Sunteți pe pagina 1din 25

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2004; 33:157181 (DOI: 10.1002/eqe.340)

Wavelet-based non-stationary seismic rocking response of


exibly supported tanks

Pranesh Chatterjee1 and Biswajit Basu2; ;


1 Department of Civil Engineering; Structural Mechanics Section; K.U.Leuven; 3001 Heverlee; Belgium
2 Department of Civil; Structural and Environmental Engineering; Trinity College; Dublin 2; Ireland

SUMMARY
The non-stationary rocking response of liquid storage tanks under seismic base excitations including
soil interaction has been developed based on the wavelet domain random vibration theory. The ground
motion has been characterized through statistical functionals of wavelet coecients of the ground accel-
eration history. The tankliquidfoundation system is modelled as a multi-degree-of-freedom (MDOF)
system with both lateral and rocking motions of vibration of the foundation. The impulsive and con-
vective modes of vibration of the liquid in the tank have been considered. The wavelet domain coupled
dynamic equations are formulated and then solved to get the expressions of instantaneous power spec-
tral density function (PSDF) in terms of functionals of input wavelet coecients. The moments of the
instantaneous PSDF are used to obtain the stochastic responses of the tank in the form of coecients
of hydrodynamic pressure, base shear and overturning base moment for the largest expected peak re-
sponses. Parametric variations are carried out to study the eects of various governing parameters like
height of liquid in the tank, heightradius ratio of the tank, ratio of total liquid mass to mass of foun-
dation, and shear wave velocity in the soil medium, on the responses of the tank. Copyright ? 2003
John Wiley & Sons, Ltd.

KEY WORDS: non-stationary; seismic excitation; wavelet; tank; rocking; soil

1. INTRODUCTION

Storage tanks in the oil and petro-chemical industries, hazardous chemical industries, nuclear
industries and water treatment and supply plants may be subjected to fatal risks and damage,
including disruption of lifelines, during seismic ground excitations. For this reason, a rigorous
safety analysis to evolve a reliable design procedure for these types of tanks against seismic-
induced vibrations is essential. The responses of uid storage tanks subjected to seismic ground
motions have been studied and reported by numerous authors in the last few decades.

Correspondence to: Biswajit Basu, Department of Civil, Structural and Environmental Engineering, Trinity College,
Dublin 2, Ireland.
E-mail: basub@tcd.ie

Received 10 June 2002


Revised 24 February 2003
Copyright ? 2003 John Wiley & Sons, Ltd. Accepted 2 June 2003
158 P. CHATTERJEE AND B. BASU

Many of the earlier studies considered the tanks to be anchored to rigid foundations [19].
Haroun [10] carried out a detailed theoretical treatment of the coupled liquidshell system
for tanks rigidly anchored to the foundations. The studies performed by Peek [11] and Peek
and Jennings [12] focused on the tank behaviour due to deterministic ground acceleration
history. None of these studies accounted for the soilstructure interaction eects on uid
storage tanks. The study made by Daysal and Nash [13] to see the soilstructure interaction
eects did not consider the radiational damping capacity of the supporting medium which
is known to have a considerable eect on the responses of the structures. The eects of
ground exibility and interaction between the tankliquid system and the supporting medium
were studied only for vertically excited systems by Veletsos and Tang [14], Haroun and
Abdel-Haz [15], Fischer and Seeber [16] etc. These studies showed that the soilstructure
interaction considerably aects the hydrodynamic eects on the tanks. Veletsos and Tang
[17] studied the unanchored tank responses due to lateral ground excitation considering the
rocking and lateral components of the base motion of the tank. The proposed analyses put
forward by Veletsos and Tang considered the decoupling eects and obtained the ultimate
responses of the tank by superposition of impulsive and convective components of lateral and
rocking motions. Malhotra [18] made a study on the exibly supported unanchored liquid-
storage tanks considering rocking motion of the tank wall and the base plate and showed that
increased thickness of the base plate below the tank reduces the values of base rotation and
base uplifting. Wunderlich et al. [19] studied the seismic response and failure mechanism
of liquid storage tanks. Tanaka et al. [20] investigated the non-linear response of cylindrical
liquid storage tanks. Some other studies on liquid sloshing in tanks, and the eect of liqueed
ground disruption on tanks, were carried out by Nagata et al. [21] and Yoshizawa et al.
[22], respectively. Most of these studies did not consider the random nature of seismic base
excitations and the induced random vibrations in the tankliquid system. Furthermore, the
ground motions consist of both amplitude and frequency non-stationarities [23]. Studies have
shown (see Basu and Gupta [2325] and Yeh and Wen [26] that the response quantities of
structures are considerably aected by the non-stationarity of ground motions. Hence, unlike
earlier studies where the tank responses were found out for deterministic or harmonic loadings
or for a uniformly modulated ground excitation, the tank responses of the soiltank system
subjected to seismic excitations in the present analysis include the eects of the amplitude and
frequency non-stationarities of the ground motions. Also, in most of the earlier studies, while
investigating the rocking responses of tanks, the uncoupled equations have been formulated
and then solved. In the present analysis, besides considering the stochastic nature of ground
excitations, a coupled analysis of impulsive and convective modes of the moving liquid along
with the lateral and rocking motions of the tank is presented.
In this paper the non-stationary lateral and rocking response of liquid storage tanks resting
directly on the soil and subjected to the lateral component of seismic base accelerations has
been formulated. A base shear at the soilfoundation interface, hydrodynamic pressure on the
tank wall and also an overturning moment at the tank base are generated. In the present study,
the random vibration analysis is carried out on the basis of wavelet analytical formulation as
proposed by Basu and Gupta [23]. The foundationsoil system has been considered to be
exible with translational and rocking components of the motion of the tank base. The tank
liquidfoundation system is modelled as a multi-degree of freedom system where the lateral
and rocking resisting forces of the soilfoundation medium are considered by appropriate
lateral and rocking impedance functions. The hydrodynamic eects are assumed to consist

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 159

of an impulsive component, representing the eect of the part of the liquid which may be
considered to move in unison with the tank wall as a rigidly attached mass, and a convective
component, representing the action of the part of the liquid that undergoes sloshing motion.
The tankliquid structural system has been modelled by linear springs and viscous dampers.
The ground motion is characterized in terms of certain statistical functionals of the wavelet
coecients of the ground acceleration time history [23]. The wavelet-based analysis is adopted
to incorporate the eects of frequency and amplitude non-stationarities of the ground motion
process on tank responses due to its unique time-frequency localization property on both
large and small scales. The functionals of the wavelet coecients are obtained from a single
realization of a ground motion process by applying statistical techniques [27]. In the present
study, the analysis is made on the basis of coupled equations including the combined eects
of translational and rocking motions of the foundation along with impulsive and convective
modes of liquid motion to get the exact solution. A parametric study has been carried out to
observe the eects of liquid height in the tank, the ratio of tank height to tank radius, shear
wave velocity in the foundation soil and the ratio of total mass of the liquid to the mass of
the foundation.

2. WAVELET-BASED THEORETICAL FORMULATION

Tankuidfoundation system
The tankuid system consists of a right circular cylindrical, rigid thin-walled tank of radius,
R, lled with an incompressible liquid of mass density, l , to a height, H . The tank wall is
assumed to be of uniform thickness, h. The tank is rigidly clamped to a rigid circular base
mat of radius, R, the same as the radius of the tank. The tank with the mat is placed on
the surface of a homogeneous elastic halfspace. The Youngs modulus of elasticity and the
mass density of the tank material are denoted by E and , respectively. The liquid in the
tank is assumed to be vibrating both in impulsive and convective modes. The total mass of
the liquid, ml , is considered to be the sum of the impulsive and the convective masses. The
damping ratio and the natural frequency of vibration of the liquid in the impulsive mode
may be assumed to be the same as that of the tank and denoted by i and !i , respectively.
Only the rst sloshing mode is considered to be contributing to the response for the vibration
analysis. For the rst sloshing mode of vibration of the convective component of the liquid,
its damping ratio and natural frequency of vibration are denoted by c and !c , respectively.
The shear modulus of elasticity, the mass density, the Poissons ratio and the velocity of shear
wave propagation in the supporting medium are denoted by Gs , s , s and Vs , respectively.
The tankliquid system is represented by a massspringdashpot model with impulsive mass
of the liquid, mi , and convective mass of the liquid, mc , in the rst sloshing mode. The two
linear springs with stinesses, Ki and Kc and the two linear viscous dampers with damp-
ing coecients Cdi and Cdc (in the sloshing mode) are attached to the masses mi and mc ,
respectively. The foundation system of the tank consists of mass, mf , with complex valued
translational and rocking impedance functions, Kx and K , respectively, due to the surround-
ing soil. The impulsive mass, mi and the convective mass, mc are concentrated at heights
hi and hc to yield the correct moments at a section immediately beneath the tank base. The
heights hi and hc give the locations at which these two masses should be placed to give the

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
160 P. CHATTERJEE AND B. BASU

xc(t)
xi(t)

mc
mi Kc
Ki

hi hc
Cdc
Cdi
Kx x(t)
R (t)

K
mf

..
xg (t)

Figure 1. MDOF model of the tankliquidfoundation system.

correct moments at a section immediately above the tank base. The schematic diagram of the
dynamical massspringdashpot model representing the tankliquidfoundationsoil model is
shown in Figure 1.

Equations of motion
Let the tankuidfoundation system, as described earlier, be subjected to a horizontal free-
eld seismic absolute ground displacement, xg (t). The absolute lateral and rocking displace-
ments of the foundation induced by the lateral free-eld ground displacement are denoted by
x(t) and R (t), respectively. The displacement of the impulsive mass of the liquid in the
tank relative to the foundation is denoted by xi (t) and the displacement of the convective
component of the liquid in the tank relative to the foundation is represented by xc (t).
The equation of motion of the tankuidfoundation system can be written as

mi (xi (t) + x(t))


 + mc (xc (t) + x(t))
 +  R (t)hi mi +  R (t)hc mc + mf x(t)
 + Qs = 0 (1)

where, the overdots represent the dierentiation with respect to time and Qs is the base shear
at the soilfoundation interface. At the no-slip condition between the soil and the foundation
this base shear is proportional to the relative deformation between the tankfoundation system
and the free-eld ground surface. The base shear may be expressed as

Qs = Kx (x(t) xg (t)) (2)

In Equation (2), Kx represents the complex valued translational impedance function for a
circular footing given by Veletsos and Tang [17] as

8Gs R
Kx = ( x + ia0 x ) (3)
2 s

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 161

where Gs = s Vs2 . The term a0 , in Equation (3) is a dimensionless frequency parameter dened
by
!R
a0 = (4)
Vs
where ! is the frequency of vibration of the foundation and  x and x are dimensionless
functions of a0 and s . Closed-form solutions of these parameters are given by Veletsos and
Verbic [28]. In this paper, the value of s has been chosen to be 13 for which  x = 1:0 and
x = 0:65. It may be noted here that when ! = 0 the complex valued impedance function
becomes real and it represents the static stiness function. In Equation (1), the expressions
for mc and hc are given by [17]:
  
2 R H
mc = 2 tanh 1 ml (5)
 1 1 H R
and

hc = hc + hc (6)

where, 1 is the rst root of the rst derivative of the Bessel function of the rst kind and
the rst order with
  
1 R 1 H
hc = 1 tanh H (7)
1 H 2 R
and
H
hc = (8)
1 HR sinh(1 HR )

On using the relations, Ki = mi !i2 and Cdi = 2mi i !i , the equation of motion for the impulsive
mass of the liquid can be written as

mi (xi (t) + x(t))


 +  R (t)hi mi + !i2 mi (xi (t) + hi R (t)) + 2i mi !i (xi (t) + R (t)hi ) = 0 (9)

with

Ci E
!i = (10)
H 

In Equation (10), Ci is a dimensionless coecient which depends on the ratio of liquid-depth


in the tank to the tank radius (i.e. HR ), the ratio of tank wall thickness to the tank radius
(i.e. Rh ), the Poissons ratio for the tank material (i.e. ), and the relative mass densities of
the liquid and the tank wall (i.e. l ). On using the relations, Kc = mc !c2 and Cdc = 2mc c !c
and on considering the convective mass of the liquid, the equation of motion may be written
as

mc (xc (t) + x(t))


 +  R (t)hc mc + !c2 mc (xc (t) + hc R (t)) + 2c mc !c (xc (t) + R (t)hc ) = 0 (11)

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
162 P. CHATTERJEE AND B. BASU

where
  
H g
!c = 2 1 tanh 1 (12)
R R
and g is the acceleration due to gravity. Another equation may be formed on considering the
moment-equilibrium of the tankliquidfoundation system as

1
2 mf R2  R (t) + K R (t) + mi hi (xi (t) + x(t)
 + hi  R (t)) + mc hc (xc (t)

 + hc  R (t)) = 0
+x(t) (13)

where, the linear rocking impedance function, K , of the foundation may be obtained from
Wolf [29] as
8Gs R3
K = ( + ia0  ) (14)
3(1 s )
with
0:3178a20
 =1 (15)
1 + 0:675a20
and
0:2612a20
 = (16)
1 + 0:675a20
corresponding to a value of the Poissons ratio of the soil, s = 13 . On dividing Equations (3)
and (14) by mf and 12 mf R2 , respectively, we get
Kx
= !lf2 ( x + ia0 x ) (17)
mf
and
K
= !rf2 ( + ia0  ) (18)
2 mf R
1 2

where, !lf and !rf are, respectively, the lateral and rocking natural frequencies of the foun-
dation with the following expressions

8Gs R
!lf = (19)
(2 s )mf
and

16Gs R
!rf = (20)
3(1 s )mf
Equations (1), (9), (11) and (13) form a set of coupled dierential equations of motion which
can be solved to obtain various types of responses of the tankliquidfoundationsoil system

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 163

based on wavelet analytical techniques, for a seismic ground acceleration, xg (t) with a given
non-stationary statistical characteristic. In the following subsection the hydrodynamic pressure
at any arbitrary point on the tank wall, the horizontal base shear and the overturning base
moment at the tank base are found by extending the wavelet based stochastic formulation of
Basu and Gupta [23] on dynamical systems.

Non-stationary response
The wavelet transform of the seismic base acceleration process, xg (t), assumed to have non-
stationary zero mean Gaussian characteristics, with respect to a wavelet basis function, ( tb
a ),
at any time instant, bi , and for a particular band, j, may be written as [30]:
  
1 t bi
W xg (aj ; bi ) = x
 g (t) dt (21)
1 aj
|aj | 2

with bi = bi and aj = j . From Equation (21), xg (t) may be reconstructed in terms of wavelet
coecients as
  K  b
xg (t) = W xg (aj ; bi ) aj ; bi (t) (22)
i j aj

where
 
2 1
K  =  (23)
4C 
and

| (!)|2
C = d! (24)
|!|

with

(!) = 1 (t)ei!t dt (25)
2

In the present analysis, the LittlewoodPaley (LP) basis function has been used for rep-
resenting the seismic base acceleration process (see Basu and Gupta [23]). The LP basis
function may be characterized by its Fourier transform as
1
(!)=  ; 6|!|6
2( 1) (26)
(!)=0; otherwise

having the time domain representation


1 sin t sin t
(t) =  (27)
 ( 1) t

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
164 P. CHATTERJEE AND B. BASU

Although the wavelet representation for the base acceleration function, xg (t), has been dis-
cussed above, these expressions hold good for the ground displacement and the structural
response functions as well. It can be observed that seismic base excitations appear in
Equation (1) in the form of ground displacement, xg (t). However, the input seismic
excitation is usually described in terms of base acceleration, xg (t). Thus, it is necessary
to relate the wavelet coecient of ground displacement with that of ground acceleration
as follows. The ground displacement in terms of wavelet coecients may be
written as
  K  b
xg (t) = W xg (aj ; bi ) aj ; bi (t) (28)
i j aj

On dierentiating Equation (28) twice, and then on Fourier transforming both sides, we obtain
  K  b
xg (!) = W xg (aj ; bi ) aj ; bi (!)(!2 ) (29)
i j aj

Also, on Fourier transforming Equation (22) we get


  K  b
xg (!) = W xg (aj ; bi ) aj ; bi (!) (30)
i j aj

On comparing Equations (29) and (30) we get


 
  K  b   K  b 1
W xg (aj ; bi ) aj ; bi (!) =
W xg (aj ; bi ) aj ; bi (!) 2
(31)
i j aj i j aj !

thus relating the wavelet coecients of ground displacement in terms of that of ground
acceleration. Now, the absolute lateral and rocking displacements of the foundation and
the relative displacements of the impulsive and convective mass of the liquid with respect
to the foundation may also be expanded in terms of wavelet co-
ecients as

  K  b
x(t) = W x(aj ; bi ) aj ; bi (t) (32)
i j aj

  K  b
R (t) = W R (aj ; bi ) aj ; bi (t) (33)
i j aj

  K  b
xi (t) = W xi (aj ; bi ) aj ; bi (t) (34)
i j aj

and
  K  b
xc (t) = W xc (aj ; bi ) aj ; bi (t) (35)
i j aj

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 165

On substituting the expressions for xg (t), x(t), R (t), xi (t) and xc (t) in Equation (1) we obtain

mi   K  b
W xi (aj ; bi )  aj ; bi (t)
mf i j aj
 
mi mc   K  b
+ + +1 W x(aj ; bi )  aj ; bi (t)
mf mf i j aj
mc   K  b
+ W xc (aj ; bi )  aj ; bi (t)
mf i j aj
 
mi  mc    K  b
+ h + h W R (aj ; bi )  aj ; bi (t)
mf i mf c i j aj
Kx   K  b
+ W x(aj ; bi ) aj ; bi (t)
mf i j aj
Kx   K  b
W xg (aj ; bi ) aj ; bi (t) = 0 (36)
mf i j aj

Further, on taking the Fourier transform of both sides of the above equation and subse-
quently multiplying both sides of it by ak ; bl (!) and on using the relation (see, Basu and
Gupta [23]):

aj ; bi (!) ak ; bl (!) = jk aj ; bi (!) ak ; bl (!)
(37)
we get
         
mi mi mc Kx mc mi  mc  Kx
P+ + +1 Q+ L+ hi + hc S = T
mf mf mf mf ! 2 mf mf mf mf ! 4
(38)

where

P= W xi (aj ; bi ) aj ; bi (!) (39)
i

Q= W x(aj ; bi ) aj ; bi (!) (40)
i

L= W xc (aj ; bi ) aj ; bi (!) (41)
i

S= W R (aj ; bi ) aj ; bi (!)
(42)
i

and

T= W xg (aj ; bi ) aj ; bi (!) (43)
i

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
166 P. CHATTERJEE AND B. BASU

On performing similar operations on Equations (9), (11) and (13) we obtain


(!i2 !2 + 2ii !i !)P + (!2 )Q + hi (!i2 !2 + 2ii !i !)S = 0 (44)
(!2 )Q + (!c2 !2 + 2ic !c !)L + hc (!c2 !2 + 2ic !c !)S = 0 (45)

and
       
mi  mi  mc  mc  1 2 K mi  2 mc  2
h P + hi h Q + hc L+ R + h h S =0
mf i mf mf c mf 2 mf ! 2 mf i mf c
(46)

respectively.

Hydrodynamic pressure, base shear and base moment


On considering only the rst sloshing mode of vibration, the total hydrodynamic pressure,
PH (t), at any point on the tank wall at any time instant is obtained by the summation of the
hydrodynamic pressures due to the impulsive component and the convective component of
the rst sloshing mode of the tankliquid and may be expressed in the following form
 + hi  R (t)} + cc (R; z){xc (t) + x(t)
PH (t) = l R cos
[ci (z){xi (t) + x(t)  + hc  R (t)}] (47)
where, ci (z) is a dimensionless function dening the heightwise variation of hydrodynamic
pressure and cc (R; z) is also a dimensionless function dening distribution of the component
of hydrodynamic pressure corresponding to the rst convective mode. These two distribution
functions, ci (z) and cc (R; z) neglect the rotational component of motion. However, the rocking
eect of the base motion is included in the above equation. In our case, as the investigation is
carried out only in the direction of the lateral seismic excitation,
in Equation (47) is assumed
to be zero degree. The expression for cc (R; z) in Equation (47) to nd out the hydrodynamic
pressure on the tank wall (i.e. r = R) becomes [17]:
2 cosh(1 Rz )
cc (R; z) = (48)
12 1 cosh(1 HR )
The eective shear, QB (t), at the tank base and the net eective overturning moment, MB (t),
at a section immediately above the tank base may be expressed as
 + hi  R (t)} + mc {xc (t) + x(t)
QB (t) = mi {xi (t) + x(t)  + hc  R (t)} (49)
and
 + hi  R (t)} + mc hc {xc (t) + x(t)
MB (t) = mi hi {xi (t) + x(t)  + hc  R (t)} (50)
PH (t), QB (t) and MB (t) may be represented in terms of the wavelet coecients respecti-
vely as
  K  b
PH (t) = W PH (aj ; bi ) aj ; bi (t) (51)
i j aj
  K  b
QB (t) = W QB (aj ; bi ) aj ; bi (t) (52)
i j aj

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 167

and
  K  b
MB (t) = W MB (aj ; bi ) aj ; bi (t) (53)
i j aj
On substituting Equations (33), (34), (35) and (51) into Equation (47) and then on Fourier
transforming and using the relation given by Equation (40), one may obtain the expression
of hydrodynamic pressure in terms of wavelet functionals as
 
W PH (aj ; bi ) aj ; bi (!) = PH (!) W xg (aj ; bi ) aj ; bi (!) (54)
i i

where, PH (!) is the frequency-dependent transfer function relating the total hydrodynamic
pressure to the ground acceleration and may be expressed as
PH (!) = l R!2 [ci (z){ i (!) + (!) + hi R (!)} + cc (R; z){ c (!) + (!) + hc R (!)}] (55)
In Equation (55), i (!) and c (!) respectively represent the transfer functions relating the
displacements of the impulsive mode and the rst convective mode of the liquid to the
ground acceleration, and R (!) represents the transfer function relating the rocking motion of
the foundation to the ground acceleration. The corresponding relationships are:
P = i (!)T (56)

L = c (!)T (57)

Q = (!)T (58)

and
S = R (!)T (59)
On performing similar operations on Equations (49) and (50) we get
 
W QB (aj ; bi ) aj ; bi (!) = Q (!) W xg (aj ; bi ) aj ; bi (!) (60)
i i

and
 
W MB (aj ; bi ) aj ; bi (!) = M (!) W xg (aj ; bi ) aj ; bi (!) (61)
i i

where, Q (!) and M (!), the frequency-dependent transfer functions relating respectively the
tank-base shear and the overturning tank-base moment to the ground acceleration, may be
expressed as
Q (!) = !2 [mi { i (!) + (!) + hi R (!)} + mc { c (!) + (!) + hc R (!)}] (62)
and
M (!) = !2 [hi mi { i (!) + (!) + hi R (!)} + hc mc { c (!) + (!) + hc R (!)}] (63)
Thus, it can be observed from Equations (55), (62) and (63) that in order to obtain the
transfer functions, PH (!), Q (!) and M (!), the other transfer functions, i (!), c (!) and

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
168 P. CHATTERJEE AND B. BASU

R (!) must be known. i (!), c (!) and R (!) may be solved from Equations (38), (44),
(45) and (46). These equations form a set of linear simultaneous algebraic equations with
complex coecients and may be expressed in matrix form as
K
x
P
m ! 4
Q f
0
[A] = T

(64)
R 0

S 0
where, [A] is a complex coecient matrix. On inverting [A], a matrix [B] with real and
imaginary components is obtained as
[B] = [Br ] + i[Bi ] (65)
where

br11 br12 br13 br14
r
b21 br22 br23 br24

[Br ] = r (66)
b31 br32 br33 br34

br41 br42 br43 b44
r

and

bi11 bi12 bi13 bi14
i
b21 bi22 bi23 bi24

[Bi ] = : (67)
bi bi32 bi33 bi34
31
bi41 bi42 bi43 b44
i

On pre-multiplying both sides of Equation (64) by [B], the expressions for P, L and S can
be obtained which leads to the expressions for the transfer functions, i (!), c (!) and R (!)
respectively as
i (!) = e(br11 + ibi11) (68)
c (!) = e(br31 + ibi31) (69)
(!) = e(br21 + ibi21) (70)

and
R (!) = e(br41 + ibi41) (71)
where, e in the above equations is given by
Kx
e= (72)
! 4 mf

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 169

Once i (!), c (!) and R (!) are known, the transfer functions, PH (!), for hydrodynamic
pressure, Q (!), for base shear, and M (!), for overturning moment at the tank base as
dened by Equations (55), (62) and (63), respectively, can be found out. Each of these
transfer functions, viz. PH (!), Q (!) and M (!) consist of a real part and an imaginary part.
On substituting the expressions of i (!), c (!) and R (!) in the expression for the transfer
function of the hydrodynamic pressure given by Equation (55) and subsequently dividing it by
Hl g (which represents the hydrostatic pressure at the tank base), we arrive at the following
equation
R  !lf 2
PH (!) = [(A 0:65a0 B) + i(B + 0:65a0 A)] (73)
Hg !
where
A = {ci (z)(br11 + br21 + hi br41 ) + cc (R; z)(br31 + br21 + hc br41 )} (74)
and
B = {ci (z)(bi11 + bi21 + hi bi41 ) + cc (R; z)(bi31 + bi21 + hc bi41 )} (75)
On taking expectation of the square of the amplitude of both sides of Equation (54), integrating
over ! and using the orthogonality relationship for the LP wavelet basis function (Basu and
Gupta [23]), we get for the j-th band

 
E[|W PH (aj ; bi )| ] = E[|W xg (aj ; bi )| ]
2 2
| PH (!)|2 | aj ; bi (!)|2 d! (76)
i i

Further, on using the time-localization property of the wavelet coecients, the instantaneous
energy for the mean square response of the hydrodynamic pressure coecient, CHP , at t = bi ,
corresponding to the band with dilation factor, aj is approximately obtained as
 
j K  b
E[|(CHP )i (!)| ]d! =
2
E[|W xg (aj ; bi )| ]
2
| PH (!)|2 | aj ; bi (!)|2 d!
aj
(77)
The instantaneous power spectral density function (PSDF) for the hydrodynamic pressure
coecient can be written as (see Basu and Gupta [23] for details):
 E[|(CHP )ij (!)|2 ]  K 
SCHP (!) = = E[|W xg (aj ; bi )|2 ]| PH (!)|2 | aj ; bi (!)|2 (78)
j b j aj
The expressions for the zeroth, rst and second moments of the PSDF in this case are given
as

m0 |t=bi = K  E[W xg (aj ; bi )2 ]I0; j (!n ; ) (79)
j

m1 |t=bi = K  E[W xg (aj ; bi )2 ]I1; j (!n ; ) (80)
j

and

m2 |t=bi = K  E[W xg (aj ; bi )2 ]I2; j (!n ; ) (81)
j

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
170 P. CHATTERJEE AND B. BASU

with
  2 
=aj
R !lf 4
I0; j = [(A 0:65a0 B)2 + (B + 0:65a0 A)2 ]d! (82)
=aj Hg !
  2 
=aj
R !lf 4
I1; j = [(A 0:65a0 B)2 + (B + 0:65a0 A)2 ]! d! (83)
=aj Hg !
  2 
=aj
R !lf 4
I2; j = [(A 0:65a0 B)2 + (B + 0:65a0 A)2 ]!2 d! (84)
=aj Hg !

and
K 
K = (85)
( 1)
For calculating the non-stationary peak factors for the largest peak response, the expression
for the probability PT (x) (that the process |xi (t)| remains below the level x within the time
interval [0; T ]) given by [31]:
  T 
PT (x) = exp (t) dt (86)
0

is considered. On discretizing Equation (86) and on using the parameter, (t)|t=bi (which is
the time-dependent rate of the Poisson process), given by the expected value of the largest
peak can be obtained from where (t)|t=bi is the time dependent rate of the Poisson process
given by
  
x2 1 exp
i 2 =2 [i (t)]1:2 xi
(t)|t=bi =
2
e i  2  (87)
 1 exp x 2 2i

the expected value of the largest peak can be obtained from


 1
E(x(1) ) = PT1 (u) du (88)
0

In Equation (87), the instantaneous rate of the zero crossings (i ) and the band-width
parameter, i , are given as

m2 |t=bi
i = (89)
m0 |t=bi
and

m21 |t=bi
i = 1 (90)
m0 |t=bi m2 |t=bi
The peak values of the base shear coecient, CBS , and the overturning base moment co-
ecient, CMB , can be found out by dividing the transfer functions, Q (!) and M (!), by
ml g and ml gH , respectively, and then proceeding similarly as done earlier in the case of the
hydrodynamic pressure coecient.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 171

3. NUMERICAL ILLUSTRATION

Characterization of ground motion


A ground motion process has been chosen for the purpose of validation of the formulation
and numerical studies. The ground motion is characterized by local averaging of wavelet
coecients calculated from a single accelerogram generated by using the SYNACC program
[3236]. This accelerogram simulates the motion recorded at the Dumbarton bridge site near
Coyote Hills during the 1989 Loma Prieta earthquake. Figure 2 shows the ground motion
corresponding to this accelerogram. The parameters considered for generating the synthetic
accelerogram for the Loma Prieta earthquake are the same as those considered by Gupta and
Trifunac [37]. The discretization parameter, b, has been taken as 0.02. The value for 
1
in Equation (23) is 2 4 (see, Basu and Gupta [23]), which can reasonably represent most
of the ground motions. In local averaging, for a particular frequency band, the total time
is divided into certain stretches, each of which corresponds to the central time period of
4
  sec for that specic frequency band. The wavelet coecients are found out at every
a j + aj
time instant (i varying from 1 to 2047) and for all frequency bands (j varying from 17 to
+4). The wavelet coecients are subsequently squared and then averaged over each stretch to
represent the value of E[W 2 xg (aj ; bi )] uniformly over that stretch. The details of this averaging
procedure may be found in Reference [27].

800

600

400
Acceleration (mm/s2)

200

-200

-400

-600

-800
0 5 10 15 20 25 30 35 40 45
t (sec)

Figure 2. Acceleration time history synthetically generated by SYNACC for the 1989 Loma Prieta
earthquake motions at the Dumbarton bridge site.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
172 P. CHATTERJEE AND B. BASU

0.4
Simulation
Wavelet

0.3
PSA (in g)

0.2

0.1

0
0.04 0.1 1 10
Tn (sec)

Figure 3. Comparison of PSA responses of the SDOF system.

Validation of wavelet-based stochastic analysis


The proposed stochastic analysis on the basis of the wavelet technique has been validated
with exact time history simulation. The time history simulation is based on the average of
the results obtained from seventeen accelerogram records generated by using the SYNACC
program, which represents the Loma Prieta motions at the Dumbarton bridge site. The peak
pseudo-spectral acceleration (PSA) response of a single degree of freedom system has been
obtained stochastically using the accelerogram shown in Figure 2 and the same has been
obtained by time history simulation of the seventeen accelerograms. The structural damp-
ing assumed was 5%. The matching in PSA responses obtained from these two methods is
shown in Figure 3. The matching seems to be quite reasonable. For the purpose of vali-
dation, we have assumed the lateral and rocking natural frequencies (i.e. !lf and !rf ) of
the foundation to be very high so that the 4-DOF model reduces to a xed-base 2-DOF
model and the number of dynamic equations in our case also reduces to two, which are as
follows:

xi (t) + !i2 xi (t) + 2i !i xi (t) = xg (t) (91)

and

xc (t) + !c2 xc (t) + 2c !c xc (t) = xg (t) (92)

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 173

The expressions for hydrodynamic pressure coecients, PH (t), base shear coecients, QBS

(t),
and overturning base moment coecients, MB (t), in this case become

PH (t) = l R[ci (z)xi (t) + cc (z)xc (t)] (93)



QBS (t) = mi xi (t) + mc xc (t) (94)

and
MB (t) = mi hi xi (t) + mc hc xc (t) (95)
Following a similar approach as described in Section 2, we can obtain the expressions of
frequency-dependent transfer functions for hydrodynamic pressure, CPH , base shear, CBS and
overturning base moment, CMB , as

!2
CPH = H
[ci (z)Hi (!) + cc (z)Hc (!)] (96)
R g

!2
CBS = [ i Hi (!) + c Hc (!)] (97)
g

and
 
!2 hi hc
CMB = i Hi (!) + c Hc (!) (98)
g H H
These expressions may then be used to obtain the largest expected peak responses.
The values of some of the parameters assumed constant for the purpose of validation as
well as for further numerical studies in the following subsection are as follows. The steel
tank is assumed to have a damping ratio, n , of 5%. The impulsive mass of the liquid (water
being considered for the present study) is assumed to move in unison with the tank during
vibration. So, the damping ratio, i of the liquid mass vibrating in impulsive mode is also
assumed to be 5%, i.e. i = n . Only the rst sloshing mode of vibration is considered, so
the value of j , i.e. 1 , is 1.841. The values of Youngs modulus of elasticity and density of
the tank material are 2:1 106 kg=cm2 and 7850 kg=m3 , respectively. The ratio of the mass
density of water to the mass density of the steel tank is kept constant at 0.127. The value
of Poissons ratio, s , and mass density, s , of soil are considered to be 13 and 1500 kg=m3 ,
respectively. The distributions of the coecients for impulsive components of hydrodynamic
pressure, ci (z), for steel tanks with Rh are shown in Figure 2 of Reference [17]. The damping
in the rst mode of sloshing is taken as 1%. The ratio of the tank wall thickness to the tank
radius, h=R, is taken as 0.001. For xed ratios of HR and Rh , the values of the corresponding
mi h
ml , Hhi , Hi and Ci have been taken from Table II of Reference [17].
Figure 4 shows the comparison between the expected largest peak pressure coecients
obtained from time history simulation and those obtained from wavelet-based stochastic for-
mulation for xed-base broad tanks (H=R = 1:0) with liquid height, H = 10 m. Figure 5 shows
the same comparison for comparatively slender tanks (H=R = 3:0) with the same liquid height.
In both cases the matching seems to be reasonably good which shows that the wavelet-based

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
174 P. CHATTERJEE AND B. BASU

1
Simulation
Wavelet

0.8

0.6
z/H

0.4

0.2

0
0 0.02 0.04 0.06 0.08 0.1
CHP

Figure 4. Comparison between variations in CHP at dierent depths along the tank wall obtained from
time history simulation and wavelet analysis at H=R = 1:0.

stochastic analysis with wavelet coecients being generated by the local averaging method
from a single accelerogram is a reliable alternative approach to time history simulation using
an ensemble of accelerograms.

Results and discussion


A parametric study on the stochastic responses of the tankuidfoundation system undergoing
rocking is carried out to study the variation in hydrodynamic pressure on the tank wall, base
shear and overturning moment at the tank base by varying the liquid height in the tank, H ,
the shear wave velocity in soil medium, Vs , the mass ratio i.e. the ratio of the total mass
of the liquid to the mass of the foundation, (= mmfl ), and the ratio of tank height to its
radius, HR .
Figures 69 represent the variation of hydrodynamic pressures on the tank wall. Figure 6
shows the variation for a liquid height of 5 m in a slender tank with H=R = 3:0 and mass ratio
6 for three dierent values of Vs (50 m=s, 200 m=s and 400 m=s). The pressure coecients are
almost the same for all Vs up to about 0:5H after which the response increases more sharply
for tanks on comparatively stier soils with shear wave velocity of 200 m=s and 400 m=s.
However, there is negligible dierence between the values of the pressure coecients corre-
sponding to Vs = 200 m=s and 400 m=s. Figure 7 plots the similar variations as in Figure 6
for a liquid height of 10 m in the tank. For tanks supported on relatively stier soils with
Vs = 200 m=s and 400 m=s the pressure coecients gradually increase from the base of
the tank wall to reach the peak at about 0:7H , after which the trend is reversed until the liquid

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 175

1
Simulation
Wavelet

0.8

0.6
z/H

0.4

0.2

0
0 0.02 0.04 0.06 0.08 0.1
CHP

Figure 5. Comparison between variations in CHP at dierent depths along the tank wall obtained from
time history simulation and wavelet analysis at H=R = 3:0.

0.8

0.6
Vs = 50 m/s
z/H

Vs = 200 m/s
0.4 Vs = 400 m/s

0.2

0
0 0.03 0.06 0.09 0.12 0.15
CHP

Figure 6. Variation in CHP at dierent depths along the tank wall


for dierent shear wave velocities in the case of H = 5 m.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
176 P. CHATTERJEE AND B. BASU

0.8

Vs = 50 m/s

0.6 Vs = 200 m/s


Vs = 400 m/s
z/H

0.4

0.2

0
0 0.01 0.02 0.03 0.04 0.05 0.06
CHP
Figure 7. Variation in CHP at dierent depths along the tank wall
for dierent shear wave velocities in the case of H = 10 m.

0.8

0.6
=4
z/H

=8
0.4

0.2

0
0 0.01 0.02 0.03 0.04 0.05 0.06
CHP
Figure 8. Variation in CHP at dierent depths along the tank wall
for dierent mass ratios in the case of H = 10 m.

surface is reached. Figure 8 depicts the variation of hydrodynamic pressure coecients at vari-
ous points on the tank wall for a liquid of height 10m in a tank with H=R = 3:0 and Vs = 300m=s
for two dierent mass ratios, viz. 4.0 and 8.0. Both the curves are almost identical in shape.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 177

0.8

0.6
H/R = 0.5
z/H

H/R = 1.0
0.4 H/R = 3.0

0.2

0
0 0.1 0.2 0.3 0.4
CHP

Figure 9. Variation in CHP at dierent depths along the tank wall


for dierent heightradius ratios in the case of H = 10 m.

A minute observation shows that for higher mass ratio the magnitudes of pressure coecients
are only negligibly less. It may be noted from the graph that in both the cases the peak value
is reached at about 0:7H after which the values decrease. Figure 9 demonstrates the eects
of heightradius ratios of the tank on the pressure coecients at dierent heights of the tank
wall with a 10 m height of liquid in the tank at Vs = 300 m=s and = 6:0. It is seen from the
gure that the magnitudes of pressure coecients are greater when the tanks are relatively
broader. There is no considerable change in the values of pressure coecients in the cases of
H=R = 1:0 and 3.0. For very broad tanks with H=R = 0:5, the values of the coecients do not
change much up to about 0:5H beyond which the values gradually decrease. On comparing
the curve corresponding to the stochastic result of Figure 5 with the curves corresponding
to Vs = 400 m=s in Figure 7 and H=R = 3:0 in Figure 9, it is seen that a higher shear wave
velocity is not the only criterion to resemble a xed-base tankliquid system. The lateral
and rocking frequencies of the foundation too are deciding parameters. In the cases of both
Figures 7 and 9, a shear wave velocity of 400 m=s alone is not high enough to transform the
system to the xed-base case and hence the values of pressure coecients are still slightly
greater than those in Figure 5. The same reason holds good in other cases too.
Figure 10 demonstrates the variation of base shear coecients with changes in the liquid
height, H , in the tank with H=R = 3:0 and Vs = 300 m=s for two mass ratios, 4.0 and 8.0.
The pattern of the curves corresponding to two mass ratios are exactly identical in nature,
both having peaks at about 3:5 m and 15 m. The values of shear coecients are very slightly
less in the case of heavier tanks ( = 8:0) as compared to those of relatively lighter ones
( = 4:0). Figure 11 plots the variation of shear coecients at the tank base with variations in

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
178 P. CHATTERJEE AND B. BASU

0.2

0.16

0.12
CBS

0.08
=4
=8
0.04

0
0 4 8 12 16 20
H (m)

Figure 10. Variation in CBS at the tank base with liquid heights in the tank for dierent mass ratios.

0.25

0.2

0.15
CBS

0.1

H/R = 0.5
0.05 H/R = 1.0
H/R = 3.0

0
0 200 400 600 800
Vs (m/s)

Figure 11. Variation in CBS at the tank base with shear wave velocities for dierent heightradius ratios.

shear wave velocities of soil medium for H = 10 m and mass ratio, = 6:0, for three dierent
heightradius ratios of the tank (0.5, 1.0 and 3.0). It can be seen that for very broad tanks
with H=R = 0:5, the response is maximum when compared to the values corresponding to other

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 179

0.16

CMB 0.12

0.08

=4
0.04 =8

0
0 4 8 12 16 20
H (m)

Figure 12. Variation in CMB at the tank base with liquid heights in the tank for dierent mass ratios.

heightradius ratios. The curve rises steeply up to a shear wave velocity of about 250 m=s
beyond which it gradually becomes at. When the tank has an H=R ratio of 1.0, or is still
slender, the corresponding curves increase steeply until Vs is about 100 m=s, beyond which
there is almost no change in the coecient values. Figures 12 and 13 show the variations in
base moment coecients against liquid heights, H , in the tanks and shear wave velocities,
Vs , of the soil medium, respectively, keeping all the other parameters the same as in Figures
10 and 11, respectively. The trends of the curves in the case of base shear coecients are
observed to hold good in the case of overturning base moment coecients.

4. CONCLUSIONS

A wavelet-based stochastic formulation for the non-stationary seismic rocking response of


liquid storage tanks with soilfoundation interaction has been presented in this study. The
non-stationary ground motion is modelled by evaluating statistical parameters of wavelet func-
tionals of a single realization of the ground motion process. These wavelet functionals are
used to analyse the liquid storage tank system with foundation interaction and obtain the
desired response quantities. The following conclusions can be drawn from the observations
about the eects of dierent parametric variations for the given ground acceleration process.
The soilstructure interaction changes the values of the pressure coecients at dierent heights
on the tank wall. The changes in mass ratio have negligible eects on the changes in the
values of the pressure, shear or moment coecients. It has also been observed that the mag-
nitude of the response quantities is generally greater in the case of broader tanks placed on
relatively stier soil.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
180 P. CHATTERJEE AND B. BASU

0.12

CMB 0.09

0.06

0.03 H/R = 0.5


H/R = 1.0
H/R = 3.0

0
0 200 400 600 800
Vs (m/s)

Figure 13. Variation in CMB at the tank base with shear wave
velocities for dierent heightradius ratios.

The major advantage of the proposed approach over the traditional approaches is that it ac-
counts for both the amplitude and frequency non-stationarities in the ground motions. Further,
the stochastic response of tanks with any desired level of condence can be obtained by using
this approach for a site with only few data available, if the ground motions are characterized
by the functional of wavelet coecients.

REFERENCES
1. Jacobsen LS. Impulsive hydrodynamics of a uid inside a cylindrical pier. Bulletin of the Seismological Society
of America 1949; 39:189 204.
2. Housner GW. Dynamic pressures on accelerated uid containers. Bulletin of the Seismological Society of
America 1957; 47:15 35.
3. Housner GW. The dynamic behavior of water tanks. Bulletin of the Seismological Society of America 1963;
53:381387.
4. Veletsos AS, Yang JY. Earthquake response of liquid storage tanks. Adv. Civil Eng. through Eng. Mech. Proc.
Eng. Mech. Specialty Conf. ASCE, Rayleigh, USA, 1977, 124.
5. Haroun MA, Housner GW. Seismic design of liquid storage tanks. Journal of Technical Councils (ASCE)
1981; 107:191207.
6. Haroun MA, Housner GW. Dynamic characteristics of liquid storage tanks. Journal of Engineering Mechanics
(ASCE) 1982; 108:783 800.
7. Haroun MA, Housner GW. Complications in free vibration analysis of tanks. Journal of Engineering Mechanics
(ASCE) 1982; 108:801 818.
8. Haroun MA, Tayel AT. Axisymmetrical vibration of tanksNumerical. Journal of Engineering Mechanics
(ASCE) 1985; 111:329 345.
9. Haroun MA, Tayel AT. Axisymmetrical vibration of tanksAnalytical. Journal of Engineering Mechanics
(ASCE) 1985; 111:346 358.
10. Haroun MA. Vibration studies and tests of liquid storage tanks. Earthquake Engineering and Structural
Dynamics 1983; 11:179 206.
11. Peek R. Analysis of unanchored liquid storage tanks under lateral loads. Earthquake Engineering and Structural
Dynamics 1988; 16:10871100.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181
NON-STATIONARY ROCKING RESPONSE OF STORAGE TANKS 181

12. Peek R, Jennings PC. Simplied analysis of unanchored tanks. Earthquake Engineering and Structural
Dynamics 1988; 16:1073 1085.
13. Daysal H, Nash WA. Soil structure interaction eects on the seismic behaviour of cylindrical liquid storage
tanks. Proceedings of the 8th World Conference on Earthquake Engineering, San Franscisco, California, V.
223 229, 1984.
14. Veletsos AS, Tang Y. Dynamics of vertically excited liquid storage tanks. Journal of the Structural Division
(ASCE) 1986; 112:1228 1246.
15. Haroun MA, Abdel-Haz EA. A simplied seismic analysis of rigid base liquid storage tanks under vertical
excitation with soilstructure interaction. Soil Dynamics and Earthquake Engineering 1998; 5:217225.
16. Fischer FD, Seeber R. Dynamic response of vertically excited liquid storage tanks considering liquidsoil
interaction. Earthquake Engineering and Structural Dynamics 1988; 16:329 342.
17. Veletsos AS, Tang Y. Soilstructure interaction eects for laterally excited liquid storage tanks. Earthquake
Engineering and Structural Dynamics 1990; 19:473 496.
18. Malhotra PK. Seismic response of soil-supported unanchored liquid storage tanks. Journal of Structural
Engineering (ASCE) 1997; 123(4):440 450.
19. Wunderlich W, Seiler C, Schwarz J, Habenberger J. Seismic response and failure mechanism of exibly
supported liquid storage tanks. 12th World Conference on Earthquake Engineering, New Zealand, 2000.
20. Tanaka M, Sakurai T, Ishida K, Tazuke H, Akiyama H, Kobayashi N, Chiba T. Proving test of analysis
method on nonlinear response of cylindrical storage tank under severe earthquakes. 12th World Conference on
Earthquake Engineering, New Zealand, 2000, No. 2585, 1 8.
21. Nagata S, Watanabe O, Kamiya A, Okamoto T. Study of earthquake ground motion to examine sloshing
in LNG in ground storage tanks. 12th World Conference on Earthquake Engineering, New Zealand, 2000,
No. 1305.
22. Yoshizawa M, Onimaru S, Uchida A, Hatanaka M. A study on liqueed ground disruption eects on liquid
storage tanks behavior. 12th World Conference on Earthquake Engineering, New Zealand, 2000.
23. Basu B, Gupta VK. Seismic response of SDOF systems by wavelet modeling of nonstationary processes. Journal
of Engineering Mechanics (ASCE) 1998; 124(10):11421150.
24. Basu B, Gupta VK. Non-stationary seismic response of MDOF systems by wavelet transform. Earthquake
Engineering and Structural Dynamics 1997; 26:1243 1258.
25. Basu B, Gupta VK. Wavelet-based non-stationary response analysis of a friction base-isolated structure.
Earthquake Engineering and Structural Dynamics 2000; 29:1659 1676.
26. Yeh C-H, Wen YK. Modeling of nonstationary motion and analysis of inelastic structural response. Structural
Safety 1990; 8(14):281298.
27. Chatterjee P, Basu B. Non-stationary seismic response of tanks with soil interaction by wavelets. Earthquake
Engineering and Structural Dynamics 2001; 30:1419 1437.
28. Veletsos AS, Verbic B. Vibrations of visco-elastic foundations. Earthquake Engineering and Structural Dy-
namics 1973; 2:87102.
29. Wolf JP. SoilStructure Interaction Analysis in Time Domain. Prentice-Hall: Englewood Clis, NJ, 1988.
30. Daubechies I. Ten Lectures on Wavelets, Society for Industrial and Applied Mathematics, Philadelphia,
Pennsylvania, 1992.
31. Vanmarcke EH. On the distribution of the rst-passage time for normal stationary random processes. Journal
of Applied Mechanics (ASME) 1975; 42:215 220.
32. Wong HL, Trifunac MD. Generation of articial strong motion accelerograms. Earthquake Engineering and
Structural Dynamics 1979; 7:509 527.
33. Lee VW, Trifunac MD. Torsional accelerograms. Soil Dynamics and Earthquake Engineering 1985; 4(3):
132139.
34. Lee VW, Trifunac MD. Rocking strong earthquake accelerograms. Soil Dynamics and Earthquake Engineering
1987; 6(2):75 89.
35. Lee VW, Trifunac MD. A note on ltering strong motion accelerograms to produce response spectra of specied
shape and amplitude. European Earthquake Engineering 1989; 3(2):38 45.
36. Trifunac MD. Curvograms of strong ground motions. Journal of Engineering Mechanics (ASCE) 1990;
116(6):1426 1432.
37. Gupta VK, Trifunac MD. A note on the eects of ground rocking on the response of buildings during 1989
Loma Prieta earthquake. Earthquake Engineering and Engineering Vibration 1993; 13(2):1228.

Copyright ? 2003 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2004; 33:157181

S-ar putea să vă placă și