Nuclear Engineering and Design 217 (2002) 120

Non-linear seismic response of base-isolated liquid storage

tanks to bi-directional excitation
M.K. Shrimali, R.S. Jangid *
Department of Ci6il Engineering, Indian Institute of Technology Bombay, Powai, Mumbai 400 076, India
Received 21 May 2001; received in revised form 8 January 2002; accepted 6 March 2002
Abstract
Seismic response of the liquid storage tanks isolated by lead-rubber bearings is investigated for bi-directional
earthquake excitation (i.e. two horizontal components). The biaxial force-deformation behaviour of the bearings is
considered as bi-linear modelled by coupled non-linear differential equations. The continuous liquid mass of the tank
is modelled as lumped masses known as convective mass, impulsive mass and rigid mass. The corresponding stiffness
associated with these lumped masses has been worked out depending upon the properties of the tank wall and liquid
mass. Since the force-deformation behaviour of the bearings is non-linear, as a result, the seismic response is obtained
by the Newmarks step-by-step method. The seismic responses of two types of the isolated tanks (i.e. slender and
broad) are investigated under several recorded earthquake ground to study the effects of bi-directional interaction.
Further, a parametric study is also carried out to study the effects of important system parameters on the effectiveness
of seismic isolation for liquid storage tanks. The various important parameters considered are: (i) the period of
isolation, (ii) the damping of isolation bearings and (iii) the yield strength level of the bearings. It has been observed
that the seismic response of isolated tank is found to be insensitive to interaction effect of the bearing forces. Further,
there exists an optimum value of isolation damping for which the base shear in the tank attains the minimum value.
Therefore, increasing the bearing damping beyond a certain value may decrease the bearing and sloshing displace-
ments but it may increase the base shear. 2002 Elsevier Science B.V. All rights reserved.
www.elsevier.com/locate/nucengdes
1. Introduction
Liquid storage tanks are strategically very im-
portant structures, since they have vital uses in
industries, nuclear power plants and are con-
nected to public life. There has been a number of
reports on damage to liquid storage tanks such as
in the Chilean earthquake (Steinbrugge and Ro-
drigo, 1963), California, Livermore earthquake
(Niwa and Clough, 1982) and San Juan earth-
quake (Manos, 1991). The earthquake motion
excites the liquid contained in the tank. A part of
the liquid moves independent of tank wall mo-
tion, which is termed as sloshing, while another
part of the liquid which moves in unison with the
rigid tank wall is known as impulsive mass. If the
exibility of the tank wall is considered then the
part of the impulsive mass moves independently
* Corresponding author. Tel.: +91-22-576-7346; fax: +91-
22-578-3557
E-mail address: rsjangid@civil.iitb.ac.in (R.S. Jangid).
0029-5493/02/$ - see front matter 2002 Elsevier Science B.V. All rights reserved.
PII: S0029- 5493( 02) 00134- 6
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 2
while the remaining accelerates back and forth
with the tank wall known as rigid mass. The
accelerating liquid, sloshing, impulsive and rigid
masses, induces substantial hydrodynamic pres-
sures on the shell wall of liquid storage tanks
which in turn generates lateral pressures (i.e. base
shear) and overturning moment. The overturning
moment produces excessive compressive stresses
at the bottom of one side of the tank. The exces-
sive stresses induce buckling of the shell of the
tank wall. The failure of liquid storage tanks is
mainly due to buckling of the tank wall, failure of
piping system and uplift of the anchorage system.
The seismic behaviour of liquid storage tank is
highly complex due to liquidstructure interac-
tion leading to a tedious design procedure from
an earthquake-resistant design point of view.
Housner (1957, 1963) developed lumped mass
model of ground supported liquid storage tank
with two-degrees-of-freedom (i.e. associated with
convective mass and impulsive mass) to investi-
gate the seismic response. It is observed that the
liquid pressure generated due to earthquake
ground motion is very important for seismic de-
sign of the tanks. Rosenblueth and Newmark
(1971) modied the expression suggested by
Housner (1963) to estimate the convective and
impulsive mass to evaluate the seismic design
forces of liquid storage tanks. Epstein (1976) has
suggested closed form expressions to calculate
lumped masses to nd seismic response of liquid
storage tanks. Haroun (1983) developed design
charts to estimate convective, impulsive and rigid
masses, assuming the liquid contained in the tank
as incompressible with irrotational ow.
The integrity of a structure can be protected
from the attack of severe earthquakes either
through the concept of resistance or isolation. In
designing a structure by resistance, it is assumed
that the earthquake forces are directly transmitted
to the structure and each member of the structure
is required to resist the maximum possible forces
that may be induced by earthquakes, based on
various ductility criteria. In the category of earth-
quake isolation, however, one is interested in re-
ducing the peak response of the structure through
implementation of certain isolation devices. Since
the forces transmitted to the superstructure have
been largely reduced, the superstructure can usu-
ally be expected to be linearly elastic, even under
the excitation of severe earthquakes. The base
isolation techniques have been developed and suc-
cessfully implemented to buildings in the past
(Kelly, 1986; Buckle and Mayes, 1990; Jangid and
Datta, 1995). However, there have been very few
studies in the past to investigate the effectiveness
of base isolation for aseismic design of liquid
storage tanks. Chalhoub and Kelly (1988) ob-
served that the sloshing response increases slightly
but the total hydrodynamic pressures decreases
substantially due to base isolation of the tanks.
Kim and Lee (1995) experimentally investigated
the seismic performance of liquid storage tanks
isolated by laminated rubber bearings under uni-
directional excitation and have shown that the
isolation is effective in reducing the dynamic re-
sponse. Malhotra (1997a) studied the seismic re-
sponse of liquid storage tanks, under
uni-directional ground motion, in which the wall
of the tank was isolated from the base plate by
horizontal exible rubber bearings. The numerical
results indicated that the isolation is quite effec-
tive in reducing the axial stresses of cylindrical
shell. The results of base-isolated liquid storage
tank indicated that the sloshing displacement is
increased but decrease in axial stress in cylindrical
shell is signicant in avoiding the buckling of the
shell (Malhotra, 1997b). Although the above stud-
ies conrm that the seismic isolation is effective in
reducing the earthquake response of the tanks,
there has been no specic study to investigate the
effects of important system parameters on the
effectiveness of base isolation for tanks under
bi-directional earthquake ground motion.
In this paper, seismic response of ground-sup-
ported liquid storage tanks isolated by the lead-
rubber bearings is investigated under two
horizontal components of earthquake ground mo-
tion. The specic objectives of the study are sum-
marized as (i) to present a method for dynamic
analysis of liquid storage tanks isolated by the
lead-rubber bearings to bi-directional earthquake
motion which incorporates the interaction be-
tween the restoring forces of the bearings in two
orthogonal directions, (ii) to study the effects of
bi-directional interaction of restoring forces on
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 3
the seismic response of liquid storage tanks. This
is achieved by comparing the seismic response of
the tanks to bi-directional excitation with consid-
eration of interaction between the restoring forces
to the corresponding response without interaction
and (iii) to investigate the inuence of important
system parameters on the effectiveness of isolation
systems. The important parameters included are:
isolation period, damping and yield strength of
the lead-rubber bearings.
2. Structural model of liquid storage tank
Fig. 1 shows the idealised structural model of
liquid storage tank mounted on isolation system.
The lead-rubber bearings are installed between the
base and foundation of the tank for isolation. The
contained continuous liquid mass is lumped as
convective, impulsive and rigid masses referred as
m
c
, m
i
and m
r
, respectively. The convective and
impulsive masses are connected to the tank wall
by corresponding equivalent spring having stiff-
ness k
c
and k
i
, respectively. The damping constant
of the convective and impulsive masses are c
c
and
c
i
, respectively. The system has six-degrees-of-free-
dom under bi-directional earthquake ground mo-
tion, two-degrees-of-freedom of each lumped
mass in two horizontal x- and y-directions. These
degrees-of-freedom are denoted by (u
cx,
u
cy
), (u
ix
,
u
iy
) and (u
bx
, u
by
) which denote the absolute dis-
placement of convective, impulsive and rigid
masses in x- and y-directions, respectively. The
tank model is assumed to have a deformable
circular cylindrical shell. The parameters of the
tanks considered are liquid height H, radius, R
and average thickness of tank wall, t
h
. The effec-
tive masses are dened in terms of liquid mass m,
from the parameters are expressed (Haroun, 1983)
as
Y
c
=1.013270.87578S+0.35708S
2
0.06692S
3
+0.00439S
4
(1)
Y
i
=0.15467+1.21716S0.62839S
2
+0.14434S
3
0.0125S
4
(2)
Y
r
=0.01599+0.86356S0.30941S
2
+0.04083S
3
(3)
where S=H/R is the aspect ratio (i.e. ratio the
liquid height to radius of the tank) and Y
c
, Y
i
,
and Y
r
are the mass ratios dened as
Y
c
=
m
c
m
(4)
Y
i
=
m
i
m
(5)
Y
r
=
m
r
m
(6)
m=yR
2
Hz
w
(7)
where z
w
is the mass density of liquid.
The natural frequencies of sloshing mass,
c
and impulsive mass,
i
are given by following
expressions (Haroun, 1983) as

i
=
P
H
'E
z
s
(8)

c
=
'
1.84
g
R

tanh

1.84
H
R

(9)
where E and z
s
are the modulus of elasticity and
density of tank wall, respectively; g is the acceler-
ation due to gravity; and P is a dimensionless
parameter expressed as Fig. 1. Structural model of base-isolated liquid storage tank.
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 4
Fig. 2. The isolation system. (a) Lead-rubber bearing and (b)
schematic diagram.
and steel plates with a central lead core. The
bearing has isotropic property, which signies
the same dynamic characteristics in all direc-
tions. The bearing is modelled such that it has
bi-linear force-deformation behaviour in hori-
zontal directions. The bearings are vertically stiff
and have initial horizontal stiffness, k
b
and vis-
cous damping, c
b
. The vertical stiffness is
derived from steel plates while parallel layers of
rubber bearings provide horizontal exibility.
The lead core yields relatively at very low shear-
ing stress leading to dissipation of seismic energy
and reduction of earthquake response
(Robinson, 1982). The non-linear bi-directional
hysteretic restoring forces in the lead-bearing are
modelled by coupled differential equations as
proposed by Park et al. (1986) which in the past
have been used by Nagarajaiah et al. (1991),
Jangid and Datta (1994). The restoring forces of
the bearings are given by
!F
bx
F
by
"
=h
k
b
0
0 k
b
n!x
b
y
b
"
+(1h)
F
y
0
0 F
y
n!Z
x
Z
y
"
(11)
where F
bx
and F
by
are the bearing forces in x-
and y-directions, respectively; h is an index
which represents the ratio of post- to pre-yield-
ing stiffness; k
b
is the pre-yielding stiffness of the
bearing; x
b
and y
b
are the relative bearing dis-
placements in x- and y-directions, respectively;
F
y
is the yield strength of bearings and the hys-
teretic components of displacement Z
x
and Z
y
are computed from following non-linear rst or-
der differential equations as
q
!Z:
x
Z:
y
"
=[G]
!x;
b
y;
b
"
(12)
where
[G] =

Ai sgn(x;
b
)Z
x
Z
x
~Z
x
2
i sgn(y;
b
)Z
y
Z
x
~Z
x
Z
y
i sgn(x;
b
)Z
x
Z
y
~Z
x
Z
y
Ai sgn(y;
b
)Z
y
Z
y
~Z
y
2
n
(13)
P=0.037085+0.084302S0.05088S
2
+0.012523S
3
0.0012S
4
(10)
3. Force-deformation behaviour of bearings
Fig. 2 shows the lead-rubber bearings consid-
ered which consist of alternative layers of rubber
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 5
where q is the yield displacement; x;
b
and y;
b
are
the relative velocities of isolation bearings in the
x- and y-directions, respectively; and i, ~ and A
are the non-dimensional parameters which control
the shape and size of force-deformation loop of
the bearing. The parameters are selected such that
the predicted response from the model matches
with experimental results. It is to be noted that
the off diagonal terms of the matrix [G] provide
the coupling or interaction between the restoring
forces of the lead-rubber bearings in two orthogo-
nal directions. It will be interesting to investigate
the effects of this interaction on the seismic re-
sponse of base-isolated tanks. This interaction
effect is ignored if the isolated tank is analyzed
under two-dimensional (2-D) idealisation in two
horizontal directions independently.
4. Governing equations of motion
The equations of motion of isolated liquid stor-
age tank subjected to earthquake ground motion
are expressed in the matrix form as
[m]{z }+[c]{z; }+[k]{z; }+{F}=[m][r]{u
g
}
(14)
where {z}={x
c
, x
i
, x
b
, y
c
, y
i
, y
b
}
T
and {F}=
{0, 0, (1h)F
y
Z
x
, 0, 0, (1h)F
y
Z
y
}
T
are the dis-
placement and restoring force vectors,
respectively; x
c
=u
cx
u
bx
and y
c
=u
cy
u
by
are
the displacements of the convective mass relative
to bearing displacements in x and y-directions,
respectively; x
i
=u
ix
u
bx
and y
i
=u
iy
u
by
are
the displacements of the impulsive mass relative
to bearing displacements in x- and y-directions,
respectively; x
b
=u
bx
u
gx
and y
b
=u
by
u
gy
are
the displacements of the bearings relative to
ground in x- and y-directions, respectively; [m],
[c] and [k] are the mass, damping and stiffness
matrix of the system, respectively; [r] is the inu-
ence coefcient matrix; and {u
g
}={u
gx
, u
gy
}
T
is
the earthquake ground acceleration vector, the u
gx
and u
gy
are the earthquake accelerations in x- and
y-directions, respectively.
The matrices [m], [c], [k] and [r] are expressed
as
[m] =

m
c
0 m
c
0 0 0
0 m
i
m
i
0 0 0
m
c
m
i
M 0 0 0
0 0 0 m
c
0 m
c
0 0 0 0 m
i
m
i
0 0 0 m
c
m
i
M

(15)
[c] =diag[c
c
, c
i
, c
b
, c
c
, c
i
, c
b
] (16)
[k] =diag[k
c
, k
i
, hk
b
, k
c
, k
i
, hk
b
] (17)
[r] =
0 0 1 0 0 0
0 0 0 0 0 1
n
T
(18)
where M=m
c
+m
i
+m
r
is the total effective mass
of the tank.
The equivalent stiffness and damping constants
of the convective and impulsive masses are ex-
pressed as
k
c
=m
c

c
2
(19)
k
i
=m
i

i
2
(20)
c
c
=2x
c
m
c

c
(21)
c
i
=2x
i
m
i

i
(22)
where x
c
and x
i
are the damping ratios of the
convective and impulsive masses, respectively.
5. Incremental solution of equations of motion
The governing equations motion of the isolated
liquid storage tank cannot be solved using the
classical modal superposition technique due to (i)
the damping in the isolation system and liquid
storage tank is different in nature because of
material characteristics and (ii) the force-deforma-
tion behaviour of the lead-rubber bearings is non-
linear. As a result, the governing equations of
motion are solved in the incremental form using
Newmarks step-by-step method assuming linear
variation of acceleration over small time interval,
Dt. The equations of motion in incremental form
are expressed as
[m]{Dz }+[c]{Dz; }+[k]{Dz; }+{DF}
=[m][r]{Du
g
} (23)
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 6
where {Dz}={Dx
c
, Dx
i
, Dx
b
, Dy
c
, Dy
i
, Dy
b
}
T
is
the incremental displacement vector; and {DF} is
the incremental hysteretic force vector of the lead-
rubber bearing expressed by
{DF}={0, 0, (1h)F
y
DZ
x
, 0, 0, (1h)F
y
DZ
y
}
T
(24)
where DZ
x
and DZ
y
are the incremental hysteretic
displacement components in x- and y-directions,
respectively.
The incremental acceleration and velocity vec-
tors {Dz } and {Dz; } over the time interval, Dt are
expressed as
{Dz }=a
0
{Dz}+a
1
{z; }
t
+a
2
{z }
t
(25)
{Dz; }=b
0
{Dz}+b
1
{z; }
t
+b
2
{z }
t
(26)
where a
0
=6/(Dt
2
); a
1
=(6/(Dt
2
)); a
2
=3;
b
0
=3/Dt; b
1
=3; b
2
=Dt/2; and the super-
script t denotes time.
The nal equations of the motion in the incre-
mental form are expressed as
[k
eff
]{Dz}={P
eff
}{DF} (27)
where [k
eff
] is the effective stiffness matrix; and
{P
eff
} is the effective excitation vector.
The matrix [k
eff
] and vector {P
eff
} are explicitly
given by
[k
eff
] =a
0
[m] +b
0
[c] +[k] (28)
[P
eff
] =[m]{r}{Du
g
}[m](a
1
{z; }
t
+a
2
{z }
t
)
[c](b
1
{z; }
t
+b
2
{z }
t
) (29)
The incremental restoring force vector involves
the incremental hysteretic displacement compo-
nents, DZ
x
and DZ
y
, which depends on the bear-
ing velocities at time, t +Dt. As a result, an
iterative procedure is required to obtain the re-
quired solution. The steps of the procedure are as
follows:
1. Assume DZ
x
=DZ
y
=0 for iteration j =1 in
Eq. (24) and solve for {Dz} by Eq. (27).
2. Calculate the incremental velocity vector using
the Eq. (26) and nd the velocity vector at
time, t +Dt by {Dz; }
t +Dt
={z; }
t
+{Dz; }. This
velocity vector will provide the velocity of the
base mass or rigid mass in x- and y-directions,
respectively (i.e. x;
b
t +Dt
and y;
b
t +Dt
).
3. Knowing the velocity of the mass in both the
horizontal directions at time t +Dt, compute
the revised incremental displacements compo-
nent, DZ
x
and DZ
y
using the third order
RungeKutta method.
4. Iterate further, until the following convergence
criteria is satised for the incremental hys-
teretic displacements component of the bear-
ing i.e.
DZ
x
j +1
DZ
x
j

Z
m
5m (30a)
DZ
y
j +1
DZ
y
j

Z
m
5m (30b)
where m is the small tolerance parameter; the
superscript denotes the iteration number; and
Z
m
is the maximum value of hysteretic dis-
placement of the bearing is expressed by
Z
m
=
' A
i+~
(31)
The nal acceleration vector is obtained from
the direct equilibrium of Eq. (14) instead of using
Eq. (25) to avoid any unbalanced forces generated
in the numerical integration scheme.
After obtaining the nal acceleration vector, the
absolute acceleration of convective and impulsive
mass in both the x- and y-directions are obtained.
The total base shear generated due to earthquake
ground motion in x- and y-directions, respec-
tively, is expressed as
F
sx
=m
c
u
cx
+m
i
u
ix
+m
r
u
bx
(32)
F
sy
=m
c
u
cy
+m
i
u
iy
+m
r
u
by
(33)
6. System parameters
The damping, stiffness and yield level of the
bearing are designed to provide the desired value
of three parameters, namely the T
b
, x
b
and F
0
expressed as
T
b
=2y
' M
hk
b
(34)
x
b
=
c
b
2M
b
(35)
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 7
F
0
=
F
y
W
(36)
where
b
=2y/T
b
is the isolation frequency; and
W=Mg is the effective weight of the liquid stor-
age tank.
The parameters, which characterize the model
of liquid storage tank, are height of liquid in the
tank (H), aspect ratio (S), damping ratio of the
convective mass (x
c
) and damping ratio of the
impulsive mass (x
i
). The isolator parameters for
the tank with lead-rubber bearings require the
specication of the parameters such as the period
of isolation (T
b
), damping ratio of the bearing
(x
b
) and yield strength level of the bearing (F
0
).
The other parameters of the lead-rubber bearing
are the yield displacement (q) and parameters of
hysteresis loop of the bearing such as A, i and ~.
However, these parameters are held constant and
values taken are: q=25 mm, A=1, i=0.5 and
~=0.5. The damping ratios for the convective
mass (x
c
) and the impulsive mass (x
i
) and are
taken as 0.5 and 2%, respectively. For the tanks
Table 1
Properties of earthquake ground motion
PGA (g) PGA (g) Component Earthquake Component
S90W 0.2144 S00E 0.3486 Imperial Valley, 1940 (El-Centro)
0.5704 Loma Prieta, 1989 (Los Gatos Presentation Center) 0.6079 N90E N00E
N90E 0.6297 N00E Kobe, 1995 (JMA) 0.8345
Fig. 3. Time variation of response of the slender tank in x-direction due to Imperial Valley earthquake (T
b
=2 s, x
b
=0.1 and
F
0
=0.05).
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 8
Fig. 4. Time variation of response of the slender tank in y-direction due to Imperial Valley earthquake (T
b
=2 s, x
b
=0.1 and
F
0
=0.05).
with steel wall the modulus of elasticity is taken as
E=200 MPa and the mass density, z
s
=7900 kg
m
3
. The ratio of post- to pre-yielding stiffness is
computed using following expression h=
b
2
Mq/
F
y
.
7. Numerical study
The seismic response of base-isolated liquid
storage tanks is investigated under two horizontal
components of earthquake ground motion. The
bi-directional interaction between the restoring
forces of the lead-rubber bearings is duly consid-
ered. The earthquake ground motions considered
are of Kobe (1995), Loma Prieta (1989) and Im-
perial Valley (1940). The peak acceleration of
different components of these ground motions is
given in Table 1. The components S90W, N90E
and N90E of Kobe, Loma Prieta and Imperial
Valley earthquake ground motions, respectively,
are applied in x-direction. The other orthogonal
components of the above mentioned ground mo-
tions are applied in the y-direction. The response
quantities of interest are base shear (F
sx
, F
sy
),
displacements of convective mass (x
c
, y
c
), impul-
sive mass (x
i
, y
i
) and isolation bearings (x
b
, y
b
).
The response of base-isolated liquid storage tank
is analyzed to study the inuence of bi-directional
interaction and system parameters on the effec-
tiveness of seismic isolation. Two different types
of tanks namely the broad and slender tanks are
considered for detailed parametric study from
Malhotra (1997a). The properties of these tanks
are: (i) aspect ratio (S) for slender and broad
tanks is 1.85 and 0.6, respectively; (ii) the height,
H, of water lled in slender and broad tanks is.
11.3 and 14.6 m, respectively; (iii) ratio of tank
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 9
wall thickness to its radius (t
h
/R) is taken 0.004
for both the tanks and (iv) the natural frequencies
of convective mass and impulsive mass for the
broad and slender tank are 0.123, 3.944 Hz and
0.273, 5.963 Hz, respectively.
The time variation of various response quanti-
ties of slender and broad tanks are shown in Figs.
36 under Imperial Valley, 1940 earthquake
ground motion. The response is shown for both
non-isolated and isolated tanks with and without
interaction of the bearing forces. The parameters
of the isolation system considered are T
b
=2 s,
x
b
=0.1 and F
y
/W=0.05. It is observed from the
gures that the isolation is quite effective in re-
ducing the base shear and impulsive displacement
of liquid storage tank signicantly. This implies
that the seismic isolation is quite effective in re-
ducing the earthquake response of liquid storage
tanks. The sloshing displacement in the isolated
slender tank is marginally increased while in
broad tank it is reduced slightly in comparison to
corresponding non-isolated tanks. Further, it is
also observed that all the seismic response quanti-
ties of slender and broad tanks with the interac-
tion of bearings forces follow the same pattern as
without interaction condition. Also, there is no
signicant difference in the peak response of the
isolated tanks obtained by interaction and no
interaction conditions. Thus, the interaction of
the restoring forces of the bearing does not have
signicant effect on the seismic response of iso-
lated tanks. The similar effects of bi-directional
interaction on the corresponding force-deforma-
tion behaviour of the lead-rubber bearing for
slender and broad tanks are depicted in Figs. 7
and 8, respectively.
The peak response quantities for the broad and
slender tanks under different earthquake ground
Fig. 5. Time variation of response of the broad tank in x-direction due to Imperial Valley earthquake (T
b
=2 s, x
b
=0.1 and
F
0
=0.05).
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 10
Fig. 6. Time variation of response of the broad tank in y-direction due to Imperial Valley earthquake (T
b
=2 s, x
b
=0.1 and
F
0
=0.05).
motion in x- and y-directions are shown in Tables
2 and 3, respectively. The parameters of bearings
considered are: T
b
=2 s, x
b
=0.1 and F
y
/W=
0.05. The percentage reduction in base shear due
to isolation of the broad tank with interaction
effect is 66.28, 74.41; 67.50, 57.96 and 65.13, 66.39
under Imperial Valley, Loma Prieta and Kobe
earthquake ground motions in x- and y-direc-
tions, respectively. The corresponding reduction
in slender tank is 68.97, 80.04; 82.23, 67.05 and
74.07, 67.16. This indicates that reduction of base
shear in slender tank is more in comparison to
broad tank. This is due to the fact that the total
base shear is mainly contributed from impulsive
component and seismic isolation is quite effective
in reducing the impulsive response of the tanks.
As the slender tanks have more impulsive mass as
compared with broad tanks, the seismic isolation
will be more effective. Further, there is signicant
increase in the sloshing displacement of liquid
mass due to isolation of the slender tank. The
effect of bi-directional interaction on the peak
response is not quite signicant as observed ear-
lier. However, the base shear and impulsive dis-
placement are reduced marginally due to
interaction of bearing forces in comparison to no
interaction condition.
In Fig. 9, variation of peak response of slender
tank is plotted against the period of isolation, T
b.
The bearing damping is considered to be 10% and
normalized yield strength as 0.05. It is observed
from the gure that the increase of exibility of
isolation system reduces the peak base shear (the
reduction is more in y-direction compared to
x-direction). The base displacement initially de-
creases due to increase in exibility of isolation
system; thereafter it increases with increase of
time period in x-direction (while in y-direction
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 11
increase in base displacement is signicant with
time period due to Loma Prieta earthquake). The
sloshing displacement is marginally decreased
with the increase of time period in both direc-
tions. Fig. 10 shows the effect of exibility of
isolation system on the peak response of broad
tank. The variation of base shear and base dis-
placement with time period of isolation system in
both the directions i.e. x- and y-directions, shows
similar trends as shown in Fig. 9. The sloshing
displacements in both the directions are not sig-
nicantly inuenced with increase in the time
Fig. 7. Force-deformation behaviour of isolation system of the slender tank under Imperial Valley earthquake (T
b
=2 s, x
b
=0.1 and
F
0
=0.05).
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 12
Fig. 8. Force-deformation behaviour of isolation system of the broad tank under Imperial Valley earthquake (T
b
=2 s, x
b
=0.1 and
F
0
=0.05).
period (except due to Kobe earthquake). This is
due to the fact that sloshing period of broad tank
is 8.13 s, which is well separated from the varia-
tion of the period considered from 1.5 to 4 s.
Hence, the seismic isolation system should be
design such that its time period is well separated
from sloshing period.
The effects of isolation damping on the peak
response of liquid storage tank are shown in Figs.
11 and 12 for both slender and broad tanks,
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 13
respectively. The period of isolation considered is
2 s and F
y
/W=0.05. The physical signicance of
damping is to dissipate seismic energy and hence
reduces bearing displacement but as a result the
acceleration transmitted to structure increases,
hence increase in base shear. The base shear rst
decreases and attains minimum value and then
increases with increase of the isolation damping.
This indicates that there exists optimum value of
isolation damping for which base shear is mini-
mum. The minimum value of base shear observed
in x-direction for slender tank at damping ratio of
0.09, 0.15 and 0.19 for Kobe, Loma Prieta and
Imperial Valley ground motions, respectively.
Similarly, damping ratio for minimum value of
base shear in y-direction for the slender tank
under the corresponding ground motions is 0.19,
0.33 and 0.19. The minimum value of base shear
observed in x-direction for broad tank at damp-
ing ratio of 0.09, 0.21 and 0.23 for Kobe, Loma
Prieta and Imperial Valley ground motions, re-
spectively. Similarly, damping ratio for minimum
value of base shear in y-direction for the broad
tank for the corresponding ground motions is
0.17, 0.25 and 0.17. For slender tanks the base
and sloshing displacements in x- and y-directions,
respectively, decrease as damping of the isolation
system increases. The variation of base displace-
Table 2
Peak response of broad and slender tanks in x-direction
Imperial Valley (1940) Earthquake Loma Prieta (1989) Kobe (1995)
Broad Slender Broad Slender Broad Tank Slender
x
c
(cm) 36.14 Non-isolated 17.50 52.05 30.47 41.54 53.08
Isolated Interaction 53.80 54.00 47.38 23.65 64.77 25.43
No interaction 54.84 53.56 45.44 23.66 62.21 26.42
Non-isolated 0.390 1.280 x
i
(cm) 0.960 2.370 0.660 1.870
0.536 0.143 0.424 Isolated 0.128 0.271 0.082 Interaction
No interaction 0.088 0.293 0.130 0.410 0.150 0.580
0.319 0.258 0.771 0.440 0.594 0.476 F
sx
/W Non-isolated
Isolated Interaction 0.099 0.087 0.137 0.143 0.154 0.166
No interaction 0.107 0.093 0.155 0.144 0.164 0.182
13.92 12.31 11.72 11.89 6.35 x
b
(cm) 7.22 Interaction Isolated
14.13 12.69 10.80 12.16 6.10 7.87 No interaction
Table 3
Peak response of broad and slender tanks in y-direction
Imperial Valley Earthquake Loma Prieta Kobe
Slender Broad Slender Tank Broad Slender Broad
Non-isolated 23.71 42.29 y
c
(cm) 142.77 55.31 37.85 23.76
44.20 Isolated 195.04 64.18 57.32 34.69 Interaction 31.95
31.20 43.78 194.71 63.97 No interaction 58.19 34.87
0.740 3.530 Non-isolated 0.910 y
i
(cm) 2.950 1.530 1.950
Isolated 0.733 0.256 0.790 Interaction 0.334 0.278 0.083
No interaction 0.088 0.292 0.335 0.790 0.260 0.760
1.056 0.609 F
sy
/W 0.883 Non-isolated 0.744 0.491 0.340
0.250 0.290 0.256 0.348 0.087 0.098 Interaction Isolated
No interaction 0.103 0.093 0.349 0.258 0.298 0.256
Isolated 20.62 y
b
(cm) 25.49 22.79 30.94 6.310 7.468 Interaction
21.13 26.08 22.32 30.96 6.320 7.470 No interaction
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 14
Fig. 9. Effect of time period on peak response of the base-isolated slender tank (x
b
=0.1 and F
0
=0.05).
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 15
ment with isolation damping for the broad tank
has trends similar to the slender tank as shown in
Fig. 11. The sloshing displacement for broad tank
is not signicantly affected by variation of isola-
tion damping (except for Kobe earthquake in
y-direction).
Fig. 10. Effect of time period on peak response of the base-isolated broad tank (x
b
=0.1 and F
0
=0.05).
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 16
Fig. 11. Effect of damping on peak response of the base-isolated slender tank (T
b
=2 s and F
0
=0.05).
The effects of yield strength of isolation system
on the peak response of slender and broad tank
are shown in Figs. 13 and 14, respectively. The
gures indicate that the base shear initially de-
creases and then increases with the increase of the
yield strength of the bearings, implying that there
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 17
exists an optimum value of bearing yield strength
for which the base shear is minimum. This is
expected that the yielding of the bearing adds the
hysteretic damping in the system and at higher
damping there is a tendency for the base shear to
increase, as observed in Figs. 12 and 13. Further,
Fig. 12. Effect of damping on peak response of the base-isolated broad tank (T
b
=2 s and F
0
=0.05).
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 18
Fig. 13. Effect of yield strength of isolation system on seismic response of the base-isolated slender tank (T
b
=2 s and x
b
=0.1).
both the bearing and sloshing displacements de-
crease with the increase of bearing yield
strength.
8. Conclusions
The seismic response of base-isolated liquid
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 19
storage tank is investigated under two horizontal
components of real earthquake ground motions.
The bi-directional interaction between the restor-
ing forces of the lead-rubber bearings is duly
considered. The response of base-isolated liquid
storage tank is analyzed to study the inuence of
Fig. 14. Effect of yield strength of isolation system on seismic response of the base-isolated broad tank (T
b
=2 s and x
b
=0.1).
M.K. Shrimali, R.S. Jangid / Nuclear Engineering and Design 217 (2002) 120 20
bi-directional interaction and system parameters
on the effectiveness of seismic isolation. The fol-
lowing conclusions are drawn from the trends of
the results of the limited parametric study:
1. Base isolation system is quite effective in re-
ducing the seismic response of the liquid stor-
age tanks.
2. The base isolation is found to be more effec-
tive for the slender tanks in comparison to the
broad tanks.
3. The peak response of base-isolated liquid stor-
age tank is not much inuenced by the bi-di-
rectional interaction of restoring forces of the
lead-rubber bearings.
4. The effectiveness of seismic isolation of the
liquid storage increases with the increase of the
exibility of isolation systems. However, the
bearing displacement also increases with the
increase of isolator exibility.
5. An optimum value of the isolation damping
exists for which the base shear in the tank
attains the minimum value. Thus, the increase
of bearing damping beyond the optimum
value will decrease the bearing and sloshing
displacement but it will increase the base
shear.
6. The bearing and sloshing displacements of the
isolated liquid storage tank decreases with the
increase of the yield strength of the bearing.
However, the base shear may get increased for
higher bearing yield strength.
References
Buckle, I.G., Mayes, R.L., 1990. Seismic isolation: history,
application and performancea world overview. Earth-
quake Spectra 6 (2), 161202.
Chalhoub, M.S., Kelly, J.M., 1988. Theoretical and experi-
mental studies of cylindrical water tanks in base isolated
structures. A report No. UCB/EERC-88/07, USA,
Berkeley.
Epstein, H.I., 1976. Seismic design of liquid storage tanks.
Journal of Structural Division, ASCE 102, 16591673.
Haroun, M.A., 1983. Vibration studies and test of liquid
storage tanks. Earthquake Engineering and Structural Dy-
namics 11, 179206.
Housner, G.W., 1957. Dynamic pressures on accelerated uid
containers. Bulletin of Seismological Society of America 1,
1537.
Housner, G.W., 1963. Dynamic behaviour of water tanks.
Bulletin of Seismological Society of America 53, 381387.
Jangid, R.S., Datta, T.K., 1995. Seismic behaviour of base
isolated buildings: A state of the art review. Journal of
Structures and Buildings, ICE 110 (2), 186203.
Jangid, R.S., Datta, T.K., 1994. Non-linear response of tor-
sionally coupled base isolated structures. Journal of Struc-
tural Engineering, ASCE 120, 122.
Kelly, J.M., 1986. A seismic base isolation: a review and
bibliography. Soil Dynamics and Earthquake Engineering
5, 202216.
Kim, N.S., Lee, D.G., 1995. Pseudo-dynamic test for evalua-
tion of seismic performance of base-isolated liquid storage
tanks. Engineering Structures 17 (3), 198208.
Malhotra, P.K., 1997a. New methods for seismic isolation of
liquid-storage tanks. Earthquake Engineering and Struc-
tural Dynamics 26, 839847.
Malhotra, P.K., 1997b. Method for seismic base isolation of
liquid storage tanks. Journal of Structural Engineering 123
(1), 113116.
Manos, G.C., 1991. Evaluation of the earthquake performance
of anchored wine tanks during the San Jan Argentina,
1977 Earthquake. Earthquake Engineering and Structural
Dynamics 20, 10991114.
Nagarajaiah, S., Reinhorn, A.M., Constantinou, M.C., 1991.
Non-linear dynamic analysis of 3-D base isolated structure.
Journal of Structure Engineering, ASCE 117, 20352054.
Niwa, A., Clough, R.W., 1982. Buckling of cylindrical liquid-
storage tanks under earthquake loading. Earthquake Engi-
neering and Structural Dynamics 10, 107122.
Park, Y.J., Wen, Y.K., Ang, A.H.S., 1986. Random vibration
of hysteretic system under bi-directional ground motions.
Earthquake Engineering and Structural Dynamics 14,
543557.
Robinson, W.H., 1982. Lead rubber hysteretic bearing suitable
for protecting structures during earthquakes. Earthquake
Engineering and Structural Dynamics 10, 593604.
Rosenblueth, E., Newmark, N.M., 1971. Fundamentals of
Earthquake Engineering. Prentice-Hall, NJ.
Steinbrugge, K.V., Rodrigo, F.A., 1963. The Chilean earth-
quake of May 1960: A structural engineering viewpoint.
Bulletin of Seismological Society of America 53, 225307.