You are on page 1of 35

Int. J. Pres. Ves.

& Piping 49 (1992) 61-95

Seismic Design Rules for Flat Bottom Cylindrical


Liquid Storage Tanks

N. J. I. A d a m s

18 Callas Rise, Wanborough, Swindon, Wiltshire, UK

(Received 6 April 1991; accepted 15 April 1991)

ABSTRACT
A set of design rules have been developed that permit a detailed
assessment to be made of large flat bottomed cylindrical bulk storage
tanks subject to seismic loading. It is proposed that these should form
an Appendix to the new British Standard BS7777. Equations and
appropriate coefficients are developed to assess the stresses in the tanks
shell due to impulsive and convective liquid mass effects, including
hoop and axial stresses.
Appropriate stress limits are proposed for both O B E and SSE events.
Lift-off of the bottom due to rocking is considered, including limits on
annular plate width and local peak foundation loads. A worked
example is included showing use of the equations that are proposed for
BS7777.

INTRODUCTION

The seismic response of storage tanks is a matter of considerable


importance in many parts of the world. Whilst the rules to be presented
are applicable to oil and refined product tanks, emphasis is placed on
refrigerated liquified gases, since the hazard associated with such a tank
failure is very much greater. From a general safety point of view, the
construction standards of storage tanks have advanced in recent years.
Many operators now use double or full containment tanks of the type
shown in Fig. 1. A detailed discourse on the general design of such
tanks is presented in E E M U A publication 147.1
61
Int. J. Pres. Ves. & Piping 0308-0161/91/$03.50 (~) 1991 Elsevier Science Publishers Ltd,
England. Printed in Northern Ireland
62 N. J. I. Adams

Roof if required

Extend
insulation
"--~
[barrier
External weather

Base -~i'---'--Outer tank shell


insulatic

_~Bottom heater

Outer stee, ~ ~ ~ U ~ u ~ ~~oof

Insulation on inside of
insulation or empty/-- "~ I=--Inner tank l~i
depending on 7 " - - ~ ' l JFJI -.'-'~Earth
embankment
\

\
\ • Bottom and wall
heater
Fig. 1. Examples of double and full containment tanks.

Depending on local conditions, tanks may be constructed on a rock


foundation, a base ring and mat over soil, or a piled foundation. In all
cases the inner tank bottom will sit on a thick layer of low density
insulating material. Where a tank is sited directly on a rock foundation,
the free surface excitation can be considered as directly imposed on the
tank base. However, when the tank is sited on soil or a piled
foundation, it is necessary to examine the soil/structure/tank response.
As an example, Fig. 2 shows a free field response spectra, whilst the
modified response spectra for tank base of a piled design is shown in
Fig. 3.
In what follows, the background is presented to rules proposed for
adoption in BS7777 for designing storage tanks to resist seismic loading.
The tank is considered to be flexible, since this has a significant effect
Seismic design rules for liquid storage tanks 63

1.o I I I ill I I I I I I I it I t I
~., 5% Damping
- 1000 years/
-- . . . . . . . . . . .
years/A
0.5 " / / "Ductility level event
o
E f~'~ Strength level event
.9
e.J
t.

U 0-2
U

I-

) o11 Recommended p r e l i m i n a r y design criteria


response spectra f o r events shown

0.051 I I I I I I I I I i ! Ill I I I
• =1.,
0.04 0.1 0-2 0-5 10 2.0 5'0 10
Response period, s

Fig. 2. Earthquake response spectra preliminary design criteria.

1"0

~) o.1
)

0"010.1 1.0 10
Frequency

Fig. 3. Design response spectra acting on tank base.


64 N. J. I. Adams

on the response period and acceleration of the liquid mass. Also,


acceptance criteria are given in terms of an operating based
earthquake, which is the event that the structure is most likely to
experience in its working life and for which damage is not intended to
occur. There is very little modern recorded data on strong ground
motions, for return periods associated with storage tank design life-
times, generally 30 years. Depending on the assumed probability of
occurrence, periods of typically 300 years or 3000 years may be used to
establish the magnitude of an event. Thus, it is considered that great
accuracy in the calculations is not of significant value. It is much more
important to have an appreciation of what may happen and recognise
the outcome, so that sound engineering principles and judgement can
be applied to the design.
The information presented is drawn from a number of publications. 2-1°
Certain additional practical aspects are introduced to arrive at what are
believed to be a satisfactory set of guidelines.

Description of problem

When the base of a tank containing liquid is subjected to seismic


excitation, the pressures exerted by the liquid on the tank shell and
foundation change in magnitude and distribution from those cor-
responding to a state of static equilibrium. These hydrodynamic
pressures result in increments in liquid pressure which vary with time
and thus induce time-dependent stresses in the tank which may
significantly affect its performance.
These pressures and stresses depend on the characteristics of the
ground motion, the properties of the contained liquids, and the
configuration and physical properties of the tank itself including the
foundation design features such as piling.
For a horizontally excited tank where the liquid surface is free, a
portion of the liquid along the walls and the bottom, up to some
prescribed depth, moves in unison with the tank as a rigidly attached
mass (impulsive), while the remaining volume moves independently
(convective), experiencing sloshing or rocking oscillations about a
horizontal axis normal to the direction of the excitation.
The proportions of the liquid mass that participate in the two types of
action depend on the configuration and dimension of the tank. For
upright, shallow, broad cylindrical tanks, the impulsive or rigidly
attached mass is generally the smaller of the two, and the convective
sloshing component is the larger. The hydrodynamic pressures as-
sociated with the two types of motion are distributed differently over
Seismic design rulesfor liquid storage tanks 65

the tank height and have different transient variations. The pressures
induced by the impulsive component are maximum near the tank base
and are associated with high frequency excitation, whereas those
induced by the convection component are maximum at the liquid
surface and are associated with low frequency excitation.

Objectives and scope

The purpose of the following sections is to provide a background to the


simplified set of equations to be applied when designing a storage tank
to BS7777. The references given provide a more detailed description of
the evolution of theories relating to the response of storage tanks
subject to seismic excitation.
The response quantities that should be examined include the hydro-
dynamic pressures induced by the liquid on the tank shell; the
associated tank forces, particularly the maximum values of the base
shear and overturning base moments; and the maximum vertical
displacement or sloshing height of the liquid surface. Two cases are
considered: the operating based earthquake (OBE) and the safe
shutdown earthquake (SSE).
In the first case, the tank is designed not to sustain any damage and
will be perfectly safe to continue in operation, although external visual
examination would be recommended.
For the case of the SSE, it is expected that damage would occur, but
without the loss of integrity. Permanent deformation and liquid in the
annular space may be anticipated for open top tanks. The tank would
not remain in continuous service without a detailed examination, which
would be expected to include entry into the tank.

Design considerations

This section summarises the more important considerations involved in


the design of liquid storage tanks for earthquakes.

Design forces and criteria


The maximum values of the dynamic forces induced in the tank by the
horizontal and vertical components of ground motion should be
combined, by the square root of the sum of the squares rule, and
superimposed on those corresponding to the hydrostatic pressure. The
stresses resulting from these total forces must be compared to the
allowable stresses.
66 N.J. L Adams

Contributing to the maximum value of the dynamic hoop force are:


(1) the hydrodynamic effects induced by the lateral and vertical
components of ground motion, and
(2) the inertia effects of the tank and the roof due to the horizontal
components of excitation.
Contributing to the maximum value of the vertical force are:
(1) the hydrodynamic effects due to the lateral component of
ground motion, and
(2) the structural inertia effects due to both the lateral and vertical
components of excitation. Except for tanks with unusually large
roof loadings, the structural inertia effects are generally small
and may be neglected.
For the OBE the maximum hoop force should be limited so as not to
cause undesirable yielding in the tank shell, and similarly the maximum
axial compressive force should not cause shell buckling. Additionally,
the shell should be stable against overturning, and there should be no
unacceptable overstressing of the supporting foundation, e.g. base
insulation.

Circumferential stress
In assessing the effect of the maximum circumferential hoop force,
provision should be made for the ability of the tank material to
withstand safely, transient excursions into the inelastic range of
deformation, particularly when the maximum forces are of a localised
nature. The shell of a welded steel tank has substantial ductility, and
this capacity should not be ignored in assessing design.

Allowable axial compressive stress


There is no general agreement regarding the magnitude of the
compressive axial stress that induces buckling in the shell of a tank
subjected to a seismic motion. This is one of the important topics
requiring additional research.
Classical theory predicts under static loading that instability will
occur when the axial compressive stress reaches
0.6Et
R
However, experiments have shown that the onset of buckling under
static loading occurs, in practice, at much lower stresses. Factors which
influence this are actual boundary conditions, eccentricity of loading,
Seismic design rules for liquid storage tanks 67

and imperfections in shell contour. In contrast, it has been found that


the actual buckling stress under dynamic loading conditions increases
above the static value.
The question of buckling in a tank shell was first addressed in API
650 Appendix E in Ref. 11, the proposal being that the axial stress is
limited to
83 000t
iT=
D
It is proposed that this limit be adopted for the OBE, since the original
intent was that a tank should not be damaged.
For the case of the SSE it is proposed that the limit is raised to the
lesser of
0.4Et
R
or the normal static design stress. The reasons for this proposal are
that:
(i) the dynamic action of the liquid will reduce any initial
imperfections in the tank shell;
(ii) the impulsive and convective forces are summed as SRSS;
(iii) buckling observed in tanks that experienced seismic loading
was notably plastic, wrinkling in the shell to annular region.
Elastic buckling from which the shell could recover would not
represent a serious problem.

Increased axial compression and overturning resistance


for unanchored tanks
As a result of horizontal excitation, the overturning moment can give
rise to uplift of the tank base.
The primary consequences of base uplift are a redistribution of the
forces at the foundation-base interface and an increase in the maximum
axial compression in the tank shell. Fundamental to these proposals is
the assumption that the maximum overturning moment for the tank is
independent of the degree of uplift developed, as shown in Fig. 4. In
reality, base uplift increases the flexibility of the tank and hence the
effective period of vibration of the tank-liquid system. This increase
may significantly decrease acceleration and hence the resulting mo-
ment. The maximum axial compression developed in the tank shell may
still increase, however, because of the reduced section available to
resist the moment.
~J.I.A~

~ M t A x

v i
J

IWs W~ ~ R --r

_ kR

e~
• cresent

Uplift and restoring conditions.

As part of the tank base lifts off the foundation, the weight of the
liquid above the raised portion is transferred by the tank shell to the
portion of the shell that remains in contact with the foundation, and
there it is reacted by an increased compressive force. At the same time
the weight of the liquid over the remaining region of the base plate is
naturally reacted directly by the foundation.

Tank foundation
The stability of the foundation should be evaluated for the combination
of the static and maximum dynamic forces transmitted by the tank base.
Due provision should be made for the fact that a high proportion of
these forces are transmitted to the foundation as concentrated line
loads through the tank shell. In particular, the local crushing force on
the low density insulation, immediately below the tank bottom, should
be checked.

Freeboard requirements
Seismic excitation leads to sloshing of the liquid, and it is necessary to
ensure sufficient freeboard exists in the tank. For a single containment
Seismic design rules for liquid storage tanks 69

tank without a suspended deck, it is clearly essential to ensure that


wave loading on the roof is acceptable. For double or full containment
tanks, with suspended decks, fluid should not reach the annular space
during an operating base earthquake.
A number of proposals are to be found for predicting wave slosh
height. Based on Housner's 3 original analysis Wozniak and MitchelP °
proposed

d = 1.24ZIC2 T 2 tanh ( 4 . 7 7 H / D ) (1)

The JapanesC 2 and New Zealand 13 Codes contain alternative proposals;


the latter being a modification based on the work of Kana. 6 From
experiments using transparent cylinders on a shaking table, it was
demonstrated that the wave height could be most accurately computed
from:

d = ( E E2)1,2 (2)

where

[2
en = ~ 3.. tanh (3..H/R) X .
] (3)

and the spectral displacement X . = a./to 2, or can be obtained from the


design response spectrum, and 3,. is the corresponding eigenvalue for a
given mode.

Seismic events

As previously stated, there is very little recorded evidence of strong


ground motions, even in highly developed countries. Figure 2 shows the
smoothed response spectra proposed in API RP2A normalised in
relation to 200 and 1000 year return periods for a site in south-east
Asia. By taking into consideration the soil properties, the tank
dimensions, flexibility and fluid effects, it is possible to calculate the
tank base acceleration, as shown in Fig. 3. Such a response spectra for
the tank base then forms the basis for the dynamic analysis and design
of the storage tank system.
A method by which the soil/structure interaction can be analysed is
described in a later section.
70 N. J. I. Adams

Modelling of tank-liquid system

For the purpose of evaluating the base shear and bending moment
induced in the tank by the hydrodynamic wall pressure, the tank-liquid
system may be modelled in the manner shown in Fig. 5. In this model,
the impulsive mass, mi, is attached rigidly to the tank at a distance, hi
from the base, and the convective masses, mj, are attached through
pairs of horizontal flexible springs, located at distances hj from the
base. The stiffness values of the springs, kj, are selected such that the
natural frequency of the jth mass is the same as jth sloshing frequency
of the liquid.

Fundamental frequency of tank-liquid system

Early research into the behaviour of tanks, subject to seismic excita-


tion, considered the tank shell as rigid. This is clearly not the case, and
whilst the rigid assumption has no marked implications for the sloshing
motion, it has been shown that it can have a significant effect on the
forces resulting from motion of the impulsive mass. It is therefore
proposed that the design shall be based on consideration of the effects
of flexibility and resulting tank-liquid interaction.
The implications of this decision are that the accelerations, applied to
the impulsive mass, become those associated with the tank-liquid
system frequencies, rather than those associated with the tank base.
These may be equal to, greater than, or smaller than the maximum
base acceleration.
The fundamental natural frequency, )Co of the tank-liquid system,
may be conveniently expressed in the form

fo = 2~----H (4)

kn kn

kc : kc

hn
mi . )

Fig. 5. Tank mass-spring model.


Seismic design rules for liquid storage tanks 71

in which E is Young's modulus of elasticity for the material of the tank


shell; Pm is its mass density; H is the depth or height of the liquid; and
C~ is a dimensionless coefficient that depends on the tank proportions,
H / R and t / R , Poisson's ratio for the tank material, and the relative
mass densities of the liquid and shell material P J P m .
/ . - - - -

c, = c . (5)

where p,, = the mass density of water, and p~ = the mass density of the
liquid under consideration and Cw can be determined from Table 1.
Equation (5) is based on the assumption that the mass of the tank itself
is negligible in comparison to that of the liquid; a condition normally
satisfied in practice.
Alternatively, Sakae et at. 14 proposed the following simple formula
for the period of the fundamental mode of a coupled liquid-tank
system.
w

)Co = X 800~Etl/3
where
A= 0.067 ( H / D ) 2 - 0.3 H / D + 0.46
w= total weight of fluid (kgf)
E= Young's Modulus of elasticity (kgf/mm 2)
D = tank diameter (mm)
tl/3 = shell thickness at xaheight from the bottom (mm)

TABLE 1
Frequency coefficient, Cw, in expression for fundamental
natural frequency, f0, of tanks full with water; v = 0.3,
Pl/P = 0.127
H/R Value of Cw

t/R = 0.0005 t/R = 0.001


0.5 0.050 6 0.071 9
0"6 0.053 8 0.076 3
0.7 0.056 4 0.079 9
0.8 0.058 7 0-082 9
0.9 0-060 5 0.085 4
1-0 0.062 0 0-087 5

1-2 0.064 1 0-090 3


1.4 0.065 0 0-091 5
1.6 0-065 1 0.091 7
72 N. J. L Adams

Hydrodynamic wall pressure

The peak values of the pressure at an arbitrary elevation and time occur
along a diametral axis in the direction of the excitation, and the zero
values occur along a normal axis. From an examination of the
expressions presented subsequently, it can be further deduced that the
impulsive pressure increases from zero at the liquid surface to a
maximum at the base, whereas the convective pressure is m a x i m u m at
the liquid surface and decreases with depth.

Impulsive pressure component


In as much as a rigid container experiences the same motion as the
ground, the impulsive pressure c o m p o n e n t may be visualised as being
due to part of the liquid vibrating synchronously with the tank wall as a
rigidly attached mass.
For tanks of very small H / R ratios,

pi(z, O, t) = c~(z)x(t)p,H cos 0 (6)

The impulsive effects for the flexible tank are obtained by replacing the
base acceleration with the pseudoacceleration function Ai(t), which
corresponds to the fundamental natural frequency of the tank-liquid
system. Thus

pi(z, O, t) = c~(z)Ai(t)plH cos 0 (7)


and the maximum impulsive base pressure is

p~ = C~o(O)A,p,H (8)

where c~(0) is given in Table 2.

Convective pressure component


The part of the liquid that does not move as a rigid body, with the tank,
experiences a sloshing or rocking motion about a diametral axis normal
to the direction of excitation. This motion can be expressed as a linear
combination of the corresponding natural modes of vibration of the
liquid.
There are an infinite n u m b e r of such modes, each associated with
vertical displacements that are antisymmetric about the axis of rotation
and which vary in the circumferential direction as cos 0. Figure 6 shows
the radial variations of these displacements for the first three modes of
vibration, normalised such that the ordinate of each curve is unity along
the tank wall.
Seismic design rules for liquid storage tanks 73

TABLE 2
Base pressure and m o m e n t coefficients, effective mass of liquid, and associated heights
for impulsive effects in rigid tanks

H/R Co(0) mi/ml hi/H Ahi/H hl/H c1(0)


0.5 0.414 0-300 0.400 1-060 1.460 0.828
0.6 0-494 0-359 0-400 0.792 1.192 0.823
0.7 0-568 0.414 0.400 0.609 1-009 0.814
0.8 0-633 0-464 0.401 0-480 0-881 0.791
0.9 0-691 0-508 0.403 0"387 0.789 0-770
1.0 0-740 0-548 0.404 0-317 0.710 0.740

1.1 0.782 0.583 0.406 0-264 0-669 0.681


1.2 0.818 0-614 0.407 0.222 0-630 0-652
1.3 0.848 0.641 0.409 0.190 0.598 0.646
1-4 0.873 0.665 0.411 0-163 0.574 0.623
1.5 0.894 0.686 0.413 0-142 0.555 0-596

The convective pressure variation with depth decreases much more


rapidly with depth for tall, slender tanks than for the shallow, broad
tanks typical to the proposed Code. This is particularly true of the
pressure components associated with the second mode of vibration.
Furthermore, the magnitude of the pressures for the higher modes are
only a small fraction of those for the fundamental mode.
The distribution of the convective pressure is given by
oo

pc(z, O, t)= ~, [ci(z)Aj(t)]plR cos 0


i=1

F i r s t mode

Second mode

Third mode
Fig. 6. Radial distributions of fluids vertical dis . . . . . . . . . . . ,
placements and pressures for first three sloshing o~8 0.4 o o,4 o.a
modes, r/ R I r/ R
74 N. J. L Adams

In order to determine the value Aj(t) it is necessary to determine the


sloshing mode frequencies.
The jth natural frequency, fj, of the sloshing mode, in cycles per
second, or Hertz, is determined from

fj = ~ ~.jg tanh

in which g is the acceleration due to gravity, and ~. has the values given
by
).1 = 1.841 Z2 = 5.331 Z3 = 8.536
Since the response of long period systems is typically characterised
by long period oscillations, the pseudoacceleration functions Aj(t )
associated with them, and hence the convective component of the
hydrodynamic wall pressure, Pc, will in general be dominated by
oscillations of a much longer period than those of the ground
acceleration, x(t), to which the impulsive pressure component is
proportional in a rigid tank. The maximum numerical value of Aj(t) is
denoted by Aj. This value may be determined from the pseudoaccelera-
tion response spectrum for the prescribed excitation using the fre-
quency fj and an associated damping factor typicaly 0.5% for a liquid.
It is proposed that for design purposes, only the first convective
component be considered, hence
Pc = cl(O)Aco)ptR
where c~(0) is given in Table 3 and Ac(~) is the pseudoacceleration
associated with fl, the fundamental sloshing mode.

M a x i m u m value o f total pressure


The hydrodynamic wall pressure at an arbitrary point may be obtained
by taking the square root of the sum of the squares of maximum
numerical value of the impulsive pressure plus the maximum modal
contributions of the convective pressure if more than one term is
considered.
However, for design purposes it is proposed that, this is reduced to
2 2__
Pmax = piV~--pc -- X/(cl(O)Aip,H) 2 + (c,(O)A¢(1)P,R) 2
That is to say, for design purposes, an evaluation of the state of stress
throughout the shell is not necessary, and it will be adequate to
evaluate only the maximum values of the circumferential, axial and
shearing stresses in the tank shell.
Seismic design rules for liquid storage tanks 75

TABLE 3
Base pressures and moment coefficients, effective mass of liquid, and associated heights
for convective effects in rigid tanks

H/R c,(O) m e ~ m , m2/ml hJH AhJH h2/H Ah2/H


0.5 0-575 0.660 0-027 0.533 1.028 0-674 0.524 0
0.6 0.500 0.608 0.023 0-545 0-674 0.712 0"025 6
0.7 0.429 0-558 0.020 0.559 0.463 0.745 0.012 8
0.8 0.365 0.511 0-017 0.574 0.328 0-772 0-006 6
0.9 0.308 0.470 0-015 0.590 0.239 0-795 0-003 4
1.0 0.259 0.432 0-014 0.606 0-177 0-814 0.001 8

1.1 0.217 0.399 0-012 0.621 0.133 0.830 0.001 0


1.2 0.182 0.370 0-011 0-637 0.101 0.844 0-0005
1-3 0.152 0.344 0-011 0.652 0.077 0.856 0-000 3
1.4 0.126 0.321 0.010 0-667 0"059 0-866 0.000 2
1.5 0.105 0.306 0.009 0-681 0.046 0-875 0-000 1

The total hydrodynamic pressures referred to are in excess of the


hydrostatic pressure, and must be added to the latter to obtain the total
pressure exerted by the liquid.

Tank forces and stresses

With the maximum values of the hydrodynamic shell pressures deter-


mined, the corresponding tank forces may be computed.

Hydrodynamic base shear


The instantaneous value of the base shear Q(t) can be obtained by
integrating the horizontal components of the impulsive and convective
pressures, over the height of the tank.
Considering only the first sloshing mode, this reduces to
Q(t) = miAi(t) + mcA~(t)
where mi/ml and m~/m~ are given in Tables 2 and 3, respectively.
Adopting the same principle used for determining the maximum
pressure, and taking the SRSS
Qmax ----[(miAi) z + (mcAc~l))2l1~

Hydrodynamic base moments


A clear distinction must be made between the hydrodynamic moment,
Mt(t), induced on a section of the tank immediately above the base,
and the moment, MS(t), induced on the foundation itself. The former is
76 N.J. L Adams

due to the pressures exerted on the tank shell, whereas the latter also
incorporates the contribution of the pressures on the tank base.
The base m o m e n t can be conveniently expressed as
Mt(t) = mihiAi(t) + m~h~Ac(t)
and
M'(t) = mihIAi(t) + m~h~A¢(t)
where h 1 = hi + Ahi and Ahi represents a contribution to the foundation
m o m e n t due to pressure exerted on the base.
The maximum values of the m o m e n t s reduce to
Mtax = ( (mihiAi) 2 + (mch~Ac(1))2} 1/2
and
M~max {(mih 1Ai)2 + (mchcAco)) 2} 1/2
=

Having established the m o m e n t at the bottom of the tank, the


maximum value of the axial compressive stress is calculated from

t
1
Ozmax= Mmax • - -
~rtR 2

Foundation toe pressure


The stability of the foundation insulation should be evaluated for the
combination of static and maximum dynamic forces induced, taking
cognizance of the fact that large proportions of these forces are
transmitted through the tank shell as almost concentrated line loads.
The forces so transmitted are those due to the dead weight and inertia
of the tank itself (which are generally negligibly small for an open top
tank) and the effects of the hydrodynamic wall pressures. In contrast,
the dead weigth of the liquid and the tank base, plus the hydrodynamic
base pressures are transmitted to the supporting m e d i u m directly as
distributed loads.
D e n h a m 15 demonstrated that when a tank is full of liquid, the shell
bulges and some rotation at the shell-annular joint occurs. This results
in lift-off of the annular plate for some distance inside the shell. Unless
a plastic hinge were to form this effect will be magnified by the
increased pressure due to seismic motion. Thus, at its worst, the
foundation pressure will only be reacted by the projection of the
annular plate beyond the shell.
The maximum vertical pressure at the tank annular plate depends
importantly on the stiffness of the foundation. However, the stiffness of
insulating concrete beneath a refrigerated storage tank is low compared
Seismic design rules for liquid storage tanks 77

to that of the steel annular plate projection. This will result in a fairly
uniform re-distribution of the shell axial stress into the foundation. On
this basis, it is proposed that the foundation pressure is given by
t
pf = O'zmax -
a

where a is the annular plate projection beyond the shell.

Surface displacement of liquid

The maximum vertical displacement of the liquid surface is required to


determine the freeboard that must be provided to avoid spillage of the
tank contents and to prevent the liquid from impacting the roof.
Considering the first two sloshing modes and taking the SRSS, eqn
(3) becomes:
d = [{d11.54 tanh ( 1 . 8 4 H / R ) } 2 + {d20.39 tanh (5.33H/R)}2] 1/2
where dl and d2 are the tank base horizontal displacements for the first
and second sloshing frequencies.

Effects of non-uniform wall thickness

The fundamental natural frequency of tanks of non-uniform wall


thickness may be determined approximately from the information
presented in the preceding section by replacing the variable wall
thickness by an average value. For this it is proposed that the thickness
is averaged only over the height hi.

Effects of vertical component of ground shaking

The vertical component of the ground acceleration, induces a hydro-


dynamic wall pressure, Pv, in addition to that induced by the horizontal
component. The pressure Pv is uniformly distributed in the circum-
ferential direction and varies in the vertical direction as well as with
respect to time.
Tank flexibility affects the magnitude, spatial and temporal variations
of the hydrodynamic pressure in essentially the same manner as for a
laterally excited tank.
The pressure pv under this assumption increases linearly from top to
bottom, and its magnitude at the base is given by
p~(z = O) = ptHA~(t)
78 N. J. I. Adams

in which Av(t) is the pseudoacceleration corresponding to the fun-


damental natural frequency of the axisymmetric, breathing mode of
vibration of the tank-liquid system, fv which is given by

fv-2~ H
where the values of Cv are as given in Table 4.
It is widely recognised that whilst seismic motion occurs in both the
horizontal and vertical direction, the vertical component is of smaller
magnitude. It is therefore proposed, in line with widely held views, that
the vertical components of motion be taken as ~ of the horizontal
values. Thus
p~ = 0"66 pIHAv
where Av is the pseudoacceleration associated with the first breathing
mode of tank vibration.
In considering the interaction of the impulsive and convective
components in the horizontal direction, it is believed to be unduly
pessimistic to sum the components directly. However, in recognition
that the resulting actions can combine, the SRSS is adopted.
When evaluating the combined effects of the horizontal and vertical
motion, it is necessary to consider the probability that the worst case
vertical and horizontal effects could be coincident.
Assessing the combined effects of vertical and horizontal components
on hoop stress in the shell and foundation pressure, it is unlikely that
they will have a significant effect on the overall design. However, when
equating the restoring and overturning moment or resistance to
sideways movement and transverse force, the vertical acceleration
could have a significant effect on design, if the peak vertical and
horizontal acceleration are taken as coincident.

TABLE 4
Coefficient Cv in expression
for fundamental frequency of
axisymmetric, breathing mode
of vibration of full tanks, v =
0.3, t/R = 0-001, PI[P = 0.127

H/R Cv
0-5 0.070 5
0.75 0.080 8
1.0 0-086 8
1-5 0-092 5
Seismic design rules for liquid storage tanks 79

Since there is already an established practice in the US Uniform


Building Code, it is proposed that this should be adopted. However, it
should be borne in mind that the building code has been developed
with a view to protecting human life and therefore the philosophy may
not be totally relevant to designing storage tanks. Clause 2312(h) of
UBC-88, states that for orthogonally uncoupled events, the combina-
tions shall be taken as 100% of one component plus 30% of the other
component. Thus combining the horizontal and vertical components the
tank should be checked for

100%H + 30%V Av(e) = 0"2H (9a)

30%H + 100%V Av(e) = 0.666H (9b)

where Av(o)is the effective vertical acceleration to be taken into account


when determining the resistance to sideways movement or the over-
turning resistance of the tank.
It should be noted that in eqn (9b) the horizontal shear force Qm~
and the overturning moment M t ~ should be adjusted to reflect the
reduced horizontal acceleration.
When addressing sideways movement of the tank, the frictional
resistance will be enhanced by the fact that the tank bottom is made up
of overlapping plates. These plates bed into the soft sand or Bitsand,
forming a series of shear steps. Given the added adhesion between the
bottom and the foundation, it does not seem unreasonable to take an
effective friction factor of 0-6 from which

Rh = 0"6Av¢o)wf> Qma~

where Rh is the resistance to horizontal movement of the tank and we is


the weight of the fluid supported by the foundation.
Similarly, the restoring moment for the tank should be adjusted to
account for the vertical acceleration, that is

MR(v) = MRA~(~)
In Ref. 13 it is specifically stated that since the interaction between
the horizontal and vertical components cannot be predicted, the effect
of the vertical component should be ignored when determining the
restoring moment. This seems to be an arbitrary decision, thus it is
proposed here that the effect be considered, but the final accountability
should be between the purchaser and tank vendor depending on the
implications for the overall design.
80 N . J . L Adams

Tank bottom uplift

The horizontal seismic motion leads to an overturning m o m e n t on the


tank, that could result in an area of the tank b o t t o m plates lifting off
the foundation. Clearly, for an anchored tank, the overturning m o m e n t
imposes tensile loads on the anchor straps, which will be reacted by the
shell.
For an anchored tank, it will be necessary to design the anchor straps
to withstand the overturning moment. In certain tank design options, it
will be necessary to sum this effect with the load produced by internal
pressure in the tank. For double or full containment tanks, there is
much to be said for not anchoring down the inner tank, when it is of the
open top design. The logic behind this is that the anchor straps must
pass through the bottom of the outer tank to be secured in the concrete
ring beam or basemat. Since there will be many straps, it must call into
doubt the ability to really achieve the desired level of double or full
containment through the outer tank bottom.
For an unanchored tank, it will be necessary to check for the extent
of lift-off. This was originally addressed 1° by considering the formation
of two plastic hinges. The mechanism of uplift is extremely complex
and not completely understood, but it involves large elastic displace-
ments, yielding at the annular-shell junction, m e m b r a n e forces in the
bottom plate and ovalling of the shell, at the very least. What follows
describes a modified version of a m e t h o d proposed by Clough 8 and
discussed in Ref. 13.
The overturning m o m e n t will be resisted by the action of the weight
of liquid which remains directly supported by the foundation and the
compression reaction at the shell-annular junction. Thus, the restoring
moment

MR = wf(R - r) + w~kR

where

w = total weight of fluid


wf = that portion of w supported by the tank foundation
ws = weight of shell and roof if applicable
Wr : W "~ W s - - W f
k R = distance from centre of compression reaction to tank centreline
0* = half angle which defines the arc of the shell b o t t o m in contact
with the foundation

See Fig. 4 for details.


Seismic design rules for liquid storage tanks 81

In essence, the analysis assumes that a crescent is formed during


lift-off, and the area remaining in contact with the foundation has a
radius r. In Ref. 13 it is suggested that MR is calculated in an iterative
manner by
(i) Assume a value of/~ = r/R
(ii) Calculate 0* = arctan (~/(1 - / u ) )
(iii) Calculate k = (2/0 .2) (1 - cos 0"), from which
(iv) MR ----gw{k(1 + ws/w) + (1 - k)/t/2 -- p~3}
The procedure is repeated by modifying/~, until the restoring moment
M R - Mm~ the overturning moment.
__ t

Alternatively, one can assume that the tank shell is rigid, in that it
retains its circular shape, and that the bottom membrane is fully elastic
and offers no resistance to lift-off. Under the circumstances a circular
area, of radius r, of the base remains in contact with the foundation. In
this case it can be shown that for large values of r/R, k can be taken as
approximately equal to 0-4R.
In reality the shell flexibility will result in a non-uniform lift-off
resulting in a larger value of k. Thus use of the above simple criteria
will give rise to a conservative estimate of the restoring moment.
However, given the construction detail of the annular/sketch plate
weld attachment, it is proposed that lift-off should not be permitted to
exceed the width of the annular plate. In this case, r/R is defined and
MR c a n be calculated directly, and should exceed Mt~x. If M R is less
than Mmax
t the tank dimensions should be modified or the width of the
annular plate increased, until the equality is achieved. Assuming that
permitted lift-off is only a small percentage of the tank radius an
average value of k can be determined to simplify the calculation
procedure.
It has been proposed 13 that an estimate of the magnitude of shell
uplift can be obtained from a modified form of an equation derived by
Cambra, 16 where it has been assumed that the shell-annular joint weld
will yield, thus

Vs=F(6-~'fybt2+Nx'Ps hi--~ - 12(~-_v2)Nj/


where
fyb is the yield strength of the annular plate material,
Nx =L~t~
Ps = hydrostatic pressure
Lb = 2R(1 - / z )
and F is the foundation stiffness factor.
82 N. J. L Adams

The bottom plate radial membrane stress frb is given by


1 2E \ 1/3
frb = ~b (3(1 "-~'V2) tbp2R2(1 ~[.~)2)
- -

Clearly, the extreme values of vs and f~ are those associated with the
limitation that uplift shall not extend beyond the width of the annular
plate.
From an examination of Fig. 7, it can be seen that the angle of
rotation of the annular plate 0p relative to the line of the shell bottom,
is given approximately by
2vs vs
0p = Lb 2R
As the tank bottom lifts off, the development of a membrane state of
stress in the annular plate is associated with the formation of a plastic
hinge in the annular plate at the shell-annular junction. The problem
arises as to what limiting value shall be applied to this rotation. In Ref.
13 it is proposed that the maximum strain be limited to 5% and that the
plastic hinge is limited to a length of 2tb from which 0p shall have a
maximum value of rotation equal to 0.2 radians. It has to be recognised
that this is very much an engineering judgement based on scant
information. Therefore, no proposals will be made on such a limit, but
it shall be left to the designer to form a judgement on a case by case
basis, taking into account the materials of construction, details of the
shell-annular junction weld and relative plate thicknesses.

Effects o f soil/structure interaction

Fundamental to the analysis procedures presented so far, are assump-


tions about the motion experienced by the base of the tank. Seismic

I_ Lb _
01~VlLb
Fi~° 7. Rotation of the bottom during uplift.
Seismic design rules for liquid storage tanks 83

data is frequently presented as the free-field motion of the ground,


which refers to the ground motion experienced at the level of the
foundation, if no structure were present. Strictly speaking, this assump-
tion is only valid for a tank supported directly on rock or other rigid
foundation materials.
For tanks supported on softer soils, the tank foundation motion may
be significantly different from the free-field ground motion and may
include a rocking component in addition to a lateral or translational
component.
Two factors are responsible for the difference in the response of a
rigidly supported tank and an elastically supported identical tank. First,
the elastically supported tank has more degrees of freedom than the
rigidly supported tank and hence different natural frequencies and
modes of vibration. Second, a substantial part of the vibrational energy
of the elastically supported tank may be dissipated into the supporting
medium by radiation of waves and by hysteretic action in the soil itself.
In most cases the free field motion will be greater than the tank
motions, but this cannot be fully assured and is sensitive to soil/geology
conditions.
With the advances that have been made in geotechnical engineering
and computational methods, a realistic approach can be adopted to
determine tank base excitation, which can also incorporate the effects
of tank base piling in deep soft soils. To this end, it is proposed that the
following may be one suitable option:

(1) Determine the free-field response spectra for the site.


(2) Define a suitable tank configuration. The estimation should
include inner and outer tank radii, height of liquid in the inner
tank for earthquake analysis and overall height of outer tank.
Plate thicknesses should be selected for mass and stiffness
calculations. The radius and thickness of the basemat should be
determined.
(3) Determine a suitable foundation scheme which is compatible
with the tank configuration (2).
(4) Determine the dynamic horizontal, vertical and rotational
stiffness of the soil/pile system. It should be assumed that the
stiffness calculated will be dependent on both the level of
excitation and frequency. Due to the variability in soil pro-
perties, it is probable that both upper and lower bound soil
properties should be considered.
(5) Determine the damping (viscous and radiation) of the soil/pile
basemat system for horizontal, vertical and rotational move-
ments. As in (4), the damping value will be dependent on both
~J.I.A~

the level of excitation and frequency. Again, the magnitude of


damping value should be bounded with confidence limits.
(6) The sloshing mass and frequency of the product should be
determined and its effective height above the basemat.
(7) For horizontal excitations the frequency of the inner tank flex
including the effective mass of the product should be
determined.
(8) A computer model which represents the mass and stiffness, of
the soil/pile/structure/product, under horizontal and rotational
excitations should be generated. The model need only in-
corporate such significant structural details that contribute to
the overall dynamic response. The soil/pile/structure interac-
tion can be modelled using horizontal and rotational springs to
ground.
(9) A similar model to (8) should be generated to represent the
response of the soil/pile/structure/product under vertical
excitation.
(10) A modal analysis should be carried out for both models (8) and
(9) to determine the frequencies of the major structural modes.
If the frequency of the dominant mode is different to what has
been assumed to determine the foundation spring stiffnesses,
the spring stiffness should be modified and the modal analysis
re-run.
(11) When satisfactory convergence is achieved in (1) the normal-
ised mode shapes should be used to calculate the percentage
damping for each mode. This can be carried out by weighting
the percentage damping for each component part of the
structure (or foundation spring) by the proportion of strain
energy in the component compared with the total strain energy
in the mode.
(12) Participation factors for each mode should be calculated to
determine the significant mode.
(13) Using the damping values calculated in (11), acceleration,
velocity and displacement spectral values should be calculated.
(14) Finally, the forces and moments in the soil springs should be
compared with the assumptions in (4) and (5), the stiffness and
damping values adjusted, if necessary, and the analysis for (8)
re-run.

Once the tank base response spectra has been established, the forces
and moments can be determined in the manner previously described.
Seismic design rules for liquid storage tanks 85

The outer tank

In systems which use an outer tank, whether for insulation contain-


ment, or double or full containment design philosophy, consideration
needs to be given to the design of the outer tank and associated fittings.
The outer tank is not assumed to contain any liquid, thus it can be
analysed by normal structural methods. If the tank has been designed
to contain the vapour pressure of the gas, it will incorporate holding
down straps. These should be checked for adequacy under combined
pressure and overturning moment.
Where the design incorporates an access tower to the tank top,
together with inlet/outlet piping, this to can be checked by normal
structural design methods. It is proposed that access towers and piping
should be erected from an extension of the tank base concrete
foundation. This ensures that all elements will, at least be subject to the
same base motion, although their periods may not be the same as those
of the outer tank. Thus motions may be out of phase with resulting
loads to the tank roof which will require consideration.

DESIGN P R O C E D U R E

(1) Calculate the fundamental frequency of the tank-liquid system


in the horizontal and vertical directions.
1000C1 /-~. ~/-~
fi = ~ and CI=C, ~ Pl

1-c 7m
where the values of C, are given in Table 1. Using the nearest
tabulated value of Cw, it can be adjusted as the square root of
the thickness-to-radius ratio, and Cv may be determined from
Table 4. The units are p(kg/m3), H(m) and E(MN/m2).
(2) Calculate the first two frequencies of the convective liquid
mass (sloshing frequency).
1(18"06 1.84H~ uz
fo~)=~ k~tanh R ¢

1 (52~29 5.33H~ v2
f~2)=~ - - t a n h R /
86 N. J. L Adams

(3) From the seismic design spectrum, determine the pseudo-


accelerations, A i , A¢(1), A¢(2) and Av, as a ratio of the
acceleration due to gravity, and the pseudodisplacements D¢o)
and D¢(2).
Calculate the value of the effective vertical acceleration,
Av(e).
Av(e) = 0.2Av
(4) Calculate the sloshing wave height from

The units of d will be those of D~.) and D~(2).


(5) Calculate the maximum base pressure due to the impulsive
mass effects.
Pi = 9"81c~(O)Aip,H10 -6 (N/mm 2)
where the value of c~(0) may be obtained from Table 2.
(6) Calculate the maximum base pressure due to the convective
mass effects.
Pc = 9"81Cl(O)Ac(1)Pt R l O - 6 (N/mm2)
where the value of c1(0) may be obtained from Table 3.
(7) Thus, the maximum hydrodynamic hoop pressure at the
bottom of the shell is:
Pb = V~2 + P 2 (N/mm 2)
(8) Calculate the maximum vertical hydrodynamic pressure
p,, = 6 . 5 3 p t H A v 1 0 -6 (N/mm 2)
(9) The maximum hydrostatic pressure at the bottom of the shell
is:
Ps = 9"81p~H10-6 (N/mmz)
(10) The maximum combined pressure at the bottom of the shell is:
Pbc = Ps + ~/pZ + p2 (N/mm 2)
(11) The maximum circumferential stress in the tank shell is:
R
OOmax= 1000pb~ ~a (N/ram2)

(Note: due to constraint effects at the shell-annular junction,


the actual maximum hoop stress will develop at approximately
the top of the first strake.)
Seismic design rulesfor liquid storage tanks 87
(12) Determine from Tables 2 and 3, the values of m J m l , mc/ml,
hi~H, hc/H and hl/H.
(13) Calculate the maximum base shear force from

amax = 9"81:tR2Hp,
[(m~ mc
Ai)2+ ( 2 A¢(,)] j
"~211/2 (N)

(14) Determine the maximum base shear stress


1
'rma~- • Qm~ (N/mm2)
lO00~rRt
where t is the thickness of the shell bottom course (in mm).
(15) Calculate the maximum base moments induced in the tank
shell in the foundation.
Mtx=9810~rR2i.i2pl[(mihiA.'~2 [m~h~ a x~211/2
[\m~H '] + \ml H e(1)/ J
[(mih' \2 / m , h , \21 'c~
M~x=9810~rR2tFp, -~-~A,) +~Ac(l, ) J (Nmm)

(16) Calculate the maximum compressive axial stress at the base of


the tank shell.
1
Ozm~ :rR2tl06. Mt~, (N/mm 2)

(17) Determine the maximum normal pressure on the foundation,


at the outer edge of the tank.
t
pf = Ozm~- (N/mm 2)
a
where a is the length of the annular plate projection in mm.
(18) Calculate the maximum restoring moment MR assuming that
lift-off is limited to a maximum distance equal to the width of
the annular plate.
w = 9.81~rR2Hp~
ws = 19.62~rRHttapml O-3
where Ht is the actual height of the tank and ta is the average
shell thickness in mm.
w(2R - La) 2
wf = 4R 2
MR = 500wfLa + 830(w + ws - wf)R (Nmm)
where La is the width of the annular plate in m.
88 N. J. I. Adams

Calculate the resistance to horizontal movement, Rh.


Rh = 0"6(1 -- Av(e))wf (N)
Check that Rh -- Qm~x
Calculate the adjusted restoring moment, MR(v).
MR(v) = Ma(1 - Av(e)) (Nmm)
Check that MR(v)- ~:~ Mmax
t
If either R, or Ma(v) is less than Qm~x or Mt~, the
implications for tank design shall be agreed between the
purchaser and tank designer.
(19) Calculate the radial membrane stress in the bottom plate, as a
result of lift-off.
1[ E
Orm = ~b 6(1 -- V2) PZ~tbLa2106
]1/3 (N/mm:)

(20) Calculate the shell lift-off

vs = 6--0--£~'Oybtb+ .psLa 500La- t~,


tTrmtb 12(1 "---~2) Orm
(ram)
(21) Using vs determine the rotation at the shell-annular junction

Op - lOOO t L a ~-~ (rads)


Having completed the evaluation of the preceding quantities for an
OBE, it is necessary to determine those quantities, dependent on

TABLE 5
Permissable design stress levels

Component Stress Transient loading

Wing loading or O B E SSE + all other


+ all other loading loading

Shell plate Tensile 0.85Yield or Yield or 0-2%


stress 0 . 2 % p r o o f stress proof stress
Compressive Para. E.5.3 of Lesser of design
stress API 650 stress E E M U A 147
or (0.4xEt/R)

Anchor Tensile or 0.85 Yield or Yield or 0-2%


compressive 0.2% proof stress proof stress
stress
Seismic design rulesfor liquid storage tanks 89

acceleration and displacement for an SSE. Sloshing wave height may


well be greater than the freeboard, and although liquid in the annular
space may be d e e m e d acceptable, some thought should be given to
impact loading on the outer tank roof.
It is proposed that the allowable stress levels against which the
suitability of the design should be judged be as shown in Table 5.

REFERENCES

1. EEMUA, Recommendations for the design and construction of refrigerated


liquified gas storage tanks. Publication No. 147, Equipment and Materials
Users Association, London, 1986.
2. Nyman, D. J. (ed.) Guidelines for the Seismic Design of Oil and Gas
Pipeline Systems. American Society of Civil Engineers (ASCE), 1984.
3. Housner, G. W., Dynamic pressures on accelerated fluid containers.
Bulletin of the Seismological Society of America, 47, 1972, 15-35.
4. Haroun, M. A. & Housner, G. W., Seismic design of liquid storage tanks.
J. of Tech. Councils of ASCE, 107 1974, 191-207.
5. Veletos, A. S. & Yang, J. Y., Earthquake response of liquid storage
tanks. Adv. in Civil Eng. Through Eng. Mechs. Proc. of the Eng. Mech.
Div. Speciality Conf., ASCE, 1977, pp. 185-99.
6. Kana, D., Seismic response of flexible cylindrical liquid storage tanks.
Nuclear Eng. and Design, 52 (1979) 185-99.
7. Haroun, M. A., Dynamic analysis of liquid storage tanks. EERL 80-04,
1980.
8. Clough, D. P., Experimental evaluation of seismic design methods for
broad cylinder tanks. EERC 77/10, 1977.
9. ASCE, Fluid/structure interaction during seismic excitation. ASCE, 1984.
10. Wozniak, R. S. & Mitchell, W. W., Basis of seismic design provisions for
welded steel oil storage tanks. API Refining 43rd Midyear Meeting, 1978.
11. Anon., Welded steel tanks for oil storage. API Standard 650.
12. Recommended practice for LNG aboveground storage. Japan Gas Asso-
ciation, December 1981.
13. Priestley, M. J. N. (ed.), Seismic design of storage tanks: recommenda-
tions of a study group. New Zealand National Society for Earthquake
Engineering, 1986.
14. Sakai, F., Sakota, H. & Ogawa, H., Seismic design oil storage tanks
related with proposal of API. Kawasaki Heavy Industry Engineering
Review, 71 (1979) 52 (in Japanese).
15. Denham, J. B., Wills, C. R. & Russel, J., A comparison of predicted and
measured stresses in a large storage tank. API Div. of Refining, May
(1968) 1034-74.
16. Cambra, F. J., Earthquake response considerations of broad liquid storage
tanks. EERC 82/25, 1982.
90 N. J. L Adams

APPENDIX

Design procedure example


Data
a = 150.0 m m A n n u l a r plate projection length
E = 207 x 103 M N / m 2 Y o u n g ' s m o d u l u s of plate material
g = 9-81 m / s 2 Acceleration due to gravity
Ht = 17.34 m T a n k shell height
H = 15-91 m Liquid level height ( H H L A )
La = 1.0m A n n u l a r plate width
R = 22.00 m T a n k shell radius
t= 11.6 m m B o t t o m plate thickness
ta = 10.3 m m Plate average thickness
tb = 10"0 m m A n n u l a r plate thickness
oyt, = 586"5 M N / m z Yield strength of annular plate material
Pl = 570.0 k g / m 3 Stored liquid density
Pm= 7800"0 k g / m 3 Plate material density
p,,, = 1000.0 k g / m 3 Water density
v= 0.3 Poisson's ratio of plate material
(1) F u n d a m e n t a l frequency of the t a n k - l i q u i d system in the
horizontal and vertical directions.

C; = C,,, P ~

Interpolating from Table 1


Cw = 0.056 86

C1 = 0"056 86 ~ 1 ~ = 0"0753
~/ 570

1000(0.0753) .~/207 x 103


f= 2at(15.91) ¥ -78--00 - 3-8811Hz
Interpolating from Table 4
Cv = 0-0797

fv 2~H
1000(0.0797) .i/207 x 103
f+- 2, (15.91) 78-0d -4.1072Hz
Seismic design rules for liquid storage tanks 91

(2) First two frequencies of the convective liquid mass (sloshing


frequency).
1 (~_0__6 (1"84H)~ v2
fcO)= ~ tanh R /
1 /18"06 (1-84(15"91)).~ lc2
fco) = ~ [ ~ t a n h 22 /
fc(1) = 0.1345 Hz
1 (~tanh (5"33H)~ '/2
fc(2)= R /

1 ~/52.29
- - ~ t a n t t . (5"33(15"91))] '/2
fc(2)----~ "~ ]

fc(2)= 0.2453 H z
(3) R e a d values of pseudoaccelerations and pseudodisplacements
from the seismic design spectrum.
f~= 3.8811 I-Iz Ai = 7.00 m/s 2
fc(,)= 0.1345 I-Iz Aco) = 0.80 m/s 2 D¢o) = 1150 m m
fc(2) ----0.2453 Hz Ac(2) = 1.45 m / s 2 Dc(2) = 640 mm
fv = 4.1072 Hz A~ = 7.00 m/s 2
As a ratio of the acceleration due to gravity, g.
Ai = 0.7136
Aco) = 0-0815
At(2) = 0" 1478
Av = 0.7136
Av(e) = 0.2Av
A~(o = 0.2(0.7136)
A~(o = 0-143
(4) Sloshing wave height.
d = ((1"54Dc(1, tanh (1"84H)~ 2 (0"39Dc(2)tanh (5"3-3H)~2~ vz
/+ R //
(1.84(15.91))~ 2
d = ((1.54(1150) tanh ~ /

(5.33(15.91))~2~ lrz
+ (0-39(640) tanh ~ / /

d = 1559.79 mm
92 N. J. I. A d a m s

(5) M a x i m u m b a s e p r e s s u r e d u e to the i m p u l s i v e mass effects.


Pi = 9 " 8 1 c ~ ( O ) A i p ~ n lO -6

From Table 2
c~(O) = 0.814
pi = 9 . 8 1 ( 0 . 8 1 4 ) ( 0 . 7 1 3 6 ) ( 5 7 0 ) ( 1 5 - 9 1 ) 1 0 -6 = 0.0517 N / m m 2
(6) M a x i m u m b a s e p r e s s u r e d u e to c o n v e c t i v e mass effects.
p~ = 9 . 8 1 c l ( O ) A c ( o R , R I O -6

From Table 3
c1(0) = 0.429
Pc = 9 . 8 1 ( 0 . 4 2 9 ) ( 0 . 0 8 1 5 ) ( 5 7 0 ) ( 2 2 ) x 10 -6 = 0.0043 N / m m 2
(7) M a x i m u m h y d r o s t a t i c h o o p p r e s s u r e at t h e b o t t o m o f t h e shell
2 2
Pb = ~ P c
Pb = ~/(0"0517) 2 + (0"0043) 2 = 0"0519 N / m m 2
(8) M a x i m u m vertical h y d r o d y n a m i c p r e s s u r e .
Pv = 6"53plHAv10 -6
Pv = 6"53(570)(15"91)(0"7136)10 -6 = 0-0423 N / m m 2
(9) M a x i m u m h y d r o s t a t i c p r e s s u r e at the b o t t o m o f t h e shell.
Ps = 9 " 8 1 p I H 10-6
1
lTmax amax
lO00~rRt

1
rmax -- (40 434 301"2075) = 50"4335 N / m m 2
1000~r(22)(11.6)
ps = 9 . 8 1 ( 5 7 0 ) ( 1 5 . 9 1 ) 1 0 -6 = 0.0890 N / m m 2

(10) M a x i m u m c o m b i n e d p r e s s u r e at t h e b o t t o m o f t h e shell.
2 2
Pbc p~ + PVrP-~b+pv
Pb~ = 0"089 + X/(0"0519)2(0"0423) 2 = 0" 1560 N / m m 2
(ii) Maximum circumferential stress in the tank shell.
R
or0... = 1 0 0 0 p ~

(22)
oom.x = 1000(0-1560) - - - 333"2039 N / m m 2
(10-3)
Seismic design rules for liquid storage tanks 93

(12) R e a d f r o m Tables 2 and 3 the following values.

m--2= 0-414
ml

m~ = 0-558
ml
hi
-- = 0.4
H

h~ 0.559
H
hi
- - = 1.009
H
(13) M a x i m u m base shear force.
m 2 mc ]211~2

Qm~ = 9-81~(22)2(15"91)(570)[(0"414)(0-7136)) 2
+ ((0-558)(0.0815))211/2
Qm~x = 4.434301.2075 N
(14) M a x i m u m base shear stress.
(15) M a x i m u m base m o m e n t s induced in the tank shell and the
foundation.

Mt,,,=98 1OAR H2pl[(~-~1~Ai


2 mihi
) 2 + [mch¢
\~11~A¢(,)] j
t
Mmax = 9810~r(22)2(15.91)2(570)[((0.414)(0.4)(0.7136))2

+ ((0.558)(0.559)(0.0815))2] lt2
t
Mma~ ----2-6015 x 1011 Nmm

M~,=9810~tR H p,[~-~-~A, + ~ll~AC(1)] j

M~ = 9810~r(22)2(15.91)2(570)[((0.414)(1.009)(0.7136) 2
+ ((0.558)(0.559)(0.0815))2] lr2
f
Mm~ = 6-4387 X 1011 N m m
(16) M a x i m u m compressive axial stress at the base of the tank shell.
1
O'~m~x= ~ R 2 t l 0 6 • Mtm~
94 N. J. I. A d a m s

1
O'zmax ~r(22)2(11.6)106. (2.6015 × 1011) = 14-7493 N / m m 2

(17) M a x i m u m normal pressure on the foundation at the o u t e r edge


of the tank.
t
pf = O'zmax
a-
(11.6)
pf = 1 4 " 7 4 9 3 ~ = 1"1406 N / m m 2
(150)
(18) M a x i m u m restoring m o m e n t MR assuming lift-off is limited to a
m a x i m u m distance equal to the width of the annular plate.
Firstly, total weight of the stored fluid,
w = 9.81~REHR~
w = 9.81~(22)2(15-91)(570) = 135 272 425.447 N
Weight of the tank shell and roof,

ws = 1 9 . 6 2 ~ R H , tapm X 10 -3
ws = 19-62~(22)(17.34)(10.3)(7800) X 10 -3
w~ = 1 889 089.252 N

Proportion of w supported by foundation,


w(2R - La) 2
Wf = 4R 2

(135 272 425.447)(2(22) - (1)) 2


wf = 4(22)2
wf = 129 193 550.956 N
Restoring m o m e n t , MR,
MR = 5 0 0 w f L a + 830(w + ws - wf)R
MR = 500(129 193 550.956)(1) + 830(135 272 425.447
+ 1 889 089.252 -- 129 193 550.956)(22)
MR = 2.100 9 X 10 u N m m
Resistance to horizontal m o v e m e n t , Rh
Rh : 0.6(1 - Av(¢))wt
gh = 0.6(0.858)(129 193 550-956)
Rh = 66 508 840"0 N
Seismic design rules for liquid storage tanks 95
Adjusted restoring moment, MR<v)
MR<v)= MR(1 - Av(O)
MR<v)= (2.1009 X 1011)(0.858)
MR(v) = 1.8026 X 1011 Nmm
(19) Radial membrane stress in the bottom plate, as a result of
lift-off.
E pZtbLa2106] It3
tb1 6(1 V2)

1 [(207 X 103) (0.089)2(10)(1)2106] 1~3


0.. = ~ L~ Z-O.-~
oH = 144.2733 N/mm 2
(20) Shell lift-off.

[ 1 lO00.psLa(5OOLa_~. 1 E )]
vs = 6--~ " Oybtb+ a~t------~b 2(1 V2)O'rm"tg -

[ (586"5)(10) 1000(0"089)(1)
Vs= L ~ + (144.2733)(10)
x 07x,0 .0)3
12(ira 0 . 3 2 ~ 3 3 ) 1 1
vs = 15.2589 mm
(21) Rotation at the shell to annular junction.
1 [2v~ v~]
Op = lO00 L La 2-R
0 = 1 [2(15(_~)589)(15"2589).]
p 1000 2(22) J
Op = 0.0302 rad.

You might also like