Winter 2010, Vol. 11, No.

4
JSEE
ABSTRACT
Available online at: http://www.iiees.ac.ir/jsee
The dynamic response of flexible five-story building supported on the variable
frequency pendulum isolator (VFPI) under bi-directional near-fault ground
motions is investigated. In order to verify the effectiveness of the VFPI, the seismic
responses are compared with the friction pendulum system (FPS) and variable
friction pendulum system (VFPS). The response of the system with bi-directional
interaction is compared with those without interaction in order to investigate the
effects of bi-directional interaction of frictional forces. Moreover, a parametric
study is carried out to critically examine the influence of important parameters on
bi-directional interaction of the frictional forces of the VFPI. From the above inves-
tigations, it is concluded that under bi-directional near-fault ground motions, the
isolator displacement in the VFPI is more than that of the VFPS and the FPS whereas
the top floor absolute acceleration and the base shear are less than that of the VFPS
and the FPS. Furthermore, if the bi-directional interactions of frictional forces of the
VFPI are ignored, the isolator displacements will be under predicted and super-
structure acceleration and base shear will be over predicted.
Seismic Performance of Variable Frequency
Pendulum Isolator under Bi-Directional
Near-Fault Ground Motions
V.R. Panchal

1
and D.P. Soni

2*
1. Professor, Dept. of Civil Engineering, Sardar Vallabhbhai Patel Institute of Technology,
Vasad-388 306, Gujarat , India
2. Assistant Professor, Dept. of Civil Engineering, Sardar Vallabhbhai Patel Institute of
Technology, Vasad-388 306, Gujarat, India,
* Corresponding Author; email: soni_svit@yahoo.com
Keywords:
Base isolation;
Superstructure flexibility;
Near-fault ground
motions; Bi-directional
interaction; VFPI; VFPS
1. Introduction
Recent years have seen a number of catastrophic
failures of structures due to severe, impulsive,
seismic events such as the 1994 Northridge earth-
quake in California, the 1995 Kobe earthquake in
Japan and 1999 Chi-Chi earthquake in Taiwan.
Failure of structures during such events seriously
hampers the relief and rehabilitation work. The
effect of severe and impulsive earthquakes on the
structures has recently received much attention and
become a significant concern for reliable aseismic
structural design. To protect structures from earth-
quake damages, seismic isolation technology has been
applied over the last three decades. This technology
is one of the most widely implemented and accepted
technologies for seismic hazard mitigation. The
fundamental concept in isolation is to reduce the
fundamental frequency of structural vibration to a
value lower than the predominant energy-containing
frequencies of the earthquake. The other purpose of
an isolation system is to provide a mean of energy
dissipation, which dissipates the seismic energy
transmitted to the system. The goal is to reduce
interstory drifts and floor accelerations to limit
damage to the structure and its contents in a cost-
effective manner.
In spite of the direct benefits of the seismic isola-
tion technology, it has been suggested that the
base-isolated buildings can be vulnerable to large
pulse-like ground motions generated at near-fault
locations [1-2]. Besides Hall et al [1] and Heaton
et al [2], several other researchers have also warned
about the vulnerability of base-isolated structures to
near-fault ground motions (for example [3-4]). Such
ground motions can be quite different than those
from the far-fault events. In particular, near-fault
ground motion records (also known as “epicentral
JSEE / Winter 2010, Vol. 11, No. 4 172
V.R. Panchal and D.P. Soni
acceleration records”) can contain large long-period
spectral components in the fault normal direction,
large short-period spectral components in the fault
parallel direction and long-duration pulses of ground
displacement and high peak ground velocities [5].
These higher spectral inputs, occurring in the neigh-
borhood of important structural periods, can result in
a structural response significantly greater than that
would occur for a typical far-fault design level event.
This concern has influenced the seismic isolation
design requirements in the Uniform Building Code,
1997 [6]. In the earlier code there were no near-fault
effects but in the recent code, near-fault effects, viz.
source type and distance dependent near-fault
factors to the customary design spectrum have been
introduced. However, it is believed that these factors
are not sufficient to solve the problem consistently,
because they pay little attention to the physical
characteristics of near-fault ground motions. Another
concern is a lack of data concerning the behaviour of
base-isolated buildings subjected to near-fault ground
motions as previous studies have focused mainly on
the seismic behavior of base-isolated buildings far
from active earthquake faults. Consequently, the
effects of these motions on buildings are not yet
understood fully.
Among various base isolation systems, the sliding
bearings are most popular due to its effectiveness
over a wide range of frequency input. The other
advantage of sliding bearings is that it ensures the
maximum acceleration transmissibility equal to the
maximum limiting frictional force. There had been
important studies on the efficiency of a variety of
sliding bearings by many researchers [7-12]. Most
of the above studies on sliding isolation systems are
based on the two-dimensional (2-D) planar model of
the isolated structure subjected to uni-directional
excitation. Such a model of the isolated structures
ignores the bi-directional interaction effects of the
frictional forces mobilized in the isolation system in
two horizontal directions. The bi-directional interac-
tion can play crucial role in the seismic response
of structures isolated with the sliding systems.
Therefore, it has received much attention and
become a significant concern for reliable aseismic
design of sliding structures. The recognition of this
fact has led several researchers to focus their study
on investigating the effects of bi-directional interac-
tion of frictional forces on the response of the
structures isolated with sliding systems [13-19]. The
review presented so far clearly shows that there have
not been many attempts to investigate the behaviour
of structures isolated with friction base isolators,
especially under bi-directional near-fault ground
motions. In view of the above, numerical studies are
carried out to understand the behaviour of structures
isolated with the VFPI under bi-directional near-fault
ground motions.
Presented in the paper is the response of
five-story building (considering flexible) isolated by
the VFPI which is investigated under bi-directional
near-fault ground motions. The specific objectives of
the study are summarized as follows:
i) To demonstrate a method for dynamic analysis
of five-story building supported on the VFPI by
duly incorporating the interaction effects of the
frictional forces of the VFPI;
ii) To compare the seismic response of building
isolated with the VFPI, FPS and VFPS in order
to measure the effectiveness of the VFPI under
bi-directional near-fault ground motions;
iii) To carry out a parametric study with a view to
investigate the influence of important parameters
on bi-directional interaction effects of frictional
forces of the VFPI. The important parameters
considered are superstructure time period,
frequency variation factor (FVF) and friction
coefficient of the VFPI; and
iv) To investigate the effects of bi-directional inter-
action of friction forces of the VFPI on the
response of the building under near-fault ground
motions (by comparing the response of the
system with and without interaction).
2. Variable Frequency Pendulum Isolator
A new isolator called the VFPI [20] incorporates
the advantages of both the friction pendulum
system (FPS) and Pure-Friction (P-F) isolators, see
Figure (1). In this isolator, the shape of the sliding
surface is non-spherical. To be more specific, its ge-
ometry has been derived from the basic equation of
an ellipse, with its semi-major axis being a linear func-
tion of sliding displacement. This is equivalent to an
infinite number of ellipses continuously transforming
into one another such that the semi-major axis is larger
for larger sliding displacement. The performance of
the VFPI is found to be very effective for a variety
of excitation and structural characteristics. The VFPI
JSEE / Winter 2010, Vol. 11, No. 4 173
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
Figure 1. Details of the VFPI [20].
is relatively flatter than the FPS, which results in
smaller vertical displacement for similar displace-
ments. This is an additional advantage of the VFPI
compared to the FPS since flatter sliding surface will
result in the generation of smaller overturning forces
in the structure. The most important properties of
this system are: i) Its time period of oscillation
depends on sliding displacement and ii) Its restoring
force exhibits softening behaviour. The isolator
geometry is such that its frequency decreases with
an increase in sliding displacement and asymptoti-
cally approaches zero at very large displacement.
As a result, the dominant frequency of excitation
and the isolator frequency are not likely to tune. The
response of structure with the FPS increases for
higher time periods, whereas the response of the
VFPI is almost independent of the structural time
period.
The restoring forces of the VFPI in the x- and
y-directions are expressed by
¹
¹
¹
;
¹
¹
¹
¹
'
¹
1
1
]
1

¸

+
¹
¹
¹
;
¹
¹
¹
¹
'
¹
·
¹
¹
¹
;
¹
¹
¹
¹
'
¹
b
b
b b
b b
y
x
by
bx
y
x

z k
z k
F
F
F
F


) ( 0
0 ) (
(1)
where
x
F and
y
F are the frictional forces of the
VFPI in the x- and y-directions, respectively; ) (
b b
z k
is the instantaneous stiffness of the VFPI;
b
z is the
resultant isolator displacement; and
b
x and
b
y are
the isolator displacement in x- and y-directions,
respectively.
The instantaneous stiffness of the VFPI [20] can
be written as
) ( ) (
2
b b b b
z M z k ω · (2)
1
1
1
]
1

¸

+ +
ω
· ω
) 2 (1 ) 1 (
) (
2
2
2
r r
z
i
b b

(3)
d
z z
r
b b
) ( sgn
·
(4)
2
2
d
gb
i
· ω
(5)
gb
d
T
i
2
2π ·
(6)
where

,
_

¸
¸
+ ·

·
N
i
i b
m m M
1
is the total mass of the
base-isolated building;
b
m is the mass of base raft;
i
m is the mass of the i
th
superstructure floor; N is
the total number of floors in the superstructure; g
is the acceleration due to gravity; b and d are semi-
minor axis and initial value of the semi-major axis
(which is greater than zero) of sliding surface; and
) ( sgn
b
z is incorporated to maintain the symmetry of
the sliding surface about the central vertical axis. The
signum function has a value of +1 for positive value
of sliding displacement and -1 for negative value of
sliding displacement; r is the non-dimensional
parameter for the sliding surface;
b
ω is the instanta-
neous frequency of the VFPI which depends on the
geometry of the sliding surface;
i
ω is the initial
frequency of the VFPI at zero isolator displacement;
and
i
T is the initial time period of the VFPI.
It can be noticed that the ratio
2
/ d b governs the
initial frequency of the isolator. Similarly, the value
of 1/d determines the rate of variation of isolator
frequency, and this factor has been defined as
frequency variation factor (FVF) [20]. It can also
be seen from Eq. (5) that the rate of decrease of
isolator frequency is directly proportional to the
FVF for a given initial frequency.
The limiting value of the frictional force, ,
s
F to
which the sliding system can be subjected in a
particular direction is expressed as
g M F
s
µ · (7)
where µ is the friction coefficient of the sliding
system.
Thus, the modeling of the VFPI is required for the
specific value of the two parameters, namely initial
time period, ,
i
T and friction coefficient, . µ
JSEE / Winter 2010, Vol. 11, No. 4 174
V.R. Panchal and D.P. Soni
3. Variable Friction Pendulum System
The VFPS [12-13] in regards of details and
operation is similar to the FPS. The difference
between the FPS and the VFPS is that the friction
coefficient of the FPS is considered to be constant
whereas the friction coefficient of the VFPS is
varied in the form of a curve. Such variation of the
friction coefficient in the VFPS can be achieved by
gradually varying the roughness of spherical surface.
The curve is chosen such that up to a certain value
of displacement the frictional force increases and
then it decreases with further displacement. This
type of curve gives the isolator initial softness for
smaller inputs, then provides stiffness for moderate
inputs, and finally for large inputs it becomes soft
again. The curve is selected with the criterion that
the isolator displacement and the base shear under
the selected near-fault ground motions decrease sig-
nificantly without much alteration to superstructure
acceleration. The equation adopted to define the
curve for the friction coefficient, µ of the VFPS is
as follows

b
z a
b
e z a
2
) (
1 0

+ µ · µ (8)
where
0
µ is the initial value of friction coefficient;
1
a and
2
a are the parameters that describe the
variation of the friction coefficient along the sliding
surface of the VFPS; ) (
2 2

b b b
y x z + · is the result-
ant isolator displacement; and
b
x and
b
y are the
displacement of the base mass relative to the ground
in the x- and y-directions, respectively.
4. Modeling and Idealization of Building Isolated
by the VFPI
Figure (2) shows the structural system under
consideration which is an idealized N-story shear-
type building resting on the VFPI. The VFPI is
installed between the base mass and foundation
of the building. The modeling of the VFPI is also
shown in Figure (2). The various assumptions made
for the system under consideration are as follows:
1. Superstructure is considered to be symmetric with
respect to two orthogonal horizontal directions
(i.e., there is no torsional coupling with lateral
movement of the system), as a result, the system
will have only the lateral degrees of freedom.
2. Floors of each story of superstructure are assumed
as rigid.
Figure 2. Modeling of multi-story building and the VFPI.
3. The force-deformation behaviour of the super-
structure is considered to be linear with viscous
damping.
4. Friction coefficient of the VFPI is assumed to be
independent of the relative velocity at the sliding
interface.
5. The VFPI is isotropic (i.e., there is same isolation
period and the coefficient of friction in two
orthogonal directions of the motion in the hori-
zontal plane).
6. The slider of the isolator is assumed to have point
contact with the sliding interface.
7. Restoring force provided by the VFPI is consid-
ered to be non-linear.
8. The frictional forces of the VFPI are assumed to
be coupled in two directions.
9. No overturning or tilting takes place in the super-
structure during sliding over the VFPI.
10. The fault normal and parallel components of
near-fault ground motion are applied in two
horizontal directions (referred as x- and y-
directions, respectively) of the building isolated
with the VFPI.
JSEE / Winter 2010, Vol. 11, No. 4 175
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
At each floor and base mass two lateral dynamic
degrees of freedom are considered. Therefore, there
are 2 x (N+1) dynamic degrees of freedom for the
N-story superstructure. The equations governing the
motion of an isolated N-story flexible shear-type
building isolated with VFPI under the two horizontal
components of earthquake excitation are expressed
as
} ]{ [ ] [ } ]{ [ } ]{ [ } ]{ [
g
z r M x K x C x M & & & & & · + + (9a)
g b bx b b
x m x k x c F x m & & & & & − · − − +
1 1 1 1
(9b)
g b by b b
y m y k y c F y m & & & & & − · − − +
1 1 1 1
(9c)
where [M], [C] and [K] are the mass, damping and
stiffness matrices of the superstructure, respectively,
of the size { }
T
N N
y y y x x x x N N ,..., , , ,..., , } { ; 2 2
2 1 2 1
· × is
the displacement vector of the superstructure
relative to the base mass;
i
x and
i
y is the lateral
displacement of the i
th
floor relative to the base
mass in x- and y-directions, respectively;
b
x& & and
b
y& &
are the acceleration of the base mass relative to the
ground in the x- and y-directions, respectively;
1
k
and
1
c are the stiffness and damping of the first
story of the superstructure, respectively; [r] is the
influence coefficient matrix;
T
b g b g g
y y x x z ) , ( } { & & & & & & & & & & + + ·
is the vector of base acceleration;
g
x& & and
g
y& & are
the earthquake ground acceleration in the x- and y-
directions;
bx
F and
by
F are the restoring forces of
the VFPI in the x- and y-directions, respectively;
T denotes the transpose; and over-dots indicate
derivative with respect to time.
5. Criteria for Sliding and Non-Sliding Phases
In a non-sliding phase (i.e., 0 · ·
b b
y x & & & & and
), 0 · ·
b b
y x & & the resultant of the frictional forces
mobilized at the interface of the VFPI is less than
the limiting frictional force ) (
2 2
s
F F F
y x
< + [21]. The
system starts sliding (i.e., 0 ≠ ≠
b b
y x & & & & and ) 0
b b
y x ≠ ≠ & &
as soon as the resultant of the frictional forces
attains the limiting frictional force. Thus, the sliding
of the system takes place if
1
2 2
·

,
_

¸
¸
+

,
_

¸
¸
s
y
s
x
F
F
F
F
(10)
Note that Eq. (10) depicts the circular interaction
between the frictional forces mobilized at the
interface of the VFPI. The system remains in the
non-sliding phase inside the interaction curve. It is
to be noted that the equations of motion of the
sliding structures in two orthogonal directions are
coupled during the sliding phases due to interaction
between the frictional forces. However, this interac-
tion effect is ignored if the structural system is
modeled as a 2-D system. In such cases, the
corresponding curve which separates the sliding and
non-sliding phases is a square as shown in Figure
(3a) by dashed lines. Further, the system changes to
non-sliding phase from the sliding phase whenever
the resultant velocity of the base mass (i.e., )
b
z&
approaches zero.
Since the frictional forces oppose the motion of
the system, the direction of the sliding of the system
with respect to the x-direction is expressed as

,
_

¸
¸
· θ

b
b
x
y

&
&
1
tan
(11)
where
b
x& and
b
y& are the velocities of the base
mass relative to the ground in x- and y-directions,
respectively.
6. Solution of Equations of Motion
The frictional forces mobilized in the VFPI are
non-linear functions of the displacement and velocity
of the system in two orthogonal directions. Also,
during the sliding phase of motion, the mobilized
frictional forces are coupled with each other by the
circular interaction (see Eq. (10)). As a result, the
equations of motion are solved in the incremental
form by employing the Newmark-β method assum-
ing linear variation of acceleration over the small time
interval, ∆t. The incremental equations in terms of
unknown incremental displacements are expressed
as
[ ] { } { } { } f P x K
eff eff
∆ ∆ + · (12)
where
eff
K is the effective stiffness matrix; } { x ∆ is
the incremental displacement vector; } {
eff
P is the
effective excitation vector; and } { f ∆ is the incre-
mental frictional force vector.
With a view to determine the incremental
frictional forces, consider Figure (3b). At time t the
frictional forces are at point A on the interaction
curve and move to point B at time t + ∆t. Therefore,
the incremental frictional forces in the x- and y-
directions, respectively, are expressed as:
t
x
t t
s x
F F f

− θ ·
+
) ( cos

∆ (13a)
JSEE / Winter 2010, Vol. 11, No. 4 176
V.R. Panchal and D.P. Soni
Figure 3. Interaction and incremental frictional forces in two
orthogonal directions of the VFPI.
t
y
t t
s y
F F f

− θ ·
+
) ( sin

∆ (13b)
where the superscript denotes the time.
Since the frictional forces are opposite to the
motion of the system, therefore, the angle
t t

∆ +
θ is
expressed in terms of the relative velocities of the
system at time t t ∆ + by

,
_

¸
¸
· θ
+
+
− +
t t
b
t t
b
t t




x
y




&
&
1
tan
(14)
Substituting for
t t

∆ +
θ in Eq. (14), the incremental
frictional forces are expressed as
( ) ( )
t
x
t t
b
t t
b
t t
b
s x
F
y x
x
F F





+
·
+ +
+
2 2
∆ ∆


& &
&
(15a)
( ) ( )
t
y
t t
b
t t
b
t t
b
s y
F
y x
y
F F





+
·
+ +
+
2 2
∆ ∆


& &
&
(15b)
In order to solve the incremental matrix Eq.
(12), the incremental frictional forces
x
f ∆ ( and
)
y
f ∆ should be known at any time interval. The
incremental frictional forces involve the system
velocities at time t + ∆ t (see Eq. (15)) which in
turn depend on the incremental displacements
b
x ∆ (
and )
b
y ∆ at the current time step. As a result, an
iterative procedure is required to obtain the required
incremental solution. The steps of the procedure
considered are as follows:
1. Assume 0 · ·
y x
F F ∆ ∆ for iteration, j = 1 in Eq.
(12) and solve for
b
x ∆ and .
b
y ∆
2. Calculate the incremental velocity
b
x& ∆ and
b
y& ∆
using the
b
x ∆ and .
b
y ∆
3. Calculate the velocities at time t + ∆ t using
incremental velocities (i.e.,
b
t
b
t t
b
x x x

& & & ∆

+ ·
+
and
)
b
t
b
t t
b
y y y

& & & ∆

+ ·
+
and compute the revised
incremental frictional forces
x
F ∆ and
y
F ∆ from
Eq. (15).
4. Iterate further, until the following convergence
criteria are satisfied for both incremental frictional
forces, i.e.,
( ) ( )
( )
ε ≤

+
j
x
j
x
j
x


F
F F



∆ ∆
1
(16a)
( ) ( )
( )
ε ≤

+
j
y
j
y
j
y


F
F F



∆ ∆
1
(16b)
where ε is a small threshold parameter. The super-
script to the incremental forces denotes the iteration
number.
When the convergence criteria are satisfied, the
velocity of the sliding structure at time t + ∆ t is
calculated using incremental velocity. In order to
avoid the unbalance forces, the acceleration of the
system at time t + ∆ t is evaluated directly from the
equilibrium of system Eq. (9). The response of
the sliding structures is quite sensitive to the time
interval, ∆ t, and initial conditions at the beginning of
sliding and non-sliding phases. For the present
study, the results are obtained with maximum time
interval, ∆ t = 0.0001sec. In order to determine the
incremental frictional forces at the sliding support,
the number of iterations in each time step is taken
as 10. At the end of each time step the phase of the
motion of the system should be checked. Further,
the sliding velocity less than 1×10-8m/sec is assumed
to be zero for checking the transition from sliding to
non-sliding phase.
JSEE / Winter 2010, Vol. 11, No. 4 177
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
7. Numerical Study
For the present study, the mass matrix of the
superstructure, [M], is diagonal and characterized
by the mass of each floor which is kept constant
(i.e., m
i
= m for i = 1 to N). Also, for simplicity
the stiffness of all the floors is taken as constant
and expressed by the parameter k. The value of k
is selected to provide the required fundamental
time period of superstructure, T
s
, as a fixed base.
The damping matrix of the superstructure, [C], is not
known explicitly. It is constructed by assuming the
modal damping ratio which is kept constant in each
mode of vibration. Thus, the superstructure and
the base mass of the isolated structural system under
consideration can be completely characterized by
the parameters namely, the fundamental time
period of the superstructure, T
s
, damping ratio of
the superstructure, ξ
s
, number of stories in the
superstructure, N, and the ratio of base mass to the
superstructure floor mass, m
b
/m. The superstructure
considered has five stories with fundamental time
period, T
s
= 0.5sec and damping ratio, ξ
s
= 2% of
critical damping. The fundamental time period and
damping ratio of the superstructure are considered
to be equal in the x- and y-directions. The mass ratio,
m
b
/m, is assumed to be unity. On the other hand, the
VFPI isolator is characterized by two parameters,
namely initial time period, T
i

, and the coefficient of
friction, µ. For all investigations, the parameters
of the VFPI are selected as b = 0.01m and d = 0.1m
(FVF 10 per m) so that it has initial time period of
2.0sec. The value of µ has been considered as 0.02.
For comparison, examples with the FPS and VFPS
isolators are also taken with isolation period of
2.0sec. The VFPS isolators are characterized by two
parameters, namely the period of the base isolation,
T
b
, and the coefficient of friction, µ. The coefficient
of sliding friction, µ, in the VFPS can be defined
by the initial time period of the VFPS, T
i

, and
the peak frictional coefficient, µ
max
. In case of the
VFPS isolator, an initial time period of 1.5sec and
a peak friction coefficient of 0.15 are chosen for
all investigations. In case of the FPS isolators, a
coefficient of sliding friction of 0.02 is selected for
all investigations. In the present study, the superstruc-
ture parameters, ξ
s
and m
b
/m, are held constant.
The response quantities of interest are the top
floor absolute acceleration (i.e.,
g b N a
x x x x & & & & & & & & + + ·
and ),
g b N a
y y y y & & & & & & & & + + · the base shear (i.e.,
bx
F and
)
by
F and the isolator displacement (i.e.,
b
x and )
b
y
in the x- and y-directions. The top floor absolute
acceleration and base shear are directly proportional
to the forces (shear force and bending moments)
exerted in the superstructure due to the earthquake
ground motion. On the other hand, the relative
displacements of the VFPI are crucial from the
design point of view of the isolator.
Six pairs of near-fault ground motions are used
as input ground motions in order to effectively study
the dynamic behaviour of building isolated with
the VFPI under bi-directional excitation. Some
characteristics of these recorded near-fault ground
motions are summarized in Table (1). From this
table, it is found that these near-fault ground
motions have a variety of PGA, PGV and PGD. It
is observed that in most of the cases, the PGA of
fault normal component is relatively higher than
that of fault parallel component. Furthermore, the
acceleration and displacement spectra of the six
ground motions for 5% damping are shown in
Figure (4). The spectra of these ground motions
indicate that the ground motions are recorded at a
firm soil or rock site. From this figure, it is found that
the average spectral acceleration and displacement
Note: PGD = Peak Ground Displacement, PGV= Peak Ground Velocity and PGA= Peak Ground Acceleration
Table 1. Some characteristics of near-fault ground motions considered in the study.
Normal Component Parallel Component
Near-Fault Ground Motions Recording Station
Duration
(sec)
PGD
(cm)
PGV
(cm/sec)
PGA
(g)
PGD
(cm)
PGV
(cm/sec)
PGA
(g)
January 17, 1994 Northridge, California Sylmar 60.000 31.1 122 0.73 9.03 53.9 0.6
January 17, 1994 Northridge, California Rinaldi 14.950 39.1 175 0.89 18.4 60.2 0.39
January 17, 1994 Northridge, California Newhall 60.000 38.1 119 0.72 17.6 49.3 0.65
June 28, 1992 Landers, California Lucerne Valley 49.284 230 136 0.71 184 70.3 0.80
October 15, 1979 Imperial Valley, California El Centro Array #5 39.420 76.5 98 0.37 150 52.5 0.55
October 15, 1979 Imperial Valley, California El Centro Array #7 36.900 49.1 113 0.46 218 55.2 0.34

JSEE / Winter 2010, Vol. 11, No. 4 178
V.R. Panchal and D.P. Soni
values of the fault normal and parallel component
is almost the same for low periods (i.e., in the range
0-0.3sec). For longer periods (i.e., beyond 0.3sec),
the spectral acceleration and displacement compo-
nent of fault normal component is significantly larger
than the fault parallel component.
7.1. Comparison of the Isolators Considering
Interaction of the Friction Forces
Table (2) shows the comparison of the peak
response quantities of the three isolators considering
the interaction of the friction forces. This table
shows that in most of the near-fault ground motions,
the responses, especially top floor absolute accelera-
tion and base shear of structures with the VFPI
are considerably reduced as compared to those
structures with the FPS and VFPS isolators, whereas
the isolator displacement of the VFPI exceeds that
of the FPS and the VFPS. This is expected as the
horizontal stiffness of the VFPI is lower than that of
the FPS and the VFPS. Such large isolator displace-
ment of the VFPI will lead to the requirement of
very large isolators, costly flexible connections for
Figure 4. Acceleration and displacement spectra of the six near-fault ground motions for 5% damping.
utilities and an extensive and expensive loss of space
for a seismic gap. Under near-fault ground motions,
this feature of the VFPI reduces its effectiveness in
comparison to the FPS and the VFPS. In order to
give further insight into the difference in the behavior
of structures with the VFPI, FPS and VFPS isola-
tors, time histories of top floor absolute acceleration
and isolator displacement are shown in Figure (5)
for Northridge, 1994 (Sylmar) and Northridge, 1994
(Rinaldi) near-fault ground motions. Similar trends
observed in Figure (5) and Table (2), are also ob-
served in Figure (6) which shows the comparison
of the hysteresis loops of the VFPI, FPS and VFPS
isolators.
7.2. Comparison between Isolator Displacement
in the x- and y-Directions
Figure (7) shows the time variation of
b
x and
b
y
of the building isolated with the VFPI (T
i
= 2.0sec
and µ = 0.02) under different near-fault ground
motions (top); and correlation between x
b
and y
b
(bottom). The peak values of x
b
and y
b
for Northridge,
1994 (Sylmar) near-fault ground motion are 63.588

JSEE / Winter 2010, Vol. 11, No. 4 179
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
Table 2. Comparison between peak response quantities of various isolators considering the interaction of the friction forces.
Figure 5. Time variation of top floor absolute acceleration and isolator displacement of five-story building isolated with the FPS
(T
b
= 2.0sec and µ = 0.02), VFPS (T
b
= 2.0sec, T
i
= 1.5 sec and µ
max
= 0.15) and VFPI (T
i
= 2.0 sec and µ = 0.02) under
Northridge, 1994 (Sylmar) and Northridge, 1994 (Rinaldi) near-fault ground motions.
Near-fault Ground Motions Building Condition
a
x && (g)
a
y && (g)
Fbx(W) Fby (W) xb (cm) yb (cm)
Isolated (FPS) 0.8564 0.4024 0.7293 0.3198 71.9780 31.7210
Isolated (VFPI) 0.1604 0.1410 0.0342 0.0303 63.5880 16.8440
Northridge, 1994
(Sylmar)
Isolated (VFPS) 0.8008 0.4739 0.4126 0.2542 40.0610 22.4930
Isolated (FPS) 0.8493 0.5163 0.7525 0.4470 72.9400 43.8890
Isolated (VFPI) 0.2037 0.1825 0.0340 0.0303 63.6420 40.4250
Northridge, 1994
(Rinaldi)
Isolated (VFPS) 0.8165 0.5370 0.5101 0.2508 50.0200 22.0930
Isolated (FPS) 0.5275 0.3107 0.4382 0.2255 42.8360 22.3720
Isolated (VFPI) 0.2426 0.1895 0.0339 0.0266 82.9760 44.8440
Northridge, 1994
(Newhall)
Isolated (VFPS) 1.0045 0.6904 0.2768 0.1994 18.6420 9.8999
Isolated (FPS) 0.4260 0.3086 0.3168 0.1546 30.9040 13.9400
Isolated (VFPI) 0.2324 0.3075 0.0354 0.0346 120.3700 72.9390
Landers, 1992
(Lucerne Valley)
Isolated (VFPS) 0.7048 0.7898 0.2897 0.1630 21.5940 10.3140
Isolated (FPS) 0.5691 0.3096 0.4782 0.2311 47.3160 22.6330
Isolated (VFPI) 0.1519 0.1973 0.0287 0.0287 123.9000 102.8500
Imperial Valley, 1979
(El Centro Array #5)
Isolated (VFPS) 0.7407 0.4165 0.2800 0.1867 23.4070 6.7835
Isolated (FPS) 0.4277 0.2927 0.3831 0.2426 37.7590 24.0500
Isolated (VFPI) 0.1239 0.1010 0.0355 0.0252 113.4200 74.9270
Imperial Valley, 1979
(El Centro Array #7)
Isolated (VFPS) 0.7162 0.8309 0.2793 0.2053 17.5720 9.9297
JSEE / Winter 2010, Vol. 11, No. 4 180
V.R. Panchal and D.P. Soni
Figure 6. Comparison between hysteresis loops of the FPS, VFPS and VFPI isolators.
Figure 7. Top: Time variation of x
b
and y
b
of five-story building isolated with VFPI (T
i
= 2.0sec and µ = 0.02) under different near-
fault ground motions. Bottom: Correlation between x
b
and y
b
.
and 16.844cm, respectively whereas for Northridge,
1994 (Rinaldi) near-fault ground motion are 63.642
and 40.425cm, respectively. From this figure, it can
be noticed that the peak isolator displacement in
fault normal direction, x
b
, is significantly larger than
the corresponding peak isolator displacement in
fault parallel direction, y
b
. This is expected due to the
fact that spectral content of fault normal component
JSEE / Winter 2010, Vol. 11, No. 4 181
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
is considerably larger than the fault parallel compo-
nent of near-fault ground motion. Furthermore, it is
observed that the response of isolated systems to fault
normal and fault parallel components are more or less
uncorrelated as the maximum displacement in the
fault normal direction occurs at a different time than
that in the fault parallel direction.
Figure (8) shows variation of peak displacements,
x
b
, y
b
, z
b
and z
m
of the five-story building isolated
with the VFPI against FVF for different types of
near-fault ground motions. The displacement, z
m
,
denotes the square root of the sum of the squares
(SRSS) of the peak values of x
b
and y
b
. As noted
earlier, the displacement due to parallel component,
y
b
, is much smaller in comparison to the correspond-
ing displacement due to the normal component,
x
b
. Furthermore, there is no significant difference
between the peak resultant displacement, z
b
, and the
corresponding displacement, x
b
, implying that the
peak resultant displacement of the isolators is mainly
contributed by the displacement due to the normal
Figure 8. Variation of the peak isolator displacements, x
b
, y
b
, z
b
, and z
m

, of five-story building isolated with the VFPI against
the FVF for various near-fault ground motions.
component of the near-fault ground motions. The
displacement, z
m
, is larger than displacement, z
b
,
confirming that the peak displacements in the
isolation system due to the normal and parallel
components of the near-fault motions do not occur at
the same time. Similar trends can be also observed
from Table (3).
7.3. Influence of Bi-Directional Interaction of the
Friction Forces
The time variation of the absolute acceleration of
superstructure (i.e.,
a
x& & and )
a
y& & and the sliding base
displacement (i.e., x
b
and y
b
) in x- and y-directions
are shown in Figure (9) for Northridge, 1994 (Sylmar)
near-fault ground motion. The response is plotted
for both considering and ignoring the interaction of
the frictional forces of the VFPI. The absolute
acceleration of the superstructure is relatively less
for considering the effects of the interaction of
frictional forces as compared to those without
interaction. Thus, the superstructure experiences
JSEE / Winter 2010, Vol. 11, No. 4 182
V.R. Panchal and D.P. Soni
Near-Fault Ground
Motions
Building
Condition
a
x&&
(g)
a
y &&
(g)
Fbx (W) Fby (W) xb (cm) yb (cm) zb (cm) zm (cm)
Non-isolated 2.4362 4.1222 1.4504 2.5888 --- --- --- ---
Isolated (No interaction) 0.2477 0.2975 0.0359 0.0359 62.1700 15.4140 64.0523 64.0523
Northridge, 1994
(Sylmar)
Isolated (Interaction) 0.1604 0.1410 0.0342 0.0303 63.5880 16.8440 64.7450 65.7810
Non-isolated 3.3692 2.2774 1.9843 1.2256 --- --- --- ---
Isolated (No interaction) 0.4108 0.3075 0.0359 0.0359 62.5110 27.9180 68.4620 68.4620
Northridge, 1994
(Rinaldi)
Isolated (Interaction) 0.2037 0.1825 0.0340 0.0303 63.6420 40.4250 64.0360 75.3950
Non-isolated 4.3521 1.5202 2.5312 0.6983 --- --- --- ---
Isolated (No interaction) 0.2835 0.3250 0.0359 0.0359 81.1890 19.4310 83.4818 83.4818
Northridge, 1994
(Newhall)
Isolated (Interaction) 0.2426 0.1895 0.0339 0.0266 82.9760 44.8440 88.4580 94.3190
Non-isolated 2.3189 1.9496 0.6026 0.5309 --- --- --- ---
Isolated (No interaction) 0.2993 0.4386 0.0359 0.0359 125.3900 13.9530 126.1639 126.1639
Landers, 1992
(Lucerne Valley)
Isolated (Interaction) 0.2324 0.3075 0.0354 0.0346 120.3700 72.9390 140.7300 140.7500
Non-isolated 1.6985 2.8422 1.0542 1.3206 --- --- --- ---
Isolated (No interaction) 0.2565 0.2615 0.0359 0.0359 147.9800 77.4810 167.0371 167.0371
Imperial Valley,
1979
(El Centro Array #5)
Isolated (Interaction) 0.1519 0.1973 0.0287 0.0287 123.9000 102.8500 153.3800 161.0300
Non-isolated 1.3789 0.9177 0.7753 0.5546 --- --- --- ---
Isolated (No interaction) 0.1578 0.2568 0.0359 0.0359 111.7700 22.6620 114.0443 114.0443
Imperial Valley,
1979
(El Centro Array #7)
Isolated (Interaction) 0.1239 0.1010 0.0355 0.0252 113.4200 74.9270 116.5300 135.9300
Table 3. Peak response quantities of non-isolated building and VFPI-isolated building (T
i
= 2sec and µ = 0.02).
Figure 9. Time history of the absolute acceleration of the superstructure and isolator displacement to the Northridge, 1994 (Sylmar)
near-fault ground motion.
JSEE / Winter 2010, Vol. 11, No. 4 183
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
less earthquake forces when the interaction of the
frictional forces is considered in the analysis. On
the other hand, the isolator displacements are rela-
tively more for considering the interaction effects
in comparison to that without interaction effects.
This is due to the fact that when the interaction is
taken into consideration the structure starts sliding
at a relatively lower value of the frictional forces
mobilized at the sliding interface (refer to the sliding
Eq. (10)), and as a result, there is more isolator
displacement. Similar trends in the response of the
building isolated with the VFPI are found in Figure
(10) for Northridge, 1994 (Rinaldi) near-fault
ground motion. This implies that there is significant
over prediction of the superstructure accelerations
and under prediction of the isolator displacements
under near-fault ground motions, if the bi-directional
interaction effects are ignored and the system is ide-
alized as a 2-D system. The under prediction of the
isolator displacement is crucial from the point of view
of designing the friction base isolators. Therefore, the
bi-directional interaction effects of frictional forces
of VFPI under near-fault ground motions must be
rigorously considered in the analysis of the structure.
Figure 10. Time history of the absolute acceleration of the superstructure and isolator displacement to the Northridge, 1994
(Rinaldi) near-fault ground motion.
Figure (11) shows the variation of the resultant
peak absolute acceleration of the superstructure
(i.e., ) ) ( ) (
2 2
y x
ax m a ax m a
& & & & + against T
s
under different
near-fault ground motions. The figure indicates that
for all values of superstructure time periods, the
absolute acceleration of the superstructure is less
for considering the interaction as compared to those
without interaction. The absolute acceleration
spectra of the superstructure without sliding support
(referred to as non-isolated) are also shown in order
to study the effectiveness of the sliding support. The
figure indicates clearly that the sliding support is quite
effective in reducing the earthquake response of the
superstructure. Further, the absolute acceleration of
the system with sliding base is less sensitive to the
time period of the superstructure in comparison with
fixed base system. Similar differences are also found
in Figures (12) and (13) and Table (3).
In Figure (14), the variation of the resultant peak
sliding base displacement is plotted against µ
for various near-fault ground motions. The figure
clearly shows that the peak isolator displacement is
significantly higher for considering the interaction as
JSEE / Winter 2010, Vol. 11, No. 4 184
V.R. Panchal and D.P. Soni
Figure 11. Effects of bi-directional interaction of frictional forces on peak absolute acceleration of the superstructure of five-story
building isolated with the VFPI (T
i
= 2.0 sec and µ = 0.02) under various near-fault ground motions.
Figure 12. Plot of peak absolute superstructure acceleration of five-story building isolated with the VFPI against the FVF for various
near-fault ground motions (considering with and without interaction).
JSEE / Winter 2010, Vol. 11, No. 4 185
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
Figure 13. Friction coefficient variation of peak absolute superstructure acceleration of five-story building isolated with the VFPI
under various near-fault ground motions (considering with and without interaction).
Figure 14. Comparison of isolator displacement of five-story building isolated by the VFPI during various near-fault ground motions
(considering with and without interaction).
JSEE / Winter 2010, Vol. 11, No. 4 186
V.R. Panchal and D.P. Soni
compared to those without interaction. Similar
differences are also observed in Figure (15). Thus,
there is a need to consider the bi-directional inter-
action effects of frictional forces on the response.
Note that similar effects of bi-directional interaction
of frictional forces for structures isolated by Teflon
sliding bearing were observed by Mokha et al [22]
and the same are further confirmed in the present
Figure 15. Plot of peak isolator displacement, z
b
, of five-story building isolated with the VFPI against the fundamental time period of
the superstructure for various near-fault ground motions (considering with and without interaction).
study for VFPI-isolated structures.
In Figure (16), the variation of ratios, R
1
and
R
2
, is plotted against the friction coefficient of
VFPI for various near-fault ground motions. The
ratio, R
1
, denotes the ratio of peak resultant isolator
displacement with interaction to the corresponding
displacement without interaction of frictional
forces. The ratio, R
1
, is an index of the bi-directional
Figure 16. Variation of ratios, R
1
and R
2
against friction coefficient of the VFPI.
JSEE / Winter 2010, Vol. 11, No. 4 187
Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions
interaction effects of frictional forces and values
significantly different from unity imply significant
interaction effects. On the other hand, values close
to unity justify the 2-D idealization of the system
and the interaction of the frictional forces may be
ignored. The ratio, R
1
, increases with the increase of
friction coefficient of VFPI. This indicates that
effects of bi-directional interaction increases with
the increase in the friction coefficient. In other words,
bi-directional interaction effects are relatively less
for the low values of the friction coefficient. This
is due to the fact that for lower value of friction
coefficient, the isolation system remains most of the
time in the sliding phase for both cases of excitation
(i.e., with and without interaction). As a result, the
difference in the sliding displacements for the two
cases is relatively less. Moreover, it is found that
these effects are strongly dependent on the input
ground motions. On the other hand, the R
2
is the
ratio of peak resultant isolator displacement, z
b
the corresponding peak isolator displacement due
to fault normal component, x
b
. The ratio, R
2
, is not
much influenced by the variation of friction coeffi-
cient. It varies in the range of 1.026-1.044. This
indicated that the resultant isolator displacement of
building isolated with VFPI under near-fault ground
motion may be obtained solely from the normal
component, with addition of about 5% to incorporate
the contribution from the parallel component. Thus,
the contribution of fault parallel component can be
neglected in calculating the peak resultant isolator
displacement, which is only marginally above the
maximum isolator displacement in the fault normal
direction.
8. Conclusions
The response of flexible five-story building
isolated with the variable frequency pendulum
isolator (VFPI) under bi-directional near-fault ground
motions is investigated using standard numerical
technique. The interaction between mobilized
frictional forces of the VFPI in two horizontal direc-
tions is duly incorporated in the governing equations
of motion of the building isolated with the VFPI. In
order to verify the effectiveness of the VFPI under
bi-directional near-fault ground motions, the seismic
responses are compared with that of the same build-
ing isolated by the variable friction pendulum system
(VFPS) and friction pendulum system (FPS). The
response of the VFPI-isolated building with interac-
tion is compared with those without interaction in
order to demonstrate the significance of the bi-direc-
tional interaction between the mobilized frictional
forces of VFPI. Furthermore, a parametric study has
been carried out to critically examine the influence
of important parameters on bi-directional effects of
frictional forces of VFPI. The important parameters
considered are the superstructure time period,
frequency variation factor (FVF) and friction coeffi-
cient of VFPI. From the trends of the numerical
results of the present study, the following conclusions
may be drawn:
v Under bi-directional near-fault ground motions,
the isolator displacement in the VFPI is more than
that of the VFPS and the FPS, whereas the top
floor absolute acceleration and the base shear are
less than that of the VFPS and the FPS.
v The peak isolator displacement of building isolated
with the VFPI under fault normal and parallel
components of near-fault ground motion are found
to be more or less uncorrelated.
v The bi-directional interaction of frictional forces
has noticeable effects on the response of the
building isolated with the VFPI. If the interaction
of the frictional forces at the sliding interface is
ignored, then the superstructure acceleration and
base shear will be overestimated and the sliding
displacement will be underestimated.
v The resultant maximum isolator displacement of
building isolated with the VFPI is mainly due to
the normal component of the near-fault ground
motions. The contribution from the parallel
component in the resultant displacement may be
ignored. The resultant maximum isolator dis-
placement can be obtained from the fault normal
component by increasing about 5% to account the
contribution of the parallel component.
v Under near-fault ground motions, the effects of
bi-directional interaction of friction forces of the
VFPI increase with the increase in the friction
coefficient.
References
1. Hall, J.F., Heaton, T.H., Halling, M.W., and Wald,
D.J. (1995). “Near-Source Ground Motion and
Its Effects on Flexible Buildings”, Earthquake
Spectra, 11, 569-605.
2. Heaton, T.H., Hall, J.F., Wald, D.J., and Halling,
JSEE / Winter 2010, Vol. 11, No. 4 188
V.R. Panchal and D.P. Soni
M.W. (1995). “Response of High-Rise and Base-
Isolated Buildings to a Hypothetical Mw 7.0 Blind
Thrust Earthquake”, Science, 267, 206-211.
3. Tena-Colunga, A. (1996). “Some Retrofit Options
for the Seismic Upgrading of Old Low-Rise
School Buildings in Mexico”, Earthquake Spec-
tra, 12, 883-902.
4. Tena-Colunga, A., Gomez-Soberon, C., and
Munoz-Loustaunau, A. (1997). “Seismic Isola-
tion of Buildings Subjected to Typical Subduction
Earthquake Motions for the Mexican Pacific
Coast”, Earthquake Spectra, 13, 505-532.
5. Jangid, R.S. and Kelly, J.M. (2001). “Base
Isolation for Near-Fault Motions”, Earthquake
Engineering and Structural Dynamics, 30, 691-
707.
6. Uniform Building Code (1997). Earthquake
Regulations for Seismic Isolated Structures,
International Conf. of Building Officials,
Whittier, CA.
7. Zhou, Q., Lu, X., Wang, Q., Feng, D., and Yao,
Q. (1998). “Dynamic Analysis on Structures
Base-Isolated by a Ball System with Restoring
Property”, Earthquake Engineering and
Structural Dynamics, 27, 773-791.
8. Hamidi, N., El Naggar, M.H., Vafai, A., and
Ahmadi, G. (2003). “Seismic Isolation of Build-
ings with Sliding Concave Foundation (SCF)”,
Earthquake Engineering and Structural Dy-
namics, 32, 15-29.
9. Tsai, C.S., Chiang, T.-C., and Chen, B.-J. (2005).
“Experimental Evaluation of Piecewise Exact
Solution for Predicting Seismic Responses of
Spherical Sliding Type Isolated Structures”,
Earthquake Engineering and Structural Dy-
namics, 34, 1027-1046.
10. Tsai, C.S., Chiang, T.-C., and Chen, B.-J. (2003).
“Finite Element Formulations and Theoretical
Study for Variable Curvature Friction Pendulum
System”, Engineering Structures, 25, 1719-1730.
11. Murnal, P. and Sinha, R. (2004). “Aseismic
Design of Structure-Equipment Systems Using
Variable Frequency Pendulum Isolator”, Nuclear
Engineering and Design, 231, 129-139.
12. Panchal, V.R. and Jangid, R.S. (2008). “Variable
Friction Pendulum System for Near-Fault Ground
Motions”, Structural Control and Health
Monitoring, 15, 568-584.
13. Panchal, V.R. and Jangid R.S. (2009). “Seismic
Response of Structures with Variable Friction
Pendulum System”, Journal of Earthquake
Engineering, 13, 193-216.
14. Panchal, V.R. and Jangid, R.S. (2008). “Seismic
Behavior of Variable Frequency Pendulum
Isolator”, Earthquake Engineering and Engi-
neering Vibration, 7, 193-205.
15. Jangid, R.S. (1996). “Seismic Response of
Sliding Structures to Bi-Directional Earthquake
Excitation”, Earthquake Engineering and
Structural Dynamics, 25, 1301-1306.
16. Jangid, R.S. (1997). “Response of Pure-Friction
Sliding Structures to Bi-Directional Harmonic
Ground Motion”, Engineering Structures, 19,
97-104.
17. Jangid, R.S. (2001). “Response of Sliding Struc-
tures to Bi-Directional Excitation”, Journal of
Sound and Vibration, 243, 929-944.
18. Matsagar, V.A. and Jangid, R. S. (2006).
“Seismic Response of Simply Supported Base-
Isolated Bridge with Different Isolators”,
International Journal of Applied Science and
Engineering, 4, 53-69.
19. Rao, P.B. and Jangid, R.S. (2001). “Performance
of Sliding Systems under Near-Fault Motions”,
Nuclear Engr. and Design, 203, 259-272.
20. Pranesh, M. and Sinha, R. (2000). “VFPI: An
Isolation Device for Aseismic Design”, Earth-
quake Engineering and Structural Dynamics,
29, 603-627.
21. Constantinou, M.C., Mokha, A.S., and Reinhorn,
A.M. (1990). “Teflon Bearing in Base Isolation
II: Modeling”, J. of Structural Engineering,
ASCE, 116, 455-474.
22. Mokha, A., Constantinou, M.C., and Reinhorn,
A.M. (1993). “Verification of Friction Model of
Teflon Bearings under Triaxial Load”, J. of
Structural Engineering, ASCE, 119, 240-261.

its geometry has been derived from the basic equation of an ellipse. FPS and VFPS in order to measure the effectiveness of the VFPI under bi-directional near-fault ground motions. the sliding bearings are most popular due to its effectiveness over a wide range of frequency input. source type and distance dependent near-fault factors to the customary design spectrum have been introduced. viz. near-fault effects. 4 . The recognition of this fact has led several researchers to focus their study on investigating the effects of bi-directional interaction of frictional forces on the response of the 172 structures isolated with sliding systems [13-19]. These higher spectral inputs. This concern has influenced the seismic isolation design requirements in the Uniform Building Code. numerical studies are carried out to understand the behaviour of structures isolated with the VFPI under bi-directional near-fault ground motions. it is believed that these factors are not sufficient to solve the problem consistently. The review presented so far clearly shows that there have not been many attempts to investigate the behaviour of structures isolated with friction base isolators. Panchal and D. Most of the above studies on sliding isolation systems are based on the two-dimensional (2-D) planar model of the isolated structure subjected to uni-directional excitation. There had been important studies on the efficiency of a variety of sliding bearings by many researchers [7-12]. occurring in the neighborhood of important structural periods. No. see Figure (1). The important parameters considered are superstructure time period. The performance of the VFPI is found to be very effective for a variety of excitation and structural characteristics. frequency variation factor (FVF) and friction coefficient of the VFPI. can result in a structural response significantly greater than that would occur for a typical far-fault design level event. The VFPI JSEE / Winter 2010. In this isolator. Such a model of the isolated structures ignores the bi-directional interaction effects of the frictional forces mobilized in the isolation system in two horizontal directions. large short-period spectral components in the fault parallel direction and long-duration pulses of ground displacement and high peak ground velocities [5]. In view of the above. the effects of these motions on buildings are not yet understood fully. The bi-directional interaction can play crucial role in the seismic response of structures isolated with the sliding systems. Consequently. This is equivalent to an infinite number of ellipses continuously transforming into one another such that the semi-major axis is larger for larger sliding displacement. the shape of the sliding surface is non-spherical.V. especially under bi-directional near-fault ground motions. Among various base isolation systems. and iv) To investigate the effects of bi-directional interaction of friction forces of the VFPI on the response of the building under near-fault ground motions (by comparing the response of the system with and without interaction).P. In the earlier code there were no near-fault effects but in the recent code.R. it has received much attention and become a significant concern for reliable aseismic design of sliding structures. The other advantage of sliding bearings is that it ensures the maximum acceleration transmissibility equal to the maximum limiting frictional force. Therefore. The specific objectives of the study are summarized as follows: i) To demonstrate a method for dynamic analysis of five-story building supported on the VFPI by duly incorporating the interaction effects of the frictional forces of the VFPI. ii) To compare the seismic response of building isolated with the VFPI. with its semi-major axis being a linear function of sliding displacement. Vol. To be more specific. However. Presented in the paper is the response of five-story building (considering flexible) isolated by the VFPI which is investigated under bi-directional near-fault ground motions. 11. 1997 [6]. Soni acceleration records”) can contain large long-period spectral components in the fault normal direction. 2. Another concern is a lack of data concerning the behaviour of base-isolated buildings subjected to near-fault ground motions as previous studies have focused mainly on the seismic behavior of base-isolated buildings far from active earthquake faults. Variable Frequency Pendulum Isolator A new isolator called the VFPI [20] incorporates the advantages of both the friction pendulum system (FPS) and Pure-Friction (P-F) isolators. because they pay little attention to the physical characteristics of near-fault ground motions. iii) To carry out a parametric study with a view to investigate the influence of important parameters on bi-directional interaction effects of frictional forces of the VFPI.

(5) that the rate of decrease of isolator frequency is directly proportional to the FVF for a given initial frequency. mb is the mass of base raft. to which the sliding system can be subjected in a particular direction is expressed as Fs = µ M g (7) (2) where µ is the friction coefficient of the sliding system. Vol. The restoring forces of the VFPI in the x. kb (zb) is the instantaneous stiffness of the VFPI. As a result. and sgn (zb) is incorporated to maintain the symmetry of the sliding surface about the central vertical axis. g is the acceleration due to gravity. the modeling of the VFPI is required for the specific value of the two parameters.and y-directions are expressed by Fbx  Fx  kb ( zb) 0   xb           = + Fby  Fy  0 kb ( zb)  yb         (1) where Fx and Fy are the frictional forces of the VFPI in the x. Ti. zb is the resultant isolator displacement. and xb and yb are the isolator displacement in x. whereas the response of the VFPI is almost independent of the structural time period. the value of 1/d determines the rate of variation of isolator frequency. The isolator geometry is such that its frequency decreases with an increase in sliding displacement and asymptotically approaches zero at very large displacement. 173 JSEE / Winter 2010.  where M =  mb +   ∑ mi   i =1 N   is the total mass of the is relatively flatter than the FPS. b and d are semiminor axis and initial value of the semi-major axis (which is greater than zero) of sliding surface. The most important properties of this system are: i) Its time period of oscillation depends on sliding displacement and ii) Its restoring force exhibits softening behaviour. This is an additional advantage of the VFPI compared to the FPS since flatter sliding surface will result in the generation of smaller overturning forces in the structure. It can be noticed that the ratio b / d 2 governs the initial frequency of the isolator. the dominant frequency of excitation and the isolator frequency are not likely to tune. µ. Details of the VFPI [20].and y-directions. 4 . The limiting value of the frictional force. N is the total number of floors in the superstructure.Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions  ωi2 2 ωb ( zb ) =   2  (1 + r) (1 + 2r)  r= zb sgn ( zb ) d gb d2      (3) (4) ωi2 = (5) 2 Ti = 2π d gb (6) Figure 1. respectively. r is the non-dimensional parameter for the sliding surface. No. It can also be seen from Eq. The instantaneous stiffness of the VFPI [20] can be written as 2 kb (zb) = M ωb (zb) base-isolated building. ωi is the initial frequency of the VFPI at zero isolator displacement.and y-directions. and Ti is the initial time period of the VFPI. and this factor has been defined as frequency variation factor (FVF) [20]. ωb is the instantaneous frequency of the VFPI which depends on the geometry of the sliding surface. The response of structure with the FPS increases for higher time periods. which results in smaller vertical displacement for similar displacements. The signum function has a value of +1 for positive value of sliding displacement and -1 for negative value of sliding displacement. namely initial time period. respectively. Fs. Thus. Similarly. 11. mi is the mass of the ith superstructure floor. and friction coefficient.

The force-deformation behaviour of the superstructure is considered to be linear with viscous damping. respectively. zb = ( xb + yb ) is the resultant isolator displacement. No.. 7.e. 9. The VFPI is installed between the base mass and foundation of the building. Restoring force provided by the VFPI is considered to be non-linear. The various assumptions made for the system under consideration are as follows: 1. Friction coefficient of the VFPI is assumed to be independent of the relative velocity at the sliding interface. as a result. the system will have only the lateral degrees of freedom. The curve is chosen such that up to a certain value of displacement the frictional force increases and then it decreases with further displacement. µ of the VFPS is as follows µ = ( µ0 + a1 zb ) e − a2 zb (8) Figure 2. Floors of each story of superstructure are assumed as rigid.V. The equation adopted to define the curve for the friction coefficient. The curve is selected with the criterion that the isolator displacement and the base shear under the selected near-fault ground motions decrease significantly without much alteration to superstructure acceleration.. and xb and yb are the displacement of the base mass relative to the ground in the x. and finally for large inputs it becomes soft again. This type of curve gives the isolator initial softness for smaller inputs. 10. Such variation of the friction coefficient in the VFPS can be achieved by gradually varying the roughness of spherical surface. The frictional forces of the VFPI are assumed to be coupled in two directions.P.e. Vol. The difference between the FPS and the VFPS is that the friction coefficient of the FPS is considered to be constant whereas the friction coefficient of the VFPS is varied in the form of a curve. there is no torsional coupling with lateral movement of the system). Soni 3. JSEE / Winter 2010. respectively) of the building isolated with the VFPI.and ydirections. No overturning or tilting takes place in the superstructure during sliding over the VFPI. 11. The VFPI is isotropic (i. Panchal and D. Modeling of multi-story building and the VFPI.and y-directions. a1 and a2 are the parameters that describe the variation of the friction coefficient along the sliding 2 2 surface of the VFPS.R. The slider of the isolator is assumed to have point contact with the sliding interface. there is same isolation period and the coefficient of friction in two orthogonal directions of the motion in the horizontal plane). 4 .The fault normal and parallel components of near-fault ground motion are applied in two horizontal directions (referred as x. Superstructure is considered to be symmetric with respect to two orthogonal horizontal directions (i. 5. Variable Friction Pendulum System The VFPS [12-13] in regards of details and operation is similar to the FPS. The modeling of the VFPI is also shown in Figure (2). 8. 2. 174 3. 4. where µ0 is the initial value of friction coefficient. 4. 6. Modeling and Idealization of Building Isolated by the VFPI Figure (2) shows the structural system under consideration which is an idealized N-story sheartype building resting on the VFPI. then provides stiffness for moderate inputs.

the sliding of the system takes place if  Fx   Fy    +   =1 F  F   s  s 2 2 [Keff ] {∆ x} = {Peff }+ {∆ f } (12) (10) Note that Eq. However. respectively.. and {∆ f } is the incremental frictional force vector. xN . to be noted that the equations of motion of the sliding structures in two orthogonal directions are coupled during the sliding phases due to interaction between the frictional forces. this interaction effect is ignored if the structural system is modeled as a 2-D system.and y-directions. {&&g} = (&&g + &&b. In such cases. the corresponding curve which separates the sliding and non-sliding phases is a square as shown in Figure (3a) by dashed lines.. {P } is the eff effective excitation vector.and ydirections. (10)). zb) approaches zero. yN}T is the displacement vector of the superstructure relative to the base mass. respectively.. Criteria for Sliding and Non-Sliding Phases In a non-sliding phase (i. during the sliding phase of motion..e. and over-dots indicate derivative with respect to time. the equations of motion are solved in the incremental form by employing the Newmark-β method assuming linear variation of acceleration over the small time interval. k1 and c1 are the stiffness and damping of the first story of the superstructure.and ydirections. the system changes to non-sliding phase from the sliding phase whenever & the resultant velocity of the base mass (i.e. Therefore. [C] and [K] are the mass.. 6. The equations governing the motion of an isolated N-story flexible shear-type building isolated with VFPI under the two horizontal components of earthquake excitation are expressed as & [M ]{&&} + [C ]{x} + [ K ]{x} = [M ] [r ]{&&g } x z & mb &&b + Fbx − c1 x1 − k1 x1 = −mb &&g x x & mb &&b + Fby − c1 y1 − k1 y1 = −mb &&g y y (9a) (9b) (9c) where [M]. (10) depicts the circular interaction between the frictional forces mobilized at the interface of the VFPI. 4 where Keff is the effective stiffness matrix. are expressed as: ∆ f x = Fs cos (θt + ∆t) − Fxt (13a) 175 .. the resultant of the frictional forces mobilized at the interface of the VFPI is less than the limiting frictional force ( Fx2 + Fy2 < Fs ) [21].and y-directions. As a result. ∆t. Thus. y2. 11. respectively. The incremental equations in terms of unknown incremental displacements are expressed as 5. y1. The system remains in the non-sliding phase inside the interaction curve. consider Figure (3b). {∆ x} is the incremental displacement vector. the mobilized frictional forces are coupled with each other by the circular interaction (see Eq. &&b and &&b x y are the acceleration of the base mass relative to the ground in the x. Further.. Fbx and Fby are the restoring forces of the VFPI in the x. T denotes the transpose. The & & system starts sliding (i. the incremental frictional forces in the x. xi and yi is the lateral displacement of the ith floor relative to the base mass in x. No.. respectively. [r] is the influence coefficient matrix. Also. Vol. the direction of the sliding of the system with respect to the x-direction is expressed as y  & θ = tan−1  b  x   &b  (11) & & where xb and yb are the velocities of the base mass relative to the ground in x. At time t the frictional forces are at point A on the interaction curve and move to point B at time t + ∆t.e. Solution of Equations of Motion The frictional forces mobilized in the VFPI are non-linear functions of the displacement and velocity of the system in two orthogonal directions. there are 2 x (N+1) dynamic degrees of freedom for the N-story superstructure.and y-directions. With a view to determine the incremental frictional forces. &&g + &&b)T z x x y y && y is the vector of base acceleration.. It is JSEE / Winter 2010. &&b = &&b = 0 and x y & & xb = yb = 0). respectively. respectively.{x} = {x1. Since the frictional forces oppose the motion of the system. of the size 2N × 2N . &&b ≠ &&b ≠ 0 and xb ≠ yb ≠ 0) x y as soon as the resultant of the frictional forces attains the limiting frictional force. xg and &&g are the earthquake ground acceleration in the x. damping and stiffness matrices of the superstructure..Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions At each floor and base mass two lateral dynamic degrees of freedom are considered. Therefore. x2..and y-directions. respectively.

JSEE / Winter 2010. the sliding velocity less than 1×10-8m/sec is assumed to be zero for checking the transition from sliding to non-sliding phase. (15)) which in turn depend on the incremental displacements (∆ xb and ∆ yb ) at the current time step. (12) and solve for ∆ xb and ∆ yb. the number of iterations in each time step is taken as 10. Iterate further. 3. the incremental frictional forces (∆ f x and 176 where ε is a small threshold parameter. In order to determine the incremental frictional forces at the sliding support. Calculate the velocities at time t + ∆ t using &t &t & incremental velocities (i.0001sec. Soni ∆ f y) should be known at any time interval. therefore.V.R. At the end of each time step the phase of the motion of the system should be checked. Interaction and incremental frictional forces in two orthogonal directions of the VFPI. (∆ Fx ) j +1 − ( ∆ Fx ) j (∆ Fx ) j Figure 3.. When the convergence criteria are satisfied. & & 2. Vol. (14). 4 . Since the frictional forces are opposite to the motion of the system. No. Panchal and D. For the present study. 11.P. The response of the sliding structures is quite sensitive to the time interval. j = 1 in Eq. xb+ ∆ t = xb + ∆ xb and t + ∆t & t & & yb = yb + ∆ yb) and compute the revised incremental frictional forces ∆ Fx and ∆ Fy from Eq. As a result. the results are obtained with maximum time interval. (15). until the following convergence criteria are satisfied for both incremental frictional forces. The steps of the procedure considered are as follows: 1. Assume ∆ Fx = ∆ Fy = 0 for iteration..e. 4. ∆ t. an iterative procedure is required to obtain the required incremental solution. Calculate the incremental velocity ∆ xb and ∆ yb using the ∆ xb and ∆ yb. Further. ≤ε (16a) t ∆ f y = Fs sin (θt + ∆ t) − Fy (13b) (∆ Fy ) j +1 − (∆ Fy ) j (∆ Fy ) j ≤ε (16b) where the superscript denotes the time. the acceleration of the system at time t + ∆ t is evaluated directly from the equilibrium of system Eq. i. The incremental frictional forces involve the system velocities at time t + ∆ t (see Eq. and initial conditions at the beginning of sliding and non-sliding phases. the velocity of the sliding structure at time t + ∆ t is calculated using incremental velocity. (12). the incremental frictional forces are expressed as ∆ Fx = Fs &t xb+ ∆ t − Fxt (x& ) + ( y& ) t + ∆t 2 b t + ∆t 2 b (15a) ∆ Fy = Fs &t yb+ ∆t (x& ) + ( y& ) t + ∆t 2 b t − Fy t +∆t 2 b (15b) In order to solve the incremental matrix Eq. (9).e. In order to avoid the unbalance forces. ∆ t = 0. the angle θt + ∆t is expressed in terms of the relative velocities of the system at time t + ∆ t by  yt + ∆t  & θt + ∆t = tan−1 b+ ∆ t   xt   &b  (14) Substituting for θt + ∆t in Eq. The superscript to the incremental forces denotes the iteration number.

000 14. The VFPS isolators are characterized by two parameters. the PGA of fault normal component is relatively higher than that of fault parallel component. N. the superstructure and the base mass of the isolated structural system under consideration can be completely characterized by the parameters namely. and the coefficient of friction. In the present study. No. Numerical Study For the present study. The superstructure considered has five stories with fundamental time period. PGV= Peak Ground Velocity and PGA= Peak Ground Acceleration JSEE / Winter 2010.89 0.and y-directions. µ.3 52. Six pairs of near-fault ground motions are used as input ground motions in order to effectively study the dynamic behaviour of building isolated with the VFPI under bi-directional excitation. On the other hand. Thus. are held constant. California El Centro Array #5 October 15. for simplicity the stiffness of all the floors is taken as constant and expressed by the parameter k. and the ratio of base mass to the superstructure floor mass. namely the period of the base isolation.e. The spectra of these ground motions indicate that the ground motions are recorded at a firm soil or rock site. Furthermore. Ti . Ts. mb/m. ξs and mb/m. damping ratio of the superstructure. namely initial time period.1 38. µ. Ts.39 0. the acceleration and displacement spectra of the six ground motions for 5% damping are shown in Figure (4). Near-Fault Ground Motions January 17. California June 28. The damping matrix of the superstructure. From this table. 1994 Northridge. it is found that the average spectral acceleration and displacement Table 1.34 October 15. The value of µ has been considered as 0.0sec. xb and yb) in the x.03 18. an initial time period of 1. mi = m for i = 1 to N). The top floor absolute acceleration and base shear are directly proportional to the forces (shear force and bending moments) exerted in the superstructure due to the earthquake ground motion. California El Centro Array #7 Note: PGD = Peak Ground Displacement. the base shear (i.1 39. Also.284 39.. It is constructed by assuming the modal damping ratio which is kept constant in each mode of vibration.5sec and damping ratio.Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions 7.65 0. [M]. Fbx and y y y y Fby ) and the isolator displacement (i. The mass ratio.46 Parallel Component PGD (cm) 9.73 0.02. is assumed to be unity. For comparison. the mass matrix of the superstructure. The coefficient of sliding friction. is diagonal and characterized by the mass of each floor which is kept constant (i.900 31. On the other hand.e.01m and d = 0.0sec.2 49. µmax. it is found that these near-fault ground motions have a variety of PGA. &&a = &&N + &&b + &&g x x x x and &&a = &&N + &&b + &&g ).02 is selected for all investigations.3 70.950 60.. Some characteristics of near-fault ground motions considered in the study. California January 17.5 49. From this figure. [C]. For all investigations. mb/m. In case of the VFPS isolator. a coefficient of sliding friction of 0.71 0. 1992 Landers. The response quantities of interest are the top floor absolute acceleration (i.420 36. ξs = 2% of critical damping. and the peak frictional coefficient. 1994 Northridge.5sec and a peak friction coefficient of 0. the fundamental time period of the superstructure. T i . Ts = 0. In case of the FPS isolators. 4 177 . 1994 Northridge. California January 17. and the coefficient of friction.5 55.4 17. is not known explicitly.and y-directions.1 230 76. the superstructure parameters. Vol. 1979 Imperial Valley.15 are chosen for all investigations. 11. California Recording Station Sylmar Rinaldi Newhall Lucerne Valley Normal Component Duration PGD PGV PGA (sec) (cm) (cm/sec) (g) 60.e. PGV and PGD. the parameters of the VFPI are selected as b = 0. the VFPI isolator is characterized by two parameters.000 49.e. µ. The fundamental time period and damping ratio of the superstructure are considered to be equal in the x. in the VFPS can be defined by the initial time period of the VFPS.6 184 150 218 PGV (cm/sec) 53. examples with the FPS and VFPS isolators are also taken with isolation period of 2. Tb . number of stories in the superstructure. Some characteristics of these recorded near-fault ground motions are summarized in Table (1). ξs. The value of k is selected to provide the required fundamental time period of superstructure. It is observed that in most of the cases.2 PGA (g) 0. the relative displacements of the VFPI are crucial from the design point of view of the isolator.6 0.1m (FVF 10 per m) so that it has initial time period of 2.1 122 175 119 136 98 113 0.9 60..72 0. as a fixed base.37 0. 1979 Imperial Valley..80 0.55 0.

This is expected as the horizontal stiffness of the VFPI is lower than that of the FPS and the VFPS. especially top floor absolute acceleration and base shear of structures with the VFPI are considerably reduced as compared to those structures with the FPS and VFPS isolators..3sec). In order to give further insight into the difference in the behavior of structures with the VFPI. 7. and correlation between xb and yb (bottom).V. whereas the isolator displacement of the VFPI exceeds that of the FPS and the VFPS. are also observed in Figure (6) which shows the comparison of the hysteresis loops of the VFPI.3sec). Panchal and D. Under near-fault ground motions.e.2.1. Comparison of the Isolators Considering Interaction of the Friction Forces Table (2) shows the comparison of the peak response quantities of the three isolators considering the interaction of the friction forces. 1994 (Sylmar) near-fault ground motion are 63.02) under different near-fault ground motions (top). FPS and VFPS isolators. Vol. this feature of the VFPI reduces its effectiveness in comparison to the FPS and the VFPS. 7. the responses. 11. Soni Figure 4.R.and y-Directions Figure (7) shows the time variation of xb and yb of the building isolated with the VFPI (Ti = 2. 1994 (Sylmar) and Northridge. 4 . values of the fault normal and parallel component is almost the same for low periods (i. the spectral acceleration and displacement component of fault normal component is significantly larger than the fault parallel component.. costly flexible connections for 178 utilities and an extensive and expensive loss of space for a seismic gap.e. time histories of top floor absolute acceleration and isolator displacement are shown in Figure (5) for Northridge. This table shows that in most of the near-fault ground motions. Comparison between Isolator Displacement in the x. in the range 0-0. Similar trends observed in Figure (5) and Table (2). 1994 (Rinaldi) near-fault ground motions. For longer periods (i. Acceleration and displacement spectra of the six near-fault ground motions for 5% damping. beyond 0.P. No. FPS and VFPS isolators.588 JSEE / Winter 2010. The peak values of xb and yb for Northridge. Such large isolator displacement of the VFPI will lead to the requirement of very large isolators.0sec and µ = 0.

7898 0. 1994 (Rinaldi) Isolated (VFPI) Isolated (VFPS) Isolated (FPS) Northridge.0346 120.4930 43.8440 9.8165 0.8360 82.15) and VFPI (Ti = 2.0342 0. 4 179 . 1994 (Newhall) Isolated (VFPI) Isolated (VFPS) Isolated (FPS) Landers.02) under Northridge.8890 40.6330 0.1867 0.4070 37.0303 0.1895 0.9760 18.9040 yb (cm) 31.3198 0.3160 0.0339 1.0287 123.4126 0.4024 0. 1994 (Sylmar) and Northridge.1239 0.1010 0.0500 74.6420 50.8008 0.1546 71.0266 0.7162 0.7407 0.8500 0.8309 0.3720 44.3075 0.5691 0.5163 0. Time variation of top floor absolute acceleration and isolator displacement of five-story building isolated with the FPS (Tb = 2.9390 10.9297 0. Near-fault Ground Motions Building Condition Isolated (FPS) Northridge.7590 6. 1992 (Lucerne Valley) Isolated (VFPI) Isolated (VFPS) Isolated (FPS) Imperial Valley.3700 0.4739 0.2037 0.2768 0.2324 0.8564 0.7525 0. 1994 (Rinaldi) near-fault ground motions.Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions Table 2.5720 Figure 5.5 sec and µmax = 0.9000 102. 1994 (Sylmar) Isolated (VFPI) Isolated (VFPS) Isolated (FPS) Northridge.0sec. Ti = 1.2542 0. JSEE / Winter 2010.1825 0.1519 0.1994 0. No. 1979 (El Centro Array #5) Isolated (VFPI) Isolated (VFPS) Isolated (FPS) Imperial Valley.0287 0.5370 0.0252 113.9270 9.5880 40.2927 0.6420 30.4165 0.3096 0.4470 0.5275 0.7210 16.2053 17.8999 13.9400 72. VFPS (Tb= 2.3140 22.0610 72.2800 0.9400 63. 11.2793 0.4277 0.4782 0.02).0200 42.6904 0.2426 0.2897 0.7835 24.0sec and µ = 0.0930 22.4382 0.0303 0.9780 63.2508 0.2255 0.4200 0.2311 21. Vol.3168 0.7293 0.8493 0.1604 0. 1979 (El Centro Array #7) Isolated (VFPI) Isolated (VFPS) &&a (g) &&a (g) Fbx(W) Fby (W) xb (cm) x y 0.0340 0.3831 0. Comparison between peak response quantities of various isolators considering the interaction of the friction forces.5940 47.1630 0.5101 0.0355 0.1410 0.3086 0.0045 0.1973 0.3107 0.2426 23.8440 22.0354 0.0 sec and µ = 0.7048 0.4250 22.4260 0.

180 JSEE / Winter 2010.P. respectively whereas for Northridge. From this figure. Panchal and D.0sec and µ = 0. is significantly larger than the corresponding peak isolator displacement in fault parallel direction. No.R.02) under different nearfault ground motions. 11. yb. Bottom: Correlation between xb and yb.642 and 40. VFPS and VFPI isolators.V. Vol. Comparison between hysteresis loops of the FPS. Figure 7. This is expected due to the fact that spectral content of fault normal component Figure 6. 1994 (Rinaldi) near-fault ground motion are 63. it can be noticed that the peak isolator displacement in fault normal direction. Soni and 16.844cm. respectively. xb.425cm. Top: Time variation of xb and yb of five-story building isolated with VFPI (Ti = 2. 4 .

Figure (8) shows variation of peak displacements. is larger than displacement.. and the corresponding displacement. xb.3. yb. zb. xb. xb. zb and zm of the five-story building isolated with the VFPI against FVF for different types of near-fault ground motions.e. of five-story building isolated with the VFPI against the FVF for various near-fault ground motions. yb. 1994 (Sylmar) near-fault ground motion. xb and yb) in x. is much smaller in comparison to the corresponding displacement due to the normal component. it is observed that the response of isolated systems to fault normal and fault parallel components are more or less uncorrelated as the maximum displacement in the fault normal direction occurs at a different time than that in the fault parallel direction. 11. zb. Influence of Bi-Directional Interaction of the Friction Forces The time variation of the absolute acceleration of && superstructure (i. zm.e. The displacement. JSEE / Winter 2010. The response is plotted for both considering and ignoring the interaction of the frictional forces of the VFPI. 7. the displacement due to parallel component. yb. zb. the superstructure experiences Figure 8. Furthermore.Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions is considerably larger than the fault parallel component of near-fault ground motion.. The absolute acceleration of the superstructure is relatively less for considering the effects of the interaction of frictional forces as compared to those without interaction. there is no significant difference between the peak resultant displacement. Similar trends can be also observed from Table (3). xb. The displacement. zm. 4 181 . No. Vol. implying that the peak resultant displacement of the isolators is mainly contributed by the displacement due to the normal component of the near-fault ground motions. As noted earlier. xa and &&a ) and the sliding base y displacement (i. denotes the square root of the sum of the squares (SRSS) of the peak values of xb and yb. Thus.and y-directions are shown in Figure (9) for Northridge. Variation of the peak isolator displacements. Furthermore. and zm . confirming that the peak displacements in the isolation system due to the normal and parallel components of the near-fault motions do not occur at the same time.

7753 0.4620 75. Soni Table 3. 11.5880 --62.0523 65.0359 0.0359 0.0359 0.0359 0.5300 135.6026 0. Time history of the absolute acceleration of the superstructure and isolator displacement to the Northridge.2568 (El Centro Array #7) Isolated (Interaction) 0. 1994 (Sylmar) Building Condition Non-isolated && xa (g) &&a (g) Fbx (W) Fby (W) y 1.0354 1.0542 0.0359 0.9800 yb (cm) --15.3800 161.0359 0.2037 0. 1992 (Lucerne Valley) Isolated (No interaction) 0.1604 0.0359 0.3521 1.0342 1.1700 63.8500 153. Peak response quantities of non-isolated building and VFPI-isolated building (Ti = 2sec and µ = 0.0266 0. Near-Fault Ground Motions Northridge.5312 0.0443 116.0340 2.7300 140.4580 --- zm (cm) --64.4818 88.0346 1.2426 0.1825 4.3206 0.1639 126.2477 0.0355 2. 1994 (Rinaldi) Isolated (No interaction) 0.1973 Non-isolated 1.2835 0.1222 Isolated (No interaction) 0.9760 --125.5110 63.0252 xb (cm) --62.7500 ----- Non-isolated 1.P.7700 113.1519 0.9530 72.9270 ----- 114.4362 4.2324 0.4200 --22.1010 167.9180 40.3789 0.2565 0.3189 1.6420 --81.1239 0.3900 120.0303 1.8440 --27. No.0523 64.7810 --68. 1994 (Newhall) Isolated (No interaction) 0.0359 0.5309 0.6620 74.5888 0.3250 Isolated (Interaction) Non-isolated 0.0303 0.02).V.9000 102.2993 0.0359 0.3075 126.1578 0.0371 167.0287 0.9390 --77. 1979 Isolated (No interaction) 0.3700 --147.4810 zb (cm) --64.3950 --83.9300 Figure 9.2256 0.0360 --83.3692 2.4504 0.9177 Imperial Valley.1895 2. 1979 Isolated (No interaction) 0. 4 .7450 --68.0371 123.4108 0. 182 JSEE / Winter 2010.5546 0.3190 --- 2.8422 Imperial Valley.1410 3.2975 Isolated (Interaction) Non-isolated 0. Panchal and D.9496 Landers.6985 2.0287 0.4310 44.R.8440 --13. 1994 (Sylmar) near-fault ground motion.0443 114.4250 --19.9843 0.2615 (El Centro Array #5) Isolated (Interaction) 0.4620 64.1890 82.2774 Northridge.3075 Isolated (Interaction) Non-isolated 0. Vol.0359 0.6983 0.0339 0.1639 140.4818 94.4386 Isolated (Interaction) 0.0359 0.0300 --111.5202 Northridge.0359 0.4140 16.

. the absolute acceleration of the system with sliding base is less sensitive to the time period of the superstructure in comparison with fixed base system. In Figure (14). The under prediction of the isolator displacement is crucial from the point of view of designing the friction base isolators. Further. Vol. if the bi-directional interaction effects are ignored and the system is idealized as a 2-D system. Therefore. (10)).e. the isolator displacements are relatively more for considering the interaction effects in comparison to that without interaction effects. Time history of the absolute acceleration of the superstructure and isolator displacement to the Northridge. ( &&a ) max+ ( &&a ) max ) against Ts under different x 2 y 2 near-fault ground motions. the bi-directional interaction effects of frictional forces of VFPI under near-fault ground motions must be rigorously considered in the analysis of the structure. 1994 (Rinaldi) near-fault ground motion. 4 183 . The figure clearly shows that the peak isolator displacement is significantly higher for considering the interaction as Figure 10. No. there is more isolator displacement. The absolute acceleration spectra of the superstructure without sliding support (referred to as non-isolated) are also shown in order to study the effectiveness of the sliding support. and as a result. This is due to the fact that when the interaction is taken into consideration the structure starts sliding at a relatively lower value of the frictional forces mobilized at the sliding interface (refer to the sliding Eq. The figure indicates clearly that the sliding support is quite effective in reducing the earthquake response of the superstructure. 11. JSEE / Winter 2010. the absolute acceleration of the superstructure is less for considering the interaction as compared to those without interaction. Similar trends in the response of the building isolated with the VFPI are found in Figure (10) for Northridge. Similar differences are also found in Figures (12) and (13) and Table (3). The figure indicates that for all values of superstructure time periods. This implies that there is significant over prediction of the superstructure accelerations and under prediction of the isolator displacements under near-fault ground motions.Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions less earthquake forces when the interaction of the frictional forces is considered in the analysis. the variation of the resultant peak sliding base displacement is plotted against µ for various near-fault ground motions. Figure (11) shows the variation of the resultant peak absolute acceleration of the superstructure (i. On the other hand. 1994 (Rinaldi) near-fault ground motion.

Soni Figure 11.0 sec and µ = 0. 4 .P.02) under various near-fault ground motions. Plot of peak absolute superstructure acceleration of five-story building isolated with the VFPI against the FVF for various near-fault ground motions (considering with and without interaction). Effects of bi-directional interaction of frictional forces on peak absolute acceleration of the superstructure of five-story building isolated with the VFPI (Ti = 2.V. Vol. Figure 12. Panchal and D.R. 184 JSEE / Winter 2010. 11. No.

Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions Figure 13. Friction coefficient variation of peak absolute superstructure acceleration of five-story building isolated with the VFPI under various near-fault ground motions (considering with and without interaction). 4 185 . Vol. Figure 14. 11. JSEE / Winter 2010. Comparison of isolator displacement of five-story building isolated by the VFPI during various near-fault ground motions (considering with and without interaction). No.

R1 and R2 against friction coefficient of the VFPI. 186 JSEE / Winter 2010. Thus. Soni compared to those without interaction. Note that similar effects of bi-directional interaction of frictional forces for structures isolated by Teflon sliding bearing were observed by Mokha et al [22] and the same are further confirmed in the present study for VFPI-isolated structures. No. of five-story building isolated with the VFPI against the fundamental time period of the superstructure for various near-fault ground motions (considering with and without interaction). denotes the ratio of peak resultant isolator displacement with interaction to the corresponding displacement without interaction of frictional forces.V. In Figure (16). R1. is plotted against the friction coefficient of VFPI for various near-fault ground motions.P. 11. Similar differences are also observed in Figure (15). zb. there is a need to consider the bi-directional interaction effects of frictional forces on the response. Variation of ratios. 4 . The ratio. the variation of ratios. R1. The ratio. Vol. is an index of the bi-directional Figure 15. Plot of peak isolator displacement. Panchal and D.R. R1 and R2. Figure 16.

J. Moreover. T. 187 . the isolator displacement in the VFPI is more than that of the VFPS and the FPS.W. it is found that these effects are strongly dependent on the input ground motions. D. R1.e. xb. (1995). 8. whereas the top floor absolute acceleration and the base shear are less than that of the VFPS and the FPS. Furthermore. In order to verify the effectiveness of the VFPI under bi-directional near-fault ground motions. the effects of bi-directional interaction of friction forces of the VFPI increase with the increase in the friction coefficient. a parametric study has been carried out to critically examine the influence of important parameters on bi-directional effects of frictional forces of VFPI. Thus. the following conclusions may be drawn: v Under bi-directional near-fault ground motions. No.. The interaction between mobilized frictional forces of the VFPI in two horizontal directions is duly incorporated in the governing equations of motion of the building isolated with the VFPI.026-1. v The resultant maximum isolator displacement of building isolated with the VFPI is mainly due to the normal component of the near-fault ground motions.J.. “Near-Source Ground Motion and Its Effects on Flexible Buildings”. which is only marginally above the maximum isolator displacement in the fault normal direction. the difference in the sliding displacements for the two cases is relatively less. Vol.. 4 response of the VFPI-isolated building with interaction is compared with those without interaction in order to demonstrate the significance of the bi-directional interaction between the mobilized frictional forces of VFPI. values close to unity justify the 2-D idealization of the system and the interaction of the frictional forces may be ignored. then the superstructure acceleration and base shear will be overestimated and the sliding displacement will be underestimated. z b the corresponding peak isolator displacement due to fault normal component. From the trends of the numerical results of the present study. and Halling.F. 569-605. v The peak isolator displacement of building isolated with the VFPI under fault normal and parallel components of near-fault ground motion are found to be more or less uncorrelated. On the other hand.H. 2. Heaton. The JSEE / Winter 2010. bi-directional interaction effects are relatively less for the low values of the friction coefficient. v Under near-fault ground motions. Hall.044. Halling. frequency variation factor (FVF) and friction coefficient of VFPI. This indicates that effects of bi-directional interaction increases with the increase in the friction coefficient. the contribution of fault parallel component can be neglected in calculating the peak resultant isolator displacement. and Wald... v The bi-directional interaction of frictional forces has noticeable effects on the response of the building isolated with the VFPI. R2. with and without interaction). As a result.F.Seismic Performance of Variable Frequency Pendulum Isolator under Bi-Directional Near-Fault Ground Motions interaction effects of frictional forces and values significantly different from unity imply significant interaction effects. This is due to the fact that for lower value of friction coefficient. the R2 is the ratio of peak resultant isolator displacement. D.. The ratio.H. Heaton. J.J. Hall. In other words. the isolation system remains most of the time in the sliding phase for both cases of excitation (i. This indicated that the resultant isolator displacement of building isolated with VFPI under near-fault ground motion may be obtained solely from the normal component. If the interaction of the frictional forces at the sliding interface is ignored. M. T. is not much influenced by the variation of friction coefficient. On the other hand.. Conclusions The response of flexible five-story building isolated with the variable frequency pendulum isolator (VFPI) under bi-directional near-fault ground motions is investigated using standard numerical technique. The contribution from the parallel component in the resultant displacement may be ignored. with addition of about 5% to incorporate the contribution from the parallel component. It varies in the range of 1. the seismic responses are compared with that of the same building isolated by the variable friction pendulum system (VFPS) and friction pendulum system (FPS). The ratio. Earthquake Spectra. 11. The important parameters considered are the superstructure time period. References 1. Wald. The resultant maximum isolator displacement can be obtained from the fault normal component by increasing about 5% to account the contribution of the parallel component. increases with the increase of friction coefficient of VFPI. 11.

Earthquake Engineering and Structural Dynamics. of Building Officials. (1990).. 4 . Science. (2003).R. and Yao. Whittier.-C. “Response of Pure-Friction Sliding Structures to Bi-Directional Harmonic Ground Motion”... 25. R. R. Soni M. R.-C. 193-216. JSEE / Winter 2010. 4. of Structural Engineering.H. Panchal. Gomez-Soberon. and Design. and Chen. R. Journal of Sound and Vibration.S. 53-69. (2006). (2000). ASCE. 568-584.P. and Jangid. P. Jangid.C. “Teflon Bearing in Base Isolation II: Modeling”. 8. G.M. 243. 13. 206-211. Earthquake Engineering and Structural Dynamics. “Response of High-Rise and BaseIsolated Buildings to a Hypothetical Mw 7. 34.. 7.M. Earthquake Spectra.S. “Variable Friction Pendulum System for Near-Fault Ground Motions”.. R. 193-205.W. “Seismic Response of Structures with Variable Friction Pendulum System”. (2004). D. 203. 259-272. X. CA. 4. and Chen. 883-902.R. 97-104..S..-J. 240-261. V. Panchal. 505-532. C. 455-474. 14. 3. 773-791.S. 929-944. International Conf..R. 1027-1046. (1993). El Naggar. B. 129-139. 18. A. and Kelly. Vol. No. Tsai. Earthquake Engineering and Structural Dynamics.. Chiang. 30. 603-627.S. “Dynamic Analysis on Structures Base-Isolated by a Ball System with Restoring Property”. C. N. 11. “Base Isolation for Near-Fault Motions”. Q. Vafai. (1996).A. Earthquake Spectra. Earthquake Engineering and Structural Dynamics. 16. (2001). Zhou. R. Jangid. Earthquake Engineering and Engineering Vibration.. 267. R. Panchal and D. Lu. 12.S. C. “Some Retrofit Options for the Seismic Upgrading of Old Low-Rise School Buildings in Mexico”.R. Q. Panchal. 9. Mokha. “Verification of Friction Model of Teflon Bearings under Triaxial Load”. A. Uniform Building Code (1997). “Seismic Isolation of Buildings Subjected to Typical Subduction Earthquake Motions for the Mexican Pacific Coast”. 29. M.. Feng. Murnal. 691707. 119. and Jangid. 25. “Aseismic Design of Structure-Equipment Systems Using Variable Frequency Pendulum Isolator”. B. (2009). Nuclear Engineering and Design. and Munoz-Loustaunau. 188 12. R.. (1996). M. Matsagar. R. A.. Wang. Tena-Colunga. Tsai. and Sinha. 27. V. Rao.S. 19. 5. “Performance of Sliding Systems under Near-Fault Motions”. 22. Earthquake Engineering and Structural Dynamics.V. Jangid. Structural Control and Health Monitoring. J. (2003). and Reinhorn. A. and Jangid. Mokha. 116. 1301-1306. Constantinou. V. Nuclear Engr. “Seismic Response of Simply Supported BaseIsolated Bridge with Different Isolators”. Engineering Structures. Engineering Structures. Journal of Earthquake Engineering. Q. 19. and Ahmadi.S.M. Jangid. of Structural Engineering. Chiang. “Seismic Behavior of Variable Frequency Pendulum Isolator”.-J. 13. 20. J. Tena-Colunga. (1995). V. M.. and Jangid.S. (1998). 1719-1730. M.C. 17. 15. 21. (1997). A. A. and Sinha. (2001). Earthquake Regulations for Seismic Isolated Structures. 15-29. “Experimental Evaluation of Piecewise Exact Solution for Predicting Seismic Responses of Spherical Sliding Type Isolated Structures”. (2008). and Reinhorn... 231.. 13. (2008). “Response of Sliding Structures to Bi-Directional Excitation”. “Seismic Response of Sliding Structures to Bi-Directional Earthquake Excitation”. 6.B. 7. Constantinou. P.S. Earthquake Engineering and Structural Dynamics. “VFPI: An Isolation Device for Aseismic Design”. R. Pranesh. J. T. 10. ASCE. (1997). 11. (2005). “Seismic Isolation of Buildings with Sliding Concave Foundation (SCF)”. International Journal of Applied Science and Engineering. A.S. “Finite Element Formulations and Theoretical Study for Variable Curvature Friction Pendulum System”. and Jangid R. (2001). 32. T. A. Hamidi.0 Blind Thrust Earthquake”. 15.S.