Research Article

Advances in Mechanical Engineering

2015, Vol. 7(5) 1–16
Ó The Author(s) 2015
A simplified mechanical model with DOI: 10.1177/1687814015582085
aime.sagepub.com
fluid–structure interaction for
rectangular tank sloshing under
horizontal excitation

Changfang Zou1,2 and Deyu Wang1,2

Abstract
Based on the spring–mass model, a novel mechanical model of sloshing with fluid–structure interaction under a horizon-
tal excitation is proposed, and the coupled dynamic equation of sloshing system is established. Considering the flexibility
of bulkhead, the effects of certain factors, such as bulkhead bending stiffness and filling ratios, on the mode of coupled
sloshing system are investigated. It is found that these factors have significant influence on the mode. By comparing the
present results with the results of ADINA based on the linear potential flow theory and published literatures, the pro-
posed coupled sloshing model is verified. The results show that the simplified rigid mass, m0 , dominates the contribu-
tions to bending moment near the bottom of a bulkhead and the spring–mass, m1 , k1 , to bending moment near the
liquid-free surface of a bulkhead. Furthermore, the computational cost is greatly reduced by using the proposed mechan-
ical model with fluid–structure interaction for a rectangular tank sloshing.

Keywords
Sloshing, fluid–structure interaction, spring–mass model, water wave mode, coupling effect

Date received: 31 October 2014; accepted: 10 March 2015

Academic Editor: Filippo Berto

Introduction
Sloshing is the phenomenon of two or more kinds of
fluid restrictedly moving in limited space. The promi-
nent feature of sloshing is the existence of free surface
and highly nonlinear characteristics. The sloshing phe-
nomenon is usually encountered in many engineering
fields, such as very large crude carrier (VLCC), lique-
fied petroleum gas (LPG), liquefied natural gas (LNG)
liquid cargo ships in non-fully loaded navigation, liquid
storage tanks in the earthquake, moving oil tank cars,
and rocket bunker tanks. When sloshing occurs, the
tank wall is subjected to long-term sloshing pressure,
which may cause structural fatigue or even ultimate fail-
ure. Sloshing can also affect motion stability and cause
disastrous accidents for liquid cargo ships, rocket, and
oil tankers. These accidents will not only cause eco-
nomic loss but also result in the environmental pollu-
tion because of oil spill. So, it is of great importance to
investigate the sloshing characteristics in engineering
problems.
1
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong
University, Shanghai, China
2
The Collaborative Innovation Center for Advanced Ship and Deep-Sea
Exploration, Shanghai, China
Corresponding author:
Deyu Wang, State Key Laboratory of Ocean Engineering, Shanghai Jiao
Tong University, Shanghai 200240, China.
Email: dywang@sjtu.edu.cn

Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License
(http://www.creativecommons.org/licenses/by/3.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (http://www.uk.sagepub.com/aboutus/
openaccess.htm).
2 Advances in Mechanical Engineering
Many researchers have studied the fluid–structure
interaction (FSI) in sloshing. The main research meth-
ods are as follows.
In the aspect of analytic solutions, the study of non-
linear analytic solution of sloshing and analytic solution
of coupled sloshing system was limited, so researchers
developed equivalent mechanical models and linear
analytic solutions based on potential flow theory for
rigid tanks of different shapes. Kim1 studied eigenvalue
of sloshing and adopted reduction integral scheme of
rotational penalty method to exclude pseudo mode.
Ibrahim2 obtained the analytic solution of sloshing for
rigid and flexible tank based on linear potential flow
theory. He also used the pendulum model and spring–
mass model to establish the simplified sloshing model
for a rigid tank and the equivalent force and moment
acting on the bulkhead were given. By means of the
spring–mass mechanical model based on linear poten-
tial theory, Livaoglu3 carried out the coupled dynamic
response calculation for partially filled tank fluid–struc-
ture–soil/foundation and analyzed the effect of founda-
tion stiffness and wall-foundation connection forms on
sloshing behavior. On the basis of linear potential the-
ory, Tabri et al.4 adopted a simplified spring–mass
model to study the coupling effect during liquid carrier
collision, and the results showed that simplified model
is reasonable by comparing with experimental data.
Tsukamoto et al.5 studied the effect of flexible connec-
tion on sloshing between moving link and wall in par-
tially filled tank based on linear potential flow theory,
and the calculation results showed that the analytic
solution is reasonable through comparison with mesh-
less moving-particle semi-implicit (MPS) method. They
also showed that the moving link can weaken sloshing
in inner fluid. Based on linear potential flow theory, Li
and Wang6 deduced the analytic solution of sloshing
pressure and velocity in a two-dimensional rectangular
tank, and they provided an accurate simplified mechan-
ical model for sloshing fluid in a rigid rectangular tank.
In the aspect of numerical methods, researchers
mainly used finite element method (FEM), meshless
method and arbitrary Lagrangian–Eulerian (ALE)
method to handle the sloshing problem with FSI.
Bucchignani et al.7 studied the motion of an incompres-
sible inviscid flow in a deformable tank. Idelsohn et al.8
used particle FEM to study a tanker sinking. They also
found that the method can provide a very advanta-
geous and efficient way for solving contact and free sur-
face problems. By means of ALE method, Zhang and
Suzuki9 established the FE model of sloshing with
fluid–structure coupling effect under ship collision
situations. By comparing Zhang’s model with linear
sloshing model and Lagrangian FE model from the
view of energy and computational costs, Zhang et al.
drew a conclusion that coupling effect has a significant
influence on dynamic response of the system and ALE
method has certain advantages for sloshing problems.
Lee et al.10 used MPS approach and FEM to simulate
sloshing in the two-dimensional rectangular tank and
dam break successfully. Kim et al.11 adopted impulse
response function approach to solve linear motion of
liquid ship while finite differential method (FDM) is
employed to discrete flow field, and they also studied
coupling effects between liquid sloshing and ship
motion. Mitra and Sinhamahapatra12 established the
coupled dynamic FE equation of sloshing for a two-
dimensional elastic tank by using Galerkin method,
and the conclusion was that the sloshing pressure in
flexible bulkhead condition was larger than that in rigid
bulkhead condition. Taking flexible tanks into consid-
eration, Degroote et al.13 numerically calculated tank
sloshing related to FSI and found that the impact pres-
sure at the bottom of the rigid cylindrical tank was
twice as much as the flexible tank. By means of ALE
approach, Kassiotis et al.14 calculated impact pressure
that acted on a nonlinear structure. Cao et al.15 numeri-
cally calculated the sloshing response with coupling
effect between fluid inside the tank and flexible wall
under the wind loads by using ALE method. The
results showed that reinforcement hoop can reduce
tank stress as well as improve the tank stability in wind
loads. In addition, Pan et al.16 employed MPS method
to simulate large deforming free surface of sloshing.
In the aspect of experimental researches, large-scale
model tests and sloshing experiments of a flexible tank
hardly have been done although lots of sloshing experi-
ments have been carried out for rigid tanks currently.
For a rigid tank, the effects of excitation parameters,
such as frequency and amplitude, on sloshing loads was
studied by Akyildiz and Unal17 and they also found
that baffles can significantly reduce fluid motion.
Nevertheless, the effect of fluid viscosity on impact
pressures was not investigated in sloshing tests. Specific
drop tests of a tank were carried out by Anghileri
et al.,18 and the effect of sloshing on tank structural
damages was studied. Compared with various numeri-
cal results obtained from FE, Eulerian, ALE, and
smoothed particles hydrodynamics (SPH) model, they
also concluded that the SPH method was the most fea-
sible for the analysis of the sloshing. Tabri et al.4 experi-
mentally investigated the effect of FSI on sloshing load
and structural deformation energy during ship colli-
sions, and they found that the deformation energy in
the wet test was only about 80% of that in the dry tests.
For a standing beam and hanging beam arranged in a
rolling tank, a coupled experiment between an elastic
beam and a sloshing liquid was carried out by Degroote
et al.,13 and the numerical method was validated by the
experimental data and the effect of grid scale on calcu-
lation results was also discussed. Carra et al.19 experi-
mentally studied the linear and geometrically nonlinear
dynamical response of a thin plate in contact with water
Zou and Wang 3

on one or both sides. They found that the plate deflec-

tion due to hydrostatic pressure played an important
role in changing the plate nonlinearity, but tests with
liquid on both sides can eliminate this effect.
By using numerical methods to study sloshing
coupled with structure, the computational scale is very
large, especially in dealing with some complex engineer-
ing problems. On the basis of above, we propose an
alternative mechanical model to investigate liquid slosh-
ing inside a flexible rectangular tank coupled to struc-
ture. For a rectangular tank subjected to a horizontal
excitation, a novel mechanical model of sloshing
involved in coupling effect is proposed. Although the
proposed model cannot focus on the details of the flow
field distribution, it can be easily used to handle the
coupling effect between equivalent sloshing liquid and
bulkhead. In this article, first, a simplified fluid–
structure coupling model for sloshing fluid in a flexible
rectangular tank is proposed, and the dynamic equation
Figure 1. Coupled sloshing system model (L = 1 m, B = 1 m,
of sloshing system is established. Second, the effects of rtank = 1200 kg=m3 , m= 0.363, E = 3 3 109 Pa).
bulkhead bending stiffness and filling ratio on coupled
modal frequencies are analyzed for this coupled literatures of Ibrahim2 and Livaoglu.3 It can be
mechanical model. Then, dynamic response analysis of expressed by the following formulae (1)–(5)
the model is carried out, and the effects of rigid mass
and convection mass on bending moment of bulkhead 8mF tanhð(2n  1)ph=lÞ
are investigated. The results are compared with the mn = ð1Þ
p3 (2n  1)3 h=l
numerical results of coupled FE model of sloshing
based on classical linear potential flow theory and other hn 1 tanhð(2n  1)ph=(2l)Þ
=  ð2Þ
results from the literatures of Bucchignani et al.,7 Lee h 2 (2n  1)ph=(2l)
et al.,10 and Löhner et al.20 The results demonstrate the
m0 X‘
mn
proposed model is reasonable. Furthermore, the pro- =1  ð3Þ
posed model has the following advantages: mF m
n=1 F

It does not need to resolve large-scale data transfer h0 mF X ‘

between the flow field and structural field.
The number of degrees of freedom in nonlinear
equation is small for the proposed mechanical slosh-
ing system during numerical calculation of fluid–
structure coupling, and the solving efficiency can be
increased greatly.
A simplified mechanical model of coupled
sloshing system
In this section, based on spring–mass mechanical model
of sloshing in a rigid tank, a simplified mechanical
model considering the coupling effect between liquid
and flexible bulkhead is proposed, and the coupled
dynamic equation is established.
A two-dimensional rectangular tank is given as
shown in Figure 1; it is of width B, height L, and liquid
filling level h.
Assuming the tank wall is rigid, sloshing liquid inside
the tank can be simplified by a series of spring–mass, as
shown in Figure 1. The model is also shown in the
mn hn
= ð4Þ
h m0 n = 1 mF h
tanh2 ð(2n  1)ph=lÞ
kn = 8mF g ð5Þ
p2 (2n  1)2 h
where m0 denotes the rigid mass moving with the tank;
mn (n = 1, 2, . . . ) and kn (n = 1, 2, . . . ) denote the equiv-
alent mass and the equivalent spring stiffness constant
of each sloshing mode, respectively; hn (n = 0, 1, . . . )
denotes the distance from the equivalent mass points to
fluid center of gravity, respectively; mF and h denote
the total mass of fluid in the tank and the fluid filling
height, respectively.
Based on the simplified spring–mass model for a
rigid tank sloshing, a novel mechanical model involving
in coupling to flexible bulkhead is proposed for a rec-
tangular tank sloshing. The bulkhead is simplified as a
plane strain beam, and its density, modulus of elasticity,
section moment of inertia, external load, and beam
bending displacement are denoted by r, E, I, p(y, t), and
ub (y, t), respectively. The simplified fluid point mass,
corresponding displacements, and spring stiffness are
4 Advances in Mechanical Engineering

denoted by mwi (i = 0, 1, . . . ), uwi (i = 0, 1, . . . ), and ∂T

= mwi u_ wi ð13Þ
kwi (i = 1, 2, . . . ), respectively. ∂u_ wi
Tank sloshing with fluid–structure coupling effect is Xn Xn Xn
∂V
investigated, as shown in Figure 1. Considering the = k1ij qjl (t) + k2ij qjr (t) + k3ij uwj ð14Þ
∂qil
flexible bulkhead, dynamic equation of fluid–structure j=1 j=1 j=1
coupling system is established by using Lagrange equa- Xn Xn Xn
∂V
tion derived from Hamilton principle and variational = k1ij qjr (t) + k2ij qjl (t) + k3ij uwj ð15Þ
∂qir
method. j=1 j=1 j=1
The total kinetic energy of coupled system is Xn Xn Xn
∂V
= k4ij uwj + k5ij qjl + k5ij qjr ð16Þ
ðL ðL ∂uwi j=1 j=1 j=1
1 1 1 X
n
T= rA(u_ bl )2 dy + rA(u_ br )2 dy + mwi (u_ wi )2
2 2 2 i=0 where
0 0
ð6Þ ðL ðL
mblij = rAci (y)cj (y)dy, k1ij = € j (y)c
EI c € i (y)dy
The total potential energy of coupled system is
0 0

ðL ðL X
n
1 00 2 1 + kt cj (yt )ci (yi )
V= EI(ubl ) dy + EI(u00br )2 dy t=1
2 2
0 0 X
n

X
n X
n k2ij = kt cj (yt )ci (yi ), k3ij =  2kj ci (yi ),
1 2
+ ki ð2uwi  ubl (yi , t)  ubr (yi , t)Þ + mwi ghi t=1
i=1
2 i=0 X
n
k4ij = 4kj ki , k5ij =  2km ki cj (ym )
ð7Þ m=1

The non-conservative force virtual work of system is ðL ð17Þ

dwnc = p(y, t)d½ubl (u, t) + ubr (u, t)dy
ðL 0
dwnc = p(y, t)dub (u, t)dy ð8Þ ðL " #
X
n
0 = p(y, t) ci (y)d½qil (t) + qir (t) dy
i=1
0
According to the superposition solution of vibration 0L 1
X
n ð
modes, the beam bending displacement can be @ p(y, t)ci (y)dyAd½qir (t) + qil (t)
expressed by =
i=1
0
X
n X
n
ubl (y, t) = ci (y)qil (t) ð9Þ = Qi d½qir (t) + qil (t)
i=1 i=1
ÐL
X
n where Qi = 0 p(y, t)ci (y)dt.
ubr (y, t) = ci (y)qir (t) ð10Þ According to Lagrange equation, we have
i=1
 
d ∂T ∂T ∂V
where A, p(y, t), ubl (y, t), ubr (y, t), qil (t), qir (t), and c(y)  + = Qi ð18Þ
denote the sectional area of the beam, the external load, dt ∂q_ il ∂qil ∂qil
 
the bending displacement of the left and right bulkhead, d ∂T ∂T ∂V
the generalized coordinates of the left and right bulk-  + = Qi ð19Þ
dt ∂q_ ir ∂qir ∂qir
head, and mode of vibration of the beam, respectively,  
as shown in Figure 1. d ∂T ∂T ∂V
 + =0 ð20Þ
Then dt ∂u_ wj ∂uwj ∂uwj

Xn Then, the following equations can be yielded by

∂T
= mblij q_ jl (t) ð11Þ equations (18)–(20)
∂q_ il j=1
X
n X
n X
n X
n

∂T X
n qjl +
mblij € k1ij qjl (t) + k2ij qjr (t) + k3ij uwj = Qi
= mbrij q_ jr (t) ð12Þ j=1 j=1 j=1 j=1
∂q_ ir j=1 ð21Þ
Zou and Wang 5

Table 1. Liquid sloshing frequencies without coupling effect (Hz).

Order Filling ratio (%)

30 40 50 60 70 80

1 0.7578 0.8142 0.8457 0.8630 0.8723 0.8773

2 1.5242 1.5288 1.5294 1.5295 1.5296 1.5296
3 1.9745 1.9747 1.9747 1.9747 1.9747 1.9747
4 2.3364 2.3365 2.3365 2.3365 2.3365 2.3365
5 2.6493 2.6493 2.6493 2.6493 2.6493 2.6493

Table 2. Tank natural frequencies without loading liquid (Hz).

Order Thickness (m)

0.01 0.013 0.016 0.02 0.024 0.027 0.03

1 2.7410 3.5631 4.3850 5.4806 6.5758 7.3969 8.2176

2 2.7410 3.5631 4.3850 5.4806 6.5758 7.3969 8.2176
3 17.173 22.316 27.453 34.288 41.106 46.205 51.290
4 17.173 22.316 27.453 34.288 41.106 46.205 51.290
5 48.070 62.433 76.753 95.759 114.64 128.71 142.70

X
n X
n X
n X
n
Numerical calculation and analysis
mbrij €qjl + k1ij qjr (t) + k2ij qjl (t) + k3ij uwj = Qi
j=1 j=1 j=1 j=1 Liquid sloshing coupling to flexible bulkheads inside a
ð22Þ two-dimensional rectangular tank is investigated in this
article. The tank is made up of plexiglass; its width (B),
X
n X
n X
n
uwj +
mwj € k4ij uwj + k5ij qjl + k5ij qjr = 0 ð23Þ height (L), density (r), Young’s modulus (E), and
j=1 j=1 j=1 Poisson’s ratio (m) are 1 m, 1 m, 1200 kg/m3, 3 3 109
Pa, and 0.363, respectively. The fluid density is 1000 kg/
Definition m3. Aiming to study sloshing problem related to cou-
pling effect, a coupled mechanical model of sloshing is
X
n X
n X
n
Mr = mbrij , Mf = mwi I, K1 = k1ij , K2 = k2ij proposed based on the spring–mass model, and modal
j=1 j=1 j=1 analysis and dynamic response calculations are carried
X
n X
n X
n X
n out under different conditions.
K3 = k3ij , K4 = k4ij , Ml = mblij , K5 = k5ij
j=1 j=1 j=1 j=1
FSI modal analysis
Then, equations (21)–(23) are expressed as In order to study the effect of FSI on the sloshing,
uncoupled p sloshing
ffiffiffiffiffiffiffiffiffiffiffi frequencies can be obtained as
2 32 3 2 32 3 2 3
Ml 0 0 q€l K1 K2 K3 ql Ql fi = (1=2p) ki =mi (i = 1, . . . , n) according to formu-
4 0 Mr 0 54 €qr 5 + 4 K2 K1 K3 54 qr 5 = 4 Qr 5 lae (1)–(5) first. We select the first five order frequencies
0 0 Mf U€f K5 K5 K4 Uf 0 (n = 5) to analyze the sloshing characteristics. The fill-
ð24Þ ing ratio ranges from 30% to 80% with an interval of
10%, and then uncoupled sloshing frequencies are cal-
where ql and qr are the generalized coordinates of the culated under different filling ratios, as shown in
left and right bulkhead, respectively; Uf is the spring– Table 1.
mass point generalized coordinates; Ql and Qr are the The bulkhead thickness is taken as follows: 0.01,
external excitations which act on the left and right bulk- 0.013, 0.016, 0.02, 0.024, 0.027, and 0.03 m, respectively,
head, respectively. In equation (24), external excitation, and natural frequencies of empty tanks for different
Q, includes force boundary and displacement bound- thickness of the bulkhead are calculated, as shown in
ary, such as external force, F, and forced displacement, Table 2.
x(t), and Q = F + m€x where m is the nodal mass. When coupling effect between liquid and tank is con-
Considering the flexibility of bulkhead for a two- sidered, modal frequencies of the coupled system are
dimensional rectangular tank sloshing, the coupled listed in Tables 3–8. The first five modes mainly repre-
dynamic equation of the proposed mechanical model is sent water wave modes, while the last five modes mainly
established and shown in equation (24). represent lower modes of tank.
6 Advances in Mechanical Engineering

Table 3. Modal frequencies of coupled system for 30% filling ratio (Hz).

Order Thickness (m)

0.01 0.013 0.016 0.02 0.024 0.027 0.03

1 0.7578 0.7578 0.7578 0.7578 0.7578 0.7578 0.7578

2 1.5159 1.5217 1.5230 1.5236 1.5239 1.5240 1.5240
3 1.9042 1.9681 1.9719 1.9733 1.9738 1.9740 1.9741
4 2.3539 2.3262 2.3335 2.3352 2.3358 2.3360 2.3361
5 2.6589 2.6247 2.6462 2.6486 2.6492 2.6494 2.6495
6 2.0213 2.7748 3.5502 4.6028 5.6681 6.4720 7.2788
7 2.0599 2.8119 3.5562 4.6046 5.6688 6.4724 7.2790
8 7.5326 10.355 13.375 17.660 22.193 25.730 29.367
9 7.5331 10.355 13.375 17.660 22.193 25.730 29.367
10 38.994 50.881 62.834 78.850 94.931 107.02 119.11

Table 4. Modal frequencies of coupled system for 40% filling ratio (Hz).

Order Thickness (m)

0.01 0.013 0.016 0.02 0.024 0.027 0.03

1 0.8044 0.8099 0.8119 0.8131 0.8136 0.8138 0.8139

2 1.4990 1.5171 1.5229 1.5258 1.5271 1.5276 1.5279
3 1.9398 1.9627 1.9689 1.9719 1.9730 1.9735 1.9738
4 2.2944 2.3250 2.3313 2.3341 2.3351 2.3355 2.3358
5 2.5894 2.6385 2.6453 2.6478 2.6487 2.6491 2.6493
6 2.7286 3.5131 4.3161 5.3984 6.4864 7.3040 8.1222
7 2.8574 3.5394 4.3257 5.4019 6.4880 7.3049 8.1228
8 8.7266 12.457 16.474 22.151 28.089 32.662 37.308
9 8.7277 12.458 16.474 22.152 28.089 32.662 37.308
10 24.815 32.743 40.908 52.139 63.727 72.630 81.693

Table 5. Modal frequencies of coupled system for 50% filling ratio (Hz).

Order Thickness (m)

0.01 0.013 0.016 0.02 0.024 0.027 0.03

1 0.7870 0.8186 0.8312 0.8383 0.8414 0.8427 0.8435

2 1.4748 1.5053 1.5169 1.5231 1.5258 1.5269 1.5276
3 1.9290 1.9556 1.9650 1.9699 1.9719 1.9727 1.9732
4 2.2936 2.3203 2.3287 2.3328 2.3344 2.3350 2.3354
5 2.6057 2.6357 2.6434 2.6469 2.6482 2.6487 2.6490
6 2.9186 3.6337 4.3991 5.4539 6.5265 7.3367 8.1497
7 3.0561 3.6776 4.4171 5.4610 6.5299 7.3389 8.1511
8 8.3565 11.955 15.851 21.389 27.211 31.710 36.293
9 8.3578 11.955 15.851 21.389 27.211 31.711 36.294
10 23.748 31.190 38.854 49.377 60.214 68.526 76.981

The parameters including liquid filling ratios and
structural stiffness influence the modal frequency of the
coupled sloshing system. Through comparison of the
results obtained from Tables 1–8, some characteristics
for coupled sloshing system are found as follows.
Liquid filling ratio has effects on the modal fre-
quency of the coupled sloshing system. It is observed
from Tables 1–8 that the first-order frequency of the
coupled sloshing system is nearly equal to the one of
the uncoupled system at 30% and 40% fill depth. But
the sixth-order frequency represented by tank in the
coupled sloshing system is smaller than that in the
uncoupled system. At 50%, 60%, and 70% fill depth,
the first-order frequencies of the coupled sloshing sys-
tem are also smaller than those in the uncoupled sys-
tem, while the sixth-order frequencies in the coupled
Zou and Wang 7

Table 6. Modal frequencies of coupled system for 60% filling ratio (Hz).

Order Thickness (m)

0.01 0.013 0.016 0.02 0.024 0.027 0.03

1 0.7211 0.7917 0.8235 0.8424 0.8510 0.8546 0.8568

2 1.4628 1.4945 1.5099 1.5194 1.5237 1.5254 1.5265
3 1.9282 1.9508 1.9616 1.9679 1.9707 1.9719 1.9726
4 2.2982 2.3180 2.3267 2.3316 2.3337 2.3345 2.3350
5 2.6146 2.6341 2.6419 2.6460 2.6477 2.6484 2.6488
6 3.1165 3.7278 4.4298 5.4352 6.4795 7.2757 8.0785
7 3.2607 3.7896 4.4583 5.4471 6.4851 7.2793 8.0809
8 6.6408 9.5106 12.649 17.172 22.007 25.798 29.706
9 6.6416 9.5108 12.649 17.172 22.007 25.798 29.706
10 24.415 31.731 39.216 49.384 59.725 67.575 75.497

Table 7. Modal frequencies of coupled system for 70% filling ratio (Hz).

Order Thickness (m)

0.01 0.013 0.016 0.02 0.024 0.027 0.03

1 0.6442 0.7469 0.7998 0.8337 0.8496 0.8563 0.8606

2 1.4589 1.4867 1.5036 1.5154 1.5211 1.5236 1.5252
3 1.9286 1.9475 1.9585 1.9660 1.9695 1.9710 1.9720
4 2.2987 2.3154 2.3245 2.3302 2.3328 2.3339 2.3346
5 2.6146 2.6320 2.6401 2.6450 2.6471 2.6479 2.6484
6 3.0081 3.5854 4.2367 5.1907 6.2008 6.9791 7.7683
7 3.1584 3.6658 4.2780 5.2090 6.2098 6.9847 7.7720
8 5.5539 7.6508 10.007 13.441 17.149 20.081 23.127
9 5.5661 7.6529 10.007 13.441 17.149 20.081 23.127
10 26.250 33.944 41.771 52.322 62.958 70.974 79.013

Table 8. Modal frequencies of coupled system for 80% filling ratio (Hz).

Order Thickness (m)

0.01 0.013 0.016 0.02 0.024 0.027 0.03

1 0.5754 0.6989 0.7702 0.8189 0.8426 0.8527 0.8593

2 1.4520 1.4786 1.4971 1.5112 1.5185 1.5217 1.5237
3 1.9184 1.9413 1.9546 1.9638 1.9683 1.9702 1.9715
4 2.2706 2.3096 2.3220 2.3293 2.3326 2.3340 2.3348
5 2.3517 2.6201 2.6355 2.6422 2.6450 2.6461 2.6467
6 2.4861 2.9243 3.5301 4.4131 5.3610 6.1005 6.8570
7 2.6745 3.0393 3.5898 4.4402 5.3746 6.1091 6.8626
8 5.6874 7.2394 9.0598 11.743 14.641 16.933 19.313
9 5.7210 7.2495 9.0633 11.744 14.642 16.933 19.313
10 29.699 38.396 47.188 58.974 70.789 79.653 88.512

sloshing system are greater than those in the uncoupled
system for more flexible bulkhead, such as the thick-
ness of the bulkhead of 0.01 m. It is also found that the
first-order frequencies and the sixth-order frequencies
in the coupled sloshing system are also smaller than
those in the uncoupled system at 80% fill depth.
Structural stiffness has effects on the modal fre-
quency of the coupled sloshing system. If thickness of
the bulkhead is smaller, namely, smaller bending stiff-
ness of the bulkhead, the coupling effect becomes stron-
ger, especially on the first two order frequencies of
fluid. When the bulkhead is thickened, such as 0.03 m,
the bending stiffness becomes larger, and the fluid
modal frequencies of coupled sloshing system are closer
to the theoretical ones of uncoupled system, namely,
coupling effect is weaker. It is also found that the water
8 Advances in Mechanical Engineering

Figure 2. Modal shape of coupled sloshing system for bulkhead: (a) filling ratio: 30%, bulkhead thickness: 0.01 m; (b) filling ratio:
30%, bulkhead thickness: 0.03 m; (c) filling ratio: 80%, bulkhead thickness: 0.01 m; and (d) filling ratio: 80%, bulkhead thickness: 0.03 m.

wave modal frequencies decrease as bending stiffness of
the bulkhead decreases. Meanwhile, the water wave
modal frequencies of the coupled sloshing system are
smaller than those of the uncoupled system for a given
liquid filling ratio.
When bending stiffness of the bulkhead becomes
larger (i.e. bulkhead thickness is greater than 0.02 m),
the sixth-order frequency represented by tank in the
coupled sloshing system first increases and then
decreases as filling ratio increases, and the first-order
frequency of non-loading liquid is the greatest. When
the bending stiffness becomes smaller (i.e. bulkhead
thickness is smaller than 0.02 m), the sixth-order fre-
quency represented by tank in the coupled sloshing sys-
tem first increases from below the sixth-order frequency
under non-loading liquid to the above of it and then
decreases the other way around as filling ratio increases.
This phenomenon illustrates that the coupling effect
has a significant influence on the mode of coupled sys-
tem. The fluid–structure coupling strength is repre-
sented by both structural stiffness and filling ratio, and
the coupled strength between structural stiffness and a
series of springs is exhibited in the non-diagonal ele-
ment of the stiffness matrix in equation (24).
Figure 2 shows the effect of bulkhead thickness and
filling ratio on the modal shapes. When the bulkhead
thickness is 0.01 m, the sixth and seventh modal shapes
represented by tank are asymmetric, such as the sixth
and seventh modal frequencies corresponding to the
fluid–structure coupling system shown in Tables 3–8.
When the bulkhead thickness is large enough, such as
0.03 m, the sixth and seventh modal shapes are sym-
metric. It illustrates that the asymmetrical characteristic
of modal shapes is more obvious when bending stiff-
ness of bulkhead is smaller, and coupling effect should
be adequately considered in this case.
Zou and Wang 9

As seen from Table 2, the first two modal frequen-

cies of empty tank are identical, and the modal shapes
are symmetrical. But the deviation between the sixth
and seventh modal frequencies represented by tank is
presented at certain filling ratio in coupled sloshing sys-
tem, as seen from Tables 3–8. The deviations lead to
asymmetry of the sixth and seventh modal shapes. As
shown in Figure 2(a) and (c), the sixth and seventh
modal shapes present the asymmetrical characteristic at
30% and 80% filling ratio, and the asymmetrical char-
acteristic is more obvious at 80% filling ratio. Although
sloshing geometric model with fluid–structure coupling
effect is symmetric, the structural modal shapes are
asymmetric when strong coupling effect is taken into Figure 3. Transient history of the horizontal displacement of
consideration. the point located on the left wall at y = 1 m, for l = L/2.
As seen from Tables 1–8, some conclusions can be
drawn:
between the results of this model and the results of the
1. Under a certain filling ratio, coupling effect literatures are carried out. Finally, this mechanical
leads to a decrease in water wave modal fre- sloshing model and FE model based on linear potential
quency, especially when bending stiffness of the flow theory are numerically simulated by commercial
bulkhead becomes smaller and filling ratio FSI code ADINA. Also, this model is verified by the
becomes higher. two comparative results and the effects of simplified
2. The parameters of filling ratio and bending stiff- mass on bulkhead bending moment are discussed.
ness of the bulkhead have an important effect A square tank (L = H = 1 m) derived from the liter-
on the first-order frequency of sloshing water ature of Bucchignani et al.7 is filled by water. The verti-
wave involving coupling effect. When the bend- cal walls are made up of steel (E = 198 3 109 Pa, m =
ing stiffness becomes larger, the first-order 0.3) and are characterized by a square section with size
water wave frequency increases as filling ratio
s = 5 cm. Free surface is subjected to an initial shape,
increases, and the first water wave frequency of
h(x, 0), under tank bottom motionlessÐ condition and
the coupled system is slightly smaller than that 1
initial potential energy, E = 1=2(rg 0 h(x, 0)dx), is
of the uncoupled system.
obtained; then, we assume that the energy totally con-
3. When the bending stiffness becomes smaller, the
verted into initial kinetic energy of spring–mass point,
first-order water wave frequency first increases
m1 , meanwhile, initial velocity (v0 = 0.022 m/s) of
(filling ratio ranges from 30% to 40%) and then
mass, m1 , is obtained. Finally, the proposed model is
decreases, and the first-order frequency of water
employed to simulate this case. For the point located
wave in the coupled system is smaller than that
on the left wall at the tip, the horizontal displacement
of the uncoupled system.
obtained from the present results and the other from
4. As filling ratio increases under smaller thickness
Bucchignani’s et al.7 is shown in Figure 3. It can be
of the bulkhead, the relative deviation of fre-
observed that both the results are essentially consistent
quencies between coupled and uncoupled sys-
despite a slight difference between the two.
tems becomes larger.
For an elastic tank in the literature of Lee et al.10
5. Considering stronger coupling effect, the first
with 50% filling ratio, its width and height are 5 and
two modal shapes represented by structure are
8 m, respectively. Young’s modulus, Poisson’s ratio,
asymmetric although sloshing geometric model
thickness, and density of the wall are 5 GPa, 0.3,
is symmetric.
0.11034 m, and 7860 kg/m3, respectively. The tank is
excited with a horizontal acceleration, a, given by a =
0.05gsin(0.4t) where g is the gravitational acceleration.
Analysis of dynamic response for the proposed
Using MPS method, numerical computation for the
coupled sloshing model tank sloshing with coupling effect is carried out by Lee
In this section, in order to validate the proposed et al.10 By using the proposed method, numerical com-
mechanical model, first, some test cases in the litera- putation of the tank sloshing is also carried out. The
tures (Bucchignani et al.,7 Lee et al.,10 and Löhner bending deflections of elastic wall are shown in
et al.20) are numerically calculated by using the pro- Figure 4. It is found that the deflections from the pro-
posed mechanical model. Then, comparative analysis posed method are consistent with those from
10
Figure 4. The bending deflection of elastic wall: (a) the results of the proposed method and (b) the result of Lee et al.10
Lee et al.10 except the right wall in the second tank of
Figure 4(a). It is also found that highly nonlinearity is
presented due to the splashing and breaking of free sur-
face, as shown in the second tank of Figure 4(b).
Because the spring–mass model is obtained from linear
potential theory in this study, this mechanical model
could not deal with the nonlinearity problem including
swirling, splashing, and breaking of free surface. As a
result, the difference of the bending deflection of the
right wall between the second tanks of Figure 4(a) and
(b) occurs when free surface presents highly
nonlinearity.
Based on volume of fluid (VOF) method, Löhner
et al.20 performed numerical computations for a rigid
tank sloshing under a given filling ratio (35%) and a
sway motion condition (x(t) = A sin(2pt/T),
A = 0.025 m, T = 1.27 s). Its length (L), width (b), and
height (H) are all 1 m. Corresponding rigid and flexible
tank are also studied by this method. For the flexible
wall, its density, elastic modulus, Poisson’s ratio, and
thickness are 1200 kg/m3, 2 3 109 Pa, 0.363, and
0.02 m, respectively.
Advances in Mechanical Engineering
The comparison of the horizontal force, Fx, calculated
by this method and by Löhner et al. is shown in Figure 5.
It is found that the transient response of the horizontal
force obtained from this method and VOF method lasted
for 40 cycles for the rigid wall. When sloshing response
turns into a steady-state regime, both the results for the
rigid wall are basically consistent although the amplitude
of this result is slightly smaller than that of Löhner’s
et al. The reason of the slight distinction is that the large-
amplitude motion of free surface is not taken into consid-
eration in the proposed model. Considering the coupling
effect, the transient response of the horizontal force
lasted for 25 cycles for the flexible wall, which is less than
that for the rigid wall. This is because part of the kinetic
energy of sloshing liquid is converted into strain energy
of the flexible wall, and the steady state of the horizontal
force is easily built up. Through comparison of the
amplitude of the horizontal force of steady state between
the rigid wall and the flexible wall, the former is larger
than the latter. The conclusion from the above analysis is
that the proposed mechanical model is reasonable for
sloshing with coupling effect.
Zou and Wang 11

Figure 5. Time history of force Fx for a tank at A/L = 0.025, T/T1 = 1.

Figure 7. Time histories of moment at p2 (30%).

Figure 6. Coupled FE sloshing model based on linear potential

flow theory is modeled by ADINA.

Aiming to investigate characteristics of dynamic

response of tank sloshing with coupling effect, the tank
bottom is subjected to a horizontal harmonic excitation
of displacement, x(t) = A sin(2pft), as shown in
Figure 1, where excitation frequency is f = 0.5 Hz, dis-
placement amplitude is A = 0.01 m, and bulkhead
thickness is 0.013 m. Filling ratio ranges from 30% to
80% with an interval of 10%. The proposed sloshing
model with coupling effect in this article and classical
coupled FE model of sloshing based on linear potential
flow theory are numerically calculated by commercial
software ADINA.
In this section, bending moments at p1 and p2 are
investigated. The positions of p1 and p2 are shown in Figure 8. Time histories of moment at p1 (30%).
Figure 1. p1 is located at the bottom of bulkhead and
p2 at an intermediate position from the free surface to ADINA. Time histories of bulkhead bending moment
bilges. for various filling ratios are shown in Figures 7–18. The
The coupled FE sloshing model based on linear relationship between filling ratios and amplitude of
potential flow theory is modeled, as shown in Figure 6. bending moment is shown in Figures 19 and 20. The
The model is calculated by the commercial FSI code solid line represents the results of the proposed
12 Advances in Mechanical Engineering

Figure 9. Time histories of moment at p2 (40%). Figure 12. Time histories of moment at p1 (50%).

Figure 10. Time histories of moment at p1 (40%). Figure 13. Time histories of moment at p2 (60%).

Figure 11. Time histories of moment at p2 (50%). Figure 14. Time histories of moment at p1 (60%).

mechanical model and the dashed line represents the

calculation results of ADINA. conditions, it can be observed that the bending moment
Through analysis of time histories of bending caused by sloshing is almost the same in the two kinds
moment (shown in Figures 7–18) under different filling of calculation models. It proves once again that the
Zou and Wang
Figure 15. Time histories of moment at p2 (70%).
Figure 16. Time histories of moment at p1 (70%).
Figure 17. Time histories of moment at p2 (80%).
proposed mechanical model is reasonable. The time
history curves are periodical. The period of bending
moment at p1 is equal to the excitation period. Also, it
is found that the time histories of bending moment at
13
Figure 18. Time histories of moment at p1 (80%).
Figure 19. Amplitude of bending moment at p1 under various
filling ratios.
p2 include some harmonic components though the
curves are periodical. As a result, it is concluded that
the bending moment near the liquid-free surface con-
tains different components of frequencies.
As seen from Figure 19, when filling ratio is no more
than 50%, the amplitude of bending moment at p2 cal-
culated by ADINA for the FE model based on classical
linear potential flow theory is greater than that from
this mechanical model in this article. When filling ratio
is above 50%, the results are on the contrary. The max-
imum deviation of the amplitude of bending moment at
p2 is 1.5 N m, and the relative deviation is 14.56%
under 70% filling ratio condition. The bending moment
at p2 nonlinearly increases as filling ratio increases. The
reason is that sloshing loading at p2 near the liquid-free
surface nonlinearly varies as filling ratio increases.
As can be seen in Figure 20, the amplitude of bend-
ing moment at p1 calculated by ADINA for the FE
model based on classical linear potential flow theory is
smaller than that of this mechanical model under all six
loading conditions. The maximum deviation of the
amplitude of bending moment is 11.5 N m, and the
14
relative deviation is 6.18% under 70% filling ratio con-
dition too. It is also found that the amplitude of bend-
ing moment at p1 basically linear increases as filling
ratio increases.
By using fast Fourier transform (FFT) technology,
time histories of the bending moment at p1 and p2 are
Figure 20. Amplitude of bending moment at p2 under various
filling ratios.
Figure 21. The amplitude spectrum analysis of bending moment at p2.
Figure 22. The amplitude spectrum analysis of bending moment at p1.
Advances in Mechanical Engineering
analyzed in frequency domain. As shown in Figure 21,
frequency component of the bending moment at p2
contains exciting frequency (0.5 Hz) and the first-order
and the second-order water wave frequency. Under all
six loading conditions, the amplitude of bending
moment corresponding to the first-order water wave
frequency is larger than that corresponding to the excit-
ing frequency. The amplitude of bending moment cor-
responding to the second-order water wave frequency
is very small and even negligible. The results reveal that
the bending moment of the bulkhead near the free sur-
face is dominant contributed by the dynamic load, and
the load is induced by the movement of liquid-free sur-
face, that is, the mass point, m1 , and spring, k1 , in this
mechanical model with coupling effect.
As can be seen in Figure 22, frequency component of
the bending moment at p1 contains excitation fre-
quency (0.5 Hz) and the first-order water wave fre-
quency. At this location, the amplitude of bending
moment corresponding to the first-order frequency is
almost negligible by comparing with that corresponding
to the exciting frequency under all six loading
Zou and Wang
conditions. The results demonstrate that the bending
moment of the bulkhead near the bottom is dominant
contributed by the inertia of rigid body mass, m0 , while
the dynamic loads caused by mass, m1 , and spring, k1 ,
have little effect on the bending moment of bulkhead
near the bottom.
Conclusion
In this article, the simplified mechanical model with
coupling effect for a rectangular tank sloshing sub-
jected to a horizontal excitation is proposed and the
coupled dynamic equation of sloshing is deduced. The
effects of bulkhead bending stiffness and filling ratios
on the mode of coupled system are investigated, and
dynamic response of the coupled system is carried out.
In this study, the main research results are summarized
as follows:
1. Coupling effect leads to a decrease in water
wave frequencies.
2. As filling ratio increases, structural modal fre-
quencies first increase and then decrease.
3. Higher filling ratio and smaller bulkhead thick-
ness cause stronger coupling effect, which can
lead to asymmetry of modal shapes for the sym-
metrical structure.
4. The simplified mass, m0 , dominates the contri-
butions to bending moment near the bottom of
the bulkhead, and the simplified spring–mass,
k1 and m1 , dominate the contributions to bend-
ing moment near the liquid-free surface of the
bulkhead.
5. By comparing the results of this mechanical
model with those of other literatures, the pro-
posed mechanical model of sloshing with cou-
pling effect is reasonable for a rectangular tank
under a horizontal excitation.
In this study, the proposed mechanical model of
sloshing is only effective for a rectangular tank sub-
jected to a horizontal excitation, since the rotational
inertia of spring–mass point and tank is not considered
for the excitation of pitching. Further researches on the
nonlinearity of sloshing under combined excitation
including heaving and pitching excitation will be car-
ried out based on the proposed mechanical model with
coupling effect.
Declaration of conflicting interests
The authors declare that there is no conflict of interests
regarding the publication of this article.
15
Funding
The work presented in this article has been carried out under
the co-support provided by the Ministry of Education and
Ministry of Finance of China (Grant No. 201335), the Key
Project of Natural Science Foundation of China (Grant No.
51239007) and the NSFC (51239007).
References
1. Kim YS. A spurious free four-node displacement-based
fluid element for fluid–structure interaction analysis. Eng
Struct 1997; 19: 665–678.
2. Ibrahim RA. Liquid sloshing dynamics—theory and appli-
cations. Cambridge: Cambridge University Press, 2005,
p.948.
3. Livaoglu R. Investigation of seismic behavior of fluid–
rectangular tank–soil/foundation systems in frequency
domain. Soil Dyn Earthq Eng 2008; 28: 132–146.
4. Tabri K, Matusiak J and Varsta P. Sloshing interaction
in ship collisions—an experimental and numerical study.
Ocean Eng 2009; 36: 1366–1376.
5. Tsukamoto MM, Cheng LY and Nishimoto K. Analyti-
cal and numerical study of the effects of an elastically-
linked body on sloshing. Comput Fluid 2011; 49: 1–21.
6. Li YC and Wang JT. A supplementary, exact solution of
an equivalent mechanical model for a sloshing fluid in a
rectangular tank. J Fluid Struct 2012; 31: 147–151.
7. Bucchignani E, Stella F and Fabio Paglia F. A partition
method for the solution of a coupled liquid–structure
interaction problem. Appl Numer Math 2004; 51:
463–475.
8. Idelsohn SR, Oñate E, Pin FD, et al. Fluid–structure
interaction using the particle finite element method.
Comput Meth Appl Mech Eng 2006; 195: 2100–2123.
9. Zhang AN and Suzuki K. A comparative study of
numerical simulations for fluid–structure interaction of
liquid-filled tank during ship collision. Ocean Eng 2007;
34: 645–652.
10. Lee CJK, Hirohisa N and Seiichi K. Fluid–shell structure
interaction analysis by coupled particle and finite element
method. Comput Struct 2007; 85: 688–697.
11. Kim Y, Nam BW, Kim DW, et al. Study on coupling
effects of ship motion and sloshing. Ocean Eng 2007; 34:
2176–2187.
12. Mitra S and Sinhamahapatra KP. 2D simulation of
fluid–structure interaction using finite element method.
Finite Elem Anal Des 2008; 45: 52–59.
13. Degroote J, Souto-Iglesias A, Paepegem WV, et al. Parti-
tioned simulation of the interaction between an elastic
structure and free surface flow. Comput Meth Appl Mech
Eng 2010; 199: 2085–2098.
14. Kassiotis C, Ibrahimbegovic A and Matthies H. Parti-
tioned solution to fluid–structure interaction problem in
application to free-surface flows. Eur J Mech B Fluid
2010; 29: 510–521.
15. Cao Y, Wang PY and Jin XL. Dynamic analysis of flex-
ible container under wind actions by ALE finite-element
method. J Wind Eng Ind Aerod 2010; 98: 881–887.
16
16. Pan XJ, Zhang HX and Sun XY. Numerical simulation
of sloshing with large deforming free surface by MPS-
LES method. China Ocean Eng 2012; 26: 653–668.
17. Akyildiz H and Unal E. Experimental investigation of
pressure distribution on a rectangular tank due to the
liquid sloshing. Ocean Eng 2005; 32: 1503–1516.
18. Anghileri M, Castelletti LML and Tirelli M. Fluid–
structure interaction of water filled tanks during the
Advances in Mechanical Engineering
impact with the ground. Int J Impact Eng 2005; 31:
235–254.
19. Carra S, Amabili M and Garziera R. Experimental study
of large amplitude vibrations of a thin plate in contact
with sloshing liquids. J Fluid Struct 2013; 42: 88–111.
20. Löhner R, Yang C and Oñate E. On the simulation of
flows with violent free surface motion. Comput Meth
Appl Mech Eng 2006; 195: 5597–5620.