d

3r International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,

28-30 June 2012, Near East University, Nicosia, North Cyprus

The influences of a liquid storage tanks and soil characteristics

under seismic loads
Naeini, S.A
Associate Professor, Civil Engineering Department, Imam Khomeini International University, Qazvin,
Iran

Valaei, A
Construction and Development of Transportation Infrastructure Company (CDTIC)

Hossen Zade, M
M.Sc. Student, Civil Engineering Department, Imam Khomeini International University,
Qazvin, Iran

KEYWORDS: Liquid storage tank, foundation, seismic behavior, wave propagation

ABSTRACT: The different seismic behavior of liquid storage tanks rather than conventional
structures make their responses more complicated. The tank seismic behavior can be improved by
modifying dynamic characteristic of tank with verifying seismic loads as well as retrofitting and
improving base ground. As a matter of fact, uses of liquid storage tanks because of the simple
construction on compact layer of soil as a foundation are very conventional. This paper focuses on a
typical steel tank on medium, loose and stiff soil and describes an evaluation of displacement of the
tank when the depth of the tank is constant. Because of the assessment of earthquake response of
such structures, finite element methods are indispensable tools since they offer a concise way of
accurately considering all these nonlinearities in the model. Three dimensional tank was modelled by
software of ABAQUS.The compact layer of soil, as the tank base, has been modelled and the
influences of tank and soil characteristics that are effective on tank seismic behavior have been
considered. The results shows that, by increasing shear strength parameter of soil, the performance of
the liquid storage tank under the case of seismic load was improved and necessity of retrofitting tank
decrease.
1 INTRODUCTION
Due to the requirement of remaining functional after a major earthquake event, the seismic
performance of liquid storage tanks has been a matter of special importance .The most common type
of tank is the vertical cylindrical tank. Damages to liquid storage tanks in past earthquakes motivated
several experimental and analytical investigations of the seismic response of vertical cylindrical
tanks. Some simplified Mass Spring Model (MSM) has been proposed for predicting tank responses,
such as base shear force and overturning moment. The most frequent failures of the steel tanks are
related to shell buckling near the bottom of the tank wall, where large compressive membrane
stresses are induced to resist the overturning moment from the earthquake shaking of the system. If
the sloshing freeboard is not properly accounted in the design procedure, sloshing of the contained
liquid can also cause yielding in the connection of the tank roof and its wall. Other types of damage,
including failure of the piping system connected to the tank (due to its inability to follow the shell
movement and rupture at the junction of the tank wall and the base plate) may appear due to
excessive plastic conditions. Nearly all the models used until recently in seismology for predicting
ground motion induced by earthquakes have been based on the assumption of linear elastic behavior
of the soil. On the other hand, for a number of years nonlinear soil implication has been routinely

339

3r International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,

28-30 June 2012, Near East University, Nicosia, North Cyprus

taken into consideration in geotechnical engineering practice (Seed & Idriss, 1984). The main reason
seismologists had in the past ignored the possibility that nonlinear phenomena could play an
important role in earthquake ground motion was that compelling evidence for nonlinear effects in the
observed motion, other than in liquefiable sites, was scarce. In the last decade, however, a number of
accelerograms have been recorded during strong earthquakes that have made it possible to infer
nonlinear response. The most common manifestations of inelastic soil behavior involve the reduction
in shear wave velocity and the increase in soil damping with increasing load (Hardin & Drnevich,
1972).In general, investigations on the seismic response of liquid storage tanks have been conducted
over the past 30 years. Housner (1954, 1957) proposed a simple MSM for computing the seismic
response of liquid storage tanks which is still widely used with certain modifications for the analysis
of rectangular and cylindrical tanks. His simplified MSM is a two degree-of freedom (DOF) system
for a rigid tank; one DOF accounting for the motion of the tank-liquid system, in which a part of the
contained fluid being rigidly attached to the tank wall (impulsive mode) and the other DOF for the
motion of the sloshing fluid effect on the tank wall (convective mode).In further studies, Housners
simplified MSM has been modified to account for the flexibility of the tank wall. Veletsos and Yang
(1976) used one mass for the impulsive component and two convective mass in their simplified
MSM .Haroun and Housner (1981) divided the impulsive mass into two parts; one part rigidly
connected to the ground and one part representing the mass participating in the relative movement
due to the deformation of the tank shell. Malhotra et al. (2000) modified the properties of the
simplified MSM proposed by Veletsos and Yang (1976) using one convective mode .Uplifting of
unanchored tanks, as well as soil structure interaction effects, has been also extensively studied by
several researchers (Natsiavas;1988, El-zeiny;1998,2003, Fisher;1979).Some of above mentioned
works have constituted the basis for the seismic design provisions for vertical cylindrical tanks in
Euro code 8-part 4.3 and American Petroleum Institute (API). In this study a typical steel tank on
loose, medium and dense sandy soil was analyzed and settlement, sliding and amplification effect for
all three types of soil were illustrated.

2. FINITE ELEMENT MODEL (FEM)

2.1. MASS SPRING MODEL (MSM)

The simplified Mass Spring Model (MSM) proposed by Malhotra et al (2000) is illustrated in Figure
1. The procedure was based on the work of Veletsos and his co-workers with certain modifications
including; i) Combining the higher impulsive modal mass with the first impulsive mode and
combing the higher convective modal mass with the first convective mode. ii) Modifying modal
heights of mass to account for the contribution of higher modes to the base overturning moment. iii)
Generalizing the formula for the impulsive period so that it could be applied to tanks of various wall
thicknesses. The effects of liquid-structure interaction and an uncoupled manner by dividing the
hydrodynamic pressure acting on the shell into two components;

340

The influences of a liquid storage tanks and soil characteristics under seismic loads
Naen, S.A., Valae, A. & Hossen Zade, M.
.

Figure 1. Simplified Mass Spring Model (MSM) proposed by Malhotra et al. (2000).

1) The impulsive pressure caused by the portion of the liquid, mi, which is rigidly attached to the
shell wall, and
2) The convective pressure caused by the portion of the liquid, mc, sloshing in the tank.
These components were then modeled as single DOF oscillators. For very large values of fluid
height to tank radius (H/r), the sloshing mass is only a small portion of the total mass. As H /r less
than unity, more than half the total mass can participate in the convective mode. The values proposed
by Malhotra for the parameters of the simplified model can be obtained from equations 1 and 2 as
well as Table 1.Where ml is the total mass of the liquid, hi and hc are the respective heights of the
resultant force of the hydrodynamic pressure due to the motion of the impulsive mi and the
convective mc masses, respectively, is the mass density of liquid, ci and cc are constant and related
to the fluid height to tank radius.
(1)
(2)
In equations 1 and 2, h is the wall thickness, E is the modulus of elasticity of the tank material, Timp
and Tcon are the mass density of the liquid, the periods of the impulsive and convective modes,
respectively,
Table 1. Parameters of the simplified MSM (Malhotra et al.2000)

In this research the tank was modeled from the Malhotras assumptions, the radius of the tank is
2.5m and the height of the liquid is 6.5m, value of fluid Height to tank Radius (H/r) is 2.6 thus the
parameters of the simplified MSM can be obtained from Table 1. The influences of the three types of
sandy soil (dense, medium, and loose) as a foundation of the tank under the specific time history
acceleration were aimed. The property of dense, medium and loose sand was illustrated in Table 2.
Table 2. Properties sands used for 3D finite element.
dense

medium

Youngs modulus (MPa)

500

200

100

Poissons ratio()

0.35

0.35

0.4

(deg)

38

30

25

Undrained shear strength

Cu(Kpa)

10

341

loose

3r International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,

28-30 June 2012, Near East University, Nicosia, North Cyprus

2.2. FEM STRATEGY

The finite element software ABAQUS was used for describing the behavior of soil. Figure 2 shows
3D finite element mesh used in this analysis. Relatively fine mesh is occupied near the surface while
a coarser mesh was used for further distance from the tank. Three types of soils were modeled by
using Mohr-Coulomb criteria .For tank-soil contact, the modeling of the tank-soil interfaces is an
important concern. Therefore one of the main issues is identifying interaction between soil and tank.
When a compressive normal pressure (p) applied on the bottom plate of the tank, tank can only
transfer shear forces along their lateral surfaces. When contact take places, according to modified
Coulombs friction theory, the relationship between shear force and normal pressure is shown in
equation 3.
=p

(3)

Where is friction coefficient and p is normal pressure that varied in each level of soil. As reported
by Jeong et al (2003) the interface friction coefficient () for sand varies from 0.4 to 0.6. Therefore,
in this study interface friction coefficient () of 0.5 for all the types of sand was adopted.

Figure 2. 3D finite element model

In order to represent the half-space soil medium with explicit finite elements, the near field is
expected is modeled with explicit 3-D brick type elements. The lateral boundary of the near field
model should be extended sufficiently far such that the outgoing wave due to the structural vibration
diminishes drastically at the boundary. To prevent any reflection of outgoing waves at the boundary,
a series of artificial viscous dampers are attached to the boundary. In ABAQUS, the approach
developed by Lysmer and Kuhlemeyer (1969) was implemented, in which viscous normal and shear
stresses are applied to the boundaries in a manner as defined in the following equations:
normal = - cdVnormal
shear = - csVtangential

(4)
(5)

Where , cd, and cs are the material density, material longitudinal and shear wave velocities of the
transmitting media. These equations reveal that the magnitude of these stresses at the boundaries is
proportional to the particle velocities in the normal (Vnormal) and in the tangential (Vtangential)
directions.
The Lysmers dampers placed on the artificial boundary are effective in reducing unwanted wave
reflections if the boundary of the finite element mesh is sufficiently far outward. However, in doing
so, the size of the near field finite element mesh is increased significantly and so is the cost of
running the dynamic analysis. As it was mentioned the unbounded or infinite medium can be
approximated by extending the finite element mesh to a far distance, where the influence of the

342

The influences of a liquid storage tanks and soil characteristics under seismic loads
Naen, S.A., Valae, A. & Hossen Zade, M.
.

surrounding medium on the region of interest is considered small enough to be neglected. This
approach calls for experimentation with mesh sizes and assumed boundary conditions at the
truncated edges of the mesh and is not always reliable. It is particularly of concern in dynamic
analysis, when the boundary of the mesh may reflect energy back into the region being modeled. A
better approach is to use infinite elements: elements defined over semi-infinite domains with
suitably chosen decay functions. Abaqus provides first- and second-order infinite elements that are
based on the work of Zienkiewicz et al. (1983) for static response and of Lysmer et al. (1969) for
dynamic response. The elements are used in conjunction with standard finite elements, which model
the area around the region of interest, with the infinite elements modeling the far-field region. As it
was shown in Figure.2 in the seismic load direction infinite element was used.

3 RESULTS AND DISCUSSION

Liquid storage tanks are important elements of lifeline and industrial facilities. The evolution of
codes and standards for the seismic design of these structures has relied greatly on observations of
tank damages during past earthquakes, yet the time lag between acquiring the information and
implementing the findings in practice has remained relatively long. Even though, current codes and
standards reflect a mature state of knowledge for tank design, recent earthquakes as well as advanced
state-of-the-art analyses continued to point out to a few overlooked issues. Both failure of the tank
and failure of the foundation of the tank were caused the necessity of tank retrofitting.
In this study, a nonlinear numerical technique based on finite element method is employed for the
seismic analysis of unanchored steel liquid storage tanks by focusing on the tank foundation. Large
differential settlement and sliding of the tank caused considerable damage and made the tank out of
work. If the estimated differential settlements are large enough to cause problems, measures can be
taken to reduce their magnitudes through treatment of the foundation. In marginal cases, it may be
more desirable to measure settlements carefully during first filling of the tank, and then institute
mitigative measures only if the predicted potentially damaging settlements are confirmed by field
observations Abaqus procedure is utilized to consider the interaction forces between tank and soil.
The complex interaction mechanism of unanchored tank base plate and soil is taken into account
with contact algorithm including friction forces.

343

3r International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,

28-30 June 2012, Near East University, Nicosia, North Cyprus

Figure 3. Time-acceleration on the ground: (a) dense, (b) medium and (c) loose sand

Figure 4. Sliding of the tank in: (a) dense, (b) medium and (c) loose sand

344

The influences of a liquid storage tanks and soil characteristics under seismic loads
Naen, S.A., Valae, A. & Hossen Zade, M.
.

Figure 5. Settlement of the tank in: (a) dense, (b) medium and (c) loose sand

As it was shown in Figure 3, 4 and 5 by increasing module of elasticity and soil strength
parameters, the measured acceleration on the ground was decreased. It means by improving the soil
strength parameters the amplification effect was reduced (Table 3).All three types of the soil couldnt
resist the tank from sliding. This point revealed that tall tank was recommended to build as an
anchored tank. On the other hands increasing the soil strength parameters had a neglect able effect on
sliding resistance, it was also reasonable because friction coefficient is constant, thus resistance force
was nearly constant. By considering the rate of settlement during the earthquake (Figure 5), although
reducing soil strength parameters caused increasing settlement of the tank, the rate of settlement
couldnt make the tank out of work.
Table 3.Maximum acceleration, sliding and settlement
dense
Acceleration (g)

medium

loose

0.8

1.1

1.5

Sliding (m)

0.0165

0.0165

0.016

Settlement (m)

0.0063

0.018

0.0285

4 CONCLUSIONS
Liquid storage tanks are vital construction that studying the behavior of them during the seismic load
is essential. As well as the behavior of soil under the tank is important, because by decreasing the
soil strength parameter the amplification effect and the settlement of the tank increase.
According to the result increasing the sliding resistance force for the tall tank is necessary and
retrofitting of the tank foundation for this case could be a solution. However it should be better
constructed tall tank as an anchored tank.
Also in this study there is no focus on the differential settlement, but for decreasing the sliding
movement of the tank some strategies necessary. To clarify retrofitting of the tank foundation, more
study is necessary.
REFERENCES
American Petroleum Institute (API), (1998). Welded Storage Tanks for Oil Storage, API 650, American
Petroleum Institute Standard, Washington D.C.
El-Zeiny, A. A., (1998), Development of Practical Design Guidelines for Unanchored Liquid Storage Tanks,
Doctoral thesis, Department of Civil and Geometrics Engineering and Construction, California State
University, Fresno.
El-Zeiny, A. A., (2003), Factors Affecting the Nonlinear Seismic Response of Unanchored Tanks,
Proceedings of the 16th ASCE Engineering Mechanics Conference, Seattle.
Euro code 8, (1998)," Design provisions of earthquake resistance of structures", Part 4: Silos, tanks and
pipelines. European Committee for Standardization, Brussels.
Fisher, F. D., (1979), Dynamic Fluid Effects in Liquid-Filled Flexible Cylindrical Tanks, Earthquake Eng.
Structure Dyn, 7, pp. 587601.
Hardin, B.O, & Drnevich, V. P, (1972)." shear modulus and damping in soils measurement and parameter
effect .J soil Mech. Found, ASCE98 (6)-603-624
Haroun, M. A. & Housner, G. W., (1981), Seismic Design of Liquid Storage Tanks, Journal of Technical
Councils, ASCE, Vol. 107, pp. 191-207.
Housner, G. W., (1954), Earthquake Pressures on Fluid Containers, Eighth Technical Report under Office of
Naval Research, Project Designation No. 081095, California Institute of Technology, Pasadena,
California.
Housner, G. W., (1957), Dynamic on Accelerated Fluid Containers, Bulletin of the Seismological
Society of America, Vol. 47, No. 1, pp. 1535.

345

3r International Conference on New Developments in Soil Mechanics and Geotechnical Engineering,

28-30 June 2012, Near East University, Nicosia, North Cyprus

Jeong, S.G.. Seo, Y.K, & Choi. K.S (2003), "Design Charts of Piled Raft Foundations on Soft Clay"
Proceedings of the 13th International Offshore and Polar Engineering Conference: Honolulu, Hawaii, USA.
Lysmer, J. & Kuhlemeyer, R. (1969), Finite Dynamic Model for Infinite Media, Journal of Eng. Mech.
Div. ASCE, EM4, pp 859-877.
Malhotra, P. K., Wenk, T. & Wieland, M., (2000), Simple Procedures for Seismic Analysis of Liquid
Storage Tanks, Structural Engineering International, IABSE, Vol. 10, No. 3, pp 197-201.
Natsiavas, S., (1988), An Analytical Model for Unanchored Fluid-Filled Tanks under Base Excitation,
ASME J. Appl. Mech., 55, pp. 648653.
Seed, H.B, & Idriss, I.M.(1984), Soil Moduli and Damping Factors for Dynamic Response Analysis of
Cohesion less Soils, Report No. UCB/EERC-8914, Earthq. Eng. Research Center, University of California,
Berkeley, CA.
Veletsos, A. S. & Yang, J. Y., (1976), Dynamics of Fixed-Base Liquid Storage Tanks, Proceedings of U.S.
Japan Seminar for Earthquake Engineering Research with Emphasis on Lifeline Systems, Tokyo, Japan, pp.
317- 341.
Zienkiewicz, JP de SR Gago & DW Kelly, (1983), "The hierarchical concept in finite element analysis",
Comput. Struct. 16.

346