EARTHQUAKE ENGINEERING AND STRUCI‘URAL DYNAMICS, VOL.

3, 3 3 4 9 (1974)

DYNAMICS OF TOWERS SURROUNDED BY WATER

c-Y.L I A W ~AND ANIL K. CHOPUS
Department of Civil Engineering, University of California, Berkeley, Califarnia, U.S.A.

SUMMARY
The effects of interaction with surrounding water on the dynamic response behaviour of cantilever tower structures
are studied. Expressions for response to harmonic ground motion in individual modes of vibration, including
hydrodynamic interaction, are presented, the accuracy of responses obtained by ignoring surface waves and
compressibility of water in the hydrodynamic solutions is evaluated, the effects of hydrodynamic interaction on
the fundamental period of vibration are studied and the commonly used ‘added mass’ approach to account for
effects of surrounding water is examined. The conclusions deduced from the results of this investigation include
the following. Interaction with surrounding water increases the fundamental period of vibration of the tower
and decreases the modal damping ratio. Compressibility of water has essentially no influence in the hydro-
dynamic effects on slender towers. The traditional definition of added mass is conceptually deficient, but is simple
to employ; the errors in this simple added mass representation are negligible for the first mode of vibration of
towers.

INTRODUCTION
The dynamic response behaviour of cantilever tower structures, such as intake towers and bridge piers, is
influenced by the surrounding water. Interaction with water causes additional dynamic forces on the structure
and also modifies its dynamic properties.
The classical theory of hydrodynamics,’ concerned with the motion of rigid bodies in incompressible
fluid, has had for many years profound influence on the point of view and assumptions adopted by researchers
concerned with earthquake analysis of structures interacting with water. The early investigations of water
pressures acting on gravity dams? on cylindrical tanks containing wate? and on cylindrical structures
surrounded by water,3 assumed the structure to be rigid, considered water to be incompressible and deter-
mined an ‘added mass’ of water which when considered moving with the structure was equivalent to the
hydrodynamic effects of the fluid.
Until a few years ago, these basic assumptions-rigid structure and incompressible fluid-had not been
systematically examined. Recent investigations4-’ concerned with concrete gravity dams demonstrated that
dynamic interaction between the flexible structure and water might have profound influence on dynamic
behaviour and response of the structure. It was further shown that significance of compressibility of water
depends on the dynamic properties of the structure relative to the reservoir of water.4 The significance of
structure-water interaction and of water compressibiIity in the earthquake response behaviour of cantilever
towers surrounded by water had not been completely investigated prior to the present study. Solutions for
hydrodynamic pressures on cylindrical towers-circular* or elliptical9 in plan-for prescribed harmonic
motions-rigid body or in a deformed shape-of the tower were obtained. Although the significance of
compressibility of water was not examined, the magnitude of added mass of water considered to be moving
with the structure, which will lead to the correct change in period of vibration due to the surrounding water,
was presented. Following the formulation and concepts developed for studying the dynamics of gravity
dams including hydrodynamic interaction? the dynamics of towers in the fundamental mode of vibration
was studied.1° The results presented were not complete enough to reach definite conclusions.
~

t Graduate Student.
t Associate Professor.
Received 19 October I973
Revised 20 December 1973
0 1974 by John Wiley & Sons, Ltd.

2 33
34 C-Y. LIAW AND ANIL K. CHOPRA

The magnitude of added mass of water which should be considered moving with the tower has also been
determined from experimentally measured changes in the period of vibration of the structure due to hydro-
dynamic interaction?‘
The main purpose of this investigation is to develop fundamental understanding of the dynamic response
behaviour of cantilever tower structures, surrounded by water, subjected to earthquake ground motion.
The vertical component of ground motion essentially causes no lateral response of slender cantilever
structures. Because the most significant stresses in slender towers are caused by lateral flexural deformations,
the effects of the vertical component of ground motion are essentially of no consequence and only the
horizontal component of ground motion is considered in this study.
Expressions for response to harmonic ground motion in individual modes of vibration of cylindrical
towers, including interaction with surrounding water, are presented. The accuracy of responses obtained
by ignoring surface waves and compressibility of water in the hydrodynamic solutions is examined. The
effects of hydrodynamic interaction on the fundamental vibration period and dynamics of the tower in the
first two modes of vibration are studied. The commonly used ‘added mass’ approach to account for effects
of surrounding water is examined.

ASSUMPTIONS AND GOVERNING EQUATIONS

For the purposes of this study, it is appropriate to simplify the problem as much as possible, yet retaining
all the features of essential importance. The structure is assumed to be linearly elastic. The geometry of the
tower is taken as cylindrical; water can then be treated as a continuum and the governing wave equation
for harmonic excitation can be solved in explicit mathematical form, which makes it convenient to examine
the significance of compressibility of water and of waves at the free surface in dynamic response of the tower.
As is well known, the response of many classes of structures, including towers, to horizontal ground motion
is primarily due to the fundamental mode of vibration. Consequently, the primary concern here is to study
tower-water interaction considering only the fundamental mode of vibration of the tower. However, limited
attention is also given to responses in higher modes.
Planar vibration of a linearly elastic cylindrical tower with circular cross-section subjected to horizontal
ground motion is considered (Figure 1). The tower may be treated most simply as a one-dimensional
structure governed by beam theory.

+%-4
Figure 1. Cylindrical tower submerged in water subjected to horizontal ground motion
 DYNAMICS OF TOWERS SURROUNDED BY WATER 35

If it is assumed that the tower can distort only in the shape of its fundamental mode of vibration in air,
the equation of motion for its response to a horizontal ground acceleration, G&), in terms of Yl(t),the
generalized co-ordinate representing the amplitude of this deformation, is
Mf Y1(t>+Ct &(?)+KT Yl(t)= -u&)Al-P,*l(t) (1)
The generalized properties denoted in this equation are defined by the following expressions, in which
the fundamental mode of the tower in lateral vibration without the surrounding water is denoted by +&),
the frequency by w1 and the damping ratio by &,

M : = /oHk(z)+l(z)adz

where m(z) represents the mass of the tower per unit of height
K: = o fM :
C: = 2f;01 Mf

and PA(?)represents the generalized load in the first mode of vibration of the tower associated with hydro-
dynamic pressures pc(8, z, t ) acting on the outside surface of the tower. It will be shown later that for planar
vibration of the tower this radial pressure varies around the circumference as cos 8 (Figure l), i.e.
P,(& z, 0 = P&O, z , o cos 8
The resultant hydrodynamic force per unit of height in the plane of vibration is

p;(z,t) = Jo2>c(~,z,t)rocosede (7)

P,*1in equation (1) then is given by

Assuming water to be linearly compressible and neglecting its internal viscosity, the small amplitude
irrotational motion is governed by the wave equation in cylindrical co-ordinates (Figure 1):

in which p(r, 8, z, t) is the hydrodynamic pressure (in excess of hydrostatic pressure) and C is the velocity of
sound in water = 4,720 fps. The hydrodynamic pressure acting radially on the outer surface of the tower,
pe(8, z, t ) = p(ro, 8,z, t). Also the following relations hold for any particle of water

g at2 ar

w a2 uz - ap
at2 az
where uryu* and uz respectively are the radial, tangential and vertical components of particle displacement of
water in the co-ordinate system shown in Figure 1; w = the unit weight of water, g = the acceleration due to
gravity.
36 C-Y. LIAW AND ANIL K. CHOPRA

For the present, only cylindrical towers are considered, i.e. the tower surface in contact with water is
vertical. The reservoir bottom is assumed to be horizontal, i.e. the depth H of water in the reservoir is
constant. It is further assumed that water extends to infinity in the radial direction from the tower.
Equation (9) together with appropriate boundary conditions at various boundaries-the outer surface of
the tower, the reservoir bottom and the free surface of water-defines the problem for the water system.

EXPRESSIONS FOR COMPLEX FREQUENCY RESPONSES

It is a property of linear time invariant systems that when the excitation is a simple harmonic motion, the
steady-state response is also simple harmonic motion at the same frequency. The amplitude and the phase
of the response generally depend on the frequency. The frequency dependence of the amplitude and the
phase is described by the complex frequency response function. The complex frequency response for a
response quantity z(t), denoted by P(w), has the property that when the excitation is the real part of eid,
the response is the real part of i(w)eid.
In studying the effects of hydrodynamic interaction on the dynamics of tower structures, it is most
appropriate and convenient to first consider harmonic ground motion and to derive the complex frequency
response functions.
The responses to harmonic ground acceleration, iio(t)= eid, can be expressed as follows:

Generalized displacement
Yl(t)= Tl(w) eiul
Generalized acceleration -
Yl(t) = Yl(w) eid = - w2 P,(w) eid
Acceleration of the tower along 0 = 0
ii(t> = [I +T1(w) +,(z)] eid
Hydrodynamic pressures
P,(& z, t> = pC(& z, w ) eid

Considering a cylindrical tower surrounded by water, subjected only to the horizontal component of ground
motion, the boundary conditions for the wave equation are:

(i) No vertical motion at the boundary z = 0, i.e.

(ii) Wave condition at the free surface

a2P aP
-
at=
(r, 8, H, t ) = -g (r, 8, H, 0

(iii) The radial component of motion of the boundary r = ro is the same as the radial
motion of the outer surface of the tower

(ra, e, z, t ) = - - [I + yl(u)+l(z)l cos 8 ei"

W

(iv) The condition of symmetry about the 0 = 0 plane I

aP 0, z, t ) = -(r,
B(r, aP T ,z, t ) = 0
ae
 DYNAMICS OF TOWERS SURROUNDED BY WATER 37

Because the governing equation as well as boundary conditions are linear, the principle of superposition
applies. The complex frequency response function for pe can therefore be expressed as

where

with k = 0 and I is the solution at r = ro of the wave equation [equation (9)] for the following boundary
conditions

The solutions Po and pl correspond to do(.) = 1 and dl(z) respectively. po is the complex frequency response
function for the hydrodynamic pressures on the tower surface when the excitation is the horizontal
acceleration of the ground and the tower is rigid. p1 is the corresponding function when the excitation is the
acceleration of the tower in the first mode of vibration and there is no ground motion.
The complex frequency response function for the response of the tower is presented next in terms of Po
and pl. Expressions for po and p1 obtained from solutions of the wave equation [equation (9)] with the
boundary conditions of equation (14) are presented later.

Structural response
The complex frequency response function for generalized modal acceleration obtained from equations
(1) and (13) is

where

and

in which the equations and boundary conditions governing pk were stated earlier. The solutions to these
boundary value problems are presented next.
For a tower with no surrounding water, d k = 0, consequently Bk = 0, and equation (15) becomes the
standard expression for structural response in the fundamental mode of vibration of any structure.
38 C-Y. LIAW AND ANIL K. CHOPRA

Hydrodynamic solutions
Po and jl obtained from steady-state solutions of the wave equation for the boundary conditions of
equation (14) are given by
h-0 Do(&, ro)cosh (a,, z) eiEora
2%H+ sinh (2%H )

ml--lh Ikm
-I:-
m-1 Am 2umH + sin (2a, H )
ro)cos (umz)eic-ro
Dm(hm,

a0
Ikm
+I:&
m a l hk 2ormH + sin (2umH)
Em(&, ro)cos (a,~)

In equation (18) k takes on the values of 0 and 1; m1is the smallest integer value of m such that a , > o/C;
with the stipulation that if m1 = 1 the first series disappears; a,, and or, are solutions of

Also,
a,tan(or,H) = --w2
g I

where +o(z) = 1 and +&) is the shape of the fundamental mode of vibration of the tower (without water),

In equations (21)-(23), J, and Y, are the Bessel functions of order n of the first and second kind respectively,
and K , is the modified Bessel function of order n of the second kind.
 DYNAMICS OF TOWERS SURROUNDED BY WATER 39

If waves at the free surface are ignored, the wave boundary condition at the free surface [equation (14b)l
is replaced by
p(r, 8, H, t ) = 0 (24)
The solution for p k for boundary conditions (14a), (14c), (14d) and (24) is a special case of equation (18)
in which the first term does not appear, equation (19b) becomes tan 01, H = 0 the solutions for which are
~ (2m - 1) 7r/2H, i.e.
0 1 , ~=

pk as presented in equations (18) and (25) is a complex valued quantity, and hence out of phase with the
excitation, for w > 01, C.
A further simplification in the solution for p k results if compressibility of water is neglected, an assumption
commonly made in many earlier investigations. The governing equation then is the Laplace equation

instead of the wave equation (9). Neglecting compressibility of water, the solution for j k ignoring surface
waves can be obtained either by solving equation (26) for the boundary conditions defined by (14a), ( I ~ c ) ,
(14d) and (24) or by substituting C = cc in equation (25)

in which = a, = (2m - 1) 7r/2H.It should be noted that p k is now independent of the excitation frequency
w and in phase with the excitation for all frequencies.

Hydrodynamic solutions-limiting cases

The cylindrical tower system approaches the two-dimensional straight gravity dam problem as ro/H tends
to infinity. The hydrodynamic solutions presented in the preceding section should therefore become identical
to the corresponding expressions obtained in earlier investigations concerned with gravity dams.4
The limiting values for various terms in equation (25), the expression for f j k neglecting surface waves but
considering compressibility of water, as ro/H approaches infinity are given by D,(X,, ro) = 4, Em(&, ro) = 4
and ei".ro = i.
The corresponding limiting expression for @ k is

where

with
(2m - 1) r
a, =
2H

which is identical to the one presented earlie9 and is apparent from equation (28) that j j k is unbounded
at a discrete set of frequencies o = a , C, m = 1,2,3, ..., the resonant frequencies for the water.
40 C-Y. LIAW AND ANIL K. CHOPRA

In contrast, the hydrodynamic pressures pk on an axisymmetric tower [equation (25)] are bounded for all
values of w including w = am C because

This lack of ynbounded resonance in pk is of considerable significance in the dynamics of tower-fluid inter-
action. This is one of the reasons why compressibility of the water has essentially no influence in the tower
response.
EFFECTS OF SURFACE WAVES
Neglect of surface waves is equivalent to assigning g = 0 in equations (14b) and (19) or equivalently w = 03
in equation (19). Surface waves are therefore not expected to be of any significance in responses at high
frequencies; however, they would possibly be significant at relatively low frequencies.
The frequency responses of total hydrodynamic force on a rigid tower-the integral of po(z,w) over
depth of water, H-computed from equations (18) and (25) are presented in Figure 2. A basic parameter
on which the hydrodynamic responses depend is ro/H;if surface waves are included an additional parameter
H also enters into the results. It is apparent that surface waves have no effect on hydrodynamic responses
except at low frequencies. The effect of surface waves on the values of Po and also of p1 at low frequencies
could significantly influence structural response [equation (1 5)] only if the fundamental frequency of the
tower lies within or close to this low frequency range. Such would very rarely be (if ever) the case because
structural frequencies are much higher. It is concluded that surface waves will be of little consequence in the
earthquake response of towers surrounded by water.

4k
FY
-\

3. 25

I I I I l l I t I l l I
0 005 010 02 05 I 2 5 10
001
EXCITATION FREQUENCY UJ+
Figure 2. Frequency responses of hydrodynamic force, rigid tower

BASIC EFFECTS OF HYDRODYNAMIC INTERACTION

How interaction with surrounding water affects the dynamic response behaviour of towers is the central
question that arises. This is investigated next by examining the structural response considering a mode of
vibration of the tower.
 DYNAMICS OF TOWERS SURROUNDED BY WATER 41

Expressions for the response of towers considering only the fundamental mode of vibration have been
presented in equation (15) in which A,, M T , B,(w) depend on the mode shape +,(z). The responses considering
only the second mode of vibration can also be computed from equation (15) if $,(z) is replaced by d2(z), the
shape of the second mode of vibration, in computing the above-mentioned quantities and w , and 5, are
replaced by w2 and c2, the frequency and damping ratio of the second mode.
In a strict sense, a normal mode of vibration of the tower (without the surrounding water) should not be
treated independently of the other modes because these modal co-ordinates do not remain uncoupled when
hydrodynamic interaction is included. The uncoupled responses presented here are, however, suitable for
the purpose of evaluating the significance of compressibility of water.

Complex frequency responses

Towers with two different values of the ratio of outside radius to height, r0/€€,= 0-05and 0.25, are
considered. The former is typical of many slender towers, whereas the latter is chosen as a rather extreme
value to examine the dependence of water compressibility effects on the ro/€isratio. The ratio of wall thick-
ness to outside radius, t/r,,, is selected as 0.2, a value typical of many towers. When water compressibility is
considered, effects of hydrodynamic interaction depend on the modulus of elasticity of concrete E, but not
when water is assumed to be incompressible. A value of 5 million psi is assumed for E. The unit weight of
concrete is taken as 155 pcf, the modal damping ratio as 0.05. The frequencies and shapes of the modes of
vibration were taken as those for uniform cantilever beams obtained by elementary beam theory.

First mode responses- Y,:

1-No surrounding woter,H=O
24[ I 2-Fully submerged tower,HSH,
2(

Excitation frequency: w / u ,
Figure 3. Frequency responses of slender tower (ro/H8= 0.05) with t/r, = 0.2. Results are the same whether compressibility
of water is included or not

The frequency response of generalized accelerations in the first two modes of vibration, treated separately,
is computed from equations (15) to (17) and presented in Figure 3 for the slender tower (ro/Hs = 0.05). The
familiar frequency response curves for one-degree-of-freedom systems describe the behaviour of the tower
42 C-Y. LIAW AND ANIL K. CHOPRA

when there is no surrounding water; in this case B k ( W ) = 0 in equation (15). Although in designing towers
the height of the structure is always selected to be greater than the maximum depth of the surrounding
water, the depth of water was taken as equal to the tower height for simplicity. The response for this case
was computed from equation (1 5) with j kcomputed from equation (25), considering compressibility of water
but ignoring effects of surface waves which, as demonstrated earlier, are of essentially no consequence.
The effect of surrounding water on the dynamic behaviour of the tower is twofold: firstly, the resonant
frequencies are lowered; secondly, the response exhibits sharper resonance implying an apparent decrease
of the modal damping ratio.
If compressibility of water is neglected, j k [equation (27)] and hence Bk in equation (15) are real valued
and independent of the excitation frequency. The frequency responses of the tower neglecting compressi-
bility of water were also computed. These results were almost identical to those obtained including water
compressibility and displayed no perceptible difference when plotted on the scale of Figure 3.
The frequency responses of generalized acceleration in the first two modes of vibration of the squatty
tower (ro/Ha= 0.25) for the three cases considered earlier, are presented in Figure 4. The effects of hydro-
dynamic interaction on response in the first mode are generally similar to those discussed earlier for the
First mode responses -p
1
1-No surrounding woter, H=O
2-Fully submerged tower, H=H,
24 Second mode respomes-p
2
3-No surroundnq water,H=O
4-Fully submerged tower,H= H,

A3

- -
' Water compressibility
- Included
--- Neglected

Excitation frequency:u/u,
Figure 4. Frequency responses of squatty tower (ro/H, = 0-25) with tlro = 0.2
slender tower. Compressibility of water has more effect in the response of the squatty tower compared to
the slender tower. Even for the squatty tower, however, the effects of compressibility of water on the first
mode response are negligibly small. The surrounding water affects the response in the second mode rather
differently in the case of the squatty tower. Firstly, hydrodynamic interaction does not cause the significant
decrease in the resonant frequency, which was observed in the case of the slender tower. Secondly, com-
pressibility of water causes drastic damping of the response especially near the resonant frequency.
 DYNAMICS OF TOWERS SURROUNDED BY WATER 43

The functions B k ( W ) are presented in Figure 5 to aid in discussing the reasons for the different type of
effect due to hydrodynamic interaction and compressibility of water in the higher mode response of the
squatty tower. The phase of B k ( W ) is zero, i.e. Bk(w) is a real valued function, for all values of the excitation
Water compressibiity
Included
Neglected

W
3
.- 0.6
c
z
E
a

0-4

0 ‘1 2 3 4 5 16 7 a
cur of tower oo5 u2of tower u2 of tower

a
Excitation f r e q u e n c y : + $ )
Figure 5. Effects of compressibility of water on functions &(w)

frequency w less than 7rC/2H no matter whether compressibility of water is recognized or not. The range of
frequencies in which hydrodynamic effects are similar to those of an ‘added mass’ includes the first two
resonant frequencies of the slender tower and the fundamental but not the second resonant frequency of the
squatty tower. In the vicinity of the second resonant frequency of the squatty tower, the phase angle of
44 C-Y. LIAW AND ANIL K. CHOPRA

Bk(w) exceeds 60 degrees when compressibility of water is considered. This implies a doininant imaginary
component which in equation (15) has the effect similar to an ‘added damping’. The result is a drastically
reduced response but essentially no change in the resonant frequency. On the other hand, the phase of
Bk(w) is zero at all frequencies when compressibility of water is ignored, which produces large errors in
response of the second mode of the squatty tower over a broad range of frequencies.

Fundamental period of vibration

One of the most important quantities governing the earthquake response of a system is its fundamental
period of vibration. The relative change in the fundamental resonant period of vibration from Tlwith no
surrounding water to Fl with surrounding water of the same depth as the height of the tower is obtained
by computing frequency response functions of the type presented earlier and noting the change in the
resonant frequency.
A careful examination of the expressions for structural response [equation (15)] and for hydrodynamic
pressures [equations (25) and (27)] reveals that the most important basic parameter controlling the
significance of compressibility of water in hydrodynamic interaction is CTJH, a dimensionless quantity
involving the period of vibration of the tower (without surrounding water), C the velocity of sound in water
and H the depth of surrounding water. The relative change in the fundamental resonant period due to
surrounding water is presented in Figure 6(a) as a function of this dimensionless parameter for a wide range
of values of the ratio r,,/kfy The most important conclusion from these results is that compressibility of water
affects the fundamental resonant frequency only if CTl/H is smaller than 4. It is apparent from Figure 6(b),
in which the immediately preceding results (Figure 6(a)) are presented as a function of the elastic modulus
for the tower, that compressibility of water has no effect in the practical range of elastic modulus of
reinforced concrete towers. This conclusion applies over a broad range of r,,/Hs which would encompass
almost all actual towers.

-
WATER COMPRESSIBILITY
INCLUDED
c
z
--- NEGLECTED
W
y 50
W
n
a \
d 3 = 0.02
w
Z
0
40 -- 005 40
I-
LL 0.10
0
0
0 30
LL
W
a
-J
a
2
w
a
s
20
2oi-
‘.I
n
2
2 IC
z
w
r
n
a I I I I I 01 I I I I I
L
u
0: 8 16 24 32 40 0 5 10 15 20 25
z CT, / ti ELASTIC M O D U L U S : E - K S I
Figure 6. Increase in fundamental period of tower due to surrounding water of depth equal to tower height

The previous results were for systems with depth of surrounding water equal to the height of the tower.
Similar computations were performed for partially full reservoirs. The increase in the fundamental period
of vibration due to surrounding water, noted from a frequency response analysis for each case, is plotted in
Figure 7 as a function of the ratio of depth of surrounding water to tower height for different values of the
 DYNAMICS OF TOWERS SURROUNDED BY WATER 45

ratio ro/Hs. Based on the preceding conclusions, water is assumed to be incompressible. The results obtained
under this assumption apply to all values of the elastic modulus. These results demonstrate:
(i) The fundamental period of vibration of the tower with a full reservoir increases by more than 25 per
cent due to interaction with surrounding water.
(ii) For typical design values of H / H , = 0.8 to 0.9, the increase in fundamental period ranges between 10
and 30 per cent depending on the ro/Hs ratio.
(iii) This increase is negligible for H/Hs less than 0.5.
(iv) The increase in the period of vibration is larger for slender towers.

e 401 0.05
0.10.

DEPTH OF WATER/HElGHT OFTOWER

Figure 7. Effect of surrounding water on fundamental period of vibration of the tower (t/r, = 0.2)

ADDED MASS REPRESENTATION

An important conclusion of the results presented above is that compressibility of water can be neglected in
analysing the dynamic response of tower structures including hydrodynamic interaction. Evidently, BL(w)
which are frequency dependent functions can be replaced by constants equal to the value of the function at
w = 0, i.e. B,(w) = B,(O). The modal equation of motion can then be directly written for arbitrary excitations
as
[Mf + BJ Yl(',(t> +
+2M,* 5, w1 r;(t)+w; ML*Y,(t) = - [ A B,] ii,ct) (30)
where Bk(0) has been denoted as B, for simplicity.
The effects of hydrodynamic interaction are equivalent to an 'added mass' Bl and 'added excitation'
B,,. From equation (16),

B, = /oHp:(z) $l(z)dz, k = 0,l

in which, from equations (17) and (23,

where Ah = a, = (2m - 1) r / 2 H and Ikm is defined in equation (20).

46 C-Y. LIAW AND ANIL K. CHOPRA

The period of vibration Fl and the damping ratio t1 for the fundamental mode of the tower including
effects of hydrodynamic interaction can, from equation (30), be expressed in terms of the corresponding
quantities Tl and 5, without surrounding water as

Surrounding water therefore has the effect of lengthening the period of vibration and decreasing the damping
ratio, These changes can be directly obtained from Figure 7.
The effects of surrounding water on the dynamics of towers in the first mode of vibration are exactly
and completely accounted for by considering

as an additional mass per unit height of the tower. It can be shown that equation (30) is also the equation
of motion for a tower in air, with mass distribution
f i ( 2 ) = m(2) +
ma(.) (35)
with mu(z) given by equation (34), constrained to be vibrating in the shape +&). The additional mass
depends on this vibration shape so that it will vary with the shape $&) of the mode of vibration of the
tower considered, which implies that no one function m,(z) will be exactly valid for all modes of vibration
of the tower. The additional mass computed from equation (34) for the first mode of a uniform tower is
presented in Figure 8.

I 0.8 -
\
N
N
w
v)
3 06-

m
Q
w 04-
Y
I?
0 0.2-

I I I I I 1 I I I
O; 02 0.4 0.6 0.8 1
0 1.2
ADDED MASS RATIO: m,tz)+$r,,~
Figure 8. Comparison of added-mass distributions

The concept of an added mass to represent the inertial influence of water in contact with a structure has
been applied for many years in many different situations, including problems in classical hydrodynamics>
ship vibrations18and earthquake effects on dams? These earlier investigations assumed the structure to be
rigid and computed an added mass to reproduce the hydrodynamic pressures on the vibrating structure.
For towers, such a concept leads to an added mass
mu(4 = Pb(4 (36)
with $*(z) given by equation (32); it is compared with equation (34) in Figure 8, The additional generalized
loading term in the first mode associated with this added mass is Jfm,(z)+l(z) dz which can be shown to be
 DYANMICS OF TOWERS SURROUNDED BY WATER 47

Jr
equal to B,, but the additional generalized mass which is m,(z) $1(z)2 dz is not equal to Bp Consequently,
the added mass defined by equation (36) is not an exact representation of the hydrodynamic effects. It
appears that this conceptual discrepancy has not been realized earlier, perhaps because most of the
investigations have been directed towards hydrodynamic effects on rigid towers. Even the investigations8-l0
which have considered flexible towers have not examined this aspect, although it has been discussed in a
recent study of hydrodynamic effects in liquid storage tanks.13
As discussed above, the equation of motion for a tower in air whose total (or virtual) mass is equal to its
actual mass plus the added mass, defined by equation (36), is not identical to that for the tower including
effects of the surrounding water. However, this added mass has the advantage that it does not depend on the
dynamic properties, such as vibration mode shapes, of the tower. It is therefore of interest to investigate
whether the added mass of equation (36) is adequate as an approximate representation of the hydrodynamic
effects. Values for the increase in periods of the first two modes of vibration due to surrounding water having
depth equal to tower height are presented in Figure 9. The added mass is taken as given by equation (34)
to compute the ‘exact’ increase in the vibration period, and by equation (36) to obtain an approximate
value. The estimated fundamental period of vibration of the tower including effects of the surrounding
water, based on the approximate curves in Figure 9, will err by less than 3 per cent. The errors in the
second mode period are higher but not excessive. The errors progressively increase with the mode number.

t
2 50 r
0

I
Fundomental mode
50 r Second mode

.S 40- Approximate
0
5
?,.0 30 - 30
D
‘5 ‘20 -
0

P * O 1t
.E I

- 0
0 005 010 0.15 0.20 0.25
rd-4
Figure 9. Evaluation of approximate added-mass representation

It has been established that the effects of surrounding water on dynamics of towers can be approximately
represented by an added mass m,(z) = p’,(z) distributed along the height. The total (or virtual) mass of the
tower f i ( z ) is simply the sum of m(z), the mass of the tower itself, and m,(z) the added mass of the water.
m,(z) has been computed from equation (36) for a range of values of the ratio ro/H. It has been normalized
with respect to the mass per unit height of the water displaced by the tower and is presented in Figure 10.
Strictly speaking, these results are only valid for towers with vertical outside walls; however, they are useful
even for tapered towers. It is recommended that the added mass at any location above the base along the
height of a tapered tower be computed from the curve for r,/H = ro(z)/Hpertaining to that location. This
simply obtained approximate value checks favourably with accurate solutions based on a finite element
analysis of the Laplace equation for fluid domains surrounding tapered towers.lP Once the added mass has
been determined from Figure 10, the virtual mass is used in a standard earthquake response analysis just as
if the tower were not submerged in water.
The earthquake response of the tower with virtual mass can, of course, be analysed by standard mode
superposition ~r0cedures.l~ Typically, only the first few modes of vibration need be included in such an
analysis. However, responses in the modes higher than the fundamental may err significantly, with the
errors increasing with mode number, because of the approximations in the added mass approach discussed
above. It is therefore most logical to use the added mass approach only to obtain estimates of earthquake
forces for purposes of design considering only the fundamental mode of vibration, and to develop a
48 C-Y. LIAW AND ANIL K. CHOPRA

procedure without the approximations of the added mass approach for purposes of accurate analysis of
earthquake re~ponses.1~

”
0 0.2 09 0.6 0.8 I.0
ADDED MASS RATIO: m,,(z) +: -rr:
Figure 10. Added-mass representing hydrodynamic effects

CONCLUSIONS
1. A rational approach to account for the effects of structure-fluid interaction in the dynamic response
behaviour of towers submerged in water has been developed.
2. Interaction with surrounding water increases the fundamental period of vibration of the tower;
associated with this is a decrease in the modal damping ratio for the structure.
3. Compressibility of water is essentially of no consequence in the hydrodynamic effects on slender towers.
The errors in the periods of vibration as well as in dynamic response of the tower resulting from neglecting
the compressibility of water are negligible. These errors become significant for the higher modes of vibration
of squatty towers.
4. Because compressibilityof water is not important, the effects of surrounding water are entirely equivalent
to an additional mass of magnitude and distribution depending on the deflected shape during vibration.
The equation governing vibration of the tower with virtual mass, equal to the sum of the actual mass and the
added mass, is identical to the equation of motion of the tower surrounded by water. This definition of
added mass is not the same as the one traditionally used. The traditional definition of added mass, based
on hydrodynamic pressures on a rigid structure, is conceptually deficient but is simple in that it does not
depend on the vibration mode shapes of the structure. The errors in the simpler added mass representation
are negligible for the first mode of vibration of towers.

ACKNOWLEDGEMENTS

This paper is based on a report14 on a research investigation sponsored and financially supported by the
Office of the Chief of Engineers, Department of the Army, Washington, D.C. Also, the authors are grateful
 DYNAMICS OF TOWERS SURROUNDED BY WATER 49

to Professor A. S . Veletsos for his comments which led to improvements in the paper, especially in the
presentation of Figure 10.

REFERENCES
1. H. Lamb, kfj~dp.o&ttUmiCS,6th edn., Cambridge University Press, 1932.
2. H . M. Westergaard, ‘Water pressures on dams during earthquakes’, Trans. A X E , 98, 418-433 (1933).
3. L. S. Jacobsen, ‘Impulsive hydrodynamics of fluid inside a cylindrical tank and of fluid surrounding a cylindrical pier’,
Bull. Seism. SOC.Am. 39, 189-204 (1949).
4. A. K. Chopra, ‘Earthquake behaviour of reservoir-dam systems’, J. Engng Mech. Ditt., ASCE, 94, 1475-1500 (1968).
5. A. K. Chopra, ‘Earthquake response of concrete gravity dams’, J. Engng Mech. Div., ASCE, 96, 443-454 (1970).
6. P. Chakrabarti and A. K. Chopra, ‘Earthquake response of gravity dams including reservoir interaction effects’, Report
No. EERC72-6, Earthq. Engng Res. Cent., University of California, Berkeley, 1972.
7. A. K. Chopra and P. Chakrabarti, ‘Dynamics of gravity dams-significance of compressibility of water and three
dimensional effects’, Znt. J. Eurthq. Engng Struct. Dyn. 2, 103-104 (1973).
8. H. Goto and K. Toki, ‘Vibrational characteristics and aseismic design of submerged bridge piers’, Proc. 3rd Wld Con$
Eurthq. Engng, 11, 107-122, New Zealand, 1965.
9. S. Kotsubo, ‘Seismic force effect on submerged bridge piers with elliptic cross-sections’, Proc. 3rd Wld Con$ Eurthq.
Engng, 11, 342-356, New Zealand, 1965.
10. A. R. Chardrasekaran, S. S. Saini and M. M. Malhotra, ‘Hydrodynamic pressure on circular cylindrical cantilever
structures surrounded by water’, 4th Symp. Eurthq. Engng, Rookee, India, 1970.
11. R. W. Clough, ‘Effects of earthquakes on underwater structures’, Proc. 2nd Wld Con$ Earthq. Engng, 11, 815-831,
Tokyo, 1960.
12. R. T. McColdrick, ‘Ship vibration’, Reporf 1451, Department of the Navy, David Taylor Model Basin, 1960.
13. A. S. Veletsos, ‘Seismic effects in flexible liquid storage tanks’, Proc. 5th Wld Conf. Eurthq. Engng, Rome, 1973.
14. C-Y. Liaw and A. K. Chopra, ‘Earthquake response of axisymmetric tower structures surrounded by water’, Report
No. 72-25, Earthq. Engng Res. Cent., University of California, Berkeley, 1973.
15. R. W. Clough, ‘Earthquake response of structures’, in Earthquake Engineering, (Ed. R. L. Wiegel) Prentice-Hall,
Englewood Cliffs, N.J., 1970.