Professional Documents
Culture Documents
SUMMARY
A building with a seismic isolation system, in an earthquake, is recognized as producing substantially smaller
accelerations and deformations compared with a building that use other systems. This type of system is there-
fore expected to better protect the building’s nonstructural components, equipment, and other contents that
are essential for the activities conducted in the building. Unlike many available studies on building responses,
only a small number of studies on a buildings’ nonstructural component responses are available, and
no study has directly addressed building performance with regard to nonstructural component protection.
This paper therefore measures the performance of various seismically isolated buildings. Specifically, the
effects of important structural parameters, namely, isolation stiffness, isolation damping ratio, and number
of stories on the response of base-isolated structures are investigated parametrically. Ground motions with
2% exceedence in 50 years Maximum Considered Earthquake (MCE) are used. Performance is compared
with that of fixed-base structures in order to present data that will be useful in justifying the more costly tech-
nology. The buildings are 3, 9, and 20 stories, represented by MDOF shear-beam models. As examples of
displacement-sensitive and acceleration-sensitive components, partition walls and ceilings are considered,
respectively. The Pacific Earthquake Engineering Research Center performance-based earthquake engineer-
ing methodology is adopted to evaluate the failure return periods of the examples based on their available
fragility curves. In addition, the curves are varied hypothetically to understand the sensitivity of the return
period to the curve features. Then, the median and dispersion of fragility curves required to satisfy the
components’ desired failure return period are obtained. Copyright © 2015 John Wiley & Sons, Ltd.
KEY WORDS: PBEE performance; continued functionality; base-isolated structure; sensitivity analysis;
nonstructural component
1. INTRODUCTION
Because of its excellent ability to protect buildings, their functions and their occupants, seismic iso-
lation has been increasingly used in Japan, China, and other seismically active countries. The main
feature of base isolation technology is that it introduces flexibility at the base story between the
superstructure and the ground. This allows for extensive concentrated movement at the base story
and reduces demands on the superstructure. In order to reduce this extensive movement at the base,
isolators are often designed to absorb energy by adding devices with energy dissipation capacity,
resulting in the addition of damping to the system. Energy dissipation can usually be achieved
by fluid viscosity (velocity-dependent damper) or hysteresis of the force-displacement relationship
(deformation-dependent damper). Reviews of several types of base isolation systems have been well
described in [1–3].
Correspondence to: Sarun Chimamphant, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama,
Kanagawa 226-8503, Japan.
E-mail: chimamphant.s.aa@m.titech.ac.jp
Several studies on seismic isolation have typically modeled buildings as two degree of freedom
(2DOF) systems [4–9]. The modeling, however, is not suitable for performance evaluation, because
it does not simulate story-by-story responses such as inter-story drifts and floor accelerations, which
can unevenly be distributed throughout the building height. It is known that story responses can con-
tain multiple modes and affect the structural and nonstructural story-by-story components, equipment,
other contents, and occupants in various ways. Clearly, modeling seismically isolated buildings using
an MDOF system is essential for performance evaluation. Several studies utilized MDOF system.
Nagarajaiah and Sun [10] studied the response of a seven-story base-isolated University of Southern
California (USC) hospital building under Northridge Earthquake. They reported that the main rea-
son for the effectiveness of base isolation is the period lengthening. Matsaga and Jangid [11] studied
the effects of isolation hysteresis (viscoelastic and bilinear) on the response of one-story and five-
story base-isolated buildings. Effective isolation periods of 2, 2.5, and 3 s and effective damping ratios
of 0.05 and 0.1 were considered in their study. They found that superstructure acceleration becomes
higher, whereas bearing displacement is smaller for bilinear isolator. Alhan and Gavin [12] studied the
influence of isolation damping and stiffness on the response of an eight-story L-shaped base-isolated
building with linear and nonlinear isolation systems. They found that increasing damping decreases
the isolator displacement but increases story drift and acceleration in the superstructure. In short, the
aforementioned studies addressed effects of hysteretic properties and irregular building plan on the
responses of the multi-story buildings from 1 to 8 stories. The present study, on the other hand, con-
siders up to a much taller buildings of 20-story and focuses on ‘performance’, which could be a more
common scale to non-engineers. Furthermore, by varying building height, isolation period, and damp-
ing ratio, the study investigates their effects on the responses in a systematic and parametric manner
unlike the aforementioned references. In order to provide comprehensive performance metrics, it is
needed to convert the story responses to the damage states of the specific structural/nonstructural com-
ponents. Moreover, the superiority of isolation systems is effectively shown, and thereby, their use can
be justified to building owners by comparing their performance metrics with those of less costly and
more common conventional structures.
Based on the discussions earlier, the present paper, for the first time, attempts performance
evaluations of seismically isolated buildings by conducting time history analyses of their MDOF
models. From among the numerous available evaluation methods, the Pacific Earthquake Engineering
Research Center (PEER) performance-based earthquake engineering (PBEE) [13–15] is selected, and
the results are explained in comparison with conventional fixed-base buildings that are evaluated in the
same manner. A key feature of the PEER methodology is the definition of performance metrics that are
relevant to decision-making for seismic risk mitigation. In brief, the PEER methodology estimates the
mean annual frequency of exceeding a specific damage state for a specific structure in a given seismic
environment. It is based on dynamic response time history analyses of a building that is subjected to a
suite of earthquake ground motions that are scaled to intensities in accordance with the seismic hazard
curve at the site.
For time history analyses, inelastic shear-beam (stick) models are utilized to represent the MDOF
systems. Stick models do not provide information on each component of the building, but they can
produce the story-by-story responses that are necessary for performance evaluation. These models are
considered appropriate, because they have small degrees of freedom and they considerably reduce com-
putational effort, which otherwise could be excessive; they also exactly represent structural properties
such as story stiffness and strength, which are closely associated with code requirements. Moreover,
even though the performance of tall, flexible, base-isolated buildings is thought to be inefficient,
they are frequently adopted in Japan. This paper, therefore, also investigates performance including
tall, flexible buildings. As examples for the performance evaluation, suspended ceilings and partition
walls, which are sensitive to acceleration and deformation, respectively, will be used. Using the PEER
methodology, the return periods of the sample nonstructural components are estimated based on their
available fragility curves. In addition, the curves are varied hypothetically to understand the return
period’s sensitivity to curve features. Then, the medians and fragility curve dispersions that are required
to satisfy the desired failure return periods of these components are obtained for both seismically
isolated and fixed-base buildings.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 7
0
wx hax wx hax
Fx0 D Vs0 and Fx D Vs (1a,b)
P
n
0 P
n
wi hai wi hai
i D1 i D1
where wi and wx D the portion of the structure’s total effective seismic weight located or assigned to
level i or x, hi and hx D the height from the base to level i or x. Additionally, a0 and a D an exponent
related to the structure period Ts 0 and Ts and Vs0 and Vs D the design base shear for fixed-base and
base-isolated structures, respectively.
According to the code, 1 6 a0 6 2 varies according to the fundamental vibration period of the
fixed-base structure, and Fx0 may not be a straight line when the building is tall. In contrast, a D 1 for
base-isolated structures, resulting in the triangular force profile. The mass of each floor mi .i D 1 to N /
is assumed to be equal, and the mass of the base floor is defined as mb . Story heights are assumed to
be equal on all stories.
K 0s u D Cd0 F 0 (2)
where F 0 D the force vector from Fx0 (Eq. 1a) and u D the displacement vector that indicates a uniform
story drift of 0.02 along the structure height. The calculated vibration period Tsj0 and mode shape 0sj
of the structure for j th mode are then obtained by solving the eigenvalue problem in Equation (3):
2 !2 3
4K 0s 2
M s 5 0sj D 0 (3)
Tsj0
where M s is the diagonal mass matrix known from the story mass that was explained earlier, repre-
senting m1 to mN . Note that the mode shape 0s1 will not be triangular if the force profile F 0 is not
triangular, that is, a0 ¤ 1. Use of a calculated period Ts 0 that is longer than the approximate period Ts 0
but shorter than 1.4Ts 0 is permitted by the code. Generally, the calculated period Ts1 0
will be longer
0
than the approximate period Ts . Hence, the reduction of design force is allowed, which will render the
structure more flexible. However, this approach is not utilized. A previous study [17] finds that most
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
8 S. CHIMAMPHANT AND K. KASAI
Mode
Structure 1 2 3 4 5 6 7
3-story 0.665 0.271 0.172 — — — —
9-story 1.579 0.633 0.400 0.293 0.232 0.192 0.164
20-story 3.034 1.182 0.746 0.546 0.431 0.356 0.304
Mode
Structure 1 2 3 4 5 6 7
3-story 0.02 0.02 0.0261 — — — —
9-story 0.02 0.02 0.0262 0.0335 0.0410 0.0487 0.0564
20-story 0.02 0.02 0.0264 0.0338 0.0415 0.0494 0.0574
of the measured building periods for moment-resisting steel-frame structures fall in the range of Ts 0
and 1.6Ts 0 . Therefore, in this study, the objective is to bring the structure to the period 1.4Ts 0 which
corresponds to the code limits [16] and observation described in [17]. To achieve this, the stiffness
0 0 2
matrix K 0s is revised by using Ts1 = min Ts1 ; 1:4Ts 0 to scale the stiffness entries. It can be seen
0
that if Ts1 < 1:4Ts 0 , the scaling factor will be 1.0, whereas if Ts1 0
> 1:4Ts 0 , the scaling factor will
be greater than 1.0, resulting in higher stiffness than was previously obtained but still giving a period
equal to 1.4Ts 0 . Nonlinear behavior is considered as will be explained in Section 2.4.
In this study, 3, 9, and 20 stories are considered. Considering a story height of 3.5 m, the approximate
formula gives Ts 0 D 0:0724h0:8 D 0:475; 1:14; 2:17 s, respectively. The calculated periods Ts1 0
from
the eigenvalue analysis are 0.85, 1.58, and 3.07 s, respectively, and the upper limit periods are 1:4Ts 0 D
0:665; 1:60; and 3:03 s, respectively. These data show that the Ts1 0
values are slightly greater than 1.4Ts 0
0
except for the nine-story building ( 1:39Ts ), all of which are consistent with the observation in [17].
Table I lists the actual natural periods of the considered fixed-base structures.
The viscous coefficients are determined considering a Rayleigh damping model, that is
C 0s D ˛ 0 M s C ˇ 0 K 0s (4)
where C 0s D the damping matrix and ˛ 0 and ˇ 0 D the mass-proportional and stiffness-proportional
damping coefficients that are to be set for the desired damping ratios at two selected vibration periods,
0 0
Ts1 and Ts2 . A superstructure damping ratio of s0 D 0:02 is assumed. Table II lists the modal damping
ratios of the considered fixed-base structures.
2
2
K s s D M s s (5)
Ts
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 9
where M s is the diagonal uniform mass matrix and s is a triangular shape, because the force profile
from the code is triangular, and the displacement vector with desired uniform drift ratio will always
produce a triangular mode shape. This obtained K s is then checked against Cd F (Eq.1b) to ensure that
the displacement satisfies the maximum uniform drift ratio of 0.015 of the story height as specified by
the code. Note that Cd is equal to RI D 2 [16]. And the superstructure is assumed to behave linearly.
As for the damping, it has been found [18, 19] that with the assumption of Rayleigh damping in
the superstructure, the first-mode damping ratio of the base-isolated structure tends to exceed the iso-
lated damping ratio, resulting in undesirable suppression of the first-mode response. Thus, stiffness
proportional damping is recommended and, as such, utilized in this study, that is
C s D ˛K s (6)
where C s D the damping matrix and ˛ D the stiffness-proportional damping coefficient that is to be
set for the desired damping ratio at Ts . A superstructure damping ratio of s D 0:02 is assumed in
order to be consistent with the fixed-base structure.
For the isolation system, base isolation periods Tb D 2; 3; and 4 s and base isolation damping
ratios b D 0.1 and 0.3, respectively, are considered to cover a wide range of base-isolation prop-
erties. Note that Tb and b are defined by assuming a rigid superstructure, and the corresponding
isolation stiffness kb becomes .m/ .2=Tb /2 when m D total mass mb C N mi . The damping of the
isolation system is idealized by linearly viscous model, and the corresponding isolation damping coef-
ficient cb becomes .m/ .4b =Tb /. This linearly viscous model gives the same period and damping
ratio for all ground motions. This study does not consider nonlinear isolation system, which could alter
superstructure behavior because period and damping ratio change according to displacement.
Tables III and IV list the natural periods and damping ratios of the considered base-isolated struc-
tures. The modal strain energy method is used to estimate the damping ratios for the real value
approach. Note also that complex eigenvalue analysis is also conducted, and that the results of the
natural periods and damping ratios are included in both tables.
Table III indicates that for short buildings, for example, three stories, the fundamental periods T1
from the complex-value approach are nearly the same as those from the real-value approach. For the
more flexible 9 and 20 stories with b D 0:1, the T1 values are still very close to those from the
real-value approach. However, when b D 0:3, T1 reduces to approximately 0.98 times owing to the
effect of nonproportional damping. Additionally for the three-story building with Tb D 2, T1 is nearly
equal to Tb . However, for the 9-story and 20-story buildings, T1 increases to 1.05 and 1.33 times,
respectively, owing to the flexibility in superstructures.
Tb D 2 s Tb D 3 s Tb D 4 s
Structure Method T1 T2 T3 T1 T2 T3 T1 T2 T3
Real 2.015 0.188 0.104 3.010 0.188 0.104 4.007 0.189 0.104
b D 0:1
Complex 2.014 0.188 0.104 3.010 0.188 0.104 4.007 0.189 0.104
3-story
Real 2.015 0.188 0.104 3.010 0.188 0.104 4.007 0.189 0.104
b D 0:3
Complex 2.010 0.188 0.104 3.006 0.189 0.104 4.005 0.189 0.104
Real 2.149 0.514 0.290 3.099 0.526 0.292 4.074 0.531 0.293
b D 0:1
Complex 2.145 0.514 0.290 3.096 0.526 0.292 4.072 0.531 0.293
9-story
Real 2.149 0.514 0.290 3.099 0.526 0.292 4.074 0.531 0.293
b D 0:3
Complex 2.108 0.511 0.291 3.068 0.526 0.293 4.049 0.531 0.293
Real 2.729 1.003 0.605 3.501 1.078 0.624 4.378 1.115 0.632
b D 0:1
Complex 2.724 1.000 0.604 3.493 1.076 0.624 4.369 1.114 0.632
20-story
Real 2.729 1.003 0.605 3.501 1.078 0.624 4.378 1.115 0.632
b D 0:3
Complex 2.666 0.962 0.585 3.405 1.055 0.619 4.282 1.104 0.631
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
10 S. CHIMAMPHANT AND K. KASAI
Tb D 2 s Tb D 3 s Tb D 4 s
Structure Method 1 2 3 1 2 3 1 2 3
Real 0.098 0.043 0.063 0.099 0.039 0.061 0.099 0.038 0.060
b D 0:1
Complex 0.098 0.043 0.063 0.099 0.039 0.061 0.099 0.038 0.060
3-story
Real 0.293 0.067 0.074 0.297 0.055 0.069 0.298 0.049 0.066
b D 0:3
Complex 0.294 0.067 0.074 0.297 0.055 0.069 0.299 0.049 0.066
Real 0.081 0.064 0.079 0.091 0.055 0.073 0.095 0.050 0.070
b D 0:1
Complex 0.081 0.064 0.079 0.091 0.055 0.073 0.095 0.050 0.070
9-story
Real 0.241 0.127 0.116 0.272 0.099 0.098 0.284 0.083 0.089
b D 0:3
Complex 0.245 0.130 0.119 0.275 0.100 0.099 0.286 0.084 0.089
Real 0.044 0.073 0.093 0.065 0.071 0.086 0.077 0.066 0.081
b D 0:1
Complex 0.044 0.072 0.093 0.065 0.071 0.087 0.077 0.066 0.081
20-story
Real 0.118 0.155 0.160 0.188 0.150 0.138 0.229 0.132 0.121
b D 0:3
Complex 0.115 0.148 0.163 0.189 0.154 0.144 0.231 0.135 0.124
From Table IV, the damping ratios from the complex-value approach are very close to those from
the real-value approach, demonstrating the reasonable accuracy of the modal strain energy method for
the considered base-isolated buildings of varying heights. In addition, the actual first-mode damping
ratios 1 indicated in the table are approximately 0.98, 0.8, and 0.4 times b for the 3-story, 9-story, and
20-story structures, respectively. 1 can be much lower than b assuming rigid superstructure. In the
present paper, b will be used to categorize base-isolated buildings, but this point must be kept in mind.
The multiple modal values listed in Tables III and IV will be useful in understanding the performance
that will be explained later in the present paper.
where SDS is the design spectral response acceleration parameter in the short period range, which is
equal to 1.13g; SD1 is the design spectral response acceleration parameter at a period of 1.0 s, which
is equal to 0.9g; R is the reduction factor, which is equal to 8; Ie is the importance factor, which is
one; Ts 0 is the superstructure period; and Ws is the total superstructure weight. From the design base
shear in Equation (7), considering 0 D 3 described earlier, the yield displacements for each story can
be obtained from K 01 0 F 0 . The structure is modeled to exhibit bilinear behavior with a post-yield
stiffness 0.1 times the initial stiffness at each story level.
For the base-isolated structure [16], the period of the superstructure above the isolation level is
approximated as Ts D 0:1N , which was mentioned previously in Section 2.3. However, unlike with the
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 11
fixed-base structure, the design superstructure shear is not determined by using Ts but instead by using
the stiffness and maximum force of the isolator at the design displacement from the code equation
as follows:
kb DD
Vs D (8)
RI
where kb is the lateral stiffness of the isolator; RI is the reduction factor, which is equal to 3/8 of the
R for the fixed-base structure
and must be less than 2 (hence, RI D 2); and DD is the design displace-
ment, which is gSD1 Tb = 4 2 B [16] where B is the damping coefficient. In this study, the isolation
system is idealized by using a viscoelastic system, and thus, isolation period Tb indicates the base
isolation period linearized from nonlinearity at a certain displacement. The response will certainly vary
according to the hysteretic behavior of the isolation system. However, in order to study the fundamen-
tal trends in base-isolated structures, an idealized viscoelastic system is utilized. Substituting DD into
Equation (8) and further manipulating it by using kb D m .2=Tb /2 , where m is the total mass of the
building including the isolation level, produces Equation (9):
kb gSD1 Tb =4 2 1 m .2=Tb /2 gSD1 Tb 1 W SD1 1
Vs D D 2
D W (9)
RI B RI .2/ B mg RI Tb B
where W is the total building weight including the isolation level and B is the damping coefficient
related to isolation damping ratio b (Section 2.4). The relationship between Vs0 and Vs is shown in the
following Equation (10):
Vs R W 1 SD1 1 Ts 0
D max ; (10)
Vs0 RI Ws B SDS Tb Tb
As mentioned earlier, R=RI D 4, SD1 =SDS D 0:794, and W=Ws D .N C 1/ =N . Hence, from
Equation (10), it is possible to observe a trend that indicates when Vs =Vs0 > 1 or vice versa by varying
Ts 0 and Tb . Table V shows the values of Vs =Vs0 for multiple structure cases.
Table V clearly shows that Vs =Vs0 increases with taller structures, smaller damping, and shorter
Tb , and vice versa. This means that base-isolated superstructures tend to be stronger than fixed-base
structures, although considering the fundamental periods, they are in fact subjected to less acceleration.
Therefore there is a very high chance that base-isolated superstructures will remain elastic. Note that
normally the effective modal mass is less than 1.0, but we assumed it to be 1.0 for comparison.
Figure 1 shows the required yield strength for both fixed-base and base-isolated structures, with the
same overstrength factor I D 0 D 3 assumed in the present study. It can be observed that the ratio
of the median of the 20 ground motions (Section 2.5) to the required fixed-base structure yield strength
levels is very large, and severe nonlinearity is expected for such buildings. However, for base-isolated
structures, the ratio is about 1.0 or less, much lower than the ratios for fixed-base structures. Note that
the required yield strengths for any base-isolated superstructure (3, 9, or 20 stories) will be the same,
and all are calculated at isolation period Tb according to the code. Because of this, the elastic behavior
of base-isolated superstructures is assumed in the present study.
Isolation damping ratio D 10%, B D 1:2 Isolation damping ratio = 30%, B D 1:7
3-story 9-story 20-story 3-story 9-story 20-story
0 0 0 0 0
Tb Ts D 0:475 Ts D 1:144 Ts D 2:167 Ts D 0:475 Ts D 1:144 Ts 0 D 2:167
2 1.765 2.118 3.792 1.246 1.495 2.677
3 1.176 1.412 2.528 0.830 0.997 1.784
4 0.882 1.059 1.896 0.623 0.748 1.338
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
12 S. CHIMAMPHANT AND K. KASAI
Figure 1. Required yield strength as designed according to the American Society of Civil Engineers code.
Figure 2. Pseudo acceleration spectra of SAC ground motions (LA21–LA40) . D 0:05 and 0:30/.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 13
where xm represents the median; ˇ represents the dispersion; M is the number of ground accelera-
tions; and rj is the unique response for each ground motion. Equation (11a,b) assume the data are
sampled from a lognormal distribution. The median is a mathematical result that indicates that one-half
of the group is higher and one-half is lower, and dispersion indicates the degree of data scatteredness.
Various combinations of isolation and superstructure properties of base-isolated structures are
subjected to ground accelerations, and their responses are discussed in this chapter.
Figure 3. (a) Median drift ratio and (b) median floor acceleration of base-isolated structures.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
14 S. CHIMAMPHANT AND K. KASAI
whereas for taller 9-story and 20-story base-isolated buildings, the median drift ratios are approxi-
mately 0.2% and 0.5%, which are 4 and 10 times greater, respectively. When superstructure period
Ts is longer, that is, the building is taller, drift ratios and peak floor accelerations become more con-
centrated at the upper stories because of the higher mode contribution. In addition, because of the
code-specified triangular force profile, the story shears at the upper stories appear to be insufficient,
resulting in low stiffness. In practice, the design of the beams for the upper stories will govern and
the strength and stiffness of the upper stories will increase, which consequently reduces such high
accelerations. For hysteretic isolation system, we found that sudden change of stiffness could cause
higher acceleration. Also, the previous work by Matsagar and Jangid [11] showed that using a bilinear
isolator indicated higher accelerations at the upper floors, even higher than viscoelastic isolator. Their
study also showed that top floor acceleration tends to increase when the building is taller regardless of
the isolation hysteresis.
Figure 4. Transfer functions for floor accelerations: (a) Afloor at the roof floor and (b) Afloor at the mid-floors
(b D 0:1 only).
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 15
Figure 4(b,d,f) shows the transfer functions when Aoutput .!/ is the acceleration from one-third the
building height, two-thirds the building height, and the top floors but only for structures with b D 0:1.
Looking at the one-third building height floors, that is, first (3 stories), third (9 stories), and seventh
(20 stories) floors, the contributions from the second and third modes decrease. However, at the two-
thirds building height, that is, 2nd (three stories), 6th (nine stories), and 14th floors, the second mode
is low, but the third mode increases. From this information, the building structure and contents, such as
partition walls and suspended ceilings, should be designed with careful attention to the higher modes
as well.
This chapter compares the median drift ratios and accelerations of fixed-base structures with those of
the base-isolated structures from the previous chapter.
Figure 5. Comparative responses of the fixed-base and base-isolated structures (b D 0:1 only): (a) median
drift ratios and (b) median floor accelerations.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
16 S. CHIMAMPHANT AND K. KASAI
period of the whole structure. Rather, it excites higher modes owing to the significant differences in
damping ratios.
In short buildings, the response can be reduced even more if the isolation system is more flexible
by increasing the isolation period Tb . For tall buildings, if the isolation period is appropriately larger
than the superstructure period, a reduced response will be observed. This indicates that adding an
isolation system will always reduce the superstructure’s drift and acceleration responses if Tb =Ts is
greater than 1.
Figure 6. Floor response spectra at the roof of the fixed-base and base-isolated structures: (a) median
acceleration at b D 0:1 and (b) median acceleration at b D 0:3.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 17
The acceleration response results in smaller reduction or even greater amount for the 9-story and
20-story structures because the isolation period Tb of 2 s is approaching the superstructure period Ts of
0.9 and 2 s. Therefore, for taller buildings, improvement is usually achieved by increasing the isolation
period Tb such that it is practical to implement from the superstructure period Ts ; improvement would
still be observed. Despite the slight improvement, this serves as a proof that adding a flexible isolation
system always improves the superstructure that rests on it when the appropriate isolation period Tb
is assigned.
In order to evaluate the performance of fixed-base and base-isolated structures, the PEER PBEE
methodology [13–15] is utilized. In this framework, seismic damage, structural, and hazard analy-
ses are conducted concurrently. There are three types of assessments [15], which are intensity-based,
scenario-based, and time-based assessments. In this study, time-based approach, which is most
rigorous, is utilized in the evaluation. The damage to the nonstructural components of drift-sensitive
partition walls and acceleration-sensitive suspended ceilings are considered. The damage state of
interest is ‘continued functionality .CF/’, meaning that a building remains usable without interruption
after an earthquake. The equations used to evaluate performance in summation form are Equations (12)
for drift-sensitive components and (13) and (14) for acceleration-sensitive components. Note that
Equation (13) uses floor acceleration as a demand parameter, which is appropriate when considering
rigid nonstructural components that are anchored to the floor, whereas Equation (14) uses component
acceleration, which is appropriate for flexible nonstructural components and can be obtained by time
history analysis of the component using floor acceleration. Because suspended ceilings are considered,
Equation (14) will be used. CF is the annual frequency of exceeding the continued functionality
damage state. Once CF is obtained, the expected return period TR (year) can be calculated by taking
the reciprocal of CF .TR D 1=CF /. The explanation of these equations is described as follows:
8 9
N <
1 X M
X ® ¯=
CF D D P ŒCFjık P ık jsj P sj (drift-sensitive) (12)
TR : ;
kD1 j D1
8 9
N <
X M
X =
1 ® ¯
CF D D P ŒCFjak P ak jsj P sj (acceleration-sensitive) (13)
TR : ;
kD1 j D1
8 9
N <
1 X XM
® ¯=
CF D D P CFjac;k P ac;k jsj P sj (acceleration-sensitive) (14)
TR : ;
kD1 j D1
The first term in each of Equations (12) and (14) indicates damage analysis based on experimental
data and damage observations of both structural and nonstructural components, and it describes the
likelihood of exceeding a certain damage state as a function of seismic demand. A number of such
fragility curves have been collected and published in the Federal Emergency Management Agency
(FEMA) P-58 [15]. However, in the present study, the fragility curves for both components are not
obtained from the database provided in [15] but from the more recently available studies described in
the succeeding paragraphs.
For the partition wall, experiments on light-gauge steel stud gypsum partition walls were conducted
recently at the University of Buffalo to assess their seismic fragility [21]. Fifty partition wall specimens,
corresponding to 22 different wall configurations, were constructed following standard construction
techniques. The fragility curves reported in that study are adopted here. The damage description cor-
responding to CF is ‘slight damage to partition walls’. For this damage state, the median drift ratio
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
18 S. CHIMAMPHANT AND K. KASAI
and dispersion are 0.35% and 0.56, respectively. Similar median drift ratio and dispersion values also
appear in the fragility curve database [15].
For the suspended ceiling, from the recent work by Motoyui et al. [22], the fragility of Japanese
ceiling systems is obtained through Monte Carlo analysis using 2D finite element analysis and the
assumed statistical variation in connection strengths from experimental testing. The period Tc D 0:31 s
and damping ratio c D 0:03 for the suspended ceiling are specified. The fragility curve that describes
‘failure of ceiling’ that was reported in their study is used. The median acceleration and dispersion
are 0.505g and 0.046, respectively. Note that the demand parameters used in all acceleration-sensitive
component fragility curves in the database [15] are peak floor accelerations a. However, for non-
rigid ceilings, component acceleration ac is generally greater than floor acceleration. Thus, component
acceleration is used for the damage evaluation, as shown in Equation (14). Also note that it is also
recognized by several works [23–25] that vertical motions could damage nonstructural components and
contents, which would affect the functionality of base-isolated buildings. On the other hand, fragility
curve for suspended ceiling does not include vertical acceleration, and we believe that the process
is underway. When this reaches a mature stage, our method can be applied to address the vertical
acceleration case, too. In such a case, however, the model of a building must be made much more
sophisticated to reflect effect of vertical acceleration at each floor, which needs to consider effects of
many connected beams and slabs. This could be included in the future work.
The second term in each of Equations (12) and (14) indicates the distributions of drift ratios and
component accelerations under the 20 ground motions defined in Section 2.5 for a specified intensity,
respectively. To cover a wide range of intensities, incremental dynamic analysis (IDA) is performed
by gradually increasing the level of seismic intensity by scaling the ground motions. The scaling is
performed at the structure’s fundamental vibration period. The intensity ranges Samin 6 Sa 6 Samax
for the IDA are selected to conform with FEMA P-58 [15]. That is, Samin .T / D 0:05g for T 6 1 s
and 0:05g=T otherwise, where T is the fundamental structure period. Meanwhile, Samax is taken at
D 0:0002, where is the annual frequency of exceeding ground shaking intensity. A total of 300
intensity levels between Samin and Samax of the 20 ground motions are used for the incremental dynamic
analysis. The structures investigated are the same as in Chapter 4, and thus, a total of 21 different T
values (3 from fixed-base and 18 from base-isolated structures) are considered. Because partition walls
and suspended ceilings are used for performance evaluation, drift ratio and component acceleration
demand parameters are of interest.
Figure 7. Incremental dynamic analysis results and probability density functions of the nine-story fixed-
base structure.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 19
Figure 7(a,c,d) shows examples of the IDA results for the nine-story fixed-base structure (T D
1:58 s) at the roof, considering drift ratio, peak floor acceleration, and the ceiling’s response accelera-
tion, respectively. Similarly, Figure 8(a,c,d) shows the results of the nine-story base-isolated structure
with Tb D 3 s and b D 0:10 (T D 3:1 s). Each curve in the figures represents the structure’s response
to a single ground motion whose intensity is increased until the specified Samax . Considering the drift
ratio at intensity Sa D 1:17g (Figure 7(a)) and 0.50g (Figure 8(a)), corresponding to the 2% in 50 years
ground motions for the fixed-base and base-isolated structures, respectively, the probability density
function can be obtained as shown in Figures 7(b) and 8(b), which will be used in the second term in
Equation (12). It is shown clearly that the median drift ratio for the base-isolated structure is signifi-
cantly lower than that for the fixed-base structure. Considering the acceleration at the same intensity
level (2%/50 years), it is clearly shown that for nonrigid suspended ceilings, the component accelera-
tion is greater than the peak floor acceleration, by approximately 2.8 times at the median values shown
in Figure 7(c,d) and 1.6 times in Figure 8(c,d). Because of the greater medians and dispersions in the
component acceleration, any acceleration-sensitive nonstructural components used in the building may
need to be stronger to compensate for such large median and dispersion values. The probability den-
sity functions can then be obtained as shown in Figures 7(e) and 8(e), which will be used in the second
term in Equation (14). Note that area under each curve in Figures 7(b,e) and 8(b,e) is equal to one.
Figure 8. Incremental dynamic analysis results and probability density functions of the nine-story base-
isolated structure (Tb D 3, b D 0:10).
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
20 S. CHIMAMPHANT AND K. KASAI
The last term in both Equations (12) and (14) indicates the probability of the occurrence of the
ground shaking intensity that was obtained from the hazard curve, which indicates annual frequency of
exceeding . The hazard curves are obtained by specifying the site’s location (longitude and latitude)
in OpenSHA [26]. In this study, the site is assumed to be located in downtown Los Angeles, where
the latitude and longitude used to specify the location are 34.053 and 118.243 degrees, respectively.
The attenuation model by Campbell and Bozorgnia [27] is used, with VS30 D 360 m/s representing
the boundary between C and D soil types, as defined in ASCE 7 (2010) [16]. Figure 9(a) shows the
hazard curves for the periods of 0.5, 1, 2, 3, 4, and 5 s. The probability density function for these curves
is approximated by the difference between the annual frequencies
of exceeding
the ground-shaking
intensities at each end of the interval divided by its interval, Saj Saj C1 = Saj C1 Saj .
The results are shown in Figure 9(b).
Now, TRı and TRa are defined, indicating the return period TR considering drift-sensitive and
acceleration-sensitive components, respectively. The higher the return period TR , the better the
performance. This chapter discusses TR for inelastic fixed-base and base-isolated structures.
Figure 10. Return periods TRı considering the partition wall’s continued functionality.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 21
Figure 11. Median component acceleration of the fixed-base and base-isolated structures.
Figure 12. Return periods TRa considering the suspended ceiling’s continued functionality.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
22 S. CHIMAMPHANT AND K. KASAI
If the ceiling’s period Tc were 0.5 s, the component accelerations at the roofs for the 3-story, 9-story,
and 20-story base-isolated structures would be 0.75g, 1.18g, and 0.79g, respectively (Figure 6). The
component acceleration ratios, Sa .Tc D 0:5/ =Sa .Tc D 0:31/, are 1.01, 1.51, and 1.49, respectively.
These ratios indicate that if the fragility curve is the same, the TRa will decrease very slightly for the
3-story structure and more significantly for the 9-story and 20-story structures from the TRa obtained
for Tc D 0:31 s. From this, it can be interpreted that if the nonstructural component’s period falls close
to one of the structure’s modal vibration periods, performance could considerably worsen.
For fixed-base structures, the return periods TRa are approximately 2, 3, and 4 years for the 3-story,
9-story, and 20-story structures, which are extremely shorter than those of the base-isolated structures.
Theese TRa indicates that there is a very high likelihood that in a few years, the suspended ceiling will
fall and disrupt the continued functionality of the building.
In the previous chapter, only a single fragility curve was considered. However, there could be multi-
ple fragility curves even for a single component because of the variety of component types and sizes.
Moreover, there are also uncertainties for fragility itself owing to the limitations of the testing con-
ditions that can accurately represent realistic behavior. Therefore, it is worth conducting sensitivity
analyses of the fragility curves. The two main statistical parameters for both partition walls and sus-
pended ceilings that are used to construct fragility curves, median xm and dispersion ˇ, are examined
for the 3-story, 9-story, and 20-story base-isolated structures with Tb D 2 s and b D 0:10. For sus-
pended ceilings, the same Tc D 0:31 s is used. The procedure in the succeeding paragraphs, however,
will apply to any case of Tc if the corresponding fragility curve is given. Note also that there is no need
to perform any structural analysis at all because only the first term in Equations (12) and (14) changes;
therefore, the results can be obtained quickly. Figures 13 and 14 show the fragility curves with varied
medians and dispersions for partition walls and suspended ceilings, respectively.
Figure 13. Fragility curves of partition walls: (a) fixed dispersion ˇ of 0.56 and (b) fixed median xm
of 0.35%.
Figure 14. Fragility curves of suspended ceilings: (a) fixed dispersion ˇ of 0.046 and (b) fixed median xm
of 0.505g.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 23
Figure 15. Sensitivity of return period TRı for 3-story, 9-story, and 20-story base-isolated structures to: (top)
median xm and (bottom) dispersion ˇ.
Figure 16. Sensitivity of return period TRa for 3-story, 9-story, and 20-story base-isolated structures to: (top)
median xm and (bottom) dispersion ˇ.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
24 S. CHIMAMPHANT AND K. KASAI
Figure 15(bottom) shows the case of varying the dispersion ˇ-values. For the three-story base-
isolated structure, compared with the TRı of the original dispersion ˇ D 0:56, the TRı for ˇ D 0:4 and
0.7 will be approximately 5.72 times and 0.32 times greater, respectively. For the nine-story structure,
they are 1.56 and 0.71 times greater, respectively, and for the 20-story structure, they are 1.13 and 0.91
times greater. Thus, TRı is highly sensitive to ˇ for the three-story structure and less sensitive for the
9-story and 20-story structures.
Figure 17. Performance curves considering the suspended ceiling continued functionality of the base-
isolated structures at the roof: (a) 3 stories, (b) 9 stories, and (c) 20 stories.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 25
Figure 18. Performance curves considering the suspended ceiling continued functionality of the base-
isolated structures at the roof: (a) 9 stories (Tb D 3 s) and (b) 20 stories (Tb D 4 s).
0.85g and 0.75g, respectively, with ˇ D 0:05 as shown in Figure 18, or xm would need to be higher if
ˇ is large. As a reminder, this is the performance curve at the roof only. If the roof is not of importance,
other floors could be selected instead.
8. CONCLUSIONS
This research focuses on studying the performance of base-isolated and fixed-base structures with
multiple combinations of isolation and superstructure properties while considering the structures’ CF.
Results for the base-isolated and fixed-base structures subjected to the MCE level ground motions (2%
in 50 years) are discussed. The effects of modal contributions are shown through the transfer func-
tions as well as the floor response spectra. Then, the PEER PBEE performance methodology is utilized
to evaluate the mean annual frequency of exceeding the CF damage, leading to the expected return
period TR , which is used as a performance measure index. This CF state is used to represent a perfor-
mance target in which the structure remains intact and the damage does not cause any interruption in
building use. Additionally, the last chapter extends the performance evaluation to cover a wide range
of fragility curves of nonstructural components to understand the sensitivity of their performance to
the fragility curves.
The vibration periods and damping ratios of the base-isolated structures that were estimated from the
real-mode and complex-mode eigenvalue analyses are very close. Comparing the isolation period Tb
with isolation damping ratio b , which are assigned based on the rigid superstructure assumption stipu-
lated by the US code, the first-mode period T1 is greater than the isolation period Tb , and the first-mode
damping ratio 1 is smaller than the isolation damping ratio b , and this can be significant for taller
buildings. The assumption of base-isolated structures’ elastic behavior is explained using a figure that
shows that base-isolated structures are actually stronger than fixed-base structures. Base-isolated struc-
tures with longer isolation periods always show smaller drift ratios and less acceleration. In contrast,
structures with higher isolation damping ratios do not necessarily show reduced responses owing to the
higher mode participation, which increases the drift ratio and acceleration responses. Compared with
fixed-base structures, however, isolation systems are highly efficient, even with tall structures.
The performance evaluation methodology is explained in detail, considering damage to partition
walls and suspended ceilings caused by story drift ratios and component accelerations. Note that the
component acceleration is higher than the floor acceleration and thus is used in the evaluation process.
Compared with fixed-base structures, performances are excellent for short base-isolated structures
and less distinct but still better for taller isolated structures. These performance results are obtained
considering a single fragility curve of the partition wall and suspended ceiling only. Sensitivity to the
fragility curves characterized by strength xm and dispersion ˇ is examined. The return period appears
to be very sensitive in the short structures and less sensitive in taller structures.
The last section demonstrates a new application of the PEER methodology to determine the neces-
sary strength xm and dispersion ˇ that will satisfy the target performance level .TR /. xm and ˇ are
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
26 S. CHIMAMPHANT AND K. KASAI
plotted with respect to TR , which is convenient for estimating desired median strengths and disper-
sions in order to produce nonstructural components. Moreover, the plot will be useful for selecting and
designing particular nonstructural components during the initial design phase.
ACKNOWLEDGEMENTS
Financial support from the Center for Urban Earthquake Engineering (CUEE), Tokyo Institute of Technol-
ogy is acknowledged. The authors are grateful to Dr Troy A. Morgan for his constructive criticism during
the initial stage of the research.
REFERENCES
1. Jangid RS, Datta TK. Seismic behavior of base-isolated buildings: a state-of-the-art review. Proceedings of the ICE
- Structures and Buildings 1995; 110(2):186–203.
2. Patil SJ, Reddy GR. State of art review - base isolation systems for structures. International Journal of Emerging
Technology and Advanced Engineering 2012; 2(7):438–453.
3. Warn GP, Ryan KL. A review of seismic isolation for buildings: historical development and research needs.
Buildings 2012; 2(3):300–325.
4. Kelly JM. The role of damping in seismic isolation. Earthquake Engineering and Structural Dynamics 1999;
28(1):3–20.
5. Makris N, Chang SP. Effect of viscous, viscoplastic and friction damping on the response of seismic isolated
structures. Earthquake Engineering and Structural Dynamics 2000; 29(1):85–107.
6. Jangid RS, Kelly JM. Base isolation for near-fault motions. Earthquake Engineering and Structural Dynamics 2001;
30(5):691–707.
7. Ordoñez D, Foti D, Bozzo L. Comparative study of the inelastic response of base isolated buildings. Earthquake
Engineering and Structural Dynamics 2003; 32(1):151–164.
8. Politopoulos I. A review of adverse effects of damping in seismic isolation. Earthquake Engineering and Structural
Dynamics 2008; 37(3):447–465.
9. Yang TY, Vamvatsikos D, Kelly JM. The influence of isolator hysteresis on equipment performance in seismic
isolated buildings. Earthquake Spectra 2010; 26(1):275–293.
10. Nagarajaiah S, Sun X. Response of base-isolated USC hospital building in Northridge earthquake. Journal of
Structural Engineering (ASCE) 2000; 126(10):1177–1186.
11. Matsagar VA, Jangid RS. Influence of isolator characteristics on the response of base-isolated structures. Engineering
Structures 2004; 26(12):1735–1749.
12. Alhan C, Gavin H. A parametric study of linear and non-linear passively damped seismic isolation systems for
buildings. Engineering Structures 2004; 26(4):485–497.
13. Moehle J, Deierlein GG. A framework methodology for performance-based earthquake engineering. In Proceedings
of the 13th World Conference on Earthquake Engineering. Vancouver: Canada, 2004.
14. Yang TY, Moehle J, Stojadinovic B, Kiureghian AD. Seismic performance evaluation of facilities: methodology and
implementation. Journal of Structural Engineering (ASCE) 2009; 135(10):1146–1154.
15. Federal Emergency Management Agency (FEMA P-58). Seismic Performance Assessment of Buildings, Applied
Technology Council: 201 Redwood Shores Parkway, Suite 240 Redwood City, California 94065, August 2012.
16. American Society of Civil Engineers (ASCE 7). Minimum Design Loads for Buildings and Other Structures. Reston:
Virginia, 2010.
17. Goel RK, Chopra AK. Period formulas for moment-resisting frame buildings. Journal of Structural Engineering
(ASCE) 1997; 123(11):1454–1461.
18. Ryan KL, Polanco J. Problems with Rayleigh damping in base-isolated buildings. Journal of Structural Engineering
(ASCE) 2008; 134(11):1780–1784.
19. Hall JF. Problems encountered from the use (or misuse) of Rayleigh damping. Earthquake Engineering and
Structural Dynamics 2006; 35(5):525–545.
20. Somerville P, Smith N, Punyamurthula S, Sun J. Development of ground motion time histories for phase 2 of the
FEMA/SAC steel project. Technical Report SAC/BD-97/04, SAC Joint Venture, Sacramento, California, October
1997.
21. Retamales R, Davies R, Mosqueda G, Filiatrault A. Experimental seismic fragility assessment of light gauge steel
studded gypsum partition walls. Proceedings of the 9th US National and 10th Canadian Conference on Earthquake
Engineering, Toronto, Canada, 2010.
22. Motoyui S, Sato Y, Macrae GA, Dhakal RP. Ceiling fragility of Japanese ceiling systems. Proceedings of the 9th
Pacific Conference on Earthquake Engineering, Auckland, New Zealand, 2011.
23. Soroushian S, Ryan KL, Maragakis M, Wieser J, Sasaki T, Sato E, Okazaki T, Tedesco L, Zaghi AE, Mosqueda G,
Alvarez D. NEES/E-defense tests: seismic performance of ceiling / sprinkler piping nonstructural systems in base
isolated and fixed base building. Proceedings of the 15th World Conference on Earthquake Engineering, Lisbon,
Portugal, 2012.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe
COMPARATIVE PERFORMANCE OF STRUCTURES 27
24. Furukawa S, Sasaki T, Sato E, Okazaki T, Ryan KL. Comparison of vertical dynamic response characteristics of two
base-isolated buildings based on full-scale shaking table test. Proceedings of the 13th World Conference on Seismic
Isolation, Sendai, Japan, 2013.
25. Furukawa S, Sato E, Shi Y, Becker T, Nakashima M. Full-scale shaking table test of a base-isolated medical facility
subjected vertical motions. Earthquake Engineering and Structural Dynamics 2013; 42(13):1931–1949.
26. Field EH, Jordan TH, Cornell CA. OpenSHA: A developing community-modeling environment for seismic hazard
analysis. Seismological Research Letters 2003; 74(4):405–419.
27. Campbell KW, Bozorgnia Y. NGA ground motion model for the geometric mean horizontal component of PGA,
PGV, PGD and 5% damped linear elastic response spectra for periods ranging from 0.01 to 10 s. Earthquake Spectra
2008; 24(1):131–171.
Copyright © 2015 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2016; 45:5–27
DOI: 10.1002/eqe