Sie sind auf Seite 1von 13

Experimental added-mass in

modal vibration of
cylindrical structures
M. R. Maheri
Department of Civil Engineering, Shiraz University, Iran

R. T. Severn
Department of Ovil Engineering, University of Bristol, Bristol, UK
(Received January 1990; revised March 1991)

Added-mass is an important parameter influencing the hydrodynamic


interaction between liquid and a structure and consequently the
hydrodynamic forces acting on that structure. For this purpose
evaluation of the added-mass has been carried out by many investigators using the classical theory of hydrodynamics for rigid-body
vibration. Using this concept, they have obtained accurate results for
the fundamental frequency of vibration only. Recent theoretical
studies have indicated that the added-mass in a flexible structure is
different from that of the same structure when considered to be rigid.
In the present paper the results of a number of experimental investigations are presented which examine the nature and magnitude
of the added-mass in flexible vibration of three cylindrical models
when containing water, as well as when surrounded by water. It is
shown that the classical added-mass is only applicable to singledegree-of-freedom bodies, whether in rigid-body motion or flexible
motion, and that the error in predicting the higher natural frequencies
of the system is due to this discrepancy.
Keywords: cylindrical structures, vibration, added mass

The presence of liquid in contact with a structure tends


to modify the dynamic properties of that structure,
namely: natural frequencies, mode shapes of vibration
and damping ratios. For thin-walled cylindrical structures such as liquid storage tanks and certain nuclear and
offshore structures these effects are of a more complex
nature than for structures with lesser flexibility. The
change in the dynamic properties of the structure is
generally attributed to an increase in the effective mass
of that structure due to the liquid, known as the
'added-mass'.
In the late eighteenth century Du Buat I published the
results of his observations on three spherical pendulums
vibrating in air and in water. He noted that in the
submerged case, the water had the same effect as increasing the mass of the sphere by about half the mass
of water which it displaced. A century later Lamb 2
presented a rigorous study of the added-mass for rigid
solids of different geometries. The added-mass concept
has since been used extensively to solve problems of
solid-fluid dynamic interaction in a number of

engineering disciplines including the ship building


industry, the space industry and earthquake engineering.
The first substantive use of the added-mass in earthquake engineering was made by Westergaard 3 which
adopted the concept to model the hydrodynamic
pressures exerted on a rigid dam. In an alternative way
of considering the hydrodynamic pressures, he suggested an equivalent mass of water (or concrete) which
when added to the mass of the dam itself would increase
its inertia (but without effecting its stiffness). Later
Jacobsen 4 and Housner 5 extended the Westergaard
approach to calculate the hydrodynamic pressures in
rigid cylindrical structures.
The earlier rigid-response assumption for practical
structures subjected to ground motion is, however, no
longer regarded as adequate in assessing their true
seismic behaviour, and the dynamic properties of structures should therefore be established. Experiments conducted on flexible structures, on the other hand indicated
that the natural frequencies of a structure in contact with
liquid are invariably less than the frequencies of that

0141-0296/92/030163-13
1992 Butterworth-Heinemann Ltd

Eng. Struct. 1992, Vol. 14, No 3

163

Modal vibration of cylindrical structures: M. R. Maheri and R. T. Severn

structure in air. A number of investigators sought the


answer in the added-mass concept and attributed the
reduction in the natural frequencies to an increase in the
effective mass of the system due to the liquid. They
regarded the increase as the same added-mass as that
evaluated for the structure when rigid. Problems of
hydrodynamic interaction in liquid storage tanks 6-9,
liquid-surrounded
cylindrical
structures 10-14 and
dam-reservoir systems 15-2 were thus solved. In their
analysis some noted, however, that the solution to the
structure-fluid system using this added-mass only yields
accurate results for the fundamental mode of vibration,
and that for the higher modes the results are considerably in error, reflecting the variability of the addedmass with the flexible mode of vibration.
In a theoretical study, Niwa and Chopra 21 utilized the
Eulerian formulation in the frequency domain to show
that the added mass as obtained for the rigid motion is
not an exact representation of the hydrodynamic effects.
The fundamental period of vibration of a watersurrounded tower, obtained using the added-mass concept, was reported in the above reference to err by about
3% compared to the correct solution. The error increased progressively with the mode number. Comparative numerical studies of the hydrodynamic effects
on dam structures carried out by the present authors, as
reported elsewhere 22, revealed similar conclusions. In
these studies the added-mass solution was checked
against the Eulerian solutions, as well as Lagrangian
solutions. Haroun and Housner 23 and Lee e t al. 24
carried out analytical studies of flexible storage tanks in
which they demonstrated that the pressure distribution
on a flexible tank is different from that of the same tank
when considered to be rigid. Haroun and Housner 25
also developed Housner's original rigid added-mass
model for cylindrical tanks 5 to include an extra
'flexible' mass due to the fundamental mode of vibration
only. The approximate masses were determined from
finite element solutions of liquid-filled tanks of different
geometries.

The added-mass effects on dams, considering their


flexibility, have also been studied in recent years 2627
and numerical methods have been developed to solve
this hydrodynamic interaction problem. Parallel studies
in the ship building industry have also considered the
interaction between the vessel and water. Jennings 28
presented a numerical method of solving the
fluid-structure coupled problem with specific application to ship hulls, taking account of their flexibility.
Considering the wide application of the added-mass
concept, evident in the number of theoretical and
analytical studies on the subject, surprisingly very few
accounts of experimental investigations, aimed at
substantiating the extent of it's participation in flexible
vibration, appear in the literature. In the present paper,
the results of a series of experimental studies are
reported which examine the nature and magnitude of the
added-mass in the modal vibration of three, groundbased steel cylindrical models. Only the added-mass
participating in the lateral modes of vibration (rigid and
flexible) is investigated. Masses associated with other
modes of the cylinders (rigid or flexible) are not considered. To evaluate the added-mass experimentally,
two approaches were sought
(i) by considering the role of the added-mass as a
parameter effecting the natural frequencies of the
cylinders and noting the changes in the frequencies
due to the liquid
(ii) by considering the correlation between the modal
accelerations and the corresponding hydrodynamic
pressures in each pure mode of vibration
Investigations were carried out on the models when only
containing water, and when only surrounded by water.
Due to practical limitations the fully submerged case
could not be studied. The first and second cylinders
(referred to in the text as small and slender cylinders,
respectively) are seamless hollow steel cylinders of
effective height-to-diameter ratios of 2.0 and 8.8, and

D----153108 ~

_= 2 6 1 ( a v e r o a e } ~_
t =2"286

571(overage)

----'"

1330
512

~235(overage)

I
L.

400

I--

Figure 1

164

"- 0 942

....

_1

b 3~,_.~

!_.~_

I -'-'-~ -20

Geometry of the cylinders. (a), Small cylinder; (b), slender cylinder; (c), large cylinder

Eng. Struct. 1992, Vol. 14, No 3

(overoge)

820

I600

Z20

M o d a l vibration o f cylindrical structures: M. R. Maheri and R. T. Severn

thickness-to-diameter ratios of 0.009 and 0.015, respectively. The third model (referred to as a large cylinder)
is basically, an open-ended, large oil barrel, purposely
manufactured without any circumferential stiffeners.
The effective height-to-diameter and thickness-todiameter ratios of this cylinder are 1.53 and 0.001,
respectively. The dimensions of the three cylinders are
shown in Figure 1.

Added-mass as a parameter effecting the


natural frequencies
Considering the free laieral vibration of a fixed-base
cylinder, the fundamental frequency of its vibration in
air, f . is give by
f~ = 1/27r(K~/M~)1/2

(1)

where Ks and Ms denote the effective stiffness and mass


of the cylinder in air, respectively. The fundamental frequency of the cylinder when in contact with water, fw,
may also be expressed as
fw = l121r(KwlMw) 1/2

Table 1 Natural frequencies of the small cylinder (Hz) n = lateral


mode number; m = circumferential mode number

Mode

Empty

100%
Water-surrounded

Water-filled

Exper.

Theor.

Exper.

Theor.

Exper.

Theor.

1
1
1
2

2
3
4
4

177.5
282.7
502.1
700.1

210.0
301.8
511.8
722.4

112.0
172.0
324.0
457.5

135.0
192.9
349.0
492.0

128.2
178.6
327.0
468.8

171.3
244.7
428.5
603.6

(2)

in which Kw and Mw are the effective stiffness and mass


of the cylinder-water system, respectively. Ignoring the
compressibility effects of water and assuming small
fluid displacements, the effective stiffness of the
cylinder in contact with water is shown to be the same
as that of the cylinder in air (i.e. Kw = Kw) 14. Therefore
from equations (1) and (2)

(3)

fdfw = (MwlM,)"2

The above expression shows that any decrease in the


natural frequency of a cylinder when in contact with
water is due only to an increase in the effective mass of
the system. Denoting the increase in the effective mass
(i.e. the added-mass) as Ma so that
Ma = Mw - M,

(4)

and the ratio of the added-mass to the actual mass of the


cylinder as o~, from equation (3) we obtain
a = MolM, = ~/fw) 2

cylinders when surrounded by water a steel water tank,


1.2 m deep and 3.6 x 4.0 m in dimension was constructed. The large concrete base of this tank was used
as a rigid foundation for mounting the cylinders. Summaries of the predominant natural frequencies evaluated
for the small, slender and large cylinders in each of the
three above-mentioned conditions are listed in Tables 1,
2, and 3, respectively. These frequencies correspond to
lateral modes of vibration (n = 1, 2, 3 . . .) associated
with
different
circumferential
wave
forms

(5)

Following the same line of argument it can be said that


the decrease in the natural frequencies of a multi-degreeof-freedom system in contact with liquid is also due only
to increases in its effective masses.
In order to study the value of the added mass as
described above, the natural frequencies of the cylindrical models in air and in water should first be
determined.

Table 2 Natural frequencies of the slender cylinder (Hz)


n = lateral mode number; m = circumferential mode number

Empty

Mode

100%
Water-surrounded

Water-filled

Exper.

Theor.

Exper.

Theor.

Exper.

Theor.

1
1
1
2
3

1
2
2
1
2

71.2
248.0
291.0
399.0
484.0

78.3
256.0
299.0
460.0
509.0

39.2
154.0
182.0
239.0
304.0

48.3
158.0
190.0
274.0
318.0

43.0
164.0
226.0
332.5

61.1
206.9
262.7
352.5
402.4

Table 3 Natural frequencies of the large cylinder (Hz) n = lateral


mode number; m = circumferential mode number
Mode
n

1
1
1
1
2

2
4
5
6
5

Empty

Water-filled

(Exper.)

(Exper.)

35.4
23.6
28.6
40.6
87.4

61.0
77.0
99.0
231.0

Table 4 Comparison between the theoretical added-mass in rigid


vibration and the added-mass in the fundamental mode obtained
using pressure/acceleration correlation

Case

Added-mass in the
fundamental mode
(kg)

Theoretical added-mass
in rigid vibration
(kg)

Water-filled
small cylinder

27.1

26.3

Water-filled
slender cylinder

24.9

24.4

Water-surrounded
small cylinder

20.9

19.8

Evaluation of natural frequencies


Natural frequencies of the cylinders when in air, when
surrounded by water and when containing water were
determined experimentally using the hammer tests 29
and forced-vibration tests. The former tests revealed the
natural frequencies, whereas their associated mode
shapes were measured in the latter tests. To test the

Eng. Struct.

1992,

Vol.

14, No 3

165

M o d a l vibration o f cylindrical s t r u c t u r e s : M. R. M a h e r i and R. T. Severn

the shell. The imperfections such as out-of-roundness


and variations in the thickness of the shell also effect the
natural frequencies of the lower wave forms.
For comparison with the experimental results, natural
frequencies of cylinders were also evaluated numerically
using the HKPROG finite element computer program 3~.
Like many recent numerical methods mentioned earlier,
this program is based on the Eulerian formulation in
which fluid motion is expressed as a pressure variable
governed by the Laplace equation. Appropriate elements
available in the program are one-dimensional, twonoded thin-shells of revolution, two-dimensional, fournoded thick-shells of revolution and two-dimensional,
six-noded fluid elements. The shell and fluid elements
are formulated specifically to discretize axisymmetric
systems for the non-axisymmetric vibrations considered
here 3~. The finite element mesh selected to model the
small cylinder in air consisted of 39 shell elements,
whereas, the small cylinder full of water was discretized
into 39 shell elements and 156 fluid elements 3. The
cylinder was also considered to be fully surrounded by
water, in which case the shell was idealized as 25 shell
elements and the fluid region was divided into 200 fluid
elements as seen in Figure 4. To avoid the problem of
reflected pressure waves, the water domain in this
analysis was assumed to extend to a radius of 1.9 m
from the shell. A number of evaluated frequencies of
this cylinder in the three conditions mentioned are listed
in Table 1. The slender cylinder in air was idealized as
80 shell elements. Selected frequencies are given in
Table 2. The cylinder was also analysed when containing 100%, 90%, 80%, 70%, 60% and 30% water and

(m = 1, 2, 3 . . .). To investigate the effects of varying


water level on the natural frequencies, hammer tests
were also performed on a slender cylinder when containing 30%, 60%, 70%, 80%, and 90% water, as well as
when surrounded by water levels 30%, 60%, 70% and
80% of its height. To give a graphic view of the effects
of the change in water level, these frequencies were plotted against the water level in Figure 2 for the watercontained condition and in Figure 3 for the watersurrounded condition. The mode shapes were evaluated
by subjecting each cylinder to steady-state forced vibration at the frequencies given by the hammer tests.
Responses were measured at 10 equidistant positions
along the height and 8, 16 and 32 positions around the
circumference for the small, slender and large cylinder,
respectively,
using the
moving accelerometer
technique 3.
The first circumferential mode shape (m = 1) of the
slender cylinder was found to be not of the expected
circular form. Also the first circumferential mode shape
of the small cylinder could not be excited in the tests
implying that its first lateral mode is of the m = 2 type
circumferential wave form. The m = 2 type circumferential mode for this cylinder in different conditions was also found to be distorted. The first resonance
frequency of the large cylinder was also found to correspond to the fourth circumferential wave pattern (m = 4),
itself of a distorted form. The inability to excite the
lower-order circumferential wave forms in thin-walled
cylinders and the fact that the lowest excitable modes are
mainly distorted versions of the theoretical wave forms
are thought to be due to the geometrical imperfections in

Hz

m=l

Hz

m=3

m=2

'

/,00" .

500 2

800

"Oo

~X

x...

300"

"0..

400"

"o

"0.

. O . ' ~ X

"0

too!

:-

' ......

0
0

o
"...

X~x~

. . . . . . . O..o," o..X.-T.x

. .
30

. . .
60
80

I00

% Full

"..X
. o , ~ X .,,,,~ 0
"" o . . ~ " . X

Eng. S t r u c t .

1 9 9 2 , V o l . 14, N o 3

''.. ,k

600

~n=~

500-

n=!--0

100
0

,
30

, , ,1,
60
80 I00
% Full

Figure 2 Effects of water level on the natural frequencies of water-filled slender cylinder. (
experimental; n, lateral mode number; m, circumferential mode number

166

~"

X-----.__X~n--l

n=3

" o o..o.~.

3o01....

200"

700

~00

I
0

11111

100

% Full
), t h e o r e t i c a l ;

( ....

o ....

),

Modal vibration of cylindrical structures: M. R. Maheri and R. T. Severn


Hz

~'~x

~00 ~

500
"o.

m=2

Hz

m=l

X~,.X,

n =2

0.

300

Hz I

m=3

~X~x

"0

o..

~00"

XX'~x,,~

"~:"

700 . . . . . "e

"0
". n--3
",
0

300 I

200.

X
"

~
X
.....

0 ....
"0..0.

200-

n=l
X ,..-.X .....v

~ X . . .
~ .

-o. ~ x
"

". .

. . . . . .
30
60 80 I00
% Water level

100
0

~x n--2
"..

~x

"~X*~ X
0, 0

x..~:.

n=2

"'o
X.~x

O.

~'~'X'~'X
"0

-C".O..o-'~

n=3
..

500-

".

0 n=2

n=l

. . . .
'J"t
30
60 80 100
~ Water level

400
0

. . . .
30
60 80
% Water level

100

Figure 3 Effects of the level of surrounding water on the natural frequencies of slender cylinder. (Key as in Figure 2)
Axis of symmetry~.~
25 s h e / / e l e m e n t s

2 0 0 f l u i d elements

, ,q

512ram

~2
L[_..

f l u i d nodes

2 6 s t r u c t u r a l nodes
8 x 237. 5 m m

.- J i/m3m0

Figure 4 Finite element idealization of water-surrounded small cylinder

Eng. Struct. 1 9 9 2 , Vol. 14, No 3

167

M o d a l v i b r a t i o n o f c y l i n d r i c a l s t r u c t u r e s : M. R. M a h e r i a n d R. T. S e v e r n

when surrounded by water assuming water levels of


30%, 60%, 70% and 80%. In the latter analyses the
radial extent of the water was considered as 1.0 m from
the axis of symmetry of the cylinder. To compare the
evaluated natural frequencies with those obtained from
the hammer tests, they are also plotted against the water
level in Figure 2 for the water-contained case and Figure
3 for the water-surrounded case. It should be noted in
Tables 1 and 2 and Figures 2 and 3 that the predicted frequencies are generally higher than the measured ones.
This reflects the overestimation of the system's stiffness
generally associated with finite element discretization.
The larger discrepancies between the finite element
prediction and the measured frequencies of lower modes
may be further explained by geometrical imperfections
in the shell affecting the lower-order circumferential
wave forms as discussed earlier.
As expected, a certain decrease in the natural frequencies of the cylinders due to water is noted in all the cases
studied. Reductions of between 31% to 39%, 37% to
40% and 54% to 63% in the experimental values were
noted for the water-filled small, slender and large
cylinders, respectively. In the case of the slender
cylinder in which different water levels were investigated, a general decreasing trend can also be
observed in the value of the natural frequencies with
increasing water level (Figures 2 and 3). This trend,
nevertheless, is not systematic and differs from mode to
mode. The varying and nonlinear decrease in frequencies indicate the frequency dependence of the
'added-mass' of water.
Comparing the experimental natural frequencies of the
cylinders when containing water and when surrounded
by water, a striking closeness in the values of corresponding frequencies is noted. The difference are all under
3.5% (most are under 1.0%) except the distorted first
mode of the small cylinder, where the difference is
6.4%. This is an interesting observation, implying the
same value of the 'modal' added-mass for the cylinders
when water-filled and when water-surrounded.

Evaluation of added-mass using equation (5)


To obtain the value of the added-mass in different
modes, the experimental natural frequencies of the
cylinders when in air and when in contact with water
were used in equation (5). The values of the added-mass
for each mode of the water-filled large cylinder as well
as for the water-surrounded small cylinder were thus
obtained. The results of these evaluations are plotted
against the frequency of vibration in Figures 5 and 6,
respectively. Both figures reveal a generally decreasing
trend in the added-mass with increasing modal frequency and shape number. The exceptions were the
values obtained for the fundamental modes which do not
follow the same trend as the higher modes. Figures 5
and 6 clearly demonstrate that the added-mass in a
flexible cylinder is a function of its natural frequencies
and mode shapes.

Added-mass as the correlation between


hydrodynamic pressures and accelerations
The modal acceleration of a cylinder against the body of
a fluid causes hydrodynamic pressures in the fluid which

168

Eng. Struct.

1992,

Vol.

14, No 3

120
(m=

I00

-~

I
I
I
I
I
I

8O
\

X rn=5

~o

.x.\ X

E
60

fro=5
rn=61
o

x" ".~,x~=s

"X. ~'~.

."
".~o~m=7

-^"

m=/

"~ .~

....~

"'0

"....ore=8

rn=8

~0

20

25

50

75

100

I25

Frequency (Hz)
Figure 5

Experimental hydrodynamic masses in the modal vibration of water-filled large cylinder. ( - - ) ,


Added mass from
pressure/acceleration;
( - ) , added mass from equation (5); n,
lateral mode number; m, circumferential mode number; (x),
n = 1; (o),n = 1

25.0

20.0

f rn=2

~X rn=3

U3
t~

E 15.0

\
~

,<

m:g
rn-~
~
o~ ,
~ ~rn=5~. --

~ m=5
_

I0.0
.'t t X'...

rn=6

5.0

0.0

200

400

600

800

1000

Frequency ( Hz)
Figure 6

Experimental hydrodynamic masses in the modal vibration of water-surrounded small cylinder. (Key as in Figure 5, also
( * ) , n = 3)

are at a maximum at the interface The magnitude of the


hydrodynamic pressure at a particular point on the interface is proportional to the acceleration of the structure
at that point. The correlation between the two

Moda/ vibration of cylindrica/ structures: M. R. Maheri and R. T. Severn

parameters (i.e. hydrodynamic pressure divided by


acceleration) is therefore related to a certain mass of
fluid which participates in the vibration of the system.
To investigate the nature and magnitude of this addedmass a series of vibration tests were also conducted on
the three cylinders in different conditions whereby the
accelerations
and
corresponding
hydrodynamic
pressures in a number of lateral modes of vibration were
measured.

Steady-state, fixed-base, forced vibration tests


The cylinders were subjected to sinusoidal vibrations at
frequencies corresponding to their natural frequencies.
A small electro-magnetic vibrator, attached to the
cylinders at their free end was found to be adequate to
excite the small and the slender cylinders in their pure
modes of vibration. The large cylinder was, however,
excited using the MAMA multi-point, forced-excitation
system 3. The cylinders were securely bolted down to
the concrete slab of the water tank, hence, vibrations at
the base were reduced to a minimum. In this way the
measured accelerations and pressures were those of
the flexible vibration of the cylinder in a particular mode
shape, corresponding to the frequency of vibration.
Tests were conducted on the cylinders when waterfilled, as well as when surrounded to their full height by
water, except for the slender cylinder in which levels of
up to only 80% were possible. For the water-filled small
cylinder, measurements were made when vibrating in its
predominant natural frequencies of 112.0, 457.5 and
854.0 Hz, corresponding to the n = 1, m = 2; n = 2,
m = 4; and n = 3, m = 5 modes, respectively. Due to
the support condition, the acceleration recorded at the
base of the cylinder in each case was negligible. The
accelerations and pressures measured along the height of
the cylinder when vibrating in the first two modes mentioned above are plotted in Figure 7. The correlations
between the measured hydrodynamic pressures and the
corresponding accelerations in each mode are also plotted in Figure 7. The correlations are obtained by
dividing the pressures by the accelerations at respective
points. The accelerations and pressures recorded around
the circumference of the cylinder were measured at the
xlO-2m
50
~5
~o
35
30
25
20
I5.
IO.
5

Accel.

Press.

Corr. curve

Accel.

/S

x
I

/
/
X
I
X
I

height of 0.45 m above the base. Results of these


measurements for a number of circumferential modes
are plotted in Figure 8. Figures 7 and 8 reveal close
similarities between the pattern of the recorded accelerations and the measured hydrodynamic pressures in any
particular mode. Similarities between the acceleration
trend and the impulsive hydrodynamic pressures due to
the modal vibration are more evident for the
measurements made around the circumference.
Eleven positions were found adequate to represent the
acceleration and the pressure distribution along the
height of the water-filled slender cylinder. Modes n = 1,
m = 2; and n = 3, m = 1 at the respective frequencies of
39.2, and 594.0 Hz were investigated. No measurable
vibrations could be noted at the base of the cylinder
during these tests, and consequently the hydrodynamic
pressures recorded at the base in each mode were
negligible. Results of the tests for the two modes mentioned above are plotted in Figure 9. This figure also
shows the correlations as evaluated in each mode
between the measured pressures and the accelerations.
The steady-state, fixed-base, forced-vibration tests on
the water-filled large cylinder were performed using the
MAMA system. In total, 12 predominant modes of
vibration, consisting of a number of 1st, 2nd and 3rd
lateral modes in association with a number of circumferential wave forms were investigated. Typical
modal accelerations and the corresponding hydrodynamic pressures recorded for these modes are plotted
in Figure 10. The correlation values for each mode are
also presented in Figure 10.
To investigate the pattern of hydrodynamic pressures
on the water-surrounded cylinders and their correlation
with the accelerations, steady-state, fixed-base, forcedvibration tests were also repeated on the small and
slender cylinders when surrounded by water. Each
cylinder was firmly fixed at its base to the central steel
plate of the water tank. The tank was, in turn, filled with
water to the described levels. The cylinders were then
subjected to sinusoidal vibration at their top and from
within. Accelerations and pressures were measured at
11 positions along the height of the small cylinder utilizing the moving transducer technique 3. In total 10 tests
were conducted measuring the accelerations and

p
,. . - X ~

I
!

I
X%

b ~
N/m 2

Corr. curve

a
g

Press.

N/m2/g

" ~." :~" ;~ ~

~
~

N/m 2

to

N/m2/g

Modal accelerations and the corresponding h y d r o d y n a m i c pressures measured along the height of the water-filled small cylinder.
(a), mode, n = 1, rn = 2, f r e q u e n c y = 1 1 2 . 0 Hz; (b), mode, n = 2, m = 4, f r e q u e n c y = 4 5 7 . 5 Hz

Figure 7

Eng. Struct. 1992, Vol. 14, No 3

169

Modal vibration of cylindrical structures: M. R. Maheri and R. T. Severn

m=2
m..-3
. ~o

"

x" ~ - 9 - - -Z~<
..... I / / ~

.....

,. . . . . x~'.

~0'~.\X".
4g

x_,f

,
j

"~ ~ -d~. --- - "


"',

..~..

"

'"

"

o
o

rn-4
.~k..

m=5

I
o

,7

"

",

0
0

.?

l|".
i..

.'.

I
o

..'

O
Q

Figure 8

Modal accelerations and the corresponding hydrodynamic pressures measured around the c;~rcumference of the water-filled
small cylinder. ( . . . x . .), pressures; ( - - o - - ) , accelerations; scale: pressure 1 m m = 2 0 0 N m - ; acceleration 1 m m = 1 . 0 g

hydrodynamic pressures for 10 modes of vibration.


Typical results obtained from these tests are plotted in
Figure 11. Because of the limited depth of the water
tank, a fully water-surrounded situation could not be
modelled for the slender cylinder. Instead, water levels
of 60% and 80% were investigated. Pressures at 6 and
8 positions on the interface were measured for the two
water levels, respectively. At each water level the
cylinder was vibrated in a number of its modes and the

170

Eng. Struct. 1992, Vol. 14, No 3

accelerations and the resultant hydrodynamic pressures


were recorded for each mode. Typical results of these
measurements are plotted in Figure 12 for the 60%
water level, and in Figure 13 for the 80% water level.
The correlation factors between the pressures and the
accelerations were also calculated for each mode as
plotted in these figures.
Results of the steady-state, fixed-base forced vibration
tests described above, demonstrate that the hydro-

M o d a l vibration o f c y l i n d r i c a l structures: M. R. M a h e r i and R. T. Severn


xlO-2m
135 "!

Accel.

Press.

?
I
o

5~

x
8
X
!

I
o
!

27

x
8
x
I

"o,

X,

X
I

P
0

Corr. c u r v e

.x~

/d
d

81

Press.
jo

108.

AcceL

Corr. c u r v e

'

~'

c5 ~
g

N/m 2

N/rn 2 / g

N/m 2

N / r n 2/g

Figure 9 Selected modal accelerations and the corresponding h y d r o d y n a m i c pressures, water-filled slender cylinder. (a), mode: n = 1,
m = 1, f r e q u e n c y = 3 9 . 2 Hz; (b), mode: n = 3, m = 1, f r e q u e n c y = 5 9 4 . 0 Hz

dynamic pressures due to the flexible response of a


cylinder in a pure mode, follow the same pattern of
distribution as the mode shape itself The exception
being at the free end, in which the pressures are zero.

x tO-2m

AcceL

82"0 t
65"6 t

328 t

Press.

,J

'o

,e ~1

Accet

82. 0 "
I

65.6.

o
I

xt

32. 8"
16.4.
O=C

N/m 2

Accet
o
/
o
o
I
o
I
o

Corr. curve

I
!
I

Press.

o
I
o
!
o
I
o
t
o

e,l

49.2.

N/m2

x10"2m

'

%.

0 L.

t
!
!

I
x

Cork curve

x
I
x
I
x
I
x

I
o
I
o
I

Accel.

o'

~9 21

Corr. curve

Press.

o
#
o
I
o
!
o
i

Also the correlation curves as shown in Figures 7 - 1 3


are of an interesting pattern in the sense that they are
very similar to the added-mass distribution pattern for
the rigid vibration. This suggests that the distribution

!
!

N / m 2/g

Corr. curve

Press.
.%

X
!

I
x
I
x

!
|

t
!

x
#

A I
IA

I
!

'x

I
!

o
|
i

I
x
IN
IN
N/m2

o
!
,

N/m2/g

(:~ c)

d:~c5 ~(:5

N/m2/g

N/m 2

Figure 10

Selected modal accelerations and h y d r o d y n a m i c pressures, water-filled large cylinder. (a), mode: n = 1, m = 5,
f r e q u e n c y = 2 8 . 6 Hz; (b), mode: n = 2, m = 7, f r e q u e n c y = 8 0 . 2 Hz; (c), mode: n = 2, m = 5, f r e q u e n c y = 8 7 . 4 Hz; (d), mode: n = 3,
m = 6, f r e q u e n c y = 1 8 5 . 2 Hz

Eng.

Struct.

1992,

Vol.

14,

No

171

M o d a l vibration o f cylindrical s t r u c t u r e s : M. R. M a h e r i and R. T. Severn

xtO-2rn Accel.
50
o
!
45"

Cofr

!0
o

5'

I
--

I
I
I
I

X
I
0

I
X

o
i

xlO-2m

Accel.

10
5
0 -C

L~

X
I
X
i

0
I
0
!
0

X
I
X

~1 1

%
%

Press,.

Corr. curve

01

X
I

io

,o,

I
X
I
,

N/m2/g

I
X
i
X

IN .,T

N/m2

0
!
0

0
!

Accel.

Corrcurve

Press.
0
I

35
30.

N/m 2

50
45
40

25
20
15

I
, , i

I
X

/
t

x
I

I0"

20"
15"

X
I

!
0

Corr. curve

Press.
0

25"

Accel.

curve

"X

I
0

40.
35"
30"

Press.

X
J

io

t
X
I
X

d
!
0

!
I

X
%

o~

t~

N/m 2

N / m 2/g

N/rn 2

N/m2/9

F i g u r e 11
S e l e c t e d m o d a l a c c e l e r a t i o n s and h y d r o d y n a m i c p r e s s u r e s , w a t e r - s u r r o u n d e d s m a , c y l i n d e r . (a), m o d e : n = 1, m = 2,
f r e q u e n c y = 1 2 8 . 2 Hz; (b), m o d e : n = 2, m = 5, f r e q u e n c y = 6 2 9 . 4 Hz; (c), m o d e : n = 1, m = 3, f r e q u e n c y = 1 7 8 . 6 Hz; (d), m o d e :

n = 2, m = 6, f r e q u e n c y = 8 7 6 . 6 Hz

pattern of the added-mass may be independent of the


type of motion or mode of vibration.
Evaluation of added-mass using pressure~acceleration
relationship

To evaluate the magnitude of the added-mass in each


mode, the correlation curve obtained from its respective
test was integrated along the height of the cylinder. The
results for a number of modes of the water-filled large
cylinder and the water-surrounded small cylinder are
also plotted against the frequency of vibration in Figures
5 and 6, respectively. Close agreements may be observed in the figures between these masses and the
added-mass evaluated previously using equation (5).
The agreement appears to increase with the increasing
mode number. The discrepancies in the value of the two
masses for the first mode is, nevertheless, very high.
This is evident both in the case of the water-filled large
cylinder and the water-surrounded small cylinder.
For a closer study of the correlation between the
measured accelerations and pressures, the correlation

172

Eng. Struct. 1992, Vol. 14, No 3

factors between the two parameters for the first lateral


mode of the water-filled small cylinder were plotted in
Figure 7. By integrating this correlation curve along the
height of the cylinder, a value of 27.1 N.s2/m (=kg) is
obtained. On the other hand, using the classical addedmass concept, the mass of water participating in the rigid
vibration of the cylinder is evaluated as 26.3 kg. Also
the correlation curve for the first mode of the waterfilled slender cylinder is given in Figure 9. The value
obtained from the integration of the correlation curve for
this mode is 24.9 kg, whereas the added-mass for rigid
vibration of this cylinder is calculated as 24.4 kg.
Similarly the integrated correlation curve for the first
mode of the water-surrounded small cylinder (Figure
11) gives a value of 20.9 kg which again is close to the
added mass evaluated for the rigid cylinder in this condition. These results are summarized in Table 4. As may
be noted, there are close agreements between the values
of the masses in the first lateral modes of the cylinders
and those of their respective added-masses in rigid-body
vibration, the similarities also exist in the pattern of
distribution of the parameters.

Modal vibration of cylindrical structures: M. R. Maheri and R. T. Severn

xtO%

Acce/.

135'"

Press.

Corr. curve

Press.

Acce/.

108.

I
!

81.

X
I
X

I
I

t
l

27-

O--

N/m 2

xlO-2rn Accel

~/m~/g

Press

Corr. curve

N/m2

Accel.

N/m~/g

Press.

'

I
t

135

54-

Corr. curve

Corr. curve

!
0

108'

81
I

I
27

o-C

Xt

/
X

/
~'~," &
~ ~
g

sX

~J

'
N/m 2

'--~%

I
x

/X

/"

g.J

dR
N/rn2/g

N/rn2

N/m2/g

Figure 12

Selected m o d a l accelerations and h y d r o d y n a m i c pressures, 6 0 % w a t e r - s u r r o u n d e d slender cylinder. (a), m o d e : n = 1, m -- 5,


f r e q u e n c y = 5 1 . 0 Hz; (b), m o d e : n = 2, m = 7, f r e q u e n c y = 2 5 1 . 0 Hz; (c), m o d e : n = 1, rn = 2, f r e q u e n c y = 1 8 0 . 0 Hz; (d), m o d e :
n = 2, m = 3, f r e q u e n c y = 6 1 3 . 0 Hz

A general conclusion drawn from the above manipulations is that the constant added-mass as noted by Du
Buat in oscillation of pendulums and later formulated by
Lamb in, for example, the rigid vibration of an infinitely
long circular cylinder submerged in water is only a true
representation of the hydrodynamic effects for singledegree-of-freedom bodies (whether in rigid-body motion
or flexible motion). In multi-degree-of-freedom bodies
the added-mass is not only a function of geometry of the
body and density of the liquid, but also a function of the
dynamic properties of the system such as natural frequencies and mode shapes. This is believed to be the
reason why previous investigators using the added-mass
concept as m rigid-vibration have obtained accurate
results only for the first mode of vibration.
A second point to note is that some investigators in the
past have used equation (5) to determine the rigid-body
mass experimentally. This expression, however, corresponds to the added-mass in a purely flexible vibration
at a particular natural frequency and as was shown
earlier (Figures 5 and 6) its use in determining the rigidbody added-mass would lead to erroneous results.

Conclusions
An accurate experimental method of evaluating the
hydrodynamic mass in flexible cylinders is found by
considering the correlation between the hydrodynamic
pressures and the corresponding accelerations.
Manipulation of the test results and the evaluated masses
leads to the following conclusions. The amount of the
hydrodynamic mass in the fundamental mode represents
the classical added-mass in rigid-body vibration, leading
to the further conclusion that the classical added-mass is
applicable to single-degree-of-freedom bodies only,
both in the rigid-body motion and in flexible motion.
.The hydrodynamic mass associated with each pure mode
is a function of the frequency and mode shape of
vibration and appears to decrease with increasing frequency and mode number. It is also concluded that,
unlike its quantity, the distribution pattern of the
hydrodynamic mass in modal (pure flexible) vibration is
independent of the frequency or mode shape, and
remains the same as the distribution pattern in rigidbody motion.

Eng. Struct. 1992, Vol. 14, No 3

173

M o d a l vibration o f cylindrical s t r u c t u r e s : M. R. M a h e r i and R. T. Severn

x~O-2m
135..

Accel.
?

Press.

Corr. curve

Accel.

I
I

I
!

?
!

54

27

~ ~

N/rn2/g

Press.

1'

/
/

IN

(5 ~

/
x

,i ~

N/m2/g

Corr. curve

(1(
,

Press.

/'

N/m2

AcceL

Corr. cur ve

27.

1
I

Accel.
q

54

%x

N/m~

81

X.,.

b ~

I
I

x..

"0~

O-a

108"

x lO -2m
135.

curve

/J:'

108"

Cork

Press.

x/

.~'

,
~1

u~

. .

tn

N / m 2/ g

N/m2

d~

c5 r~
g

N/rn 2

r~

N / m 2/g

Figure 13 Selected modal accelerations and h y d r o d y n a m i c pressures, 8 0 % w a t e r - s u r r o u n d e d slender cylinder. (a), mode: n = 1, m = 1,
f r e q u e n c y = 4 3 . 0 Hz; (b), mode: n = 3, m = 2, f r e q u e n c y = 3 3 2 . 5 Hz; (c), mode: n = 2, m = 2, f r e q u e n c y = 2 2 6 . 0 Hz; (d), mode:
n = 4, m = 2, f r e q u e n c y = 8 4 6 . 5 Hz

References
1 Stelson, T. E. and Mavis, F. T. 'Virtual mass and acceleration in
fluids', Trans. ASCE, 1957, 122, Paper No. 2870
2 Lamb, H. Hydrodynamics, Cambridge University Press, 1932 (6th
edn)
3 Westergaard, H. W. 'Water pressures on dams during earthquakes'
Trans ASCE, 1933, Paper No. 1835
4 Jacobsen, L. S. 'Impulsive hydrodynamics of fluid inside a cylindrical tank and of fluid surrounding a cylindrical pier' Bull. Seism.
Soc. Amer., 1949, 39, t 8 9 - 2 0 4
5 Housner, G. W. 'Dynamic pressures on accelerated fluid containers',
Bull. Seism. Soc. Amer., 1957, 47, 15-35
6 Baron, M. L. and Skalak, R. 'Free vibrations of fluid-filled cylindrical shells' Proc. ASCE, 1962, EM3
7 Shaaban, A. S. and Nash, W. A. 'Finite element analysis of a
seismically excited liquid' Report NSF/RA-760261, Univ. of
Massachusetts, Amherst, Mass., 1976
8 Balendra, T. and Nash, W. A. 'Earthquake analysis of a cylindrical
liquid storage tank with a dome by finite element method', University
of Massachusetts, Amherst, Mass., 1978
9 Kalnins, A. 'Free vibration of rotationally symmetric shells' J.
Acoust. Soc. of Amer., 1964, 36, 1355-1365
10 Clough, R. W. 'Effects of earthquakes on underwater structures',
Prof. 2nd Worm Conf. on Earthq. Eng., 1960, Tokyo, Japan
11 Laird, A. D. K. 'Water effects on flexible oscillating cylinders',
Proc. ASCE, 1962, W~/3, 125-137
12 Goto, H. and Toki, K. 'Vibrational characteristics and aseismic
design of submerged bridge piers', Proc. 3rd World Conf. Earthq.
Eng., New Zealand, 1965, 2, 107-122

174

Eng.

Struct.

1992,

Vol.

14,

No

13 Skop, R. A., Ramberg, S. E. and Ferer, K. M. 'Added mass and


damping forces on circular cylinders', Proc. ASME, 1976, Paper No.
76-PET-3
14 Chandrasekaran, A. R., Saini, S. S. and Malhotra, M. M. 'Virtual
mass of submerged structures', Proc. ASCE, 1972, HY5,
8 8 7 - 896
15 Kotsubo, S. 'Dynamic water pressure on. dam during earthquakes',
Proc. 2nd World Conf. on Earthq. Eng., Japan, 1960
16 Zienkiewics, O. C. 'Hydrodynamic pressures due to earthquakes', J.
Water Power, Sept. 1964, 382-388
17 Bustamente, J. I. and FIores, A. 'Water pressure on dams subjected
to earthquakes', Proc. ASCE, 1966, EM5'
18 Nath, B. 'Coupled hydrodynamic response of a gravity dam', Proc.
Inst. of Cir. Eng., 1971, 48, 245-257
19 Fin, W. D. L. and Varoglu, E. 'Dynamic of dam-reservoir systems',
J. Comput. Struct., 1973, 3, 913-924
20 Chakrabarti, P. and Chopra, A. K. 'Earthquake analysis of gravity
dams including hydrodynamic interaction', Earthq. Eng. and Struct.
Dyn., 1973, 2, 143-160
21 Niwa, C. and Chopra, A. K. 'Earthquake response of axisymmetric
tower structures surrounded by water', EERC Report No. 73-25,
Oct. 1973
22 Maheri, M. R., Taylor, C. A. and Blakeborough, A. 'Comparative
numerical studies of the hydrodynamic effects on dam structures',
Proc. 8th European Conf. Earthq. Eng., Lisbon, 1986, 3,
6.8/25 -6.8/32
23 Haroun, M. A. and Housner, G. W. 'Dynamic characteristics of
liquid storage tanks', Proc. ASCE, 1982, E M 5
24 Lee, S. C., Liaw, C. Y. and Tung, C. C. 'Earthquake response of
sea-based storage tanks', Appl. Ocean Res., 1983, 5, 150-157

Modal vibration of cylindrical structures: M. R. Maheri and R. T. Severn


'5 Haroun, M. A, and Housner, G. W. 'Seismic design of liquid storage
tanks', Proc. ASCE, 1981, TC1
!6 Hall, 1. F. and Chopra, A. K. 'Hydrodynamic effects in the dynamic
response of concrete gravity dams', Earthq. Eng. and Struct. Dyn.,
1982, 10, 333-345
'.7 Hall, J. F. and Chopra, A. K. 'Dynamic analysis of arch dams ineluding hydrodynamic effects', ASCE, 1983, 109, EM1
28 Jennings, A. 'Added mass for fluid-structure vibration problems', Int. J. for Num. Meth. Fluids, 1985, 5, 817-830

29 Maheri, M. R. and Severn, R. T. 'Impulsive hydrodynamic pressures


in ground-based cylindrical structures', J. Fluids Struct., 1989, 3,
555 -577
30 Maheri, M. R. 'Hydrodynamic investigations of ground-based cylindrical structures and other structure-fluid systems', PhD Thesis,
Dept. of Civil Eng., Univ. of Bristol, 1987
31 Karadeniz, H. 'The theoretical and experimental dynamic analysis of
thin shells of revolution', PhD Thesis, Dept. of Civil Eng., Univ. of
Bristol, 1976

Eng. Struct. 1992, Vol. 14, No 3

175

Das könnte Ihnen auch gefallen