22nd International Conference

ENGINEERING MECHANICS 2016

Svratka, Czech Republic, 9 – 12 May 2016

INFLUENCE OF WALL FLEXIBILITY OF LIQUID STORAGE

TANKS ON HYDRODYNAMIC PRESSURES INDUCED DURING
SEISMIC ACTIVITY

M. Musil*, M. Sivý**

Abstract: The paper deals with the seismic analysis of the circular vertical ground-supported tanks containing
liquid with the aim to determine the distribution of hydrodynamic pressures along the tank height during
seismic excitation. Circular tanks are used for storage of liquids in various industrial sectors. Hence, it is a
request for satisfactory performance during dynamic loadings. In general, these tanks are usually made of
reinforced concrete or steel. For determination of distribution of hydrodynamic pressures there must be
distinguished between rigid and flexible tanks due to different responses of tank-liquid systems. In cases of
flexible tanks, the wall flexibility is taken into account. The paper is also dedicated to compare results of
distribution of hydrodynamic pressure for both cases assuming models with identical geometric proportions.
For computation and subsequent comparison Eurocode 8 standard and other procedures are used.

Keywords: circular liquid storage tank, hydrodynamic pressures, flexibility, seismic excitation,
Eurocode 8

1. Introduction
Liquid storage tanks are important components in various industrial sectors and are used as part of
lifeline transmission and distribution systems. The main purpose of these systems is to store a variety of
liquids (e.g. with toxic explosive nature) before subsequent treatment or utilization. Liquid storage tanks
can be subjected to loadings of various nature which may threaten ordinary operation. In addition to static
loading, which is usually represented by hydrostatic pressure, tanks may be subjected to the dynamic effects
(e.g. seismic excitation) as well. There are many negative consequences which may be caused by
dynamically loaded systems and can take one of the following ways: buckling of shells due to excessive
axial compression taking the form of a bulge called elephant’s foot; roof damage caused by sloshing of
upper zone of contained liquid with inadequately designed freeboard between liquid surface and roof etc.
There were proposed procedures for seismic resistance of tank-liquid systems. One of the basic seismic
characteristics is the distribution of hydrodynamic pressures along the tank height. Time-dependent
hydrodynamic pressures induce time-dependent stresses in tanks which can significantly affect their
performance. Hence, estimate of distribution and magnitude of hydrodynamic pressures during loading are
crucial in tank design in order to preserve their functionality before, during and after seismic event. These
pressures and relative stresses depend on type of motion, liquid properties, proportions and material of tank.

2. Basic concepts and theory

Following the past earthquakes like El Centro, San Fernando and others, there was a huge effort to
describe tank behavior in a more common way. The most widely used procedure for evaluating seismic
effects in cylindrical liquid storage tanks is the one based on a spring-mass equivalent model (Housner,
1954). The proposed method shows that hydrodynamic pressures can be expressed as the sum of two
contributions called impulsive and convective, separately. The impulsive pressures are those associated
with inertial forces produced by accelerations of the walls of the container and are directly proportional to

* Doc. Ing. Miloš Musil, CSc.: Institute of Applied Mechanics and Mechatronics, Faculty of Mechanical Engineering, Slovak
University of Technology, Námestie slobody 17; 812 31, Bratislava; SK, [email protected]
** Ing. Martin Sivý: Institute of Applied Mechanics and Mechatronics, Faculty of Mechanical Engineering, Slovak University of

Technology, Námestie slobody 17; 812 31, Bratislava; SK, [email protected]

433
these accelerations. The convective pressure is produced by oscillation of liquid and is a consequence of
the impulsive pressure. However, Housner’s method is limited to the rigid tanks. It is usually
unconservative to consider all tanks to be rigid due to the wall deflection. Whereas walls of rigid tanks
move in unison as the ground, motion of walls of flexible tanks is different. The wall deflection
configuration of these tanks do not remain the same but vary from the base to the top of the tank (Veletsos,
1974). From procedures it follows, maximum acceleration in a given direction of a point on a rigid tank is
the same as maximum acceleration of the ground, whereas in cases of flexible tanks it can be smaller than,
equal to or greater than the maximum ground acceleration depending on tank flexibility. All these
knowledge are reflected in seismic characteristics (base shear, overturning moment, stresses etc.). Tank
flexibility affects the hydrodynamics effects and these may be increased significantly. Following
investigation in the past, there were proposed procedures which underlie international standards in relation
to design of structure for seismic resistance (e.g. Eurocode 8).
Following subchapters of paper are focused on the comparison of procedures for estimate of distribution of
hydrodynamic pressures in rigid and flexible tanks in accordance with Eurocode 8 and other approaches.
An investigated liquid storage tank is illustrated in Fig. 1. It is the cylindrical tank of radius R (8 m), height
H (16 m) and thickness t (0,004 m). Tank is filled to the height HL (16 m) with liquid (water). It is assumed
that the base of the tank is continuously fixed to gravel-sand foundation, which is excited by a horizontal
component of ground acceleration of 1 ms-2 with component importance factor III in accordance with
Eurocode 8.

z
y

H x
HL
θ
R

𝑥 (𝑡)

Fig. 1: Liquid storage tanks under horizontal seismic excitation

2.1. Hydrodynamic pressures in rigid tanks

During lateral base excitation, tank wall is subjected to the lateral hydrodynamic pressure. For rigid
tanks in accordance with Eurocode 8, hydrodynamic pressures consists of two components analogous to
Housner’s theory. The spatial-temporal distribution of the rigid impulsive pressure is given by expression
𝑝i (𝜉, 𝜍, 𝜃, 𝑡) = 𝐶i (𝜉, 𝜍)𝜌L 𝐻L 𝐴g (𝑡) cos 𝜃 (1)
where Ci gives the distribution along the height of pi
(−1)𝑛 𝑣
𝐶i (𝜉, 𝜍) = 2 ∑∞
𝑛=0 cos(𝑣𝑛 𝜍)𝐼1 ( 𝑛 𝜉)
𝐼1′ (𝑣𝑛 ⁄𝛾 )𝑣𝑛2 𝛾

in which
2𝑛+1 𝐻L 𝑟 𝑧
𝑣𝑛 = 2
π; 𝛾= 𝑅
; 𝜉 = 𝑅; 𝜍 = 𝐻
L

𝐼1′
𝐼1 and represent a modified Bessel function of first kind and its derivative, 𝐴g (𝑡) denotes the ground
acceleration time-history, 𝜉, 𝜍 are non-dimensional geometric parameters and 𝛾 denotes slenderness
parameter (liquid height 𝐻L to tank radius 𝑅). The part of liquid that does not move as a rigid body with the
tank experiences sloshing effect. This motion can be expressed as a linear combination of the corresponding
natural modes of liquid vibration. The spatial-temporal distribution of the convective pressure component
is given by expression
𝑝c (𝜉, 𝜍, 𝜃, 𝑡) = 𝜌L ∑∞
𝑛=1 𝜓𝑛 cosh(𝜆𝑛 𝛾𝜍) 𝐽1 (𝜆𝑛 𝜉)cos 𝜃𝐴c𝑛 (𝑡) (2)

3
434
where
2𝑅
𝜓𝑛 =
(𝜆2𝑛 −1)𝐽1 (𝜆𝑛 ) cosh(𝜆𝑛 𝛾)

J1 is a Bessel function of the first order and 𝜆𝑛 are values for which the first derivative of J1 is zero, Acn (t)
is the acceleration time-history of SDOF oscillator with natural frequency ωcn. To withstand roof damage
caused by sloshing in the tank, a sufficient freeboard is required. The sloshing wave height may be given
from the following expression
𝑟
2 𝐽1 (𝜆𝑛 𝑅) 𝑆e (𝑇c𝑛 )
𝑑(𝑟, 𝜃, 𝑡) = 𝑅 ∑∞
𝑛=1 𝜆2 −1 𝐽 (𝜆 ) 𝑔
cos 𝜃 (3)
𝑛 1 𝑛

where Se (Tcn ) is the elastic spectral acceleration at first convective frequency. In Fig. 2, distributions of
maximum impulsive and convective pressure along the tank height of a described model are shown.

a) b)

Fig. 2: Distribution of impulsive and convective pressures in the model of rigid tank

2.2. Hydrodynamic pressures in flexible tanks

According to the standard Eurocode 8, hydrodynamic pressure of flexible tanks is usually expressed as
the sum of three contributions namely as rigid impulsive, flexible and convective. Flexible contribution
satisfies the conditions of radial velocity of liquid along the height to be equal to the deformation velocity
of the tank wall, zero velocity at the tank bottom and zero pressure at the free surface of liquid. Due to weak
coupling between the second and the third component of hydrodynamic pressure, the flexible component
may be determined independently. Weak coupling is a consequence of large differences between natural
frequencies of sloshing liquid and tank-liquid system. The rigid impulsive and sloshing component in (1)
and (2) remain unaffected. Assuming the modes of vibration f (ς), the flexible pressure distribution on the
walls takes form
𝑝f (𝜍, 𝜃, 𝑡) = 𝜌L 𝐻L 𝜓 cos 𝜃 ∑∞
𝑛=0 𝑑𝑛 cos(𝑣𝑛 𝜍) 𝐴f𝑛 (𝑡) (4)
where
1
∫0 𝑓(𝜍) cos(𝑣𝑛 𝜍)𝑑𝜍 𝐼1 (𝑣𝑛 ⁄𝛾 )
𝑑𝑛 = 2
𝑣𝑛 𝐼1′ (𝑣𝑛 ⁄𝛾 )
1 𝜌 𝑡(𝜍)
∫0 𝑓(𝜍)[𝜌 s 𝐻 +∑∞ ′
𝑛=0 𝑏𝑛 cos(𝑣𝑛 𝜍)]𝑑𝜍
𝜓= 1
L L
𝜌 𝑡(𝜍)
∫0 𝑓(𝜍)[𝜌S 𝐻 𝑓(𝜍)+∑∞
𝑛=0 𝑑𝑛 cos(𝑣𝑛 𝜍)]𝑑𝜍
L L

(−1)𝑛 𝐼1 (𝑣𝑛 ⁄𝛾 )
𝑏𝑛′ = 2
𝑣𝑛2 𝐼1′ (𝑣𝑛 ⁄𝛾 )

ρS is the mass density of the tank, t (ς) is the tank thickness and Afn (t) is the pseudo-acceleration function
corresponding to the natural frequency of the tank-liquid system. In most cases of flexible tanks, the
pressure pf provides predominant contribution to the total pressure, due to the fact, that while the rigid
impulsive pressure (1) varies with ground acceleration, the flexible component varies with response
acceleration Afn (t), which may be significantly greater than maximum ground acceleration. Procedure for
flexible tanks proposed by A. S. Veletsos (Veletsos, 1974) assumes the tank-liquid system vibrates in a
fixed configuration along its height with no distortion of its cross-section. For deflection, it takes three
π z
different configurations between base and free surface liquid, like half-sine wave function ψA = sin ( 2 H ),
L

4
435
 z π z
linear function ψB = H and reversed half-sine function ψC = 1- cos ( 2 HL
), where z represents coordinate
L
along the tank height. Fig. 3 a) represents mode of vibration of the tank-liquid system when liquid oscillates
in unison with the tank and represents unfavorable response to the system which behaves like uniform
cantilever flexural beam. Pressure distribution of flexible contribution of impulsive pressure along the tank
height is shown in Fig. 3 b). For estimate of pressure distribution, maximum response acceleration of the
tank-liquid system having the period of impulsive mode is assumed. It compares pressure distributions
according to the Eurocode 8 and Veletsos’s approach. Good conformity can be seen when comparing
Eurocode 8 and deflection configuration ψA . Using another mentioned deflection configuration in
Eurocode 8 computations, similar results as those shown in Fig. 3 b) will be attained.

a) b)

ψC

ψB

ψA
Eurocode 8

Fig. 3: Impulsive mode and distribution of flexible pressure contribution in the model of flexible tank

3. Conclusions
The aim of this paper was to estimate hydrodynamic pressures in the model of liquid storage tank and
to compare acquired results when assuming rigid and flexible tank walls, respectively. It can be concluded
that impulsive pressure in rigid tank increases from zero at liquid surface to a maximum at the base, whereas
convective pressure reaches maximum at the liquid surface and decreases with depth. It was shown that
flexibility of tank wall has impact on the impulsive liquid that experiences accelerations that are greater
than peak ground acceleration. Therefore there is a different impulsive pressure distribution assuming
additional flexible component which has tendency to shift a peak ordinate from the base to the top as 𝛾
increases. Convective pressure due to a weak coupling is not affected by wall flexibility. For design
purposes, these effects can be evaluated by considering the tank to be rigid. To sum up, the base shear and
overturning moment of flexible tanks calculated by assuming the tank as rigid can be non-conservative.

Acknowledgement
The paper was supported by grant from Grant Agency of VEGA no. 1/0742/15 entitled "Analysis for
seismic resistance of liquid storage tanks with nonlinear and time-dependent parameters".

References
Eurocode 8 (1998) Design Provisions of Earthquake Resistance of Structures, Part 4: Silos, Tanks and Pipelines.
European Committee for Standardization, Brussels.
Housner, G. W. (1954) Earthquake Pressures on Fluid Containers. Eight Technical Report under Office of Naval
Research. California Institute of Technology, Pasadena, California.
Ibrahim, R. A. (2005) Liquid Sloshing Dynamics Theory and Applications. Cambridge University Press, Cambridge,
UK, 948 p. ISBN 05-111-2492-9.
Malhotra, P. K., Wenk, T. & Wieland, M. (2000) Simple Procedure for Seismic Analysis of Liquid Storage Tanks, in
Structural Engineering, IABSE, Vol. 10, No.3, pp. 197-201.
Musil, M. & Sivý, M. (2015) Dynamic Analysis of Liquid Storage Tank Using FE Method and Results Comparison
with Analytical Models, in: ISET 2015, pp. 13-21.
Veletsos, A. S. (1984) Seismic Response and Design of Liquid Storage Tanks. Guidelines for the Seismic Design of
Oil and Gas Pipeline Systems. Technical Council on Lifeline Earthquake Engineering (ASCE), New York, USA.

5
436