single mode and this mode shape remains unchanged even after yielding occurs, and (2)
the invariant lateral force distribution can represent and bound the distribution of inertia
forces during an earthquake. These assumptions would only be approximate after
structure yields, but several research investigations [Saiidi and Sozen, 1981; Miranda,
1991; Lawson et al., 1994; Fajfar and Fischinger, 1988; Krawinkler and Seneviratna,
1998; Kim and DAmore, 1999; Maison and Bonowitz, 1999; Gupta and Krawinkler,
1999, 2000; Skokan and Hart, 2000] have led to good estimates of seismic demands.
However, such satisfactory predictions of seismic demands are mostly restricted to low-
and medium-rise structures in which inelastic action is distributed throughout the height
of the structure [Krawinkler and Seneviratna, 1998; Gupta and Krawinkler, 1999].
None of the invariant force distributions can account for the contributions of
higher modes to response, or for a redistribution of inertia forces due to structural
yielding and the associated changes in the vibration properties of the structure. To
overcome these limitations, several researchers have proposed adaptive force
1
distributions that attempt to follow more closely the time-variant distributions of inertia
forces [Fajfar and Fischinger, 1988; Bracci et al., 1997; Gupta and Kunnath, 2000].
While these adaptive force distributions may provide better estimates of seismic demands
[Gupta and Kunnath, 2000], they are conceptually complicated and computationally
demanding for routine application in structural engineering practice. Attempts have also
been made to consider more than the fundamental vibration mode in pushover analysis
[Paret et al., 1996; Sasaki et al., 1998; Gupta and Kunnath, 2000; Kunnath and Gupta,
2000; Matsumori et al., 2000].
A modal pushover analysis (MPA) procedure that includes the contributions of

several modes of vibration, thus providing superior accuracy in estimating inelastic
seismic demands on buildings, while retaining the conceptual simplicity and
computational attractiveness of the procedure with invariant force distribution, has
recently been developed [Chopra and Goel, 2001, 2002]. Although it is somewhat
intuitive for inelastic buildings and has been tested for only a few buildings [Goel and
Chopra, 2002], it seems reasonable because the lateral force distribution for each mode
appears to be the most rational choice among all invariant distributions of forces and it
provides results for elastic buildings that are identical to the well-known response
spectrum analysis (RSA) procedure. Being an approximate method, however, it should
obviously be evaluated comprehensively before practical application to building
evaluation and design.
1.2 OBJECTIVES
The objectives of this report are to
(1) Investigate the basic premise that the roof displacement of a multistory building can
be determined from the deformation of an equivalent inelastic SDF system
(2) Identify and develop more understanding of the underlying assumptions and
approximations in MPA
(3) Evaluate the accuracy of the MPA procedure in estimating the seismic demands on
moment-resisting-frame (MRF) buildings for a wide range of building heights (and
2
periods), design ductility factors, and vertical variation of story stiffness and
strength
(4) Evaluate the MPA procedure and NSP, or pushover analysis, using invariant lateral
force distributions specified in FEMA-356 on a comparative basis
(5) Present simplified versions of the MPA procedure for practical application to
building evaluation and design with the seismic hazard defined in terms of median
or smooth response spectrum
(6) Develop a convenient equation for the ratio of deformations of bilinear and
corresponding linear SDF systems to facilitate the determination of target roof
displacement
1.3 REPORT ORGANIZATION
First, methods to estimate seismic demands on elastic and inelastic multistory buildings
are reviewed in Chapter 2, where the derivation and underlying assumptions of MPA
procedure are also presented. Chapter 3 describes structural systems analyzed, the
ensemble of ground motions considered, and response statistics presented in this report;
bias and dispersion of approximate procedure are defined here. The accuracy of the
target roof displacement estimated by an equivalent SDF system is examined in terms of
bias and dispersion, and the assumptions underlying this approximate method are studied
in Chapter 4. Chapter 5 evaluates the accuracy of the MPA procedure by examining its
bias and dispersion in estimating seismic demands on sixty vertically regular frames
covering a wide range of building heights (and periods) and design ductility factors
designed according to lateral force distribution in the International Building Code [ICC,
2000]. The accuracy of MPA procedure for vertically-irregular frames is evaluated in
Chapter 6; also identified are the irregular frames where MPA does not work well.
Chapter 7 presents a comparative evaluation of MPA and pushover analysis using FEMA
force distributions in estimating seismic demands for the vertically regular and
irregular frames. The MPA procedure is simplified in Chapter 8 for practical application
where seismic hazard is defined in terms of a response spectrum and the bias in the MPA
procedure with additional simplifications is documented. Finally, the seismic
3
deformations of bilinear SDF systems are comprehensively evaluated in Chapter 9 and an
equation for the ratio of deformations of bilinear and corresponding linear SDF systems
is developed to facilitate the determination of target roof displacement.
4
2. METHODS TO ESTIMATE SEISMIC DEMANDS OF
MULTISTORY BUILDINGS
2.1 INTRODUCTION
Conventional dynamic analysis and modal pushover analysis (MPA) procedures to

determine seismic demands for elastic and inelastic buildings are presented in this
chapter. First, two versions of modal analysis, response history analysis (RHA) and
response spectrum analysis (RSA), for linearly elastic systems are reviewed. Then,
standard equations of motion for inelastic MDF systems are expressed in terms of elastic
modal coordinates. Although, these modal equations are not uncoupled in contrast to
elastic systems, their coupling is shown to be weak and thus neglected to develop the
uncoupled modal response history analysis (UMRHA) procedure. The peak modal
responses, which can be determined by a pushover analysis for each mode, are then
combined according to an appropriate modal combination rule.
2.2 ELASTIC MULTISTORY BUILDINGS
2.2.1 Modal Response History Analysis (RHA)
The differential equations governing the response of a multistory building to horizontal

earthquake ground motion u g ( t ) are as follows:
mu + cu + ku = m u g ( t ) (2.1)
where u is the vector of N lateral floor displacements relative to the ground, m, c, and k
are the mass, classical damping, and lateral stiffness matrices of the system; each element
of the influence vector is equal to unity.
The right side of Eq. (2.1) can be interpreted as effective earthquake forces:
p eff ( t ) = m u g ( t ) (2.2)
The spatial distribution of these forces over the height of the building is defined by the
vector s = m and their time variation by u g ( t ) . This force distribution can be expanded
as a summation of modal inertia force distributions s n [Chopra, 2001: Section 13.12]:
5
N N
m = s n = n mn (2.3)
n =1 n =1
where n is the nth natural vibration mode of the structure (see Figs. 2.1 and 2.2), and
Ln
n = Ln = nT m M n = nT mn (2.4)
Mn
The effective earthquake forces can then be expressed as

N N
p eff ( t ) = p eff ,n ( t ) = s nu g ( t ) (2.5)
n =1 n =1
The contribution of the nth mode to s and to p eff ( t ) is:
s n = nmn p eff ,n ( t ) = s n u g ( t ) (2.6)
The response of the MDF system to p eff,n ( t ) is entirely in the nth-mode, with no
contributions from other modes. The equations governing the response of the system are
mu + cu + ku = s n u g ( t ) (2.7)
By utilizing the orthogonality property of modes, it can be demonstrated that none of the
modes other than the nth mode contribute to the response. Then the floor displacements
are
u n ( t ) = n qn ( t ) (2.8)
where the modal coordinate qn ( t ) is governed by
qn + 2 nn qn + n2 qn = n u g ( t ) (2.9)
in which n is the natural vibration frequency and n is the damping ratio for the nth
mode. The solution qn ( t ) can readily be obtained by comparing Eq. (2.9) to the equation
of motion for the nth-mode elastic SDF system, an SDF system with vibration
propertiesnatural frequency n and damping ratio n of the nth-mode of the MDF
system, subjected to u g ( t ) :
6
Dn + 2 nn Dn + n2 Dn = u g ( t ) (2.10)
Comparing Eqs. (2.9) and (2.10) gives
qn ( t ) = n Dn ( t ) (2.11)
and substituting in Eq. (2.8) gives the floor displacements
u n ( t ) = n n Dn ( t ) (2.12)
Any response quantity r ( t ) story drifts, internal element forces, etc.can be
expressed by
rn ( t ) = rnst An ( t ) (2.13)
where rnst denotes the modal static response, the static value of r due to external forces
s n , and
An ( t ) = n2 Dn ( t ) (2.14)
is the pseudo-acceleration response of the nth-mode SDF system [Chopra, 2001; Section
12.1]. The two analyses that lead to rnst and An ( t ) are shown schematically in Fig. 2.3.
Equations (2.12) and (2.13) represent the response of the MDF system to
p eff, n ( t ) . Therefore, the response of the system to the total excitation p eff ( t ) is
N N
u ( t ) = u n ( t ) = nn Dn ( t ) (2.15)
n =1 n =1
N N
r ( t ) = rn ( t ) = rnst An ( t ) (2.16)
n =1 n =1
This is the classical modal RHA procedure wherein Eq. (2.9) is the standard modal
equation governing qn ( t ) , Eqs. (2.12) and (2.13) define the contribution of the nth-mode
to the response, and Eqs. (2.15) and (2.16) reflect combining the response contributions
of all modes. However, these standard equations have been derived in an unconventional
way. In contrast to the classical derivation found in textbooks [e.g., Chopra, 2001:
7
Sections 12.4 and 13.1.3], we used the modal expansion of the spatial distribution of the
effective earthquake forces. This concept provides a rational basis for the modal pushover
analysis procedure presented later.
2.2.2 Modal Response Spectrum Analysis (RSA)
The peak value ro of the total response r ( t ) can be estimated directly from the response
spectrum for the ground motion without carrying out the response history analysis (RHA)
implied in Eqs. (2.9)-(2.16). In such a response spectrum analysis (RSA), the peak value
rno of the nth-mode contribution rn ( t ) to response r ( t ) is determined from
rno = rnst An (2.17)
where An is the ordinate A (Tn , n ) of the pseudo-acceleration response (or design)
spectrum for the nth-mode SDF system, and Tn = 2 n is the natural vibration period of
the nth-mode of the MDF system.
The peak modal responses are combined according to the Square-Root-of-Sum-

of-Squares (SRSS) or the Complete Quadratic Combination (CQC) rules. The SRSS rule,
which is valid for structures with well-separated natural frequencies such as multistory
buildings with symmetric plan, provides an estimate of the peak value of the total
response:
1/ 2
N
ro rno2 (2.18)
n =1
2.2.3 Modal Pushover Analysis (MPA)
To develop a pushover analysis procedure consistent with RSA, we note that static
analysis of the structure subjected to lateral forces
f no = nmn An (2.19)
will provide the same value of rno , the peak nth-mode response as in Eq. (2.17) [Chopra,
2001: Section 13.8.1]. Alternatively, this response value can be obtained by static
8
analysis of the structure subjected to lateral forces distributed over the building height
according to
s*n = mn (2.20)
and the structure is pushed to the roof displacement, urno , the peak value of the roof
displacement due to the nth-mode, which from Eq. (2.12) is
urno = nrn Dn (2.21)
where Dn = An n2 . Obviously Dn and An are available from the response (or design)
spectrum.
The peak modal responses, rno , each determined by one pushover analysis, can be
combined according to Eq. (2.18) to obtain an estimate of the peak value ro of the total
response. This modal pushover analysis (MPA) for linearly elastic systems is equivalent
to the well-known RSA procedure (Section 2.2.2).
2.3 INELASTIC MULTISTORY BUILDINGS
2.3.1 Nonlinear Response History Analysis (NL-RHA)
For each structural element of a building, the initial loading curve is idealized as bilinear,
and the unloading and reloading curves differ from the initial loading branch. Thus, the
relations between lateral forces f s at the N floor levels and the lateral displacements u are
not single valued, but depend on the history of the displacements:
f s = f s ( u, sign u ) (2.22)
With this generalization for inelastic systems, Eq. (2.1) becomes
mu + cu + f s ( u,sign u ) = m u g ( t ) (2.23)
The standard approach is to solve directly these coupled equations, leading to the exact
nonlinear response history analysis (RHA).
Although classical modal analysis (Section 2.2.1) is not valid for inelastic
systems, it is useful for later reference to transform Eq. (2.23) to the modal coordinates of
9
the corresponding linear system. Each structural element of this elastic system is defined
to have the same stiffness as the initial stiffness of the structural element of the inelastic
system. Both systems have the same mass and damping. Therefore, the natural vibration
periods and modes of the corresponding linear system are the same as the vibration
properties of the inelastic system undergoing small oscillations (within the linear range).
Expanding the displacements of the inelastic system in terms of the natural

vibration modes of the corresponding linear system we get
N
u ( t ) = n qn ( t ) (2.24)
n =1
Substituting Eq. (2.24) in Eq. (2.23), premultiplying by Tn , and using the mass- and
classical damping-orthogonality property of modes gives
Fsn
qn + 2 nn qn + = nu g ( t ) n = 1, 2, N (2.25)
Mn
where the only term that differs from Eq. (2.9) involves
Fsn = Fsn ( q,sign q ) = Tn f s ( u,sign u ) (2.26)
This resisting force depends on all modal coordinates qn ( t ) , contained in q , implying
coupling of modal coordinates because of yielding of the structure.
Equation (2.25) represents N equations in the modal coordinates qn . Unlike Eq.

(2.9) for linearly elastic systems, these equations are coupled for inelastic systems.
Simultaneously solving these coupled equations and using Eq. (2.24) will, in principle,
give the same results for u ( t ) as obtained directly from Eq. (2.23). However, Eq. (2.25)
is rarely solved because it offers no particular advantage over Eq. (2.23).
2.3.2 Uncoupled Modal Response History Analysis (UMRHA)
Neglecting the coupling of the N equations in modal coordinates [Eq. (2.25)] leads to the
uncoupled modal response history analysis (UMRHA) procedure. This approximate RHA
procedure is the preliminary step in developing a modal pushover analysis procedure for
inelastic systems.
10
The spatial distribution s of the effective earthquake forces is expanded into the
modal contributions s n according to Eq. (2.3), where n are now the modes of the
corresponding linear system. The equations governing the response of the inelastic
system to p eff,n ( t ) given by Eq. (2.6b) are
mu + cu + f s ( u,sign u ) = s nu g ( t ) (2.27)
The solution of Eq. (2.27) for inelastic systems will no longer be described by Eq. (2.8)
because qr ( t ) will generally be nonzero for modes other than the nth mode,
implying that other modes will also contribute to the solution. For linear elastic
systems, however, qr ( t ) = 0 for all modes other than the nth-mode; therefore, it is
reasonable to expect that the nth mode should be dominant even for inelastic systems.
These assertions are illustrated numerically in Figs. 2.4a and 2.4b for elastic and
inelastic, respectively, 6-story frames with T1 = TL , which will be described in detail later
in Section 3.2.1. Equation (2.27) was solved by nonlinear RHA, and the resulting roof
displacement history was decomposed into its modal components. The modal
decomposition of the roof displacement for the first three modes due to an earthquake
ground motion (LMSR Record No.5, described later in Section 3.3) demonstrates that
bg
because the elastic frame does not yield, the response to excitation p eff,n t is all in the
nth-mode (Fig. 2.4a). On the other hand, the inelastic 6-story frame, which is designed
for SDF-system ductility factor=6, yields when subjected to the same ground motion, and
the modes other than the nth-mode contribute to the response. The second and third
bg
modes start responding to excitation p eff,1 t at about 4.9 sec, the instant the structure
first yields; however, their contributions to the roof displacement are only 3% and 0.3%,
respectively, of the first mode response (Fig. 2.4b). The first and third modes start
bg
responding to excitation p eff,2 t at about 4.8 sec, the instant the structure first yields;
however, their contributions to the roof displacement are 41% and 13%, respectively, of
the second mode response (Fig. 2.4b).
11
Approximating the response of the structure to excitation p eff,n ( t ) by Eq. (2.8),
substituting Eq. (2.8) in Eq. (2.27) and premultiplying by Tn gives Eq. (2.25), except for
the important approximation that Fsn now depends only on one modal coordinate, qn :
Fsn = Fsn ( qn ,sign qn ) = Tn f s ( qn ,sign qn ) (2.28)
With this approximation, solution of Eq. (2.25) can be expressed by Eq. (2.11) where
Dn ( t ) is governed by
Fsn
Dn + 2 nn Dn + = u g ( t ) (2.29)
Ln
and
Fsn = Fsn ( Dn ,sign Dn ) = Tn f s ( Dn ,sign Dn ) (2.30)
is related to Fsn ( qn ,sign qn ) because of Eq. (2.11).
Equation (2.29) may be interpreted as the governing equation for the nth-mode
inelastic SDF system, an SDF system with (1) small amplitude vibration properties
natural frequency n and damping ratio n of the nth-mode of the corresponding linear
MDF system; (2) unit mass; and (3) Fsn Ln Dn relation between resisting force Fsn Ln
and modal coordinate Dn defined by Eq. (2.30). Although Eq. (2.25) can be solved in its
original form, Eq. (2.29) can be solved conveniently by standard software because it is of
the same form as the SDF system excited by ground acceleration u g ( t ) , and the peak
value of Dn ( t ) can be estimated from the inelastic response (or design) spectrum
[Chopra, 2001: Sections 7.6 and 7.12.1]. Introducing the nth-mode inelastic SDF
system also permitted extension of the well-established concepts for elastic systems to
inelastic systems. Compare Eqs. (2.25) and (2.29) to Eqs. (2.9) and (2.10); note that Eq.
(2.11) applies to both systems.
Solution of the nonlinear Eq. (2.29) formulated in this manner provides Dn ( t ) ,
which substituted into Eq. (2.12) gives the floor displacements of the structure associated
with the nth-mode inelastic SDF system. Any floor displacement, story drift, or another
12
deformation response quantity r ( t ) is given by Eqs. (2.13) and (2.14), where An ( t ) is
now the pseudo-acceleration response of the nth-mode inelastic SDF system. The two
analyses that lead to rnst and An ( t ) are shown schematically in Fig. 2.5. Equations (2.13)
and (2.14) represent the response of the inelastic MDF system to p eff, n ( t ) , the nth-mode
contribution to p eff ( t ) . Therefore the response of the system to the total excitation
p eff ( t ) is given by Eqs. (2.15) and (2.16). This is the UMRHA procedure.
Properties of the nth-mode Inelastic SDF System
To determine the Fsn Ln Dn relation in Eq. (2.29), the relationship between lateral
forces f s and Dn in Eq. (2.30) should be determined by nonlinear static analysis of the
structure as the structure undergoes displacements u = Dn n with increasing Dn .

However, most commercially available software can not implement such displacement-
controlled analysis. An alternative approach, which is an approximation, is to conduct a
force-controlled nonlinear static analysis of the structure subjected to lateral forces
distributed over the building height according to Eq. (2.20). When implemented by
commercially available software, such nonlinear static analysis provides the so-called
pushover curve, which is a plot of base shear Vbn against roof displacement urn . A
bilinear idealization of this pushover curve for the nth-mode is shown in Fig. 2.6a. At
the yield point, the base shear is Vbny and roof displacement is urny .
To convert this Vbn urn pushover curve to the Fsn Ln Dn relation, the two sets
of forces and displacements are related as follows:
Vbn urn
Fsn = Dn = (2.31)
n nrn
Equation (2.31) enables conversion of the pushover curve to the desired Fsn Ln Dn
relation shown in Fig. 2.6b, where the yield values of Fsn Ln and Dn are
13
Fsny Vbny urny
= Dny = (2.32)
Ln M n* nrn
in which M n* = Ln n is the effective modal mass [Chopra, 2001: Section 13.2.5]. The two
are related through
Fsny
= n2 Dny (2.33)
Ln
implying that the initial slope of the curve in Fig. 2.6b is n2 . Knowing Fsny Ln and Dny
from Eq. (2.32), the elastic vibration period Tn of the nth-mode SDF system is computed
from
1/ 2
Ln Dny
Tn = 2 (2.34)
F
sny
This value of Tn , which may differ from the period of the corresponding linear system,
should be used in Eq. (2.29). In contrast, the initial slope of the pushover curve in Fig.
2.6a is kn = n2 Ln , which is not a meaningful quantity. Properties of the first three
modes of inelastic SDF systems determined from modal pushover curves for the 6-
story frame example are shown in Fig. 2.7.
Underlying Assumptions and Accuracy
The approximate solution of Eq. (2.27) by UMRHA is compared with the exact
solution by nonlinear RHA, both for the inelastic 6-story frame with SDF-system
ductility factor of 6 subjected to LMSR record no. 5; this large ductility factor was
chosen to ensure that the structure deforms well beyond its linear elastic limit. Such
comparison for roof displacement and top-story drift is presented in Figs. 2.8a and 2.8b,
respectively. The errors are slightly larger in drift than in displacement, but even for this
level of inelastic action, the errors in either response quantity are only a few percent.
These errors arise from the following assumptions and approximations: (1) the
coupling among modal coordinates qn ( t ) arising from yielding of the system [recall Eqs.
14
(2.25) and (2.26)] is neglected; (2) the superposition of responses to p eff,n t bg
(n = 1,2 N ) according to Eq. (2.15) is strictly valid only for linearly elastic systems; and
(3) the Fsn Ln Dn relation is approximated by a bilinear curve to facilitate solution of
Eq. (2.29) in UMRHA. Although several approximations are inherent in this UMRHA
procedure, when specialized for linearly elastic systems it is identical to the RHA
procedure of Section 2.2.1. The overall errors in the UMRHA procedure for roof
displacement are documented in the examples presented in Section 4.3.
Step-by-step UMRHA Procedure
The inelastic response of an N-story building with plan symmetric about two orthogonal
axes to earthquake ground motion along an axis of symmetry can be estimated as a
function of time by the UMRHA procedure just developed, which is summarized next as
a sequence of steps; details are available in Chopra and Goel [2001: Appendix A]:
5. Compute the natural frequencies, n , and modes, n , for linearly-elastic vibration of
the building.
6. For the nth-mode, develop the base-shear roof-displacement ( Vbn urn ) pushover
curve for the force distribution s*n [Eq. (2.20)].
7. Idealize the pushover curve as a bilinear curve with post-yield stiffness ratio n (Fig.
2.6a). Details of the implementation in this study can be found in Appendix A.
8. Convert the idealized pushover curve to the Fsn Ln Dn relation (Fig. 2.6b) by
utilizing Eq. (2.32).
9. Compute the deformation history, Dn (t ) , and pseudo-acceleration history, An ( t ) , of
the nth-mode inelastic SDF system (Fig. 2.5b) with force-deformation relation of
Fig. 2.6b.
10. Calculate histories of various responses by Eqs. (2.12) and (2.13).
11. Repeat Steps 2 to 6 for as many modes as required for sufficient accuracy. Typically,
the first two or three modes will suffice.
15
12. Combine the modal responses using Eqs. (2.15) and (2.16) to determine the total
response.
13. Calculate the peak value, r o , of the total response r ( t ) obtained in Step 8.
2.3.3 Modal Pushover Analysis (MPA)
A pushover analysis procedure is presented next to estimate the peak response rno of the
inelastic MDF system to effective earthquake forces p eff ,n ( t ) . Consider a nonlinear static
analysis of the structure subjected to lateral forces distributed over the building height
according to s*n [Eq. (2.20)], with the structure is pushed to the roof displacement urno .
This value of the roof displacement is given by Eq. (2.21) where Dn , the peak value of
Dn ( t ) , is now determined by solving Eq. (2.29), as described in Section 2.3.2;
alternatively, it can be determined from the inelastic response (or design) spectrum
[Chopra, 2001: Sections 7.6 and 7.12]. At this roof displacement, the pushover analysis
provides an estimate of the peak value rno of any response rn ( t ) : floor displacements,
story drifts, joint rotations, plastic hinge rotations, etc.
This pushover analysis, although somewhat intuitive for inelastic buildings, seems
reasonable. It provides results for elastic buildings that are identical to the well-known
RSA procedure (Section 2.2.2) because, as mentioned earlier, the lateral force
distribution used possesses two properties: (1) it appears to be the most rational choice
among all invariant distribution of forces; and (2) it provides the exact modal response
for elastic systems.
The response value rno is an estimate of the peak value of the response of the
inelastic system to p eff,n ( t ) , governed by Eq. (2.27). As shown in Sections 2.2.2 and
2.2.3, for elastic systems, rno also represents the exact peak value of the nth-mode
contribution rn ( t ) to response r ( t ) . Thus, we will refer to rno as the peak modal
response even in the case of inelastic systems.
16
The peak modal responses rno , each determined by one pushover analysis, are
combined using an appropriate modal combination rule, e.g. Eq. (2.18), to obtain an
estimate of the peak value ro of the total response. This application of modal combination
rules to inelastic systems obviously lacks a theoretical basis. However, it seems
reasonable because it provides results for elastic buildings that are identical to the well-
known RSA procedure described in Section 2.2.2.
Step-by-step MPA Procedure
The peak inelastic response of a building to earthquake excitation can be estimated by the
MPA procedure just developed, which is summarized next as a sequence of steps; details
are available in Chopra and Goel [2001: Appendix B].
Steps 1 to 4 of the MPA are same as those for UMRHA.
5. Compute the peak deformation, Dn , of the nth-mode inelastic SDF system (Fig.
2.5b) with force-deformation relation of Fig. 2.6b by solving Eq. (2.29), or from the
inelastic response (or design) spectrum.
6. Calculate the peak roof displacement urno associated with the nth-mode inelastic
SDF system from Eq. (2.21).
7. At urno , extract from the pushover database values of other desired responses, rno .
8. Repeat Steps 3 to 8 for as many modes as required for sufficient accuracy.

Typically, the first two or three modes will suffice.
9. Determine the total response by combining the peak modal responses using the
SRSS combination rule of Eq. (2.18). From the total rotation of a plastic hinge,
subtract the yield value of hinge rotation to determine the hinge plastic rotation.
We will refer to this MPA procedure as Procedure A. Procedures B and C, which

eliminate the need to perform response history analysis of inelastic SDF systems in Step
6 by obtaining the peak deformations from a response spectrum and equation for inelastic
deformation ratios, will be introduced later in Chapter 8.
17
1 2 3
6 6 6
5 5 5
Story 4 4 4
3 3 3
2 2 2
1 1 1
0 0 0
1 0 1 1 0 1 1 0 1
Figure 2.1 The first three vibration modes of a 6-story frame used in this study.
s s1 s2 s3
6
4
Story
3 = + + + ...
0
m 0 m m 0 m m 0 m m 0 m
Figure 2.2 Modal expansion of the effective earthquake force for a 6-story frame (m =
story mass).
Forces
sn
An(t )
n, n
rnst
ug(t )
(a) Static Analysis of (b) Dynamic Analysis of

Structure SDF System
Figure 2.3 Conceptual explanation of modal response history analysis of elastic MDF
systems. (Source: Chopra and Goel, 2001)
18
(a) Elastic
p (t) = s (t) p (t) = s (t)
eff,1 1 g eff,2 2 g
5 1
u (cm) Decomposed Mode 1 Decomposed Mode 1
ur1 (cm)
0 0 0.00
r1
5 4.63 1
5 1
Decomposed Mode 2 Decomposed Mode 2
u (cm)
ur2 (cm)
0 0.00 0
r2
0.63
5 1
5 1
u (cm)
ur3 (cm)
0 0.00 0 0.00
r3
5 1
0 10 20 30 0 10 20 30
Time (sec) Time (sec)
(b) Inelastic
peff,1(t) = s1g(t) peff,2(t) = s2g(t)
5 1
3.14 Decomposed Mode 1 Decomposed Mode 1
u (cm)
ur1 (cm)
0.28
0 0
r1
5 1
5 1 0.68
u (cm)
ur2 (cm)
0 0
0.10
r2
5 1
5 1
u (cm)
ur3 (cm)
0.01 0.09
0 0
r3
5 1
0 10 20 30 0 10 20 30
Figure 2.4 Modal decomposition of the roof displacement of (a) elastic and (b) inelastic 6-
story frames (with T1 = TL ) due to p eff,1 (t ) = s1u g (t ) and p eff,2 (t ) = s 2u g (t ) ,
where u g (t ) is the ground acceleration record no. 5 in LMSR ensemble.
19
Forces
sn
Unit mass
An(t )
n, n, Fsn / Ln
rnst
ug(t )
(a) Static Analysis of (b) Dynamic Analysis of

Structure Inelastic SDF System
Figure 2.5 Conceptual explanation of uncoupled modal response history analysis of

inelastic MDF systems. (Source: Chopra and Goel, 2001)
Vbn (a) Idealized Pushover Curve Fsn / Ln (b) Fsn / Ln Dn Relationship
Idealized
Vbny nkn Vbny / Mn
*
2
1 n n
1
Actual
kn 2
n
1 1
ur n Dn
ur n y Dny = ur n y / n r n
Figure 2.6 Properties of the nth-mode inelastic SDF system from the pushover curve.
(Source: Chopra and Goel, 2001)
20
Mode 1
70
60
Base shear, V (kips)

50
Vb1y=59.75 ur1y=0.606
b1
40
30 Fs1y/L1=23.64 D1y=0.440
20
=3.75%
1
10
0
0 0.5 1 1.5 2 2.5
Mode 2
40
30
b2
20
Vb2y=29.40 ur2y=0.114
Fs2y/L2=82.14 D2y=0.217
10
=21.48%
2
0
0 0.05 0.1 0.15 0.2 0.25
Mode 3
30
25
20
b3
15
V =28.22 ur3y=0.036
b3y
10 Fs3y/L3=227.04 D3y=0.181
5 =23.08%
3
0
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045
Roof displacement, urn (in)
Figure 2.7 Modal pushover curves and properties of the first three modes inelastic SDF
systems determined from the modal pushover curves of the 6-story frame
example.
21
(a) Roof displacement
(i) nonlinear RHA (ii) UMRHA
5 5
3.06 n=1 3.06 n=1
ur1 (cm)
ur1 (cm)
0 0
5 5
5 5
n=2 n=2
ur2 (cm)
ur2 (cm)
0.836 0.753
0 0
5 5
5 5
n=3 n=3
ur3 (cm)
ur3 (cm)
0.110
0 0
0.109
5 5
0 5 10 15 20 25 30 0 5 10 15 20 25 30
(b) Topstory drift

(i) nonlinear RHA (ii) UMRHA
0.3 0.3
n=1 0.134
n=1
0.120
(%)
(%)
0 0
r1
r1
0.3 0.3
0.3 0.230 0.3
n=2 0.194 n=2
(%)
(%)
0 0
r2
r2
0.3 0.3
0.3 0.3
n=3 n=3
(%)
(%)
0 0
r3
r3
0.072 0.068
0.3 0.3
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure 2.8 (a) Roof displacement and (b) top-story drift of an inelastic 6-story frame
( =6 and T1 = TL ) due to p eff ,n ( t ) = s n u g (t ) , n = 1, 2, and 3, where
u g (t ) = LMSR Record No. 5, (i) exact solution by nonlinear RHA; and (ii)
approximate solution by UMRHA.
22
3. STRUCTURAL SYSTEMS, GROUND MOTIONS, AND
RESPONSE STATISTICS
3.1 INTRODUCTION
Two groups of generic one-bay planar frames, which are appropriate for modeling
buildings with plan symmetric about two orthogonal axes subjected to earthquake ground
motion along an axis of symmetry, are analyzed in this study. The first group of frames
defined as regular is designed to have gradual variation of story stiffness and story
strength according to lateral forces specified the International Building Code [ICC,
2000]. The other group of frames defined as vertically irregular is obtained by
introducing stiffness or/and strength irregularity into the regular frame. These structural
systems are described first, followed by the ensemble of 20 ground motions and statistical
analysis.
3.2 STRUCTURAL SYSTEMS
3.2.1 Regular Frames
Regular frames considered are two sets of generic one-bay frames of six different
heights: 3, 6, 9, 12, 15, and 18 stories (Fig. 3.1). Each set corresponds to the fundamental
vibration period T1 equal to
TL = 0.028H 0.8 or TU = 0.045 H 0.8 (3.1)
which define the mean-minus- and mean-plus-one-standard-deviation of measured

periods of steel moment-resisting frames [Goel and Chopra, 1997]; H is the height of the
frame in feet.
The height-wise distribution of stiffness is defined to achieve equal drifts in all

stories under the lateral forces specified in the International Building Code [ICC, 2000]:
wi hik
Fi = Vb N
(3.2)
w h
j =1
j
k
j
23
1 T1 0.5sec

where k = (T1 + 1.5 ) / 2 0.5 < T1 < 2.5sec (3.3)
2 T1 2.5sec

and Fi , wi , and hi are the lateral force, story weight, and elevation of the i th floor; Vb is
the total base shear. The mass, m , lumped at each floor is 200 kips, the story height, h ,
is 12 feet, and the beam span, L , is 24 feet. Assuming that the second moment of cross-
sectional area for each beam and its supporting columns in the story below are the same,
numerical values for the flexural rigidities of structural elements were selected such that
the fundamental vibration period T1 is equal to values given by Eq. (3.1). The
fundamental periods obtained from Eq. (3.1) are plotted versus frame height in Fig. 3.2
and all modal periods are listed in Table 3.1.
The frames are designed according to the strong-column weak-beam philosophy

(Fig. 3.3); therefore, plastic hinges form only at beam ends and the base of the first-story
columns. The columns in other stories are assumed to remain elastic. All plastic hinges
have rigid-plastic moment-rotation relationships (Fig. 3.4a) with post-yield stiffness
equal to 3% of joint-rotation elastic stiffness under symmetrical loading (Fig. 3.4b). The
yield moments of plastic hinges are selected such that yielding occurs simultaneously at
all plastic hinges under the IBC lateral force distribution [Eq. (3.2)]. The yield base shear
is selected as Vby = ( Ay g ) W , where W is the total weight of the frame, and Ay is the
median (over 20 ground motions) pseudo-acceleration for an SDF system with vibration
period Tn = T1 and a ductility factor = 1, 1.5, 2, 4, and 6 (Fig. 3.15b). Thus ten different
designs (two values of T1 and five values of ) are considered for each frame height,
leading to a total of 60 frames. Note that the frames designed based on the median
spectrum for =1 will not necessarily remain elastic when subjected to some of the 20
ground motions, so a separate set of elastic systems is also analyzed.
The Rayleigh damping matrix, c , is defined to obtain damping ratio of 5% in the

1st and j th modes of vibration, where j is selected by iteration so that the weighted
average of modal damping ratios is closest to 5%. The weighting factors used are the
modal contribution factors of base shear, widely known as modal mass participation
24
factors. The selected values are, j =2 for 3-story frame; 3 for 6-story and 9-story frames
with T1 = TU ; and 4 for 9-story frame with T1 = TL , 12-, 15-, and 18-story frames. Table
3.2 lists all modal damping ratios of all regular frames and Figure 3.5 shows the
variation of modal damping ratio against modal frequency for an example of 6-story
frame with T1 = TU .
3.2.2 Vertically Irregular Frames
Forty-eight irregular frames, all 12-stories high, were considered to account for three
types of irregularities introduced in eight different locations along the height using two
modification factors, described next. Three types of irregularities in the heightwise
distributions of frame properties were considered: stiffness irregularity (KM), strength
irregularity (SM), and combined-stiffness-and-strength irregularity (KS). Various
irregular frames are obtained by modifying the stiffness or/and strength of the reference
frame. To obtain a soft or stiff story, the story stiffness was divided or multiplied by a
modification factor; and to obtain a weak or strong story, the story strength was divided
or multiplied by a modification factor. Two values of the modification factor were
considered: MF=2 or 5. For each of the three types of irregularity, the following eight
cases were investigated: (1) soft or/and weak top story; (2) stiff or/and strong top story;
(3) soft or/and weak mid-height story; (4) stiff or/and strong mid-height story; (5) soft
or/and weak first story; (6) stiff or/and strong first story; (7) soft or/and weak lower half
of structure; and (8) stiff or/and strong lower half of structure.
Stiffness-Irregular Frames
Figure 3.6 shows the ratio of story stiffness, and of story strength of stiffness-irregular
frames to the corresponding properties of the regular frame; KMj denotes stiffness-
irregularity case j (=1, 28). A total of 16 stiffness-irregular frames are considered
corresponding to the 8 cases mentioned above and 2 values of the modification factor.
The stiffness of a story was modified by changing the stiffness of the columns in
that story and the beam they support. To ensure meaningful comparison of seismic
demand on regular and irregular frames, their fundamental vibration period, yield base
shear, and damping properties were kept the same. Modifying the stiffness of one or more
25
stories by the factor MF obviously affects the vibration period. To maintain the same
period as for the regular frame, all story stiffnesses were scaled uniformly, causing the
ratio of story stiffnesses of irregular and regular frames to be different than MF, as seen
in Fig. 3.6.
The pushover curves using the IBC force distribution show that stiffness
irregularity may influence the initial slope significantly but the yield strength only
slightly (Fig. 3.7a). All story strengths were scaled uniformly to obtain an irregular
frame with the same yield base shear as the regular frame (Fig. 3.7b). Note that the
post-yield stiffness of irregular frames can be slightly different than the regular frame.
The Rayleigh damping matrix for an irregular frame is defined to maintain the
modal damping ratio equal to 5% in the 1st and 4th modes, as for the regular frame.
Strength-Irregular Frames
Figure 3.8 shows the ratio of story stiffness and of story strength of strength-irregular
frames to the corresponding properties of the regular frame; SMj denotes strength-
irregularity case j. The story stiffnesses, fundamental period, and damping matrix of
strength-irregular frames were kept the same as for the regular frame. The strength of a
story was modified by changing only the strength of the beam at the top of the story
(recall that the columns are assumed to remain elastic). However, in cases 5 to 8, where
strength of the first story was modified (Cases 5-8), the strength of columns in the first
story, which are designed to hinge at the base, was also changed. The yield base shear of
a frame with story strengths modified as described differs from the regular frame (Fig.
3.9a). All story strengths were scaled uniformly to obtain an irregular frame with the
same yield base shear as the regular frame (Fig. 3.9b), causing the ratio of the story
strengths of irregular and regular frames to be different than the modification factor, as
in Fig. 3.8.
Stiffness-and-Strength-Irregular Frames
Figure 3.10 shows the ratio of story stiffness and of story strength of stiffness-and-
strength-irregular frames to the corresponding properties of the regular frame; KSj
denotes combined stiffness-and-strength irregularity case j. Each frame is designed by
26
modifying the story stiffnesses and damping matrix as described earlier for the stiffness-
irregular frame, and the story strengths as described earlier for the strength-irregular
frame. Pushover curves for these frames are shown in Fig. 3.11.
3.3 GROUND MOTIONS
The seismic excitation is defined by a set of 20 Large-Magnitude-Small-distance (LMSR)

records (listed in Table 3.3). These ground motions were obtained from California
earthquakes of magnitude ranging from 6.6 to 6.9 recorded at distances of 13 to 30 km on
firm soil (NEHRP site class D). The ground acceleration, velocity and displacement time
histories of the LMSR ensemble are shown in Figs. 3.12, 3.13, and 3.14, respectively.
The constant-ductility pseudo-acceleration and yield-deformation spectra for each of
these ground motions for ductility factor =1 (elastic), 1.5, 2, 4, and 6 are shown in
Figs. 3.15a and 3.16a, respectively. These inelastic spectra were developed for bilinear
SDF systems with post-yield stiffness equal to 3% of initial stiffness, chosen to be
consistent with the pushover curves due to first-mode force distribution. Note that
although this set of 20 records is binned for the given earthquake scenario, there is large
variability from record to record. The median spectra are presented in Figs. 3.15b and
3.16b together with 16th-84th percentile spectra shown by shading area between them.
3.4 RESPONSE STATISTICS
The dynamic response of each structural system to each of the 20 ground motions was
determined by the two procedures described in the previous chapter: nonlinear RHA
(Section 2.3.1) using DRAIN-2DX [Allahabadi and Powell, 1988], and MPA (Section
2.3.3). Gravity load effects were not considered because they can not be meaningfully
defined for generic one-bay frames. The exact peak value of structural response, r,
determined by nonlinear RHA is denoted by rNL-RHA , and the approximate value from
MPA by rMPA . From these data for each ground motion, a response ratio is determined:
*
rMPA = rMPA rNL-RHA . An approximate method is invariably biased in the sense that the
median of the response ratio differs from one, underestimates the median response if the
ratio is less than one, and provides an overestimate if the ratio exceeds one.
27
Presented in this paper are the median values x , defined as the geometric mean,
of n ( = 20 ) observed values ( xi ) of rMPA , rNL-RHA , and rMPA
*
; and the dispersion measure
of rMPA
*
defined as the standard deviation of logarithm of the n observed values:
1/ 2
n n 2
ln xi ( ln xi ln x )
x = exp i =1 ; = i =1 (3.4)
n n 1

For small values, e.g., 0.3 or less, the above dispersion measure is close to the coefficient
of variation. This measure will be referred to as dispersion in subsequent sections.
Equations (3.4a) and (3.4b) are logical estimators for the median and dispersion,
especially if the data are sampled from lognormal distribution [Benjamin and Cornell,
1970], which is known to be appropriate for peak earthquake response of structures.
An advantage of using the geometric mean as the estimator of median is that the
ratio of the median of rMPA to the median of rNL-RHA is equal to the median of the ratio
*
rMPA , i.e., the bias of MPA in estimating the median responses is equal to the median of
bias in estimating response to individual excitation.
When presenting such response data for inelastic response of the selected systems,
their response assuming elastic behavior is also considered. In this case the nonlinear
RHA procedure specializes to linear RHA and the MPA procedure reduces to standard
response spectrum analysis (RSA); thus, the latter response ratio is written as:
*
rRSA = rRSA rRHA .
Number of modes considered in MPA and RSA
Unless specified otherwise, sufficient numbers of modes were included in MPA and
RSA to ensure that the sum of the modal contribution factors [Chopra 2001, Section
12.10] for the top-story drift exceed 95%: 2 for 3-story frame, 3 for 6-story frame, 4 for
9- and 12-story frames, and 5 for 15- and 18-story frames.
28
SDF-system estimate
Including only the first mode contribution in the MPA and RSA procedures, i.e.
retaining only the first term in Eq. (2.18), provides the SDF-system estimate of response.
The peak value of structural response is denoted by rSDF , and the response ratio is written
* *
as: rSDF = rSDF rNL-RHA for inelastic frames, or rSDF = rSDF rRHA for elastic frames.
29
Table 3.1 Modal vibration period, T n , of "regular" frames
(a) T 1=T L (b) T 1=T U

Modal vibration period, T n (sec) Modal vibration period, T n (sec)
Mode Number of stories Number of stories
3 6 9 12 15 18 3 6 9 12 15 18
1 0.492 0.857 1.186 1.492 1.784 2.064 0.791 1.377 1.905 2.398 2.867 3.317
2 0.164 0.323 0.458 0.580 0.693 0.799 0.262 0.512 0.723 0.912 1.094 1.271
3 0.074 0.178 0.268 0.348 0.422 0.491 0.118 0.282 0.423 0.548 0.667 0.782
4 0.109 0.177 0.238 0.294 0.346 0.173 0.280 0.375 0.465 0.551
5 0.073 0.125 0.174 0.219 0.261 0.117 0.198 0.274 0.346 0.415
6 0.054 0.092 0.132 0.169 0.205 0.086 0.145 0.208 0.268 0.326
30
7 0.071 0.103 0.135 0.165 0.112 0.163 0.214 0.264
8 0.057 0.083 0.110 0.136 0.091 0.131 0.174 0.217
9 0.047 0.068 0.091 0.114 0.075 0.108 0.144 0.182
10 0.058 0.077 0.097 0.092 0.122 0.155
11 0.049 0.066 0.083 0.079 0.105 0.133
12 0.043 0.057 0.073 0.069 0.092 0.116
13 0.051 0.064 0.081 0.102
14 0.045 0.057 0.072 0.091
15 0.041 0.051 0.065 0.082
16 0.046 0.074
17 0.042 0.068
18 0.039 0.063
Table 3.2 Modal damping ratio, n , of "regular" frames
(a) T 1=T L (b) T 1=T U

Modal damping ratio, n (%) Modal damping ratio, n (%)
Mode Number of stories Number of stories
3 6 9 12 15 18 3 6 9 12 15 18
1 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
2 5.0 3.8 3.4 3.4 3.5 3.5 5.0 3.8 4.0 3.4 3.5 3.5
3 8.9 5.0 3.9 4.0 4.0 4.0 8.9 5.0 5.0 4.0 4.0 4.0
4 7.3 5.0 5.0 5.0 5.0 7.3 6.8 5.0 5.0 5.0
5 10.4 6.6 6.4 6.3 6.2 10.4 9.2 6.4 6.3 6.2
6 14.0 8.8 8.2 7.9 7.7 13.9 12.2 8.2 7.9 7.7
31
7 11.2 10.3 9.7 9.3 15.6 10.3 9.7 9.3
8 13.8 12.7 11.7 11.2 19.3 12.7 11.7 11.2
9 16.7 15.3 14.1 13.2 23.3 15.2 14.1 13.2
10 18.0 16.6 15.5 17.9 16.6 15.5
11 21.0 19.3 18.0 20.7 19.2 17.9
12 24.1 22.1 20.6 23.6 22.0 20.6
13 25.1 23.3 24.9 23.3
14 28.1 26.1 27.8 26.1
15 31.2 29.1 30.7 29.0
16 32.1 31.9
17 35.2 34.9
18 38.2 37.8
Table 3.3 List of ground motions in LMSR ensemble
.. .
R* u go u go u go
No. Earthquake Name M Location Record
(km) (cm/s/s) (cm/s) (cm)
1 1989 Loma Prieta 6.9 Agnews State Hospital LP89agw 28.2 169 25.9 12.6
2 1989 Loma Prieta 6.9 Capitola LP89cap 14.5 435 29.2 5.5
3 1989 Loma Prieta 6.9 Gilroy Array #3 LP89g03 14.4 360 44.7 19.3
4 1989 Loma Prieta 6.9 Gilroy Array #4 LP89g04 16.1 208 37.9 10.1
5 1989 Loma Prieta 6.9 Gilroy Array #7 LP89gmr 24.2 221 16.4 2.5
6 1989 Loma Prieta 6.9 Hollister City Hall LP89hch 28.2 242 38.5 17.7
7 1989 Loma Prieta 6.9 Hollister Diff. Array LP89hda 25.8 274 35.6 13.0
8 1989 Loma Prieta 6.9 Sunnyvale - Colton Ave. LP89svl 28.8 203 37.3 19.1
32
9 1994 Northridge 6.7 Canoga Park - Topanga Canyon NR94cnp 15.8 412 60.7 20.3
10 1994 Northridge 6.7 LA - N Faring Rd NR94far 23.9 268 15.8 3.3
11 1994 Northridge 6.7 LA - Fletcher Dr NR94fle 29.5 236 26.2 3.6
12 1994 Northridge 6.7 Glendale - Las Palmas NR94glp 25.4 202 7.4 1.8
13 1994 Northridge 6.7 LA - Hollywood Stor FF NR94hol 25.5 227 18.2 4.8
14 1994 Northridge 6.7 La Crescenta - New York NR94nya 22.3 156 11.3 3.0
15 1994 Northridge 6.7 Northridge-Saticoy St NR94stc 13.3 361 28.9 8.4
16 1971 San Fernando 6.6 LA - Hollywood Stor Lot SF71pel 21.2 171 14.8 6.3
17 1987 Superstition Hills 6.7 Brawley SH87bra 18.2 153 13.9 5.3
18 1987 Superstition Hills 6.7 El Centro Imp. Co. Center SH87icc 13.9 351 46.3 17.6
19 1987 Superstition Hills 6.7 Plaster City SH87pls 21.0 182 20.6 5.4
20 1987 Superstition Hills 6.7 Westmorland Fire Station SH87wsm 13.3 169 23.5 13.1
* The closest distance to fault rupture
3-story 6-story 9-story 12-story 15-story 18-story
Figure 3.1 Generic one-bay 3, 6, 9, 12, 15, 18-story frames used in this study.
3.5
3
T =0.045H0.80
U
Fundamental period, T1 (sec)
2.5
1.5
0.80
TL=0.028H
1
0.5
0
0 3 6 9 12 15 18 21
Number of stories
Figure 3.2 Fundamental period TL=0.028H0.8 and TU=0.045H0.8, which define the mean-
minus- and mean-plus-one-standard-deviation of measured periods of steel
moment-resisting frames (SMRF) [Goel and Chopra, 1997].
33
Figure 3.3 Beam-hinge model (Source: Nassar and Krawinkler, 1991).
(a) (b)
Plastic-hinge moment
M
EI, L

M
M 6 EI
k p = 3%kel kel = =
L
My 1
Interior beam modeled as an

elastic beam element
Plastic-hinge rotation
Figure 3.4 (a) Moment-rotation relationship of a plastic hinge is rigid-plastic with post-
yield stiffness kp=3%kel; (b) kel is the end-rotation stiffness of the interior
beam, which is modeled as an elastic beam element.
34
20
18
16
Modal damping ratio, (%)

14
n 12
10
0
0 10 20 30 40 50 60 70
Modal frequency, , (rad/sec)
n
Figure 3.5 Variation of modal damping ratio n against modal frequency n of a 6-story
frame with T1=TU.
35
(a) MF=2
KM1 KM2 KM3 KM4 KM5 KM6 KM7 KM8
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2
Story stiffness of irregular frame Story stiffness of reference regular frame
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2
Story strength of irregular frame Story strength of reference regular frame
(b) MF=5
KM1 KM2 KM3 KM4 KM5 KM6 KM7 KM8
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Figure 3.6 Ratio of story stiffness and of story strength of stiffness-irregular frames to
the corresponding properties of the regular frame for modification factors,
MF=2 and 5.
36
(a)
Base shear / Total weight (%)

1.5
0.5
0
0 0.05 0.1 0.15 0.2 0.25 0.3
(b)
2
1.5
1 Base case
KM1
KM2
KM3
KM4
0.5
KM5
KM6
KM7
KM8
0
0 0.05 0.1 0.15 0.2 0.25 0.3
Roof displacement / Total height (%)
Figure 3.7 Pushover curves of regular and stiffness-irregular frames: (a) before and
(b) after story strength is scaled uniformly so that the yield base shear is the
same as the regular frame; MF=2.
37
(a) MF=2
SM1 SM2 SM3 SM4 SM5 SM6 SM7 SM8
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2
(b) MF=5
SM1 SM2 SM3 SM4 SM5 SM6 SM7 SM8
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Figure 3.8 Ratio of story stiffness and of story strength of strength-irregular frames to
the corresponding properties of the regular frame for modification factors,
MF=2 and 5.
38
(a)

1.5
0.5
0
0 0.05 0.1 0.15 0.2 0.25 0.3
(b)
2
1.5
1 Base case
SM1
SM2
SM3
SM4
0.5
SM5
SM6
SM7
SM8
0
0 0.05 0.1 0.15 0.2 0.25 0.3
Figure 3.9 Pushover curves of regular and strength-irregular frames: (a) before and
(b) after story strength is scaled uniformly so that the yield base shear is the
same as the regular frame; MF=2.
39
(a) MF=2
KS1 KS2 KS3 KS4 KS5 KS6 KS7 KS8
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2 0 1 2
(b) MF=5
KS1 KS2 KS3 KS4 KS5 KS6 KS7 KS8
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
12
11
10
9
8
Story
7
6
5
4
3
2
1
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Figure 3.10 Ratio of story stiffness and of story strength of stiffness-and-strength-irregular

frames to the corresponding properties of the regular frame for
modification factors, MF=2 and 5.
40
(a)

1.5
0.5
0
0 0.05 0.1 0.15 0.2 0.25 0.3
(b)
2
1.5
1 Base case
KS1
KS2
KS3
KS4
0.5
KS5
KS6
KS7
KS8
0
0 0.05 0.1 0.15 0.2 0.25 0.3
Figure 3.11 Pushover curves of regular and stiffness-and-strength-irregular frames: (a)

before and (b) after story strength is scaled uniformly so that the yield base
shear is the same as the regular frame; MF=2.
41
LP89agw (No. 1) LP89cap (No. 2) LP89g03 (No. 3) LP89g04 (No. 4)
500
g (cm/s/s)
500
LP89gmr (No. 5) LP89hch (No. 6) LP89hda (No. 7) LP89svl (No. 8)

500
g (cm/s/s)
500
NR94cnp (No. 9) NR94far (No. 10) NR94fle (No. 11) NR94glp (No. 12)
500
g (cm/s/s)
500
NR94hol (No. 13) NR94nya (No. 14) NR94stc (No. 15) SF71pel (No. 16)
500
g (cm/s/s)
500
SH87bra (No. 17) SH87icc (No. 18) SH87pls (No. 19) SH87wsm (No. 20)
500
g (cm/s/s)
500
0 20 40 0 20 40 0 20 40 0 20 40
Time (sec) Time (sec) Time (sec) Time (sec)
Figure 3.12 LMSR ensemble of 20 ground motions: ground accelerations.
42
100
g (cm/s) 50
50
100

100
50
g (cm/s)
50
100
100
50
g (cm/s)
50
100
100
50
g (cm/s)
50
100
100
50
g (cm/s)
50
100
0 20 40 0 20 40 0 20 40 0 20 40
Figure 3.13 LMSR ensemble of 20 ground motions: ground velocities.
43
50
ug (cm)
0
50

50
ug (cm)
50
50
ug (cm)
50
50
ug (cm)
50
50
ug (cm)
50
0 20 40 0 20 40 0 20 40 0 20 40
Figure 3.14 LMSR ensemble of 20 ground motions: ground displacements.
44
(a) (b)
Elastic spectra, =1 Elastic spectra, =1
1.5 1.5
A (g) 1 1
0.5 0.5
0 0
0 1 2 3 4 0 1 2 3 4
Constantductility spectra, =1.5 Constantductility spectra, =1.5
1 1
Ay (g)
0.5 0.5
0 0
0 1 2 3 4 0 1 2 3 4
Constantductility spectra, =2 Constantductility spectra, =2
0.8 0.8
0.6 0.6
Ay (g)
0.4 0.4
0.2 0.2
0 0
0 1 2 3 4 0 1 2 3 4
0.4 0.4
Ay (g)
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 1 2 3 4 0 1 2 3 4
0.4 0.4
0.3 0.3
Ay (g)
0.2 0.2
0.1 0.1
0 0
0 1 2 3 4 0 1 2 3 4
Vibration period, Tn (sec) Vibration period, T (sec)
n
Figure 3.15 Constant-ductility pseudo-acceleration spectra for LMSR ensemble of ground

motions: (a) individual records and (b) 16th, 84th, and 50th (median)
percentile values; ductility factor, =1, 1.5, 2, 4, and 6; strain-hardening
ratio, =3%; damping ratio, =5%.
45
(a) (b)
Elastic spectra, =1 Elastic spectra, =1
80 80
60 60
D (cm)
40 40
20 20
0 0
0 1 2 3 4 0 1 2 3 4
Constantductility spectra, =1.5 Constantductility spectra, =1.5
40 40
30 30
D (cm)
20 20
y
10 10
0 0
0 1 2 3 4 0 1 2 3 4
20 20
15 15
D (cm)
10 10
y
5 5
0 0
0 1 2 3 4 0 1 2 3 4
15 15
D (cm)
10 10
y
5 5
0 0
0 1 2 3 4 0 1 2 3 4
10 10
D (cm)
5 5
y
0 0
0 1 2 3 4 0 1 2 3 4
Vibration period, Tn (sec) Vibration period, Tn (sec)
Figure 3.16 Constant-ductility yield-deformation spectra for LMSR ensemble of ground

motions: (a) individual records and (b) 16th, 84th, and 50th (median)
percentile values; ductility factor, =1, 1.5, 2, 4, and 6; strain-hardening
ratio, =3%; damping ratio, =5%.
46
4. ESTIMATING TARGET ROOF DISPLACEMENT
4.1 INTRODUCTION
The seismic demands are often computed by nonlinear static analysis of the structure
subjected to monotonically increasing lateral forces with an invariant height-wise
distribution until the target value of roof displacement is reached. This roof displacement
value is estimated from the earthquake-induced deformation of an inelastic SDF system
derived from the pushover curve.
This chapter investigates the basic premise that the roof displacement of a
multistory building can be determined from the deformation of an SDF system. For this
purpose, the roof displacement estimated by an inelastic SDF system is compared to the
exact value determined rigorously by nonlinear response history analysis (RHA),
without introducing any of the approximations underlying the simplified methods for
estimating the deformation of an SDF system. The statistics of the SDF-system estimate
of roof displacement are presented for the regular frames described in Chapter 3.
Results of similar investigation that includes SAC buildings can be found in [Chopra et
al, 2001].
4.2 ELASTIC SYSTEMS
Shown in Fig. 4.1 are the median and dispersion of the ratio ( ur* ) for elastic regular
SDF
frames plotted against the fundamental vibration period (or number of stories); results are
presented for both period values: T1 = TL and T1 = TU . This ratio starts very close to 1.0
for the 3-story frame and decreases to 0.85 for the 15-story frame (Fig. 4.1a), indicating
that the SDF estimate, ( ur )SDF , is biased in the sense that it underestimates the roof
displacement and that this bias increases for taller (or longer period) frames. The SDF
system consistently underestimates the roof displacement because it ignores the higher
mode contributions that are increasingly significant as the fundamental period lengthens
[Chopra, 2001, Chapter 18]. For the same reasons, dispersion starts at close to zero for
the 3-story frame and increases to 0.16 for the 15-story frame with T1 = TL , or 0.15 for the
18-story frame with T1 = TU (Fig. 4.1b).
47
When higher-mode contributions are included in RSA, the median of the ratio
(u )
*
r RSA becomes closer to 1.0 compared to ( ur* )
SDF
, indicating that the biasalthough
still an underestimationhas decreased (see Fig. 4.1a). Because the peak modal response
for each mode is computed exactly by RHA [Eq. (2.17)], the remaining bias is entirely
due to approximations associated with the modal combination rule [Eq. (2.18)]. While
this source of approximation is well known, it should be noted that the bias is consistently
an underestimation. The dispersion of roof displacement is also reduced when higher
mode contributions are included.
While the median and dispersion of the displacement ratio (u )

*
r SDF are two
important sample statistics, data for individual ground motions are also of interest. For
this purpose, histograms of the 20 values of the ratio are plotted in Figs. 4.2 and 4.3 for
period values: T1 = TL and T1 = TU , respectively. Note that while the SDF-system provides
an accurate estimate of displacement of the 3-story frame for every ground motion, it
underestimates the displacement of the 6-story (with T1 = TU ) and taller frames for a large
majority of excitations. This estimate can be alarmingly small for a few excitations. The
smallest values of ( ur* ) encountered are 0.96, 0.65, 0.62, 0.76, 0.70, and 0.57 for 3, 6,
SDF
9, 12, 15, and 18-story frames with T1 = TU , respectively.
Figures 4.4 and 4.5 present similar histograms for ( ur* ) . The ( ur* ) ratios are
RSA RSA
larger compared to ( ur* ) in Figs. 4.2 and 4.3 because the higher-mode contributions are
SDF
included using the SRSS modal combination rule; thus the bias becomes smaller.
However, the dispersion is not significantly reduced.
To better understand the reasons for this large underestimation of roof

displacement, Fig. 4.6 shows the time variation of roof displacement due to individual
vibration modes, its total values [Eq. (2.15)], and the ( ur )RSA value [Eq. (2.18)] for the 6-
story frame due to two of the 20 ground motions considered. Consistent with popular
belief, the first mode for one of the excitations is dominant (Fig. 4.6a), therefore the SDF
estimate, ( ur )SDF , of roof displacement is essentially exact (4.71 versus 4.67 cm.). For the
48
other excitation (Fig. 4.6b), however, while the displacement due to the first mode is
largest among all modes, the relative contributions of various modes and how they
combine is such that the exact peak response (5.23 cm) is much larger than the first mode
value (3.41 cm); the SDF system underestimates the roof displacement by 35%. If the
contributions of the first three modes are included in RSA, the error is reduced slightly
and the roof displacement is underestimated by 26% (3.85 versus 5.23 cm).
4.3 INELASTIC SYSTEMS
The median and dispersion of the ratio (u ) *

r SDF are plotted versus the fundamental
vibration period (or number of stories) in Fig. 4.7 and versus the design ductility factor,
, in Fig. 4.8; results are presented for both period values: T1 = TL and T1 = TU . This
median ratio starts very close to 1.0 for 3-story frames irrespective of the design ductility
factor, , but increasingly differs from 1.0 and becomes increasingly dependent on as
T1 becomes longer (see Figs. 4.7a and 4.8a). The SDF-system estimate, ( ur )SDF , is biased
as expected, but the nature and magnitude of this bias depends on . For smaller , the
SDF-system method underestimates the roof displacement; this bias increases for longer-
period systems (or taller frames) just as in the case of elastic systems (Fig. 4.1a). The
situation is reversed for larger ; for = 6 the SDF-system method overestimates the
roof displacement, and this bias increases for taller frames (Figs. 4.7a and 4.8a). For
intermediate values of , the ratio ( ur* ) is closer to one, implying that the SDF-system
SDF
estimate of roof displacement is relatively more accurate for frames of all heights. The
dispersion tends to increase for taller frames for every value of (Fig. 4.7b). It is
smallest for elastic systems and tends to increase with the design ductility factor, but this
trend is not perfect (see Fig. 4.8b).
Including higher mode contributions according to the MPA procedure (Figs.

4.9a and 4.10a) obviously increases the estimate ( ur )MPA of the roof displacement
relative to the SDF-system estimate ( ur )SDF , thus the ( ur* ) plot is shifted up (compare
MPA
Figs. 4.7a and 4.9a). As a result, MPA overestimates the roof displacement except for
49
elastic or nearly elastic cases where it underestimates to a lesser degree than the SDF-
system estimate. Generally, this overestimation is modest, except for combinations of
very long periods and large design ductility values. Including higher mode
contributions to the roof displacement reduces the dispersion significantly for lower
values of and, to a lesser degree, for larger values of (compare Figs. 4.8b and
4.10b).
Shown in Figs. 4.11 and 4.12 are the histograms of the 20 values of the ratio
(u )
*
r SDF together with the range of values and median value of this ratio for each of the
six frames with design ductility factor = 6 for period values: T1 = TL and T1 = TU ,
respectively. The SDF-system estimate of roof displacement can be alarmingly small for
individual ground motions for frames as low as 6 stories and, of course, for taller frames.
The smallest values of ( ur* ) encountered are 0.89, 0.72, 0.73, 0.66, 0.72, and 0.75 for
SDF
3, 6, 9, 12, 15, and 18 story-frames (with T1 = TU ), respectively. The SDF-system

estimate can also be surprisingly large for a few excitations, especially for taller frames.
The largest values of ( ur* ) observed are 1.40, 1.62, 1.46, 1.38, 1.58 and 1.88 for 3, 6, 9,
SDF
12, 15, and 18-story frames (with T1 = TU ). A comparison of Figs. 4.11 and 4.12 to Figs.
4.2 and 4.3 indicates that the ( ur* ) ratio varies over a much wider range for inelastic
SDF
systemsand good accuracy occurs less oftencompared to elastic systems. The

histograms of ratio ( ur* ) shown in Figs. 4.13 and 4.14 demonstrate that the range of
MPA
values does not narrow significantly, implying that even when higher mode
contributions are included large error can occur in roof displacement estimates for
individual ground motions.
To investigate the large error in the SDF-system estimate of roof displacement,

the error due to each assumption in UMRHA described in Section 2.3.2 is examined
separately. While the coupled nonlinear equation [Eq. (2.27)] must be solved to
bg
determine the exact roof displacement due to p eff,n t , the first assumption implies that
an approximate result can be obtained from Eq. (2.12), with Dn ( t ) determined by
50
nonlinear RHA of the nth-mode inelastic SDF system, governed by Eq. (2.29). Figure
2.8a already compared this approximate solution for a 6-story frame with = 6 to the
exact result for n = 1, 2, and 3. Errors in this SDF-system method are only a few
percent, although the frame deforms well into the inelastic range and undergoes
considerable permanent drift. This demonstrated accuracy of an SDF-system method to
bg
estimate the roof displacement response to p eff,n t has been confirmed for buildings
ranging from 3 to 18 stories, SDF-system ductility varying from 1 to 6, and 20 ground

motions.
The second assumption implies that superposition [Eq. (2.15)] of the exact roof
bg
displacements urn ( t ) due to p eff,n t n = 1, 2, 3determined by nonlinear RHA of
the MDF system [Eq. (2.27)]will provide a good approximation to the exact response
of the MDF system to peff ( t ) [Eq. (2.2)]. Based on Fig. 2.8a, this implies that the
superposition of the approximate urn ( t ) determined using Dn ( t ) of the nth-mode
inelastic SDF system should provide a good approximation to the exact value;
however, this is not always the case. The response history of modal contributions, the
combined value determined by UMRHA, and the exact response by nonlinear RHA are
presented for the 6-story frame designed for = 6 due to three of the 20 ground motions
in Fig. 4.15; also included is the ( ur )MPA value determined from Eq. (2.18). In the first
case the first mode contribution is dominant; yielding causes very little drift of the first-
mode SDF system away from its zero-displacement position in spite of the large design
ductility, and the SDF-system estimate of the roof displacement is very close to the
exact value determined by nonlinear RHA (see Fig. 4.15a). In the second case the first
mode contribution is dominant, but the yielding-induced permanent drift is much
smaller than seen in the exact response by nonlinear RHA (see Fig. 4.15b).
Consequently, the SDF system underestimates the roof displacement by 28%. In the third
case the first mode contribution remains dominant, but the yielding-induced permanent
drift in the first mode SDF system is larger than seen in the results of nonlinear RHA
(Fig. 4.15c). Consequently, the SDF system overestimates the roof displacement by 62%.
51
For the latter two ground motions, little if any improvement is achieved by
including higher mode contributions according to the UMRHA procedure. This
persistent discrepancy implies that the second assumption identified earlier in developing
the UMRHA procedure is not always valid; it works in Fig. 4.15a but not in Figs. 4.15b
or 4.15c. When it does not work, the roof displacement ( ur )MPA estimated by MPA is
also inaccurate (see values noted in Fig. 4.15). In principle, this estimate should be less
accurate than the UMRHA result because it contains additional modal combination
errors, however, that is not always the case because errors due to various approximations
can cancel or reinforce each other.
52
(a) Median
1.4 1.4
T1=TL SDF T1=TU
RSA
1.2 1.2
Median u *
r
1 1
0.8 0.8
0.6 0.6
0 1 2 3 0 1 2 3
(b) Dispersion
0.3 0.3
T =T T =T
1 L 1 U
Dispersion of u *
r
0.2 0.2
0.1 0.1
0 0
0 1 2 3 0 1 2 3
Fundamental period, T (sec) Fundamental period, T (sec)
1 1 1 1
0.5
0 0.5
0
3 6 9 12 15 18 3 6 9 12 15 18
Number of stories Number of stories
Figure 4.1 (a) Median and (b) dispersion of ur* ( )SDF and (ur* )RSA versus fundamental
vibration period T1 for elastic regular frames.
53
(a) 3story frame (b) 6story frame
20 20
Range = 0.99 to 1.02 Range = 0.92 to 1.09
Median = 1.005
Median = 0.991
Number of records 15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(c) 9story frame (d) 12story frame

20 20
Range = 0.66 to 1.02 Range = 0.66 to 1.06
Median = 0.915
Median = 0.915
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(e) 15story frame (f) 18story frame

20 20
Range = 0.55 to 1.06 Range = 0.61 to 1.02
Median = 0.859
Median = 0.886
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(ur*)SDF (ur*)SDF
Figure 4.2 Histograms of ratio ur* ( )SDF for elastic regular frames with T = T ; range 1 L
of values and median value of this ratio are noted.
54
20 20
Range = 0.96 to 1.02 Range = 0.65 to 1.08
Median = 0.996
Median = 0.928
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.62 to 1.02 Range = 0.76 to 1.10
Median = 0.892
Median = 0.923
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.70 to 1.00 Range = 0.57 to 1.04
Median = 0.850
Median = 0.867
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(ur*)SDF (ur*)SDF
Figure 4.3 Histograms of ratio ur* ( )SDF for elastic regular frames with T = T ; range 1 U
55
20 20
Range = 0.99 to 1.02 Range = 0.93 to 1.11
Median = 1.006
Median = 1.000
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.81 to 1.04 Range = 0.82 to 1.23
Median = 0.947
Median = 0.965
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.74 to 1.08 Range = 0.74 to 1.15
Median = 0.928
Median = 0.949
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(ur*)RSA (ur*)RSA
Figure 4.4 Histograms of ratio ur* ( )RSA for elastic regular frames with T = T ; range 1 L
56
20 20
Range = 0.96 to 1.03 Range = 0.73 to 1.08
Median = 0.998
Median = 0.961
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.76 to 1.08 Range = 0.81 to 1.22
Median = 0.946
Median = 0.968
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.79 to 1.02 Range = 0.70 to 1.09
Median = 0.916
Median = 0.941
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(ur*)RSA (ur*)RSA
Figure 4.5 Histograms of ratio ur* ( )RSA for elastic regular frames with T = T ; range 1 U
57
(a) Record No. 16 (b) Record No. 05
10 10
ur1 (cm)
ur1 (cm)
0 0
3.41
4.71
Mode 1 Mode 1
10 10
10 10
ur2 (cm)
ur2 (cm)
0 0
1.27 1.78
Mode 2 Mode 2
10 10
10 10
ur3 (cm)
ur3 (cm)
0.150
0 0
0.197
Mode 3 Mode 3
10 10
10 10
(u ) = 4.88 5.23 (u ) = 3.85
r RSA r RSA
ur (cm)
ur (cm)
0 0
4.67
RHA RHA
10 10
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Figure 4.6 Modal contributions to roof displacement of elastic 6-story frame to LMSR
ground motions: (a) Record No. 16 and (b) Record No. 5; RSA estimate of roof
displacement is also noted.
58
(a) Median
1.4 1.4
T1=TL Elastic T1=TU
SDF = 1
1.2 1.5 1.2
Median of (u *)
2
r
4
1 6 1
0.8 0.8
0.6 0.6
1 2 3 1 2 3
(b) Dispersion
0.3 0.3
T =T T =T
1 L 1 U
SDF
Dispersion of (u *)
r
0.2 0.2
0.1 0.1
0 0
1 2 3 1 2 3
1 1 1 1
0.5
0 0.5
0
3 6 9 12 15 18 3 6 9 12 15 18
Figure 4.7 (a) Median and (b) dispersion of ur* ( )SDF versus fundamental vibration period
T1 for inelastic regular frames.
59
(a) Median
1.4 1.4
T =T T =T
Median of (ur*)SDF 1 L 1 U
1.2 1.2
1 N= 3 1
6
9
0.8 12 0.8
15
18
0.6 0.6
Elastic
Elastic
1 2 4 6 1 2 4 6
(b) Dispersion
0.3 0.3
T =T T =T
1 L 1 U
Dispersion of (ur*)SDF
0.2 0.2
0.1 0.1
0 0
Elastic
Elastic
1 2 4 6 1 2 4 6
SDFsystem ductility factor, SDFsystem ductility factor,
Figure 4.8 (a) Median and (b) dispersion of ur*

SDF
( )
versus SDF-system ductility factor
for inelastic regular frames of 3, 6, 9, 12, 15 and 18 stories.
60
(a) Median
1.4 1.4
T1=TL Elastic T1=TU
MPA = 1
1.2 1.5 1.2
2
Median of (u *)
r
4
1 6 1
0.8 0.8
0.6 0.6
1 2 3 1 2 3
(b) Dispersion
0.3 0.3
T =T T =T
1 L 1 U
MPA
Dispersion of (u *)
0.2 0.2
r
0.1 0.1
0 0
1 2 3 1 2 3
1 1 1 1
0.5
0 0.5
0
3 6 9 12 15 18 3 6 9 12 15 18
Figure 4.9 (a) Median and (b) dispersion of (ur* )MPA versus fundamental vibration
period T1 for inelastic regular frames.
61
(a) Median
1.4 1.4
T =T T =T
Median of (ur*)MPA 1 L 1 U
1.2 1.2
1 N= 3 1
6
9
0.8 12 0.8
15
18
0.6 0.6
Elastic
Elastic
1 2 4 6 1 2 4 6
(b) Dispersion
0.3 0.3
T =T T =T
1 L 1 U
Dispersion of (ur*)MPA
0.2 0.2
0.1 0.1
0 0
Elastic
Elastic
1 2 4 6 1 2 4 6
Figure 4.10 (a) Median and (b) dispersion of ur*

MPA
versus SDF-system ductility factor ( )
for inelastic regular frames of 3, 6, 9, 12, 15 and 18 stories.
62
20 20
Range = 0.96 to 1.16 Range = 0.72 to 1.44
Median = 1.041
Median = 1.001
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.72 to 1.29 Range = 0.81 to 1.33
Median = 0.990
Median = 1.064
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.68 to 1.55 Range = 0.75 to 1.59
Median = 1.135
Median = 1.100
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(ur*)SDF (ur*)SDF
Figure 4.11 Histograms of ratio ur* ( )SDF for regular frames with design ductility factor
= 6 and T1 = TL ; range of values and the median value of this ratio are noted.
63
20 20
Range = 0.89 to 1.40 Range = 0.72 to 1.62
Median = 1.014
Median = 1.010
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.73 to 1.46 Range = 0.66 to 1.38
Median = 1.082
Median = 1.088
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.72 to 1.58 Range = 0.75 to 1.88
Median = 1.184
Median = 1.191
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(ur*)SDF (ur*)SDF
Figure 4.12 Histograms of ratio ur* ( )SDF for regular frames with design ductility factor
= 6 and T1 = TU ; range of values and the median value of this ratio are
noted.
64
20 20
Range = 0.96 to 1.16 Range = 0.74 to 1.45
Median = 1.041
Median = 1.008
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.78 to 1.31 Range = 0.85 to 1.45
Median = 1.016
Median = 1.105
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.71 to 1.64 Range = 0.78 to 1.69
Median = 1.183
Median = 1.164
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(u *) (u *)
r MPA r MPA
Figure 4.13 Histograms of ratio ur*

MPA
for regular frames with design ductility factor ( )
= 6 and T1 = TL ; range of values and the median value of this ratio are noted.
65
20 20
Range = 0.89 to 1.41 Range = 0.73 to 1.64
Median = 1.016
Median = 1.031
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.75 to 1.51 Range = 0.77 to 1.42
Median = 1.118
Median = 1.148
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

20 20
Range = 0.78 to 1.69 Range = 0.86 to 1.96
Median = 1.260
Median = 1.289
Number of records
15 15
10 10
5 5
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(u *) (u *)
r MPA r MPA
Figure 4.14 Histograms of ratio ur*

MPA
for regular frames with design ductility factor ( )
= 6 and T1 = TU ; range of values and the median value of this ratio are
noted.
66
(a) Record No. 15 (b) Record No. 5 (c) Record No. 14
20 5 10
Mode 1 3.06 Mode 1 Mode 1
ur1 (cm)
0 0 0
12.33 5.25
20 5 10
20 5 10
Mode 2 Mode 2 Mode 2
ur2 (cm)
0.75
0 0 0
2.35 0.84
20 5 10
20 5 10
Mode 3 Mode 3 Mode 3
ur3 (cm)
0.12
0 0 0
0.34 0.11
20 5 10
20 5 10
(ur)UMRHA (cm)
(u ) =12.57 UMRHA 3.16 UMRHA (u ) =5.32 UMRHA

r MPA r MPA
0 0 0
12.84 (u ) =3.14 5.64

r MPA
20 5 10
20 5 4.23 10
(ur)NLRHA (cm)
NLRHA NLRHA
0 0 0
3.24
12.41 NLRHA
20 5 10
0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30
Time (sec) Time (sec) Time (sec)
Figure 4.15 Response histories of roof displacement of a 6-story frame ( =6) due to three
ground motions: individual modal responses, combined response from
UMRHA, and exact response from nonlinear RHA; parts (a) and (c) are for
frames with T1 = TU and part (b) is for T1 = TL ; MPA estimate of roof
displacement is also noted.
67
5. EVALUATION OF MPA PROCEDURE FOR VERTICALLY
REGULAR FRAMES
5.1 INTRODUCTION
The recently developed MPA procedure has been tested for only a few buildings [Goel
and Chopra, 2002]. Being an approximate method, however, it should obviously be
evaluated comprehensively before practical application to building evaluation and design.
The objective of this chapter is to expand the previously obtained results and evaluate the
accuracy of the MPA procedure for a wide range of buildings and ground motion
ensembles. The structures analyzed in this chapter are the regular frames of six
different heights: 3, 6, 9, 12, 15, and 18 stories and five strength levels corresponding to
SDF-system ductility factor of 1, 1.5, 2, 4, and 6, described in Section 3.2.1. Each frame
is analyzed for twenty ground motions. The median values of seismic demands from the
MPA and nonlinear response history analysis (RHA) procedures are compared, and the
bias and dispersion in the estimate of demand from the MPA procedure are presented and
analyzed. Results for story drifts and their interpretations are emphasized because they
are better related to structural damage; however, floor displacement results are included
and briefly discussed.
5.2 NONLINEAR RESPONSE HISTORY ANALYSIS RESULTS
Figures 5.1 and 5.2 show the median values of story-drift demands determined by
nonlinear RHA over the height of 60 regular frames for both period values: T1 = TL and
T1 = TU , respectively. Also included are results from elastic analyses of each frame
implemented by linear RHA. The distribution of story-drift demandswhich approach
1%over the frame height is as follows:
is essentially uniform for 3-story frames;
becomes increasingly non-uniform for taller frames;
increases at upper stories of elastic frames where the contribution from higher modes
is significant;
depends significantly on the SDF-system ductility factor, ; note that base-story drift
68
is essentially independent of , but top-story drift decreases as increases or as the
strength of the structure decreases.
Median values of floor displacements for these frames are plotted in Figs. 5.3 and 5.4.
The deflected shapes of short frames are closer to a straight line than those of taller
frames. The roof displacement tends to decrease as the strength decreases (or
increases).
5.3 COMPARISON OF MPA AND NONLINEAR RHA RESULTS
The MPA procedure has been implemented for each of the 60 frames and for each of the
20 excitations, and contributions from several modes were considered. The combined
values of story drifts were computed including one, two, or three modes. Figures 5.5
and 5.6 show these median values of story-drift demands for 18 ( T1 = TU ) of the 60
frames, together with the results of nonlinear RHA obtained from Fig. 5.2. Also included
are the results of linearly-elastic analysis, in which nonlinear RHA reduces to linear RHA
and MPA is equivalent to RSA.
Note that the first mode alone is inadequate in estimating story drifts (see Figs.
5.5 and 5.6); however, the estimates are much more accurate by introducing the response
contributions due to the second mode for 3, 6, and 9-story frames, and second and third
modes for 12, 15, and 18-story frames. With two or three modes included, the
height-wise distribution of story drifts estimated by MPA is generally similar to the
exact results detailed in the preceding section. The MPA predicts reasonably well the
changing height-wise variation of demand with building height and SDF-system ductility
factor.
A similar investigation comparing floor displacements determined by MPA and

nonlinear RHA procedures demonstrated that the first mode alone is, however,
sufficient in estimating these response quantities (Figs. 5.7 and 5.8).
5.4 BIAS AND DISPERSION OF MPA PROCEDURE
Figures 5.9 and 5.10 show the median of the drift ratio *MPA for the 60 regular frames
(both period values: T1 = TL and T1 = TU ) plotted over frame height, including a sufficient
69
number of modes to ensure that the sum of the modal contribution factors [Chopra, 2001:
Section 12.10] for the top-story drift exceed 95%: 2 for 3-story frame, 3 for 6-story
frame, 4 for 9- and 12-story frames, and 5 for 15- and 18-story frames. These results
permit the following observations:
The MPA procedure is least biased for 3-story frames and relatively more biased for
taller frames. For a fixed value of design ductility factor, , this trend is imperfect
and will be discussed further.
For a fixed frame height, the bias in MPA does not display any systematic
dependence on the SDF-system ductility factor, , and will be examined further
The MPA procedure accurately estimates seismic demands for 3- and 6-story frames;
the bias is less than 10% and 20% for 3- and 6-story frames, respectively.
The MPA procedure estimates drift demands for 9, 12, 15, and 18-story frames with
less than 25% bias in most cases; however, it underestimates demand by 30% in
upper stories for cases with low design ductility ( = 1.5 and 2) cases and
overestimates demand by more than 50% in the top-three stories of the 15- and 18-
story building for =6.
For the same frame height (number of stories), MPA procedure for frames with
longer period ( T1 = TU ) in Fig. 5.10 tends to be more biased than frames with shorter
period ( T1 = TL ) in Fig. 5.9 because higher mode contributions are more significant in
longer period structures.
Figures 5.11 and 5.12 show the dispersion of the drift ratio *MPA plotted over the height
of the same 60 frames. The dispersion is smallest (less than 15%) in 3-story frames but
tends to increase for taller frames as the contribution from the higher-modes becomes
more significant. The dispersion is smaller than 30% for most taller frames, but it exceeds
40% in the upper stories of 12, 15, and 18-story frames. The dispersion tends to increase
in the upper stories because higher-mode contributions are more significant.
To demonstrate how bias and dispersion measures relate to the accuracy of the
MPA procedure, the MPA estimate of story drift MPA is plotted against the exact
70
value NL-RHA in Fig. 5.13. Results are presented for top-, middle-, and bottomstory
drifts of 12-story frames with =2, 4, and 6 and T1 = TU ; each data point corresponds to
one ground motion. The diagonal line represents MPA = NL-RHA ; a data point above this
line implies that MPA overestimates demand for the particular excitation and a data point
below this line implies that MPA underestimates demand. The median value and
dispersion of the drift ratio *MPA are also noted.
Note Figure 5.13b, which considers top-story drift and an SDF-system ductility
factor =4; the bias is very small (3.4%), but the dispersion is quite large (0.358) because
the data points scatter over a wide range from underestimation to significant
overestimation. The MPA estimate is significantly inaccurate for several excitations,
although the median drift ratio *MPA is very close to 1. Compare this case to Fig. 5.13e
for 6th-story drift and the same =4, where the bias is still small and similar in
magnitude as the previous case, but the dispersion is smaller (0.249), indicating that MPA
estimates the demand accurately more often than the previous case. This demonstrates
that an approximate procedure, such as MPA, is more likely to be accurate for an
individual excitation if both its bias and dispersion are small.
Next, Figure 5.13a considers the top-story drift and an SDF-system ductility
factor =2. The bias is large (27%), but the dispersion is relatively small (0.213). In
this case, MPA underestimates consistently by about 27% the top-story drift demand for
many ground motions; the MPA estimate corrected by the factor of 1/0.735 (1.36) would
very often be close to the exact drift demand, NL-RHA , determined by nonlinear RHA.
Recall that the bias and dispersion of MPA estimate of drift in upper stories of tall
frames with large are very large, e.g., *MPA =1.28 and MPA =0.38 for the top-story
drift of 12-story frame with =6. Figure 5.13c shows the case where both bias and
dispersion are large, and clearly the MPA procedure does not provide a reasonable
estimate for many ground motions. Figures 5.13b and 5.13c support the impression that
MPA prediction tends to be unreliable when dispersion is larger than 30%.
71
On the other hand, Figures 5.13d to 5.13i show that MPA can estimate story drift
at the mid-height and at the first story very well for most excitations.
Figure 5.14 plots the median of the drift ratio *MPA for the top and bottom stories
versus fundamental period (or number of stories) for fixed for both period values:
T1=TL and T1=TU. Also shown is the median of *RSA for elastic frames, which will be
discussed later. For fixed , consistent with the well-known trend in higher mode
responses for elastic systems [Chopra, 2001: Chapter 18], the bias in the MPA procedure
tends to increase for longer-T1 (or taller) frames, although these trends are not perfect.
However, the largest bias over all (shown by the shaded band) increases systematically
as T1 gets longer (or frames are taller).
Plotted in Fig. 5.15 is the dispersion of the drift ratio for the top and bottom
stories versus fundamental period (or number of stories) for fixed ; results are presented
for both period values: T1=TL and T1=TU. The dispersion increases with period
although the trend is not perfectbecause the higher-mode contributions to seismic
demand become more significant, therefore the accuracy of MPA deteriorates. As
previously observed, both bias and dispersion of the bottom-story drift are smaller than
those of the top-story drift, indicating that the former response can be more accurately
estimated by MPA, even for individual excitation.
Figure 5.16 plots the median of the drift ratio *MPA for the top and bottom stories
versus the SDF-system ductility factor for fixed fundamental period (or number of
stories); results are presented for both T1 values. Intuition would suggest that the
accuracy of MPA should deteriorate for increasing , however, the bias in the MPA
procedure is not systematic and varies depending on . The bias found in taller frames
is more dependent on ductility factor than in shorter frames. For tall frames and small ,
top-story drifts are underestimated and bottom-story drifts are overestimated. The
situation is reversed when increases; top-story drifts are overestimated and bottom-
story drifts are underestimated. These results demonstrate that the MPA procedure is
more accurate for moderate values of than for smaller or larger values of .
72
The dispersion of the drift ratio *MPA varies unsystematically between 0.1 and 0.3,
depending on the SDF-system ductility factor (Fig. 5.17). Although this dispersion
MPA of error in MPA for inelastic frames is much larger than the dispersion RSA for
elastic frames, MPA increases as increases from 1 to 2, but remains roughly constant
for greater than 2. However, the dispersion MPA for the top-story drifts of tall frames
(N 12) continues to increase as increases, becoming as large as 0.4 for =6,
indicating that MPA estimates may be unreliable for individual ground motions, as
demonstrated earlier in Fig. 5.13c.
Similar results for bias and dispersion of MPA in estimating floor displacements
are plotted over the height of the 60 regular frames in Figs. 5.18-5.21. The bias and
dispersion for floor displacements is much smaller compared to story drifts, indicating
that floor displacements can be estimated by MPA more accurately than story-drift
demands. They tend to (1) increase with frame height (number of stories) and (2) vary
unsystematically with , as noted earlier for story-drift demands.
5.5 COMPARATIVE EVALUATION OF MPA AND RSA PROCEDURES
The MPA procedure applied to elastic systems is equivalent to the standard response
spectrum analysis (RSA) because now the modal coordinates are indeed uncoupled and
superposition of modal responses is perfectly valid. The RSA procedure, implemented in
most commercial software, has become a standard analytical tool for the structural
engineering profession. The principal approximation in RSA uses modal combination
rules to combine the peak modal responses to estimate the total response. As these errors
are tacitly considered acceptable by the profession, next we compare the bias in the MPA
procedure for inelastic systems with that in the RSA procedure for elastic systems.
For this purpose, elastic analysis of each of the 60 frames was implemented by
RSA and RHA methods. The median of the story-drift ratio *RSA , shown in Figs. 5.7,
5.8, 5.12, and 5.14, and the dispersion of the drift ratios, shown in Figs. 5.9, 5.10, 5.13,
and 5.15, lead to the following conclusions. :
RSA underestimates slightly the elastic response for shorter frames and up to 18% in
73
the upper stories of taller frames. The bias in the RSA procedure is smaller in lower
stories, larger in the upper stories, and increases gradually over the frame height (see
Figs. 5.7 and 5.8).
The dispersion of the drift ratio for elastic frames determined by the RSA procedure
increases with the fundamental period T1 (Fig. 5.13).
These trends in variation of bias and dispersion in RSA of elastic frames with T1 and
over frame height are similar to earlier observations from MPA results for inelastic
frames.
The MPA is more biased than RSA for most of the 60 frames and for most of the
design ductility values because of the aforementioned additional approximations
involved in the MPA procedure (Figs. 5.7, 5.8, and 5.12).
The height-wise variation of MPA bias is more irregular compared to RSA, especially
for the larger values of (Figs. 5.7 and 5.8).
The dispersion of the drift ratio for inelastic systems (determined by the MPA
procedure) is larger than for elastic systems (determined by RSA procedure) because
of the additional approximations in the MPA procedure.
Similar to the preceding observations regarding relative accuracy of MPA and RSA
procedures in estimating story-drift demands, the RSA procedure has less bias and
dispersion than the MPA procedure in estimating floor displacements (Figs. 5.18-5.21).
74
(a) 3story frames (b) 6story frames
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0 0.5 1 1.5 0 0.5 1 1.5
(c) 9story frames (d) 12story frames

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.5 1 1.5 0 0.5 1 1.5
(e) 15story frames (f) 18story frames

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.5 1 1.5 0 0.5 1 1.5
Story drift (%) Story drift (%)
Figure 5.1 Median story-drift demands determined by nonlinear RHA for five designs of
each of 3, 6, 9, 12, 15, and 18-story frames (with T1 = TL ) corresponding to
SDF-system ductility factor =1, 1.5, 2, 4, and 6. Results for elastic frames
are also included.
75
3 6
2 4
Story
1 2
G G
0 0.5 1 1.5 0 0.5 1 1.5

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.5 1 1.5 0 0.5 1 1.5

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.5 1 1.5 0 0.5 1 1.5
Story drift (%) Story drift (%)
Figure 5.2 Median story-drift demands determined by nonlinear RHA for five designs of
each of 3, 6, 9, 12, 15, and 18-story frames (with T1 = TU ) corresponding to
are also included.
76
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Displacement/Total height (%) Displacement/Total height (%)
Figure 5.3 Median floor displacements determined by nonlinear RHA for five designs of
each of 3, 6, 9, 12, 15, and 18-story frames (with T1 = TL ) corresponding to
are also included.
77
3 6
2 4
Story
1 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Displacement/Total height (%) Displacement/Total height (%)
Figure 5.4 Median floor displacements determined by nonlinear RHA for five designs of
each of 3, 6, 9, 12, 15, and 18-story frames (with T1 = TU ) corresponding to
are also included.
78
(a) 3story frame (b) 6story frame (c) 9story frame
Elastic Elastic Elastic
3 6 9
8
5
7
2 4 6
Story
5
NLRHA 3
4
MPA
1 1 "mode" 2 3
2 "modes" 2
1
3 "modes" 1
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
=2 =2 =2
3 6 9
8
5
7
2 4 6
Story
5
3
4
1 2 3
2
1
1
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
=4 =4 =4
3 6 9
8
5
7
2 4 6
Story
5
3
4
1 2 3
2
1
1
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
=6 =6 =6
3 6 9
8
5
7
2 4 6
Story
5
3
4
1 2 3
2
1
1
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
Story drift (%) Story drift (%) Story drift (%)
Figure 5.5 Median story-drift demands determined by MPA with variable number of
modes and nonlinear RHA for 3, 6, and 9-story frames (with T1 = TU ), each
designed for =2, 4, and 6. Results for elastic frames are also included.
79
12 15 18
16
10 12 14
8 12
9
Story
10
6 NLRHA
8
6 MPA
4 6 1 "mode"
3 4 2 "modes"
2
2 3 "modes"
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
=2 =2 =2
12 15 18
16
10 12 14
8 12
9
Story
10
6
8
6
4 6
3 4
2
2
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
=4 =4 =4
12 15 18
16
10 12 14
8 12
9
Story
10
6
8
6
4 6
3 4
2
2
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
=6 =6 =6
12 15 18
16
10 12 14
8 12
9
Story
10
6
8
6
4 6
3 4
2
2
G G G
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
Figure 5.6 Median story-drift demands determined by MPA with variable number of
modes and nonlinear RHA for 12, 15, and 18-story frames (with T1 = TU ),
each designed for =2, 4, and 6. Results for elastic frames are also included.
80
3 6 9
8
5
7
2 4 6
Story
5
3 NLRHA
4
MPA
1 2 3 1 "mode"
2 2 "modes"
1
1 3 "modes"
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
=2 =2 =2
3 6 9
8
5
7
2 4 6
Story
5
3
4
1 2 3
2
1
1
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
=4 =4 =4
3 6 9
8
5
7
2 4 6
Story
5
3
4
1 2 3
2
1
1
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
=6 =6 =6
3 6 9
8
5
7
2 4 6
Story
5
3
4
1 2 3
2
1
1
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
Displacement/Total height (%) Displacement/Total height (%) Displacement/Total height (%)
Figure 5.7 Median floor displacements determined by MPA with variable number of
modes and nonlinear RHA for 3, 6, and 9-story frames (with T1 = TU ), each
designed for =2, 4, and 6. Results for elastic frames are also included.
81
12 15 18
16
10 12 14
8 12
9
Story
10
6 NLRHA
8
6 MPA
4 6 1 "mode"
3 4 2 "modes"
2
2 3 "modes"
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
=2 =2 =2
12 15 18
16
10 12 14
8 12
9
Story
10
6
8
6
4 6
3 4
2
2
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
=4 =4 =4
12 15 18
16
10 12 14
8 12
9
Story
10
6
8
6
4 6
3 4
2
2
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
=6 =6 =6
12 15 18
16
10 12 14
8 12
9
Story
10
6
8
6
4 6
3 4
2
2
G G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
Displacement/Total height (%) Displacement/Total height (%) Displacement/Total height (%)
Figure 5.8 Median floor displacements determined by MPA with variable number of
modes and nonlinear RHA for 12, 15, and 18-story frames (with T1 = TU ),
each designed for =2, 4, and 6. Results for elastic frames are also included.
82
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Storydrift ratio *MPA or *RSA Storydrift ratio *MPA or *RSA
Figure 5.9 Median story-drift ratios *MPA for 3, 6, 9, 12, 15, and 18-story frames (with
T1 = TL ), each designed for =1, 1.5, 2, 4, and 6. Results *RSA for linearly
elastic frames are also included.
83
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Storydrift ratio *MPA or *RSA Storydrift ratio *MPA or *RSA
Figure 5.10 Median story-drift ratios *MPA for 3, 6, 9, 12, 15, and 18-story frames (with
T1 = TU ), each designed for =1, 1.5, 2, 4, and 6. Results *RSA for linearly
elastic frames are also included.
84
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of *MPA or *RSA Dispersion of *MPA or *RSA
Figure 5.11 Dispersion of story-drift ratios *MPA for 3, 6, 9, 12, 15, and 18-story frames
(with T1 = TL ), each designed for =1, 1.5, 2, 4, and 6. Dispersion of *RSA for
linearly elastic frames is also included.
85
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of *MPA or *RSA Dispersion of *MPA or *RSA
Figure 5.12 Dispersion of story-drift ratios *MPA for 3, 6, 9, 12, 15, and 18-story frames
(with T1 = TU ), each designed for =1, 1.5, 2, 4, and 6. Dispersion of *RSA for
linearly elastic frames is also included.
86
(a) =2, 12 story (b) =4, 12thstory (c) =6, 12thstory
th
3 3 3
2.5 * =0.735 2.5 *MPA=1.034 2.5 *MPA=1.282

MPA
2 2 2
MPA (%)
1.5 1.5 1.5
1 1 1
0.5 0.5 0.5

=0.213 =0.358 =0.383
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
(d) =2, 6 story (e) =4, 6thstory (f) =6, 6thstory

th
3 3 3
2.5 *MPA=1.035 2.5 *MPA=0.966 2.5 *MPA=1.000

2 2 2
MPA (%)
1.5 1.5 1.5
1 1 1
0.5 0.5 0.5

=0.252 =0.249 =0.166
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
(g) =2, 1ststory (h) =4, 1ststory (i) =6, 1ststory

3 3 3
2.5 *MPA=1.126 2.5 *MPA=0.891 2.5 *MPA=0.838

2 2 2
MPA (%)
1.5 1.5 1.5
1 1 1
0.5 0.5 0.5

=0.159 =0.137 =0.138
0 0 0
0 1 2 3 0 1 2 3 0 1 2 3
NLRHA (%) (%) (%)
NLRHA NLRHA
Figure 5.13 Plots of MPA-estimate of story-drift demand, MPA , versus exact value
NL-RHA for top-, middle-, and bottom-stories of 12-story frames with
T1 = TU and =2, 4, and 6.
87
(a) Topstory drift; T =T (c) Topstory drift; T =T
1 L 1 U
Median drift ratio *MPA or *RSA

1.6 1.6
Elastic
1.4 = 1 1.4
1.5
1.2 2 1.2
4
1 6 1
0.8 0.8
0.6 0.6
0.4 0.4
0 1 2 3 0 1 2 3
(b) Bottomstory drift; T =T (d) Bottomstory drift; T =T

1 L 1 U
Median drift ratio *MPA or *RSA
1.6 1.6
1.4 1.4
1.2 1.2
1 1
0.8 0.8
0.6 0.6
0.4 0.4
0 1 2 3 0 1 2 3
1 1 1 1
0.5
0 0.5
0
3 6 9 12 15 18 3 6 9 12 15 18
Figure 5.14 Median story-drift ratio *MPA and *RSA for top and bottom stories of frames
(with T1 = TL and T1 = TU ) plotted against fundamental vibration period, T1 ,
(and number of stories) for fixed .
88
Dispersion of drift ratio *MPA or *RSA Dispersion of drift ratio *MPA or *RSA
1 L 1 U
0.5 0.5
Elastic
= 1
0.4 0.4
1.5
2
0.3 4 0.3
6
0.2 0.2
0.1 0.1
0 0
0 1 2 3 0 1 2 3
(b) Bottomstory drift; T1=TL (d) Bottomstory drift; T1=TU

0.5 0.5
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0 0
0 1 2 3 0 1 2 3
Fundamental period, T1 (sec) Fundamental period, T1 (sec)
1
0.5 1
0.5
0 0
3 6 9 12 15 18 3 6 9 12 15 18
Figure 5.15 Dispersion of story-drift ratios *MPA and *RSA for top and bottom stories of
frames (with T1 = TL and T1 = TU ) plotted against fundamental vibration
period, T1 , (and number of stories) for fixed .
89
(a) Topstory drift; T1=TL (c) Topstory drift; T1=TU
RSA
1.6 1.6
or *
N= 3
1.4 1.4 6
MPA 9
1.2 1.2 12
15
Median drift ratio *
1 1 18
0.8 0.8
0.6 0.6
0.4 0.4
Elastic
Elastic
1 2 4 6 1 2 4 6

1 L 1 U
RSA
1.6 1.6
or *
1.4 1.4
MPA
1.2 1.2
Median drift ratio *
1 1
0.8 0.8
0.6 0.6
0.4 0.4
Elastic
Elastic
1 2 4 6 1 2 4 6
Figure 5.16 Median story-drift ratio *MPA and *RSA for top and bottom stories of 3, 6, 9,
12, 15, and 18-story frames (with T1 = TL and T1 = TU ) plotted against SDF-
system ductility factor, .
90
RSA
1 L 1 U
or *
0.5 0.5
N= 3
6
MPA 0.4 0.4
9
Dispersion of drift ratio *
12
0.3 15 0.3
18
0.2 0.2
0.1 0.1
0 0
Elastic
Elastic
1 2 4 6 1 2 4 6
RSA

1 L 1 U
or *
0.5 0.5
MPA
0.4 0.4
Dispersion of drift ratio *
0.3 0.3
0.2 0.2
0.1 0.1
0 0
Elastic
Elastic
1 2 4 6 1 2 4 6
Figure 5.17 Dispersion of story-drift ratios *MPA and *RSA for top and bottom stories of
3, 6, 9, 12, 15, and 18-story frames (with T1 = TL and T1 = TU ) plotted against
SDF-system ductility factor, .
91
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Floordisplacement ratio u*MPA or u*RSA Floordisplacement ratio u*MPA or u*RSA
*
Figure 5.18 Median floor-displacement ratios uMPA for 3, 6, 9, 12, 15, and 18-story frames
(with T1 = TL ), each designed for =1, 1.5, 2, 4, and 6. Results uRSA

*
for
linearly elastic frames are also included.
92
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Floordisplacement ratio u*MPA or u*RSA Floordisplacement ratio u*MPA or u*RSA
*
Figure 5.19 Median floor-displacement ratios uMPA for 3, 6, 9, 12, 15, and 18-story frames
(with T1 = TU ), each designed for =1, 1.5, 2, 4, and 6. Results uRSA

*
for
linearly elastic frames are also included.
93
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of u*MPA or u*RSA Dispersion of u*MPA or u*RSA
*
Figure 5.20 Dispersion of floor-displacement ratios uMPA for 3, 6, 9, 12, 15, and 18-story
frames (with T1 = TL ), each designed for =1, 1.5, 2, 4, and 6. Dispersion of

*
uRSA for linearly elastic frames is also included.
94
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of u*MPA or u*RSA Dispersion of u*MPA or u*RSA
*
Figure 5.21 Dispersion of floor-displacement ratios uMPA for 3, 6, 9, 12, 15, and 18-story
frames (with T1 = TU ), each designed for =1, 1.5, 2, 4, and 6. Dispersion of

*
uRSA for linearly elastic frames is also included.
95
6. EVALUATION OF MPA PROCEDURE FOR VERTICALLY
IRREGULAR FRAMES
6.1 INTRODUCTION
The seismic response of vertically irregular frames, the subject of numerous research
investigations, were reviewed in two recent, comprehensive investigations by
Valmundsson and Nau [1997] and Al-Ali and Krawinkler [1998], both studies
considering mass, stiffness, and strength irregularities separately and in various
combinations. The first of these investigations focused on evaluating building code
requirements for vertically irregular frame buildings, whereas the latter emphasized the
effects of vertical irregularities on heightwise variation of seismic demands and behavior
of frame buildings. It was found that among the four types of irregularity, the effect of
mass irregularity is the smallest, the effect of strength irregularity is larger than the effect
of stiffness irregularity, and the effect of combined-stiffness-and-strength irregularity is
the largest. The roof displacement was shown to be a stable parameter not affected
significantly by vertical irregularities [Al-Ali and Krawinkler, 1998].
Both of these comprehensive investigations were based on idealized frames

designed according to the strong-beam-weak-column philosophy. Such a column-hinge
model is likely to exaggerate the effects of irregularity by restricting the redistribution of
yielding to the story that yields first and forms a story mechanism. Therefore, this paper
studies the effects of vertical irregularities on seismic demands of frame buildings using
the more realistic strong-column-weak-beam frame or beam-hinge model; mass
irregularity is not considered here because its effects are known to be small [Al-Ali and
Krawinkler, 1998].
By studying the bias and dispersion of this approximate procedure, MPA has been
shown to be accurate enough in estimating seismic demands for seismic evaluation of
such regular buildings (Chapter 5). As mentioned earlier, because vertical irregularities
significantly influence the seismic demands on buildings, the next logical step is to
determine whether MPA can estimate seismic demands on irregular buildings to a degree
of accuracy sufficient for practical application. Furthermore, as all pushover analyses
96
aim to detect any deficiency in the structure that results in localizing large seismic
demands, MPAs potential in this regard remains to be evaluated
The objectives of this chapter are as follows: (1) to study the influence of vertical
irregularities in the stiffness and strength distribution, separately and in combination, on
seismic demands of strong-column-weak-beam frames by comparing the median seismic
demands on irregular and regular frames computed by nonlinear response history
analysis (RHA) for an ensemble of ground motions; and (2) to evaluate the accuracy of
MPA in estimating seismic demands and detecting weakness in vertically irregular
frames by documenting the bias and dispersion of the ratio of the seismic demands on
irregular frames determined by MPA procedure to their exact values computed by
nonlinear RHA.
The results presented in this chapter show that the effects of vertical irregularities
on seismic demands on beam-hinge model of frames are significantly different than those
reported using the less realistic column-hinge model [Al-Ali and Krawinkler, 1998]. It is
also demonstrated that the MPA procedure has a similar degree of accuracy for
estimating seismic demands for many types of irregular frames as it does for regular
frames. In addition, the MPA procedure detects which stories will be subjected to large
seismic demands; irregular frames for which MPA does not work well are identified.
6.2 EFFECT OF IRREGULARITY ON STORY-DRIFT DEMANDS
Figure 6.1 presents median story-drift demands ( NL-RHA ) determined by nonlinear RHA
for all cases and three types of irregularity with MF=2 and compares them to the
regular frame. As expected, vertical irregularity in stiffness or strength influenced the
heightwise variation of story-drift demands. For each of the eight cases, the three types
of irregularity influenced the heightwise variation of story drifts similarly, with the
effects of strength irregularity being larger than stiffness irregularity, and the effects of
combined-stiffness-and-strength irregularity being the largest among the three. This
observation agrees with conclusions of Al-Ali and Krawinkler [1998].
Introducing a soft and/or weak story (Fig. 6.1: Cases 1, 3, and 5) increases the
drift demands in the modified and neighboring stories and decreases the drift demands in
97
other stories. On the other hand, introducing a stiff and/or strong story (Fig. 6.1: Cases 2,
4, and 6) decreases the drift demands in the modified and neighboring stories and
increases the drift demands in other stories. Cases 7 and 8 will be discussed later. These
trends differ from the results reported by Al-Ali and Krawinkler [1998], where the drift
demand was affected to a greater degree, but only in the soft and/or weak story; their
column-hinge model restricts redistribution of seismic demands to adjacent stories.
To illustrate how significantly irregularity effects drift demands, Fig. 6.2 presents
the ratio of the median drift demands ( NL-RHA ) of irregular and regular frames; the
difference between this ratio and unity indicates the effect of irregularity. Essentially
independent of the location of the irregular story, the drift demand in a soft and/or weak
story (Cases 1, 3, and 5) with MF=2 increases due to stiffness (KM), strength (SM) and
combined-stiffness-and-strength (KS) irregularity by about 15%, 25%, and 40%,
respectively. Similarly, the percentage reduction in drift demand at the stiff and/or strong
story (Cases 2, 4, and 6) with MF=2 was also essentially independent of the location of
the irregular story. As shown in Fig. 6.2, however, the effect of an irregular story on drift
demands at stories further away from the irregular story is strongly dependent on its
location: It is significant when the irregular story is near the base (Cases 5 and 6) but
almost negligible when the irregular story is near the top (Cases 1 and 2).
A soft and/or weak lower half of the frame (Case 7 in Figs. 6.1 and 6.2) increases
slightly the drift demands for those stories, but significantly decreases the drift demands
in stories in upper half of the frame. In contrast, a stiff and/or strong lower half of the
frame (Case 8) reduces the drift demands for those stories but significantly increases the
drift demands in upper half of the frame. These observations imply that drift demands
occurring in the upper stories are much more sensitive to irregularity than the drift
demands in the lower stories.
As expected, Fig. 6.1 (Cases 7 and 8) shows that the drift is amplifiedrelative to
the regular framein the weaker story adjacent to the discontinuity in strength at mid-
height (irregularity types SM and KS). Note that this amplification is distributed over all
weaker stories, not the concentrated amplification in one story as predicted by the less
realistic column-hinge model [Al-Ali and Krawinkler, 1998].
98
Figure 6.3 presents median story-drift demands determined by nonlinear RHA of
the stiffness-and-strength-irregular frames for two values of modification factor (MF=2
and 5) and compares them to the median drift demands for the regular frame. As
expected, the effect of irregularity for MF=5 is much larger than for MF=2.
6.3 EFFECT OF IRREGULARITY ON FLOOR DISPLACEMENTS
Figure 6.4 compares median values of floor displacements determined by nonlinear-RHA

for three types of irregular frames with MF=2 to that of the regular frame. As long as
the irregularity is in the mid- or upper stories (Cases 1-4), all three types of irregularity
have very little influence on floor and roof displacements. In contrast, irregularity in the
base story or lower half of the building (Cases 5-8) significantly influences the
heightwise variation of floor displacements (Fig. 6.5). As observed by Al-Ali and
Krawinkler [1998], while the roof displacement is normally insensitive to vertical
irregularity (Cases 1-6); however, it is significantly different for frames with a soft and
weak lower half (Case KS7), or a stiff and strong lower half (Case KS8). Figure 6.6,
which plots the median floor displacements of stiffness-and-strength-irregular frames
with MF=2 and 5, shows that the effects of irregularity on floor displacements are even
larger when MF=5.
Although the results of Cases 5-8 in Figs. 6.4 and 6.6 may seem counter-intuitive
in that the roof displacements of irregular frames with soft and/or weak first story or
lower half are smaller than the regular frame, whereas the roof displacements of
irregular frames with stiff or/and strong first story or lower half are larger than the
regular frames, these results can be rationalized by recognizing that the irregular
frames were scaled to have to same vibration period and yield base shear as the regular
frames. Thus, a frame with soft or/and weak lower half is not softer or weaker than the
regular frame in an overall sense, but its lower half is softer and/or weaker relative to
the upper half, compared to the regular frame.
6.4 BIAS AND DISPERSION OF MPA PROCEDURE
The median story-drift demands determined by MPA on regular framesincluding 1, 2

or 3 modesare compared to the results of nonlinear RHA (Fig. 6.7a). The MPA
99
considering only the first mode is inadequate in estimating story drifts demands in
upper stories, where contributions of higher modes are known to be significant, even in
elastic systems [Chopra, 2001, Chapter 18]. With higher mode contributions included,
however, MPA follows the heightwise variation of drift demands reasonably well.
Examining the median and dispersion of the ratio *MPA indicates that the bias in MPA
(Fig. 6.7b) and its dispersion (Fig. 6.7c) decreases as higher modes are included;
however, even with 3 modes included, MPA tends to underestimate the drift demands
in all stories of this frame except the top one. This underestimation is similar to a trend
found in response spectrum analysis of elastic frames and does not disappear even if all
modes are included. In brief, MPA underestimates or overestimates the demand
depending on the height, vibration period, and design ductility (Section 5.4).
Figure 6.8 presents the median of the ratio *MPA of drift demands obtained from
MPA (including 4 modes) and from nonlinear RHA for all eight cases and three types
of irregularity with MF=2 and compares them to results for the regular frame. As
mentioned earlier, the difference between median *MPA and unity represents the bias in
the MPA procedure. These results demonstrate that the bias in the MPA procedure does
not increase, i.e., its accuracy does not deteriorate, in spite of irregularity in stiffness,
strength or both stiffness and strength, provided the irregularity is in the top story or mid-
height story (Cases 1-4).
The MPA procedure is less accurate for structures where the irregularity is in the
first story or in the lower half of the frame (Cases 5-8). The bias is significantly larger
for (i) the lower stories of an irregular frame with strong or stiff-and-strong first story
(Cases SM6 and KS6) compared to the regular frame; (ii) the upper stories of frame
with a soft, weak, or soft-and-weak lower half (Cases KM7, SM7, KS7); and (iii) the
lower stories of a frame with stiff, strong, or stiff-and-strong lower half (Cases KM8,
SM8, KS8). Note that the bias in the MPA procedure for frames with a soft, weak, or
soft-and-weak first story (Cases KM5, SM5, KS5) is about the same as found for the
regular frame.
The larger bias in the MPA estimates suggests that the MPA procedure is not
appropriate for estimating story drift demands for irregular frames corresponding to
100
Cases 6-8. To better understand the limitations of the MPA procedure, Figs. 6.9 and 6.10
compare the median story drift demands determined by the MPA including 1, 2, or 3
modes to the results of nonlinear RHA. As suggested by the results in Fig. 6.8, the
MPA procedure provides accurate drift demands for the irregular frame cases 1-4 (Fig.
6.9). It is seen to detect the concentration of drift demand in lower stories due to the
presence of a weak and/or soft first story (Fig. 6.10: Case 5) and estimates reasonably
well the drift demands in all stories. Similarly, the MPA procedure identifies the larger
drift demands in lower stories of frames with weak and/or soft lower half (Fig. 6.10: Case
7); the drift estimates are accurate enough for practical application, although they are less
accurate compared to the regular frame. Although the MPA procedure detects the
larger drifts in the upper stories of frames with a stiff and/or strong first story (Fig. 6.10:
Case 6), it may underestimate significantly the smaller drift demands in the first few
stories of the frame. Although this underestimation by the MPA procedure becomes
larger in frames with stiff and/or strong lower half (Fig. 6.10: Case 8), MPA still detects
larger drifts in the upper stories of such frames and estimates them to a useful degree of
accuracy. The overall impression that emerges is that the MPA procedure is able to
identify the stories with the largest drifts, even for Cases 5-8, and hence able to detect
critical stories in such frames with the caveat that it may underestimate significantly the
smaller drift demands in other stories.
Also presented in Figs. 6.9 and 6.10 are the drift demands computed by MPA
considering only the first mode of vibration. Clearly, even for irregular frames, the
first mode alone is inadequate in estimating the drift demands in the upper stories
where the contributions of higher modes are significant.
Figure 6.11 presents the median of the drift ratio *MPA for stiffness-and-strength-
irregular frame cases KS1-KS4, KS6, and KS8, and two values of the modification
factor, MF=2 and 5 and for the reference regular frame. Note that the bias in the MPA
procedure for irregular frames with MF=5 is not much larger than when MF=2 or the
regular frame indicating that even when this irregularity is extreme the MPA procedure
works well for frames with a soft or/and weak story or with a stiff and/or strong story at
mid-height or at the top of the frame. The bias in MPA procedure for frames with a stiff-
101
and-strong first story (Case KS6) or with stiff-and-strong lower half of the frame (Case
KS8), which was shown to be unacceptably large when MF=2 (Fig. 6.8), is even worse
when MF=5. However, the MPA procedure is able to detect the critical stories in such
frames even when the irregularity is extreme, MF=5. Frames with a soft-and-weak first
story (Case KS5) or with soft-and-weak lower half (Case KS7) are not included because
the second-mode pushover analysis is unable to reach the target displacement when
MF=5; this difficulty arises because the frame is unrealistically weak in lower stories.
This difficulty in MPA procedure is documented in Appendix B. .
Figure 6.12 presents the dispersion of *MPA for all 8 cases and three types of
irregularity with MF=2 and compares them to the regular frame. The dispersion of
*MPA for irregular frames is similar to that found for the regular frame, except that it is
much larger for the lower stories of Cases 6 and 8 of strength- and stiffness-and-strength-
irregular frames, for which the bias in MPA is also large (Fig. 6.8). Except for these two
cases, the MPA procedure should be similarly reliable in estimating the seismic demands
of irregular and regular frames due to an individual ground motion.
Figure 6.13 shows the dispersion of *MPA of the combined-stiffness-and-strength

irregular frame Cases KS1-KS4 for two values of the modification factor (MF=2 and 5)
and for the regular frame. The dispersion for MF=5 is roughly similar to that for M=2
in Cases KS2-4, but larger in Case KS1, implying that the MPA procedure is similarly
reliable in estimating the seismic demands due to an individual ground motion even for
highly irregular frames, provided that the irregular story is at mid-height or upper part of
the frame.
The bias and dispersion of MPA in estimating floor displacements of irregular

frames are plotted in Figs. 6.14-6.17. These results show that the bias increases, i.e., the
accuracy deteriorates for frames with strength or stiffness-and-strength irregularity in the
first story or in the lower half (Cases SM5-SM8 and KS5-KS8); this is similar to earlier
observation from the results for story-drift demands. The MPA procedure is significantly
biased in estimating floor displacements in lower stories of irregular frames with strong-
or stiff-and-strong first story or lower half (Cases SM6, SM8, KS6, and KS8). Note that
the roof displacement can still be estimated accurately in all irregular frames considered,
102
even when MF=5; the bias is always less than 16% and the dispersion is less than 0.22.
This suggests that the large bias is not due to errors in estimating roof displacements.
103
Case 1: Soft or/and weak top story Case 2: Stiff or/and strong top story
12 12
Regular
10 KM 10
8 SM 8
Story
KS
6 6
4 4
2 2
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Case 3: Soft or/and weak midheight story Case 4: Stiff or/and strong midheight story
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
st st
Case 5: Soft or/and weak 1 story Case 6: Stiff or/and strong 1 story
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Case 7: Soft or/and weak lowerhalf stories Case 8: Stiff or/and strong lowerhalf stories
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Story drift, NLRHA (%) Story drift, NLRHA (%)
Figure 6.1 Median story-drift demands determined by nonlinear RHA for regular
frame and stiffness-, strength-, and combined-stiffness-and-strength-irregular
frames denoted by KM, SM, and KS, respectively, with modification factor,
MF=2.
104
12 12
10 10
8 8
Story
6 6
4 KM 4
SM
2 2
KS
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
st
Case 5: Soft or/and weak 1 story Case 6: Stiff or/and strong 1st story
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(NLRHA)Irregular / (NLRHA)Regular (NLRHA)Irregular / (NLRHA)Regular
Figure 6.2 Ratio of the median story-drift demands of regular and irregular frames.
Three types of irregularitystiffness-, strength-, and combined-stiffness-and-
strengthare included with modification factor, MF=2.
105
Case KS1: Softandweak top story Case KS2: Stiffandstrong top story
12 12
10 10
8 8
6 6
4 Regular 4
MF=2
2 2
MF=5
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Case KS3: Softandweak midheight story Case KS4: Stiffandstrong midheight story
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
st st
Case KS5: Softandweak 1 story Case KS6: Stiffandstrong 1 story
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Case KS7: Softandweak lowerhalf stories Case KS8: Stiffandstrong lowerhalf stories
12 12
10 10
8 8
6 6
4 4
2 2
G G
0 0.5 1 1.5 2 0 0.5 1 1.5 2
Story drift, (%) Story drift, (%)
NLRHA NLRHA
Figure 6.3 Median story-drift demands determined by nonlinear RHA for regular
frame and combined-stiffness-and-strength-irregular frames with
106
12 12
Regular
10 KM 10
8 SM 8
Story
KS
6 6
4 4
2 2
G G
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Floor displacement/Total height (%) Floor displacement/Total height (%)
Figure 6.4 Median floor displacements determined by nonlinear RHA for regular
frame and stiffness-, strength-, and combined-stiffness-and-strength-irregular
frames denoted by KM, SM, and KS, respectively, with modification factor,
MF=2.
107
12 12
10 10
8 8
Story
6 6
4 KM 4
SM
2 2
KS
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
(uNLRHA)Irregular / (uNLRHA)Regular (uNLRHA)Irregular / (uNLRHA)Regular
Figure 6.5 Ratio of the median floor displacements of regular and irregular frames.
Three types of irregularitystiffness-, strength-, and combined-stiffness-and-
strengthare included with modification factor, MF=2.
108
12 12
10 10
8 8
6 6
4 Regular 4
MF=2
2 2
MF=5
G G
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
st st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
12 12
10 10
8 8
6 6
4 4
2 2
G G
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Floor displacement/Total height (%) Floor displacement/Total height (%)
Figure 6.6 Median floor displacements determined by nonlinear RHA for regular
frame and combined-stiffness-and-strength-irregular frames with
109
(a) (b) (c)
12 12 12
10 NLRHA 10 10
8 MPA 8 8
1 "mode"
Story
6 2 "modes" 6 6
3 "modes"
4 4 4
2 2 2
0 0 0
0 0.5 1 1.5 0 0.5 1 1.5 2 0 0.2 0.4 0.6
Story drift (%) Storydrift ratio * Dispersion of *
MPA MPA
Figure 6.7 (a) Median story-drift demands determined by MPA and by nonlinear RHA,
(b) median drift ratio *MPA , and (c) dispersion of *MPA for the reference 12-
story regular frame including 1, 2, and 3 modes.
110
12 12
10 10
8 8
Story
6 6
Regular
4 KM 4
SM
2 2
KS
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
st st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Storydrift ratio, *MPA Storydrift ratio, *MPA
Figure 6.8 Median story-drift ratios *MPA for regular frame and stiffness-, strength-,
and combined-stiffness-and-strength-irregular frames with modification
factor, MF=2.
111
KM1 SM1 KS1
12
10
Story 8
6 NLRHA
MPA
4 1 "mode"
2 "modes"
2
3 "modes"
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
KM2 SM2 KS2
12
10
8
Story
6
4
2
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
KM3 SM3 KS3
12
10
8
Story
6
4
2
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
KM4 SM4 KS4
12
10
8
Story
6
4
2
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
Figure 6.9 Median story-drift demands of stiffness-, strength-, and stiffness-and-strength-

irregular frame cases 1-4 with modification factor, MF=2, determined by
MPA including 1, 2, and 3 modes and by nonlinear RHA.
112
KM5 SM5 KS5
12
NLRHA
10 MPA
Story 8 1 "mode"
2 "modes"
6 3 "modes"
4
2
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
KM6 SM6 KS6
12
10
8
Story
6
4
2
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
KM7 SM7 KS7
12
10
8
Story
6
4
2
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
KM8 SM8 KS8
12
10
8
Story
6
4
2
G
0 0.5 1 1.5 2 0 0.5 1 1.5 2 0 0.5 1 1.5 2
Figure 6.10 Median story-drift demands of stiffness-, strength-, and stiffness-and-strength-

irregular frame cases 5-8 with modification factor, MF=2, determined by
MPA including 1, 2, and 3 modes and by nonlinear RHA.
113
12 12
10 10
8 8
Story
6 6
4 Regular 4
MF=2
2 2
MF=5
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Storydrift ratio, * st
Case KS6: Stiffandstrong 1 story
MPA
12
10
8
6
4
2
G
0.4 0.6 0.8 1 1.2 1.4 1.6
Case KS8: Stiffandstrong lowerhalf stories
12
10
8
6
4
2
G
0.4 0.6 0.8 1 1.2 1.4 1.6
Storydrift ratio, *
MPA
Figure 6.11 Median story-drift ratios *MPA for regular frame and combined-stiffness-
and-strength-irregular frames (Cases KS1-KS4, KS6, and KS8) with
114
12 12
10 10
8 8
Story
6 6
Regular
4 KM 4
SM
2 2
KS
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
st st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of *MPA Dispersion of *MPA
Figure 6.12 Dispersion of story-drift ratios *MPA for regular frame and stiffness-,
strength-, and combined-stiffness-and-strength-irregular frames with
modification factor, MF=2.
115
12 12
10 10
8 8
Story
6 6
4 Regular 4
MF=2
2 2
MF=5
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of * st
Case KS6: Stiffandstrong 1 story
MPA
12
10
8
6
4
2
G
0 0.2 0.4 0.6
12
10
8
6
4
2
G
0 0.2 0.4 0.6
Dispersion of *
MPA
Figure 6.13 Dispersion of story-drift ratios *MPA for regular frame and combined-
stiffness-and-strength-irregular frames (Cases KS1-KS4, KS6, and KS8) with
116
12 12
10 10
8 8
Story
6 6
Regular
4 KM 4
SM
2 2
KS
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
st st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Floordisplacement ratio, u*MPA Floordisplacement ratio, u*MPA
*
Figure 6.14 Median floor-displacement ratios uMPA for regular frame and stiffness-,
strength-, and combined-stiffness-and-strength-irregular frames with
modification factor, MF=2.
117
12 12
10 10
8 8
Story
6 6
4 Regular 4
MF=2
2 2
MF=5
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Floordisplacement ratio, u* st
MPA Case KS6: Stiffandstrong 1 story
12
10
8
6
4
2
G
0.4 0.6 0.8 1 1.2 1.4 1.6
12
10
8
6
4
2
G
0.4 0.6 0.8 1 1.2 1.4 1.6
Floordisplacement ratio, u*MPA
*
Figure 6.15 Median floor-displacement ratios uMPA for regular frame and combined-
stiffness-and-strength-irregular frames (Cases KS1-KS4, KS6, and KS8) with
118
12 12
10 10
8 8
Story
6 6
Regular
4 KM 4
SM
2 2
KS
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
st st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of u*MPA Dispersion of u*MPA
*
Figure 6.16 Dispersion of floor-displacement ratios, uMPA , for regular frame and
stiffness-, strength-, and combined-stiffness-and-strength-irregular frames
with modification factor, MF=2.
119
12 12
10 10
8 8
Story
6 6
4 Regular 4
MF=2
2 2
MF=5
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Dispersion of u* st
MPA Case KS6: Stiffandstrong 1 story
12
10
8
6
4
2
G
0 0.2 0.4 0.6
12
10
8
6
4
2
G
0 0.2 0.4 0.6
Dispersion of u*MPA
*
Figure 6.17 Dispersion of floor-displacement ratios uMPA for regular frame and
combined-stiffness-and-strength-irregular frames (Cases KS1-KS4, KS6, and
KS8) with modification factors, MF=2 and 5.
120
7. COMPARATIVE EVALUATION OF MPA AND PUSHOVER
ANALYSIS USING FEMA FORCE DISTRIBUTIONS
7.1 INTRODUCTION
The accuracy of the nonlinear static procedure (NSP) using the lateral force distributions
specified in the FEMA-356 document [BSSC, 2000], now standard in engineering
practice, and the MPA procedure is evaluated in this chapter using both regular and
irregular frames. The dynamic response of each structural system to each of the 20
LMSR ground motions was determined by nonlinear RHA, MPA, and pushover analysis
using four FEMA-356 force distributions, described in detail in BSSC [2000] and
summarized in the Section 7.2. In this comparative evaluation of analysis procedures, the
target roof displacement for FEMA analysis is not determined by FEMA method, but
taken equal to the roof displacement determined by MPA for each ground motion to
eliminate the discrepancy in how the target roof displacement is determined.
7.2 FEMA FORCE DISTRIBUTIONS
FEMA-356 requires that at least two force distributions be considered in the nonlinear
static procedure (NSP). The first is selected from one of the following:
(1) Fundamental mode (1st mode) distribution, which is s1* [Eq. (2.20)]
(2) Equivalent Lateral Force (ELF) distribution, which is the same as the IBC lateral
force [Eq. (3.2)]
(3) SRSS distribution, which is defined by the lateral forces back-calculated from the
story shears determined by RSA (Section 2.2.2) including sufficient number of
modes to capture at least 90% of the total mass.
The second distribution is either the Uniform distribution, where story force is
proportional to story mass, or an Adaptive distribution [Bracci et al., 1997; Fajfar and
Fischinger, 1988; Gupta and Kunnath, 2000]. All four FEMA force distributions
excluding the Adaptive distribution will be considered in the comparative evaluation in
this chapter. Fig. 7.1 shows the four FEMA force distributions for a 9-story regular
frame with T1 = TU .
121
7.3 REGULAR FRAMES
Figure 7.2 shows the median values of story drift demands, NL-RHA , FEMA , and MPA
determined by nonlinear RHA, pushover analysis using FEMA-force distributions, and
the MPA procedure, respectivelyfor six3, 6, 9, 12, 15 and 18-storyregular
frames designed for a SDF-system ductility factor = 4 and with T1 = TU . Results are
presented for each of the four FEMA-356 lateral force distributions up to the target
displacement determined by MPA, and the range of drift demands has been shaded. Note
that because of the way these frames were designed (see Section 3.2.1), the drift demand
produced by the ELF distribution is uniform over the height of all regular frames.
These results indicate that the force distribution that gives the largest drift demand
depends on the story location along the building height. The uniform force distribution
always leads to the largest drifts in the lower stories; the SRSS distribution gives largest
drifts in the upper stories, and the ELF distribution in the middle stories. The first mode
force distribution does not lead to largest drift in any story of taller buildings because of
significant higher mode contributions to response that are ignored.
A comparison between FEMA and NL-RHA in Fig. 7.2 demonstrates the limitation
of the approximate procedure. None of the FEMA distributions leads to drifts that are
close to the exact results from nonlinear RHA; however, the envelope of the four sets
of FEMA results provides better results, which overestimates the exact demands in
some cases but underestimates them in others. Comparing MPA with FEMA and NL-RHA
demonstrates that MPA provides story drifts that are close to their exact values and are
much more accurate compared to the FEMA force distributions, individually or their
envelope value (see Fig. 7.2).
To clearly and comprehensively illustrate the errors in approximate procedures

FEMA and MPAthe median of story drift ratios *FEMA and *MPA are presented in Fig.
7.3 for all thirty regular frames with T1 = TU . The FEMA force distributions lead to
gross underestimation of story drifts. These may be as small as one-tenth or one-half of
that predicted by nonlinear RHA in upper and lower stories, respectively.
122
While the bias in the MPA procedure tends to increase for longer period frames
and larger SDF-system ductility factors, these trends are not perfect, as discussed in
Section 5.4. Figure 7.3 demonstrates that MPA is almost always more accurate in
estimating drifts in all stories of all frames than all of the FEMA force distributions. The
bias in MPA is generally well below 25%, but for rare exceptions, e.g., upper story drifts
in 15- and 18- story frames with =4 and 6. Note that in these cases the FEMA results
are even more biased. The individual FEMA force distributions are systematically biased
in estimating the story drifts. The uniform distribution consistently overestimates the
demand in lower stories but underestimates it in upper stories; the ELF distribution
consistently underestimates the demand in stories near the top and near the base; the
SRSS distribution consistently underestimates demands in lower two-thirds of the height.
In contrast, MPA overestimates the demand in lower stories of some frames but
underestimates it for others, depending on the fundamental period (or height) of the frame
and its design ductility factor (Fig. 7.3). Similarly, the demand in upper stories may be
underestimated for some frames but overestimated for others. The envelope of the four
estimates corresponding to the four FEMA force distributions is less biased than the
individual estimates, but is almost always more biased than MPA.
Figure 7.4 shows the dispersion of the drift ratio *MPA and *FEMA plotted over the
height of the same 30 regular frames. Both dispersions are smallest in 3-story frames
and tend to increase for taller frames, especially in their upper stories where higher mode
contributions are more significant in seismic demands. The FEMA dispersion is
significantly worse (larger) than the MPA dispersion in almost all stories of all frames; it
can be unacceptably large in upper stories of even short frames for which pushover
analyses are believed to be effective, e.g., 6-story frames.
An approximate procedure is more likely to be accurate in estimating the seismic

demands due to an individual excitation in cases where both its bias and dispersion are
small (Section 5.4); therefore, the MPA procedure with its smaller bias and dispersion is
more reliable than FEMA force distributions. Although the FEMA dispersion in upper
stories of some tall frames determined by pushover analysis using the uniform force
123
distribution is smaller than the MPA dispersion, the FEMA estimate is not necessarily
reliable because its bias was observed to be very large (Fig. 7.3).
7.4 IRREGULAR FRAMES
The FEMA and MPA procedures applied to irregular frames are compared and evaluated
in Figs. 7.5 and 7.6. Figure 7.5 shows the median values of story drift demands NL-RHA ,
FEMA , and MPA for eight stiffness-and-strength-irregular frames with MF=2. Figure 7.6
presents the median values of story drift ratios *FEMA and *MPA for all 24 irregular
frames with MF=2. As before, results are presented for each of the four FEMA-356
lateral force distributions and the range of data has been shaded. These results lead to the
following observations:
As for found for regular frames, none of the FEMA force distribution leads to
story drifts that follow the height-wise variation of exact results, and the individual
FEMA force distributions are systematically biased in estimating the story drifts. In all
cases the uniform distribution underestimates the drift demand in upper stories, but
overestimates it in the lower stories; the ELF distribution underestimates demands in
stories near the top and near the base; and the SRSS distribution underestimates demands
in lower two-thirds of the height. Individual FEMA results and the envelope of the four
sets of FEMA results are generally less accurate for many irregular frames compared to
the regular 12-story frame (Figs. 7.2 and 7.3). Despite being an envelope, it significantly
underestimates the drift demands in upper stories of several irregular frames (Cases SM2,
SM5, SM7, KS2, KS5, and KS7); however, it does significantly overestimate the demand
in lower stories of other frames (Cases KM1, KM4, KM6, KM8, SM1, SM4, SM5, KS1,
and KS4).
The MPA procedure maintains its superiority for almost all of the irregular
frames. The MPA estimate is almost always more accurate compared to all of the FEMA
force distributions and to the envelope of the four FEMA estimates. The only exceptions
are the lower stories of irregular frames with a strong and stiff-and-strong first story
(Cases SM6 and KS6) or lower half (Cases SM8 and KS8), where the uniform
distribution provides an exceptionally accurate estimate of demand, but is worse than
124
MPA in the upper stories; furthermore the other FEMA distributions provide results that
are worse than MPA near the top of the frame. Thus, pushover procedures seem
incapable of providing accurate estimates of median seismic demands for such systems,
and nonlinear RHA seems necessary.
Figure 7.7 shows the dispersion of the drift ratio *MPA and *FEMA plotted over the
height of the 24 irregular frames. Overall, these results give the impression that the
dispersion in the MPA estimate is generally smaller than for the four FEMA estimates,
although counter examples can be observed in individual stories and force distributions.
This observation combined with the earlier observation that MPA is less biased
demonstrates that the MPA procedure will provide more reliable estimates of seismic
demands due to individual excitations compared to FEMA force distributions. However,
the uniform force distribution in FEMA leads to impressively smaller dispersion for
drifts in lower stories of irregular frames with strong and stiff-and-strong lower half
(Cases SM8 and KS8), but is worse than MPA near the top of the frame. As noted earlier
pushover analysis procedures are not especially effective for such systems, which should
be analyzed by nonlinear RHA.
125
0.191 0.260 0.326 0.111
0.173 0.213 0.183 0.111
0.154 0.170 0.087 0.111
0.134 0.130 0.044 0.111
0.114 0.096 0.050 0.111
0.093 0.065 0.079 0.111
0.070 0.040 0.096 0.111
0.048 0.020 0.086 0.111
0.024 0.006 0.049 0.111
(a) 1st Mode (b) ELF (c) SRSS (d) Uniform
Figure 7.1 FEMA-356 force distributions for a 9-story regular frame with T1 = TU : (a)
1st Mode, (b) Equivalent lateral force (ELF), (c) SRSS, and (d) Uniform
distributions.
126
3 6
2 4
Story
NLRHA
MPA 3
1 FEMA 2
1st Mode
ELF
SRSS 1
Uniform
G G
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Story drift, NLRHA, MPA, or FEMA (%) Story drift, NLRHA, MPA, or FEMA (%)
Figure 7.2 Median story-drift demands determined by nonlinear RHA, MPA, and
pushover analysis using four FEMA force distributions for regular frames
with =4 and T1 = TU .
127
=1 =1.5 =2 =4 =6
3
MPA
Story 2 FEMA
1st Mode
ELF
1 SRSS
Uniform
4
Story
6
Story
12
9
Story
15
12
Story
9
6
3
18
15
12
Story
9
6
3
G
0 1 2 1 2 1 2 1 2 1 2
*MPA or *FEMA
Figure 7.3 Median story-drift ratios *MPA and *FEMA for thirty regular frames with
T1 = TU : 3, 6, 9, 12, 15, and 18-story in combination with =1, 1.5, 2, 4, and 6,

all.
128
=1 =1.5 =2 =4 =6
3
MPA
Story 2 FEMA
1st Mode
ELF
1 SRSS
Uniform
4
Story
6
Story
12
9
Story
15
12
Story
9
6
3
18
15
12
Story
9
6
3
G
0 0.2 0.4 0.6 0.2 0.4 0.6 0.2 0.4 0.6 0.2 0.4 0.6 0.2 0.4 0.6
Dispersion of *MPA or *FEMA
Figure 7.4 Dispersion of story-drift ratios *MPA and *FEMA for thirty regular frames
with T1 = TU : 3, 6, 9, 12, 15, and 18-story in combination with =1, 1.5, 2, 4,
and 6.
129
12 12
NLRHA
10 10 MPA
8 8 FEMA
Story
1st Mode
6 6 ELF
SRSS
4 4 Uniform
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
st st
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Story drift, NLRHA, MPA, or FEMA (%) Story drift, NLRHA, MPA, or FEMA (%)
Figure 7.5 Median story-drift demands determined by nonlinear RHA, MPA, and
pushover analysis using four FEMA force distributions for combined-
stiffness-and-strength-irregular frames with MF=2.
130
12
KM1 KM2 KM3 KM4
9 MPA
Story 6 FEMA
1st Mode
ELF
3 SRSS
Uniform
12
SM1 SM2 SM3 SM4
9
Story
12
KS1 KS2 KS3 KS4
9
Story
12
KM5 KM6 KM7 KM8
9
Story
12
SM5 SM6 SM7 SM8
9
Story
12
KS5 KS6 KS7 KS8
9
Story
G
0 1 2 1 2 1 2 1 2
*MPA or *FEMA *MPA or *FEMA *MPA or *FEMA *MPA or *FEMA
Figure 7.6 Median story-drift ratios *MPA and *FEMA for all irregular frames
considered: Cases 1-8 of stiffness-, strength-, and combined-stiffness-and-
strength-irregular frames with MF=2.
131
12
KM1 KM2 KM3 KM4
9 MPA
Story 6 FEMA
1st Mode
ELF
3 SRSS
Uniform
12
SM1 SM2 SM3 SM4
9
Story
12
KS1 KS2 KS3 KS4
9
Story
12
KM5 KM6 KM7 KM8
9
Story
12
SM5 SM6 SM7 SM8
9
Story
12
KS5 KS6 KS7 KS8
9
Story
G
0 0.2 0.4 0.6 0.2 0.4 0.6 0.2 0.4 0.6 0.2 0.4 0.6
Dispersion of * or *
MPA FEMA
Figure 7.7 Dispersion of story-drift ratios *MPA and *FEMA for all irregular frames
considered: Cases 1-8 of stiffness-, strength-, and combined-stiffness-and-
strength-irregular frames with MF=2.
132
8. PRACTICAL IMPLEMENTATION OF MPA
8.1 INTRODUCTION
The MPA procedure described in Section 2.3.3, referred to earlier as Procedure A, and
evaluated in Chapters 5 and 6, is implemented as if the earthquake hazard is given in
terms of ground motion records. The target roof displacement is determined from the
response history analysis of inelastic SDF system due to each ground motions and the
accuracy of MPA Procedure A was evaluated based on the median and dispersion of
*
response ratio rMPA calculated for each record. In this chapter, the MPA procedure is
simplified to facilitate its implementation in engineering practicewhere the earthquake
hazard is defined in term of a smooth design spectrum corresponding to a selected
exceedence probabilityand the accuracy of this simplified procedure is documented.
The MPA procedure can be simplified in two stages, resulting in Procedures B and C:
8.2 MPA PROCEDURE B
In the MPA Procedure A, the seismic demand due to each (say, ith) ground motion is
determined by calculating ( Dn )i , (urno )i , (rno )i , and (rMPA )i , and then the median of
(rMPA )i (i=1,2,3) gives rMPA . The first simplification estimates the median value of
modal seismic demands rno directly from the deformation D n of the nth-mode
inelastic SDF system, which was determined from the median spectrum for the ensemble
of ground motions. The resulting Procedure B consists of the first five steps of Procedure
A plus the following steps:
6. Repeat step 5 for all excitations and obtain ( Dn )i for each excitation.
7. Calculate D n , the median value of ( Dn )i , by Eq. [3.4(a)].
8. Calculate the median peak roof displacement urno associated with the nth-mode
inelastic SDF system from
urno = nrn D n (8.1)
133
9. Extract other desired responses, rno , from the pushover database values at roof
displacement urno .
10. Repeat Steps 3 to 9 for as many modes as required for sufficient accuracy;
usually the first two or three modes will suffice.
11. Determine the total response rMPA by combining the peak modal responses rno
using appropriate modal combination rule, e.g., the SRSS combination rule:
1/ 2
J
rMPA = rno2 (8.2)
n =1
Figure 8.1 compares the results obtained from Procedures A and B. The floor
displacements and story drifts of an 18-story frame determined by Procedure A due to
each of the 20 excitations together with their median value are compared with the median
demand determined by Procedure B. When only one mode is included, the two versions
of the MPA procedure give identical results for the median seismic demand for elastic
frames (Figs. 8.1a and 8.1b) because the peak value (rno )i of any response quantity r due
to the ith ground motion is related linearly to ( Dn )i :
( rno )i = rnstn2 ( Dn )i (8.3)
where rnst is the nth-modal static response [Chopra, 2001: Section 18.3]. Equation
[3.4(a)] gives the median response as
r1o = r1st12 D1 (8.4)
where n=1 for one-mode response. The left-hand side of Eq. (8.4) is the median value of
demand as determined by Procedure A, whereas the right-hand side of Eq. (8.4) is the
demand corresponding to D1 as determined by Procedure B. However, Both Procedures
A and B give different results for inelastic systems because ( rno )i , determined by
nonlinear pushover analysis, is not related linearly to the ( Dn )i after the structure yields,
which is confirmed by the results for the 18-story frame with = 4 (see Figs. 8.1c and
8.1d).
134
When several modes are included, the median demand from Procedure B is no
longer identical to the median demand from Procedure A, even for elastic frames (see
Figs. 8.2a and 8.2b) because the median peak value of total response is computed by
applying Eq. [3.4(a)] to the total [after SRSS modal combination in Eq. (2.18)] responses
to individual excitations in Procedure A, but by combining the median values of peak
modal responses in Procedure B [Eq. (8.2)], and the median and SRSS operations do
not commute. As expected, the discrepancy between the two procedures is larger for
inelastic systems (compare Figs. 8.2c and 8.2d to Figs. 8.2a and 8.2b).
8.3 MPA PROCEDURE C
Procedure B requires nonlinear RHA of the nth-mode inelastic SDF system (step 5) for
each ground motion. Procedure C avoids this computation by determining D n from the
median deformation spectrum for inelastic SDF systems for constant yield-strength-
reduction-factor Ry [Chopra, 2001: Section 8.4]. Steps 5-7 in Procedure B to determine
D n are replaced by the following steps:
5. Compute the yield strength reduction factor Ryn for the nth-mode inelastic SDF
system from
D n ,elastic
Ryn = (8.5)
Dny
where D n ,elastic is the spectral ordinate of the median elastic deformation spectrum
at period Tn; Dny is the yield deformation of the nth-mode inelastic SDF system
obtained in Step 4.
6. Compute the peak deformation of the nth-mode inelastic SDF system with
Ry = Ryn for every ground motion and determine the median deformation, D n . A
plot of D n against Tn is the median deformation spectrum for Ry = Ryn and
damping ratio n .
7. Obtain D n from this median deformation spectrum at period Tn .
135
This additional approximation in Procedure C is examined in Fig. 8.3a, which presents
deformation response spectra ( Dn )i for the ith ground motion, i=1 to 20, for an inelastic
SDF system with yield deformation Dny = D n ,elastic 5. [In the MPA procedure, recall that
Dny is determined from the pushover curve (step 4) and ( Dn )i are used to calculate
(urno )i in Eq. (2.21) of Procedure A.] For all excitations the selected Dny is the same
because it is a property of the nth-mode SDF system (see Fig. 2.5). Constant Dny
implies that the strength reduction factor ( Ryn )i = ( Dn ,elastic )i / Dny varies with the
excitation because the elastic deformation demand ( Dn ,elastic )i is excitation-dependent.
(However, the median value of Ryn will be equal to 5 for this choice of Dny .) Also
shown in Fig. 8.3a is the median deformation D n spectrum, which is used in Eq. (8.1) of
Procedure B to calculate urno . In Procedure C, D n is computed for an inelastic SDF
system with Ryn fixed for all ground motions, not with a fixed Dny , by determining the
median of the constant- Ry spectra for individual excitations. Figure 8.3b compares such
a spectrum with the median spectrum from Fig. 8.3a, demonstrating that D n can be
estimated from the constant- Ry deformation spectrum with errors less than 10%.
8.4 COMPARATIVE EVALUATION OF THREE VERSIONS OF THE MPA

PROCEDURE
The median drift ratios *MPA determined by the MPA Procedures A, B, and C are
presented here. In Procedure A, the median drift ratio is computed from the drift ratios
( )
*
MPA i determined for individual excitations (i=1, 220). This result is identical to the
ratio of the two values of median demand MPA and NL-RHA , determined by MPA and
nonlinear-RHA procedures, respectively, because the geometric mean is chosen to
estimate the median value. Only the latter calculation is possible in Procedures B and C
because they provide only the median demand; no information is available for the
demand due to individual excitations.
136
To compare the bias in these approximate procedures, the median drift ratios
*MPA are presented for elastic and inelastic 12-story frames with =1, 1.5, 2, 4, and 6
(see Fig. 8.4) and for 3, 6, 9, 12, 15, and 18-story frames with =4 (see Fig. 8.5).
Comparing the median drift ratio obtained from Procedures A and B against 1.0 shows
that the bias due to the additional approximation in Procedure B can be as much as 20%-
30% in addition to the bias already inherent in Procedure A. The additional bias in
Procedure B increases for taller frames (Fig. 8.5) and for larger (Fig. 8.4).
Comparing results obtained from Procedures B and C indicates that the two
procedures provide identical results for elastic systems (Fig. 8.4a). The two sets of results
are similar for inelastic systems over the entire range of building height (Fig. 8.5) and
SDF-system ductility factor (Fig. 8.4), implying that the two procedures are biased
similarly. Therefore, the median deformation for each modal inelastic SDF system can
be estimated with reasonable accuracy from the constant- Ry deformation spectrum. Note
that the discrepancy between the two procedures is related to the difference between the
exact value of D n and its approximate value from the constant- Ry spectrum (Fig. 8.3). If
this difference for the combination of T1 and considered is large, seismic demands
computed using these two procedures could differ significantly even for shorter
buildings.
To summarize the bias in the practical version (Procedure C) of MPA procedure,

median drift ratio *MPA for all 60 regular frames ( T1 = TL and T1 = TU ) are plotted over
the height in Figs. 8.6 and 8.7. It is seen to be only slightly larger than the bias in
Procedure A (Figs. 5.9 and 5.10), thus demonstrating that this simplified version of MPA
still provides estimates of story drift demands accurate enough for practical applications.
8.5 ESTIMATION OF SDF-SYSTEM DEFORMATION
Procedure C in the MPA requires determining the median value D n of the peak
deformation of the nth-mode inelastic SDF system. D n may be computed from the
constant- Ry deformation spectrum, as was implemented in developing the results
137
presented in Figs. 8.6 and 8.7. However, the seismic hazard is usually defined by a
median (or some other percentile) design spectrum for elastic systems, which provides
the median deformation D n ,elastic of the corresponding elastic system. The median
deformation D n of the inelastic system can then be estimated from
D n = CR D n ,elastic (8.6)
where the ratio CR of peak deformations of inelastic and corresponding elastic system
depends on period Tn , yield-strength reduction factor Ry , post-yield stiffness ratio n ,
and damping ratio, . This deformation ratio is comprehensively investigated in the next
chapter; a simple equation of CR for conveniently estimating deformation of inelastic
SDF system is also developed.
138
(a) (b)
18 18
16 Elastic 16 Elastic
14 14
12 12
Story
10 10
8 8
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 0 0.5 1 1.5
Floor displacement/Total height (%) Story drift (%)
(c) (d)
18 18
16 =4 16 =4
14 14
12 12
Story
10 10
8 8
6 Procedure A: 6
4 Individual Excitations 4
Median
2 Procedure B 2
G G
0 0.2 0.4 0.6 0.8 0 0.5 1 1.5
Figure 8.1 Floor displacement and story drift demands for two 18-story frames (elastic
frame and inelastic frame with =4) determined by versions A and B of the
MPA procedure including only one vibration mode.
139
(a) (b)
18 18
16 Elastic 16 Elastic
14 14
12 12
Story
10 10
8 8
6 6
4 4
2 2
G G
0 0.2 0.4 0.6 0.8 0 0.5 1 1.5 2
(c) (d)
18 18
16 =4 16 =4
14 14
12 12
Story
10 10
8 8
6 Procedure A: 6
4 Individual Excitations 4
Median
2 Procedure B 2
G G
0 0.2 0.4 0.6 0.8 0 0.5 1 1.5
Figure 8.2 Floor displacement and story drift demands for two 18-story frames (elastic
frame and inelastic frame with =4) determined by versions A and B of the
MPA procedure including five vibration modes.
140
(a)
40
^
Median D
n
35 Individual (Dn)i
30
25
Dn (cm)
20
15
10
0
0 0.5 1 1.5 2 2.5 3 3.5 4
(b)
12
Procedure B
10
Procedure C
8
Median Dn (cm)
^
0
0 0.5 1 1.5 2 2.5 3 3.5 4
Natural vibration period, T (sec)
n
Figure 8.3 (a) Deformation spectra of individual LMSR ground motions and their
median spectrum for inelastic SDF systems with yield deformation
Dny = D n ,elastic 5. (b) Comparison of median D n spectrum used in Procedures
B and C. All spectra are for n =5% and strain-hardening ratio n =3.72%.
141
(a) Elastic (b) =1
12 12
Procedure A
10 Procedure B 10
Procedure C
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
(c) =1.5 (d) =2

12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
(e) =4 (f) =6
12 12
10 10
8 8
Story
6 6
4 4
2 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Figure 8.4 Story-drift ratio *MPA determined by versions A, B, and C of the MPA
procedure for inelastic 12-story frames (with T1 = TU ) designed for =1, 1.5,
2, 4, and 6. Results for linearly elastic frames are also included.
142
3 6
Procedure A
Procedure B 5
Procedure C
2 4
Story
1 2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Figure 8.5 Story-drift ratio *MPA determined by versions A, B, and C of the MPA
procedure for inelastic 3, 6, 9, 12, 15, and 18-story frames (with T1 = TU )
designed for =4.
143
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Figure 8.6 Story-drift ratio *MPA determined by version C of the MPA procedure for
inelastic 3, 6, 9, 12, 15, and 18-story frames (with T1 = TL ) designed for =1,
1.5, 2, 4, and 6. Results for linearly elastic frames are also included.
144
3 6
2 4
Story
Elastic 3
= 1
1 1.5 2
2
4 1
6
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

9 12
8
10
7
6 8
Story
5
6
4
3 4
2
2
1
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6

15 18
16
12 14
12
9
Story
10
8
6
6
3 4
2
G G
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
Figure 8.7 Story-drift ratio *MPA determined by version C of the MPA procedure for
inelastic 3, 6, 9, 12, 15, and 18-story frames (with T1 = TU ) designed for =1,
1.5, 2, 4, and 6. Results for linearly elastic frames are also included.
145
9. INELASTIC DEFORMATION RATIOS FOR SDF BILINEAR
SYSTEMS
9.1 INTRODUCTION
The target roof displacement in all of the pushover procedures is determined from the
peak deformation of an inelastic single-degree-of-freedom (SDF) system with its force-
deformation relation defined from the pushover curve. This has led to renewed interest in
the relationship between the peak deformations of inelastic and corresponding linear SDF
systems, um and uo , respectively, a problem first studied by Veletsos and Newmark
[1960]. Following convention, um and uo notations will be adopted in this chapter
instead of Dn and Dn ,elastic in Chapter 8.
The inelastic deformation ratio, if expressed as a function of elastic vibration

period Tn and ductility factor , can be used to determine the inelastic deformation of a
new or rehabilitated structure where global ductility capacity can be estimated. Veletsos
and Newmark [1960] proposed and evaluated two approaches: equal-deformation rule,
i.e., um / uo =1; and equating the strain energy for elastic and inelastic systems, leading to
um / uo = / 2 1 . This ratio was shown to differ significantly in different spectral (or

frequency) regions, with the equal deformation rule being valid only in the displacement-
sensitive spectral region [Veletsos et al., 1965]. The equal deformation rule was also
recommended for the velocity-sensitive spectral region [Veletsos and Vann, 1971].
With the accumulation of strong motion records and ease of computation, it

became possible to investigate the statistics of the inelastic deformation ratio over large
ensembles of ground motions. Response data for 124 excitations and bilinear systems
with 3% post-yield stiffness ratio demonstrated that the inelastic deformation ratio was
strongly dependent on the elastic vibration period in the acceleration-sensitive spectral
region and that the period beyond which the equal-deformation rule is valid depends on
the site conditionsrock, alluvium, or soft soiland on the ductility factor [Miranda,
1991 and 1993]. A more recent investigation based on 264 records concluded that the
earthquake magnitude and rupture distance have little influence on the inelastic
deformation ratio [Miranda, 2000]; however, at short periods (0.1 to 1.3 sec), this ratio
146
for fault-normal near-fault records with forward directivity effects was shown to be much
larger compared to far-fault records [Bez and Miranda, 2000]. Regression analyses of
these data led to an equation, which is restricted to elastoplastic systems, for the inelastic
deformation ratio as a function of elastic vibration period and ductility factor [Miranda,
2000].
The inelastic deformation ratio, if expressed as a function of elastic vibration

period and yield-strength reduction factor Ry (= strength required for the structure to
remain elastic yield strength of the structure), can be used to determine the inelastic
deformation of an existing structure with known lateral strength. For reinforced concrete
structures, researchers have developed an equation that specifies the range of Ry and Tn
where the equal-deformation rule is valid [Shimazaki and Sozen, 1984], established the
short-period range where the inelastic deformation ratio is sensitive to both strength and
stiffness of the system, and developed an upper bound equation for the deformation [Qi
and Moehle, 1991]. An evaluation of the FEMA-273 procedure for estimating
deformation demonstrated that the ratio of the mean inelastic and elastic deformations
exceeds unity if Ry exceeds 5, regardless of Tn [Whittaker et al., 1998]. Response data
for 216 ground motions recorded on NEHRP site classes B, C, and D demonstrated that
the inelastic deformation ratio is influenced little by soil condition, by magnitude if
Ry < 4 (but significantly for larger Ry ), or by rupture distance so long as it exceeds 10
km [Ruiz-Garcia and Miranda, 2002]. Regression analysis of these data led to an

equation for the inelastic deformation ratio as a function of Tn and Ry ; this equation is
restricted to elastoplastic systems.
Researchers have demonstrated that the inelastic deformation ratio in the

acceleration-sensitive spectral region is reduced because of post-yield stiffness [Veletsos,
1969; Qi and Moehle, 1991], and increased due to stiffness degradation [Clough, 1966;
Qi and Moehle, 1991; Song and Pincheira, 2000] and pinching [Gupta and Kunnath,
1998; Gupta and Krawinkler, 1998; Song and Pincheira, 2000] of the hysteresis loop.
However, others have concluded that the influence of post-yield stiffness on the ductility
demand for constant-strength systems is not significant, especially at longer periods
147
[Clough, 1966] and that mean responses of constant-ductility systems can be
conservatively estimated using the elastoplastic model [Riddell et al., 2002]; it is
noticeableof the order 10-20%but not predominant on inelastic strength demands
[Nassar and Krawinkler, 1991].
Alternatively Ry - - Tn relations can be used to estimate the peak deformation of
an inelastic SDF system [Chopra and Goel, 1999; Fajfar, 2000], but this indirect method
is biased towards underestimating the deformation [Miranda, 2001].
The objective of this chapter is to present the median and dispersion of the
inelastic deformation ratio of SDF systems for several ensembles of ground motions,
representing large or small earthquake magnitude and small or large distance, NEHRP
site classes B, C, and D (all firm sites); near-fault ground motions are also included. Two
sets of results are presented, one for systems with known ductility factor and the other for
systems with known yield-strength reduction factor. The earthquake responses of bilinear
systems are investigated comprehensively, but degradation of stiffness or strength, or
pinching of the hysteresis loop is not considered. Investigated is the influence of
earthquake magnitude and distance, site class, and near-fault ground motions on the
inelastic deformation ratio. Presented are two equationsin terms of or Ry for the
inelastic deformation ratio, valid for all ground motions ensembles considered, in a form
convenient for application to design or evaluation, respectively, of structures.
9.2 INELASTIC DEFORMATION RATIO: THEORY
9.2.1 Bilinear Systems
The initial loading curve for a system with bilinear force-deformation relation
f s ( u ,sgn u ) is shown in Fig. 9.1. The elastic stiffness is k and the post-yield stiffness is
k , where is the post-yield stiffness ratio; elastoplastic systems ( = 0 ) are included

but not systems with negative post-yield stiffness ( < 0 ). The yield strength is f y and
the yield deformation u y . Unloading and reloading of the hysteretic system occurs
without any deterioration of stiffness or strength. Within the linearly elastic range the
system has a natural vibration period Tn (frequency n = 2 / Tn ) and damping ratio .
148
The yield strength reduction factor, Ry , is defined by
f o uo
Ry = = (9.1)
f y uy
where f o and uo are the minimum yield strength and yield deformation required for the
structure to remain elastic during the ground motion, or the peak response values for the
corresponding linear system. The peak force in the inelastic system is f m (Fig. 9.1).
The peak or absolute (without regard to algebraic sign) maximum deformation of the
bilinear system is denoted by um and its ductility factor by
um
= (9.2)
uy
It can be shown that the inelastic deformation ratio, defined as the ratio of deformations
of inelastic and corresponding linear systems, is
um
= (9.3)
uo R y
9.2.2 Computing Inelastic Deformation Ratio
The equation governing the deformation u ( t ) of an SDF system of mass m due to
earthquake acceleration u g (t ) is written in two alternative forms:
u + 2nu + f s / m = u g (t ) or u t + 2nu + f s / m = 0 (9.4)
where u t = u + u g . The deformation responses of an inelastic systemdefined by period
Tn , post-yield stiffness ratio , and damping ratio and its corresponding linear
system are determined by numerical solution of Eq. (9.4a) to obtain the peak
deformations um and uo , respectively, leading to the inelastic deformation ratio.
This ratio will be denoted by
um um
C = or CR = (9.5)
uo uo
149
Consistent with Mirandas notation [Miranda, 2001], the subscripts and Ry in Eq.
(9.5) represent systems with known ductility capacity or known yield strength defined
by the reduction factor Ry , respectively. The inelastic deformation ratio CR for a system
of known yield strength can be determined directly from the computed peak deformations
um and uo of inelastic and elastic systems; the corresponding ductility demand is given
by Eq. (9.3), wherein Ry is readily available from Eq. (9.1). An iterative procedure is
necessary to determine the inelastic deformation ratio C for a specified ductility factor
because the yield strength corresponding to a selected cannot be determined directly

[Chopra, 2001; Section 7.5]. Because C is not unique, as more than one value of f y
may yield the same , following convention the smallest f y is chosen that gives the
largest C [Chopra, 2001; Section 7.5]. The median value and dispersion of C and CR
is then computed for each of the nine ensembles of excitations described later.
9.2.3 Limiting Values of Inelastic Deformation Ratio
This ratio can be determined analytically for two limiting cases: Tn tends to zero and Tn
tends to infinity. The first implies a very-short-period system that, for a fixed mass, is
essentially rigid; thus u ( t ) 0, u ( t ) 0, and u t ( t ) u g (t ) . Equation (9.4b) then gives
f s (t ) = mu g (t ) and its peak value is f so = mu go (where the subscript o denotes the peak
value), which is independent of the force-deformation relation and hence valid for
bilinear and corresponding linear systems. Thus for the limiting case of Tn =0, f m = f o , an
equal-force rule based on Fig. 9.1 can be expressed as
f y + ( 1) f y = f o
Dividing by f y gives
1 + ( 1) = Ry (9.6)
Substituting Eq. (9.6) in Eq. (9.3), the deformation ratio um / uo for Tn =0 systems can be
expressed as a function of or of Ry :
150
1 Ry 1
L = or LR = 1 + (9.7)
1 + ( 1) Ry
Equations (9.7a) and (9.7b) can be interpreted as the limiting values of um / uo as Tn
tends to zero for systems with constant or constant Ry , respectively, independent of
damping. For elastoplastic systems ( = 0 ), L = and LR = ; thus, the limiting value
of um / uo for systems with Tn = 0 is for constant- systems but is unbounded for
constant- Ry systems. In passing, note that the prevailing view that Ry =1 at Tn =0,
independent of , is valid only for elastoplastic systems; Eq. (9.6) indicates that Ry is
slightly larger for bilinear systems.
Figures 9.2a and 9.2b are plots of Eqs. (9.7a) and (9.7b) for various values of
and Ry , respectively. For the limiting case of Tn =0, um is greater than uo , and um / uo
increases with (Fig. 9.2a) and with Ry (Fig. 9.2b)i.e., decreasing yield strength.
This ratio rapidly increases with Ry near Ry = 1 , implying that the deformation of the
inelastic system is much larger even if its strength is only slightly below that required for
the structure to remain elastic, an observation first made by Veletsos et al. [1965]. Post-
yield stiffness reduces the um / uo ratio below its value for elastoplastic systems, which as
mentioned earlier is for constant- systems but is unbounded for constant- Ry
systems; in the latter case the ratio is bounded if the system has the slightest post-yield
stiffness. Post-yield stiffness is much more influential in reducing um / uo for constant- Ry
systems compared to constant- systems. Thus, much difference between the C and
CR would be expected in the short-period range.
The ratio of deformations of inelastic and corresponding linear systems with very
long period Tn can be determined by physical reasoning. The mass of such an extremely
flexible system would be expected to stay essentially stationary while the ground below
moves. Thus, its peak deformation would tend to the peak ground displacement, u go , as
Tn tends to infinity, independent of the force-deformation relation, and, therefore, valid
151
for bilinear and corresponding linear systems. Thus for the limiting case of Tn = ,
um = uo = u go and
C = C R = 1 (9.8)
which is the well-known equal-displacement rule [Veletsos and Newmark, 1960].

Equations (9.7) and (9.8) are independent of the excitation.
9.3 GROUND MOTIONS AND ELASTIC RESPONSE SPECTRA
Seven ensembles of far-fault ground motions, each with 20 records, were included in this
investigation. The first group of ensembles, denoted by LMSR, LMLR, SMSR, and
SMLR represent four combinations of large (M=6.6-6.9) or small (M=5.8-6.5) magnitude
and small (R=13-30 km) or large (R=30-60 km) distance1. These ground motions and
their parameters are presented in Tables 3.3, 9.1, 9.2, and 9.3 for LMSR, LMLR, SMSR,
and SMLR ensembles, respectively. All these records except one correspond to NEHRP
site class D; the Morgan Hill-San Juan Bautista (MH84sjb) record in the SMLR ensemble
corresponds to NEHRP site class C. The second group of three ensembles is categorized
by NEHRP site classes B, C, or D. These ground motions were recorded during
earthquakes with magnitudes ranging from 6.0 to 7.4 at distances2 ranging from 0 to 120
km. Ground motions in this group and their parameters are listed in Tables 9.4-9.6. The
eighth ensemble includes two horizontal components (fault-normal and fault-parallel) of
15 near-fault ground motions, recorded during earthquakes of magnitudes ranging from
6.2 to 6.9 and at distances ranging from 0 to 9 km. These ground motions were all
recorded on firm soil (NEHRP site class D) or rock; the rock motions have been modified
to reproduce soil site conditions [Somerville, 1998]. Ground motions in this ensemble
and their parameters are shown in Tables 9.7 and 9.8.
The median response spectrum for each ensemble of ground motions was
determined. The one for the LMSR ensemble is presented in Fig. 9.3 as a four-way
1
Record-to-source distance is defined as the closest distance to the fault rupture zone except for two
records: Point Mugu-Port Hueneme (PM73phn) in the SMSR ensemble and Borrego-El Centro Array#9
(BO42elc) in the SMLR ensemble where the hypocentral distance is reported as 25 and 49 km,
respectively.
2
Horizontal distance from the edge of horizontal projection of the fault rupture area to the site
152
logarithmic plot. The idealized version of the response spectrumshown in dashed
linewas constructed according to the procedure described in Riddell and Newmark
[1979], where the spectrum is divided logically into three period ranges [Chopra, 2001;
Section 6.8]. The long-period region to the right of point d, Tn > Td , is called the
displacement-sensitive region; the short-period region to the left of point c, Tn < Tc , is

called the acceleration-sensitive region; and the intermediate-period region between
points c and d, Tc < Tn < Td , is called the velocity-sensitive region. Similar plots for the
other ground motion ensembles are shown in Figs. 9.4 to 9.11, and Ta , Tb , Tc , Td , Te and
T f of all ensembles are shown in Table 9.9.
The median pseudo-acceleration spectra normalized to the peak ground

acceleration of the ensemble are presented in Fig. 9.12. Clearly, the spectral shapes for
the LMSR, LMLR, SMSR, and SMLR ensembles are very similar (Fig. 9.12a), and for
the site class B, C, and D ensembles are similar, but differ from the LMSR shape in the
period range 0.3 to 0.8 sec (Fig. 9.12b). However, the shapes of median response spectra
from the near-fault fault-normal (NF-FN) and near-fault fault-parallel (NF-FP) ensembles
are significantly different than the LMSR ensemble (Fig. 9.12c).
9.4 DEFORMATION OF SYSTEMS WITH KNOWN DUCTILITY
Figure 9.13 presents the median value of C for the LMSR ensemble as a function of Tn
for fixed damping ratio =5%; all results presented in this chapter are for this damping
ratio. The results for fixed post-yield stiffness ratio (Fig. 9.13) permit the following
observations on how the degree of inelastic action indicated by influences the
relationship between um and uo in the various spectral regions. In the acceleration-
sensitive region, um uo at Tn = Tc but um exceeds uo increasingly for shorter periods
and larger , indicating greater inelastic action, and approaches the limit given by Eq.
(9.7a) as Tn tends to zero. In the velocitysensitive region, um uo and is affected very
little by . In the displacement-sensitive region, um < uo for systems in the period range
Td to T f where um decreases for increasing ; however, for systems with periods longer
153
than T f , um uo (essentially independent of ) and um approaches uo [Eq. (9.8)] as Tn
tends to infinity, independent of .
The results for fixed ductility factor (Fig. 9.14) indicate that post-yield stiffness
reduces the deformation of bilinear systems relative to elastoplastic systems over the
entire period range except for Tn > T f where um uo . While this observation is valid over
the range of =0 to 10%, it may not be valid for larger values of the post-yield stiffness
ratio, e.g., in the period range Td to T f , um will eventually increase for large enough
values of to uo at =1. Computed from the data of Fig. 9.13, the ratio of median um
for bilinear and elastoplastic systems is plotted against Tn for fixed and three values of
(Fig. 9.15a) and for fixed and four values of (Fig. 9.15b). Figure 9.15 indicates
that the percentage reduction in deformation due to post-yield stiffness is roughly
constant over a wide period range, increases slightly for very short periods ( Tn < Ta ), but
disappears for very long periods ( Tn > T f ). The reduction in deformation is roughly
independent of period because the degree of inelastic action, as indicated by the value,
is kept constant; consistent with intuition, greater reduction is achieved at larger ductility
factors. The reduction of deformation due to post-yield stiffness is modest for realistic
values of and ; e.g., for =4 and =3% the deformation is reduced by less than
15% over the entire period range (Fig. 9.15a). These results support the view that for
systems with known and small ductility, elastoplastic models provide a usefully
conservative estimate of deformation [Riddell and Newmark, 1979; Riddell et al., 2002]
and is consistent with the earlier result that Ry is affected little by post-yield stiffness
[Nassar and Krawinkler, 1991].
The dispersion of C for the LMSR ensemble is plotted against Tn for fixed
and four values of (Fig. 9.16a) and for fixed with four values of (Fig. 9.16b).
Here the dispersion is (1) zero as Tn tends to zero or infinity because the limiting value of
C [Eqs. (9.7a) and (9.8)] is independent of the ground motion; (2) roughly similar over
a wide range of periods from Tb to Te ; (3) increases with (Fig. 9.16a) consistent with
154
intuition; and (4) affected very little by post-yield stiffness over the range of
considered (Fig. 9.16b), contrary to intuition.
9.5 DEFORMATION OF SYSTEMS WITH KNOWN Ry
Figure 9.17 presents the median value of CR for the LMSR ensemble as a function of Tn
for = 5% . The results for fixed post-yield stiffness ratio (Fig. 9.17) permit the
following observations on how the yield strength influences the relationship between the
deformations um and uo of inelastic and elastic systems in the various spectral regions. In
the acceleration-sensitive region, starting from um uo at Tn = Tc , um exceeds uo
increasingly for shorter periods where um is very sensitive to the yield strength,
increasing as the yield strength is reduced; um approaches the limit given by Eq. (9.7b) as
Tn tends to zero. Very short period systems ( Tn < Ta ), even systems with strength only
slightly smaller than the minimum strength required for the systems to remain elastic
(e.g., Ry =1.5), experience deformations much larger than the elastic deformation. Just as
in the case of constant- systems, um uo in the velocity-sensitive region and is

essentially independent of the yield strength. In the displacement-sensitive region, the
relationship between um and uo is similar to that observed for constant- systems:
um < uo for systems in the period range Td to T f where um decreases as strength is
reduced; um uo , essentially independent of strength, for periods longer than T f ; and um
approaches uo [Eq. (9.8)] as Tn tends to infinity, independent of yield strength. The
results for fixed (or constant) Ry presented in Fig. 9.18 indicates that post-yield stiffness
reduces the deformation of a bilinear system relative to its value for elastoplastic systems.
Although the deformation is reduced over the entire period range, the ratio of
median um for bilinear and elastoplastic systems demonstrates that post-yield stiffness
reduces the deformation only moderately in the velocity and displacement-sensitive
regions; e.g. for Ry = 4 and Tn > Tc this reduction is less than 13%, 17%, and 22% for
= 3, 5, and 10%, respectively (Fig. 9.19a). In these regions, the percentage reduction in
155
deformation due to post-yield stiffness is insensitive to the period because the median
ductility demand is roughly constant over this period range (Fig. 9.19b). However, the
deformation is reduced considerably in the acceleration-sensitive region, especially for
periods shorter than Tb .
The constant- Ry plots in Fig. 9.19a demonstrate that post-yield stiffness is much
more effective in reducing deformation in the acceleration-sensitive region than observed

from the constant- plots of Fig. 9.15a. Such is the case because ductility demands for
systems with a selected Ry (say, Ry = 4 ) are much larger than the Ry (say, 4) value in
the acceleration-sensitive spectral region (Fig. 9.19b); and larger implies larger
reduction in deformation due to post-yield stiffness (Fig. 9.15b). This observation implies
that ignoring post-yield stiffness in estimating deformation is too conservative for seismic
evaluation of existing structures with periods in the acceleration-sensitive region. While
factual and seemingly significant, this observation may not be important for real
structures because they are usually strong enough to remain elastic for most ground
motions.
The dispersion of CR for the LMSR ensemble is plotted against Tn for fixed
and four values of Ry (Fig. 9.20a) and for fixed Ry with four values of (Fig. 9.20b).
Here the dispersion is: (1) zero as Tn tends to zero (except for elastoplastic systems,
= 0 , Fig. 9.20b) or as Tn tends to infinity because the limiting values of CR [Eqs.

(9.7b) and (9.8)] are independent of the ground motion; (2) larger for weaker (larger Ry )
systems over a wide range of periods except for Tn < Tb where this trend is reversed; (3)
essentially independent of post-yield stiffness ratio (Fig. 9.20b) for periods in the velocity
and displacement-sensitive regions; and (4) reduced significantly with the slightest post-
yield stiffness in the very short-period ( Tn < Tb ) portion of the acceleration-sensitive
region.
156
9.6 INFLUENCE OF EARTHQUAKE MAGNITUDE AND DISTANCE
The computations that led to the results of the preceding section were repeated for the
LMSR, LMLR, SMSR, and SMLR ground motion ensembles. As observed earlier, the
shapes of the four median response spectra are very similar (Fig. 9.12a) and their Tc
values0.43, 0.46, 0.41, and 0.44 sec, respectivelyare also close. This similarity
suggests that the median C versus Tn and CR versus Tn functions should be quite
similar. This expectation is confirmed by the deformation ratio-period plots for the four
ensembles shown in Figs. 9.21a and 9.21b.
9.7 INFLUENCE OF FIRM SITE CLASSES
Similar computations were repeated for three ensembles of ground motions recorded on
firm sites: NEHRP site classes B, C, and D. As observed earlier, the shapes of the three
median response spectra are similar (Fig. 9.12b) and their Tc values0.32, 0.33, and
0.40 sec, respectivelyare also similar. Although the Tc values for the three site classes
vary over a wider range compared to when earthquake magnitude and recording distance
were varied, the median C versus Tn curves (Fig. 9.22a) and CR versus Tn curves (Fig.
9.22b) for the three site classes are very similar; they are close to the LMSR result in
spite of the difference in their spectra (Fig. 9.12b).
9.8 NEAR-FAULT GROUND MOTIONS
As mentioned earlier, the median response spectra for the fault-normal and fault-parallel
components of near-fault ground motions display a significantly different shapewith
much different Tc =0.84 and 0.62 seccompared to far-fault ground motions, e.g., the
LMSR ensemble (Fig. 9.12c). Because of these differences, the inelastic deformation
ratios C and CR for near-fault ground motions is significantly different than far-fault
motions (Figs. 9.23a and 9.24a), which is consistent with earlier work reported by Bez
and Miranda [2000]. This systematic difference between the values of CR for near-fault
and far-fault ground motions in the acceleration-sensitive spectral region is primarily due
to the difference between the period Tc values for the two sets of excitations, as
157
demonstrated by replotting these data against the normalized vibration period Tn / Tc in
Figs. 9.23b and 9.24b, first introduced by Chopra and Chintanapakdee [2001]. Now the
C values for the far-fault ground motions and bothfault-normal and fault-parallel
components of near-fault ground motions are close in the acceleration-sensitive ( Tn < Tc )

region of the spectrum, and also in the velocity and displacement-sensitive spectral
regions, Tn > Tc (Fig. 9.23b); the same observation applies also to CR (Fig. 9.24b).
9.9 ESTIMATING DEFORMATIONS OF INELASTIC SYSTEMS
Simplified equations for inelastic deformation ratios C and CR would obviously
facilitate estimation of the deformation of inelastic SDF system because the deformation
of the corresponding linear system is readily known from the elastic design spectrum.
Such an equation for C could be used to determine the inelastic deformation of a new or
rehabilitated structure where the global displacement ductility capacity can be estimated.
Similarly an equation for CR could be used to determine the inelastic global and local
deformations of existing structures with known lateral strength. Such equations are
developed in this section.
9.9.1 Equation for C
Miranda [2000] developed the following equation for elastoplastic systems:
1
1

( )
C = 1 + 1 exp 12T 0.8

(9.9)
Alternatively, the deformation of inelastic systems can be determined from Eq. (9.3)
using Ry - - Tn relations, which have been developed by several researchers. The earliest
of these relations for elastoplastic systems [Veletsos and Newmark, 1960], and consistent
with Newmark-Hall inelastic design spectra [Newmark and Hall, 1982], is available in
Fig. 7.11.2 of Chopra [2001]. Substituting this relation in Eq. (9.3) gives
158
Tn < Ta

C = / 2 1 Tb < Tn < Tc ' (9.10)

1 Tn > Tc
wherein Tc ' = Tc 2 1 / and equations for the transitional period regions are not
included, but can be found in [Chopra and Goel, 1999]. Substituting the relation
developed by Krawinkler and Nassar [1992], based on earthquake response of bilinear
systems, in Eq. (9.3) gives
1/ c
C = c ( 1) + 1 (9.11) (9.11a)
where
Tna b
c(Tn , ) = + (9.11b)
1 + Tna Tn
and the numerical coefficients depend on the post-yield stiffness ratio, : a = 1 and
b = 0.42 for = 0 ; a = 1 and b = 0.37 for = 2% ; a = 0.8 and b = 0.29 for = 10% .
Figure 9.25a compares Eqs. (9.9), (9.10), and (9.11) for =0 with the median
value of C determined earlier (Fig. 9.13) for elastoplastic systems subjected to the
LMSR ensemble of ground motions. Only results for ductility factor of 6 (the largest
value considered) are presented because it is the severest test of the C -equations. Even
for this extreme case, Eqs. (9.9)-(9.11) provide excellent estimates of the inelastic
deformation for elastoplastic systems, irrespective of the fact that they were developed
for ground motion ensembles different than the LMSR ensemble; the latter two equations
are based on an indirect method, with the bias identified by Miranda [2001]. A similar
comparison for =10% (Fig. 9.25b) demonstrates that Eq. (9.11) (specialized for
= 10% ) is satisfactory over a wide range of periods but overestimates C at very short
periods because as Tn tends to zero, Eq. (9.11) approaches a limiting value that is
valid only for elastoplastic systemsinstead of Eq. (9.7a) derived earlier. Because Eqs.
(9.9) and (9.10) were developed from response data for elastoplastic systems, as expected
they overestimate C for bilinear systems; note that Eq. (9.9) is based on mean value of
159
C , which is larger than the median value for probability density function that are
skewed to the right.
Presented next is an improved equation that fits the median C data for any
ensemble of ground motions, but ignores the data showing C <1 over the period range
Td to T f (Fig. 9.13) and satisfies the limiting values of C at Tn =0 and Tn = [Eqs.
(9.7a) and (9.8)]. Such a function for C has been derived (Appendix C) in terms of
ductility factor and the normalized period Tn / Tc :
1
a Tn
d
C = 1 + ( L 1) + b + c
1
(9.12)
Tc
where L is given by Eq. (9.7a). The numerical parameters a, b, c, and d were
determined from the response data by the algorithm (fmins.m) in MATLAB [1997] to
minimize the residual error, defined as the sum of the squares of differences between the
actual data and the regression equation, but with a penalty for underestimation: a = 72,
b =2.2, c = 1.6, and d = 1.5 using the LMSR data and a = 105, b = 2.3, c = 1.9, d = 1.7
using the data for four (LMSR, LMLR, SMSR, and SMLR) ensembles; these parameter
values are independent of post-yield stiffness ratio, . Further information is available in
Appendix C.
Figure 9.26 shows that Eq. (9.12) with the first set of parameter values and
Tc = 0.43 sec closely matches the median C computed for the LMSR and near-fault
ground motions. The excellent agreement with the LMSR data is expected because the
parameters were determined from those data. Most impressive is that the same
parameters provide a good match with the near-fault data, which has been achieved
because Eq. (9.12) is developed as a function of Tn / Tc , the normalized period, instead of
Tn . Figure 9.27 shows that Eq. (9.12) using the second set of parameter values provides a
modest overestimate of the median C computed for a wide range of system parameters
and all four ground motion ensembles. A more complicated equation would be necessary
160
to match the C <1 data in the Td -to- T f period range; however, these data should be re-
examined with P- effects included before developing such an equation.
9.9.2 Equation for CR
Ruiz-Garcia and Miranda [2002] developed the following equation for elastoplastic
systems
1 1
CR = 1 + + ( R 1) (9.13)
a (T / Ts )b c
where a=50, b=1.8, c=55 and Ts =0.75, 0.85, or 1.05 for NEHRP site class B, C, or D,
respectively.
Alternatively, the deformation of inelastic systems can be determined from Eq.

(9.3) using Ry - - Tn relations, as demonstrated earlier for constant-ductility systems.
Substituting the Ry - - Tn relations, for elastoplastic systems, consistent with [Newmark
and Hall, 1982] in Eq. (9.3), gives
Tn < Ta
2
CR = ( Ry + 1) / 2 Ry Tb < Tn < Tc ' (9.14)

1 Tn > Tc
and using the relation developed by Krawinkler and Nassar [1992] for bilinear systems
gives
CR =
1
Ry
1 c
( )

1 + c R y 1 (9.15)
where c is given by Eq. (9.11b).
Figure 9.28a compares Eqs. (9.13), (9.14), and (9.15) for =0 with the median
value of CR determined earlier (Fig. 9.16) for elastoplastic systems subjected to the
LMSR ensemble of ground motions; results are presented only for Ry = 4 . Equations
(9.13)-(9.15) provide good estimates of the inelastic deformation for elastoplastic systems
161
with initial period Tn longer than Tb , although they were developed for ground motion
ensembles different than the LMSR ensemble. Surprisingly, a similar comparison for
=10% (Fig. 9.28b) demonstrates that surprisingly Eq. (9.15) (specialized for =10%)
is not much better than Eqs. (9.13) and (9.14), which were intended for elastoplastic
systems. All the relations overestimate CR at very short periods because they approach
infinity as Tn tends to zero, instead of the bounded value given by Eq. (9.7b). Because
Eqs. (9.13) and (9.14) were developed from response data for elastoplastic systems, as
expected they overestimate CR for bilinear systems.
Presented finally is an equation that fits the median CR data for any ensemble of
ground motions, but ignores the data showing CR <1 over the period range Td to T f (e.g.,
Fig. 9.17) and satisfies the limiting values of CR at Tn =0 and Tn = [Eqs. (9.7b) and
(9.8)]. Such a function for CR has been derived (Appendix C) in terms of the yield-
strength reduction factor Ry and the normalized period Tn / Tc :
1
a Tn
d
CR = 1 + ( LR 1) + b + c
1
(9.16)
Ry Tc

Observe that the form of this equation is the same as Eq. (9.12) for C where L is
replaced by LR and by Ry . Using the same procedure and criteria, nonlinear
regression analysis of the LMSR data led to a = 63, b = 2.3, c = 1.7, and d = 2.3, and of
the data for four ensembles of ground motion led to a = 61, b = 2.4, c = 1.5, and d = 2.4.
Figure 9.29 shows that Eq. (9.16) using the first set of parameter values agrees
well with the median CR computed for the LMSR and near-fault ground motions, which
is expected because the parameters were determined from those data. Most impressive is
that the same parameters provide a good match with the near-fault data, which has been
achieved because Eq. (9.16) is developed as a function of Tn / Tc , the normalized period,
instead of Tn . Figure 9.30 shows that Eq. (9.16) with the second set of parameter values
162
provides a good, generally modestly conservative estimate of the median CR for all four
ensembles.
163
Table 9.1 List of ground motions in LMLR ensemble
.. .
R* u go u go u go
1 1968 Borrego Mountain 6.8 El Centro Array #9 BM68elc 46.0 56 13.2 10.1
2 1989 Loma Prieta 6.9 APEEL 2E Hayward Muir School LP89a2e 57.4 167 13.7 3.9
3 1989 Loma Prieta 6.9 Fremont - Emerson Court LP89fms 43.4 138 12.9 8.3
4 1989 Loma Prieta 6.9 Halls Valley LP89hvr 31.6 132 15.4 3.3
5 1989 Loma Prieta 6.9 Salinas - John & Work LP89sjw 32.6 110 15.7 8.0
6 1989 Loma Prieta 6.9 Palo Alto - SLAC Lab LP89slc 36.3 191 37.4 10.0
7 1994 Northridge 6.7 Covina - W. Badillo NR94bad 56.1 98 5.7 1.2
8 1994 Northridge 6.7 Compton - Castlegate St NR94cas 49.6 134 7.0 2.2
9 1994 Northridge 6.7 LA - Centinela St NR94cen 30.9 315 22.9 5.5
164
10 1994 Northridge 6.7 Lakewood - Del Amo Blvd NR94del 59.3 135 11.2 2.0
11 1994 Northridge 6.7 Downey - Co Maint Bldg NR94dwn 47.6 155 13.6 2.3
12 1994 Northridge 6.7 Bell Gardens - Jaboneria NR94jab 46.6 67 7.7 2.5
13 1994 Northridge 6.7 Lake Hughes #1 NR94lh1 36.3 85 9.4 3.7
14 1994 Northridge 6.7 Lawndale - Osage Ave NR94loa 42.4 150 8.0 2.6
15 1994 Northridge 6.7 Leona Valley #2 NR94lv2 37.7 62 7.2 1.6
16 1994 Northridge 6.7 Palmdale - Hwy 14 & Palmdale NR94php 43.6 66 8.4 2.0
17 1994 Northridge 6.7 LA - Pico & Sentous NR94pic 32.7 183 14.2 2.4
18 1994 Northridge 6.7 West Covina - S. Orange Ave NR94sor 54.1 62 5.9 1.3
19 1994 Northridge 6.7 Terminal Island - S Seaside NR94sse 60.0 190 12.0 2.3
20 1994 Northridge 6.7 E Vernon Ave NR94ver 39.3 150 10.1 1.8
Table 9.2 List of ground motions in SMSR ensemble
.. .
R* u go u go u go
1 1979 Imperial Valley 6.5 Calipatria Fire Station IV79cal 23.8 77 13.3 6.2
2 1979 Imperial Valley 6.5 Chihuahua IV79chi 28.7 265 24.8 9.1
3 1979 Imperial Valley 6.5 El Centro Array #1 IV79e01 15.5 137 15.8 10.0
6 1979 Imperial Valley 6.5 Cucapah IV79qkp 23.6 303 36.3 10.7
7 1979 Imperial Valley 6.5 Westmorland Fire Station IV79wsm 15.1 108 21.9 10.0
8 1980 Livermore 5.8 San Ramon - Eastman Kodak LV80kod 17.6 75 6.1 1.7
9 1980 Livermore 5.8 San Ramon Fire Station LV80srm 21.7 39 4.0 1.2
165
10 1984 Morgan Hill 6.2 Agnews State Hospital MH84agw 29.4 31 5.5 2.0
11 1984 Morgan Hill 6.2 Gilroy Array #2 MH84g02 15.1 159 5.1 1.4
12 1984 Morgan Hill 6.2 Gilroy Array #3 MH84g03 14.6 191 11.2 2.4
13 1984 Morgan Hill 6.2 Gilroy Array #7 MH84gmr 14.0 111 6.0 1.8
14 1973 Point Mugu 5.8 Port Hueneme PM73phn NA 110 14.8 2.6
15 1986 N. Palm Springs 6.0 Palm Springs Airport PS86psa 16.6 184 12.1 2.1
16 1987 Whittier Narrows 6.0 Compton - Castlegate St WN87cas 16.9 325 27.1 5.0
17 1987 Whittier Narrows 6.0 Carson - Catskill Ave WN87cat 28.1 41 3.8 0.8
18 1987 Whittier Narrows 6.0 Brea - S Flower Av WN87flo 17.9 113 7.0 1.2
19 1987 Whittier Narrows 6.0 LA - W 70th St WN87w70 16.3 148 8.6 1.5
20 1987 Whittier Narrows 6.0 Carson - Water St WN87wat 24.5 102 9.0 1.9
Table 9.3 List of ground motions in SMLR ensemble
.. .
R* u go u go u go
1 1942 Borrego Mountain NA El Centro Array #9 BO42elc NA 67 3.9 1.4
2 1983 Coalinga 6.4 Parkfield - Cholame 5W CO83c05 47.3 128 9.9 1.3
3 1983 Coalinga 6.4 Parkfield - Cholame 8W CO83c08 50.7 96 8.5 1.5
4 1979 Imperial Valley 6.5 Coachella Canal #4 IV79cc4 49.3 126 15.6 3.0
5 1979 Imperial Valley 6.5 Compuertas IV79cmp 32.6 183 13.8 2.9
6 1979 Imperial Valley 6.5 Delta IV79dlt 43.6 233 26.0 12.0
7 1979 Imperial Valley 6.5 Niland Fire Station IV79nil 35.9 107 11.9 6.9
8 1979 Imperial Valley 6.5 Plaster City IV79pls 31.7 56 5.4 1.9
9 1979 Imperial Valley 6.5 Victoria IV79vct 54.1 163 8.3 1.0
166
10 1980 Livermore 5.8 Tracy - Sewage Treatment Plant LV80stp 37.3 72 7.6 1.8
11 1984 Morgan Hill 6.2 Capitola MH84cap 38.1 97 4.9 0.6
12 1984 Morgan Hill 6.2 Hollister City Hall MH84hch 32.5 70 7.4 1.6
13 1984 Morgan Hill 6.2 San Juan Bautista MH84sjb 30.3 35 4.4 1.5
14 1986 N. Palm Springs 6.0 San Jacinto Valley Cem PS86h06 39.6 62 4.4 1.2
15 1986 N. Palm Springs 6.0 Indio PS86ino 39.6 63 6.6 2.2
16 1987 Whittier Narrows 6.0 Downey - Birchdale WN87bir 56.8 293 37.6 4.9
17 1987 Whittier Narrows 6.0 LA - Century City CC South WN87cts 31.3 51 3.5 0.6
18 1987 Whittier Narrows 6.0 LB - Harbor Admin FF WN87har 34.2 70 7.3 0.9
19 1987 Whittier Narrows 6.0 Terminal Island - S Seaside WN87sse 35.7 41 3.9 1.0
20 1987 Whittier Narrows 6.0 Northridge - Saticoy St WN87stc 39.8 115 5.1 0.8
Table 9.4 List of ground motions in NEHRP site class B ensemble
.. .
Station R* u go u go u go
No. Earthquake Name M Location
No. (km) (cm/s/s) (cm/s) (cm)
1 1992 Landers 7.3 Silent Valley, Poppet Flat 12206 51.3 49 3.8 3.5
2 1992 Landers 7.3 Twentinine Palms Park Maintennance Bldg 22161 41.9 79 3.7 2.2
3 1992 Landers 7.3 Amboy 21081 69.2 146 20.0 7.5
4 1989 Loma Prieta 6.9 Point Bonita 58043 88.1 71 13.7 3.4
5 1989 Loma Prieta 6.9 Piedmont, Piedmont Jr. High Grounds 58338 77.2 81 9.1 3.3
6 1989 Loma Prieta 6.9 San Francisco, Pacific Heights 58131 80.5 60 14.3 4.4
7 1989 Loma Prieta 6.9 San Francisco, Rincon Hill 58151 78.5 89 11.5 4.1
8 1989 Loma Prieta 6.9 Hollister-SAGO vault 1032 30.6 6 8.7 NA
9 1989 Loma Prieta 6.9 South San Francisco, Sierra Point 58539 67.6 103 8.4 2.5
167
10 1994 Northridge 6.7 Mt Wilson, CIT Seismic Station 24399 36.9 229 7.6 NA
11 1994 Northridge 6.7 Antellope Buttes 24310 48.6 100 4.3 2.4
12 1994 Northridge 6.7 Wonderland 90017 22.7 172 11.8 2.8
13 1994 Northridge 6.7 Wrightwood, Jackson Flat 23590 42.5 55 5.0 0.7
14 1994 Northridge 6.7 San Gabriel E. Grand Ave. 90019 41.7 256 9.7 2.8
15 1994 Northridge 6.7 Littlerock-Brainard Can 23595 46.9 7 6.3 1.3
16 1971 San Fernando 6.6 Lake Hughes, Array Station 9 127 23.0 119 4.9 4.4
17 1986 N. Palm Springs 6.0 Silent Valley, Poppet Flat 12206 23.7 107 3.6 NA
18 1987 Whittier Narrows 6.0 Mt Wilson, CIT Seismic Station 24399 22.1 171 4.1 NA
19 1987 Whittier Narrows 6.0 Los Angeles, Gritfith Park Observatory 141 22.3 134 15.2 NA
20 1979 Imperial Valley 6.5 Superstition Mountain 286 26.0 189 9.2 6.2
* The closest distance to the horizontal projection of the rupture
Table 9.5 List of ground motions in NEHRP site class C ensemble
.. .
1 1979 Imperial Valley 6.5 El Centro, Parachute Test Facillity 5051 14.0 200 14.6 8.4
2 1971 San Fernando 6.6 Pasadena, CIT Kresge Cal Tech Seismo Lab 266 22.0 199 11.3 NA
3 1971 San Fernando 6.6 Lake Hughes, Array Station 12 128 17.0 346 15.8 NA
4 1971 San Fernando 6.6 Castaic Old Ridge Route 110 26.0 309 20.0 NA
5 1989 Loma Prieta 6.9 Gilroy #6, San Ysidro Microwave site 57383 19.9 167 13.9 NA
6 1989 Loma Prieta 6.9 Saratoga, Aloha Ave. 58065 12.4 494 41.5 17.1
7 1989 Loma Prieta 6.9 Gilroy, Gavilon college Phys Sch Bldg 47006 10.9 349 29.2 5.8
8 1989 Loma Prieta 6.9 Santa Cruz, UCSC 58135 12.5 433 21.6 9.7
9 1989 Loma Prieta 6.9 San Francisco, Dimond Heighs 58130 75.9 111 14.2 4.2
168
10 1984 Morgan Hill 6.2 Gilroy Gavilon college Phys Scl Bldg 47006 16.0 95 3.4 NA
11 1984 Morgan Hill 6.2 Gilroy #6, San Ysidro Microwave Site 57383 11.5 215 11.2 2.5
12 1986 N. Palm Springs 6.0 Fun Valley 5069 15.8 129 10.6 1.4
13 1987 Whittier Narrows 6.0 Castaic, Old Ridge Route 24278 77.3 67 4.2 NA
14 1987 Whittier Narrows 6.0 Riverside, Airport 13123 57.8 57 1.3 NA
15 1952 Kern County 7.4 Pasadena, CIT Athenaeum 475 109.0 52 9.1 3.2
16 1952 Kern County 7.4 Santa Barbara, Courthouse 283 85.0 129 19.4 NA
17 1994 Northridge 6.7 Littlerock, Brainard Canyon 23595 47.9 71 6.0 1.4
18 1994 Northridge 6.7 Castaic Old Ridge Route 24278 24.6 557 51.8 NA
19 1994 Northridge 6.7 Lake Hughes #1, Fire station #78 24271 37.7 85 9.4 3.7
20 1994 Northridge 6.7 Rancho Palos Verdes, Hawthorne Blvd. 14404 53.8 71 5.0 NA
Table 9.6 List of ground motions in NEHRP site class D ensemble
.. .
1 1992 Landers 7.3 Yermo, Fire Station 22074 26.3 240 51.3 44.7
2 1992 Landers 7.3 Palm Springs, Airport 12025 28.2 87 13.8 8.4
3 1992 Landers 7.3 Pomona, 4th and Locust FF 23525 117.6 66 12.8 7.6
4 1994 Northridge 6.7 Los Angeles, Hollywood Storage Bldg. 24303 24.8 381 22.2 NA
5 1989 Loma Prieta 6.9 Gilroy 2, Hwy 101 Bolsa Road Motel 47380 12.1 394 33.7 8.3
6 1989 Loma Prieta 6.9 Gilroy 3, Sewage Treatment Plant 47381 14.0 532 35.4 NA
7 1989 Loma Prieta 6.9 Hayward, John Muir School 58393 58.9 166 13.9 3.7
8 1989 Loma Prieta 6.9 Agnews, Agnews State Hospital 57066 27.0 163 31.7 20.1
9 1989 Loma Prieta 6.9 Oakland, 2 story 58224 76.3 238 37.8 9.1
169
10 1987 Whittier Narrows 6.0 Los Angeles, Hollywood Storage Bldg. 24303 23.8 201 8.9 NA
11 1987 Whittier Narrows 6.0 Downey, County Maintennance Bldg 14368 16.2 193 29.3 4.1
12 1987 Whittier Narrows 6.0 Century City, LA Country Club South 24390 29.6 67 4.2 0.7
13 1987 Whittier Narrows 6.0 Rancho Cucamonga, Law and Justice Ctr. 23497 45.5 55 0.7 NA
14 1979 Imperial Valley 6.5 El Centro #13, Strobel Residence 5059 22.0 136 14.2 7.9
15 1979 Imperial Valley 6.5 Calexico, Fire Station 5053 10.6 270 18.2 NA
16 1952 Kern County 7.4 Los Angeles, Hollywood Storage PE Lot 135 107.0 58 6.1 NA
17 1984 Morgan Hill 6.2 Gilroy #7, Mantnilli Ranch, Jamison Rd 57425 13.7 183 6.7 NA
18 1984 Morgan Hill 6.2 Gilroy #3 Sewage Treatment Plant 47381 14.4 190 12.2 3.4
19 1971 San Fernando 6.6 Los Angeles, Hollywood Storage Bldg. 135 23.0 207 20.5 NA
20 1971 San Fernando 6.6 Vernon, Cmd Terminal 288 33.5 105 17.5 NA
Table 9.7 List of near-fault ground motions in fault-normal component (NFFN) ensemble
.. .
R* u go u go u go
1 1992 Erzincan 6.9 Erzincan Station EZ92erzi 2.0 424 119.2 42.3
2 1979 Imperial Valley 6.5 El Centro Array 6 IV79ar06 1.2 424 109.9 57.8
3 1979 Imperial Valley 6.5 Meloland IV79melo 0.0 372 117.4 43.6
4 1995 Hyogo-Ken-Nanbu 6.9 Kobe Station KB95kobj 3.4 1067 160.2 40.1
5 1995 Hyogo-Ken-Nanbu 6.9 Port Island KB95kpi1 6.6 426 100.3 49.7
6 1995 Hyogo-Ken-Nanbu 6.9 Takatori Station KB95tato 4.3 771 173.8 56.0
170
7 1989 Loma Prieta 6.9 Lexington Dam LP89lex 6.3 673 178.6 56.8
8 1989 Loma Prieta 6.9 Los Gatos LP89lgpc 3.5 704 172.8 65.1
9 1984 Morgan Hill 6.2 Anderson Dam MH84andd 4.5 436 26.6 3.7
10 1984 Morgan Hill 6.2 Coyote Dam MH84clyd 0.1 712 70.1 10.2
11 1994 Northridge 6.7 Newhall, LA County Fire Station NR94newh 7.1 709 119.1 34.1
12 1994 Northridge 6.7 Rinaldi Receiving Station NR94rrs 7.5 873 174.8 39.8
13 1994 Northridge 6.7 Sylmar Converter Station NR94scs ? 577 130.9 65.4
14 1994 Northridge 6.7 Sepulveda NR94spva 8.9 715 62.6 16.0
15 1994 Northridge 6.7 Sylmar County Hospital NR94sylm 6.4 718 122.2 31.0
* The closest distance to the fault
Table 9.8 List of near-fault ground motions in fault-parallel component (NFFP) ensemble
.. .
R* u go u go u go
1 1992 Erzincan 6.9 Erzincan Station EZ92erzi 2.0 448 58.1 29.5
2 1979 Imperial Valley 6.5 El Centro Array 6 IV79ar06 1.2 343 63.3 27.9
3 1979 Imperial Valley 6.5 Meloland IV79melo 0.0 262 27.7 16.2
4 1995 Hyogo-Ken-Nanbu 6.9 Kobe Station KB95kobj 3.4 564 72.4 15.9
5 1995 Hyogo-Ken-Nanbu 6.9 Port Island KB95kpi1 6.6 136 25.3 10.0
6 1995 Hyogo-Ken-Nanbu 6.9 Takatori Station KB95tato 4.3 416 63.7 23.3
171
7 1989 Loma Prieta 6.9 Lexington Dam LP89lex 6.3 363 68.6 25.4
8 1989 Loma Prieta 6.9 Los Gatos LP89lgpc 3.5 449 91.1 36.7
9 1984 Morgan Hill 6.2 Anderson Dam MH84andd 4.5 275 28.1 4.9
10 1984 Morgan Hill 6.2 Coyote Dam MH84clyd 0.1 844 78.3 15.5
11 1994 Northridge 6.7 Newhall, LA County Fire Station NR94newh 7.1 638 49.6 15.3
12 1994 Northridge 6.7 Rinaldi Receiving Station NR94rrs 7.5 381 59.6 21.6
13 1994 Northridge 6.7 Sylmar Converter Station NR94scs ? 641 103.8 36.8
14 1994 Northridge 6.7 Sepulveda NR94spva 8.9 696 69.9 15.6
15 1994 Northridge 6.7 Sylmar County Hospital NR94sylm 6.4 584 53.9 9.0
* The closest distance to the fault
Table 9.9 Periods defining spectral regions of all ensembles
Ensemble Ta Tb Tc Td Te Tf
Name (sec) (sec) (sec) (sec) (sec) (sec)
LMSR 0.034 0.16 0.43 2.44 8.0 20.4
LMLR 0.430 0.16 0.46 1.82 5.9 24.6
SMSR 0.035 0.13 0.41 2.03 7.0 24.8
SMLR 0.041 0.16 0.44 1.36 6.4 22.7
B 0.035 0.13 0.32 2.44 8.4 17.7
C 0.042 0.17 0.33 1.34 9.1 18.2
D 0.043 0.17 0.40 1.97 11.3 19.1
NFFN 0.043 0.41 0.84 2.58 3.8 14.0
NFFP 0.031 0.30 0.62 2.24 3.4 14.0
172
fs
f
o
f
m
f k
y
0
0 uy uo um u
Figure 9.1 Bilinear force-deformation relationship of an inelastic SDF system and the
corresponding elastic system.
(a) (b)
50 50
40 =0% 40
30 30 LR= for =0 3%
=3%
20 20
L =u /u at T =0
L =u /u at T =0
=5% 5%
n
n
=10%
10 10 10%
o
o
5 5
m
m
4 4
3 3
R

2 2
1 1
1 2 3 4 5 10 1 2 3 4 5 10
Ductility factor, Strength reduction factor, Ry
Figure 9.2 Influence of post-yield stiffness ratio on limiting values of inelastic

deformation ratio as elastic vibration period Tn tends to zero: (a) L and (b)
LR .
173
10
Spectral Regions
Acceleration Velocity Displacement
sensitive sensitive sensitive
0
10
c d
go
Pseudovelocity, V/
10
10
1
b
e
1
go
/u
1
D
A/

0.
f
1
go
1
0.
0.1
a
0.
Ta T T T T T
01
b c d e f
0.034 0.16 0.43 2.44 8.0 20.4
0.1 1 10
Natural vibration period, Tn (sec)
Figure 9.3 Median elastic response spectrum for the LMSR ensemble of far-fault
earthquake ground motions, shown by a solid line, together with an idealized
version in dashed line and spectral regions; =5%.
10
Spectral Regions
0
10
c d
go
Pseudovelocity, V/
10
10
1
b e
1
go
/u
1
D
A/

0.
1
go
1
0.
0.1
a f
0.
Ta T T T T T
01
b c d e f
0.043 0.16 0.46 1.82 5.9 24.6
0.1 1 10
n
Figure 9.4 Median elastic response spectrum for the LMLR ensemble of far-fault
174
10
Spectral Regions
0
10
c d
go
Pseudovelocity, V/
10
10
1
b e
1
go
/u
1
D
A/

0.
1
go
1
0.
0.1
f
a
0.
Ta T T T T T
01
b c d e f
0.035 0.13 0.41 2.03 7.0 24.8
0.1 1 10
Figure 9.5 Median elastic response spectrum for the SMSR ensemble of far-fault
10
Spectral Regions
0
10
c d
go
Pseudovelocity, V/
10
1
10
b
e
1
go
/u
1
D
A/

0.
1
go
0.1
1
0.
a
f
0.
Ta T T T T T
01
b c d e f
0.041 0.16 0.44 1.36 6.4 22.7
0.1 1 10
n
Figure 9.6 Median elastic response spectrum for the SMLR ensemble of far-fault
175
10
Spectral Regions
0
10
go
c d
10
Pseudovelocity, V/
10
1
b
e
1
go
/u
1
D
A/

0.
f
1
go
1
0.
0.1
a
0.
01
Ta T
b
T
c
T
d
T
e
T
f
0.035 0.13 0.32 2.44 8.4 17.7
0.1 1 10
Figure 9.7 Median elastic response spectrum for the B ensemble of far-fault earthquake
ground motions, shown by a solid line, together with an idealized version in
dashed line and spectral regions; =5%.
10
Spectral Regions
0
10
c d
go
Pseudovelocity, V/
10
10
1 b
1
go
/u
1
e
A/

0.
1
go
1
0.
0.1 a f
0.
01
Ta T
b
T
c
T
d
T
e
T
f
0.042 0.17 0.33 1.34 9.1 18.2
0.1 1 10
n
Figure 9.8 Median elastic response spectrum for the C ensemble of far-fault earthquake
176
10
Spectral Regions
0
10
c d
go
10
Pseudovelocity, V/
10
1
b
1
go
/u
1
D
e
A/

f
0.
1
go
1
0.
0.1 a
0.
01
Ta T
b
T
c
T
d
T
e
T
f
0.043 0.17 0.40 1.97 11.3 19.1
0.1 1 10
Figure 9.9 Median elastic response spectrum for the D ensemble of far-fault earthquake
177
10
Spectral Regions
0
10
go
c d
Pseudovelocity, V/
1
e
10
b
10
1
go
/u
1
D
f
A/

0.1
0.
1
go
1
0.
a
Ta T
b
T
c
T T
d e
T
f
0.
0.043 0.41
01
0.84 2.583.8 14.0
0.1 1 10
Figure 9.10 Median elastic response spectrum for the NFFN ensemble of near-fault
10
Spectral Regions
0
10
c d
go
Pseudovelocity, V/
e
10
1
10
b
1
go
/u
1
f
A/

0.
1
0.1
go
1
0.
a
Ta T T T T T
0.
01
b c d e f
0.031 0.30 0.62 2.24 3.4 14.0
0.1 1 10
n
Figure 9.11 Median elastic response spectrum for the NFFP ensemble of near-fault
178
3
(a) LMSR
2.5 LMLR
SMSR
SMLR
2
1.5
0.5
go
0
Median normalized pseudoacceleration, A/
0 0.5 1 1.5 2 2.5 3

3
(b) LMSR
2.5 B
C
D
2
1.5
0.5
0
0 0.5 1 1.5 2 2.5 3
3
(c) LMSR
2.5 NFFN
NFFP
1.5
0.5
0
0 0.5 1 1.5 2 2.5 3
n
Figure 9.12 Comparison of median pseudo-acceleration response spectra among (a)

LMSR, LMLR, SMSR, and SMLR ensembles of ground motions; (b) LMSR
and site class B, C, and D ensembles; and (c) far-fault (LMSR), and near-fault
fault-normal (NFFN) and near-fault fault-parallel (NFFP) ensembles; =5%.
179
(a) =0% (b) =3%
10 10
Ta Tb Tc Td Te T=1.5
f
Ta Tb Tc Td Te Tf
=2
5 =4 5
4 =6 4
Median C
3 3
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) =5% (d) =10%

10 10
Ta Tb Tc Td Te Tf Ta Tb Tc Td Te Tf
5 5
4 4
Median C
3 3
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Period, Tn (sec) Period, Tn (sec)
Figure 9.13 Influence of ductility factor on the median of inelastic deformation ratio
C for bilinear systems (with post-yield stiffness ratio =0, 3, 5, and 10%)
subjected to the LMSR ensemble of ground motions.
180
(a) =1.5 (b) =2
10 10
Ta Tb Tc Td Te=0%
Tf Ta Tb Tc Td Te Tf
=3%
5 =5% 5
4 =10% 4
Median C
3 3
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) =4 (d) =6
10 10
5 5
4 4
Median C
3 3
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Figure 9.14 Influence of post-yield stiffness ratio on the median of inelastic deformation
ratio C for bilinear systems subjected to the LMSR ensemble of ground
motions, presented for fixed values of ductility factor =1.5, 2, 4, and 6.
181
(a) =4 (b) =3%
1.5 1.5
Ta Tb Tc Td T =3%
T Ta Tb Tc Td T =1.5
T
e f e f
=5% =2
=10% =4
(um)(um)=0%
(um)(um)=0%
=6
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Figure 9.15 Ratio of median deformations um for bilinear and elastoplastic systems: (a)
influence of post-yield stiffness ratio at ductility factor =4 and (b)

influence of at =3%.
(a) =3% (b) =4

1 1
T T T T T T=1.5 T T T T T =0%
T
a b c d e f a b c d e f
=2 =3%
0.5 =4 0.5 =5%
Dispersion of C
0.4 =6 0.4 =10%

0.3 0.3
0.2 0.2
0.1 0.1
0.05 0.05
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Figure 9.16 Dispersion of inelastic deformation ratio C for bilinear systems subjected to
the LMSR ensemble of ground motions: (a) influence of ductility factor at

post-yield stiffness ratio =3% and (b) influence of at =4.
182
(a) =0% (b) =3%
100 100
Ta Tb Tc Td Te RT=1.5
yf Ta Tb Tc Td Te Tf
Ry=2
Ry=4
R =6
Median CR
y
10 10
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) =5% (d) =10%

100 100
Median CR
10 10
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100

Figure 9.17 Influence of strength reduction factor Ry on the median of inelastic
deformation ratio CR for bilinear systems (with post-yield stiffness ratio =0,
3, 5, and 10%) subjected to the LMSR ensemble of ground motions.
183
(a) Ry=1.5 (b) Ry=2
100 100
Ta Tb Tc Td Te=0%
Tf Ta Tb Tc Td Te Tf
=3%
=5%
=10%
Median CR
10 10
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) Ry=4 (d) Ry=6

100 100
Median CR
10 10
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100

Figure 9.18 Influence of post-yield stiffness ratio on the median of inelastic deformation
ratio CR for bilinear systems subjected to the LMSR ensemble of ground
motions for fixed values of strength reduction factor Ry =1.5, 2, 4, and 6.
184
(a) (b)
2
T T T T T T T T T T T =3%
T
a b c d e f 100 a b c d e f
=5%
1 =10%
(um) (um)=0%
0.5

0.4 10
0.3
0.2 =3%
=5%
=10%
0.1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Period, Tn (sec) Period, T (sec)
n
Figure 9.19 Influence of post-yield stiffness ratio on (a) ratio of median deformations
um for bilinear and elastoplastic systems and (b) ductility demand; Ry =4.
(a) =3% (b) Ry=4

2 2
Ta Tb Tc Td Te RT=1.5
yf Ta Tb Tc Td Te=0%
Tf
1 Ry=2 1 =1%
Ry=4 =3%
Dispersion of CR
R =6 =5%
0.5 y 0.5 =10%
0.4 0.4
0.3 0.3
0.2 0.2
0.1 0.1
0.05 0.05
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Figure 9.20 Dispersion of inelastic deformation ratio CR for bilinear systems subjected to
the LMSR ensemble of ground motions: (a) influence of strength reduction
factor Ry at post-yield stiffness ratio =3% and (b) influence of at fixed
Ry =4.
185
(a) =4 (b) Ry=4
50 50
40 LMSR 40 LMSR
30 30
LMLR LMLR
20 SMSR 20 SMSR
SMLR SMLR
10 10
Median CR
Median C
5 5
4 4
3 3
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Figure 9.21 Comparison of inelastic deformation ratio for LMSR, LMLR, SMSR, and
SMLR ensembles of far-fault ground motions: (a) C for =4 and =3%;
(b) CR for Ry =4 and =3%.
(a) =4 (b) Ry=4

50 50
40 LMSR 40 LMSR
30 30
B B
20 C 20 C
D D
10 10
Median CR
Median C
5 5
4 4
3 3
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Figure 9.22 Comparison of inelastic deformation ratio for LMSR and site class B, C, and
D ensembles of far-fault ground motions: (a) C for =4 and =3%; (b) CR
for Ry =4 and =3%.
186
(a) (b)
5 5
4 LMSR 4 LMSR
NFFN NFFN
3 NFFP 3 NFFP
Median C
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Period, Tn (sec) T /T
n c
Figure 9.23 Comparison of inelastic deformation ratio C for far-fault LMSR and near-
fault ground motions plotted versus (a) elastic vibration period Tn and (b)
normalized period Tn / Tc ; both plots are for ductility factor =4 and post-
yield stiffness ratio =3%.
(a) (b)
100 100
LMSR LMSR
NFFN NFFN
NFFP NFFP
Median CR
10 10
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100

Period, T (sec) T /T
n n c
Figure 9.24 Comparison of inelastic deformation ratio CR for far-fault LMSR and near-
fault ground motions plotted versus (a) elastic vibration period Tn and (b)
normalized period Tn / Tc ; both plots are for Ry =4 and post-yield stiffness
ratio =3%.
187
(a) =0% (b) =10%
10 10
LMSR Data
Miranda
5 NewmarkHall 5
4 KrawinklerNassar 4
3 3
C
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Figure 9.25 Comparison of inelastic deformation ratio C estimated by equations
developed by various researchers and LMSR data for ductility factor =6:
(a) elastoplastic systems, and (b) post-yield stiffness systems ( =10%).
188
(a) =0%, =2 (b) =10%, =2
10 10
LMSR
NFFN
5 Proposed C 5

4 4
3 3
C
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) =0%, =6 (d) =10%, =6

10 10
5 5
4 4
3 3
C
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
Tn/Tc Tn/Tc
Figure 9.26 Comparison of inelastic deformation ratio C estimated by proposed equation
with computed data for LMSR-far-fault and near-fault (fault-normal

component) ground motions for elastoplastic ( = 0 ) and bilinear ( = 10% )
systems; and =2 and 6.
189
(a) =0%; =2 (b) =10%; =2
10 10
LMSR
LMLR
5 SMSR 5
4 SMLR 4
Proposed C

3 3
C
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) =0%; =6 (d) =10%; =6

10 10
5 5
4 4
3 3
C
2 2
1 1
0.5 0.5
0.01 0.1 1 10 100 0.01 0.1 1 10 100
T /T Tn/Tc
n c
Figure 9.27 Comparison of inelastic deformation ratio C estimated by proposed equation
with computed data for LMSR, LMLR, SMSR, and SMLR far-fault ground
motions for elastoplastic ( = 0 ) and bilinear ( = 10% ) systems; and =2
and 6.
190
(a) =0% (b) =10%
100 100
LMSR Data
Ruiz GarciaMiranda
NewmarkHall
KrawinklerNassar
10 10
CR
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100

Figure 9.28 Comparison of inelastic deformation ratio CR estimated by equations
developed by various researchers and LMSR data for Ry =4: (a) elastoplastic
systems, and (b) bilinear systems ( =10%).
191
(a) =0%; R =2 (b) =10%; R =2
y y
100 100
LMSR
NFFN
Proposed CR
10 10
CR
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) =0%; R =6 (d) =10%; R =6

y y
100 100
10 10
CR
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100

Tn/Tc T /T
n c
Figure 9.29 Comparison of inelastic deformation ratio CR estimated by proposed equation

with computed data for LMSR-far-fault and near-fault (fault-normal
component) ground motions for elastoplastic ( = 0 ) and bilinear ( = 10% )
systems; and Ry =2 and 6.
192
(a) =0%; R =2 (b) =10%; R =2
y y
100 100
LMSR
LMLR
SMSR
SMLR
Proposed C
R
10 10
CR
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100
(c) =0%; R =6 (d) =10%; R =6

y y
100 100
10 10
CR
1 1
0.01 0.1 1 10 100 0.01 0.1 1 10 100

Tn/Tc T /T
n c
Figure 9.30 Comparison of inelastic deformation ratio CR estimated by proposed equation

with computed data for LMSR, LMLR, SMSR, and SMLR far-fault ground
motions for elastoplastic ( = 0 ) and bilinear ( = 10% ) systems; and Ry =2
and 6.
193
10. CONCLUSIONS
The objectives of this research investigation were to evaluate the accuracy of the modal
pushover analysis (MPA) procedure in estimating seismic demands for a wide range of
planar moment-resisting frame and to develop a simpler version of the MPA procedure
for practical implementation.
The estimation of target roof displacement of a multistory frame from the

deformation of equivalent inelastic SDF system, investigated in Chapter 4, has led to the
following conclusions:
Elastic Regular Frames
1. When compared to exact values obtained from rigorous nonlinear RHA, the first-
mode SDF system underestimates the median value of roof displacement over an
ensemble of ground motions. The bias and dispersion of the displacement ratio
(u )
*
r SDF increases for longer-period frames, median values as low as 0.850 for the 15-
story frame, implying that the SDF-system underestimates roof displacement by 15%.
The SDF-system estimate of roof displacement due to individual excitations can be
alarmingly inaccurate.
2. When higher mode contributions to response are included in RSA, the bias and
dispersion of the displacement ratio (u )*
r RSA reduces; the remaining bias and
dispersion is due to approximations associated with modal combination rules.
Inelastic Regular Frames
3. The first mode SDF system estimate of the median roof displacement is biased, as
expected, but the nature and magnitude of this bias depends on how far the structure
is driven into the inelastic range, characterized by an overall ductility demand , For
larger , the SDF-system method overestimates the median roof displacement, and
this bias increases for longer-period systems. The situation is reversed for small ;
194
the SDF-system method underestimates roof displacement, and this bias increases for
longer-period systems.
4. The median values of ( ur* ) ranged from 0.85 to 1.19, implying an underestimation
SDF
by 15% to an overestimation by 19%, respectively.
5. The dispersion of the displacement ratio ( ur* ) tends to increase for taller frames for
SDF
every value of .
6. This large discrepancy arises because for individual grounds motion the SDF system
may underestimate or overestimate the yielding-induced permanent drift in the
exact response determined by nonlinear RHA.
7. While this discrepancy is not improved significantly by including higher mode

contributions, the MPA procedure has the advantage of reducing the dispersion in the
roof displacement and the underestimation of the median roof displacement for elastic
or nearly elastic cases at the expense of increasing slightly the overestimate of roof
displacement of buildings responding far into the inelastic range.
Evaluation of the accuracy of the MPA procedure in estimating the seismic

demands for 60 regular frames due to an ensemble of 20 ground motions (presented in
Chapter 5) has led to the following conclusions:
1. If a sufficient number (2 or 3) of modes are included, the height-wise variation of

story drifts determined by MPA is generally similar to the exact results from
nonlinear RHA. The first mode alone, which is the basis for pushover procedures
currently used in engineering practice, does not adequately estimate seismic demands.
2. While the bias in the MPA procedure tends to increase for longer-period frames and
larger SDF-system ductility factors, these trends are not perfect. The dispersion of
error is smallest in 3-story frames and tends to increase for taller frames because
higher-mode contributions are more significant in their responses. The MPA is more
accurate in estimating seismic demands due to an individual excitation if its bias and
195
dispersion are both small; its estimate is unreliable when the dispersion is larger than
30%.
3. The RSA procedure consistently underestimates the response of elastic structures,

slightly underestimates for shorter frames, and underestimates up to 18% in the
upper-story drifts of taller frames. The bias and dispersion in MPA estimates of
seismic demands for inelastic frames are usually larger than for elastic systems
(determined by RSA procedure), and their height-wise variation is more irregular
because of the additional approximations in the MPA procedure.
Prior to evaluating the accuracy of MPA in estimating the seismic demands for
irregular frames, the effects of vertical irregularitystiffness, strength, or combined-
stiffness-and-strength irregularityon seismic demands of strong-column-weak-beam
frames (beam-hinge model) were investigated in Chapter 6, leading to the following
conclusions:
1. The three types of irregularities similarly influence the heightwise variation of story
drifts, with the effects of strength irregularity being larger than stiffness irregularity,
and the effects of combined-stiffness-and-strength irregularity being the largest
among the three.
2. Introducing a soft and/or weak story (Cases 1, 3, and 5) increases the story drift
demands in the modified and neighboring stories and decreases the drift demands in
other stories. On the other hand, a stiff and/or strong story (Cases 2, 4, and 6)
decreases the drift demand in the modified and neighboring stories and increases the
drift demands in other stories.
3. Drift demands in the upper stories are much more sensitive to irregularities in the
lower stories than the response of lower stories is affected by irregularities in the
upper stories.
4. While the roof displacement is usually insensitive to vertical irregularity (Cases 1-6),
it is significantly different for frames that are stiffness-and-strength irregular in their
lower half (Cases KS7 and KS8). Irregularity in the base story or lower stories
196
(Cases 5-8) has significant influence on the heightwise distribution of floor
displacements.
Concerning accuracy of the MPA procedure in estimating seismic demands for

vertically irregular frames (presented in Chapter 6), the following conclusions were
reached:
1. The bias in the MPA procedure does not increase, i.e., its accuracy does not
deteriorate, in spite of irregularity in stiffness, strength, or stiffness and strength
provided the irregularity is in the top story or mid-height story (Cases 1-4).
2. The MPA procedure can be more biased, i.e. less accurate, relative to the regular
frame in estimating the seismic demands of frames with strong or stiff-and-strong
first story; soft, weak, or soft-and-weak lower half; stiff, strong, or stiff-and-strong
lower half. In contrast, the bias in the MPA procedure for frames with soft, weak or
soft-and-weak first story is about the same as for the regular frame.
3. In spite of the larger bias in estimating drift demands for some stories in Cases 6-8,
the MPA procedure identifies the stories with largest drift demands and estimates
them well, detecting the critical stories in such frames.
4. The dispersion of *MPA for irregular frames is similar to the regular frame, except
in the lower stories of the frames with a strong first story (Cases SM6 and KS6) or
with strong lower half (Cases SM8 and KS8), for which the bias in MPA is also large.
5. The MPA procedure provides reasonably accurate seismic demands also for irregular
frames, except for those with a strong first story or strong lower half. The seismic
demands for such irregular frames should be determined by nonlinear RHA.
The comparative evaluation of MPA and pushover analysis using FEMA force
distributions based on the story-drift demands has led to the following conclusions:
1. MPA is almost always more accurate in estimating drifts in all stories of all frames
than all of FEMA distributions. The envelope of the four estimates corresponding to
197
the four FEMA force distributions is less biased than the individual estimates but is
almost always more biased than MPA. Generally well below 25%, the bias in MPA
exceeds this limit in the upper story drifts in 15-and 18-story frames designed for a
ductility factor of 4 or 6. Seismic demands for such frames should be determined by
nonlinear RHA, not by pushover procedures.
2. The MPA procedure maintains its superiority for almost all of the irregular frames.
The MPA is almost always more accurate compared to all of the FEMA force
distributions and to the envelope of the four FEMA estimates. Neither of the
approximate procedures, however, can estimate seismic demands accurately for
vertically irregular frames with strong or stiff-and-strong first story or lower half;
seismic demands for such systems should be determined by nonlinear RHA.
3. An approximate procedure is likely to be accurate in estimating the seismic demands

due to an individual excitation if both its bias and dispersion are small. In this sense,
the MPA procedure is more reliable than FEMA-356 force distributions, although
pushover procedures are not recommended in conjunction with individual excitations.
A practical version of the MPA procedure has been developed (in chapter 8) to
estimate seismic demands for inelastic systems with earthquake hazard defined by a
median (or some other percentile) design spectrum for elastic systems. In this case, the
deformation of each modal inelastic SDF system is estimated from the deformation of
the corresponding elastic system multiplied by CR , the ratio of deformations of inelastic
and corresponding elastic SDF systems.
The investigation of the median and dispersion of the ratio of deformations

denoted by C for system with known ductility factor and by CR for systems with
known yield strength reduction factor Ry of bilinearly inelastic and corresponding
linear systems over eight ensembles of ground motions (presented in Chapter 9) has led
to the following conclusions:
198
1. Equations have been derived for L and LR , the limiting values of C and CR as the
elastic vibration period Tn of the system tends to zero. L depends on and , the
post-yield stiffness ratio; LR depends on Ry and . For the special case of
elastoplastic systems, L = and LR = . Post-yield stiffness is much more
influential in reducing LR from an unbounded value to a finite valuethan L . For
the limiting case of Tn = , physical reasoning leads to C = C R = 1 , the well-known
equal deformation rule [Veletsos and Newmark, 1960].
2. The median values of C (and CR ) for the seven ensembles of far-fault ground
motions confirm known results that C (and CR ) exceed 1 in the acceleration-
sensitive spectral region, increasing with (and Ry ); 1 in the velocity-sensitive
and displacement-sensitive regions, essentially independent of (and Ry ); except
that they fall below 1 in the period range Td to T f of the latter region, decreasing for
increasing (and Ry ). For very short-period systems ( Tn < Ta ), CR is very sensitive
to the yield strength and can be very large even if the strength of the system is only
slightly smaller than that required for it to remain elastic.
3. In contrast to the prevailing view, based largely on earthquake response of SDF

systems for fixed ductility factor, ignoring post-yield stiffness in estimating
deformation is too conservative for seismic evaluation of existing structures with
known Ry in the acceleration-sensitive region.
4. The dispersion of C is zero as Tn tends to zero or infinity; roughly similar over a
wide period range, Tb to T f ; increasing with ; and affected little by post-yield
stiffness.
5. The dispersion of CR is zero as Tn tends to zero (except for elastoplastic systems) or
Tn tends to infinity is larger for weaker systems over a wide range of periods except
199
for Tn < Tb where this trend is reversed; and is essentially independent of post-yield
stiffness ratio for the range of considered in the velocity- and displacement-
sensitive regions, but is reduced significantly with the slightest post-yield stiffness in
the short-period portion of the acceleration-sensitive region.
6. The median C versus Tn plots for four ground motion ensembles denoted by
LMSR, LMLR, SMSR, and SMLRrepresenting four combinations of large (M =

6.6-6.9) or small (M = 5.8-6.5) magnitude and small (R = 13-30 km) or large (R = 30-
60 km) distanceare very similar; the CR - Tn plots of the four ensembles are also
very close.
7. The median C - Tn and CR - Tn plots are essentially independent of NEHRP site
classes B, C, and D, all of which are firm soil sites.
8. In the acceleration-sensitive spectral region, the median C - Tn and CR - Tn plots for
near-fault ground motions are systematically different than far-fault ground motions;
however, when plotted against normalized period Tn / Tc (where Tc is the period
separating the acceleration- and velocity-sensitive regions), they become very similar
in all spectral regions.
9. Simplified equations for inelastic deformation ratios of C and CR for elastoplastic
and bilinear systems have been developed as functions of Tn / Tc , and or Ry ,
respectively. These equations satisfy the limiting values: L and LR at Tn =0, and 1
at Tn = . Determined by regression analysis of the C and CR data for four
ensembles (LMSR, LMLR, SMSR, and SMLR) of far-fault ground motions, the
numerical values for the parameters in these equations are also valid for near-fault
ground motions. Thus the equations presented for C and CR should be valid for
almost any ensemble of ground motions recorded at firm sites. The equation for C
should be useful in estimating the inelastic deformation of a new or rehabilitated
200
structure where the global ductility capacity can be estimated. Similarly, the equation
for CR should be useful for existing structures with known lateral strength.
This investigation has not evaluated the accuracy of MPA procedure for buildings
subjected to near-fault ground motions or ground motions recorded on soft soils,
buildings with stiffness or strength degradation, e.g. reinforced concrete structures, or
buildings with unsymmetrical plan responding in coupled lateral-torsional motion, in
which torsion effects will occur. All these issues should be investigated in future
research.
201
REFERENCES
Al-Ali, A. A. K., and Krawinkler, H. (1998). Effects of vertical irregularities on seismic

behavior of building structures, Report No. 130, John A. Blume Earthquake
Engineering Center, Stanford University, California, 198 pp.
Allahabadi, R., and Powell, G. H. (1988). DRAIN-2DX user guide, Report No.
UCB/EERC-88/06, Earthquake Engineering Research Center, University of
California, Berkeley, California.
Applied Technology Council (1996). Seismic evaluation and retrofit of concrete

buildings, Report No. ATC-40, Redwood City, California.
Bez, J. I., and Miranda, E. (2000). Amplification factors to estimate inelastic

displacement demands for the design of structures in the near field, CD-Rom
Proceeding of the Twelfth World Conference on Earthquake Engineering, Paper
No. 1561, Auckland, New Zealand.
Benjamin, J. R., and Cornell, C. Allin (1970). Probability, Statistics, and Decision for
Civil Engineers, McGraw-Hill, New York, 684 pp.
Bracci, J. M., Kunnath, S. K., and Reinhorn, A. M. (1997). Seismic performance and
retrofit evaluation for reinforced concrete structures, Journal of Structural
Engineering, ASCE, 123(1):3-10.
Building Seismic Safety Council (1997). NEHRP Guidelines for the Seismic
Rehabilitation of Buildings, FEMA-273, Federal Emergency Management
Agency, Washington, D.C.
Building Seismic Safety Council (2000). Prestandard and Commentary for the Seismic
Rehabilitation of Buildings, FEMA-356, Federal Emergency Management
Agency, Washington, D.C.
Fajfar, P., and Fischinger, M. (1988). N2a method for nonlinear seismic analysis of
regular structures, Proceeding of the Ninth World Conference on Earthquake
Engineering, Tokyo-Kyoto, Japan, 5:111-116.
Chopra, A. K., and Goel, R. K. (1999). Capacity-demand-diagram methods based on

inelastic design spectrum, Earthquake Spectra, 15(4):637-656.
Chopra, A. K. (2001). Dynamics of Structures: Theory and Applications to Earthquake

Engineering, Second Edition, Prentice Hall, Englewood Cliffs, New Jersey, 844
pp.
Chopra, A. K., and Chintanapakdee, C. (2001). Comparing response of SDF systems to

near-fault and far-fault earthquake motions in the context of spectral regions,
Earthquake Engineering and Structural Dynamics, 30(12):1769-1789.
202
Chopra, A. K., and Goel, R. K. (2001). A Modal Pushover Analysis Procedure to
Estimating Seismic Demands for Buildings: Theory and Preliminary Evaluation,
PEER Report 2001/03, Pacific Earthquake Engineering Research Center,
University of California, Berkeley, California.
Chopra, A. K., Goel, R. K., and Chintanapakee, C. (2001). Statistics of SDF-System

estimate of roof displacement for pushover analysis of buildings, PEER Report
2001/16, Pacific Earthquake Engineering Research Center, University of
California, Berkeley, California.
Chopra, A. K., and Goel, R. K. (2002). Modal pushover analysis procedure for
estimating seismic demands for buildings, Earthquake Engineering and
Structural Dynamics, 31(3):561-582.
Clough, R. W. (1966). Effect of stiffness degradation on earthquake ductility

requirements, Report No. SEMM 66-16, Dept. of Civil Engineering, University
of California, Berkeley, California, 67 pp.
Fajfar, P. (2000). A nonlinear analysis method for performance-based seismic design,

Earthquake Spectra, 16(3):573-592.
Fajfar, P., and Fischinger, M. (1988). N2a method for nonlinear seismic analysis of
regular structures, Proceeding of the Ninth World Conference on Earthquake
Engineering, Tokyo-Kyoto, Japan, 5:111-116.
Goel, R. K., and Chopra, A. K. (1997). Period formulas for moment-resisting frame
buildings, Journal of Structural Engineering, ASCE, 123(11):1454-1461.
Goel, R. K., and Chopra, A. K. (2002). Evaluation of MPA procedure using SAC
buildings, Journal of Structural Engineering, ASCE, submitted for publication.
Gupta, A., and Krawinkler, H. (1998) Effect of stiffness degradation on deformation

demands for SDOF and MDOF structures, Proceeding of the Sixth U.S. National
Conference on Earthquake Engineering, Earthquake Engineering Research
Institute, Oakland, California.
Gupta, B., and Kunnath, S. K. (1998). Effect of hysteretic model parameters on inelastic
seismic demands, Proceeding of the Sixth National Conference on Earthquake
Engineering, Earthquake Engineering Research Institute, Oakland, California.
Gupta, A., and Krawinkler, H. (1999). Seismic demands for performance evaluation of
steel moment resisting frame structures (SAC Task 5.4.3), Report No. 132, John
A. Blume Earthquake Engineering Center, Stanford University, California.
Gupta, A., and Krawinkler, H. (2000). Estimation of seismic drift demands for frame
structures, Earthquake Engineering and Structural Dynamics, 29:1287-1305.
203
Gupta, B., and Kunnath, S. K. (2000). Adaptive spectra-based pushover procedure for
seismic evaluation of structures, Earthquake Spectra, 16(2):367-392.
International Code Council. 2000 International Building Code, Falls Church, Virginia,
2000.
Kim, S., and DAmore, E. (1999). Pushover analysis procedure in earthquake

engineering, Earthquake Spectra, 15:417-434.
Krawinkler, H., and Nassar, A. A. (1992). Seismic design based on ductility and
cumulative damage demands and capacities, In: Nonlinear Seismic Analysis and
Design of Reinforced Concrete Buildings, Eds. P. Fajfar and H. Krawinkler. New
York: Elsevier Applied Science.
Krawinkler, H., and Seneviratna, G. D. P. K. (1998). Pros and cons of a pushover

analysis of seismic performance evaluation, Engineering Structures, 20(4-
6):452-464.
Kunnath, S. K., and Gupta, B. (2000). Validity of deformation demand estimates using
nonlinear static procedures, Proceeding of the U.S. Japan Workshop on
Performance-Based Engineering for Reinforced Concrete Building Structures,
Sapporo, Hokkaido, Japan.
Lawson, R. S., Vance, V., and Krawinkler, H. (1994). Nonlinear static pushover
analysiswhy, when and how?, Proceeding of the Fifth U.S. National
Conference on Earthquake Engineering, 1:283-292.
Maison, B., and Bonowitz, D. (1999). How safe are pre-Northridge WSMFs? A case
study of the SAC Los Angeles Nine-Story Building, Earthquake Spectra,
15(4):765-789.
MATLAB: The Language of Technical Computing (1997). Version 5.0, The Mathworks
Inc., Natick, Massachusetts.
Matsumori, T., Otani, S., Shiohara, H., and Kabeyasawa, T. (1999). Earthquake member
deformation demands in reinforced concrete frame structures, Proceeding of the
U.S.-Japan Workshop on Performance-Based Earthquake Engineering
Methodology for R/C Building Structures, Maui, Hawaii, 79-94.
Miranda, E. (1991). Seismic evaluation and upgrading of existing structures, Ph.D.

thesis, Department of Civil Engineering, University of California, Berkeley,
California.
Miranda, E. (1993). Evaluation of site-dependent inelastic seismic design spectra,

Journal of Structural Engineering, ASCE, 119(5):13191338.
Miranda, E. (2000). Inelastic displacement ratios for structures on firm sites, Journal of
Structural Engineering, ASCE, 126(10):11501159.
204
Miranda, E. (2001). Estimation of inelastic deformation demands of SDOF systems,
Journal of Structural Engineering, ASCE, 127(9):1005-1012.
Nassar, A. A., and Krawinkler, H. (1991). Seismic demands for SDOF and MDOF
systems, Report No. 95, John A. Blume Earthquake Engineering Center,
Stanford University, California, 204 pp.
Newmark, N. M., and Hall, W. J. (1982). Earthquake Spectra and Design, Earthquake
Engineering Research Institute, Berkeley, California, 103 pp.
Paret, T. F., Sasaki, K. K., Eilbekc, D. H., and Freeman, S. A. (1996). Approximate
inelastic procedures to identify failure mechanisms from higher mode effects,
Proceeding of the Eleventh World Conference on Earthquake Engineering, Paper
No. 966, Acapulco, Mexico.
Qi, X., and Moehle, J. P. (1991). Displacement design approach for reinforced concrete
structures subjected to earthquakes, Report No. EERC 91/02, Earthquake
Engineering Research Center, University of California, Berkeley, California.
Riddell, R., and Newmark, N. M. (1979). Statistical analysis of the response of

nonlinear systems subjected to earthquakes, Structural Research Series No. 468,
Department of Civil Engineering, University of Illinois, Urbana, Illinois, 291 pp.
Riddell, R., Garcia, J. E., and Garces, E. (2002). Inelastic deformation response of
SDOF systems subjected to earthquakes, Earthquake Engineering and Structural
Dynamics, 31(3):515-538.
Ruiz-Garcia, J., and Miranda, E. (2002). Inelastic displacement ratios for evaluation of
existing structures, to appear in Earthquake Engineering and Structural
Dynamics.
Saiidi, M., and Sozen, M. A. (1981). Simple nonlinear seismic analysis of R/C
structures, Journal of Structural Division, ASCE, 107(ST5):937-951.
Sasaki, K. K., Freeman, S. A., and Paret, T. F. (1998). Multimode pushover procedure
(MMP)A method to identify the effects of higher modes in a pushover
analysis, Proceeding of the Sixth National Conference on Earthquake
Engineering, Earthquake Engineering Research Institute, Oakland, California.
Shimazaki, K., and Sozen, M. A. (1984). Seismic drift of reinforced concrete

structures, Technical Research Report of Hazama-Gumi Ltd., Tokyo, 145166.
Skokan, M. J., and Hart, G. C. (2000). Reliability of nonlinear static methods for the
seismic performance prediction of steel frame buildings, Proceeding of the
Twelfth World Conference on Earthquake Engineering, Paper No. 1972,
Auckland, New Zealand.
205
Somerville, P. G. (1998). Development of an improved representation of near-fault
ground motions, Proceeding of SMIP98 Seminar on Utilization of Strong-Motion
Data, California Division of Mines and Geology, Oakland, California, 1-20.
Song, J. K., and Pincheira, J. A. (2000). Spectral displacement demands of stiffness- and
strength-degrading systems, Earthquake Spectra, 16(4):817-851.
Valmundsson, E. V., and Nau, J. M. (1997). Seismic response of building frames with
vertical structural irregularities, Journal of Structural Engineering, ASCE,
123(1): 30-41.
Veletsos, A. S., and Newmark, N. M. (1960). Effect of inelastic behavior on the

response of simple systems to earthquake motions, Proceeding of the Second
World Conference on Earthquake Engineering, Tokyo, Japan, Vol. II, 895912.
Veletsos, A. S., Newmark, N. M., and Chepalati, C. V. (1965). Deformation spectra for
elastic and elastoplastic systems subjected to ground shock and earthquake
motion, Proceeding of the Third World Conference on Earthquake Engineering,
Wellington, New Zealand, Vol. II, 663682.
Veletsos, A. S. (1969). Maximum deformations of certain nonlinear systems,

Proceeding of the Fourth World Conference on Earthquake Engineering, Chilean
Association on Seismology and Earthquake Engineering, Santiago, Chile, Vol. II,
pp. A4-155 to A4-170.
Veletsos, A. S., and Vann, W. P. (1971). Response of ground-excited elastoplastic

systems, Journal of Structural Division, ASCE, 97, ST4:1257-1281.
Whittaker, A., Constantinou, M., and Tsopelas, P. (1998). Displacement estimates for
performance-based seismic design, Journal of Structural Engineering, ASCE,
124(8):905912.
206
APPENDIX A BILINEAR IDEALIZATION
In UMRHA and MPA presented in Chapter 2, the force-deformation ( Fsn Ln Dn )

relation of equivalent inelastic SDF system is obtained from the pushover curvebase
shear-roof displacement ( Vbn urn ) relation (Fig. 2.6). The pushover curve often consists
of multi-linear segments as a result of successive yielding at different locations in the
structure. Such multi-linear curve is usually idealized as bilinear curve to facilitate
solution of Eq. (2.29). The implementation of bilinear idealization in this report adopted
the criterion specified in FEMA-356 that the first linear segment shall intersect the actual
curve at 60% of the (idealized) yield force; however, this criterion alone can not uniquely
define a bilinear curve. Therefore, we adopted another widely used criterion that the
strain energy (area under the curve) associated with the peak response has to be the same
as for the actual curve. To impose the second criterion, the target roof displacement is
needed, but not yet known, so the idealization process needs to implemented iteratively:
(1) Assume a trial target displacement so that area under the pushover curve can be
calculated
(2) Obtain a bilinear curve that satisfies the two criteria (by any optimization
algorithm, e.g., fmins.m in MATLAB [1997])
(3) Convert the idealized pushover curve to the Fsn Ln Dn relation (Fig. 2.6b) by
utilizing Eq. (2.32)
(4) Compute the peak deformation, Dn , of the nth-mode inelastic SDF system (Fig.
2.5b) with force-deformation relation of Fig. 2.6b by solving Eq. (2.29), or from
the inelastic response (or design) spectrum
(5) Calculate the peak roof displacement urno associated with the nth-mode
inelastic SDF system from Eq. (2.21)
(6) Repeat step (2)-(5) until the peak roof displacement is equal to the value in the
previous iteration.
207
The next example demonstrates that the idealized bilinear curve is sensitive to the
target roof displacement assumed in Step 1. Fig. A.1 shows the pushover curve (dashed
line) for the second mode of an 18-story regular frame with =6 and T1 = TU . After
the above iteration procedure for LMSR record no. 10, 20, and 8, the target roof
displacement converges to 0.90, 2.13, 5.27 in., respectively, as shown on the curve.
Idealized bilinear curves corresponding to peak response for each record are also shown.
The yield base shears vary among the three records (=11.2, 12.0, and 13.2 kips) because
the target displacements for idealization are different; the post-yield stiffness ratios are
also different.
If one does not iterate the above procedure, but assumes a large target
displacement in bilinear approximation, e.g. 5.27, which is appropriate only for record
no. 8, and solve Eq. (2.29) for an SDF system having this bilinear relation subjected to
LMSR record no. 10 or 20, one would obtain target roof displacements equal to 1.27 or
1.98, respectively, which are quite different from the value that is otherwise obtained
from iteration (0.90 and 2.13). This indicates that, without the iteration presented above,
target displacement varies randomly with the arbitrarily assumed target displacement in
bilinear curve idealization; therefore the iteration should be implemented.
25
Record no.8
20
Base shear, Vby2 (kips)
15
Record no.20
Record no.10
10
0
0 1 2 3 4 5 6 7
Roof displacement, u (in)
r2
Figure A.1 Pushover curve for the second mode of an 18-story regular frame with
=6 and T1 = TU ; idealized bilinear versions are shown in solid lines.
208
APPENDIX B DIFFICULTY IN MPA
In Chapter 6, the second-mode pushover analysis is unable to reach the target

displacement for frames with a weak first story (Cases SM5 and KS5) or weak lower half
(Cases SM7 and KS7) when MF=5. This difficulty arises because the frame is
unrealistically weak in lower stories.
The force distribution s*n [Eq. (2.20)] for higher modes ( n >1) consists of forces
in both forward and backward directions (e.g. Fig. B.1a); thus, the story shear does not
increase monotonically from top to bottom (e.g. Fig.B.1b). At some locations along the
height of the frame, the story shear, which equals to the summation of forces in all stories
above, is in the opposite direction to the target roof displacement. When yielding occurs
in such story, the large plastic deformation in this story can cause the roof displacement
to decrease while the applied forces with invariant distribution are increasing. This is
demonstrated in Figure B.2a, which shows the deflected shape of the irregular frame with
a weak first story (Case SM5) under the second-mode force distribution s*2 just before
and after yielding, which occurs at the first story. Figure B.2b shows the pushover curve
of this case and it is seen that the target roof displacement can not be reached because
after yielding occurs the roof displacement decreases as base shear increases.
Although MPA estimate for such irregular frames can not be obtained because of
this difficulty, MPA detects the deficiency, which is the very weak first story or lower
half, in the structure.
209
(a) (b)
12 12
s*2
10 10
8 8
Story
Story
6 6
4 4
2 2
0 0
0.5 0 0.5 1 0.5 0 0.5 1
Story force / Base shear Story shear / Base shear
Figure B.1 (a) Force distribution s*2 and (b) the corresponding story shear of the
irregular frame with a weak first story (Case SM5).
(a) (b)
12 16
14
10
Base shear (kips)
12
8
10
Story
6 8
V =8k
b 6
4 10 k
12 k 4
2 14 k
2
16 k
0 0
0.2 0.1 0 0.1 0.2 0.3 0 0.1 0.2 0.3
Floor displacement (in) Roof displacement (in)
Figure B.2 (a) Deflected shape and (b) pushover curve of pushover analysis of the
irregular frame with a weak first story (Case SM5) under s*2 force
distribution.
210
APPENDIX C REGRESSION EQUATIONS FOR C AND CR
The convenient equations in Section 9.9 to facilitate the estimation of C or CR as a
function of Tn / Tc , or Ry , and was developed by the following approach. First,
observe that C or CR approaches the limiting value L or LR [Eq. (9.7)] as Tn / Tc tends
to zero, and approaches one as Tn / Tc tends to infinity. It is found that equations
1 1
Tn
B
Tn
B
C = 1 + ( L 1) + A
1
or CR = 1 + ( LR 1) + A
1
(C.1)
Tc Tc
can satisfy the limiting values for C and CR , where parameters A and B are functions
of or Ry , and . The variation of parameters A and B against or Ry , and is
examined by performing regression analysis finding the values of A and B that

minimize the sum of the square of errorsdifference between the actual C or CR value
and Eq. (C.1)for fixed values of or Ry , and . Such values for A and B in Eq.
(C.1a) for C are shown in Table C.1 and C.2. Parameter A depends strongly on , and
its variation with can be fitted by
a
A= +c (C.2)
b
Parameter B does not vary significantly with or , so it is assumed to be independent

of and :
B=d (C.3)
Substituting Eqs. (C.2) and (C.3) in the Eq.(C.1) gives

1
a Tn
d
C = 1 + ( L 1) + b + c
1
(C.4)
Tc
The numerical parameters a, b, c, and d were determined from the response data by the
algorithm (fmins.m) in MATLAB [1997] to minimize the residual error, defined as the
211
sum of the squares of differences between the actual data and the regression equation, but
with a penalty for underestimationmultiplying the error by a penalty factor, say 5,
when the error is underestimation to obtain a regression equation that is slightly
conservative: a = 72, b =2.2, c = 1.6, and d = 1.5 using the LMSR data and a = 105,
b = 2.3, c = 1.9, d = 1.7 using the data for four (LMSR, LMLR, SMSR, and SMLR)
ensembles; these parameter values are independent of post-yield stiffness ratio, .
A similar derivation for CR gives
1
a Tn
d
CR = 1 + ( LR 1) + b + c
1
(C.5)
Ry Tc

Nonlinear regression analysis of the LMSR data led to a = 63, b = 2.3, c = 1.7, and
d = 2.3, and of the data for four ensembles of ground motion led to a = 61, b = 2.4,
c = 1.5, and d = 2.4.
212
Table C.1 Parameter A in Eq. (C.1a) for each value of and

0% 1% 2% 3% 5% 10%
1.5 812.41 953.51 1095.51 1254.29 1629.94 2946.12
2 103.91 127.43 155.04 190.09 265.93 562.66
3 20.56 26.68 33.63 41.24 57.52 107.92
4 8.82 11.78 14.97 18.32 25.16 41.39
5 5.54 7.46 9.51 11.47 15.09 22.77
6 4.11 5.64 7.06 8.42 10.75 15.24
8 2.87 3.99 4.96 5.79 7.02 9.21
Table C.2 Parameter B in Eq. (C.1a) for each value of and

0% 1% 2% 3% 5% 10%
1.5 3.413 3.492 3.564 3.630 3.756 4.034
2 2.758 2.868 2.974 3.083 3.262 3.652
3 2.217 2.374 2.509 2.628 2.814 3.156
4 1.910 2.094 2.239 2.360 2.546 2.836
5 1.769 1.959 2.110 2.223 2.388 2.653
6 1.692 1.894 2.036 2.145 2.302 2.545
8 1.633 1.839 1.975 2.081 2.218 2.439
213

Eerc 03 03

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Eerc 03 03

Uploaded by

Copyright:

Available Formats

REPORT NO.

EERC 2003-03 EARTHQUAKE ENGINEERING RESEARCH CENTER

EVALUATION OF THE MODAL PUSHOVER

A Report on Research Conducted Under

UNIVERSITY OF CALIFORNIA, BERKELEY

A Report on Research Conducted Under Grant No. CMS-9812531

Earthquake Engineering Research Center

Comparing the median values of story-drift demands determined by MPA to those

Comparative evaluation of the accuracy of MPA and pushover analysis using

Finally, the MPA procedure is simplified to facilitate its implementation in

influence of post-yield stiffness on the inelastic deformation ratios C and CR is

investigated comprehensively. All data is interpreted in the context of acceleration-

deformation of new or rehabilitated structureswhere the global ductility capacity can

This research investigation is funded by the National Science Foundation under

2. METHODS TO ESTIMATE SEISMIC DEMANDS OF MULTISTORY

3. STRUCTURAL SYSTEMS, GROUND MOTIONS, AND RESPONSE

5. EVALUATION OF MPA PROCEDURE FOR VERTICALLY REGULAR

6. EVALUATION OF MPA PROCEDURE FOR VERTICALLY IRREGULAR

7. COMPARATIVE EVALUATION OF MPA AND PUSHOVER ANALYSIS

8. PRACTICAL IMPLEMENTATION OF MPA.................................................... 133

9. INELASTIC DEFORMATION RATIOS FOR SDF BILINEAR SYSTEMS... 146

9.6 Influence of Earthquake Magnitude and Distance............................................ 157

10. CONCLUSIONS...................................................................................................... 194

1.1 BACKGROUND AND MOTIVATIONS

This pushover analysis procedure is obviously based on two major assumptions:

A modal pushover analysis (MPA) procedure that includes the contributions of

The objectives of this report are to

1.3 REPORT ORGANIZATION

Conventional dynamic analysis and modal pushover analysis (MPA) procedures to

2.2 ELASTIC MULTISTORY BUILDINGS

2.2.1 Modal Response History Analysis (RHA)

The differential equations governing the response of a multistory building to horizontal

as a summation of modal inertia force distributions s n [Chopra, 2001: Section 13.12]:

The effective earthquake forces can then be expressed as

The contribution of the nth mode to s and to p eff ( t ) is:

s n = nmn p eff ,n ( t ) = s n u g ( t ) (2.6)

where the modal coordinate qn ( t ) is governed by

Comparing Eqs. (2.9) and (2.10) gives

and substituting in Eq. (2.8) gives the floor displacements

Any response quantity r ( t ) story drifts, internal element forces, etc.can be

2.2.2 Modal Response Spectrum Analysis (RSA)

rno = rnst An (2.17)

where An is the ordinate A (Tn , n ) of the pseudo-acceleration response (or design)

the nth-mode of the MDF system.

The peak modal responses are combined according to the Square-Root-of-Sum-

2.2.3 Modal Pushover Analysis (MPA)

urno = nrn Dn (2.21)

2.3 INELASTIC MULTISTORY BUILDINGS

2.3.1 Nonlinear Response History Analysis (NL-RHA)

With this generalization for inelastic systems, Eq. (2.1) becomes

Expanding the displacements of the inelastic system in terms of the natural

classical damping-orthogonality property of modes gives

Fsn = Fsn ( q,sign q ) = Tn f s ( u,sign u ) (2.26)

This resisting force depends on all modal coordinates qn ( t ) , contained in q , implying

coupling of modal coordinates because of yielding of the structure.

Equation (2.25) represents N equations in the modal coordinates qn . Unlike Eq.

is rarely solved because it offers no particular advantage over Eq. (2.23).

2.3.2 Uncoupled Modal Response History Analysis (UMRHA)

Fsn = Fsn ( qn ,sign qn ) = Tn f s ( qn ,sign qn ) (2.28)

is related to Fsn ( qn ,sign qn ) because of Eq. (2.11).

Solution of the nonlinear Eq. (2.29) formulated in this manner provides Dn ( t ) ,

Properties of the nth-mode Inelastic SDF System

structure as the structure undergoes displacements u = Dn n with increasing Dn .