Documente Academic
Documente Profesional
Documente Cultură
Abstract: An essential requisite in performance-based contribution of higher modes. Because structural dam-
seismic design is the estimation of inelastic deformation age is directly related to local deformations, a more ratio-
demands in structural members. An increasingly popu- nal approach for seismic evaluation should be based on
lar analytical method to establish these demand values is inelastic displacements rather than elastic forces and sev-
a “pushover” analysis in which a model of the building eral articles on the subject of displacement-based seismic
structure is subjected to an invariant distribution of lat- design can be found in the literature (Bertero et al., 1991;
eral forces. Although such an approach takes into consid- Nassar and Krawinkler, 1991; Moehle, 1992; Priestley
eration the redistribution of forces following yielding of and Calvi, 1997; Aschhiem and Black, 2000; Chopra and
sections, it does not incorporate the effects of varying dy- Goel, 2002).
namic characteristics during the inelastic response. Simple Advances in displacement-based seismic design
modal combination schemes are investigated in this article have, in many ways, contributed to the progress of
to indirectly account for higher mode effects. Because the performance-based design. The emergence of FEMA-
modes that contribute to deformations may be different 356 (2000) has now laid the foundation for the de-
from the modes that contribute to forces, it is necessary to velopment of future performance-based seismic codes.
identify unique modal combinations that provide reliable Inherent in FEMA-356 is the assumption that a nonlin-
estimates of both force and deformation demands. The ear static analysis is more reliable than a linear static
proposed procedure is applied to typical moment frame procedure, and that a regular building with a dominant
buildings to assess the effectiveness of the methodology. first mode response in the elastic state is generally not
It is shown that the envelope of demands obtained from influenced by higher modes. A nonlinear static proce-
a series of nonlinear static analysis using the proposed dure (NSP) incorporates nonlinear material characteris-
modal-combination-based lateral load patterns results in tics in representing the force–displacement response of
better estimation of inter-story drift, a critical parameter the structure. A mathematical model of the building, that
in seismic evaluation and design. includes all significant lateral force-resisting elements,
is subjected to a monotonically increasing “invariant”
lateral force pattern until a predetermined target dis-
placement is reached or the building is on the verge of
1 BACKGROUND AND INTRODUCTION
incipient collapse. The stiffness of elements is revised as
yielding occurs. The capacity curve of the structure is
Modern seismic design is a force-based procedure. How-
determined in terms of base shear versus control node
ever, there is general consensus both in the engineer-
displacement. The internal forces and deformations in
ing and the research community that force-based design
the structural elements are then evaluated. FEMA-356
procedures do not account for force redistribution in
recommends using at least two lateral load patterns that
elements following yielding, and they do not take into
approximately bound the likely distribution of the inertia
consideration the influence of changing dynamic charac-
forces: a uniform load pattern or a modal load pattern.
teristics of the system, particularly those arising from the
The modal load pattern is an approximate representa-
∗ Towhom correspondence should be addressed. E-mail: skkunnath@ tion of the inertial forces in the elastic range, whereas the
ucdavis.edu. uniform load pattern represents their likely distribution
C 2004 Computer-Aided Civil and Infrastructure Engineering. Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, USA,
and 9600 Garsington Road, Oxford OX4 2DQ, UK.
Modal combinations for nonlinear static analysis of building structures 247
in the inelastic range assuming that the structure has mitting a comparison of demand versus capacity. A con-
formed a soft story. Using this basic procedure, numerous tentious issue in CSM is the representation of damping
methods have been developed in an effort to minimize when developing the demand curve to account for in-
the errors and approximations of using a static analysis to elastic effects. When the structure responds inelastically
reproduce a response similar to a dynamic analysis while under the action of seismic forces, the input energy is
maintaining its simplicity. Obviously, a static procedure dissipated by viscous damping and yielding of the struc-
cannot account for inertia effects, damping, ground mo- ture. The net effect of yielding is to efficiently increase
tion characteristics, and effects of system degradation. the overall damping of the system. This total damping is
termed equivalent viscous damping. The determination
of the damping due to yielding of the system is an iter-
1.1 Overview of current pushover procedures
ative process because it involves computation of energy
The term “pushover” analysis of structures is a modern dissipated by damping and the strain energy stored in
variation of the well-known “collapse” analysis that is the system.
based on classical plastic analysis of frame structures. The displacement coefficient method (DCM) has been
However, unlike classical plastic analysis where ulti- adopted by NEHRP in their prestandard for seismic re-
mate strength (typically for gravity loads) is of main habilitation of buildings (FEMA-356, 2000) as the pre-
interest, pushover analysis aims at characterizing the ferred method to determine the expected maximum
lateral strength as well as local deformations in the displacement (or target displacement) for the nonlin-
structure. The concept gained prominence after its in- ear static analysis procedure. The target displacement
troduction in the capacity spectrum method (CSM) by (δ t ) is computed by modifying the spectral displacement
Freeman (1978). CSM uses a pushover analysis to es- (Equation (2)) of an equivalent SDOF system as follows:
tablish the base shear versus control node displacement
Te2
and then converts these quantities into spectral accel- δt = CoC1 C2 C3 Sa g (5)
eration and spectral displacement that is plotted in 4π 2
acceleration-displacement response spectrum (ADRS) The factors Co , C1 , C2 , and C3 are modification factors
format. Sasaki et al. (1998) extend the basic CSM ap- that account for spectral displacement, inelasticity, hys-
proach to account for higher modes. Load patterns teresis shape, and P- effects, respectively. T e is the ef-
based on higher modes are used to generate a series of fective fundamental period computed at a specified se-
pushover curves. The following relationships provide the cant stiffness of the system and T i is the elastic funda-
conversion to ADRS format: mental period. Because the nonlinear static response of
the structure is extremely sensitive to choice of load pat-
Sa,n = αn (Vn /W) expressed in units of g (1)
tern, FEMA-356 recommends using at least two load
c,n patterns that approximately bound the distribution of
Sd,n = (2)
βn c,n the inertia forces along the building height in a seismic
N event. The first is a profile based on lateral forces that
i=1wi i,n
2
αn = 2 (3) are proportional to the total mass at each level called the
N uniform pattern. The second pattern can be a triangular
i=1 wi i,n
one that is dependent on the fundamental period of the
N
wi i,n structure in the direction under consideration or a pat-
βn = i=1
N
(4) tern resulting from a modal combination using SRSS or
i=1 wi i,n
2
CQC modal response combination.
where The inability of regular pushover methods to identify
possible failure mechanisms due to higher mode effects
Sa,n is the spectral acceleration for mode n,
in structures with large periods has led engineers to look
Sd,n is the spectral displacement for mode n,
for other methods that minimize this shortcoming. The
V n is the base shear for mode n,
multi-mode pushover (MMP) described earlier (Sasaki
W is the seismic weight of building,
et al., 1998) is an attempt to introduce higher modes
wi is the seismic weight of floor at level i,
by considering multiple pushover curves derived from
i,n is the modal amplitude at level i for mode n, and
different modal force patterns. The adaptive pushover
c,n ,c,n are displacement and modal amplitude of
method (APM) developed by Gupta and Kunnath (2000)
control node for mode n, respectively.
uses a load pattern that changes depending on the in-
The next step in CSM is to convert the standard re- stantaneous dynamic properties of the system and con-
sponse spectrum (which represents the demand side of siders as many modes as necessary to capture significant
the equation) into ADRS format as well, thereby per- higher-mode effects. An elastic response spectrum for
248 Kunnath
the site-specific ground motion to be used is established 1.2 Unresolved issues in pushover methods
so that the modal forces can be determined for vari-
Pushover methods have been the subject of numer-
ous steps. Story forces, as described in Equation (6), are
ous studies as it found its way into performance-based
then computed following an eigenvalue analysis of the
evaluation documents such as FEMA-356 and ATC-
model.
40. However, the ability of a static procedure to pre-
dict dynamic response raises several significant ques-
Fi j = j φi j Wi Sa ( j) (6)
tions that must be addressed before they can be used
reliably in performance-based seismic evaluation. Kun-
The modal participation factor for the jth mode is j =
i=N nath and John (2000) identify several inconsistencies in
i=1 mi φi j , φ ij is the mass-normalized ( M = 1)
T
the different pushover procedures currently used to es-
mode shape value at level i for mode j, mi is the mass
timate seismic demands. Iwan (1999) raises questions
of story level i, and N is the number of stories. A static
about the validity of pushover methods for pulse-like
analysis is then performed using story forces correspond-
near-fault ground motions. The same study also suggests
ing to each mode independently, resulting in push-and-
that restrictions need to be placed on the use of equiv-
pull forces at different levels. The incremental element
alent damping in CSMs. The most significant issue with
forces, deformations, and story drifts are then computed
a pushover analysis using an inverted triangular or uni-
by SRSS combination of the respective modal quantities
form lateral load pattern is that it fails to account for
for the current step. The accumulated member forces
certain critical higher mode contributions thereby under-
are compared with their respective yield values at every
estimating drift demands in the mid and upper stories.
step. If element yielding is detected, the member and
Kunnath and Gupta (2000) investigated the differ-
global stiffness matrices are updated and a new eigen-
ent lateral load patterns recommended in FEMA-356.
value analysis is carried out. This process continues until
In their study, a typical eight-story reinforced concrete
the specified target displacement is reached.
office building was first subjected to a strong recorded
A number of alternative methods for pushover anal-
ground motion. Details of the building and the ground
ysis have also been investigated. These include methods
motion used are described in Section 3 of the article.
where deformation levels and/or the stiffness state deter-
The resulting nonlinear time–history response was con-
mine the load pattern. Fajfar and Fischinger (1988) sug-
sidered to be the benchmark solution. The same build-
gested using story forces proportional to the deflected
ing model was then analyzed using a pushover analy-
shape of the structure. Eberhard and Sozen (1993) used
sis with all three lateral load patterns recommended in
force patterns based on mode shapes derived from the
FEMA-356. The target displacement for the pushover
secant stiffness at each load step and were able to demon-
analyses was the maximum computed roof displacement
strate the effectiveness of the method for shear-wall
in the time–history evaluation. Figure 1 shows the peak
structures. An enhanced capacity-spectrum-based ap-
proach, wherein demands and capacities are estimated
at the story level as opposed to overall base shear ver- 8 8
sus top story displacement, was investigated by Bracci NTH NTH
7 7 NSP-1
et al. (1997). Other variations of these approaches can be NSP-1
NSP-2 NSP-2
found in published articles by Saiidi and Sozen (1981), 6 6 NSP-3
NSP-3
Qi and Moehle (1991), Biddah et al. (1995), and Kilar
STORY LEVEL
STORY LEVEL
5 5
and Fajfar (1997).
An important fact that emerges from the above discus- 4 4
sion is that there exists the need to predict expected de- 3 3
formations across the height of the structure using some
reliable analytical procedure. 2 2
Given the fact that a pushover method is a static 1 1
method which does not require the selection of ground
0 0
motions and the modeling effort in building a computer 0 10 20 30 40 50 0 1 2 3 4
model of a building is significantly less than for a time– DISPLACEMENT (CM) STORY DRIFT (% )
history method, engineers are more comfortable with
pushover procedures than nonlinear time–history meth- Fig. 1. Peak displacement and drift profile using different
ods. Given this appeal of pushover procedures, it is rea- analytical methods (NTH: nonlinear time–history analysis;
sonable to assume that pushover analyses will become NSP-1: pushover analysis using inverted triangular load;
commonplace in seismic evaluation compared to time– NSP-2: uniform load; NSP-3: lateral loads derived from
history methods. modal combination).
Modal combinations for nonlinear static analysis of building structures 249
Fig. 2. Plastic hinge locations predicted using various analysis methods (circles represent plastic hinges).
displacement profile and the peak inter-story drift pro- mands are then combined to study the overall response
file obtained with the different analyses. In this case, the of a building structure. An interesting alternative to com-
peak displacements are generally well represented by bining modal contributions was used by Matsumori et al.
the uniform load pattern (NSP-2). The remaining two (1999), wherein the combination is performed by nu-
patterns underestimate the displacements at almost all merically adding and subtracting the contribution of dif-
levels. The plot of peak inter-story drift, on the other ferent modes. To better estimate maximum earthquake
hand, clearly highlights the inability of “all” nonlinear demands, they used two patterns of story shear distri-
static methods to predict this critical deformation pa- butions: the sum of two modal story shears and the dif-
rameter. The importance of the inter-story drift demands ference of two modal story shears. In their study, they
in the upper stories is further demonstrated in Figure 2, considered only the first and second modes. Their article
which shows the predicted plastic hinges in the frame. suggests that the pushover analysis was carried out till
The potential for the formation of a story mechanism the first mode displacement was equal to the maximum
in the upper stories can be overlooked with traditional earthquake response and that separate pushover anal-
pushover techniques. yses were carried out for each mode. The application
An important consideration in evaluating a pushover of the methodology on two building models produced
method, therefore, is its ability to predict inter-story good correlation with demands resulting from nonlinear
drifts rather than roof displacements. Consequently, the time–history analyses.
concept of a roof ductility factor is not meaningful in the
design or assessment of structures because the control- 2.1 Rationale for the methodology
ling failure mechanism may be a local story mechanism.
Based on these and similar findings from related stud- The central idea behind the proposed modal combi-
ies, it appears that lateral load patterns such as those nation scheme is best illustrated through an example.
specified in FEMA-356 are adequate only for structures Returning to the example of the eight-story frame struc-
with a first-mode dominant response. Further, current ture considered in Section 1.2, the difficulties in using
FEMA guidelines to establish the limits of a pushover static procedures to replace dynamic processes becomes
procedure are also flawed because they are based on a obvious when examining a snapshot of the time–history
modal approximation in the undamaged, elastic state of response of inter-story drift for the building. Figure 3 dis-
the structure. plays the variation in inter-story drift for approximately
There is growing evidence that FEMA-356 prescribed
pushover analyses are becoming commonplace in seis- 8
mic evaluation. The introduction of nonlinear static 6 1st
STORY DRIFT (CM)
loading can obviously vary as a function of time both in response. If the first three modes were being combined,
terms of amplitude and spatial distribution. The objec- the following combinations would be used:
tive of deriving an expression of the form given by Equa-
F j = α1 1 m1 Sa (ζ1 , T1 ) ± α2 2 m2 Sa (ζ2 , T2 )
tion (12) is to separate the spatial distribution from the
time-varying amplitude function. ± α3 3 m3 Sa (ζ3 , T3 ) (15)
The next step is to introduce features of the earth-
The procedure, therefore, requires multiple pushover
quake loading. Because the procedure being developed
analyses, wherein a range of modal load patterns are
is a static one, the most appropriate form of earth-
applied. In each case, the pattern itself is invariant. To
quake loading that can be considered is a response
arrive at estimates of deformation and force demands,
spectrum. The spatial distribution of lateral forces to
it is necessary to consider peak demands at each story
be used in conjunction with a pushover analysis is ap-
level and then establish an envelope of demand values
proximated in terms of the peak modal contributions, as
for use in performance-based evaluation.
follows:
Two questions that obviously arise from the above for-
mulations are: precisely what modes should be included
fn = n mn Sa (ζn , Tn ) (13)
in the combination and how are the modification fac-
where Sa is the spectral acceleration for the given earth- tors to be assigned? Preliminary studies carried out on
quake loading at a frequency corresponding to the pe- structures ranging in height from 4 to 20 stories indicate
riod, T and damping ratio, ζ for mode n. that the number of modes to be included is a function of
The modal forces computed using Equation (13) will the height of the structure. While a single (fundamental)
represent the contributions to mode n only. In the ap- mode is adequate for low-rise structures, more modes
proach proposed by Chopra and Goel (2002), it is as- need to be included for taller structures. No more than
sumed that the inelastic response can also be approx- three modes were necessary to obtain a conservative
imated by modal superposition because the nth mode envelope of the demand values in all cases considered.
is expected to be dominant, even for inelastic systems. Sample results and pertinent observations from an ongo-
The advantage with such an approach is that an inelas- ing study are reported here to illustrate the methodology
tic response spectrum may then be used to estimate the and to highlight features of the procedure that will en-
peak inelastic displacements for each mode. The validity hance our understanding of pushover analyses in general
of the procedure, though illustrated for a single building and modal combination techniques in particular.
and a single earthquake, needs to be demonstrated for a
variety of structural configurations and varying ground
motion characteristics. 3 STRUCTURAL MODELS AND GROUND
The modal combination procedure involves identify- MOTIONS
ing appropriate modes to include in the analysis and the
manner in which the combination will be carried out. In 3.1 Building details
general, the spatial variation of the applied forces will be An eight-story and a 16-story reinforced concrete frame
computed from the following expression: building were analyzed to investigate important features
of the proposed modal combination procedure. The plan
nn
view for both buildings is shown in Figure 5. The building
Fj = αmmmm Sa (ζm, Tm) (14)
m=n1,n2
Table 1
Details of members for various frames used in validation study.
Table 2
Dynamic properties of buildings
0.4 6 6
Story Level
0 5
Story Level
5
-0.4 4
4
-0.8
3
3
2
-1.2 2
0 2 4 6 8 10 1
TIME (SEC) 1 Mode 1 0
0 -600 -300 0 300 600
2.5
5% damped spectra 0 300 600
SPECTRAL ACCL. (g)
0.5
Pushover analyses were carried out using the two
0.0 modal patterns shown in Figure 7 independently. Next,
0 1 2 3 4 5 the modal combinations shown in Figure 8 were used as
PERIOD (SEC) the applied lateral forces. In each case, the lateral loads
were incrementally applied (as a small fraction of the
Fig. 6. Time history and spectrum of ground motion used in loads shown) until the roof displacement was the same as
benchmark analysis. that obtained in the nonlinear time–history analysis. The
resulting displacement profiles are shown in Figure 9.
resulting eigenvectors, the relative modal contributions The displacement amplification at the top story levels is
(which is equivalent to the relative modal mass contribu- obvious in the second mode response for the first case
tion) were compared. It was determined that the modal and in the modal difference combination for the latter.
participation in the first two modes was 55.7 and 21.0%, The main problem with the responses obtained from the
respectively, thus contributing to almost 80% of the to- independent modal pushovers is the need to develop a
tal response. Addition of the third mode would include procedure to combine the individual responses.
about 90% of the modal contributions to the response. If
Equation (13) is used to estimate story forces to be used
in the lateral load analysis, the resulting spatial distribu- 8 Mode 1 + Mode 2 8
tion is displayed in Figure 7. Only the first two modes are
7 7
shown. Next, the proposed modal combination scheme is
used. For the eight-story structure under consideration, 6 6
only the first two modes are used. A closer observation
Story Level
5
Story Level
F j = 1 m1 Sa (ζ1 , T1 ) ± 2 m2 Sa (ζ2 , T2 ) (16) Lateral Force (kN) Lateral Force (kN)
The resulting modal forces to be used in the pushover Fig. 8. Spatial distribution of lateral forces using modal
analyses are shown in Figure 8. combination.
254 Kunnath
Story Level
5
5 16 16
Story Level
Each Mode
Combination 14 14
4
Time History
12 12
3 10 10
Story Level
Story Level
2 8 8
6 6
1
4 4
Mode (1) + (2)
Mode 1
0 2 2
Mode 2 Mode (1) - (2)
0.000 0.025 0.050 0.075
0 0
Interstory Drift Ratio -400 0 400 800 -500 0 500 1000
Lateral Force (kN) Lateral Force (kN)
Fig. 10. Envelope of peak drift estimates for different
approaches. Fig. 11. Lateral forces used in pushover procedures.
Modal combinations for nonlinear static analysis of building structures 255
Story Level
factors may be needed to better approximate the con-
8 8
tribution of higher modes. Rather than using random
6 6 combinations that improve the drift estimates, a system-
atic approach needs to be developed so as to identify
4 4 features of the different modes that are critical to the
Time History
Time History
2
Mode 1
Mode 2 (a)
response. The basis for such a method is explored in the
2 Mode (1) + (2)
Mode 2 (b)
Mode (1) - (2) next section.
0 0
0 100 200 300 0 50 100 150 4.2.1 Advanced combination schemes. The most critical
Displacement (cm) Displacement (cm) aspect of structural response that needs to be understood
if static procedures are to be used to predict a dynamic
Fig. 12. Comparison of displacement profile using different process is the mechanics of inter-story drift and the fac-
methods.
tors that influence its magnitude. The next two figures
attempt to examine some of the features of inter-story
Displacement estimates resulting from the pushover drift. Figure 14 presents the variation of inter-story drift
analysis using second mode forces alone overestimate at each level throughout the time–history response for
displacements at the lower levels and underestimate the eight-story structure. Figure 15 does the same for
them at upper levels. Results of the modal combination the 16-story building. The peak drifts in the lower levels
consisting of the difference of modes generally underes- occur at approximately the same time during the same
timate the floor displacements at all levels. A comparison cycle of response. This cycle also leads to the first in-
of the maximum inter-story drifts using the different ap- elastic excursion in the system. Higher modes do not
proaches (Figure 13) identifies several shortcomings in seem to play a significant role early in the response. The
all methods. drift ratios at the upper levels happen later in the re-
The use of independent modes to estimate lateral sponse and suggest the role of inelastic behavior in mod-
forces can make the task of interpreting the response ifying the modal contributions of the higher modes. A
a challenging task. Drifts are grossly overestimated at similar phenomenon is observed in the response of the
most levels. Even if the modal contribution of the sec- 16-story building (Figure 15). This implies that a modal
ond mode is incorporated when estimating the response, combination resulting from an eigenvalue analysis of the
the resulting drifts will provide a poor measure of the ac- structure in the elastic state is probably valid only for
predicting the drifts at the lower levels. To properly in-
16 Time History
Mode 1
16 Time History corporate higher mode effects, the distribution of modal
Mode 2 (a)
Mode 2 (b)
Mode (1) + (2) forces in the inelastic state of the system is needed. To fa-
14 14 Mode (1) - (2)
cilitate a better understanding of the variation of modal
contributions, eigenvalue solutions were obtained at sev-
12 12
eral discrete steps throughout the time–history analysis
10 10 for both buildings. The duration of the record used in
Story Level
Story Level
1st Level
0.04 2nd Level
0.02 3rd Level
-0.01
-0.03
-0.05
4th Level
0.04
5th Level
0.02 6th Level
0.00
-0.02
-0.04
7th Level
0.02
8th Level
0.01
0.00
-0.01
-0.02
1st Level
0.000
2nd Level
3rd Level
-0.050
4th Level
0.000
5th Level
6th Level
-0.050
0.050
7th Level
0.000 8th Level
9th Level
-0.050
0.040
10th Level
0.000 11th Level
12th Level
-0.040
8 8-STORY BUILDING M M*
My
MODAL PERIODS (SEC)
7 ko
6 T1
T2
5
T3
4
T4
3 T5 χ Μy
2 Φ ko
T6
1
0
0 2 4 6 8 10 12
TIME (SEC)
8 s
16-STORY BUILDING
MODAL PERIODS (SEC)
7
6 T1
α Μy
T2
5
T3
4 Fig. 17. Hysteresis model used in nonlinear time–history
T4
3 analysis.
T5
2 T6
1 Figure 16 suggests that a snapshot of the lateral force
0 distribution at two stages of the analysis should be ade-
0 2 4 6 8 10 12 quate to capture the participation of the different modes.
TIME (SEC) An additional factor to keep in mind is the earthquake
demands at these modal periods. The fundamental (first
Fig. 16. Shifts in the first six modal periods during mode) period experiences a significant change and a cor-
earthquake response. responding decrease in seismic demands, whereas the
higher modal periods are not altered to the same extent.
However, earthquake response is characterized by This leads to the conclusion that when considering higher
cyclic action, which returns the system to a more stable
Table 3
behavior. Clearly, the variation in dynamic characteris-
Modal combinations used to develop drift profile shown in
tics is a function of the material and hysteretic models
Figure 18
used in the analysis. A more complex model with mul-
tiple paths will result in significantly more variations. In
Time Modal Modal
the present study, the multi-parameter hysteretic model
Story history combination combination
(Figure 17) in IDASS was used. This model uses several level IDR envelope value case
control parameters to establish the rules under which
inelastic loading reversals take place. For example, α, 16 0.0138 0.032 α 1 M1 + M2
which can be expressed as a function of the deformation, 15 0.0229 0.044 α 1 M1 + M2
controls the amount of stiffness loss; φ and χ control the 14 0.0288 0.049 α 1 M1 + M2
initiation and degree of pinching; and the slope s and the 13 0.0272 0.049 α 1 M1 – M2
12 0.0298 0.047 α 1 M1 − M2
change in expected peak strength (M to M∗ ) controls
11 0.0346 0.042 α 1 M1 − M2
the softening due to system deterioration. For the anal- 10 0.0438 0.034 M1 − M2
ysis of the buildings in this study, the following param- 9 0.0464 0.043 α 1 M1 + α 2 M2 + α 3 M3
eters were used: α = 0.5 (nominal degradation); = 0; 8 0.0450 0.048 α 1 M1 + α 2 M2 + α 3 M3
χ = 1.0 (no pinching); and an energy-based strength 7 0.0469 0.044 α 1 M1 + α 2 M2 + α 3 M3
deterioration parameter β = 0.01 (see IDASS manual, 6 0.0488 0.039 α 1 M1 − α 2 M2 + α 3 M3
Kunnath, 1995). Hence, the inelastic behavior model 5 0.0481 0.047 α 1 M1 − α 2 M2 + α 3 M3
used in the present analysis is reasonably complex and 4 0.0467 0.049 α 1 M1 − α 2 M2 + α 3 M3
the frequency variation in the different modes shown in 3 0.0469 0.049 α 1 M1 – M3
Figure 16 is representative of the changes in the dynamic 2 0.0435 0.049 α 1 M1 + α 2 M2 − α 3 M3
characteristics of the building as it responds inelastically. 1 0.0286 0.034 α 1 M1 + α 2 M2 − α 3 M3
It must be pointed out, however, that eigenvalue samples Note: M1 = Mode 1; M2 = Mode 2; M3 = Mode 3; α 1 = 0.08;
were taken only at discrete time intervals of 0.25 seconds. α 2 = 0.46; and α 3 = 0.46.
258 Kunnath