Ocean Engineering 168 (2018) 60–82

Contents lists available at ScienceDirect

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

Coupled analysis of integrated dynamic responses of side-by-side offloading T

FLNG system
Dongya Zhaoa,c, Zhiqiang Hub,∗, Ke Zhoua,c, Gang Chena,d, Xiaobo Chene, Xingya Fenge
a
State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai, China
b
School of Engineering, Newcastle University, Newcastle upon Tyne, NE1 7RU, UK
c
Collaborative Innovation Center for Advanced Ship and Deep-Sea Exploration, Shanghai Jiao Tong University, Shanghai, China
d
Marine Design & Research Institute of China, Shanghai, China
e
Deepwater Technology Research Centre, Bureau Veritas, Singapore

A R T I C LE I N FO A B S T R A C T

Keywords: Integrated dynamic responses of FLNG system in side-by-side offloading operation are investigated numerically
FLNG system and experimentally in this paper. A numerical code is developed based on potential flow theory to predict the
Coupling interaction interactions between connected vessels' motions and liquid sloshing in the time domain. The impulsive response
Sloshing function (IRF) method is adopted in the resolution for 6 DOF vessels' motions, nonlinear sloshing in liquid tanks
Potential flow theory
is solved using boundary element method (BEM), and connection system including hawsers and fenders is nu-
Model test
merically modeled as linear response system. A series of model tests are conducted to validate the feasibility of
the numerical code. Hydrodynamic interaction between the vessels and shielding effects under different wave
directions are analyzed; sloshing effects on the vessels' motions and on the loads of connection system are
investigated; the sensitivities of the vessels' motions and loads to connection system stiffness are discussed. It is
found that the vessels' motions are significantly affected by the hydrodynamic interactions between vessels and
sensitive to wave directions for shielding effects. The radiation forces of the adjacent vessel tend to amplify
vessel's motions and LNG carrier is more likely to be affected by FLNG's radiation forces for their difference in
displacement. In addition, compared with solid loading condition, liquid loading vessel tends to have decreased
natural roll frequency and have increased sway motions in the frequency region that higher than the natural
sloshing frequency. The two sloshing responses peaks appear in the natural roll frequencies and natural sloshing
frequencies, which are respectively mainly excited by vessels' roll and sway motions. Besides, in low fill con-
ditions, the sloshing loads contribute to larger sway motions in low fill conditions for the natural sloshing
frequencies are closer to the main response frequency region of vessels. Sloshing nonlinearity gets obvious in the
conditions with low fill conditions and large wave amplitude, while the motion responses of vessels have slight
nonlinearity with the increases of sloshing nonlinearity when no violent sloshing with wave break is excited.
Furthermore, small stiffness of the connection system has slight influences on the vessels' motions, and resonant
motions can be excited when the natural frequency of the connection system located in the wave frequency
region.

1. Introduction configuration in offloading operation. In the hydrodynamic perspective,

the side-by-side arrangement of a FLNG system involves interactions
The development of floating liquefied natural gas (FLNG) system between multiple bodies, sloshing in liquid tanks and resonance in the
that can produce, process and store liquefied natural gas (LNG) offshore narrow gap between vessels. Besides, the connection system adopted to
has been stimulated by the increasing demand of natural gas. One restrict the relative motions between vessels also plays a crucial role,
challenge for the operation of FLNG system is LNG offloading to an LNG and attentions should also be paid to the loads on the connection
carrier. Considering the efficiency and the limitation of flexible pipeline system.
in transferring LNG under extremely low temperature, side-by-side Hydrodynamic interactions among floating bodies in proximity exist
configuration is more economically attractive than tandem in many applications in ocean engineering and have received many

∗
Corresponding author.
E-mail address: [email protected] (Z. Hu).

https://doi.org/10.1016/j.oceaneng.2018.08.016
Received 13 August 2017; Received in revised form 7 May 2018; Accepted 9 August 2018
Available online 11 September 2018
0029-8018/ © 2018 Published by Elsevier Ltd.
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

researchers’ attentions. Kodan (1984) analyzed two parallel floating simulate coupled ship motion and nonlinear liquid sloshing in a tank;
structures using strip theory. Inoue and Islam (1999) calculated mo- experiments were conducted to validate the numerical program.
tions of multiple floating offshore structures with consideration of the Newman (2005) proposed that unless coupling ship motions and
effects of connectors and mooring lines in the frequency domain. sloshing loads in the time domain, the liquid tank could be regarded as
Buchner et al. (2001) developed a numerical model to predict the hy- an extension of the exterior wet surface. Thus, no iteration between
drodynamic responses of paralleled moored vessels in the time domain, internal sloshing loads and external wave loads was needed, but it
where hydrodynamic coefficients were prepared in the frequency do- would result in larger size of matrix equation. Mitra et al. (2012) stu-
main. Newman (2001) presented numerical and analytical results in the died coupling between three dimensional sloshing and ship motion in
research of multiple floating bodies, first and second order interaction different sea conditions, sloshing in tank was simulated based on po-
effects were considered. Choi and Hong (2002) and Hong et al. (2005) tential flow and good agreements with experimental results were
applied high order boundary element method (HOBEM) to simulate achieved. Research conducted by Zhao et al. (2015) proved potential
side-by-side vessels in the time domain. To research the coupling me- flow with an artificial damping model can give accurate and highly
chanism of two adjacent vessels, Koo and Kim (2005) compared three efficient simulation of sloshing in the time domain.
different numerical methods in combining motion equations. Zhu et al. Although multi-body floating system and liquid sloshing in vessels
(2008) developed a numerical program to simulate hydrodynamic in- have been widely investigated, these two parts were mostly considered
teractions of three-dimensional multiple floating structures in the time separately. The fully coupled analysis of liquid loading FLNG system in
domain based on potential flow theory, the governing equations were side-by-side configuration was rarely conducted. This study aims to
solved in time domain directly rather than using IRF. Pessoa et al. develop an accurate and efficient numerical code to predict the dy-
(2015) conduced numerical and experimental research on two floating namic responses of side-by-side arranged FLNG system. The coupling
structures connected with mooring lines and fenders. First order re- between sloshing in liquid tanks and vessels’ motions are solved in the
lative motions and loads on connection system were investigated. Zhao time domain, and hydrodynamic interactions between the vessels as
et al. (2012) carried out numerical simulation of FLNG system in par- well as motion responses under the restriction of connection system are
allel arrangement with time-domain analysis code SIMO. Xu et al. investigated. A series of experimental tests are conducted to validate
(2015) studied coupled motions of three barges in side-by-side ar- the numerical code. The research results reveal hydrodynamic inter-
rangement; a numerical program was developed and validated by test actions are sensitive to wave direction and small motions are excited
results. Based on relative motions between vessels, loads on connection when the vessel is located in the lee side. Sloshing loads will change the
system were obtained by assuming the connection system reacts quasi- natural roll frequencies of vessels and lead to increased sway motions in
statically (Hong et al., 2005; Zhao et al., 2014; Pessoa et al., 2015). the frequency region higher than that of natural sloshing frequency.
For floating vessels in proximity, resonance in the gap between
vessels is also a complicated phenomenon and is of practical im- 2. Mathematical formulation
portance. Molin (2001) derived analytical solution of resonance in
moon pools. Sun et al. (2010) studied the first and second order re- In this study, an FLNG system with parallelly arranged FLNG vessel
sonant properties between barges in parallel. The results of model tests and LNG carrier is considered, and each vessel has two liquid tanks. In
showed that resonant frequencies would shift depending on whether the numerical model, coordinate systems are defined as shown in Fig. 1.
the barges were fixed or floating. Zhao et al. (2017) experimentally The global coordinate system O − XYZ is earth-fixed with Z axis
investigated the resonant sloshing responses in the gap between two pointing upward. Vessel-fixed coordinate systems O1 − X1 Y1 Z1 and
fixed barges. In the calculation of hydrodynamic coefficients in the O2 − X2 Y2 Z2 are respectively located in the gravity center of FLNG and
frequency domain, potential flow theory tends to over-predict resonant LNG carrier, with OX and OY pointing to the bow and portside of the
responses in the gap and generates inaccurate hydrodynamic coeffi- vessels. The FLNG-fixed coordinate system vessel O1 − X1 Y1 Z1 coincides
cients for no viscosity was accounted. Huijsmans et al. (2001) applied a with the earth-fixed coordinate O − XYZ initially. The four tank-fixed
rigid lid in the gap to suppress the unrealistic prediction. Chen (2005) coordinate systems o − xyz are located in the free surface center in-
modified the free surface condition between vessels by introducing an itially and parallel with the vessel-fixed coordinate system.
artificial damping term to suppress resonant fluid in the gap. This
method had been used in sloshing problems to account for energy
2.1. Vessels in side-by-side configuration
dissipation (Faltinsen, 1978). Lu et al. (2011) and Yao and Dong (2015)
also applied a dissipating parameter to study the wave elevation in the
Vessel motions in waves are regarded as linear system that response
gap, and sensitivities to the parameter value were discussed. It was
linearly to wave excitation. In the time domain calculation, impulsive
found that the artificial damping term can suppress unrealistic resonant
response function method based on frequency-domain results is used in
sloshing effectively, but the selection of proper value still needs much
dealing with radiation forces, and solution of vessel motions in the
effort.
frequency domain is required.
Moreover, sloshing in the liquid tanks in FLNG system makes ves-
The linear incident wave potential can be written in a general form
sels' motion responses much different from solid loading condition. To
as:
date, many methods have been adopted in the predication of sloshing in
liquid loading vessel. Lee et al. (2007) and Nam et al. (2009) calculated ζa g
sloshing in a liquid tank through a Navier-Stokes solver, and coupled ϕI (x , y, z , t ) = e kz sin[k (x cos β + ysinβ ) − ωt − σ ]
ω (1)
the sloshing loads with the ship motion solved by linear potential flow.
Jiang et al. (2015) used the open source computational fluid dynamics where ζa , ω , k , β , σ are the wave amplitude, frequency, wave number,
(CFD) code OpenFOAM to investigate the couplings in liquid loading heading angle and phase shift, respectively. The potential in the flow
vessels in the time domain and found impact loads due to violent field can be regarded as the superposition of incident wave potential,
sloshing had less effect on ship's global motion responses. CFD methods radiation potentials and diffraction potential:
solving Navier-Stokes equations tend to have low efficiency in dealing ϕ = ϕI + ϕRiF + ϕRiL + ϕD (2)
with FLNG system with large liquid tanks and various fill conditions.
Malenica et al. (2003) considered dynamic coupling of liquid loading where ϕI is the incident wave potential, and ϕRiF ϕRiL
are respectively the
vessel in the frequency domain, and linear potential theory was adopted radiation potential corresponding to the 6° motions of FLNG and LNG
in the solving both sloshing and vessel's seakeeping properties. carrier; ϕD is the diffraction potential. In the flow field, the potential ϕ
Rognebakke and Faltinsen (2003) applied a multimodal method to satisfies the Laplace equation:

61
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 1. Definition of coordinate systems.

∇2 ϕ = 0 (3) On the wet surface of FLNG SF

On the free surface, the linearized free surface condition is ∂ϕRiF F ∂ϕRiL
= ξ˙i niF =0
∂n ∂n (9)
∂ϕ ω2
− ϕ=0
∂z g (4) On the wet surface of LNG carrier SL

To account for energy dissipation in the gap between vessels and ∂ϕRiF ∂ϕRiL L
=0 = ξ˙i niL
suppress unrealistic resonance in the gap, a damping-lid is applied in ∂n ∂n (10)
the gap zone (shown in Fig. 2). For free surface in the gap, a damping
The hydrodynamic coefficients as added mass μij , radiation damping
force proportional to particle velocity is added (Chen, 2005):
λij and wave forces in the frequency domain are computed using panel
ε program HydroStar developed by Bureau Veritas (2010). Wave dif-
f = − ∇ϕ
ω (5)
fraction forces and radiation forces in the time domain can be obtained
The free surface condition in the gap zone is modified as using IRF method. According to Newton's law, motion equation can be
expressed as:
∂ϕ ω2 ω2
− ϕ − iε ϕ = 0 F FF F
∂z g g (6) ⎡ Mij + μij (∞) μijFL (∞) ⎤ ⎡ ξ¨i (t ) ⎤
⎢ ⎥⎢ ⎥
where ε is the damping-lid coefficient and normally need to be de- ⎢ μijLF (∞) MijL + μijLL (∞) ⎥ ⎢ ξ¨ L (t ) ⎥
⎣ ⎦⎣ i ⎦
termined by experimental calibration. F
⎡ kij (t − τ ) kij (t − τ ) ⎤ ⎡ ξ˙i (τ ) ⎤
F FL F FL
On the seabed: t ⎡ Cij Cij ⎤ ⎡ ξ F (t ) ⎤
+ ∫0 ⎢ LF ⎥ ⎢
k (t − τ ) kijL (t − τ ) ⎢ ξ˙ L (τ ) ⎥
⎥ d τ + ⎢ ⎥ ξ L (t ) ⎥
C LF CijL ⎢
∂ϕ ⎣ ij ⎦⎣ i ⎦ ⎣ ij ⎦⎣ ⎦
=0
∂n (7)
FiF (t ) ⎤
=⎡
⎢ FL (t ) ⎥, i, j = 1,2, ⋅⋅⋅6
On the vessels’ wet surface SF and SL , the diffraction potential and (11)
⎣ i ⎦
incident wave potential have the relationship:
F = Fwave (t ) + Fsloshing (t ) + Fconnect (t ) (12)
∂ϕD ∂ϕ
=− I
∂n ∂n (8) where Mij , μij (∞) , kij and Cij are the total mass of vessels (including
liquid mass), infinite added mass, retardation function, and hydrostatic
The coupling radiation can be decomposed as two cases: FLNG
restoring coefficients. Fsloshing (t ) and Fconnect (t ) are forces on vessels
moves with LNG carrier fixed and LNG carrier moves with FLNG fixed.
caused by liquid sloshing in tanks and connection system that including
Then, the boundary conditions of radiation potential on vessels’ wet
hawsers and fenders.
surfaces can be given as:
Retardation function kij stands for memory effects of past motions
and can be calculated based on radiation damping coefficients:
2 ∞
kij (t ) =
π
∫0 λij (ω)cos(ωt ) dω
(13)
Where λij is radiation damping in frequency domain. For the interaction
between vessels, the memory effects can last for a much longer time
than single vessel. 300 frequencies are adopted in the integral with
truncation frequency equals 3.0 rad/s. Added mass in infinite frequency
and retardation function has the relationship as:
1 ∞
μij (∞) = μij (ω) +
ω
∫0 kij (t )sin(ωt ) dt
(14)
Viscous damping is non-negligible for vessel's roll motion, in this
study, a commonly used equivalent linear damping coefficient is in-
Fig. 2. Model of vessels in the calculation of frequency coefficients. troduced as:

62
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

according to Bernoulli equation:

P ∂φ du dv dw 1
− = − ω × r⋅(∇φ + V) + x +y +ζ + ∇φ
ρ ∂t dt dt dt 2
⋅∇φ + k θgr0 (21)

φa = ∂φ / ∂t is defined in this study and the solution of φa can be regarded

as a similar boundary value problem as the solution of φ . In tank-fixed
coordinate system, the governing equations of φa can be obtained as:
In the fluid domain
∇2 φa = 0 (22)

On the free surface

Fig. 3. Schematic of numerical tank model.
du dv dw 1
φa = −x −y −ζ − ∇φ⋅∇φ − k θgr0 + (ω × r)⋅(∇φ + V)
v dt dt dt 2
b44 = 2γ (M44 + μ44 (∞)) C44 (15)
(23)
Where the γ is the ratio of viscous damping to critical damping coef- On wall boundaries
ficient. Soft numerical springs are used to restrict vessel's surge, sway
and yaw motions so that monotonously increasing or decreasing mo- ∂φt
= n⋅[ω˙ × r + 2ω × Vr + ω × (ω × r) − (∇φ⋅∇) ∇φ] + n
tions can be prevented. The natural frequency of soft spring can be ∂n
calculated as Ciispring /(Mii + μii (∞)) , which are smaller than first mode ⋅∇ [(ω × r)⋅(∇φ + V)] (24)
response frequencies of vessel motions and sloshing.
where Vr is the velocity of fluid observed in tank-fixed coordinate
system.
2.2. Liquid sloshing
After solving φa , sloshing loads on vessels can be obtained. The time-
derivative of potential on the free surface is obtained as:
Internal sloshing problem in liquid tanks is solved based on poten-
tial theory by assuming that the fluid is perfect and flow irrotational. In dφ ∂ζ ∂φ du dv dw 1
= −x −y −ζ − ∇φ⋅∇φ − k θgr0 + (ω × r)
calculation, numerical model of rectangular liquid tank with length B , dt ∂t ∂y dt dt dt 2
width W and water depth D is built, as shown in Fig. 3, where tank 1 is
⋅(∇φ + V) (25)
taken for example.
For the convenience of coupling with vessel motion, sloshing pro- For the convenience of mesh updating during calculation, free sur-
blem is solved in tank-fixed coordinate. Velocity potential ϕ is de- face is updated vertically in the tank-fixed coordinate:
composed as ϕ = φ + xu + yv + zw . Boundary value problem of ve-
dζ ∂φ ∂ζ ∂φ ∂ζ ∂φ
locity potential φ in tank-fixed coordinate system can be derived as: = − − + (ω × r)⋅∇ζ
In the fluid domain dt ∂y ∂x ∂x ∂y ∂y (26)

∇2 φ = 0 (16) Artificial damping model is used to account for energy dissipation

caused by fluid viscosity. A damping force proportional to particle ve-
On wall boundaries locity with an opposite direction is assumed on the free surface. With
∂φ damping coefficient μ , Euler's equation can be expressed as:
= (ω × r)⋅n
∂n (17) Dw 1
= − ∇p − g∇ζ − μ∇φ
where [u, v, w] and ω are transitional velocity and rotational velocity of Dt ρ (27)
excitation on tank corresponding to the vessel's six degree motions. n is The sloshing model is solved numerically in BEM, details about
outward vector and normal to tank walls, r is the distance from rotation numerical simulation can be found in Zhao et al. (2018).
center to boundary.
On the free surface SF , dynamic and kinematic conditions are
2.3. Connection system
∂φ du dv dw 1
− ω × r⋅(∇φ + V) + x +y +ζ + ∇φ⋅∇φ + kθ gr0 = 0
∂t dt dt dt 2 The relative motions between vessels are restricted by connection
(18) system to guarante the safety of offloading arm and avoid collision
between vessels during the offloading operation. Loads on connection
∂ζ ∂φ ∂ζ ∂φ ∂ζ ∂φ
+ ⋅ + ⋅ − − ω × r0⋅∇ζ = 0 system Fconnect including Fhawser and Ffender are related to the relative
∂t ∂x ∂x ∂y ∂y ∂z (19)
motions between vessels. Hawsers and fenders are numerically simu-
cos β cos γ − cos β sin γ sin β lated as linear stiffness systems. A sketch of hawsrs and fenders are
⎡ ⎤ presented in Fig. 4 to illustrate the solution of Fconnect .
⎢ sin α sin β cos γ − sin α sin β sin γ − sin α cos β ⎥
⎢ ⎥ The hawser can only be stretched and tension on hawser can be
kθ = ⎢ + cos α sin γ + cos α cos γ ⎥ given as:
⎢ ⎥
⎢ − cos α sin β cos γ cos α sin β sin γ cos α cos β ⎥ Fhawser = khawser⋅δlhawser ⋅nhawser δlhawser > 0
⎢ ⎥
⎣ + sin α sin γ + sin α cos γ ⎦ (20) Fhawser = 0 δlhawser < 0 (28)

where ∂ , β and γ are the roll, pitch and yaw motions angles of the where khawser and δlhawser are the stiffness coefficient and elongation of
vessel, g = [0 0 g ], r0 = [0 0 ζ ], ζ is the free surface elevation de- hawser, nhawser stands for the hawser direction that from FLNG to LNG
fined in the tank-fixed coordinate system. carrier. For each hawser, pF (x , y, z ) and pL (x , y, z ) are points where the
Initial conditions in free surface can be expressed as hawser is attached to FLNG vessel and LNG carrier. The distance be-
φ = −xu − yv − zw and η (x , y, 0) = 0 . After the solution of velocity tween pF and pL is the corresponding hawser length lhawser . The motion
potential, hydrodynamic pressure in wall boundaries can be achieved of pF and pL in global coordinate system O − XYZ can be described as

63
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

motion of FLNG and LNG carrier, respectively.

The fender can only be compressed and reaction force can be given
as

Ffender = kfender⋅δlfender ⋅n δlfender ≤ 0 (38)

where δlfender is the deformation of fender under compression.

Unlike hawsers that attached to the two vessels, the fenders are
floating in the free surface between vessels. In this study, fender is
simplified to be fixed to FLNG vessel on the free surface position PF′ and
perpendicular to the FLNG hull surface. Thus, position p′L where fender
reaction force exerts on LNG vessel varies when relative motions be-
tween vessels occur. Assuming normal direction of fender in initial
condition nfender , the fender length κ and coordinates of p′L stasify the
following relationship:

ξiF + kθF r Ffender + κkθF nfender = ξiL + kθL r Lfender + rrel i = 1,2,3 (39)

on which, p′L (y ) is known for p′L located in the starboard of LNG

carrier. p′L (x ) , p′L (z ) and fender length κ can be solved according to
this linear equation set.
After the solution of fender length, fender force can be obtained
according to (38). Moment exerted on FLNG and LNG carrier by fender
can be given as:
−1
MFfender = r Ffender × (kθF ) FFfender (40)
Fig. 4. Schematic of connection system.
−1
MLfender = r Lfender × (kθL) FLfender (41)
pF = ξiF + kθF r hawser
F
i = 1,2,3 (29)

pL = ξiL + kθL r hawser

L
i = 1,2,3 (30)
2.4. Coupling solution
where kθL
and kθL
are related to vessels' rotational motions, and have the
same form as kθ in equation (20). The superscript ‘F’ and ‘L’ stand for The complex coupling effects in FLNG system make the direct so-
FLNG and LNG carrier, respectively. The elongation and normal di- lution of (11) in the time domain rather complicated. In the present
rection of hawser can be obtained as: study, an iteration strategy is adopted to couple the responses of liquid
sloshing and connection system with vessels’ motions.
δlhawser = ⎯p⎯⎯⎯⎯⎯
→ −l
F pL initial (31) In the iteration process, as shown in Fig. 5, vessel motions serve as
⎯p⎯⎯⎯⎯⎯p→ the excitation on the liquid tank, and the relative motions between
F L vessels determine the loads on connection system. The sloshing and
n = ⎯⎯⎯⎯⎯⎯
pF p→ L (32) connection system loads work as the excitation forces on vessels' mo-
tions. The wave action part is solved using IRF method that described in
where linitial is the length of hawser at the initial condition in still water
Section 2.1 and is not affected by the time domain calculation. The
and can be solved as:
solution of liquid sloshing and connection system parts are described in
L F
linitial = r hawser + rrel − r hawser − lpretension (33) Section 2.2 and 2.3, respectively. In the time domain calculation,
F L iteration calculation between vessels’ motions and loads from sloshing
where r hawserand are the coordinates of pF and pL that described
r hawser
and connection system is conducted in each step until convergence is
in vessel-fixed coordinate systems O1 − X1 Y1 Z1 and O2 − X2 Y2 Z2 ,
obtained, which means the balance between forces and motions in (11)
rrel = O1 O2 is the distance between origin of O1 − X1 Y1 Z1 and
is achieved. After the iteration calculation, both vessel motions and
O2 − X2 Y2 Z2 that described in global coordinate system O − XYZ .
internal sloshing are updated, then calculation proceeds the next time
lpretension is the elongation of hawser that caused by the pretension of
step.
hawser, and can be given as
It should be noted Fsloshing (t ) is solved in tank-fixed coordinate
Fpretension system and needs to be transferred to the coordinate system of vessel
lpretension =
khawser (34) motions. In addition, inertia forces of liquid have been considered in the
first term in left side of equation (11) and need to be excluded from
Then, forces applied on FLNG and LNG carrier by hawser can be
Fsloshing (t ) in the coupling calculation.
given as
F L
Fhawser = −Fhawser = Fhawser (35)
3. Model tests
The hawser forces need to be transferred to vessel-fixed coordinate
system before the calculaiton of the moment applied on vessels. Thus, A model test was conducted to validate the proposed numerical
the moment exerted on FLNG and LNG carrier can be given as code. The model test was conducted in Deepwater Offshore Basin at
F F −1 Shanghai Jiao Tong University at a model scale of 1:60. The basin is
Mhawser = r hawser × (kθF ) Fhawser
F
(36) 50 m long, 40 m wide and 10 m in depth. Wave makers are installed in
L L −1 the two adjacent sides of the basin and oblique waves can be produced.
Mhawser = r hawser L
× (kθL) Fhawser (37)
The other two adjacent sides are installed with wave absorber equip-
where kθF and kθL are in the form of (20), and correspond to the rotation ment to prevent the wave reflection.

64
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 5. Flow chart of the numerical code in the simulation of FLNG system.

3.1. Vessels model 3.3. Sea states

Arrangement of FLNG system in side-by-side configuration is pre- White noise waves are adopted in the model tests to obtain FLNG
sented in Fig. 6. The FLNG and LNG carrier are placed in parallel with system responses in a wide frequency range. The white noise waves
initial gap 6 m. The two vessels have similar profiles but different size. used have significant wave amplitude of 1.5 m to meet the linear wave
For each vessel, two plexiglass rectangular tanks are installed symme- assumption. Wave period ranges from 5 s to 30 s to cover the main
trically about the mid-ship section. The main particulars of vessels and response periods of vessels’ motion and liquid sloshing in tanks.
tanks are listed in Table 1, FLNG and LNG carrier are respectively Different wave directions (defined in Fig. 7) including beam waves (90
66.7% and 60% filled. Solid loading tests are also included in the ex- and 270°), oblique waves (135 and 225°) and heading wave (180°) are
periment for comparison. The solid and liquid loading conditions have considered in the model tests. White noise waves are calibrated before
same inertia particulars when assume the liquid in tanks is frozen. the model test by measuring wave elevation histories in the center of
Wave probes are installed in the liquid tanks close to the bow (T1 to basin using wave probe. After FFT calculation, the measured wave
T10 in Fig. 6 (a)) to capture sloshing wave elevation in representative spectra are obtained and comparison with target wave spectra is pre-
positions. Wave probes T11 to T13 are fixed to vessels to measure wave sented in Fig. 8.
elevation around vessels.
In the model testes, vessels are restricted with soft horizontal 4. Comparison and validation
mooring lines (shown in Fig. 7) to prevent motions caused by drift
forces. The soft stiffness of the horizontal mooring lines make sure the The developed numerical code is validated in this section. The nu-
natural frequency of mooring system is very low and has slight effects merical code in predicting liquid sloshing under forced excitation,
on vessels’ first order motions. The horizontal mooring forces are not coupling between sloshing and ship motions, interactions between two
considered in the numerical model described in section 2 for only first vessels are separately validated by previous studies and experimental
order wave forces are adopted. results. All the numerical simulations are conducted in the time do-
main; some results are normalized by excitation amplitude or wave
amplitude and presented in the form of response amplitude operators
3.2. Connection system (RAO).

A simplified connection system that consists of four hawsers and

4.1. Liquid sloshing
two fenders are adopted in this study. Both the hawsers and the fender
are simulated as linear stiffness system. Though the nonlinearity of
In this subsection, numerical predictions of sloshing under forced
fender will become significant under large deformation, when the
excitations are validated. Natural sloshing frequencies of three dimen-
compression on fender is limited, the linearized fender mode has been
sional rectangular tank described in Fig. 3 can be expressed analytically
proved to be feasible (Zhao et al. (2017)) for the less violent motions of
as:
FLNG system. The connection system is attached to the port of FLNG
and starboard of LNG carrier, respectively. Hawsers are located in the
m2 n2 ⎞ ⎛ m2 n2 ⎞ ⎞
upper deck and fenders are located in the water line position and fixed ωmn = π ⎛ 2 +
⎜
2
tanh ⎜π ⎛ 2 +
⎟ ⎜
2 ⎟
⎟

⎝ B W ⎠ B W
to FLNG vessel. Coordinates and stiffness of hawsers and fenders are ⎝ ⎝ ⎠⎠ (42)
listed in Table 2. In the initial condition, the hawsers are stretched with
where m , n are positive integers. For sloshing cases with different tanks
a pretension of 350 kN, this pretension is balanced by the reaction
size and excitation, convergences analyses are conducted to guarantee
forces of fenders and static restore forces of vessels.
the mesh size and time step convergence. Firstly, sloshing responses in
three dimensional liquid tank are compared with calculation results of
Wu et al. (1998) in FEM method. Parameters of tank are B = W = 4 ,

65
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 6. Profiles of vessels in side-by-side arranged FLNG system (unit: mm).

Table 1 forced excitations is validated. A two dimensional tank subjected to

Main particulars of vessels and LNG tanks. rotational excitation solved by Nakayama and Washizu (1981) is
Main particulars
chosen. The tank has parameters as B = 0.9m , D = 0.6m . Rotational
excitation on the tank is applied in cosine form as Aθx (t ) = θx cos(ωθx t ) ,
FLNG LNG carrier where θx = 0.8deg and ωθx = 5.5rad/ s . In the proposed numerical pro-
gram, a numerical tank with B × W × D = 0.9m × 0.3m × 0.6m is built
Length (m) 213.9 171.2
Width (m) 44.8 35.8
and mesh numbers are set as 18 × 6 × 12 . Wave elevation histories in
Depth (m) 25.5 20.4 the corner (B /2, −D /2) are presented in Fig. 10 and good agreement is
Draft (m) 10.8 9 obtained. Moreover, sloshing loads on tank wall due to rotational ex-
Displacement (t) 98923.1 52821.3 citation are also validated. Experimental and numerical studies con-
Roll gravity radius (m) 16 10.2
ducted by Chen et al. (2013) are chosen. Dimensions of the liquid tank
Pitch gravity radius (m) 60 50
Yaw gravity radius (m) 60 50 are B × W × D = 1m × 1m × 0.3m and mesh numbers are set as
Vertical gravity center (m) 13.8 12 15 × 15 × 5 in the present simulation. Rotational excitation on the tank
Tank length (m) 36 29.44 is in sinusoidal form as Aθx (t ) = θx sin(ωθx t ) with θx = 5deg . Two cases
Tank width (m) 36 29.44 with ωθx = 0.95rad/ s and ωθx = 3.09rad/ s are selected. Pressure load
Tank height (m) 27 24
histories on the left wall that 0.1m form initial free surface are captured.
Fig. 11 shows that present numerical results in ωθx = 0.95rad/ s agree
displacements of tank in x and y directions are in sinusoidal form as well with experimental and numerical results obtained by Chen et al.
d x (t ) = Ax sin(ωx t ) and d y (t ) = Ay sin(ωy t ) with Ax = 0.0372 , ωx = ω10 , (2013). When excitation increases to ωθx = 3.09rad/ s , larger sloshing is
Ay = 0.0182 , ωy = ω01. Mesh numbers along tank's width, length and induced and pressure discrepancy appears for the free surface of liquid
water depth are set as 14 × 14 × 8, and time step is set as 0.01s. Wave are not so regular as in mild sloshing case. The general tendency and
elevation histories in two corners are presented in Fig. 9 and good amplitude of sloshing pressure are well simulated, which proves the
agreements are obtained. validation of the proposed numerical program.
Secondly, free surface elevation in sloshing under rotationally It should be noted that the applicability of proposed numerical

66
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 7. Mooring configuration side-by-side arranged vessels.

Table 2 conducted using experimental results. Test results of solid and liquid
Main particulars of FLNG vessel and LNG tank. loading conditions serve as the validation of vessel motion calculation
Location in FLNG (FLNG- Location in LNG carrier (LNG Stiffness
and coupling calculation, respectively.
fixed coordinate system) carrier-fixed coordinate Firstly, experimental tests conducted by Rognebakke and Faltinsen
system) (2003) are selected for comparison. In Rognebakke's experiment, a
rectangular hull contains two liquid tanks is restricted to sway motion
x (m) y (m) z (m) x (m) y (m) z (m) K (kN/m)
in waves. Responses of solid loading condition and liquid loading
Hawser 1 90.0 22.4 14.7 −74.0 −17.9 11.4 4689.0 condition with two tanks 0.186 m filled are selected in the comparison
Hawser 2 15.0 22.4 14.7 −50.0 −17.9 11.4 2690.0 with numerical results. Mesh numbers along the length, width and
Hawser 3 −15.0 22.4 14.7 50.0 −17.9 11.4 2690.0 water depth of each liquid tank are set as 8 × 18 × 10 . Fig. 12 shows the
Hawser 4 −90.0 22.4 14.7 74.0 −17.9 11.4 4689.0
comparison of sway RAOs, and good agreements are obtained. Then,
Fender 1 46.0 22.4 −3.0 46.0 −17.9 −1.8 885.6
Fender 2 −46.0 22.4 −3.0 −46.0 −17.9 −1.8 885.6 experimental results of the solid and liquid loading FLNG vessel under
beam sea condition are used. Comparison of vessel's motions and
sloshing wave elevation in tank are presented in Fig. 13. μ = 0.1 is
selected to account for energy dissipation in sloshing, and good
agreements in vessel's motions and sloshing responses are obtained. The
good agreements between numerical and experimental results prove
that the proposed numerical code can give good prediction of the
coupling in liquid loading vessels. It can be pointed out that sway
motions of liquid loading decrease to almost 0 in the natural sloshing
frequencies. The two sloshing peaks appear near natural roll frequency
and natural sloshing frequency, it can be deduced the sway and roll
motions are respectively the main excitation of sloshing peaks.

4.3. Resonant responses in the gap zone

Damping coefficient ε on the free surface of the gap zone is crucial

to hydrodynamic coefficients of side-by-side arranged FLNG system. In
this study, wave elevation in the gap that sensitive to damping value is
selected to validate the value of ε .
Experimental results of wave elevation in T12 (shown in Fig. 6(a))
Fig. 8. Spectra of white noise wave in beam sea condition. are used to validate selection of ε . As wave probe T12 is fixed to the
larboard of FLNG vessel, wave elevation caused by vessel's motions
program is limited to sloshing with single valued free surface profile. need to be excluded based on the wave probe locations. Different values
Violent sloshing that includes wave breaking and overturning are not of ε are applied in the calculations in frequency domain, comparisons of
considered in the present study. Thus, extreme shallow loading condi- wave elevation in the middle of the gap with experimental results are
tions that tank bottom may be out of water are also beyond the cap- presented in Fig. 14. Three wave directions are selected to check the
ability of the present model. sensitivity of damping value to wave directions. Response peaks that
correspond to the natural frequencies of gap resonances can be ob-
served, and the damping lid model can suppress the resonant wave
4.2. Coupling of liquid loading vessel elevation in the gap successfully. The damping lid has slight effects on
responses in the low frequency range that away the resonant frequency.
Validation of coupling calculaiton of liquid loading single vessel is Though the RAOs of wave elevation differ obviously under different

67
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

hydrodynamic coefficients of vessels in frequency domain, ε = 0.05 is

selected.
The resonant frequencies in the gap obtained from the numerical
and experimental results have good agreements. Molin, 2001; Molin
et al., 2002) gave the general solution of resonant frequencies in open-
ended moonpool. For moonpool has the size of length L, width B and
draft D, resonant frequencies can be calculated as:

1 + Jmn tanh(νmn H )
ωmn = gνmn
Jmn + tanh(νmn H ) (43)
2
where νmn = λm2 + μn2 , λm = mπ / L , μm = nπ / B . m and n stand for order
of resonant mode in longitudinal and lateral direction, respectively. In
the narrow gap with B / L is rather small, response along the width of the
gap is in the form of piston type. With the increase of B response
translate into types with different modes. For FLNG system, the gap is
rather small compared with length of vessels. Resonant frequencies in
that case correspond to n = 0 in (43) and Jm0 can be written as (Molin
et al., 2002):

2 ⎧ 1 r2
Fig. 9. 3D sloshing wave elevation histories.
Jm0 =
mπ 2r ⎨
∫0 u2
⎡1 + 2u + (u − 1)cos(mπu)
u2 + r 2 ⎣
⎩
3 1 1 + cos θ ⎫
− sin(mπu) ⎤ du − + 1 + 2r ln
mπ ⎦ sin θ 1 − cos θ ⎬
⎭ (44)

where r = B / L , tan θ = 1/ r .
In Table 3, gap resonant frequencies under four wave directions are
presented. Analytical results are also obtained for comparison, in which
B = 6m and L = 160m is set as the length of LNG carrier. Considering
Neumann condition defined in the bottom of gap Molin (2001), it is
more reasonable to define D = 9m according to the draft of LNG carrier.
Table 3 shows the shift of resonant frequencies exists between the
analytical and experimental results. Gap resonance between floating
bodies tends to have larger frequencies for the radiation effects of
vessels, and this tendency can also been found in model tests conducted
by Sun et al. (2010). Besides, the difference size between the two ves-
sels will cause complex diffraction effects, which also contribute to the
differences between the discrepancies in Table 3.

Fig. 10. Wave elevation histories due to rotational excitation.

4.4. Vessels’ motions in side-by-side configuration

Using hydrodynamic coefficients with considering damping lid in

the gap zone, vessels' motions under different wave directions are va-
lidated. Experimental results of solid loading FLNG system under ob-
lique (135°) and head waves are selected to validate the numerical
code. RAOs of vessels' motions are presented in Fig. 15 and Fig. 16,
good agreements between experimental and numerical results are ob-
tained, which indicates the frequency coefficients calculated can
guarantee good results in time domain. The frequency region smaller
than 0.2 rad/s is not considered as horizontal mooring system plays
dominant role in this region, especially for vessels' sway, swan and yaw
motions. The vessels' motions are considered in linear range, and the
viscous model in roll mode can give good prediction of the vessels’ roll
motion responses in the natural roll frequencies for the nonlinear vis-
cous is not obvious.

5. Discussion

In this section, dynamic responses of the FLNG system are in-

vestigated based on numerical and experimental results. Firstly, hy-
Fig. 11. Pressure histories due to rotational excitation. drodynamic interactions between vessels under different wave direc-
tions, sloshing effects on vessels’ motions and loads on connection
wave directions, good agreements between numerical and experimental system are analyzed without considering the connection system for the
results can be obtained when ε = 0.05. In the calculation of convergence of analysis. Then, responses of connection system in-
cluding sloshing effects are studied.

68
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 12. RAOs of sway motion in solid and liquid loading conditions.

5.1. Hydrodynamic interaction between vessels retardation function of single vessel case decays much faster than that
in side-by-side configuration. As retardation function will decay to
For side-by-side arranged FLNG vessel and LNG carrier, the ex- zeros gradually, convolution calculation in equation (11) form the be-
istence of the adjacent vessel can change the boundary condition in the ginning (t = 0) is not always necessary. But the calculation should in-
fluid field, which will affect the hydrodynamic coefficients of vessels clude relatively longer time in side-by-side conditions than in single
and wave forces on vessels. In the single vessel case, the radiation wave vessel conditions.
can transmit to infinity and disappear; in the side-by-side arranged Hydrodynamic interactions between vessels are sensitive to the
vessels, the radiation wave of the moving vessel will be reflected by the wave direction. Fig. 18 shows the motion responses of vessels under
adjacent vessel and radiation energy dissipates slowly. Roll motion two beam sea conditions without consideration of connection system.
retardation function are compared for FLNG vessel in single vessel FLNG vessel locates in the weather side and lee side in 90° and 270°
condition and side-by-side configuration. As shown in Fig. 17, headings, respectively. It can be seen that vessel motions tend to have

Fig. 13. RAOs of vessel motions and sloshing elevation in T1 under beam sea.

69
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 14. Comparison of wave elevation in the gap center under different wave directions.

Table 3 larger responses when the vessel is located in the weather side. For
Resonant frequencies in gap under different wave directions. sway and heave motions, sheltering effects mainly lies in the high fre-
Mode ω10 ω30 ω50 quency region; for roll motion, a slight response amplitude decrease
appears in the natural roll frequency when vessels are located in lee
Ana (D = 9m) 0.840 0.979 1.122 side. In addition, no obvious responses are induced near the frequency
of gap resonance, which indicates the resonance in the gap has less
Exp(135deg) 0.910 1.072 1.191
Exp(180deg) 0.896 1.064 1.198
effect on vessels’ motions.
Exp(225deg) 0.905 1.033 1.189 The effects of radiation wave caused by the motions of the adjacent
vessel are investigated. RAOs of the vessels' motion under beam sea
(90°) are plotted in Fig. 19. The differences between the two cases with
adjacent vessel free floating and fixed reflect the radiation wave effects.

Fig. 15. RAOs of vessels' motions under quarter wave (135°).

70
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 16. RAOs of vessels' motions under heading wave.

FLNG is located in the weather side and wave excitation on LNG carrier
decreases for the shielding effects, and the free floating FLNG can excite
much larger roll motion of LNG carrier. The RAOs of FLNG's roll mo-
tions show the radiation wave produced by the roll motion of LNG
carrier contributes less to the roll motion of FLNG. The reason could be
that the roll motion of LNG carrier is decreased for the shielding effects,
besides, the differences in displacement also makes LNG carrier more
sensitive to the motion of adjacent vessel than FLNG.

5.2. Sloshing effects

Sloshing effects on the responses of FLNG system are investigated in

Fig. 17. Retardation function of FLNG in roll mode. this section. Damping coefficient μ = 0.1 is adopted in the artificial
damping model that described in 2.2. Vessels’ motion and sloshing
It can be seen that the RAOs of vessel's motions are much smaller when loads on vessels are discussed. Sloshing effects in different filling levels
the adjacent vessel is fixed, especially in the sway and have modes, are compared and nonlinear effects are also considered.
which indicate the radiation caused by the vessel's motions can amplify
the adjacent vessel's motion responses significantly. In the roll mode, 5.2.1. Liquid loading effects
when FLNG is fixed, the roll motion of LNG carrier is much small as the Comparisons of motion responses in solid and liquid loading con-
ditions under beam sea are conducted. The FLNG and LNG carrier are

Fig. 18. Numerical RAO results of vessels' motions under beam sea conditions.

71
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 19. Numerical RAO results of vessels' motions without radiation potential under beam sea condition (90°).

Fig. 20. Numerical RAO results of vessels' motions in solid and liquid loadings under beam sea condition (90 deg).

66.7% and 60% filled, respectively; the corresponding first mode nat- 0.58 rad/s in the solid loading conditions to 0.42 rad/s and 0.52 rad/s
ural sloshing frequencies of liquid tanks in FLNG and LNG carrier are in the liquid loading conditions, respectively. Fig. 21 shows the RAOs of
0.886 rad/s and 1.018 rad/s, respectively. Sway, heave and roll motions sloshing wave elevations and sloshing loads in the sway and roll modes.
of vessels are presented in Fig. 20. It is indicated that numerical results Two peaks corresponding to natural roll motion frequency and natural
have good agreements with experimental results in both solid and li- sloshing frequency can be observed from sloshing elevations. It can be
quid loading conditions. The sway motions of vessels in liquid loading pointed out that the two peaks of sloshing elevation exert larger
condition have small response amplitudes in the natural sloshing fre- sloshing loads in roll mode and sway mode, respectively. As revealed in
quency and larger responses in the frequency region higher than the Section 4.2, the two sloshing peaks around the natural roll motion
natural sloshing frequency. The heave motions of vessels are less af- frequencies and natural sloshing frequencies are mainly excited by the
fected by liquid sloshing. The roll motions of vessels in liquid loading vessels’ roll motion and sway motion, respectively.
conditions have smaller natural frequencies and response amplitudes Responses of liquid loading vessels under 270° beam sea condition
than those in solid loading conditions. The natural roll motion fre- are presented in Fig. 22 for comparison. As solid loading cases shown in
quencies of FLNG and LNG carrier decrease from 0.48 rad/s and Fig. 18, for the shielding effects, the vessel tends to have larger heave

72
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 21. Numerical RAO results of sloshing elevations and sloshing loads under beam sea condition (90 deg).

and sway motion responses when it is located in the weather side. But Phase shifts of sloshing loads to wave forces are presented in Fig. 23 to
this tendency cannot be found from the roll motions of vessels in liquid investigate the sloshing contributions to the motions of vessels. The
loading condition. The RAOs of roll motion of the vessels have slight phase shifts in sway and roll modes have similar distributions except in
differences with the change of wave direction, which indicates the the frequency regions lower than the natural roll motion frequencies.
sloshing effects on vessels’ roll motion are much significant than the The phase shifts are also affected by the wave directions. According to
interaction effects between vessels under different wave directions. the phase shifts in roll mode, for the vessels located in the weather side,

Fig. 22. Numerical RAO results of vessels' motions in liquid loading under beam sea conditions.

73
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 23. Numerical phase shifts between wave forces and sloshing loads for vessels in liquid loading condition.

the sloshing loads contribute to roll motions positively in the frequency The RAOs of the vessels' motions in different filling levels are shown
region smaller than the natural roll frequency and negatively in the in Fig. 24. The sway motions tend to have larger responses in the low
frequency region between natural roll frequencies and the natural filling levels, especially for LNG carrier. The reason is the low filling
sloshing frequencies. The phase shifts change from positive to negative level tanks have smaller natural sloshing frequencies and larger
in the natural sloshing frequencies. For the vessels in the lee side, the sloshing can be excited when the natural sloshing frequencies are close
phase shifts in the frequency regions between the natural roll fre- to the main response frequency regions of vessels' motions. As shown in
quencies and the natural sloshing frequencies are smaller than those of Fig. 25, low filling level cases have larger sloshing responses in the
vessels located in the weather side, and the phase shifts change from frequency region between the natural roll frequency and natural
positive to negative in the natural roll motion frequencies. Moreover, in sloshing frequency. The natural frequencies of FLNG's roll motion vary
the frequency regions higher than natural sloshing frequencies, phase with the change of filling levels. Compared with solid loading condi-
shifts are much smaller except for LNG carrier located in lee side. As tion, the natural roll motion frequency of FLNG decreases to 0.42 rad/s
significant sloshing loads in sway mode are excited in the frequency in 66.7% filling level and increases to 0.52 rad/s in 33.3% filling level;
regions higher than the natural sloshing frequencies, the RAO peaks of the natural roll motion frequency of LNG carrier decreases from
sway motion in these regions can be explained. 0.58 rad/s in solid loading condition to 0.52 rad/s in 60% filling level
and increases to 0.61 rad/s in 24% filling level. It can be seen that
natural roll frequencies tend to be larger than solid loading conditions
5.2.2. Filling level effects in low fill condition and smaller than solid loading conditions in high
Responses of the FLNG system in different filling levels are discussed filling level. Fig. 25 shows sloshing loads in roll mode are much small in
here. Four fill cases are selected, as shown in Table 4. Each vessel has the low fill conditions, because sloshing loads in roll mode are mainly
two filling levels. First mode sloshing frequencies in liquid tanks are caused by static pressure related to the liquid mass rather than dynamic
presented.

Table 4
Natural sloshing frequencies in liquid tanks under different filling levels.
Fill case A B C D

FLNG LNG carrier FLNG LNG carrier FLNG LNG carrier FLNG LNG carrier

Filling level 66.7% 60% 66.7% 25% 33.3% 60% 33.3% 25%
Natural frequency (rad/s) 0.886 1.018 0.886 0.814 0.749 1.018 0.749 0.814

74
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 24. Numerical RAO results of vessels' motions in different filling levels.

Fig. 25. Numerical RAO results of sloshing elevations and loads on vessels in different filling levels.

75
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

pressure as in sway mode. profiles in the LNG carrier liquid tank at the maximum wave elevation.
The sloshing in liquid tanks can also affect the hydrodynamic in- The free surface is much flat in the case 0.61 rad/s as the sloshing is
teraction between vessels. The motion and sloshing responses of FLNG caused by the inclination of tank related to the LNG carrier's roll mo-
in same filling level have slight change with the variance of the filling tion. In the case 0.84 rad/s, the free surface demonstrates first order
level of LNG carrier, while the variance of filling level of FLNG can mode and obvious nonlinearity can be observed. But the sloshing
cause different responses of LNG carrier in the same filling levels. As nonlinearity does not change the sloshing loads exerted on vessels
shown in Fig. 24, with different filling levels of the FLNG, LNG carrier much, as shown in Fig. 28. The reason is that sloshing loads on tanks are
has different sway motions in fill Cases B and D and has different roll decided by the differences of wave elevation in the two sides of tank,
motions in Case A and Case C, which correspond to the conclusion the nonlinear sloshing with higher crest and lower trough contributes
drawn above that sloshing has more effects on the sway motion in the less to the total loads except in the sloshing cases with strong non-
low filling level conditions and has more effects on the roll motion in linearities. Thus, the RAOs of liquid loading vessels' motions shown in
the high filling level conditions for the large liquid mass. As vessels in Fig. 26 do not demonstrate nonlinear properties under increased wave
different filling levels have the same inertia when liquid in tanks is amplitude. When the wave amplitude continues to increase, the
frozen, this kind of interaction can be regarded as the change of ra- sloshing will become much violent and wave breaking will occur, which
diation effects caused by liquid sloshing on the adjacent vessel. The beyond the limitation of the potential flow theory. Research conducted
effects of FLNG's motions on LNG carrier are more obvious for their by Jiang et al. (2015) shows smaller roll motion RAOs are induced
differences in displacement. when the sloshing gets violent.

5.2.3. Sloshing nonlinearity 5.3. Connection system responses

Nonlinear free surface is adopted in the sloshing calculation, and the
nonlinearity of sloshing will become obvious with the increase of wave Responses of the FLNG system with connection system are discussed
amplitude, especially for the shallow filling level conditions. Fill Case D in this section. Restriction effects of the connection system on the
is selected and a smaller wave amplitude equals 0.5A is added. Fig. 26 vessels’ motion and loads on connection system are analyzed. FLNG
shows the RAOs of the vessels' motions under different wave ampli- system is restricted with soft horizontal spring lines rather than single-
tudes, the motion RAOs are slightly changed with the increase of wave mooring system in this study. Thus, weather-vaning property will not
amplitude. The wave elevation histories in T1 and T6 under regular occur as in practice and loads on connection system are mainly related
waves are plotted in Fig. 27. Regular wave frequencies (0.52, 0.61, to the wave excited motions of vessels.
0.77, 0.84 rad/s) around the natural roll motion frequencies and The responses of FLNG system with connection system in solid
sloshing frequencies of vessels in the fill Case D are selected. It can be loading condition are considered. The motion RAOs of the vessels under
seen sloshing wave elevation in the natural roll frequencies has no quarter (135°) wave are presented in Fig. 30, the numerical results have
obvious nonlinearity, while sloshing near the natural sloshing fre- good agreements with experimental results. It can be seen that the surge
quencies demonstrate obvious nonlinear properties with the increase of motions have large value in 0.39 rad/s. RAOs of the loads on hawsers
wave amplitude. As discussed above, sloshing around the natural roll the relative motions between the two vessels are shown in Figs. 31 and
frequencies are mainly excited by tank inclination and is linearly re- 32, respectively. Corresponding to peaks in the surge motions, peaks
lated to vessels' roll motion; while sloshing around the natural sloshing values of loads on hawsers and relative motions in sway mode also
frequencies are dynamically excited by sway motion and has nonlinear appear in 0.39 rad/s. It can be deduced that the resonant responses of
relationships with excitation amplitude. Fig. 29 show the free surface connection system in sway mode are excited. For the FLNG system

Fig. 26. Responses of vessels under different wave amplitudes.

76
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 27. Histories of sloshing elevations.

Fig. 28. Histories of sloshing loads.

Fig. 29. Free surface of sloshing in the liquid tank of LNG carrier.

consists of two floating vessels connected with springs, natural fre- observed from experimental results that LNG carrier has larger sway
quency of the connection system can be calculated as: and yaw motions, which may be caused by the second wave forces and
cannot be found in the numerical results.
K connect (M1 + M2) Sensitivities to the stiffness of the connection system are checked to
ω=
M1 M2 (45) optimize the motion responses of the vessels. Fig. 33 shows the RAOs of
where K connect stands for stiffness of connection system in specific di- vessels’ motions with different connection system stiffness defined
rection, M1 and M2 are the mass (including added mass) of FLNG and based on K shown in Table 2. With the stiffness of the connection
LNG carrier, respectively. Resonant responses will occur when the system decreased to K connect = 0.25K , resonant frequency in surge mode
natural frequency of connection system meet with wave frequency. reduces to half with much smaller peak response. In the case
Relative motions in other modes are not as obvious as in surge mode; K connect = 0.1K , the natural frequency of connection is much small and
the two peaks in the relative roll motion respectively correspond to the no resonant responses are induced. The motions of vessels are slightly
natural roll frequencies of FLNG and LNG carrier. Moreover, it can be affected by the connection system for the small stiffness. When

77
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 30. RAOs of vessels' motion with connection system.

Fig. 31. RAOs of loads on hawsers.

78
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 32. RAOs of relative motions between vessels.

Fig. 33. Sensitivities of the vessels' motions to the stiffness of connection system.

79
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 34. Loads on hawsers for connection system with different stiffness.

Fig. 35. RAOs of vessels' motions in solid and liquid loadings.

connection system has large stiffness K connect = 10K , motions in surge, correspondingly.
heave and roll modes are decreased significantly. The reason could be Sloshing effects on the connection system are analyzed by com-
that the two vessels are connected more closely for the larger stiffness paring responses of FLNG system in solid and liquid loading conditions.
and motions of vessel are more affected by the adjacent vessel through Stiffness of the connection system is set as K connect = 0.1K . The RAOs of
the connection system. For the large stiffness, the roll motion responses vessels’ motions and loads on the hawsers are presented Fig s 35 and 36.
of FLNG and LNG carrier have same natural roll frequency, which is As demonstrated in Section 5.1, for liquid loading condition, larger
larger than the natural roll frequency of the FLNG and smaller than the surge motions are induced by sloshing loads in the frequency region
natural roll frequency of the LNG carrier in the case without connection higher than natural sloshing frequency, and larger loads on hawsers can
system. Moreover, sway and yaw motions have increased responses in be found in liquid loading condition from Fig. 36. Roll motions have
the case K connect = 10K . The reason could be that the resonance of the decreased natural roll frequency in the liquid loading condition. As
connection system in sway mode is excited. Fig. 34 shows the loads on sway motion couples closely with roll motions, obvious frequency shift
the hawsers with different stiffness of the connection system. With the between solid and liquid loading conditions in sway motion RAO can be
increase of the stiffness, the loads on the hawsers increase observed. As a result, the peak value of loads on Hawser 1 appears in a

80
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Fig. 36. Loads on hawsers for vessels in solid and liquid loadings.

lower frequency. Vessels in liquid loading condition also have smaller side-by-side mooring to an FPSO. In: The Eleventh International Offshore and Polar
pitch motion RAOs, while the response amplitudes are rather small and Engineering Conference. International Society of Offshore and Polar Engineers.
Chen, X.B., 2005. Hydrodynamic analysis for offshore LNG terminals. In: Proceedings of
its influences on hawsers are not obvious. the 2nd International Workshop on Applied Offshore Hydrodynamics, Rio de Janeiro.
Chen, Z., Zong, Z., Li, H.T., Li, J., 2013. An investigation into the pressure on solid walls
6. Conclusion in 2D sloshing using SPH method. Ocean Eng. 59, 129–141.
Choi, Y.R., Hong, S.Y., 2002. An analysis of hydrodynamic interaction of floating multi-
body using higher-order boundary element method. In: The Twelfth International
A high efficiency numerical code is developed to investigate the Offshore and Polar Engineering Conference. International Society of Offshore and
coupled responses of side-by-side arranged FLNG system. Experimental Polar Engineers.
Faltinsen, O.M., 1978. A numerical nonlinear method of sloshing in tanks with two-di-
tests are conducted to validate the numerical code and provide more mensional flow. J. Ship Res. 22 (3).
perceptual understanding of the coupling mechanism. Based on nu- Hong, S.Y., Kim, J.H., Cho, S.K., Choi, Y.R., Kim, Y.S., 2005. Numerical and experimental
merical and experimental studies, the hydrodynamic interactions be- study on hydrodynamic interaction of side-by-side moored multiple vessels. Ocean
Eng. 32 (7), 783–801.
tween vessels, sloshing effects on vessels’ motions, and responses of the
Huijsmans, R.H.M., Pinkster, J.A., De Wilde, J.J., 2001. Diffraction and radiation of
connection system are investigated. The following conclusions are waves around side-by-side moored vessels. In: The Eleventh International Offshore
drawn: and Polar Engineering Conference. International Society of Offshore and Polar
Engineers.
Inoue, Y., Islam, M.R., 1999. Comparative study of numerical simulation and the ex-
1) Numerical program developed in this study based on potential flow perimental results for a parallelly connected FPSO and LNG in waves. In: The Ninth
is reliable and of feasible efficiency in the prediction of liquid International Offshore and Polar Engineering Conference. International Society of
loading FLNG system responses, the damping-lid applied in the gap Offshore and Polar Engineers.
Jiang, S.C., Teng, B., Bai, W., Gou, Y., 2015. Numerical simulation of coupling effect
zone can give good results in hydrodynamic coefficients calculation. between ship motion and liquid sloshing under wave action. Ocean Eng. 108,
2) Hydrodynamic interactions between vessels are sensitive to wave 140–154.
directions for shielding effects, vessels will subject to larger motion Kodan, N., 1984. The motions of adjacent floating structures in oblique waves. J. Energy
Resour. Technol. 106 (2), 199–205.
responses when located in weather side. Increased motion responses Koo, B.J., Kim, M.H., 2005. Hydrodynamic interactions and relative motions of two
can also be excited by the radiation wave of the adjacent vessel. floating platforms with mooring lines in side-by-side offloading operation. Appl.
3) The sloshing peaks in liquid loading vessels under beam sea mainly Ocean Res. 27 (6), 292–310.
Lee, S.J., Kim, M.H., Lee, D.H., Kim, J.W., Kim, Y.H., 2007. The effects of LNG-tank
excited by motions in the roll and sway modes, respectively. sloshing on the global motions of LNG carriers. Ocean Eng. 34 (1), 10–20.
Sloshing effects on vessels' sway motion can increase in the low Lu, L., Teng, B., Sun, L., Chen, B., 2011. Modelling of multi-bodies in close proximity
filling level for the natural sloshing frequencies are more close to under water waves—fluid forces on floating bodies. Ocean Eng. 38 (13), 1403–1416.
Malenica, S., Zalar, M., Chen, X.B., 2003. Dynamic coupling of seakeeping and sloshing.
vessels' main response frequency region. With the increase of wave
In: 13th International Offshore and Polar Engineering Conference. ISOPE, Honolulu,
amplitude, nonlinear sloshing becomes obvious near the natural HI, pp. 25–30 May.
sloshing frequency; while vessels' motions do not demonstrate ob- Mitra, S., Wang, C.Z., Reddy, J.N., Khoo, B.C., 2012. A 3D fully coupled analysis of
vious nonlinearity. nonlinear sloshing and ship motion. Ocean Eng. 39, 1–13.
Molin, B., 2001. On the piston and sloshing modes in moonpools. J. Fluid Mech. 430,
4) Resonant responses of vessels can be excited when the resonant 27–50.
frequency of connection system equals wave frequency. The vessels' Molin, B., Remy, F., Kimmoun, O., Stassen, Y., 2002. Experimental study of the wave
motions can be reduced significantly when the stiffness of the con- propagation and decay in a channel through a rigid ice-sheet. Appl. Ocean Res. 24
(5), 247–260.
nection system is rather large and interactions between the vessels Nakayama, T., Washizu, K., 1981. The boundary element method applied to the analysis
increase correspondingly. of two‐dimensional nonlinear sloshing problems. Int. J. Numer. Meth. Eng. 17 (11),
1631–1646.
Nam, B.W., Kim, Y., Kim, D.W., Kim, Y.S., 2009. Experimental and numerical studies on
Acknowledgement ship motion responses coupled with sloshing in waves. J. Ship Res. 53 (2), 68–82.
Newman, J.N., 2001. Wave effects on multiple bodies. Hydrodyn. Ship Ocean. Eng. 3,
This work was supported by State Key Lab of Ocean Engineering 3–26.
Newman, J.N., 2005. Wave effects on vessels with internal tanks. In: Proceedings of the
Shanghai Jiao Tong University and financially supported by China
20th Workshop on Water Waves and Floating Bodies, Spitsbergen, Norway.
National Scientific and Technology Major Project (Grant No. Pessoa, J., Fonseca, N., Soares, C.G., 2015. Numerical study of the coupled motion re-
2016ZX05028-006-009); all of these supports are gratefully acknowl- sponses in waves of side-by-side LNG floating systems. Appl. Ocean Res. 51, 350–366.
Rognebakke, O.F., Faltinsen, O.M., 2003. Coupling of sloshing and ship motions. J. Ship
edged by the authors.
Res. 47 (3), 208–221.
Sun, L., Taylor, R.E., Taylor, P.H., 2010. First-and second-order analysis of resonant
References waves between adjacent barges. J. Fluid Struct. 26 (6), 954–978.
Veritas, B., 2010. Hydrostar for Experts User Manual.
Wu, G.X., Ma, Q.W., Taylor, R.E., 1998. Numerical simulation of sloshing waves in a 3D
Buchner, B., Van Dijk, A., De Wilde, J., 2001. Numerical multiple-body simulations of tank based on a finite element method. Appl. Ocean Res. 20 (6), 337–355.

81
D. Zhao et al. Ocean Engineering 168 (2018) 60–82

Xu, X., Yang, J.M., Li, X., Xu, L.Y., 2015. Time-domain simulation for coupled motions of Zhao, D.Y., Hu, Z.Q., Chen, G., Lim, S., Wang, S.Q., 2018. Nonlinear sloshing in rectan-
three barges moored side-by-side in floatover operation. China Ocean Eng. 29, gular tanks under forced excitation. Int. J. Nav. Architect. Ocean Eng. 10 (5),
155–168. 545–565.
Yao, C.B., Dong, W.C., 2015. Modeling of fluid resonance in-between two floating Zhao, W.H., Yang, J.M., Hu, Z.Q., 2012. Hydrodynamic interaction between FLNG vessel
structures in close proximity. J. Zhejiang Univ. - Sci. 16 (12), 987–1000. and LNG Carrier in side by side configuration. J. Hydrodyn. Ser. B 24 (5), 648–657.
Zhao, D.Y., Hu, Z.Q., Chen, G., 2015. An investigation on two dimensional nonlinear Zhao, W.H., Yang, J.M., Hu, Z.Q., Tao, L.B., 2014. Prediction of hydrodynamic perfor-
sloshing in rectangular tank. In: ASME 2015 34th International Conference on Ocean, mance of an FLNG system in side-by-side offloading operation. J. Fluid Struct. 46,
Offshore and Arctic Engineering. American Society of Mechanical Engineers 89–110.
V001T01A045-V001T01A045. Zhu, H.R., Zhu, R.C., MIAO, G.P., 2008. A time domain investigation on the hydro-
Zhao, D.Y., Hu, Z.Q., Chen, G., 2017. Experimental investigation on dynamic responses of dynamic resonance phenomena of 3-D multiple floating structures. J. Hydrodyn. Ser.
flng connection system during side-by-side offloading operation. Ocean Eng. 136, B 20 (5), 611–616.
283–293.

82