Ocean Engineering 38 (2011) 1161–1173

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

Effects of backfill on seismic behavior of rectangular tanks

Ramazan Livaoglu a,n, Tufan Cakir b, Adem Dogangun a, Mustafa Aytekin c
a
Department of Civil Engineering, Uludağ University, 16059 Bursa, Turkey
b
Department of Civil Engineering, Gümüs- hane University, 29000 Gümüs- hane, Turkey
c
Department of Civil Engineering and Architecture, University of Bahrain, Kingdom of Bahrain

a r t i c l e i n f o abstract

Article history: The reliability and/or stability of the lifeline structures against failure under seismic loads are of critical
Received 13 July 2010 concern, and must be studied carefully. Therefore, the main objective of this paper is to demonstrate
Accepted 15 May 2011 the commonly encountered backfill effects on the dynamic response of rectangular tanks. However,
Editor-in-Chief: A.I. Incecik
only the exterior wall of the tank which interacts with both the backfill and fluid is tackled, as each part
Available online 8 June 2011
of the structure shows considerable differences in terms of both the load bearing mechanisms and the
Keywords: geometrical and positional differences. Finite element analyses are employed, taking into consideration
Rectangular tanks the fluid–wall–backfill interaction. The analyses are conducted to observe whether or not both backfill
Fluid–tank wall–backfill interaction and wall behavior can be affected by variation of the internal friction angle. For that purpose, some
Internal friction angle
comparisons are made on vertical displacements of the backfill, roof displacements, stress responses,
Dynamic response
etc., by means of internal friction angle variations of the backfill from 251 to 401. Consequently, it is
observed that the variations on maximum vertical displacements are affected considerably. In contrast,
the maximum stress responses are affected partially. However, the inertial effects of the backfill show
that pseudo-static approximations may be insufficient to understand the dynamic behavior of the
backfill–wall–fluid system.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction A multitude of studies have been carried out on the seismic

behavior of liquid storage tanks, most of them are concerned with
Liquid storage tanks are extensively used across the world and ground level cylindrical tanks. However, the behavior of rectan-
the design of these structures is an important topic of research for gular tanks during seismic loading has been studied by very few
civil engineers. Again, the devastating effects of the earthquakes researchers. The first report on the analytical and experimental
make the problem more complicated compared to the static observations of rigid rectangular tanks under simulated horizon-
design procedure for liquid storage tanks. Therefore, the stability tal earthquake excitation was given by Hoskins and Jacobsen
of the liquid storage tanks under earthquake conditions must be (1934). Subsequently, the spring–mass analogy was used to
studied carefully. The wall movement causes distortion or even model the fluid in a rectangular container by Graham and
collapse of the system. Although this form of failure is not as Rodriguez (1952). Housner (1963) proposed a widely accepted
dramatic as other types of earthquake damages, the seismic analytical model, wherein the hydrodynamic pressures were
behavior of these types of structures is an important design separated into impulsive and convective components. This model
problem in the seismic regions. Otherwise, earthquake damage was mostly used for the horizontal motion. An extended applica-
to tanks may take place in several forms and result in a variety of tion of Housner’s procedure, in the sense of a practical design rule,
unwanted consequences, such as, shortage of drinking water, was given by Epstein (1976). From various backfill geometries
large fires, and substantial economical losses (Livaoglu, 2008). The encountered in practice, Livaoglu et al. (2007), investigated to see
recent earthquakes in Turkey, like the 1999, Mw ¼7.4 Kocaeli and whether the dynamic behavior of the rectangular tank walls was
the 1999 Mw ¼7.2 Düzce earthquakes, have added important affected or not, and they emphasized that the backfill system,
dimensions to this problem. As seen in city centers, the drinking based on elastic FEM, had considerable effects on the dynamic
water tanks are of utmost importance. behavior of tank walls. Livaoglu (2008), evaluated the dynamic
behavior of a rectangular tank considering the fluid–structure–
n
soil/foundation interaction altogether by means of changing soil/
Corresponding author. Tel.: þ902242941961; fax: þ 902242941903.
E-mail addresses: [email protected] (R. Livaoglu),
foundation conditions, and concluded that the displacements and
[email protected] (T. Cakir), [email protected] (A. Dogangun), base shear forces were affected by soil stiffness. Hashemi et al.
[email protected] (M. Aytekin). (2010) conducted a free vibration analysis of vertical rectangular

0029-8018/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.oceaneng.2011.05.017
1162 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173

Mindlin plates resting on Pasternak elastic foundation, which appeared to be satisfactory provided the inertia of the wall was
were fully or partially in contact with fluid on one of their sides, taken into account.
and investigated them for different combinations of boundary Experiences from damage to soil-retaining structures showed
conditions. They also purposed a numerical approximation having that few cases of damage or collapse to soil-retaining walls had
high accuracy and small computational cost. Dynamic analyses of been reported and the retaining systems supporting unsaturated
rectangular tanks were carried out using a Lagrangian fluid finite soils had performed well during many recent earthquakes except
element (Dogangun et al., 1996). Chen and Kianoush (2005) to quay walls and bridge abutments (Nazarian and Hadjian,
conducted a parametric study stating that flexibility of a tank 1979). Waterfront structures, especially quay walls, seemed to
wall should be considered when calculating hydrodynamic pres- have been more susceptible to failures because of the added
sure. Kianoush and Chen (2006) also studied the importance of effects of hydrodynamic pressures (Nazarian and Hadjian, 1979).
the vertical component of ground motion on the overall seismic In the Düzce Earthquake (1999) with a magnitude of 7.2
behavior of the rectangular tank, and they suggested that the (Mw ¼7.2), a rectangular-drinking water tank, with a volume of
vertical component of ground motion should be considered 10,000 m3, failed moderately and lost its container capability.
especially for tanks close to the field zone, for not experiencing Most of the earth-retaining structures can be classified into
the earlier mentioned unwanted events. Flexibility of the tank two groups—displacement-governed walls and force-governed
walls and hydrodynamic pressure acting on the wall were walls. However, the exterior walls of drinking water structures
considered by some researchers (Chen and Kianoush, 2005; are neither displacement-governed nor force-governed walls,
Kianoush and Chen, 2006; Dogangun et al., 1997; Kim et al., because of the displacement sensitivity at the construction point
1998; Dogangun and Livaoglu, 2004). Similarly, Kim et al. (1996) of the wall and force sensitivity at the design process of the wall.
carried out a study including wall flexibility and they stated that The problem investigated here is also quite specific, as the wall is
due to comparisons made with the 3D hybrid boundary element- an imposed finite medium of the backfill, different from that
finite element method, a two-dimensional model may represent given in literature and the other type of retaining structures. On
the fluid–structure interaction effect of the rectangular tank. the other hand, there is no literature that proposed the analysis of
To determine the dynamic characteristics of rectangular tanks, a backfill–rectangular drinking water storage tank’s wall–fluid
experimental studies were conducted by Minowa (1984), system under the combined actions of forces induced by fluid and
Koh et al. (1998), Faltinsen et al. (2003), and Akyildiz and Unal soil interactions. The mentioned studies so far also show that
(2005, 2006). The influence of the direction of the earthquake relatively little work has been done on rectangular storage tanks.
motions on the hydrodynamic loads of a fluid-filled rectangular Some studies for waterfront retaining walls were carried out,
tank and elevation of the associated amplitudes of the water considering both hydrodynamic pressures and seismic forces
surface were investigated by Isaacson and Ryu (1998). Park et al. acting simultaneously on the wall, in which pseudo-static meth-
(2000) investigated the dynamic behavior and seismic design ods were proposed (Ahmad and Choudhury, 2008; Choudhury
consideration of base isolated pool-type tanks for the storage of and Ahmad, 2007). The present studies on the design purpose of
nuclear spent fuel assemblies using numerical analysis and drinking water tanks given in the literature are not satisfactory, so
shaking table experiments. In addition, there were no recorded new studies with different approaches to the problem are
studies of rectangular tank uplift behavior, but Priestley et al. extremely essential to understand the seismic behavior of the
(1986) suggested a procedure, which was based on the principles system under consideration. Therefore, to appreciate the signifi-
for cylindrical tanks. cance of a backfill and the variation of its friction angle, finite
The problem of earth pressure on a retaining structure is element analyses that cover backfill–tank wall–fluid system have
commonly encountered in the design process. In technical litera- been performed by considering both the backfill and fluid
ture, the design trend can basically be broadly divided into three interaction.
categories based on the approach and the theory used by various
investigators: (1) fully plastic (static) solution; (2) solution based
on elastic wave theory; and (3) solution based on elastoplastic 2. Description of the rectangular tank system under
and nonlinear theory. Almost all the investigations in these fields consideration
up to the early 80 s were widely evaluated on state-of-the-art, by
Nazarian and Hadjian (1979). The pioneering work, which In this study, the structural properties of the prismatic rein-
was commonly used and known as the Mononobe–Okabe method forced concrete rectangular storage tank, with a container capa-
(M–O) (see Kramer, 1996) was developed by Okabe (1924) and city of 10,000 m3, which is frequently constructed in Turkey, are
Mononobe and Matsuo (1929), in which they used an extension of considered, as shown in Fig. 1. For that purpose, a Kökc- üoğlu
the Coulomb wedge theory in dry cohesionless soil by considering typical drinking water tank constructed in Samsun, Turkey, has
the pseudo-static seismic accelerations that included additional been selected as a reference structure. The rectangular tank under
vertical and horizontal seismic inertial forces of the soil. Since consideration has two main divisions. The length and the width of
then, a great deal of research has been performed to evaluate its the structure are 61 and 39.2 m, respectively. The total height of
adequacy and to improve it. Seed and Whitman (1970) focused on the reference tank is 7.4 m from the bottom of the foundation.
a limit-state design and they used the modified Mononobe–Okabe The other characteristics such as the dimensions of the tank and
analysis, an extension of the Coulomb–Rankine sliding wedge the foundation system are shown in Fig. 1. Here, it is suitable to
theory. Mylonakis et al. (2007) proposed a solution that was say that for the other tanks with different capacities, the exterior
essentially an approximate yield-line approach, based on the wall geometry and structural properties may be approximately
theory of discontinuous stress fields and took into account identical; however, the fluid container geometry may vary from
different parameters from the others. However, these methods each other due to container capacity and design differences.
of analysis only gave satisfactory results if the wall displacements Therefore, the dynamic behavior of another tank with different
were sufficient to fully mobilize the shear strength. Therefore, container capacity constructed in another place may differ both in
Richards and Elms (1979) suggested a new design approach, terms of the local soil and design properties. In the analyses,
based on the initial choice of specified limiting wall displacement, Young’s modulus, Poisson’s ratio, and the weight of concrete per
for a displacement-governed wall, like gravity walls. They showed unit volume are taken to be 28,000 MPa, 0.2 and 25 kN/m3,
that for gravity walls, the Mononobe–Okabe analysis thus respectively. The container is filled with water, with a density of
 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173 1163

0.5 m

2m
6.0 m
A

3 9.
A

61.0 m

Construction joint 0.8 m 6.1 m

A-A profile with respect to symmetry axis 3.0 m 0.9 m 1.6 m

9.60 9.40 9.40

7.50 cm
gravel
5.5 m (max)

20 cm
backfill
6.0 m soil
30 cm
Leakage slab
preventing tape

0.8 0.9 m
1.6 0.9 3.0 4.55 1.4 0.5 1.4 6.10 1.4 0.5 1.4 4.725 2.55 0.9

30.5m

Fig. 1. Considered tank and the exterior wall properties.

600
E-W COMPONENT
Ground acceleration (m/s2)

400

200

-200
amax=3.22 m/s2
-400

-600
0 5 10 15 20 25 30 35 40 45
Time (s)

Fig. 2. Horizontal ground acceleration recorded at YPT station.

1000 kg/m3. In such structures, the maximum service level (liquid According to this study, the corner frequencies were found at
level) taken into account is 5.5 m. 0.75, 0.78, and 0.61 Hz for NS, EW, and UD components, respec-
In the seismic analysis, the Kocaeli earthquake motion (1999) tively. It is worth stating here that to investigate the relation
with a magnitude of 7.4 (Mw ¼7.4), recorded at Yarimca site (YPT) between the dominant frequencies of the ground motion and the
is considered. The horizontal YPT-EW (east–west) acceleration fundamental frequencies of the systems examined, which is out of
history shown in Fig. 2 is used in the dynamic analysis of the the scope of this study, is required in case the problem is
tanks. The peak horizontal ground acceleration at the YPT station ascertained as eigenvalue problem. Thus, it can be realistically
was 0.32 g. The frequency analyses for N–S, E–W, and U–D determined how the structure under investigation can be affected
components of YPT signals were carried out by Durukal (2002). from the ground motion.
1164 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173

The exterior wall of the rectangular tank considered in this (2007) is developed mainly for building structures, so its basic
study interacted with the various soil types encountered in provisions could be used to design or evaluate the adequacy of the
practice. To evaluate the dynamic response of the system, dry buildings. According to almost all the codes about tanks and/or
cohesionless, silty, sand soil type (SM) with four different internal earth-retaining structures, the stability and design of the walls
friction angles, as shown in Table 1, was considered. are estimated based on pseudo-static approaches: The Indian
The nonlinear time history analyses were carried out by using the Standard code of practice uses the M–O method of analysis and
above-mentioned system. Rayleigh damping was used in the ana- does not consider permissible displacement. Contrary to this, the
lyses. The seismic analysis of the system was carried out for each of Eurocode 8, 1993 version had used the Richards-Elms displace-
the four internal friction angles in case of, ‘‘full of fluid’’ and ‘‘without ment based model (Choudhury et al., 2004). However, the present
fluid’’. Similar analyses were also carried out for no backfill in case of version of EC-8 part-5 (2003) includes the M–O method. Further-
‘‘full of fluid’’ and ‘‘no fluid’’. All these models and their properties more, EC-8 part-4 (2006) requires that in partially backfilled or
that are considered during the analysis are given in Table 1. buried tanks, permanent loads include, in addition to the weight
of the structure, the weight of the earth cover and any permanent
external pressures due to groundwater. Similarly ACI 350.3
3. Backfill–exterior wall–fluid interaction model (2001) suggests that dynamic earth pressures will be taken into
account when computing the base shear of a partially or fully
Currently there is no Turkish code available that is directly buried liquid-containing structure, and when designing the walls,
related to the design of tanks. The Turkish Earthquake Code the dynamic backfill forces will not be relied upon to reduce the

Table 1
Properties and abbreviations of the considered model and the specifications.

Name of the model F BF25 BF30 BF35 BF40 W B25 B30 B35 B40
Backfill  ’ ’ ’ ’  ’ ’ ’ ’
Fluid ’ ’ ’ ’ ’     

Soil properties
f  25 30 35 40  25 30 35 40
E (MPa)  20 20 20 20  20 20 20 20
u  0.3 0.3 0.3 0.3  0.3 0.3 0.3 0.3
g (kN/m3) 19 19 19 19 19  19 19 19 19

E: Young’s modulus, f: internal friction angle; u: Poisson’s ratio,  : symbolize nonexistence, ’: symbolize existence.

Drucker-Prager
-σ3 φ > 0 σ1 = σ2 = σ3

3c cot φ

-σ 2
-σ1 1.6 m

Soil Finite Element

12 m Structural Finite Element
y z
F
x

(Dn; Fn) Fluid Finite Element

Unidirectional-coupled nodes between

Nonlinear-Unidirectional Element structural and fluid elements
3D - Artificial Boundary

Fig. 3. The considered backfill–structure–fluid interaction model.

 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173 1165

dynamic effects of the stored liquid or vice versa. However, the As is known, the simulation of the infinite medium in the
codes of the tanks do not address the specific methods on how the numerical method is a very important topic for the dynamic soil–
effect of backfill can be taken into account. structure interaction problems. In some cases, the backfill under
Modeling the backfill–structure–fluid system, the finite element consideration may also be finite, and for the other circumstance the
method is used as shown in Fig. 3. The considered system is assumed dimension perpendicular to the wall may be larger than the dimen-
to be supported on a rock foundation. The structural wall is modeled sion of the wall. Therefore, the general method of treating this
with the solid elements having three degrees-of-freedom at every problem is to divide the infinite medium into the near field
node for a roof system with quadrilateral shell element (four nodes, (truncated layer), which includes the irregularity as well as the
six degrees-of-freedom per node), also with an additional mass of non-homogeneity of the soil adjacent to the structure, and the far
cover. To model a backfill–wall interaction, a unidirectional element field, which is simplified as an isotropic homogeneous elastic
with nonlinear generalized force-deflection capabilities is used in the medium (Wolf and Song, 1996). The near field is modeled using
analysis. The element has a longitudinal or torsional capability in finite elements and the far field is treated by adding some special
1-D, 2-D, or 3-D applications. The longitudinal option is a uniaxial artificial boundaries or connecting some special elements. The more
tension–compression element with up to three degrees of freedom at appropriate approximations include utilization of the artificial and/or
each node: with translations in the nodal x, y, and z directions transmitting boundaries. Furthermore, the reflecting and radiation
(ANSYS, 1994). Natural backfill behind the exterior wall of the effects of the propagating waves from the structure-foundation layer
structure interacts with the wall in compression, but no interaction may be avoided by means of these types of boundaries. There are
is assumed in tension. Therefore, an attempt has been made to different types in frequency or time domain, with different sensitiv-
model this interaction in this scope, through the study. Next, the ities. First, Lysmer and Kuhlemeyer (1969) developed a viscous
unidirectional nonlinear element is used, having very rigid compres- boundary, using the one-dimensional beam theory and this theory
sion characteristics, tensionless on the interaction face of the has been commonly used with FEM (Livaoglu and Dogangun, 2007).
backfill–wall system. It is also seen that no bending and torsion are In this study, the viscous boundary is used for three dimensions, to
assumed. It must be emphasized here that the vertical friction consider the radiational effect of the seismic waves through the soil
between the wall and backfill is ignored. In order to characterize medium in a direction perpendicular to the normal of wall (Fig. 3).
the fluid–rectangular tank–backfill model and determine the seismic To analyze all the aspects of the seismic response from the
behavior of the system, a transient dynamic analysis was carried out backfill–wall–fluid system is currently impossible due to the complex
by means of the ANSYS commercial package program. All the interacting phenomena and the inherent variability and uncertainties
elements mentioned above are available in ANSYS; the fluid ele- of the soil properties. In addition, soil behavior is very sensitive in
ments are specially formulated to model fluid contained within a general when it is exposed to earthquake-induced motion. Thus, the
container having no net flow rate. Mathematical details on the elastoplastic and/or perfectly plastic behavior of the backfill soil are
modeling of a fluid and the bounded media can be found in another frequently observed in the soil–structure interaction (SSI), especially
study of the authors (Livaoglu and Dogangun, 2007). for a system subjected to the lateral force or a system excited by

Table 2
The obtained peak responses and their occurrence times for full fluid systems.

Model name F BF25 BF30 BF35 BF40

t (s) Value t (s) Value t (s) Value t (s) Value t (s) Value

uvi (m) – – 16.0  0.4563 16.0  0.3711 16.0  0.2654 16.0  0.2626
uve (m) – – 16.0  0.4267 16.0  0.3347 16.0  0.2277 16.0  0.2269
ur (m) 5.0 0.0028 4.95 0.0078 4.95 0.0079 4.95 0.0079 4.95 0.0079
usl (m) 10.0  1.2573 9.95  1.2613 9.95  1.2613 9.95  1.2613 9.95  1.2613
usr (m) 10.0 1.2591 9.95 1.2631 9.95 1.2631 9.95 1.2630 9.95 1.2630
Sze (MPa) 12.35  1.9754 4.95 5.3383 4.95 5.2959 4.95 5.2890 4.95 5.2890
Szi (MPa) 12.35 1.7916 4.95  5.5741 4.95  5.7034 4.95  5.6167 4.95  5.6992
Sxe (MPa) 12.35  0.0530 11.2  0.1025 4.95 0.0467 4.95 0.0461 4.95 0.0461
Sxi (MPa) 12.35  0.0491 11.2  0.0929 12.3  0.0456 7.45  0.0424 7.45  0.0424

uvi: maximum vertical displacements at top of backfill adjacent point on wall face; uve: maximum vertical displacements at outer point of top of backfill; ur: maximum
horizontal roof displacement of wall; Sze: maximum calculated stress in z direction of exterior face (backfill side) of the wall; Szi: maximum calculated stress in z direction
of interior face (fluid side) of the wall; Sxe: maximum calculated stress in x direction of exterior face (backfill side) of the wall; Sxi: maximum calculated stress in the x
direction of interior face (fluid side) of the wall; usl and usr: maximum sloshing vertical displacements at left- and right-hand sides of the fluid medium.

Table 3
The obtained peak responses and their occurrence times for empty systems.

Model name W B25 B30 B35 B40

t (s) Value t (s) Value t (s) Value t (s) Value t (s) Value

uvi (m) – – 16.0  0.3966 16.0  0.2885 16.0  0.1875 16.0  0.0881
uve (m) – – 16.0  0.3727 16.0  0.2638 16.0  0.1601 16.0  0.0797
ur (m) 5.0 0.0019 5.0 0.0068 5.0 0.0069 5.0 0.0068 5.0 0.0065
Sze (MPa) 5.0 1.1182 5.0 5.7839 5.0 5.2646 5.0 5.7735 5.0 4.9202
Szi (MPa) 5.0  1.4136 5.0  6.0802 5.0  6.1427 5.0  6.0698 5.0  5.8052
Sxe (MPa) 5.0 0.0293 15.65  0.0357 4.9  0.0204 5.0  0.0051 5.0 0.0054
Sxi (MPa) 5.0 0.0138 5.0 0.0781 5.0 0.0786 5.0 0.0772 5.0 0.0734
1166 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173

seismic actions. Lateral responses are generally the most significant Fig. 4 shows that the displacement responses of the backfill
parts of the SSI interaction. In view of all the reasons mentioned gradually increase with time as expected, however, for a considered
herein, the Drucker–Prager material model is used in the modeling of soil type huge residual displacements are calculated, that is, for the
the backfill soil medium in this study. BF25 model, the maximum displacement reaches 0.4563 and
0.4267 m at the adjacent node of the backfill top to the wall and
exterior face of the backfill top, respectively. Although, for the B25
4. Discussion of the results model, with the same backfill characters, in case of nonexistence of
the fluid, the result is estimated at 0.3966 and 0.3727 m, respectively.
Tables 2 and 3 report the maximum vertical displacements, roof The results show yielding in the backfill as the soil material is mostly
displacements, and stress values at both the exterior and interior affected by the decrements in the internal friction angle values. The
faces of the rectangular tank wall and their occurrence times following statements can be given from the results estimated and
obtained from the analyses of empty and full containers, taking into comparisons given. The backfill response is notably amplified in the
consideration the variation of the internal friction angle of the lower values of the internal frictional angles by fluid interactions.
backfill soil. For the tanks that are full of water at the service level, Hence, the effect of the fluid interaction on the stability of the backfill
the maximum sloshing vertical displacements and their occurrence is obvious. The tank wall response is amplified when the fluid exists
time are given for both the left- and right-hand sides. Their within the container. When similar evaluations are performed for
implications are discussed a little later in this text. To design this higher values of the internal frictional angles, the comparison of the
system and to obtain the results, the effect of the fluid and backfill evaluations strictly show the effects of the fluid interaction, that is, for
are included in the models, which in turn result in time consuming the BF40 and B40 models. The responses of vertical displacements for
and computationally expensive dynamic analysis. the BF40 and B40 models are calculated as 0.2626 and 0.0881 m,
respectively. This result shows that the deviation due to the effect of
fluid interaction may reach 198% increment. Therefore, it must be
4.1. Vertical displacements of the backfill stated that during the design process of the backfill stability, fluid
interaction of the system must be taken into account.
To evaluate the calculated vertical displacements of a backfill,
which represents the soil behavior during seismic action, eight
different analyses, by varying of internal friction angles, were 4.2. Lateral and roof displacements of the wall
carried out in two cases. In the first case it was full of fluid, and in
the second case it was empty. The computed peak responses and The maximum computed displacements along the height of
the corresponding times where the maximum vertical displace- the exterior wall of rectangular tanks are shown in Fig. 5. Fig. 5a
ment occurs at the backfill interior (uvi), which means the shows the comparisons with regard to the backfill and fluid effect
displacement of the node at the contact face between the medium on the models. For this purpose, results from the BF40, B40, F, and
and the wall, and the vertical displacements on the exterior face W models are given relative to each other. The most important
node (uve) at top of the backfill soil are given in Fig. 4. As it can be point from the comparison of the illustrations is that backfill
seen, normally all the maximum responses are obtained at the existence or assuming the backfill effect on the design notably
end of the duration of seismic action (t¼16 s). increases the displacement response of the system, that is, while

uvi
uve

0.00 0.00
Vertical displacement of backfill (m)
Vertical displacement of backfill (m)

-0.05 -0.05
-0.10 -0.10
-0.15 -0.15
-0.20 -0.20
B25(uvi) BF25(uvi)
-0.25 B25(uve) -0.25 BF25(uve)
-0.30 B30(uvi) -0.30 BF30(uvi)
B30(uve) BF30(uve)
-0.35 -0.35
B35(uvi) BF35(uvi)
-0.40 B35(uve) -0.40 BF35(uve)
-0.45 B40(uvi) -0.45 BF40(uvi)
B40(uve) BF40(uve)
-0.50 -0.50
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s) Time (s)

Fig. 4. Calculated maximum vertical displacement time histories at the top of the backfill (a) in case of an empty tank and (b) tank full of fluid.
 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173 1167

7 7 7

6 6 6

5 5 5
Height (m)

4 4 4

3 3 3

2 2 2

B25 BF25
1 BF40 1 B30 1 BF30
B40 BF35
F B35
W B40 BF40
0 0 0
0.000 0.002 0.004 0.006 0.008 0.010 0.000 0.002 0.004 0.006 0.008 0.010 0.000 0.002 0.004 0.006 0.008 0.010
Displacement (m) Displacement (m) Displacement (m)

Fig. 5. Heightwise variation of maximum calculated displacements of the exterior wall of the rectangular tank (a) in the case of four different models (b) deviations due to
internal friction angle for the empty tanks (c) deviations due to internal friction angle for the tanks full of fluid.

0.009
t=5s u r =0.0068m (B25) B25
0.008 t=5s u r =0.0069m (B30) B30
Roof displacements (m)

t=5s u r =0.0068m (B35) B35

0.007
B40
0.006 t=5s ur=0.0065m (B40)

0.005
0.004
0.003
0.002
0.001
0.000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)

Fig. 6. Variation in the roof displacements in time for different internal friction angles of backfill in case of the empty container system.

the maximum roof displacement is estimated as 1.9 mm for B25 and B30, respectively. This displacement is decreased to 6.5 mm
model W, where only the wall is modeled in case of nonexistence for model B40. For the empty tank, this decrement may be ignored, so
of both fluid and backfill, for model BF25, where both fluid and the deviation in the model investigated here reaches 5% only. As the
backfill are included, the response is estimated as 7.8 mm. It is top value of the friction angle considered in this investigation is 401,
observed that when the interaction effects of both fluid and which is a very high value for soils used as backfill, in practice, it is
backfill are taken into account, a dramatic increase of almost rarely possible to meet such a soil type. Thus, from the engineering
310% takes place. In case of only the existence of fluid for the point of view, these variations can be accepted with no influence on
F model, this response is calculated as 2.8 mm. As can be seen the design.
from these comparisons, not properly taking into account the Similarly, the time history responses of the system for non-
fluid and/or, especially, the backfill causes one to underestimate existence of fluid (Fig. 6), and for the existence of it (Fig. 7) are
the response and this highly affects the design process. The compared due to the internal friction angle. It seems that almost
reason for this conclusion is that the tanks are displacement- all the responses coincide during excitation. Maximum roof
sensitive-structures, especially during the design processes, as the displacement deviations also show that their maximum values
tank sites are in areas having strong ground motion risks. and occurring times are almost same (Figs. 6 and 7).
However, when the internal friction angle increases from In order to evaluate the fluid interaction effect on the lateral
251 to 401, maximum lateral displacements of the wall are not displacement of the wall system, the lateral displacement’s height-
considerably affected as seen in the given illustrations (Fig. 5b and wise variation can be compared as shown in Fig. 8. The existence of
c). It is worth stating here that in case of a tank full of fluid, the fluid appears to cause an increment in the lateral displacement of the
backfill effect variations on the wall due to the internal friction angle system with resulting higher displacements especially for the lower
are mostly absorbed by the fluid. However, some deviations for the values of the internal frictional angle (Fig. 8b). While the maximum
other case occur in the displacements of the wall. For instance, the roof displacement is estimated as 6.8 mm for the B25 model, for the
maximum roof displacement for the wall modeled in case of BF25 model, considering the fluid interaction, this value reaches
nonexistence of fluid is estimated as 6.8 and 6.9 mm for models approximately 7.8 mm. The fluid interaction causes up to 15%
1168 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173

0.009
t=4.95s ur=0.0079m (BF35) t=4.95s ur=0.0078m (BF25) BF25
0.008

Roof displacements (m)

t=4.95s ur=0.0079m (BF30) BF30
BF35
0.007 BF40
t=4.95s ur=0.0079m (BF40)
0.006
0.005
0.004
0.003
0.002
0.001
0.000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00
Time (s)
Fig. 7. Variation in the roof displacements in time for different internal friction angles of backfill in case of the container system full of fluid.

7 7

6 6

5 5

4 4
Height (m)

Height (m)

3 3

2 2

1 BF40 1 BF25

B40 B25

0 0
0.000 0.002 0.004 0.006 0.008 0.010 0.000 0.002 0.004 0.006 0.008 0.010
Displacement (m) Displacement (m)

Fig. 8. Height wise variation of maximum calculated displacements of the exterior wall (a) between BF40 and B40 and (b) between BF25 and B25.

0.009
t=4.95s u r =0.0079m (BF40) BF40
0.008
Roof displacements (m)

B40
0.007
0.006
t=5s ur =0.0065m (B40)
0.005
0.004
0.003
0.002
0.001
0.000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)

Fig. 9. Variation of the calculated roof displacement time histories between B40 and BF40.

increment of displacement. On the other hand, as a result of A comparative variation of roof displacement as a function of time
comparison between the B40 and BF40, it shows that this deviation for BF40 and B40 models is shown in Fig. 9. While the above-
can reach an increment of almost 22%. mentioned ones are merely realistic examples, not covering all the
 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173 1169

possible cases, their trend is clear: fluid action on the wall tends to and their effect amplifies the response of the system. In some cases
increase the wall displacement considerably. There are different this may not be an increase in the amplification. However, for the
parameters of the system that explain amplification on displacement. system having relatively short fundamental periods such as the walls
For example, when the fluid within the container is included, with the considered here, dynamic reactions are mostly amplified during the
impulsive mass of fluid simultaneously acting on the wall and system, seismic action due to fluid interaction. However, it is not always
the fundamental period is increased. Furthermore, fluid sloshing possible to say so. Increments in the fundamental periods of the
causes extra hydrodynamic pressure adding to the impulsive ones, system may cause decrements in the system displacement response,

2.0
BF25
Sloshing displacements (m)

1.5 BF30
BF35
1.0 BF40
0.5
0.0
-0.5
-1.0
-1.5
t=9.95s uS =1.26m (BF25,BF30,BF35,BF40)
-2.0
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)

Fig. 10. Calculated vertical sloshing displacement time histories of the fluid within the container.

1000000
B25
0 B30
B35
-1000000 B40

-2000000
Szi (N/m2)

-3000000

-4000000

-5000000 t=5s Szi=6.08MPa (B25)

-6000000 t=5s Szi=6.07MPa (B35)

t=5s Szi=6.14MPa (B30)
t=5s Szi=5.81MPa (B40)
-7000000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)
1000000

-1000000

-2000000
Szi (N/m2)

-3000000

-4000000
t=4.95s Szi=5.57MPa (BF25)
-5000000 BF25
t=4.95s Szi=5.62MPa (BF35) BF30
-6000000 t=4.95s Szi=5.70MPa (BF30) BF35
t=4.95s Szi=5.70MPa (BF40) BF40
-7000000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)

Fig. 11. Comparative variations of the calculated stress histories at interior nodes of the wall in case of (a) empty tanks and (b) full of fluid.
1170 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173

especially for long-period systems, due to the frequency contents of more clearly. Therefore, these results, to some extent, show that
the seismic actions and flexibility of the foundation soil. properly considering the backfill and fluid interaction effect on
Many codes estimate the dynamic earth pressures according to the design of such a structure is crucial.
the potential of the wall to deform. For example, for the seismic
analysis of bridge abutments, the Regulatory Guide E39/93 4.3. Sloshing displacement of fluid
proposes three different cases for the calculation of the dynamic
earth pressures, depending on the ratio between the expected As a result of the analyses of five different models, almost the
(or allowable) displacement at the top of the wall U to its height H same results are obtained, where deviations in the internal
(Gazetas et al., 2004). These allowable displacement ratios are friction angle of the backfill system do not change the maximum
generally between 0.05% and 0.10%. In the present study, the sloshing response of the fluid inside the container. Sloshing
calculated maximum response for the BF40 model is 7.9 mm and response time histories are shown in Fig. 10. As can be seen from
the ratio is estimated as 0.13%. However, for the B40 model the it, the time histories coincide. The predicted maximum sloshing
displacement and its ratio are 6.5 mm and 0.108%, respectively. displacements are 1.26 m (at 9.95 s) for both the left- and right-
This result shows that when the fluid interaction is considered, hand sides for the fluid, in all the systems analyzed in these
the estimated maximum value for the considered excitation investigations. As the designer tends to leave only a gap of 0.5 m
exceeds allowable displacements. Indeed, the displacements are from the maximum liquid level in the tank investigated here, the
almost in the boundary of the maximum displacement level for sloshing of fluid may cause the sloshing pressure on the roof. This
the system in an empty condition. Similarly if one investigates the sloshing pressure, however, can be evaluated as ignorable for
result from the nonexistence of the backfill for tanks full of fluid, such a roof system which have a stiff and heavy reinforced plate
these ratios maximally reach the 0.04% that are left; lower than supporting gravity effects of very thick roof coverings.
the minimal ratio limit for ground motion is taken into account in
this study. Naturally it is possible to say that for the different 4.4. Stresses in the wall
ground motion having more peak response and high frequency
content, these displacements can potentially increase and then The computed stress responses and their variations in time at
the limiting values of the displacements again can be exceeded both the exterior and the interior face of the rectangular tank wall

7000000
t=5sSze=5.78MPa (B25) B25
6000000 t=5sSze=4.92MPa (B40) B30
t=5sSze=5.77MPa (B35)
B35
5000000 B40

4000000 t=5sSze=5.26MPa (B30)

Sze (N/m2)

3000000

2000000

1000000

-1000000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)
7000000
t=4.95s Sze=5.34MPa (BF25) BF25
6000000 BF30
t=4.95s Sze=5.29MPa (BF35)
t=4.95s Sze=5.30MPa (BF30) BF35
5000000 t=4.95s Sze=5.29MPa (BF40) BF40

4000000
Sze (N/m2)

3000000

2000000

1000000

-1000000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)

Fig. 12. Comparative variations of the calculated stress histories at the exterior node of the wall in case of (a) empty tanks and (b) full of fluid.
 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173 1171

are given in the subtitle. The maximum stress values obtained at Maximum response value of the stresses, similar to the interior ones,
the exterior face of the tank wall in the z direction generally tend generally tends to decrease when the internal friction angle increases.
to decrease with increasing of the internal friction angle from 251 This decrease between the B25 to B40 is calculated as 15% approxi-
to 401. Fig. 11 shows that the obtained maximal stresses of the mately. In theory, the concrete that is accepted has no tension
interior node of the wall in the z direction, due to internal friction strength. However tension strength is approximately 10% of the
angle, are almost the same in the time of occurrence and compressive strength. Hence, for concrete, such a tension amplitude
amplitude of the stresses. Estimated stresses in the z direction may cause it to crack on the tension side of the section, but reinforced
(Szi) take place at the interior nodes of the wall, at the level of concrete bearing capacity is fully sufficient for almost all material
0.9 m from the top of the foundation, as tension. Maximum classes used in practice in the other side, because the C20 concrete
stresses obtained for the B25 and BF30 models, are with 6.08 class, which has 20 MPa compressive strength (fc) and S420 steel,
and  5.70 MPa pressures, respectively. Depending on the inter- which has 420 MPa yield strength (fy), are generally used in practice
nal frictional angle, maximal deviation between B25 and B40 is for such a type of structure. Furthermore, it means that the tensile
calculated as 4.6%. From the engineering point of view, this crack would be growing in the side of the backfill due to the effect of
maximal decrease can be practically ignored. Furthermore, the earth pressure, as the maximum tensile strength for this concrete
amplitude of the stresses calculated in the pressure shows that class can reach 2 MPa (fct). On the other hand in the design processes
this does not have an important value, so it does not exceed the this value is assumed to be 1.33 MPa, which is the tensile design
compressive design strength of the concrete used in practice. strength of concrete (fctd). Although this situation does not threaten
Fig. 12 shows comparative variations of the stress histories at the the stability of the wall, some measures must be taken to prevent the
exterior nodes of the wall for tanks that are empty and full of fluid. leakage of fluid. Still the wall section bearing capacity is without

a
3000000
t=12.35s Szi=1.79MPa (F)
t=5s Szi=1.41MPa (W)
2000000

1000000

-1000000
Szi (N/m2)

-2000000

-3000000

-4000000

-5000000 W
-6000000 F
t=5s Szi=6.08MPa (B25)
B25
-7000000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)

b 1000000

-1000000

-2000000
Szi (N/m2)

-3000000

-4000000

-5000000
B25
t=4.95s Szi=5.57MPa (BF25)
-6000000
t=5s Szi=6.08MPa (B25) BF25
-7000000
0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00
Time (s)

Fig. 13. Comparative variations of the calculated stress histories at the interior node of the wall for the models (a) between W, F, and B25 and (b) between B25 and BF25.
1172 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173

doubt sufficient to carry the earthquake-induced force considered in interaction on the sloshing response can be ignored in the
this investigation. evaluation of the seismic behavior of the system.
Both Figs. 11 and 12 show that when the fluid within the 2. The soil interaction effect is clearly more visible in the case of
container is taken into account, the amplitude deviations of the nonexistence of the fluid. According to the results of the
stress histories differ from the ones obtained from empty tanks. analyses performed, the internal friction angle affects the
While the maximal values of the stress response are close to each system (the backfill–wall–fluid model), as the displacement
other for the exterior and interior nodes of B systems their maximal response generally decreases to some extent, while the inter-
values are obtained on the tensile and pressure sides, respectively. nal friction angle has higher values. This also shows that by
On the other hand in BF systems, it can be clearly seen that stresses taking into account fluid interaction, the response of the whole
calculated for the exterior node pass through the pressure side and system is absorbed.
for the interior nodes pass through the tensile side in the same time 3. To consider the tank is empty, full of fluid or to evaluate the
interval (i.e., between 7 and 8 s). These reversible stresses and situation of nonexistence of backfill causes a completely
increases on its amplitude prove that fluid interaction amplifies the different seismic behavior of the wall. While both exist, their
wall vibration during seismic action. inertial effect amplifies the system response considerably.
The maximum stress values obtained at both the exterior When one only considers both the ‘‘full of fluid’’ and ‘‘empty
corner face and exterior mid face of the tank wall, in the case’’ (with no backfill) comparatively, the response of the
x direction, decrease with an increase in the internal friction system is reversible and its response values are lower than in
angle, as an absolute value. Furthermore, their occurrence times other circumstances; however, the system displacement
are almost the same for the empty (B) and full tank (BF) systems. response and stress deviations are one side of the response,
It is worth stating that the calculated stresses are very small when that is, the pressure or tension side for stresses, when the
compared with the bearing capacity of the tank wall material in backfill is considered, as backfill prevents the reversible
tension. When the pressure strength is not exceeded due to the response of the system and pushes the wall toward the fluid
excitations considered here, the shear capacity of the system is or empty medium.
sufficient, as the stresses estimated are almost 10% or more, that 4. There are similar backfills having likely material and geome-
is, less than their pressure ones. Moreover, when it is taken into trical properties for almost all tanks in practice; the allowable
account that the shear strength of the concrete system has displacements are exceeded for the systems imposed with
generally 35–80% capacity of the compressive strength, the seismic motion considered in this study. This shows that not
above-mentioned idea would be more verifiable. taking into account the backfill effects causes the system
As expressed earlier, well-known codes suggest that dynamic response to be underestimated, and misleads the results in
earth pressures will be taken into account when computing the base the design process of the tank wall.
shear of a partially or fully buried liquid-containing structure and 5. When static or pseudo-static methods are used, misleading
when designing the walls, the dynamic backfill forces will not be results may be obtained, as these approximations amplify the
relied upon to reduce the dynamic effects of the stored liquid or vice response of the empty reference system, with no interaction
versa. If designers model the system in a static condition or use the effect. However, based upon these approximations, the system
pseudo-static model for the dynamic effects, as proposed by most of behavior with interactions is completely different from the
the literature about the earth retaining structure, only then would reference system.
they add the earth pressure to the wall at a different level. Indeed
Fig. 13 proves that these approaches, considering the backfill effect,
are not adequate for the walls subjected to both backfill and fluid Acknowledgment
interactions. Inertial effects of the backfill are most pronounced on
comparative variations of the model given in Fig. 13. For example if The present work is supported by Grant-in-Aid for Scientific
the designer ignores the backfill inertial effects and their dynamic Research (Project no. 105M252) from the Scientific and Techno-
characteristics and assumes the wall subjected to a pseudo-static logical Research Council of Turkey (TÜB_ITAK).
force, then the response of the W model or F model with the fluid
interaction effect, is more amplified. However, the response of the
system stresses may take place more differently, as shown in Fig. 13a,
References
and the stresses reverse, that is, while the maximum response of Sze is
1.12 MPa for model W, the maximum response for the B25 and BF25
Ahmad, S.M., Choudhury, D., 2008. Pseudo-dynamic approach of seismic design for
models are calculated as 5.78 and 5.34 MPa, respectively. waterfront reinforced soil–wall. Geotextiles and Geomembranes 26, 291–301.
Akyildiz, H., Unal, E., 2005. Experimental investigation of pressure distribution on
a rectangular tank due to the liquid sloshing. Ocean Engineering 32,
1503–1516.
5. Conclusions Akyildiz, H., Unal, E., 2006. Sloshing in a three-dimensional rectangular tank:
numerical simulation and experimental validation. Ocean Engineering 33,
To evaluate the seismic behavior of the backfill–rectangular 2135–2149.
American Concrete Institute (ACI 350.3-01/350.3R-01), 2001. Seismic Design of
tank–fluid system, a finite element model, including both the Liquid-Containing Concrete Structures (ACI 350.3-01) and Commentary
fluid–structure and soil–structure interaction effects and non- (350.3R-01). ACI Committee 350, Environmental Engineering Concrete Struc-
linear interaction effect of the soil and contact nodes between the tures, December.
ANSYS, 1994. In: Peter, K. (Ed.), Theory Manual 12th ed. SAS IP, Inc. 1266.
soil and the wall, is proposed. The proposed model can be used for Chen, J.Z., Kianoush, M.R., 2005. Seismic response of concrete rectangular tanks for
determining the internal forces, sloshing effect, and inertial liquid containing structures. Canadian Journal of Civil Engineering 32,
effects of the backfill behind the wall in the design process. 739–752.
Choudhury, D., Sitharam, T.G., Subba Rao, K.S., 2004. Seismic design of earth-
However, FEM analysis with this procedure needs relatively more
retaining structures and foundations. Special section: geotechnics and earth-
computer memory capacity and CPU time. As a consequence of quake hazards. Current Science 87, 1417–1425.
this study, the following results must be given: Choudhury, D., Ahmad, S.M., 2007. Stability of waterfront retaining wall subjected
to pseudo-static earthquake forces. Ocean Engineering 34, 1947–1954.
Dogangun, A., Durmus, A., Ayvaz, Y., 1996. Static and dynamic analysis of
1. The sloshing response is not affected by the backfill interaction rectangular tanks by using the Lagrangian fluid finite element. Computers
considered in this study. Therefore, the effect of the backfill and Structures 59, 547–552.
 R. Livaoglu et al. / Ocean Engineering 38 (2011) 1161–1173 1173

Dogangun, A., Durmus, A., Ayvaz, Y., 1997. Earthquake analysis of flexible Kramer, S.L., 1996. Geotechnical Earthquake Engineering. Prentice-Hall, Engle-
rectangular tanks by using the Lagrangian fluid finite element. European wood Cliffs, NJ.
Journal of Mechanics A—Solids 16, 165–182. Livaoglu, R., Dogangun, A., 2007. Effect of foundation embedment on seismic
Dogangun, A., Livaoglu, R., 2004. Hydrodynamic pressures acting on the walls of behavior of elevated tanks considering fluid–structure–soil interaction. Soil
rectangular fluid containers. Structural Engineering and Mechanics—An Inter- Dynamics and Earthquake Engineering 27, 855–863.
national Journal 17 (2), 203–214. Livaoglu, R., Cakir, T., Dogangun, A., 2007. Investigation of backfill effects on
Durukal, E., 2002. Critical evaluation of strong motion in Kocaeli and Düzce seismic behavior of rectangular tanks. In: International Symposium on
(Turkey) earthquakes. Soil Dynamics and Earthquake Engineering 22, Advances in Earthquake and Structural Engineering, Isparta, Turkey, October
589–609. 24–26, pp. 478–490 (in Turkish).
Epstein, H.I., 1976. Seismic design of liquid storage tanks. Journal of the Structural Livaoglu, R., 2008. Investigation of seismic behavior of fluid–rectangular tank–soil/
Division—ASCE 102, 1659–1673. foundation systems in frequency domain. Soil Dynamics and Earthquake
Eurocode 8, 2003. Design of structures for earthquake resistance—Part 5: Engineering 28, 132–146.
foundations, retaining structures and geotechnical aspects. Final Draft, Eur- Lysmer, J., Kuhlemeyer, R.L., 1969. Finite dynamic model for infinite media.
opean Committee for Standardization, Brussels, Belgium, December. Engineering Mechanics Division Journal—ASCE 95, 859–877.
Eurocode 8, 2006. Design of structures for earthquake resistance—Part 4: silos, Minowa, C., 1984. Experimental studies of seismic properties of various type water
tanks and pipelines. Final Draft, European Committee for Standardization, tanks. In: Proceedings of the Eighth WCEE, San Francisco, pp. 945–952.
Brussels, Belgium, January. Mononobe, N., Matsuo, H., 1929. On the determination of earth pressures during
Faltinsen, O.M., Rognebakke, O.F., Timokha, A.N., 2003. Resonant three-dimen- earthquakes. In: Proceedings of the World Engineering Congress, vol. 9, Tokyo,
sional nonlinear sloshing in a square-base basin. Journal of Fluid Mechanics
Japan, pp. 179–187.
487, 1–42.
Mylonakis, G., Kloukinas, P., Papantonopoulos, C., 2007. An alternative to the
Gazetas, G., Psarropoulos, P.N., Anastasopoulos, I., Gerolymos, N., 2004. Seismic
Mononobe–Okabe equations for seismic earth pressures. Soil Dynamics and
behaviour of flexible retaining systems subjected to short-duration moder-
Earthquake Engineering 27, 957–969.
ately strong excitation. Soil Dynamics and Earthquake Engineering 24,
Nazarian, H.N., Hadjian, A.H., 1979. Earthquake-induced lateral soil pressures on
537–550.
structures. Journal of the Geotechnical Engineering Division—ASCE 105,
Graham, E.W., Rodriguez, A.M., 1952. Characteristics of fuel motion which affect
1049–1066.
airplane dynamics. Journal of Applied Mechanics 19, 381–388.
Okabe, S., 1924. General theory of earth pressure and seismic stability of retaining
Hashemi, H.S.H., Karimi, M., Taher, H.R.D., 2010. Vibration analysis of rectangular
wall and dam. Journal of the Japan Society of Civil Engineers 10 (6),
Mindlin plates on elastic foundation sand vertically in contact with stationary
fluid by the Ritz method. Ocean Engineering 37, 174–185. 1277–1323.
Hoskins, L.M., Jacobsen, L.S., 1934. Water pressure in a tank caused by simulated Park, J.H., Koh, H.M., Kim, J.K., 2000. Seismic isolation of pool-type tanks for the
earthquake. Bulletin of the Seismological Society of America 24, 1–32. storage of nuclear spent fuel assemblies. Nuclear Engineering and Design 199,
Housner, G.W., 1963. Dynamic behavior of water tanks. Bulletin of the Seismolo- 143–154.
gical Society of America 53, 381–387. Priestley, M.J.N., Davidson, B.J., Honey, G.D., Hopkins, D.C., Martin, R.J., Ramsay, G.,
Isaacson, M., Ryu, C.S., 1998. Directional effects of earthquake-induced sloshing in 1986. Seismic Design of Storage Tanks: Recommendation of a Study Group.
rectangular tanks. Canadian Journal of Civil Engineering 25, 376–382. The New Zealand Society for Earthquake Engineering, New Zealand.
Kianoush, M.R., Chen, J.Z., 2006. Effect of vertical acceleration on response of Richards, R., Elms, D.G., 1979. Seismic behavior of gravity retaining walls. Journal
concrete rectangular liquid storage tanks. Engineering Structures 28, 704–715. of the Geotechnical Engineering Division—ASCE 105 (GT4), 449–464.
Kim, J.K., Koh, H.M., Kwahk, I.J., 1996. Dynamic response of rectangular flexible Seed, H.B., Whitman, R.V., 1970. Design of earth retaining structures for dynamic
fluid containers. Journal of Engineering Mechanics—ASCE 122 (9), 807–817. loads, Proceedings of the Speciality Conference on Lateral Stresses in the
Kim, J.K., Park, J.Y., Jin, B.M., 1998. The effects of soil structure interaction on the Ground and Design of Earth Retaining Structures. ASCE, Ithaca, New York, pp.
dynamics of 3-D flexible rectangular tanks. In: Proceedings of the Sixth East 103–147.
Asia-Pacific Conference on Structural Engineering and Construction, Taipei, Turkish Earthquake Code (TEC), 2007. Specification for Structures to be built in
Taiwan, January 14–16. Disaster Areas. Ministry of Public Works and Settlement Government of
Koh, H.M., Kim, J.K., Park, J.H., 1998. Fluid–structure interaction analysis of 3-D Republic of Turkey.
rectangular tanks by a variationally coupled BEM–FEM and comparison with Wolf, J.P., Song, C.H., 1996. Finite-element Modelling of Unbounded Media. John
test results. Earthquake Engineering and Structural Dynamics 27, 109–124. Wiley & Sons, Chichester.