SEISMIC RESPONSE OF SOIL-SUPPORTED UNANCHORED

LIQUID-STORAGE TANKS

By Praveen K. Malhotra,' Member, ASCE

ABSTRACT: A systematic study is made of the effects of base uplifting on the seismic response of cylindrical
liquid-storage tanks that are supported directly on flexible soil foundations. First. a detailed investigation is made
of the effects of system parameters on the uplifting resistance and energy dissipation capacity of the partially
uplifted base plate. It is shown that: (1) the hydrodynamic base pressures reduce the uplifting resistance as well
as the energy dissipation capacity of the base plate; (2) the uplifting resistance increases with increase in the
thicknesses of the base plate and the tank: wall. and the stiffness of the foundation soil; and (3) the energy
dissipation capacity increases with an increase in the base-plate thickness and the yield level of plate material,
and reduces with increase in foundation stiffness and the thickness of the tank wall. Next. an efficient method
is presented for the dynamic response analysis of flexibly supported unanchored tanks. It is shown that the
flexibility of the foundation reduces the overturning base moment, and reduces significantly the axial compressive
stresses in the tank: wall. but these reductions are accompanied by increased values of plastic rotations and (in
some cases) base uplifting and hoop compressive stresses.

INTRODUCTION was simulated by placing a 2.54 cm (I in.) thick rubber pad

Numerous studies have been conducted on the seismic re-
sponse of ground supported liquid-storage tanks. Whereas, the
initial studies by Jacobsen (1949) and Housner (1963) were
concerned with the hydrodynamics of fl uid in a rigid tank. the
later studies by Veletsos and Yang (1977), Haroun and Hous-
ner (1981), Veletsos and Tang (1990), among others, explored
the effects of fluid-structure and fluid-stfUcture-foundation in-
teraction for fully anchored, flat-bottom tanks. In reality. how-
ever, complete anchorage is not always warranted or feasible;
as a result, a large number of tanks are either unanchored or
only partially anchored at their base. During intense ground
shaking these tanks experience partial base uplifting and re-
spond in a nonlinear manner. Recent studies on the response
of unanchored tanks supported on rigid concrete mat founda-
tions have shown that base uplifting influences significantly
the dynamic response of tanks and leads to axial compressive
stresses in their walls that are substantially higher than those
in similarly excited fixed-base systems (peek 1986; Natsiavas
1987; Haroun and Badawi 1988; Lau and Clough 1989; Mal-
hotra and Veletsos 1994, 1995).
In practice, many unanchored tanks are supported directly
on flexible soil foundations . When subjected to earthquake
ground shaking. these tanks uplift on one side and penetrate
their flexible foundation on the opposite side; the resulting
response is therefore highly nonlinear. Such tanks have sus-
tained damage in the fonn of: ( I) the failure of the piping
connections to the wall, caused by large base uplifting: (2)
rupture at the plate-shell junction, caused by excessive joint
stresses; (3) buckling of the tank wall, caused by large axial
compressive stresses; and (4) failure of the soils underneath,
caused by excessive foundation penetrations (Hanson 1973;
Smoots 1973; USDOC 1973; Gates 1980; Manos and Clough
1985 ; "Northridge" 1995).
Manos and Clough (1982), Cambra ( 1982), Sakai et al.
(1988), and Akiyama and Yamaguchi (1988) conducted static-
tilt and shaking table tests on scale models of flexibly sup-
ported unanchored tanks. The flexibility of the foundation soils
under the base plate. It was shown that the foundation flexi-
bility reduces significantly the axial compressive stresses, but
increases base uplifting, foundation penetration. and hoop
compressive stresses in the tank wall.
The current study is an analytical approach to the dynamic
analysis of flexibly supported unanchored tanks. An important
step in this direction-the response analysis of partially up-
lifted base plate under static loading-was the subject of an
earlier paper (Malhotra 1995). This paper has a twofold ob-
jective: (I) to present an in-depth study of the effects of system
parameters on the uplifting resistance and energy dissipation
capacity of the base plate under dynamic loading; and (2) to
present an efficient method for the seismic response analysis
of flexibly supported unanchored tanks.
SYSTEM CONSIDERED
Shown in Fig. I(a), the system considered is a cylindrical
tank of radius R, filled to a height H with a liquid of mass
density PI. Presumed resting on a Winkler foundation of sub-
grade modulus K, the tank is excited by a unidirectional hor-
izontal ground motion x,(t), the intensity of which is sufficient
to induce rocking of its wall and partial uplifting of its flex ible
base plate, as shown in Fig. I(b). The maximum width of the
uplifted portion of the base plate is denoted by L. Points in
the tank are specified with a cylindrical coordi nate system (r,
(.J (bJ
+
-
['-"l
L - i,(t)

lAdjunct Fac. , Dept. of Civ. Engrg., California State Univ., Sacra-

mento, CA 95819-6029.
NOie. Assoc iate Editor: Nicholas P. Jones. Discussion open until Sep-
tember I , 1997. To extend the closing date one month, a written request
must be filed with the ASCE Manager of Journals. The manuscript for
this paper was submitted for review and possible publication on April 24,
1995. This paper is part of the ]ourlUll of Structural Engineering, Vol.
123. No.4. April. 1997. O ASCE, ISSN 0733-9445/97/0004-0440-04501
$4.00 + $.50 per page. Paper No. 10577. FIG. 1. System Considered

440 / JOURNAL OF STRUCTURAL ENGINEERING / APRIL 1997


~
b 1 n/2 - 1
T2
3
FIG. 2. Base Plate Model
<1>, z), the origin of which is taken at the center of the base
plate.
UPLIFTING RESISTANCE A ND DAMPING OF
BASEPLATE
Problem Stat ement
In its "at rest" condition, the base plate is subjected to a
unifonn hydrostatic pressure p = PlgH on its surface and a
unifonn line load W/2'ITR along its bou ndary, where W is the
weight of the tank wall. In an earthquake, the base plate is
subjected also [Q a hydrodynamic pressure Pd on its surface
and an overturning moment Mr transmitted to its boundary by
the wall of the tank. As a result of these forces, the base of
the tank rotates by an angle IjI and a portion of the base plate
uplifts, as shown in Fig. I(b). Of interest herein is the Mr -
tV relationship.
Model and Solution Method
The plate is represented by n unifonnly loaded, flexibly sup-
ported, semiinfinite beams that are connected at their ends to
the cylindrical wall of the tank, as shown in Fig. 2. For the
sake of clarity, only a few beams are shown in this figure. It
is assumed that the plate boundary remains in one flat plane
at all times. A detailed analysis of the accuracy of the model
may be found elsewhere (Malhotra 1995; Malhotra and Yelet-
sos 1995).
The Mr - 4t relationship is established by considering small
increments of base rotation .6.\jJ. For each increment, the ver-
tical displacement at the end of each beam is obtained first,
the uplifting force at the end of each beam compu ted next
from the beam analysis (Malhotra 1995), and the moment re-
sultant of the uplifting forces computed in the end to yield a
value of Mr. Beams in this model are loaded unifonnly by the
hydrostatic base pressure only. The effect of hydrodynamic
base pressures is considered later.
Numerical Result s
A representative plot of the Mr - \jJ relationship for three
different load cycles of size IIjI I = 0.2°, 0.35°, and 0.5" is
shown in Fig. 3(a). The moment Mr is expressed as a per-
centage of W,R. where WI = -rrR2p is the total weight of the
liquid. For the results shown, the base plate thickness h =
RI2,OOO, the nonnalized yield stress rr,lp = \,800, and the
nonnalized Young's modulus of elasticity Elp = 1.5 X 10'.
The plate is constrained at its boundary by the tank wall of
unifonn thickness h, = R/I ,OOO and weight W = 0.015W,. The
slope of the Mr - IjI plot represents the uplifting stiffness of
the base plate, whereas the area enclosed within a load cycle
represents the hysteretic energy lost by plastic yielding at the
plate boundary. It is desired to separate these two effects.
For increasing values of MT (loading), the plastic hinging at
the plate boundary starts at <I> = 0° and 180° (not necessarily
simultaneously) when the radial bending moment, per unit cir-
cumferential length of the plate, becomes equal to the yield
moment
" ' h'
M, =- 4- (I )
where uy = yield stress of the plate material. With further in-
crease in M T , the plastic hinging spreads in the circumferential
direction. The direction of the yield moment at 4> = 0° is
shown in Fig. 4(a). The loading path after plastic hinging is
along the upper S-shaped curve in Fig. 3(a). For decreasing
values of MT (unloading), the plastic hinging begins once again
at <P = 0° and 180°, but the direction of yield moment is re-
versed in this case, as shown in Fig. 4(b). The unloading path
is along the lower S-shaped curve in Fig. 3(a). Since the load-
ing and unloading paths are nearl y parallel to each other, th ey
represent the same stiffness. The size of the hysteresis loop,

(a) (b)
+ (c)
8

IOOMrO
W/R

-4

-8
-0.5 0 0 .5 -0. 5 0 0.5 -0.5 0 0 .5

Base Rot ation , '" ( d egrees )

FIG. 3. Plots of: (a) Moment-Rotation Diagram; (b) Skeleton Stiffness; (e) Hysteresis Loops

JOURNAL OF ST RUCTURAL ENGINEERING I APRI L 1997 / 441

 Ca) Loading Cb) Unloading in most cases the fundamentaJ impulsive mode alone provides
satisfactory results. The hydrodynamic wall pressures under
Shell these assumptions may be expressed as (Veletsos and Tang
1990)
p,(z, 01>. t) = -a(z)p,RA (t)cos 01> (3)
where a(z) = dimensionless fu nctio n that defines the height-
wise variation of pressures; and A(t) = instantaneous pseu-
doacceleration of a single degree of freedom (SDOF) model
Base plate
of the tank-liquid system. The hydrodynamic wall pressures
can be integrated to obtain the following expression for the
overturning base moment M T •
M,
MT(t) = miiA(t) (4)
FIG. 4. Plastic Hinging at Plate Boundary at 01> = ()" [Fig. 1(a)) where m = effective or modal mass of the liquid; and Ii =
height of the resul tant of hydrodynamic wall pressures. The
TABLE 1. Values of Es , ED. and ~II for Three Load Cycles hydrodynamic base pressures may be expressed as
Shown in Fi g 3 (8 )
pir, 01>, t) - -a(r)p,RA(t)cos <I> (5 )
10'Es 10 sEo
1" I
(d eg rees)
(1 )
W, R
(2)
W, R
(3)
'"
(percent)
(4)
where a(r) = di mensionless function with defines the radial
variation of pressures. Values of m , Ii, and a are reported in
literature as functions of the properties of the tank and the
0.2 94 89 7.5
0.35 216 167 6.2 contained liq uid (Haroun and Housner 1981; Veletsos and
0.5 363 236 5.2 Tang 1990); m and ii are usually expressed as fractions of the
totaJ liquid mass ml and liquid height H, respectively.
Note: Results are for Khlp = 2; hlR = 0.0005; h.lR = 0.001; a~/p =
1,800; £/p = 1.5 x 106 ; and WI W/ = 0.015 . The net pressure on the base plate, obtained by superim-
posing the hydrodynamic pressure Pd on the hydrostatic pres-
sure P = p,gH, is given by
hence the amount of energy dissipated, is controlled by th e
p(r, <1>, t) = p,gH - a (r )p,RA(t)cos <I> (6)
distance between the two paths. The distance reduces as My
reduces; in the limiting case when My = a the loading and The pseudoacceieration A(t) in (6) is eliminated by making
unloading paths overlap and the hysteresis loop disappears use of (4), to give
completely, as shown in Fig. 3(b). The plot of Fig. 3(b) is,
therefore, the skeleton stiffness of the base plate, which is MT(r)
p(r, 01>, t) = p,gH - a (r )p,R - -- cos <I> (7)
obtained by assuming a pin condition (zero moment) at the mh
plate-shell junction. The damping curve (hysteresis loop) is
obtai ned by subtracting the skeleton stiffness from the plot of Solution Method
Fig. 3(a); it is shown in Fig. 3(c). The circumferential variation of base pressures is accounted
The area Es under the skeleton stiffness curve [Fig. 3(b)] for by assigning different values of pressure to different beams
denotes the elastic strain energy, while the area ED withi n the in the base plate model of Fig. 2. The pressure on the ith
hysteresis loop [Fig. 3(c)J denotes the energy d issipated in a
beam is
load cycle. Values of Es and ED for the three load cycles are
shown in Table 1. Also shown in this table are the val ues of MT(t)
p ,(r, t) = p,gH - c«r)p,R - _ - cos
[21T
- (i - I) ] (8)
the quantity mh n

{ _ 100Eo On account of the variation of a in the radial direction, the

(2)
h - 41TEs intensity of pressure varies along the length of each beam. This
variation may, however, be neglected because the uplifted
where 'h = perce ntage effective viscous damping due to hys-
teretic action (Chopra 1995; pages 94 - 100). If the tank wall
width of the plate is only a small fraction of the tank radius
R. In this analys is, the val ue of pressure corres pondin g to T =
were rigid , 'h
would denote the contribution of the base plate
to the overall system damping. For a real tank, the effect of
R is used for the entire length of each beam. This leads to the
following expression for the pressure on the ith beam .
thi s contribution is reduced because only a pan of the total
deformation takes place in the base plate, while the remaining
p,(r) = { I - ~
M T(r)
- - cos - [27r (i - I) ]} p,gH (9)
takes place in the tank wall. In Table I, both Es and ED are W,R n
seen to increase with increase in the size of the load cycle; the
latter, however, increases at a slower rate. The hysteretic in which
damping 'h,
therefore, reduces with increase in the size of the a (R)
( 10)
load cycle. ~ = (mi m,)(iilH )(HIR)'
BASE UPLIFTING UNDER HYDRODYNAMIC LOADING The solution method for dynamic uplifting is similar to that
for static uplifting, except that the pressure on various beams
Hydrodynam ic Pressures needs to be recomputed at the end of each rotation increm ent
Il.IjJ using (9) and (10).
In a seismically excited tank, the hydrodynamic pressures
are generated by the impulsive action of the liquid movi ng
Numerical Results
rigidly with the tank wall, and the convecti ve action of the
liquid moving in sloshing modes near the free surface. Of the Results were obtained for different values of the hydrod y-
two, the impUlsive action usually dominates the response, and namic parameter r3 , subgrade modulus K, base plate th ic kness
442 I JOURNAL OF STRUCTURAL ENGINEERING I APRIL 1997
 (a) (b) (c)
8
{! = 5 (! = 15

-4

-8 -0.5 a 0.5 -0.5 a 0.5 -0.5 a 0.5

Base Rotation, 'if; (degrees)
FIG. 5. Effect of Hydrodynamic Base Pressures on Skeleton Stiffness and Hysteresis Loop

(a) (b) (c)

8
K.hl p = 1 "hi p = 2 Khl p = 100

-8 -0.5 a 0.5 -0.5 a 0.5 -0.5 a 0.5

Base Rotation , 'if; (degrees)
FIG. 6. Effect of Foundation Stiffness on Skeleton Stiffness and Hysteresis Loop

(a) (b) (c)

8
hi R = 0.00025 hlR = 0.0005 hlR = 0.001

-8 -0.5 a 0.5 -0.5 a 0.5 -0.5 a 0.5

Base Rotat ion , 'if; (degrees)
FIG. 7. Effect of Base Plate Thickness on Skeleton Stiffness and Hysteresis Loop

h, shell thickness hs . and yield level of plate material crr Se-
lected plots of skeleton stiffness and hysteresis loop are pre-
sented in Figs. 5-7, and the values of E" ED, and {, in Table 2.
For a typical tank, a.(R) - 0.5 (Veletsos and Tang 1990).
For steel tanks with height to radius ratio HIR between 0.5
and 3, mimi ranges from 0.3 to 0.7, and IiIH from 0.4 to 0.55.
The value of 1>, determined from (10), therefore ranges from
0.14 for a very tall tank to 16 for a very broad tank. The plots
in Fig. 5 show that as f3 increases from 0 to 15 , the skeleton
stiffness reduces. This is because the hydrodynamic base pres-
sures assist uplifting by acting upward on the uplifted side of
the base plate and downward on the contact side. The size of
the hysteresis loop (ED in Table 2) reduces by 57%, as I> in-
creases from 0 to 15; the corresponding reduction in the value
of ~, is 46%.
In Fig. 6, an increase in K is shown to increase sign ificantl y

JOURNAL OF STRUCTURAL ENGINEERING 1 APRIL 1997 / 443

 h~~1
TA BL E 2. Effect of System Parameters o n Values of E• • Eo.
and., fo r Load Cyc le of S ize 1",1 = 0.5"

P arameter
105 E$
W,R
105 £
W, R
0 .,
(percent) m
c,

r
(1 ) (2) (3) (4)
~=O 363 236 5.2
~=5 338 202 4.8
~ = 15 292 101 2.8
Khlp = 1 286 202 6.0
Khlp = 2 338 202 4.8

I,·
Khlp = 100 551 257 3.7
hiR' = 0.0025 252 98 3.1
hiR' = 0.005 338 202 4.8
hiR' = 0.01 410 280 5.4
h,IR b = 0.005 288 173 4.8
h,IR b = 0.01 338 202 4.8
h,/R' = 0.02 418 209 4.0
u 7 1p = 1,800 338 202 4.8
u 7 1p = 2,700 338 247 5.8 (-) L.... ;;,(t) (b) L.... ;;,(t )
u ,lp = 3,600 338 259 6.1
FIG. 8. Modet of Tank-liquid System
Note: unless noted otherwise, 13 = 5; Khlp = 2; hlR = 0.0005; h,IR =
0.001; u ,lp = 1,800; Elp = 1.5 X IO's; and WIW/ = 0.015.
-For these results. Khlp = I. 2, and 4. respectively. milo + c(';o - Ii.j, ) + k(u o - IiljJ) = -mx, (t) (lla;
"For these results, WIW, = 0.0075, 0.015, and 0.03 , respectively.
whereas the equilibrium of base moments requires that

the skeleton stiffness. Although the size of the hysteresis loop (C(';o - Ii.j,) + k(uo - IiljJ)]1i = MT(IjJ) + c • .j, (lIb)
'h
also increases with increase in K, the damping ratio actually
reduces as seen in Table 2. It should be pointed out that the
where Uo = overall horizontal displacement of the mass relative
to the moving base (overdot denotes differentiation with re-
values of Khlp = I and 100 imply that the base plate, when spect to time t); C ::: damping coefficient for the tank in its
subjected to the hydrostatic pressure p, settles by h and hlI 00, fixed-base condition; c.. = soil damping coefficient; k ::: stiff-
respectively. These two values of Kh /p are expected to rep- ness of the superstructure; and MT (~) the moment in the base
resent a soft soil and a rigid concrete mat foundation, respec- spring, as a function of the time-dependent rotation 41.
tively. Eqs. (lla) and (lIb) are solved incrementally, assuming
In Fig. 7, an increase in the base plate thickness is shown linear relationship between each moment increment !l.MT and
to increase both the skeleton stiffness and the size of the hys- each rotation increment A~ , i.e.
teresis loop. In Table 2, a fourfold increase in h is shown to
increase ~, by 76%. Values in Table 2 show that the skeleton (1 2)
stiffness increases significantly with increase in the shell thick-
ness h,. Although the size of the hysteresis loop (ED) increases, where k~ ::: instantaneous value of the rotational spring stiff-
the value of 'hactually reduces. With increase in the yield
level (j y the skeleton stiffness remains unchanged. but the size
ness. With the prefix A used to represent a small increment
for each remaini ng response as well, (Ila) and (lIb) may be
of the hysteresis loop increases. as seen in Table 2. A twofold written as
increase in (j1 results in a 28% increase in ~ h'
[
m
o 0]
0
{l:J.ilo }
lil:J.ijJ
+ [ c
-c c +
-c]
c.lli'
{1:J.' ;0}
lil:J..j,
SE IS MIC R ESPO NSE ANA LYS IS
System Model + [k -k k
-k] {~uo}
+ k,lli' hl:J. IjJ
= _{ mI:J.X,(I)}
0 ( 13)

The hydrodynam ic pressures in a flat-bottom tank, respond- The preceding set of equations may be reduced to a single
ing in its fundamental impulsive mode of vibration, are in- differential equation by making use of the following approx-
duced by the translational and rocking motion of the tank wall, imate relationship between Auo and !l.41, which is established
as we ll as by th e rocking motion of the tank base plate. For from the second part of (l3) neglecting the effect of damping
an uplifting tank, th e latter contribution to the hydrodynamic forces
pressures can be ignored because in this case, as seen later,
only about 5-10% of the total area of the base plate actually ( 14)
participates in the rocking motion. Under these assumptions
the system may be represented by the model shown in Fig.
8(a), in which the mass m represents the portion of the liquid In ( 14), kr = the effective stiffness of the uplifting system; it
assoc iated with the fundamental impulsive mode, and Ii is the is given by
height of the resultant of the hydrodynamic wall pressures.
The rotational spring at the base represents the rocking resis- I I Ft'
tance of the base plate. The rotational damper accounts for (in - :::- + - (IS)
k. k k>JI
an approximate manner) the effects of soil internal damping
and the radiation damping. On substituting (l4) into ( 13) and premultiplying the resulting
expression by the transpose of the vector on the right side of
Method of Ana lysis (l4), one obtains
mAuo + c.Auo + kr 6.u o = -mAx,(t) ( 16)
The equilibrium of forces on the mass in Fig. 8(a) requires
that which is the incremental equation of motion for the SDOF
444 1 JOURNAL OF STRUCTURAL ENGINEERIN G 1 APR IL 1997
model shown in Fig. 8(b). In ( 16), c, = the effective viscous k,h
damping of the uplifting system; it is given by IlIjI = - Ilu. (21)
k.
c~ = C (k'k )' + hic, ( kk')'
1- (17 ) Corresponding to the new value of the base rotation 1.lI, new
values of the o verturning base moment M T , uplifting of plate
Since the stiffness is inversely proportional to the squ are of boundary, radial separation between the base plate and foun-
the period, the stiffness ratio k/k may be expressed as dation , plastic rotations at plate boundary, axial and hoop com-
pressive stresses in tank wall, and shear force in tank wall are
computed from the base plate analysis discussed earlier.
(18 )
Numeri cal Results
where T = period o f the fixed-base system ; and t == effective
period of the uplifting system. Substitution of (18) into ( 17) Description of Tank, Conditions of Support, and
gives Ground Motion

c- - c
, - (tiT)'
+ -c, ( 1- -
h'
I )'
(tiT)'
Note that the effective damping c, depends on the period elon-
gation t i T. For a system that derives its flexibility from the
superstructu re o nl y, t IT = 1; which gives Cr = c. For a system
that derives its flexibility from the rotational spring only, f IT
;:: 00; which gives Ct ;:: c,/f? The percentage effective viscous
damping {, is given by
100c,t
{ = --
r 4m'Tr
It should be noted thaI th e contributi on of the hys teretic damp-
ing in the base plate all) is not included in ~t.
Eq. ( 16) is solved for Il u. at uniform time steps II I = 0.0002
s by the linear acceleration method (Clough and Penzien 1993;
pages 121-132). The results changed imperceptibly when Ilt
was halved to 0.0001 s or doubled to 0.0004 s. The extremely
small time step was needed to capture accurately the effects
of sudden changes in the stiffness of the sys tem due to up-
lifting of the base plate and yielding at the plate boundary.
The incremental base rotation a1.ll is computed next using the
second part of (14), i.e.
(19) A 15.2 m (50 ft) radius steel tank, filled with water to a
height of 12.2 m (40 ft) is selected for a detailed analysis. The
shell of the tank has a maximum thickness h, = 1.37 cm (0.54
in.) at the base; and the base plate is of uniform thickness h
= 1.04 cm (0.4 1 in .). The material properties for the lank are:
Young 's modulus of elasticity E = 200 GPa (29 X 10' ksi ),
yield stress cry = 248 MPa (36 ksi), and Poisson's ratio v =
0.3. The unit weights of water and the tank material are: Pig
= 9.81 kN/m' (62.4 pcf) and pg = 77 kN/m' (490 pef), re-
spectively. The total weight of tank wall and roof W = 1.41
(20) MN (31 6 kips). The system parameters for the first impulsive
mode of vibration are obtai ned from Veletsos and Tang ( 1990)
for an assumed value of the equivalent uniform thickness of
the shell h, = 1.14 cm (0.45 in.). These parameters are: mg =
39.4 MN (8,85 5 kips), h = 4.9 m (16.2 ft), and T = 0.22 s.
The damping factor c is such that the system exhibits 2%
critical dampi ng in its fixed-base condition. The soil damping
faclOr c. = 237 MN'm's (2.1 X 10' kip-in.-s), which upon
using (19) and (20), together wi th an assumed period elo n-
gation of tiT = 2, gives an effecti ve damping ratio {, = 5%
critical.
The tank is examined for these three conditi ons of suppon:
(I) fully anchored on a rigid foundation; (2) unanchored on a

(a) (b)
120 6
60 3
0 0
-60 -3
=-6.
:;;;
0
8
-120
120 r Unanchored on rigid base,
-6
3
~
-
~

60 0 i!S
c:
Q) ~
E 0 -3 i!!
0
:::;; US
Cl
-60 -6 0;
c:
·E -1 20 -9 ~
:J
t
Q)

<3
120
r Unanchored on flexible b a s \
6
60 3
0 0
-60 -3
-120 -6
0 2 3 4 5 6 7 0 2 3 4 5 6 7
Time (sec) Time (sec)
FIG. 9. Histories of Overturnin g Base Moment and Axial Stress In Tank Wall for Three Different Cond itio ns of Support

JOURNAL OF STRUCTURAL ENGINEERING / APR IL 1997/ 445

rigid foundation; and (3) unanchored on a flexible foundation.
The subgrade modulus for the flexible foundation is K = 54.3
N/cm) (200 pci), which is representative of a compacted gravel
fill; a value I ()() times larger, K = 5.43 kN/cm' (20,000 pci),
is assumed for the rigid foundation.
The ground motion considered is the first 6.3 s of the N-S
compo nent of the 1940 EI Centro, Calif. , earthquake ground
motion record, scaled to a peak value of O.4g.
Response Time Histories
The plots of the overturning moment Mr for the three dif-
ferent conditions of support are shown in Fig. 9(a). A com-
parison between the top and the middle plot shows that, for a
rigidly supported tank, a change from fully anchored to unan-
chored condition causes the response period to elongate from
0.22 s to about 0.4 s, and the peak base moment to reduce by
nearly 25%. Higher damping in the uplifting system is partly
responsible for this reduction. An estimate of damping in the
uplifting system is given later. A comparison between the mid-
dle and the bottom plot in Fig. 9(a) shows that a change from
rigid to flexible fou ndation causes the response period to elon-
gate further to about 0.55 s and the overturning base moment
to reduce by an additional 10%.
Shown in Fig. 9(b) are the plots of the axial stress at the
base of the tank wall computed at <p = 180° [Fig. I(a)]; neg-
ative values imply compression. A comparison between the
top and the middle plot shows that, for a rigidly supported
tank, a change from fully anchored to unanchored condition
causes the axial compressive stress to increase to nearly four
times. This dramatic increase is due to considerably small con-
tact between the wall and the foundation of a rigidly supported
uplifting tank. A comparison between the middle and the bot-
tom plot in Fig. 9(b) shows that, for an unanchored tank, a
change from rigid to flexible foundation causes the axial com-
signifies a tendency of the joint between the wall and the base
plate to increase from 900 [Fig. 4(a)1. whereas a negative value
signifies a tendency of the joint to reduce from 90° [Fig. 4(b)].
Several cycles of large plastic rotations at plate-shell junction
may cause the joint to rupture. The plots in Fig. lO(b) show
that the maximum positive plastic rotation is nearly the same
for both tanks, but the flexibly supported tank also experiences
significant negative plastic rotation, and is therefore more sus·
ceptible to rupture at the plate-shell junction. Note that at the
end of the shaking, some plastic rotation remains at the plate-
shell junction.
Moment-Rotation Diagrams and Overall System Damping
The plots of the relationship between the overturning mo-
ment and base rotation are shown in Fig. 11. The hysteresis
loops for the flexibly supported tank are bigger, due 10 greater
plastic action in the base plate of that tank.
An estimate of the overall system damping is obtained by
adding the hysteretic damping 10 the effective viscous damping
" . The hysteretic damping obtained from (2) is: " = 4% for
the rigidly supported system (for !1jJ! = 0.25°) ; and ,, = 5.4%
for the flexibly supported system (for !1jJ! = 0.3°). Since only
a portion of the total defonnation takes place in the base plate,
the net hysteretic damping for the overall system should be
smaller than Sit. Further, since the deformations are inversely
proportional to the stiffness, the net hysteretic damping s~ is
given by
,; = " k,F.' (22)
k.
which, upon making use of (1 5) and (I8), gives
,; = " [1 - (t/~)'] (23)

pressive stress to reduce to less than one-third.
Shown in Fig. 100a) are the plots of base uplifting at q, =
180° for rigidity and flexibly supported unanchored tank. The
maximum uplifting is nearly the same for both tanks, but the
flexibly supported tank also experiences significant foundation
penetration ( - ve uplift). Small nonzero values of penetrations
for the rigid foundation are on account of the very high, yet
finite, value of the subgrade modulus K for that foundation.
Shown in Fig. I O(b) are the plots of plastic rotation at the
plate boundary at <p = 180°. A positive value of plastic rotation
For the rigidly supported tank, f IT - 1.8 ; which gives St =
4% [from (J9) and (20)] and G = 2.8%. For the flexibly sup-
ported tank, tiT - 2.5%; which gi ves " = 7.6% and ,; =
4.5%. The overall system damping (', + ,;) for the rigidly
and flexibly supported tank is therefore 6.8 and 12.1 %, re-
spectively.
Circumferential Distribution of Responses
The plots of circumferential distribution of base uplift, ra-
dial separation, and plastic rotation are shown in Fig. 12, and

(a) (b)
6 15
runanChored on rigid base
4 10
(i)
2 5 OJ
~
Ol
§. 0 0 OJ
;E ~
g-2 -5 <=
a
OJ

'"'"
6
r Unanchored on flexible base"""\ 15
~
a:
CD
4 10 ()

~
2 5 '"
1i:
0 0

-2 0 2 3 4 5 6 7 0 1 2 3 4 5 6 7 -5
Time (sec) Time (sec)
FIG. 10. Histories of Base Uplifting and Plastic Rotation at Plate Boundary for Rigidly and Flexibly Supported Unanchored Tank

446 1 JOURNAL OF STRUCTURAL ENG INEERING 1 APRIL 1997

those of axial stress, hoop stress, and shear force are shown
in Fig. 13. The plots are at an instant of time when the tank's
respo nse is maximum. The dashed line represe nts the footprint
of the tank wal l and the solid line represents the magnitude of
the response; positive values are plotted radially inward.
The rigidly supported tank experiences an uplift of 13.2 cm
(S.2 in.) at 4> = 180° and practically no foundatio n penetration
at 4> = 0°. The flexibly supported tank experiences an uplift
of IS.5 cm (6.1 in.) and a foundati on penetration of 4.1 em
( 1.6 in.). The maximum separation between the base plate and
the foundation is 104 cm (4 1 in.) for the rigidly supported,
and 94 cm (37 in.) for the flexibly supported tank. These val-
ues are significantly small compared to the radius of the tank
which is IS.2 m (SO ft). The net area of the uplifted portion
of the base plate is about 5% of the total base area for each
tank. The maximum positi ve plastic rotation, at <J> = 180°, is
12.3° fo r the rigidly supported, and 14.5° for the flexibly
supported tank; the flexibly supported tank also experiences a
negative plastic rotation of 7 .2° at 4> = 0°,
The axial force generated by the moment Mr and the weight
of tank W is distributed over an arc of central angle equal to
40° for the ri gidly supported, and 1 \30 for the flexibly sup-
ported tank (Fig. 13). The maximum axial compressive stress
for the rigidly supported tank is S7 .2 MPa (8.3 ksi), which is
more than three times the val ue of 17.9 MPa (2.6 ksi) for the
flexibly supported tank.
As the base of the tank uplifts, the tendency of the tank wall
to move radially inward induces hoop stress cr~ 4> in the wall
(Malhotra 1995). The hoop compressive stress is maximum on
the uplifted side of the tank and nearly zero on the contact
side; values of 14S MPa (2 \.1 ksi) for the ri gidly supported,
and 190 MPa (27.6 ksi ) for the flexibly supported tank, are
quite significant. Nonunifonn distribution of th e axial com-
pressive stress at the base of the tank wall induces shear force

(a) (b)

80 80
Rigid foundation Flexible foundation
:iF
c. K = 20000 pci " = 200 pci
:;;:
0
0 40 40
0
:::.
E
Q)
E 0 0
0
;:;;
C>
c
·c
:; -40 -40
1::
Q)

<3
-80 -80
-0.4 -0.2 0 0.2 0.4 -0.4 -0.2 0 0.2 0.4
Base Rotation (degrees) Base Rotation (degrees)

,' ,.- . . .. '· ·- -· · ·

FIG. 11. Base Moment-Rotation Diagrams

(a) Rigidly Supported

..............
,,
,
.., "" ""
•,
5.~·:- t-
0 G
•,,
,.--
,, ,
41
,,
. ................. ,
, ...
, ,,
,,
,
" , ,' .. ............. .. .. ........ "
Base Uplift (in.) Radial Separation (in.) Plastic Rotation (deg)

(b) Flexibly Supported

",--,',' -.. " ,.
.. .......... ..
.... ,,\
•, ,,
•, ,
-1.6 6 . 1 ,:-- l- -7.2
, , ,,
,,
,"
..-
"' ........ '"

Base Uplift (in.) Radial Separation (in.) Plastic Rotation (deg)

FIG. 12. Circumferential Plots of Responses at Base of Rigidly and Flexibly Supported Unanchored Tank

JOURNAL OF STRUCTURAL ENGINEERING 1 APRIL 1997 1447

 (a) Rigidly Supported
.. ..........
,,
,, ,,
,,
-8 .3 - :, 40·
, ,,
,,
,......
.... _-- _.... " "
Axi al Stress (ksi) Hoop Stress (ksi) Shear Force (kips)

(b) Flexibly Supported

,\ . . ---- .. ... . " ,,

,,
, ,
-2.6 , ••
,• ••
,,
...... _........ '
Axial Stress (ksi) Hoop Stress (ksi) Shear Force (kips)
FIG. 13. Circumferential Plots of Responses at Base 01 Rigid ly and Flexib ly Support ed Unanchored Tan k

TABLE 3. Effects of Ground Motion Intensity on Maximum Responses

Ii,
(g)
(1 )
M,
103 kN ' m
(10' kip-ft)
(2)
~
(degrees)
(3)
em (in.,.
(4)
w,
em (in .)b
(5)
L
em (in.)
(6)
(degrees)·
(7)
9,
(degrees)!>
(8)
"u
MPa (ksi)
(9)
" ..
MPa (ksi)
(10)
v.,
MN (kips)
(1 1)
0.2 70.2 (51.8) 0 .10 3.3 (1.3) -2.0 (-0.8) 56 (22) 2.6 -1.2 -10.3 (-1.5) -11.0 (- 1.6) 1.63 (367)
0.3 82.7 (61.0) 0 .22 8.6 (3.4) -3.0 (-1.2) 76 (30) 8. 1 -3.9 -13 .8 (-2.0) -67.6 (-9.8) 1.96 (440)
0.4 95.8 (70.7) 0 .37 15.5 (6.1) -4. 1 (- 1.6) 94 (37) 14.5 -7.2 -17.9 (-2.6) -190.3 (-27.6) 2.30 (518)
0.5 114.1 (84.2) 0.58 25 .2 (9.9) -5.6 ( -2 .2) 119 (47) 21.8 -11.4 - 23.4 (-3.4) -4 13.2 (-59.9)' 2.75 (618)
Note: Results are for H :: 12.2 m (40 ft); R = 15.2 m (50 ft); w"", 1.41 MN (316 kips); h.:z: 1.37 em (0.54 m.); h = 1.04 em (0.4 1 tn.): and ~ = 200 pel.
-Maxi mum + lU' vaJue
bMaximum -ue value.
cStress eltceeds yield value.

V"'t on vertical sections of the wall (Mal hotra 1995). Due to
symmetry. the shear force is zero on vertical sections of the
tank wall passing through 4> = 0° and 180°; on a section in-
between it altains a maximum value of 2.81 MN (631 kips)
for the rigidly supported and 2.30 MN (518 kips) for the flex-
ibly supported tank.
Effects of Ground Motion Intensity
The sensitivity of the responses of the flexibly supported
unanchored tank to the intensity of ground shaking was in-
vestigated by considering scaled versions of the EI Centro
ground mOlion record. The effects of varying the frequency
content of the ground motion is not examined here. Table 3
lists the maximum tank responses for different values of
the peak ground acceleration x" ~ The responses most sensitive
to change in intensity are the base rotation tV. base uplifling
w 1( + ve), foundation pe netratio n WI( - ve) , plastic ro tations
9, and the hoop compressive stress CT<tllb; these responses in-
crease at a rate faster than the peak ground acceleration X"
The responses th at are relatively less sensitive to changes in
intensity are the overturning base moment Mr. radial separa-
tion L, axial compressive stress CT w and the shear force V<t> x;
these responses increase at a rate slowe r than the peak ground
acceleration Xg •
In Table 3, the hoop stress CT<t>41 exceeds the yield value of
248 MPa (36 ksi) for x, = O.5g. This is on accounl of the
assumption made in the analysis that the tank wall behaves in
a li near-elastic manner, which is not true fo r this particular
448 / JOURNAL OF STRUCTURAL ENGINEER ING / APRIL 1997
case. The results are still shown to identify trends in the var-
ious responses.
Effec ts of Subgrade Modulus
The results for progressively increasing flexibility of the
foundation (reducing values of subgrade modulus K) are pre-
sented in Table 4. An increase in foundation flexibility red uces
the upli fting resistance of the base plate (increases the effective
period of the syste m), and allows greater contact between the
wall and the foundation after uplifting. As a result, the over-
turning base moment Mr reduces, the base rotation $ increases,
and the axial compressive stress in the tank wall CT~ : reduces.
In addition, the values of the base uplifting, foundation pe n-
etration, plastic rotations, and hoop compressive stresses in-
crease, and the values of the radial separation L. and the shear
force V4l : red uce. Respon ses mos t sensitive to c hanges in foun -
dati on flexibility are the axial compressive stresses in the wall
and the negative plastic rotations in the base plate.
Effects of Base Plate and Wall Thicknesses
The responses of the flexibly supported tank were computed
for several different val ues of the base plate thickness h. Re-
sults showed that an increase in base plate thickness is asso-
ciated wi th: ( 1) red uced values of base rotati on, base uplifting,
fo undation penetration, plastic relations, and hoop compres-
sive stress; and (2) increased values of the overturni ng base
moment M T , radial separation L, axial compressive stress CT w
and shear force V4It • Most significant effect of an increase in
 TABLE 4 Effects of Subgrade Modulus K on Maximum Responses
M, w, e,
K 103 kN'm ~ L a" a" V"
(pci) (10' kip-ft) (degrees) em (in.)- em (in .)!> em (in.) (degreest (degrees)!> MPa (ksi) MPa (ksi) MN (kips)
(1 ) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11 )
20,000 104.2 (76.9) 0.26 13.2 (5.2) 0.5 ( 0.2) 104 (41) 12.3 0.1 57 .2 ( 8.3) 145.4 ( 21.1 ) 2.81 (631)
1,000 102.5 (75.6) 0.32 15.2 (6.0) -1.8 (-0.7) 102 (40) 14.2 -4 .0 -27.5 (-4.0) -185.4 (-26.9) 2.59 (582)
500 100.7 (74.3) 0.34 15.5 (6.1) -2.8 (-1.1 ) 99 (39) 14.5 -5.4 -23.5 ( -3.4) -193.0 (-28.0) 2.50 (561)
200 95 .8 (70.7) 0.37 15.5 (6.1) -4 .1 ( -1.6) 94 (37) 14.5 -7.2 -17.9 (-2.6) -190.3 (- 27 .6) 2.30 (518)
100 94.2 (69.5) 0.43 17.0 (6.7) - 6.1 (-2.4) 91 (36) 15 .7 -9.7 -15 .8 (-2.3) -219.2 (-31.8) 2.20 (495)
Note: Resuils are for H = 12.2 m (40 ft); R = 15.2 m (50 ft); W = 1.41 MN (316 kips); h, = 1.37 cm (0.54 In .); h = 1.04 cm (0.4 1 In.); and x, - 0.4 g.
"Maximum + tie value.
~Maxi mum - tit' value.

base plate thickness is to reduce the negative plastic rotation
at the plate boundary,
An increase in the thickness of the tank wall hs has a three-
fold effect: ( I) it reduces the fixed-base period of the system;
(2) it increases the weight of the tank wall. hence the resistance
to uplifting; and (3) it further increases the resistance to up-
lifting by increasing the resistance against inward movement
at the plate boundary. The results obtained for several different
values of the shell thickness showed that an increase in shell
thickness is associated with reduced values of base rotation,
base uplifting, positive plastic rotation, axial compressive
stress, and hoop compressive stress. and an increased value of
the negati ve plastic rotation. Responses not particularly sen-
sitive to changes in shell thickness are the overturning base
moment M r , foundation penetration w!( - ve), radial separation
L , and shear force V.,.
CONCLUSIONS
Seismic response of unanchored liquid-storage tanks, sup-
ported directly on flexible soil foundations. has been examined
using an analytical approach. First, a detailed insight has been
provided into the uplifti ng stiffness and damping of the base
plate under hydrodynamic loading. Next, an efficient method
has been presented for the dynamic response analysis of tanks
under seismic loading. The trends observed in the numerical
results are in agreement with those observed in past experi-
mental studies. The following conclusions are based on the
results presented in this paper.
For unanchored tanks, the hysteretic dampi ng due to plastic
yielding in the base plate may range from 2.5 to 5%. The value
of the damping reduces with increase in the size of the load
cycle.
The hydrodynamic base pressures reduce the uplifting stiff-
ness and energy dissipation capacity of the base plate. The
effect of hydrodynamic base pressures is significant for broad
tanks only.
Increase in the foundation stiffness increases the uplifting
stiffness of the base plate, but reduces its energy dissipation
capacity. An increase in the thick ness of the base plate or the
tank wall increases the uplifting stiffness of the base plate, but
only an increase in base plate thickness increases its energy
dissipation capacity. Increase in the yield level of the base
plate material increases the energy dissipation capacity of the
base plate. but does not affect its uplifting stiffness.
Base uplifting significantly reduces the magnitude of the
hydrodynamic forces generated by ground shaking. The flex-
ibility of the supporting foundation further reduces the hydro-
dynamic forces.
Unlike rigidly supported tanks, base upl ifting for flexibly
supported tanks is not associated with significant increase in
the axial compressive stress in the lank wall. Unanchored tanks
on flexible soil foundations are, therefore, less likely to ex-
perience "elephant-foot" type buckling of their walls as com-
pared to similar tanks on rigid concrete .mat foundations.
The flexibility of the supporting soils may allow significant
foundation penetration and may lead to greater values of base
uplifting and hoop compressive stresses in the tank walL Due
to the nonlinear behavior of soils (not considered in this
study), soil supported tanks may experience uneven and per-
manent senlement around the perimeter.
The foundation flexibility leads to greater plastic rotations
at the plate boundary and, therefore, greater dissipation of en-
ergy due to hysteretic action. For the same reason, the likeli-
hood of rupture at the plate-shell junction is greater for a flex-
ibly supported tank, The radial separation between the base
plate and the foundation is not significantly affected by
changes in foundation flexibility. The shear force in the tank
wall reduces as the foundation flexibility increases.
Responses most sensitive to changes in intensity of ground
shaking are the base rotation, base uplifting, foundation pen-
etration, plastic rotations at plate boundary, and hoop com-
pressive stresses in tank walL Responses relatively less sen-
sitive to changes in intensity are the overturning base moment,
radial separation between base plate and foundation, axial
compressive stresses in tank wall. and shear force in tank wall.
An increase in either the base plate thickness or the tank
wall thickness reduces the base rotation and the uplifting of
plate boundary. In addition, an increase in plate thickness re-
duces the foundation penetration and plastic rotations, and an
increase in wall thickness reduces the axial and hoop com-
pressive stresses in tank wall.
APPENDIX I, REFERENCES
Akiyama, N., and Yamaguchi, H. (1988). "Experimental study on lift-
off behavior of flexible cylindrical tank." Proc., 9th World Con! on
Earthquake Engrg., Vol. 6, Tokyo- Kyoto. Japan, 655-660.
Cambra. F. (1982). "Earthquake response considerations of broad liquid
storage tanks." Rep. EERC 82-25. Earthquake Engrg. Res. Clf., Univ.
of California. Berkeley, Calif.
Chopra, A. K. (1 995). Dynamics of structures: theory and applications
to earthquake engineering. Prentice-Hall, Inc., Englewood Cliffs, N.J.
Clough, R. W., and Penzien , J. ( 1993). Dynamics of structures, 2nd Ed.,
McGraw-Hill Book Co., Inc .. New York. N.Y.
Gates. W. E. (1980). "Elevated and ground-supported steel storage tanks.
reconnaissance report. Imperial County. California Earthquake of Oc-
tober IS , 1979." Earthquake Engrg. Res. Inst.. Oakland, Calif.. 65-
73.
Hanson. R. D. (1973). "Behavior of liquid-storage tanks. the Great
Alaska earthquake of 1964." Nat. Academy of Sci., Washington. D.C..
Vol. 7, 331-339.
Harou n, M. A., and Badawi, H. S. (1988). "Seismic behavior of unan-
chored ground-based cyli ndrical tanks." Proc., 9th World Con! on
Earthquak.e Engrg., Vol. 5, Tokyo-Kyoto, Japan, 643 -648.
Haroun, M. A., and Housner, G. W. (1981). "Dynamic interaction of
liquid storage tanks and foundation soil." Proc., 2nd ASCEIEMD Spec.
Coni on Dyn. Response of Struct., ASCE. New York, N.Y.. 346-360.
Housner. G. W. (1963). "The dynamic behavior of water tanks ." Bull.
Seismological Soc. of Am .• 53(2). 381-387.
Jacobsen, L. S. (1949). " Impulsive hydrodynamics of fluid inside a cy-
lindrical tank and of fluid surrounding a cylindrical pier." Bull. Seis-
mological Soc. of Am., 39(3), 189-203.
Lau. D. T.. and Clough, R. W. (1989). "Static tilt behavior of unanchored

JOURNAL OF STRUCTURAL ENGINEERING / APRIL 1997 / 449

 cylindrical tanks ." Rep. EERC 89- 11 . Earthquake Engrg. Res. etc., c = viscous damping of fixed-base system;
Univ. of California. Berkeley, Calif. Ce effective viscous damping of uplifting system defined by
Malhotra. P. K. (1995). "Base uplifting analysis of flexibly supported Eq. (17);
liquid-storage tanks." J. Earthquake Engrg. Sirucr. Dyn. , 24(12), c... soil damping coefficient;
1591-1607. E Young's modulus of elasticity;
Malhotra. P. K. , and Veletsos. A. S. ( 1994). "Uplifting response of unan-
chored liquid-storage tanks." J. Siruer. Engrg., ASCE, 120(12),3525-
ED ; dissipated energy in a load cycle [Fig. 3(c)};
3547. Es = strain energy associated with base plate uplifting [Fig.
Malhotra. P. K., and Veletsos, A. S. (1995). "Seismic response of unan- 3(b)};
chored and partially anchored liquid-storage tanks ." Rep. TR·J05809, g = acceleration due to gravity;
Electric Power Res . lnst.. Palo Alto, Calif. H ; height of liqu id in tank;
Manos, G. C., and Clough, R. W. (1982).• 'Further study of the earthquake h = thickness of base plate;
response of a broad cylindrical liquid-storage tank model." Rep. EERC h= height of the resultant of hydrodynamic wall pressures;
82·07. Earthquake Engrg. Res. Ctr., Univ. of California. Berkeley, Calif. hs = thickness of shell at the base of tank;
Manos, G. c., and Clough. R. w. ( 1985). "Tank damage during the May k = stiffness of fixed-base system;
1983 Coalinga Earthquake. " J. Earthquake Engrg. Siruct. Dyn., 13(4), ke = effective stiffness of uplifting system;
449-466. k. instantaneous stiffness of rotational spring representi ng
Natsiavas, S. ( 1987). " Response and failure of fluid-filled tanks under base
excitation," PhD thesis. California Inst. of Tech no!.. Pasadena, Calif. the base plate;
"Northridge earthquake of January 17, 1994, reconnaissance report." L radial separation between base plate and foundation [Fig.
(1 995). 95-01, Earthquake Engrg. Res. Inst., J. F. Hall. ed., Oakland, I(b)};
Calif., Vol. I, 162-168. Mr overturning base moment due to hydrodynamic wall pres -
Peek. R. (1986). " Anal ysis of unanchored liquid storage tanks." Rep. sures;
EERL 86·0 1, California In st. of Tech no!., Pasadena, Calif. M, plastic moment capacity per unit circumferential length of
Sakai, F., Isoe. A. . Hirakawa. H. , and Mentani, Y. ( 1988). "Experimental base plate;
study on uplifting behavior of flat-based liquid storage tanks without m = fundame ntal impulsive mass o f system;
anchors." Proc., 9th World Con/. on Earthquake Engrg., Va!. 6, Tokyo- m, mass of liquid in tank;
Kyoto, Japan, 649-654. PlgH = hydrostatic base pressure;
Smoots, Y. A. ( 1973). "Observed effects on foundations of structures, San P
Fernando Earthquake of February 9, 1971." U.S. Dept. of Commerce,
R radi us of tank:
Nat. Oceanic and Atmospheric Admin" Washington, D.C., Vol. I, 805 T period of fixed-base system;
- 807. t ; effective period of uplifting system;
U.S. Department of Commerce (DOC) ( 1973). "Earthquake damage to overall horizo ntal displacement of mass with respect to
water and sewage facili ties, San Fernando earthquake of February 9, moving base [Fig. S(a)};
1971." Nat. Oceanic and Atmospheric Admin ., Washington, D.C., Vo!' v~ : shear force on vertical section of tank wall;
2, 135-138. W = weight of tank wall;
Veletsos. A. S., and Tang. Y. (1990). "Soil-structu re interaction effects for w, total liquid weight;
laterally excited liquid-storage tanks." J. Earthquake Engrg. Struct. w, ; uplift at plate boundary [Fig. I(b)};
Dyn., 19(4), 47 3-496. x, peak horizontal ground acceleration;
Ve letsos. A. S .. and Yang. J. Y. (1977). "Earthquake response of liquid z= vertical distance from center of base plate;
storage tanks." Proc.. Ad\}. in Civ. Engrg. through Engrg. 13 hydrodynami C base pressure parameter defined by Eq.
Mech.. Engrg. Meeh. Div. Spec. Con! , ASCE, New York, N.Y.. 1-
( 10);
24.
~ hysteretic damping due to yielding at plate boundary;
a l plastic rotation at plate boundary;
APPENDIX II, NOTATION K subgrade modulus of foundati on;
yield stress of tank. material ;
The following symbols are used in chis paper.' :; axial stress in tank wall;
= hoop stress in tank wall;
circumferential coordi nate [Fig. l (a)]; and
A (c) = pseudoacceleration of SOOF model of tank· liquid system; ; rOlalion of tank base [Fig. I (b)}.

450 I JOURNAL OF STRUCTURAL ENGINEERING I APRIL 1997