Applied Mathematical Modelling 34 (2010) 24262436

Contents lists available at ScienceDirect

Applied Mathematical Modelling

journal homepage: www.elsevier.com/locate/apm

Importance of the exural and membrane stiffnesses in large

deection analysis of oating roofs
R. Shabani a,*, S. Tariverdilo b, H. Salarieh c, G. Rezazadeh a
a

Mech. Eng. Dep., Faculty of Eng., Urmia Univ., Urmia, Iran

Civil Eng. Dep., Faculty of Eng., Urmia Univ., Urmia, Iran
c
Faculty of Mech. Eng., Sharif Univ. of Tech., Tehran, Iran
b

a r t i c l e

i n f o

Article history:
Received 7 April 2009
Received in revised form 4 November 2009
Accepted 6 November 2009
Available online 18 November 2009
Keywords:
Floating roof
Variational principle
Large deection

a b s t r a c t
Applying integrated variational principles on uid and deck plate to the large deection
analysis of oating roofs, this paper investigates the signicance of the exural and membrane components in the formulations of the deck plate. Integrated variational principles
facilitate the treatment of the compatibility of deformation between oating roof and supporting liquid. Analysis results show that different assumptions about deck plate formulation commonly used in the literature, results in considerably different deection and stress
patterns on the oating roof. The results show that modeling of the deck plate as a exural
element rather than the membrane, by eliminating the need for nonlinear analysis, gives
reasonable results for deections and stresses in the deck plate. Finally, to check the results
of the variational formulation, employing Bessel functions and ignoring membrane stiffness an approximate solution is derived and its results compared with those of the variational formulation. This comparison shows that the approximate solution closely follows
the variational formulation.
2009 Elsevier Inc. All rights reserved.

1. Introduction
Floating roofs are used in the petroleum industries for storage of liquid hydrocarbons in atmospheric storage tanks. By
reducing the evaporation of storage materials, oating roofs provide better protection against possible ignition of the vapors
by sparks generated by different sources such as cigarette smoking, earthquake, static electricity, etc. [1]. In addition to economic benet by preventing the evaporation of valuable products, these roofs are also helpful in reduction of the environmental pollution caused by evaporation. The conventional oating roofs can be categorized into two types: single deck and
double deck. The single deck oating roofs, which are the subject of this study, consist of a circular deck plate and a circumferential pontoon and in some cases a central pontoon. In these roofs in addition to the pontoon buoyancy, the buoyancy
provided by the liquid under deck plate is also essential in the equilibrium of the system.
Due to large deection of the deck plates near the edges, it was anticipated that large deection analysis would be important in the analysis of the deck plate. With this in mind, different researchers adopted different formulations for the deck
plate in their analyses. Mitchell [2] investigated the stability of the pontoon in oating roofs. Ignoring exural deformation
and applying large deection formulation on deck plate, he evaluated the forces exerted on the pontoon from the deck plate.
Then with the assumption of small displacements for pontoon, he investigated the possibility of in plane, out of plane and
torsional buckling of the pontoon due to forces exerted on it by the deck plate. Epstein [3] used the shooting method with
* Corresponding author. Tel.: +98 441 2972947; fax: +98 441 2972901.
E-mail address: r.shabani@urmia.ac.ir (R. Shabani).
0307-904X/$ - see front matter 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.apm.2009.11.008

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

2427

RungeKutta numerical integration technique to solve the boundary value problem of the deection of a oating roof. He
used a large deection formulation ignoring exural stiffness of the deck plate. Yuan et al. [4] introduced a method based
on an equivalent rst order ordinary differential equation to solve the deection and stress distribution in oating roofs with
circumferential and central pontoons. They derived the equivalent formulation in the state space and then proposed a solution algorithm to solve the resulting rst order ordinary differential equations. They used a large deection formulation
including exural deformation. Nerantzaki et al. [5] developed a boundary element formulation to study the effect of the
ponding of rainwater on oating roofs, modeled as a membrane. Sun et al. [6] after deriving the formulation of the load
and deformation, proposed an iterative method to modify the buoyancy forces due to supporting liquid and applied load
due to rainwater accumulation on the top of the deck plate. Nagata et al. [7] and Ohmatsu [8] analyzed a rectangular large
oating structure using a semi-analytical approach based on eigenfunction expansions in the depth direction. Seto and Ochi
[9] developed a hybrid element method (as a combination of the nite element method and innite element) for hydroelastic analysis of a large oating structure in stepped-depth conguration. Iijima et al. [10] analyzed the hydroelastic behavior
of semi-submersible type oating structures using their program named VODAC.
Variational principles greatly simplify treating the compatibility of deformation between the deck plate and the contained uid. Isshiki and Nagata [11] by integrating Hamiltons variational principle for plate and Kelvins principle for liquid
derived the HamiltonKelvin principle. Then by introducing the velocity potential as an alternative variable instead of the
uid velocity, they developed different schemes of so-called HamiltonDirichlets principles. Nagata et al. [12] and Ohmatsu
[13] successfully applied this variational formulation to analyze an elastic oating plate. Investigating the possible causes of
sinking of the oating roofs during Niigata earthquake, Sakai et al. [14] used the variational principle to study the sloshing
behavior of oating roofs. By considering only exural stiffness, they developed a linear formulation to study the effect of
presence of the oating roof, on the sloshing behavior of the contained liquid during earthquakes.
As discussed above, different researchers adopted different hypotheses about the deck plate in their formulations. Some
(e.g. [3]) ignoring the exural stiffness of the deck plate used the large deection formulation to develop their derivations,
while some others (e.g. [12]) to have a linear formulation ignored the deck plate membrane stiffness and developed their
formulation by taking into account only the exural stiffness. On the other hand some other peoples (e.g. [4]) considered
the exural and membrane stiffnesses altogether in their derivations.
Extending the use of variational formulation to the nonlinear case, this paper derives the large deection formulation of
the oating roofs with single deck and circumferential pontoon, by applying variational principle simultaneously on the deck
plate and the liquid. Then by imposing some simplifying assumption the importance of the exural and membrane components on the deck plate deection and stresses are evaluated. It is shown that retaining the exural stiffness, while ignoring
the membrane stiffness results in good estimate for the deck plate deections and stresses. By this way, it is possible to evaluate these deections and stiffnesses using simple analyses employing commercial linear nite element codes.
2. The variational principle
Floating roofs essentially consist of a circular deck plate and an outer circumferential pontoon as is shown in Fig. 1. Applying variational principles simultaneously on the deck plate and the liquid facilitates the problem of imposing the compatibility of deformation between the oating roof and the supporting liquid. It also eliminates the need for two-phase analysis
of the system as a coupled eld. In the following derivations, two main assumptions are made. The rst is that the supporting
liquid and the oating roof are in full contact, which seems reasonable for a stationary liquid. The second is the assumption
of rigid pontoon. This hypothesis is justied by the dimension of the pontoon being large in comparison with the deck plate.
Assuming that the plate and the uid are always in contact with each other, the Lagrangian of the deck plate (Ld) and uid
(Lf) reduces to [11]

C
L

Pontoon

T
b
a

Cp
Deck Plate

Fig. 1. Schematic diagram of a oating roof with circumferential pontoon.

2428

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

L Ld1 Ld2 Lf

1
rrr err rhh ehh rrh erh dV 
2

Vd

Z
Ad

qd gwhdA

1
2

qgw2 dA;

Ad

where rij and eij denote the deck plate stresses and strains, Vd and Sd are deck plate volume and area (deck plateuid interface), q and qd are uid and deck plate density, h is deck plate thickness and w shows the plate deection relative to the
pontoon. Note that the plate Lagrangian includes the strain energy of the deck plate and its weight, and the uid Lagrangian
takes into account the buoyancy force exerted on the deck plate by uid. Using the principle of virtual work and equating the
variation of Ld and Lf to the variation of the work done by pontoon buoyancy force (Wp), we have

dLd Lf dW p :

Variation of the strain energy component of deck plate Lagrangian becomes

dLd1

rrr derr rhh dehh rrh derh dV:

Considering the large deection theory and assuming that the in-plane displacements are innitesimal and ignoring nonlinear terms in straindisplacement relation due to this in plane displacements (von Karman theory), the straindisplacement
relation reads as [15]

 2
@ur 1 @w
;

@r 2 @r

2
1 @uh ur 1 1 @w
ehh
;

r @h
r 2 r @h
1 @ur @uh uh 1 @w @w
erh
:
;


r @h
r @r @h
@r
r

err

where the displacements are

@w
;
@r
z @w
:
uh u0h 
r @h

ur u0r  z

Here u0h and u0r denote the tangential and radial displacements in the neutral axes of the deck plate. Taking into account the
axisymmetry of the solution, Eq. (4) takes the following form:

 2
@u0r
@ 2 w 1 @w
;
z 2
@r
2 @r
@r
0
u
z @w
ehh r
;
r @r
r
erh 0

err

Due to large stiffness of the pontoon in comparison with the deck plate, ignoring any deformation in the pontoon, we use the
assumption of rigid pontoon. To calculate the work done by the pontoon buoyancy, we need to know the level of free surface
(T). Considering the equilibrium of the pontoon in the vertical direction, we have
2

2pbQjrb T  C p q  qd g pa2  b F p ;

where Q denote shear in the deck plate. Considering the axisymmetry of solution, the moment and shear in the plate could
be evaluated using the following equations:
2

1 dw

2
r dr
dr
!
2
d d w 1 dw
Q

dr dr 2
r dr
M

d w

Employing pontoon buoyancy force (Fp) from Eq. (7), the variation of the work done by the pontoon buoyancy will be

dW p F p dw:

Now to apply an approximate Ritz solution, assume that w and ur are functions of known interpolation functions fi and gi,
then we have

I
X
i1

Bi fi r;

u0r

J
X

ci g i r;

i1

where Bi and ci are the unknown coefcients.

10

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

2429

Substituting w and ur in the plate strain energy and evaluating the variational equilibrium, we will have
J
I
I X
I X
I X
I
X
X
X
Q ij qgU ij Bj
glki Bl ck
jlkmi Bl Bk Bm
j1

l1 k1 m1

l1 k1

J
I X
X

klki Bl ck Q i ;

11

l1 k1
J
X

Hij cj

I X
I
X

j1

llki Bl Bk

l1 k1

I X
I
X

hlki Bl Bk 0;

l1 k1

where

Q ij 2pD

rfj00 rfi00 r

glki

2pEh
1  m2

jlkmi

fl0 rg 0k rfi0 rrdr:

fi rrdr F p fi b

pEh
1  m2
2

klki

Q i 2pqd gh

2pEh m
1  m2

U ij 2p

1 0
m
m
f rfi0 r fj00 rfi0 r fi00 rfj0 r dr;
r2 j
r
r

12

fl0 rfk0 rfm0 rfi0 rr dr

fl0 rg k rfi0 r dr:

fj rfi rrdr:

Hij

2pEh
1  m2



1
m
m
r g 0j rg 0i r 2 g j rg i r g 0j rg i r g j rg 0i r dr;
r
r
r

b
0

llki

pEh
1  m2

hlki

pEhm
1  m2

b
0

fl0 rfk0 rg 0i rrdr;

1 0
f rfk0 rg i rrdr;
r l

where D is dened as
3

Eh
:
121  m2

13

Solving the resulting set of nonlinear equations, it is possible to calculate the deck plate deformation and stresses.
To satisfy the axisymmetric boundary conditions, the assumed interpolation functions for w and ur should satisfy the following conditions:

w;r r; hjr0 0;

u0r rjr0 0:

14

To satisfy these boundary conditions the following interpolation functions are used in this study

fn r cos

npr 
;
b

g n r sin

npr 
:
b

15

The resulting set of nonlinear equations solved using the NewtonRaphson method. The solution requires an iterative
scheme since the free surface level (T), is not know priory. Assuming a value for T an iterative scheme could be developed
deriving new estimation for it using Eq. (7). The scheme is sound and with few iteration, it is possible to reduce the order of
the error to acceptable level.

2430

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

3. Simulation results
To assess the impact of different formulations on the global and local response of a typical oating roof, the deection and
stress distributions of the deck plate are calculated for formulations including membrane and exural stiffnesses, exural
stiffness only, and nally membrane stiffness only. To obtain the results for the case of exural stiffness only, the nonlinear
terms in the strain equations (Eq. (4)) have been ignored and the results with membrane stiffness only is calculated by ignoring the second terms on the right hand side of the displacement equations (Eq. (5)). To assess the deection and stress pattern of the deck plate as affected by the presence of the pontoon, two examples are considered in this study. The data used in
these examples are shown in Table 1, where there is a smaller pontoon size and consequently smaller pontoon buoyancy in
the second example.
Fig. 2 depicts the deck plate deection for tanks of 20, 40 and 60 m diameters evaluated assuming different stiffness
schemes. To have a better resolution on the deections resulting from the use of different schemes, the deections for
20 m diameter deck is depicted again in Fig. 3. As can be seen, the formulation with membrane stiffness results in the largest
deections, while the deection for the case that includes membrane and exural stiffness is the smallest one. Considering
only the exural stiffness results in at most 5% increase in the deection, while with the membrane stiffness the increase in
deection is about 22%.
Fig. 4 shows the stress on the bottom side of the deck plate for decks of aforementioned diameters, calculated with
different assumptions. The stress pattern for decks of different diameters are similar. Again to have more insight on the difference in the stress pattern due to the use of different stiffness schemes, the stress distribution for 20 m diameter deck is
reproduce (with better resolution) in Fig. 5. Examining the total stress at the bottom face of the deck plate shows that in the
case of exural stiffness the resulting stress pattern and its magnitude are very similar to that predicted using the assumption of exural and membrane stiffnesses and in fact the difference between these cases is insignicant. At the same time the
difference in the stress pattern and magnitude as calculated by both of aforementioned methods and those calculated by the
membrane stiffness is substantially different. In another word, using the assumption of membrane stiffness, the resulting
deection pattern is misleading and the resulting estimate of the stress magnitude in the deck plate (particularly near
the pontoondeck interface) will be grossly unconservative.
Table 1
The data used in the study.
Pontoon cross sectional dimension in the rst example
Pontoon cross sectional dimension in the second example
Pontoon thickness
Pontoon depth below deck plate, Cp
Liquid density, q
Deck plate thickness, h
Deck plate density, qd
Deck plate Poissons ratio, m
Deck plate modulus of elasticity, E

5  0.4 m
2  0.4 m
5 mm
0.1 m
700 kg/m3
5 mm
7800 kg/m3
0.3
2.1  1011 N/m2

flexural stiffness only

membrane and flexural stiffness
membrane stiffness only

w (m)

-0.01

-0.02

pontoon
width

-0.03

-0.04

10

20

30

r (m)

40

50

60

Fig. 2. Comparison of the deck plate deection using different formulations, for decks of different diameters in the rst example.

2431

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

flexural stiffness only

membrane and flexural stiffness
membrane stiffness only

w (m)

-0.01

-0.02

pontoon width

-0.03

-0.04

10

15

r (m)

20

Fig. 3. Comparison of the deck plate deection using different formulations, for 20 m diameter deck in the rst example.

flexural stiffness only

membrane and flexural stiffness
membrane stiffness only

10
5

stress (MPa)

0
-5
-10

pontoon
width

-15
-20
-25
-30

10

20

30

r (m)

40

50

60

Fig. 4. Comparisons of the stress distributions in the bottom face of the deck plate, calculated using different stiffness assumptions for decks of different
diameters in the rst example.

To have an assessment of the effect of the pontoon buoyancy on the deection and stress pattern of deck plate, we considered a second example, where smaller pontoon is used. In this example, the same parameters as in rst example are used
(see Table 1) with the exception of the pontoon dimension. Fig. 6 depicts the deection pattern of the deck plate for this case.
To have a closer look at the situation, the deection pattern for deck plate in the case of 20 m diameter deck is replotted in
Fig. 7. Comparing these gures with those of Figs. 2 and 3, it is clear that pattern of deection for both examples are similar.
Again we have insignicant differences between the case of exural stiffness and the case of exural and membrane stiffnesses. It should be noted that the difference in the deection magnitude for the rst and the second examples originates
from the different value of the T, required for equilibrium in these examples. Fig. 8 depicts the stress pattern for the second
example. Decreasing the size of the pontoon, the stress magnitude in the deck plate near pontoondeck interface, decreases
signicantly, however the pattern of stress is similar for both examples, and for both cases ignoring the exural stresses is
grossly unconservative. Also the accuracy of the solution in both examples for the scheme that only considers the exural
stiffness is very good.

2432

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

flexural stiffness only

membrane and flexural stiffness
membrane stiffness only

10
5

stress(MPa)

0
-5

pontoon width

-10
-15
-20
-25
-30

10

15

r(m)

20

Fig. 5. Comparisons of the stress distributions in the bottom face of the deck plate, calculated using different stiffness assumptions for 20 m diameter deck
in the rst example.

flexural stiffness only

membrane and flexural stiffness
membrane stiffness only

pontoondeck
interface

w (m)

-0.01

-0.02

10

20

30

r (m)

40

50

60

Fig. 6. Comparison of the deck plate deection using different formulations, for decks of different diameters in the second example.

The results show that different assumptions regarding the stiffness formulation of the deck plate could result in substantially different patterns of deection and stress on the deck plate. Considering only the membrane stiffness could be extremely unconservative and at the same time, it yields excessively large estimate of the actual deection. On the other hand
reviewing Figs. 2 and 3 shows that designing deck plate assuming only the exural stress will be conservative for both stress
and deection evaluation. The linearity of governing equations in this case also increases its attractiveness.
To develop an approximate solution to the problem, consider the governing equation for an axisymmetric oating plate
ignoring the membrane stiffness
2

1 d
D

2
r dr
dr

!
1 dw
qd t d g  qwg:

2
r dr
dr
2

d w

16

By introducing the following notations

qg
D

1
l

r
x :
l

17

2433

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

flexural stiffness only

membrane and flexural stiffness
membrane stiffness only

pontoon-deck interface

w (m)

-0.01

-0.02

10

20

r (m)

Fig. 7. Comparison of the deck plate deection using different formulations, for 20 m diameter deck in the second example.

flexural stiffness only

membrane and flexural stiffness
membrane stiffness only

stress (MPa)

-5

pontoondeck
interface

-10

-15

10

20

30

r (m)

40

50

60

Fig. 8. Comparisons of the stress distributions in the bottom face of the deck plate, calculated using different stiffness assumptions for decks of different
diameters in the second example.

Eq. (16) reads

! 2
!
4
1 d
d w 1 dw
l

q td g:
2
2
x dx
x dx
D d
dx
dx
2

18

Decomposing the solution to particular (wp) and homogenous (wh) ones, we have

w wp wh :

19

The particular solution will be

4

wp

l
q td g:
D d

20

2434

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

Now with the notation

!
1 d
;

2
x dx
dx
2

r2x

21

the homogenous form of Eq. (18) becomes

r2x r2x wh wh 0:

22

By introducing variable n as

p
n x i:

23

And rewriting the differential equation with respect to n, we have

r2n r2n wh  wh 0;

24

however, this is equivalent to

r2n r2n wh r2n wh  r2n wh wh 0;

25

which can be satised if w satises the following equation:

r2n wh wh 0

26

or
2

d wh
dn2

1 dwh
wh 0;
n dn

27

this is the Bessel differential equation of rst kind. To derive another set of solution for Eq. (24), this equation could be
rewritten as

r2n r2n wh  r2n wh r2n wh  wh 0;

28

which can be satised if w satises the following equation

r2n wh  wh 0

29

or
2

d wh
dn2

1 dwh
 wh 0:
n dn

30

Note that this equation could be derived from Eq. (27) by substituting ni instead of n in that equation. Finally the combined
solution will be [16]

 p 
 p 
 p 
 p 
wh A1 J 0 x i A2 J 0 ix i A3 Y 0 x i A4 Y 0 ix i ;

31

where J0 and Y0 are Bessel functions of rst and second kinds. In order to express the solution in terms of real functions, we
dene w as a linear summation of four real functions [17]

wh B1 berr=l B2 beir=l B3 kerr=l B4 keir=l;

32

where we have

berx realJ 0 xeip=4 ;

33

beix imag J 0 xeip=4 ;

kerx 
keix

p
2

p
2


realY 0 xeip=4 imagJ 0 xeip=4 ;


imagY 0 xeip=4  realJ 0 xeip=4 :

The third and forth terms rapidly tends to zero as r increases, therefore it is customary to ignore these terms in the solution
of the nite plates [18]. Therefore for the solution of this fourth order differential equation, we need to have two boundary
conditions and another equation to determine the level of free surface. Applying the following boundary conditions, it is possible to determine the constants Bi and free surface level T

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

2435

variational formulation
approximate solution by Bessel functions
0

w (m)

-0.01

-0.02

pontoon width

-0.03

-0.04

10

15

20

r (m)
Fig. 9. Comparison of the deck plate deection in rst example for 20 m deck diameter as calculated using variational formulation and approximate
solution using Bessel functions.

wjrb 0;

34

dw
0;
dr rb
2

Q jrb T  C p q  qd g pa2  b =2pb:

The last equation is a restatement of the shear equilibrium in the pontoondeck interface. This equation also provides a
means for iterative solution of the resulting set of equations. First, a value for the free surface level (T) is considered, then
applying the rst two boundary conditions of Eq. (34), an estimation of the deck plate deection is obtained. Finally after
calculating the shear in the deck plate using Eq. (8), it is possible to re-evaluate the free surface height (T) employing the
last equation in Eq. (34). In general, few iterations are enough to obtain an acceptable accuracy.
Fig. 9 depicts the deection pattern as calculated using the approximate solution employing Bessel functions and using
the variational formulation with exural and membrane stiffnesses, for rst example and for 20 m deck diameter. There is
good correlation between approximate and variational solutions. Due to good correlation between deection pattern as calculated by aforementioned methods, the stress distribution pattern for approximate solution employing Bessel functions closely follows those of variational formulation, and there are negligible differences between two solutions in this regard.
4. Conclusions
Simultaneous application of the variational principle on the deck plate and the supporting liquid, the large deection formulation of the system composed of the oating roof and the liquid is derived. Simulation results show that different
assumptions regarding the stiffness of the deck plate, results in signicantly different patterns of stress and deection in
the oating roof. It is shown that the best possible simplication will be to use only the exural stiffness for the deck plate,
which results in a reasonable estimate of the stress and deection in the deck plate. In the end, for analyzing the oating roof
considering only exural stiffness an approximate solution using Bessel functions is developed. Comparison of the results of
this approximate solution to those of the simplied variational formulation, reveals that the approximate solution closely
follows the simplied variational solution results.
References
[1]
[2]
[3]
[4]
[5]
[6]

J.I. Changa, C.-C. Linb, A study of storage tank accidents, J. Loss Protect. Process Ind. 19 (1) (2006) 5159.
G.C. Mitchell, Analysis and stability of oating roofs, J. Eng. Mech. ASCE 99 (10) (1973) 10371052.
H.I. Epstein, Stress and displacements for oating pan roofs, Comput. Struct. 15 (4) (1982) 433438.
S. Yuan, J. Wang, H. Zhong, Analysis of oating roofs by ODE-solver method, J. Eng. Mech. ASCE 124 (10) (1998) 11291134.
M.S. Nerantzaki, T. John, J.T. Katsikadelis, Ponding on oating membranes, Eng. Anal. Bound. Elem. 27 (2003) 589596.
X. Sun, Y. Liu, J. Wang, Z. Cen, Strength and buoyancy analyses of oating roof with continuous beams and dome frames under rainwater loading, J.
Pressure Vessel Technol. ASME 129 (1) (2007) 7382.

2436

R. Shabani et al. / Applied Mathematical Modelling 34 (2010) 24262436

[7] S. Nagata, H. Yoshida, T. Fujita, H. Isshiki, The analysis of wave-induced response of an elastic oating plate in a sea with a breakwater, in: Proceedings
of 8th International Offshore and Polar Engineering Conference, Montreal, Canada, vol. I, 1998, pp. 223230.
[8] S. Ohmatsu, Numerical calculation of hydroelastic behavior of pontoon type VLFS in waves, in: Proceedings of 17th International Conference on
Offshore Mechanics and Arctic Engineering, Lisbon, Portugal, OMAE98-4333, 1998.
[9] H. Seto, M. Ochi, A hybrid element approach to hydroelastic behavior of a very large oating structure in regular waves, in: M. Kashiwagi, W.
Koterayama, Ohkusu (Eds.), Hydroelasticity in Marine Technology, RIAM, Kyushu University, 1998, pp. 185193.
[10] K. Iijima, K. Yoshida, H. Suzuki, Hydroelastic analysis of semisubmersible type VLFS capable of detailed structural analysis, in: M. Kashiwagi, W.
Koterayama, M. Ohkusu (Eds.), Hydroelasticity in Marine Technology, RIAM, Kyushu University, 1998, pp. 211218.
[11] H. Isshiki, S. Nagata, Variational principles related to motion of an elastic plate oating on a water surface, in: Proceeding of the 11th International
Offshore and Polar Engineering Conference, Stavanger, Norway, 2001, pp. 190197.
[12] S. Nagata, H. Yoshida, T. Fujita, H. Isshiki, The analysis of the wave induced responses of an elastic oating plate, in: Proceeding of the 16th
International Conference on Offshore Mechanics and Arctic Engineering, vol. VI, 1997.
[13] S. Ohmatsu, Numerical calculation of hydro-elastic responses of pontoon type VLFS, J. Soc. Naval Architect. Jpn. 182 (1997) 329340.
[14] F. Sakai, M. Nishimura, H. Ogawa, Sloshing behavior of oating-roof oil storage tanks, Comput. Struct. 19 (12) (1984) 183192.
[15] M. Amibili, Nonlinear Vibration and Stability of Shells and Plates, Cambridge University Press, 2008.
[16] S. Timoshenko, Theory of Plates and Shells, McGraw-Hill, 1959.
[17] G.N. Watson, Theory of Bessel Functions, Cambridge University Press, 1922.
[18] E. Ventsel, T. Krauthammer, Thin Plates and Shells: Theory, Analysis and Applications, Marcel Dekker Inc., 2001.