DYNAMIC PRESSURES ON ACCELERATED FLUID CONTAINERS

By G. W. HOUSKER

ABSTRACT
An analysis is presented of the hydrodynamic pressures developed when a fluid container is sub-
jected to horizontal aecelerations. Simplified formulas are given for containers baving twofold
symmetry, for da.ms with sloping faces, and for flexible retaining walls. The analysis includes both
impulsive a.nd convective fluid pressures.

I NTRODUCTIO)i"

THE DYNAMIC fluid pressures developed during an earthquake are of importance in

the design of structures such as dams and tanks. The first solution of such a problem
was that by Westergaard (1933), who determined the pressures on a rectangular,
vertical dam subjected to horizontal acceleration. J acobsen (1949) solved the cor-
responding problem for a cylindrical tank containing fluid and for a cylindrical pier
surrounded by fluid . Werner and Sundquist (1949) extended Jacobsen's work to
include a rectangular container, a semicircular trough, a triangular trough, and a
hemisphere. Graham and Rodriguez (1952) gave a very t horough analysis of the
impulsive and convective pressures in a rectangular container. Hoskins and Jacob-
sen (1934) determined impulsive fluid pressures experim entally, and Jacobsen and
Ayre (195]) gave the resu lts of similar measurements. Zangar (1953) presented
the pressures on dam faces as measured on an electric analog.
The foregoing analyses wore all carried out in the same fashion , which requires
finding a solution of La Place's equation that satisfies the bound ary conditions. With
these known solutions as checks on accuracy, it is possible to deri ve satisfactory
solutions by an approximate method which avoids partial-differential equations and
infinite series and presents solutions in simple forms. The approximate method ap-
peals to physical intuition and makes it easy to visualize the fluid motion, and it
thus seems particularly suitable for engineering applications. To introduce the
method, the problem of t he rectangular tank is treated in some detail; applications
to other types of containers are treated Illore concisely.
The more exact analyses show that the pressures can be separated into impulsive
an d convective parts. The impUlsive pressures arc those associated with the forces
of in ertia produeed by impulsive movements of the walls of the container, and the
pressures developed arc directly proportional to the acceleration of the container
walls. The convective pressures are those produced by the oscillation of the fluid
and are thus the comlOquences of the impulsive pressures. In the following analysis
the impulsive and convective pressurcs a,re examincd separately, t he fluid is assumed
to be incompressible, and Huid displacements are assumed to bc sma.ll.

IMP ULSIVE PRESSURES

Consider a container ,vith vertical side walls and horiwntal bottom that is sym-
metrical \vith respect to the vertical x- y and z-y planes. Let the walls of the con-
tainer be given an impulsive acceleration Uo in the x direction. This will generate
Manuscript received for publication November 17, 1955.
l'51
 16 BULIJETIN Ot' THE SEISMOLOGICAL SOCIETY OF AMERICA

fluid acceleration u, v in the x, y directions and may also generate an acceleration

component tV in the z direction. For a rectangular tank tV is obviously zero, and
Jacobsen ( 1949) showed that for a cylindrical Lank tV is also zero. Tn what follows
it will be assumed that the ratio of 1.0 to u is either exactly zero or at least so small
that tV may be neglected. Physically, this is equi valent to having the fluid restrained
by t hin, vertical membranes, spaced dz apart, which force the fluid motion to take

I~X

Fig. 1.

--1 ~"
;r
Lr ]"
...c:

[I. >u+dUdx
. U dx I. u .I
Fig. 2. Fig. 3.

place in the x, y plane only. It is then sufficient to consider the impulsive pressures
generated in a lamina of fluid.
Consider a lamina of Auid of unit thickness, figure 1, and let the walls be given a
horizontal acceleration Uo. The initial effect of this acceleration is to impart a hori-
zontal acceleration to the fluid and also a vertical component of aeeeleration. This
action of the fluid is similar to that which would result if tbe horizontal component,
U, of fluid velocity were independent of the y coordinate j that is, imagine the fluid
to be constrained by thin, massless, vertical membranes free to move in the x direc-
tion, and let the membranes be originally spaced a distance dx aps..rt. When the
walls of the container are given an acceleration, the membranes will be accelerated
with the fluid, and fluid will also be squeezed vertically with respect to t he mem-
branes. As shown in figure 2, the fluid constrained between two adjacent mem-
 DYKAMIC PRESSURES ON ACCELERATl::D FLUID CON'r AINERS 17
branes is given a vertical velocity
du
v ~ (II - y) fix (I)

Sin ce the flu id is incompressible, the accelerations satisfy the same equation, so

. (h ) du (Ia)
v= -Ydx
T he pressure in t he fl uid is then given by
a~
:..r:.:
_ _pI) (2)
iJy -

where p is the density 01 tho ftuid. The total horizontallorce on one membrane is

p = i lopdY (3)

These equations may be '''Titten

d1i
v = (h - y)-
fix

P = - p J,f ' (h d du
- y) d: dy ~ - ph'(y/ h - Hy/ h)') -dx (4)

P = - ph'
i' o
du du
(y/ h - t(y/ !!)') - dy = - ph'/ 3 -
dx dx

The acceleration u is determi ned fro m the horizontal motion of t he fluid con-
tained botwoon two membranes. The slice of fluid shown in figure 2 will be acceler-
ated in the x direction if t he pressures on t he two faces differ. The equation of
motion is
dP fix = -phdx':'
dx
Using the value of P from equation (4) gives
d'U 3 .
dxt - h2U = 0 (5)

a.nd the solution of this equation is

-x -x
it. = C1 cos h V3 h + C, sinh V3 h (6)

Equations (4) and (6) determine the fl uid pressures, and they are strictly applicable
on ly when t he surface of thc fluid is horizontal, bu t if consideration is restricted to
small displacements of fluid the equations may be used ~vcn when the surface of thc
fl uid has been excited into motion, that is, equations (4) give the im pulsive flu id
pressures, p(t), correspond ing to arbitrary acceleration 1io(t).
If the container is slender, having h > 1.5i, somewhat better results are obtained
18 BULLETIN OF TIlE SEIS MOLOGICAL SOCIETY QP AM.ERICA

--1f-+- -x
w

~a
-, I
" IT
~ I

R R
I I
Fig.4.

by applying equations (4) to the upper portion, h' = 1.5l, of the fluid only and con-
sidering the fluid below this point to move as a. completely constrailled fluid exert-
ing a wall pressure PU! = pliLo (sec fig. 3). At a depth of l.5l the moment exerted on
the horizontal plane by the fluid above is approximately equal to the moment
(t pi4:,P) exerted on the same plane by the constrained fluid below which implies
that the generation of fluid velocity is restricted essentially to the fluid in the upper
part of a slender container.
CONVECTIVE PRES5UR}]S
When the walls of a fluid container are subjected to accelerations, the fluid itself
is excited into oscillations and this motion produces pressures on the walls and floor
of the container. To examine the first mode of vibration of the fluid consider con-
straints to be provided by horizontal, rigid membranes, free t.o rotate, as shown in
figure 4. Let u, v, w be the x, y, z components of fluid velocity, and describe the
constraints on the flow by the following equations :

a(ub) ~ _bav
ax ay
v = xO (7)

aw
oz
~ _(au
ax
+ av)
ay
 DYNAMIC PRESSURES ON ACCELERATED FLUID CONTAINERS 19
where band 0 are as shown in figure 4. These equations state, respectively, that the
fluid at a given x, y moves with a uniform u, that all the fluid at a given x, y moves
with the same v, and that continuity of flow is preserved. In a manner similar to
that of the preceding section the appropriate equations of motion could be written
for the p articular shape of container under conslderation. A general solutlon,
applicable to any shape (twofold symmetry) can be deduced as follows. From the
preceding equations
U = - - - lOllI'
b iJy -R
xbdx
(8)
w = z -b'2 - ae IX xbdx
b iJy _R

where b' . db j dx. The total kinetic energy is thus:

T = (' I,nI"{ + (ae)' (I'

tp J 0 _R -b x 8
2 2 1b2 ay _R xb dx )' ( + (b')')}
1 b
Z2 dx dy dz

~ tp I: {r,lI' + K (:D,} dy
where

I, ~ i x' dA

(9)

The potential energy of the fluid is

V = tpg8 11 2 f x 2 dx dz = tpg811 21s

By Hamilton's Principle

or
0'11) o8dxdt + f " p ((aii)
f "J"(' p (l,e - K;Ji K a + gl,8, ) o8,dt ~ 0
" 0 Y . Y " t,

This gives the two equati.ons

a'8 _ r, II ~ 0
iJIl K
(10)
a'at' (a8)
ay, + g
I.
K 8, ~ 0
20 BULLETI~ OF TIlE SEISMOLOGICAL SOCIETY Q}o' AMERICA

From which there is obtained by integration

. h
8m ~-' y
"
Q - fA K . ",t
sm
sm
. h If. h
VI(
(lOa)

These are the equations for the free oscillation and the natural frequency of the
fundamental mode of vibration. For a container of specified shape, such as rect-
angular, circular, elliptical, etc., it is necessary to evaluate only the integrals I z
andK.
The pressure in t he fluid is given by

~~ -pW ap
-~ -pu
az ax
p~ (JI)

Q= f'
_R
xb dx

Knowing p, the forces and moments on t he walls and floor of the container can be
determined readily.
RECTANGULAR CONTAINER
J'or a rectangular container of unit width as shown in figure], the boundary condi-
tions for the impulsive pressures are it = ~ at x = l, for which equation (6) gives

cosh V3~
h
U == Un - - - - - " (12)
cosh V3 Lh
Equations (4) then give
sinh V3 ~
p~ -pU,h V3 (y/h _ !(y/h)2) _ _--:;h
cosh V 3 L
It (13)

. h' sinh V3 ~
- pUo 3 - ---';1
cosh V3 -h
The wall acceleration, Uo, thus produces an increase of pressure on one wall and a
decrease of pressure on the opposite wall of
 DYNAMIC l~RESSURES ON ACCELERATED FLUW CO~T.Ul\"ERS 21
- - !
p. - pu,h(y/h - !(y/ h)') V3 tanh V3 h (14)

a nd produces a pressure on the bottom of the tank

. ;13 sinh Va f.
- pu,h - 2 - - ---71 (15)
cos h v3':'
h
The total force acting on one wall is

p~
h'
puo - - tanh (16)
V3
and its resultant acts at a distance above the bottom

h, ~ -3 h (17)
8
It is seen that the over-all effect of the fluid on the walls of the contain er is the
same as if a fraction) 2P -7- 2lhp~) of the total mass of the fluid were fastened rigidly
to the walls of the container at a height 3/ 8 h above the bottom. Th e magnitude of
this equivalent mass) kf o) is
- I
tan h V3 Ii
M, ~ M ! (18)
V3 -
It
where M is the total mass of the fluid.
The total moment exerted on the bottom of the tank is

f" xp. dx ~ _ pu,h'! (I _ tanh ~3 f. ) (19)

-, V3 -
II
Including this) the correct total moment on the tank is given when the equivalent
mass M 0 is at an elevation above the bottom of

3 hl+3
h'~8 ( 4( va k_I-I)) (20)
tnn h V3 Ii
The accuracy of the preceding analysis can be judged by comparison with the
values computed by Graham and Rod riguez (1952). Equation (18) gives an M,
slightly larger than that computed by these authors with maximum error less than
2.5 per cent, and equation (20) gives an ". slightly smaller than theirs with. maxi-
mum erro r less than 2 per cent. It may thu s be concluded that for the rectangular
tank the errors introduced by the approximation of equ ation (1) are negligible so
far as engineering purposes arc concerned .
22 BOJ...LETIN OF THE SEISJ.fQr,OGJCAT, SOCIF.TY 01<' AMERICA.

In the case of free oscillations of the fluid in the fundamental mode for a rect-
angular tank of unit width, equations (9) are

>'x
lz =
f _I ,
2
dx = - p
2
3

K = 2 f >'2 (f>'~ dx.)'

- I _ I
4
dx = 15 l5

thus .Jrt = ~ y and equations (lOa) are

sinh @U
.
v - (jh
~i' I sm
.
wt
. h 15 h
sm 'i"2 Y (21)

{5g I5h
w' = ~i Y tanh ~2 y
The velocity at any point in the fluid is given by
l2 _ x 2 dB
U=
2 dy
v = Ox
The pressure in the fluid is given by

(22)

The pressure exerted on the wall of the container, (x = l), is

~"2 T
-

5 1I
[2 f5 cosh .
pw = P"3 \}"2 2
w (J,. sm wt (23)
. h
SID ~5- -h
2 I
The f cree exerted on ono wall is

p 3'I' CJJ2(h sin wl (24)

The total force, 2P, exerted on the tank by the fluid is the same as would be pro-
d need by an equivalent mass M 1 that is spring mounted as shown in figure 5. If Ml
 DYNAntIC pnESSURES ON ACCIU,ERATED FLUID CONTAINERS 23
oscillates with displacement Xl the for ce against the tank and the kinetic energy of
t he mass a.re as follows:
Xl = Al sin wt
FI = -M,Alw2 sinwl (25)
T = ~M1 A12W2 sin 2 wt

Comparing these with the corresponding equations for the oscillating fluid it is
seen that
h
0, - -- ' ' ' - --
I. ~ tanh I. ~
'.I'll '.I-Z I
(26)
III I = ,,(I @.z. tanh I.~)
'" 3 '-12 h '.12 I

"'.
1'1 ,
h,

~
1"1,
hi
/ '\
1-
"--/Mo
~

Fig. 5.

The elevation of l'v/l above the bottom of the tank is determined so that it pro-
duces the same momen t as t he fluid. Considering only the moment of the fluid
pressures on the walls (neglecting the pressures on the bottom), there is obtained

h, = h (1 __15h --"1~_)
15 h
(27)

'.1 2 I tanh '.12 I

When the pressures exerted on the bottom arc also taken into account the height is

15 k
cosh "2 Y - 2~
h, = h 1 - ---:::--'-''----'--::- (28)
(
~ ~sinh ~ ~
24 BUI.JLETIN OF THE SEl SMOI.OGICAL SOCIETY OF AMERICA

Comparing with the exact soluti.on of Graham and Rodriguez, it is found that
equation (21) gives a value for w2 that is slightly too large with a maximum en"or
less than 1 per cent; equation (26) gives a value of M\ slightly too large with a
maximum effor less than 2 per cent.
As shown in figure 5, the over-all effect of the fluid upon the container is the same
as a system consisting of the container , a fixed IUass MOl and spring-mounted masses
M Il Maj etc. It will be noted that the formulas for the higher unsymmetrical
(n = 1, 3, 5 ... ) modes arc the same as for t he first mode if l is replaced by li n.
The response of the system shown in Figure 5 when the container is subjected to
arbitrary horizontal acceleration can be computed readily. }'rom the motion of MIl
the oscillation of the fluid in the fundamental mode can be determined from
equation (26), which gives the relation between Al and ()". The actual displacement
of the water surface is determined from equation (22) I which at y h gives

(29)

This pressure is produced by t he weight and inertia force of the fluid above the
plane y = h. The depth d of water above this plane is thus

(3 0)

CYI,INDRlCAL C ON'.L'AINER

Consider a cylindrical tank as shown in figure 6, subjected to a horizontal accelera-

t ion ~ and let the fluid be constrained between fixed membranes parallel to the
x axis. J acobsen (1949) has shown that an impulse Uo does not generate a velocity
component tV in the fluid so t hat in tills case the membranes do not actually in.tro-
duce a constraint. Each slice of fluid may thus be treated as if it were a narrow
rectangular tank and the equations of t he preceding section "\\-ill apply. The pressure
exerted against the wall of the tank is, from equ ation (14),

p. ~ - rMoh(yj h - !(yjh)') V3 tan h ( V3 ~ cos ) (3 1)

The pressure on the bottom of the tank is

Va sinh V3~
Pb ~ - rMoh 2 ----"; (32)
cOBh V3 {
h
The preceding expressions are not convenient for calculating the total force exerted
by the fluid. The follo wing modification gives very accurate values for R/h small
and somewhat overestimates the pressure when Rl h is not small.
- - R
p . ~ - puoh(yjh - !(yj h)') V3 cos tanh V3 h (31)
 DYNAMIO PRESSURES ON ACCELERATED FLUID CONTAINERS 25
From this expression the rosultant force exerted on the wall is
-R
tanh y3 h
p ~
J.'J." p . cos q, R dq, dy ~ -pUo 7r R,'lh _R
y3 -
h
(33)

from which it is seen that the force exerted is the same as if an equivalent ma.ss 11'10
were moving with the tank/ where
-R
tanh y3 h
M. ~ M R (34)
ya -h
"
u.

1J
I L-
h

Fig. 6.

Comparing with Jacobsen (1949)/ it is found that equation (34) overestimates Mo

with a maximum error less than 4 per cent.
To exert a moment equal to that exerted by the fluid pressure on the wall, the
mass Mo should be at a height above the bottom

h. ~ -3 h (35)
8
26 BULLETIN 01<' THF. SEISMOLOGJCAT~ SOCIl!:'IY OF A};fERICA

If the moment exerted by the pressures on the tank bottom are included, the equiva-
lent mass, M o, must be at a height

3h(1
h,~ "8 + 34 ( Y3~_R )
tanh y3 Ii
)
1 (~ < 15) (36)

to produce tbe proper total moment on the tank. Comparing with Jacobsen (1949)
it is found that equation (36) underestimates ho \vith a maximum error less than
6 per cent.
The free oscillations of the fluid (first mode) are determined from equations (21),
etc. For the cylindrieal tank
I _ 7rR~
,- 4

(37)

sm
"h
, 1my
"0 -
8 ~ 8, 8 R

8m "h!!7h --
8 R
Comparing with the exact solution, Lamb (1932), it is found that equation (37)
slightly overestimates w 2 with a maxllnum error Jess than 1 per cent.
From equations (J 1) the pressure in the fluid is given by

p~
fl'
-P"3 (x
n-ii1 (x)' x(z)') ao
R -4R R
1
ay
(38)

~ ~).
aB
ay

The pressure on the wall is

(
- !! - 27 1 cosh
8 R s inh
\Is
[27 ~
\Is
R

R
()hw2
"
sm wt

cos! <P 8in 2

- -3- - - 4 -
) cos~ (39)

The resultant horizontal force exerted on the wall is

(40)
12M "
= 11 lYO" SIll wt
 DYNAloIlC PRESSURES ON ACCELERATED FI.UID CONTAINERS 27
This force is the same as that produced by an equivalent mass M I oscillating in a
horizontal plane with motion

XI = Al sin wl

M, ~ M!
4
(!!)' !27 R tanh !27 ~
12 \Is h \Is R
(41)

In order that }'f 1 exert the same moment as the fluid pressure on the wall it should
be at an elevation above the bottom of

The pressure exerted on the bottom of the tank is

Pb ~ - pw' ~ sinh ~ ~ (~ - ~ ~)' - ~ ~ (:J) 8, sin wt (42)

8 R
This exerts a moment about the z axis equal to

Including this, the correct total moment on the tank is produced when

h, ~ h (I _ cosh JJ ~ - w) (43)
!27 ~ smh !27 h
\Is R \Is R
EI,UPTICAL TANK

Proceeding in the same way as for the cylindrical tank, the impulsive pressure on
the wall is given by equation (14)
- - /
p. ~ pU,h(yl h - Hyl h)') v3 tanh V3 Ii (44)

with a similar expression for acceleration in the direction of the y axis.

28 BULLETI N 01<' THE SE:ISMOLOGI CAf.J SOCIETY OJ.' AMERICA

For oscillations of the fluid, equations (21) apply and fo r the first mode about the
minor axis

where 2a is the maj or axis of t he ellipse and 2b is t he minor axis. For hla small this
reduces to

Comparing this ,vith the exact solution, Jeffreys (1924), it is found that w is slightly
overestimated with a maximum error less than 1 per cent.

-x
H

'J
Fig. 7. Fig. 8.

COMPOSI'l'E TA N KS

Symmetrical tanks form ed of composite shapes such as that shown in figu re 7 will
have impulsive pressures given by equation (14) and oscillations described by
equations (21). The ,t ank shown in figure 7 has

1
'J
~ 2Rl'
12
+ 71"
R' (!4 + \R
( f- + -)')
371"
(46)

K, ~ Rl' {0 .233 (f)' + 0 .627 (~)' + 1377 ()' + 0197 ()'

+ 0.131 ~ + 0 .0!66}
RECTANGULAR DAM

For a dam with slopin g rectangular face and constraints on the flow as shown in
figure 8, the impulsive pressures a.re given by the following equations:
 DYNAMIC PRESSURES ON ACCl:LERATED FLUID CONTAINERS 29

du
v ~ (h - y) dx + u cos </>
u ~ . CXp ( - V3 x/h)
~~
ay (47)

P ", .
= puoh~ { ---:
L - -2-
COS</>}
v3

Fig. 9.

The resultant horizontal force 011 the dam is

F, ~ ,,".Ii' _._1_
sm</>
{_lV3__ cos2 "'} (48)

For DO ~ 1> ~ 55, equation (48) overestimates Fh by 6.5 per cent; for < 55 the
accuracy of the preceding equation decreases and a different approximation must
be used, as given below.,
When rP < 55 the fluid may be divided into two regions as shown in figure 9,
where a rigid membrane lies along the x axis and has a horizontal acceleration cUo
such that the pressure for ce on each side of the membrane is the same. In the region
to the left of the x axis the foHowing equations describe the flow:

au
v~ (h - y) ax + cu. cos '"
(49)
-pv
 30 BULLETIN O:W THE SEISMOLOGICAL SOCIETY OF AMERICA

Applying Hamilton's Principle to the total kinetic energy in this region leads to
the equation

x -d'"
d:t;2
+ d
3x!!U - 3 tan 2 rp
dx
.
U = 3c Uo cos tan rp (50)

The appropriate solution is

_ c cos <1 (::)" + c cos <I>}

tan <I> l tan

a ~ VI + 3 tan' <I> - I

The pressure in the fluid is

In the region to the right of the x axis, figure 9,

v ~ (l - x) dduy + U cos. <I>

- p(lx - ! x 2) du + xU cos if> (52)

dy

l'd l'
p=p ( -~+-ucos )
3 dy 2

u~u,exp(-V3~)
Equating the pressure forces on the two sides of the membrane lying along the x
axis determines the value of the constant

I
c~
cosel> a a+2
2fftan'</> + cos</> - tan</> a + 2

ff ~ (~3 - c~<I
Thc pressure on the inclined wall of the dam is thus

p. ~ - "aoIl {(* - ~ (*) ') ta: </> (I - c ~~~' :) + c * cos <I> (53)
 DYNAMIO PRESSURES ON ACCELERATED FLUID CONTAINERS 31
where au is the horizontal acceleration of the inclined face. The resultant hori zontal
force exerted against the inclined faee is

F, ~ _ pa,lI' {_ a_ (I - c - -q,)q, + - cooq,

tan q,
cos
tan
c
2
} (54)

Equation (54) overestimates Ph. with a maximum error of 6 per cent at (jJ = 25.

FJ,EXIBLE RE'l'AIN"ING W ALL

Au approximate analysis ma.y be made of the effect of wall flexibility on 'w ater
pressures. Suppose the fluid is retained by a vertical cantilever wall which is stiff
to the degree that wave propagation in the wall may be neglected. As ShO\\Tll in
figure] 0, let the flu id be restrained by membranes whose shapes are proportional
to the deflected shape of the wall. For a sinusoidal vibration the horizontal velocity
of a fl uid particle is u fey) sin wt, and the vertical velocity is

v = ~: ih fCy) dy sin wt (55)

Applying Hamilton's Principle, there is obtained

(56)
A ~ .r: (f(y))' dy

The pressure on the wall is

p. ~ ""oW' ~ f.'f' fey) dy dy sin wt (57)

and the result.an t force on the wall is

p ~ ""ow' ~ f.'f.'f' fey) dy dy dy sin wt (58)

For a wall of uniform cross section, if we approximate thc actual pressure by

32 BULLETIN OF 'l'HE SlSMQU)GLCATJ SOCIk.'TY OF AMERICA

po sin (r./2) (y/h) and compute the deflected :shape of the wall, there is obtained

fey) ~ u. (1 - ~ + Il s in ~ ~)
(59)
p h'
il --u.
-
m'EI
x
f(~ )
PI! (/

if
"
Ut(~1
h

U.
Fig. 10.

where P is the total force exerted on the wall . 'Vith thisf(y) the pressure and force
arc computed to be

Pw ~ phu,,"' V3 ~: ~~::~~! ~:~~~: ( l - il) (*- t (*)') +(;) 'Il sm ~ *)

p ~ ph',",,"' /1 - 1.68il + 1.l 8~' (1 _ O.22{l) (60)
V3 -V 1 + 2.413 + 1.6313
The latter equation may be written

1 - 1.68{l + 1.18{l' (1 - O.22m

1+ 2.441l + 1.63il' Il

For a given w, that is, a given iV, this equation gives the appropriate value of /3.
Figure 11 gives a graph of N vs {l and also shows how Ihe lolal force on Ihe wall is
reduced by wall flexibility. For a rigid wall, EI = co, the preceding equation over-
estimates P by 6 per cent.
 DYNAMIC PRESSURES ON ACCELERATED FLUID CONTATNERS 33
10

9 1\
8
\
1 l\
V IO%.
'\
'\.

4 \ "'- r-....
\ fJN I'--- r--...
3

z. I~
~ -...
o
O.Z 0.4 0.8 1.0
~
F ig. 11.

NUMJ<iRICA.L ExAMPU:;

Consider a cylindrical tank of 40 ft. radius and 25 ft. depth of water. Suppose the
tank to be given a horizontal acceleration in the x direction of Uo = 0. 1(1 and let this
acceleration persist for 1/ 4 second and during the following 1/ 4 second let Uo =
- O.1g, so that the tank comes to rest with a total displacement of 0.2 ft. During
t his time the impulsive pressure on the wall of the tank is given by equation (31),
which for the present problem takes the form

P. 0(62.4)(0.1)(25) (*- t (*)) tanh ( va ~ cos )

From equation (34) the total impulsive force exerted on the tank is

- 40
tanh v325
P - 0.1 W _ 40 - 0.36 W - 280,000 Ibs.
v 3 25
34 BULLETIN OF' THE SEISMOLOGICAL SOCIETY OP AMERICA

The osciUations Bet up in the fluid a.rc determined from equations (37), etc.

g!27 !27 25
w' ~ 40'lis tanh 'lis 40 ~ 2.6
-2~ = 3.9 secon d d
speno 0 f VIb ratlOn
.
w
The force exerted by the fluid during the oscillations is the same as that exerted by
a simple oscillator (figure 5) I which has

M, ~ M 4"
j (11) '
12
/27 40
'lis !27 25
25 tanh 'lis 40 ~ 0.50 M

Since the natural period (3 ,9 sec.) is long as compared with the Y2 sec. duration of
the ground acceleratioIl, the net effect of the O.lg acceleration is to generate an
initial displacement of 0.2 ft. of M i relative to the tank \vall an d t he consequent
motion of M 1 is
Xl = 0.2 sin wt

The maximum Xl is 0.2 ft., and hence from equation (41) the maximum OJ. is

11
0, ~ 0.OG5 g 12
!27 40
'lis 1 !27
tan h 'lis
25 .
40 ~ O.OOS fad,.""

Th e pressu re on the wall at this (h. is given by equations (37) and (39). The total
force exerted on the wall is given by equation (40) .

p ~ ii (0.5 Mg)(O.OOS) sin wI ~ 33,000 sin wt

The amplitude of the water surface oscillations can be determined froth equations
(30) and (39) . The fluid pressure against the wall at y = hand <p = 0 is found fro~
equation (39) to be 171bs. per sq. ft., and from equation (30) the amplitude of the
water surface is
17
d~ ( .. ) ~ 0.2S ft.
p g - xU"

In summary, during the first U second the fluid exerts a force of 280,000 lbs.
against the tank wall, in the negative x direction ; during the next 7i second the
280,000 Ibs. force is in the + x direction; following this the only force against the
wall is the oscillating force the amplitude of which is 30,000 1bs. and the period 3.9
sec. During this oscillation the maximum amplitude of the water surface is 0.28 ft .
above the horizontal position.
 DYNAMIC PRESSUIlES ON ACCELERATED FLUID CONTA INERS 35

REFI<JRFJNCES

II . Jeffreys (1924), "Free Oscillations of \Vater in an Elliptical Lake," Pmc. London Math. Soc.,
VoL 23, 1924.
H . Lamb (1932), H ydrodynamics (Cambridge Univ. P ress, 1932).
H . M. Wcstf'..rgaard (933), "Water Pressures on Dams during Earthquakes," Tram. Am. Soc.
Civ. Eng., Vol. 98, 1933.
L . M. H oskins and L. S. J acobsen (1934), "Water Pressure in a Tank Caused by It Simulated
Earthquake, Bull. Seism. Soc. Am., Vol. 24, 1934.
L. S. Jacobsen (1949), "Impulsive Hydrodynamics or Fluid Inside a Cylindrical Tank l\nd of It
Fluid Surrounding a Cylindrical.Picr," BuU. Seism. Soc. Am., Vol. 39, 1949 .
.P. W. Werner and K. J. Sundquist (1949), "On Hydrodynamic Rar~hquake Effects," Tram. Am.
Geophys. Union, Vol. 30, 1949.
L. S. Jacobsen and R. S. Ayrc (195 J), "Hydrodynamic Experiments wit h Rigid Cylindrical Tanks
Subjected to Transiellt Motiolls," Bun. Seism. Soc. Am., Vol. 41, ]951.
E. W. Graham and A. M. Rodriguez (1952), "Characterist.ics of Fuel Motion Which Affect Air-
plane Dynamics," Jour. Applied Mechanics, Vol. 19, No. 3, 1952.
C. N. Zaogar (1953), "Hydrodyna.mic Pressures on Dams D ue to H orizontal Earthquakes," Proc.
Soc. Ezper. Stress Analysis, Vol. 10, No.2, ] 953.
G. W. H ousner (1954), Earthquake Pressures on Fluid Containers (California Institute of '1'cch~
nology, 1954).

DIVISION OF' ENGI NEF..RUrG,

CAl.Jf'ORNIA INS'J'J1'UTE OF TFCHNOT,OGY,
PASA.Dl:NA, CALU'.