Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

CHARACTERISTICS OF HEAT TRANSFER DURING COOLING DOWN PROCESS IN A

SINGLE CARGO TANK OF LNG CARRIER
(DOI No: 10.3940/rina.ijme.2020.a3.588)

J J Deng, L Y Song, and J Xu, Zhejiang Ocean University, Zhoushan, The School of Port and Transportation Engineering,
China, B Liu, Shijiazhuang Tiedao University, Shijiazhuang, The School of Mechanics Engineering, China, J S Lu, and
J W Zhang, Zhejiang Ocean University, Zhoushan, The School of Port and Transportation Engineering, China

KEY DATES: Submitted: 14/06/19; Final acceptance: 21/06/20; Published: 07/10/20

SUMMARY

A deep understanding of heat transfer characteristics is essential in evaluating risk and putting forward any option for the
Liquefied Natural Gas (LNG) tank cooling down process. A novel Computational Fluid Dynamics (CFD) model was built to
perform the flow and heat transfer simulation of the process. The predicted results agreed well with the test data from a prototype
LNG tank. Then the heat transfer characteristics of the process were analysed. It was found that the vapour temperature and
density were linearly varying and became stable after 2.3 hours. A sudden pressure drop risk was identified during the process,
which will cause the inwards collapse risk of the invar membrane. Then the proposals to prevent the risks of the inwards
collapsing membrane are presented. The heat transfer characteristics of the vapour and different membrane layers were analysed
in detail, and if the suggested option was to be implemented this could save about 39% of LNG consumed.

NOMENCLATURE vm Mixture’s mass-averaged velocity (m s-1)

vdr,d Drift velocity of phase d (m s-1)
A Contiguous area (m2)
Cp Specific heat capacity (J kg-1K-1) Yk Dissipation of k (kg m-1 s-3)
Dω Cross-diffusion term of ω (kg m-3 s-2) Yω Dissipation of ω (kg m-3 s-2)
Em The volume average energy of mixture (J m-3)
Abbreviation
F External body forces (N)
BOG Boiling-Off Gas
Gk Generation of k (kg m-1 s-3) CFD Computational Fluid Dynamics
Gω Generation of ω (kg m-3 s-2) CCS Cargo Containment System
hs The sensible enthalpy of solid material (J kg-1) LNG Liquefied Natural Gas
h Sensible enthalpy (J kg-1) RMS Root Mean Square
j Either phase v or phase d SST Shear Stress Transport
K Heat transfer coefficient (W m-2 K-1) Tcf Trillion Cubic Feet
keff Effective heat transfer coefficient (W m-2 K-1)
ks The conductivity of solid material (w m-1 K-1) Subscript
L The hydraulic diameter (m) m Mixture
mvd Phase transitions from vapour to droplet (kg s-1) d Droplet
mdv Phase transitions from droplet to vapour (kg s-1) evap Evaporated
m1 Injected LNG masses for each hour (kg) s Solid
m2 Evapourated LNG masses for each hour (kg) sec Secondary barrier layer
Mw Molecular weight of the gas (kg kmol-1) pri Primary barrier layer
N Number of samples use Utilization
Nu Nusselt number v or vap Vapour
Pop Operating pressure (Pa)
P Static pressure (Pa) Greek symbols
R Universal gas constant (J m-1 K-1) α Volume fractions of phase
Q Net heat transfer rate between layers (W) γ The relaxation time factor(s-1)
Q1 Heat transfer rate to the current layer (W) ε Accumulated root-mean-square
Q2 Residual heat of the current layer (W) η The utilization coefficient
Q3 Heat transfer rate from the current layer (W) ηuse Cooling energy utilization ratio
ST Energy source term due to the phase transition ηevap LNG droplet evaporated ratio
(W) λ Fluid thermal conductivity (W m-1 K-1
T Temperature (K) μ Viscosity of the fluid (Pa s)
T1 Correlation data of temperature (K) Γk Effective diffusivity of k (Pa s)
T2 Baseline data of temperature (K) Γω Effective diffusivity of ω (Pa s)
Tref Reference temperature (K) ρ Density (kg m-3)
ΔT Temperature difference of adjacent layers (K) g The gravitational acceleration (m s-2)
t Cooling down time (s)

©2020: The Royal Institution of Naval Architects A-231

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

1. INTRODUCTION The primary advantage of the LPM is time-saving.

However, its prediction tends to have a greater derivation
The intense push to pursue cleaner energy resources from the measured values because of its oversimplified
during the last decade for the purpose of environmental assumptions. Lu (2012) conducted an unsteady 3D
protection has resulted in the consumption of natural gas numerical simulation of the cooling down process of
worldwide is projected to increase from 120 trillion cubic membrane cargo tanks by using the versatile CFD
feet (Tcf) in 2012 to 203 Tcf in 2040 (EIA, 2016). The software ANSYS-FLUENT. The temperature field,
amount of the LNG transported by ship carriers has grown velocity field, coolant particle streamline and vapour
seven times faster than the amount of LNG transported by volume fraction contour of the fluid inside the cargo tank
pipeline, and it is predicted that by 2035 its proportion will were systematically investigated. However, the heat
reach approximately 50% (BP 2017), compared to 32% transfer between the liquid in the cargo tank and the walls
today. There are 452 LNG carriers being operated around of the tank has not yet been considered and the latent heat
the world, and 112 LNG carriers have been ordered to be caused by the phase transition has been ignored.
built. The expansion of LNG carriers is expected to
continue in the future due to the sharp increase in the The Lee model (Lee, 1980) has been widely utilized to
demand for LNG transportation (Raju et al, 2016; Ekanem model the phase transition (from liquid to vapour) model
Attah and Bucknall, 2015). to date, and it has been applied in the CFD simulations (Lu
et al, 2016a; Zhang and Wu, 2014; Liu et al, 2014).
When using carriers to transport LNG, safety issues must However, the derivation of the results varied widely from
be considered (Lee et al, 2011; Harris, 1993; Horvat, 2018; the measured data because of the improper equation used
Fulford and Slatter, 1988). The aim of the cooling down to calculate the state of the gas. Moreover, only the
operation is to cool down the tanks in accordance with thermal conductivity process of insulation materials was
specifications. This operation is carried out immediately presented in their studies, and the risks of the vapour
after the completion of the gassing-up in which the inert pressure, temperature histories, and optimization
gas is purged with BOG to prevent the high freezing temperature were not considered.
material from blocking the valves and equipment (WMT
Limited 2009; Jia et al 2013; CHINA CLASSIFICATION Two major issues have been widely ignored in the existing
SOCIETY 2014). Rapid cooling could significantly models and the approaches to predict the thermodynamics
weaken the strength of the construction (Zhu et al 2018b; associated with the cooling down operation. The first issue
Zhu et al, 2018a), which might leads to a critical condition is the suitable gas law, which could accurately predict the
of the cargo containment system (CCS) (Luo, 2011). pressure risk inside the tank. The second issue involves
operation economics. The negative gauge pressure due to
To ensure safe operations, it is necessary to understand the violent evaporation of the LNG droplets should be
the heat transfer mechanisms of the cooling down accurately predicted because it could result in the inward
process (Castillo and Dorao 2013; Wang et al 2018; Qu collapsing of the invar membranes since the barriers are
et al, 2018; Yan et al, 2016; Krikkis, 2018). No located on the outside of the tank without any support from
experimental test data or comprehensive theoretical the inside of the invar membranes. The pressure
analysis regarding this topic has been reported due to the oscillation inside the tank during the rapid cooling down
extremely high cost and the complicated process. process could destroy the vapour header, the LNG header,
Therefore, numerical simulations were chosen as the and the tank. The conservation of time and the resources
basic tools to investigate the flow and heat transfer in the by the cooling down operation leads to the improvement
tanks (Peng et al, 2019; Saleem et al, 2018a; Saleem et of the economics (Wilson, 1974). This also justifies the
al, 2018b). The scientific literature on numerical necessity of developing a reliable and efficient numerical
simulations for LNG tanks cover different methods, approach to predict the thermodynamics associated with
including the lumped parameter method (LPM) (Luo, the cooling down operation (Al-Sharafi et al, 2017),
2011; Cui, 2001; Li, 1996) and computational fluid contributing to the decision-making process for an
dynamics (CFD) simulations (Wang et al, 2010; Lu et al, optimized cooling down plan.
2016a; Zhang and Wu 2014; Lee et al, 2015; Lu, 2012).
Cui (2001) and Li (1996) developed a cooling down In this paper, a CFD model to simulate the flow and heat
prediction technology for Moss spherical tanks (Niu, transfer of the cooling down process is presented. The
2017) (which are the product by Moss, Norway) based incompressible ideal gas law is applied to compute the
on the LPM. This method assumed that the temperatures density and pressure accurately. The temperature,
of the fluid inside the cargo tanks and that of the tank density, and pressure of the vapour and the temperature
wall were uniforms with zero thermal gradient. The histories of the insulation layers are then analysed. The
equation of the thermal equilibrium was based on the inwards collapsing risk of the invar membranes due to
principle of energy conservation between the vapour the negative gauge pressure is identified, and proposals
inside the tank and the barriers of the tank, and the are presented. The heat transfer characteristics of the
equation was solved through iteration. This method primary and secondary barriers are also discussed, and
fundamentally ignored the temperature variations inside based on the discussion, and the proposal of suggested
the cargo tanks and the CCS. option is introduced.

A-232 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

2. MODEL AND VERIFICATION outside of the inner-hull is assumed to be constant at 300

K. (v) Only half of the cargo is modeled because of the
2.1 PHYSICAL MODEL symmetry of the physical and geometrical characteristics.

An LNG carrier tank of 14,000m3 capacity is shown

schematically in Figure 1. The X-Y plane is the horizontal
plane, with the X-axis oriented along the length of the tank.
The cross section is an octagon in the Y-Z plane. The
characteristic dimensions are also shown in this figure. The
tank is supported by the LNG CCS. Generally, an LNG CCS
consists of a thin, flexible membrane called the primary
membrane, which is in contact with the cargo, a layer of
plywood boxes filled with Perlite called the primary
insulation, a second flexible membrane similar to the first one
called the secondary membrane, and a second layer of boxes
also filled with Perlite in contact with the inner hull called the
secondary insulation (WMT Limited, 2009; Lu et al, 2016b).
In these layers, the primary and secondary insulation layers
act as barriers, and they are mainly insulation layers.
Figure 1: Physical model of the liquid cargo tank
The two invar membrane layers are the leakage prevention
layers whose thermal expansion coefficients is very small, The material properties of the model and the LNG flow rate
its thermal expansion coefficients are about one-tenth of the at different cooling times are shown in Table 1 (Choi et al,
thermal expansion coefficients 304 type’s stainless steel. 2012) and Table 2 (WMT Limited, 2009), respectively.
The thicknesses of the primary and secondary barriers are
230mm and 300mm, respectively. The coolant LNG 2.2 MATHEMATICAL MODEL
droplets are sprayed from the liquid header which is defined
as the velocity inlet in the model. The vapour header, which The mixture model is chosen because it is a perfect
is defined as the pressure outlet, is located in the middle of substitute for the full Eulerian multiphase model when
the liquid header. The liquid header and vapour header are there is a wide distribution of the particulate phase. In this
located in the middle of the roof of the tank. process, there are two phases: the vapour occupies most
volume of the tank, it is set as the primary phase, and the
It was feasible to construct a simplified computational liquid droplets are set as the secondary phase. The vapour
domain to reduce the computational run-time by taking and droplet phases are subscripted as v and d, respectively.
advantage of the following facts that: (i) The thicknesses The mixture phase, subscripted as m, is the combination
of the tank primary barrier invar membranes are very of the vapour and droplet phases.
small (0.7 mm) and their thermal conductivity is very large.
Thus, the invar membranes are ignored in the heat-transfer The multi-phase mixture model solves the continuity,
model. (ii) The physical properties of the materials of the momentum, and energy equations for the mixture and the
insulation layers and those of the carrier’s inner-hull are volume fraction equations for the secondary phase/phases
steady. (iii) The vapour density is in accordance with the (Patankar, 1980; Veersteg and Malalasekera, 1995). The
incompressible ideal gas law, the vapour density governing equations include (1) the mass conservation of
difference is 170% when the temperature reduces from the mixture:
300 K to 110 K, whereas the vapour density difference is
only 1.8% when the pressure changes from 7000 Pa to
9000 Pa. therefore the effect of pressure change on the
m
+ ( m vm )=0 (1)
t
vapour density can be ignored. (iv) The temperature of the

Table 1. The material properties of fluids and solids

Density Specific heat Viscosity Thermal conductivity Latent heat
Material ( kg / m 3 ) ( J / (kg k) ) ( kg / m s ) ( w / (m k) ) ( kJ / kg )

LNG --- 3408 0.1176 0.186 511.15

Fluid 1.087
Vapour variable 2222 variable ---
× 10−5
Barriers Expanded Perlite 50 487 --- variable ---
Inner-hull Steel 8130 480 --- 13.8 ---

©2020: The Royal Institution of Naval Architects A-233

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

Where ρm is the density of the mixture in kg/m3, 𝑡 is the time kg/(m3.s2) respectively , Yk and Yω are the dissipation of k
in s, vm is the mixture’s mass-averaged velocity in m/s. in kg/(m.s3) and the dissipation of ω in kg/(m3.s2)
respectively.
The volume fraction equations for the primary and
secondary phases are described as follows: A mass transfer model describing the process of
evaporation and condensation that was introduced by Lee
(Lee, 1980) has been proven to be robust. In this model,
( ad d )
+ ( ad d vm ) = − ( ad d vdr,d ) + ( mvd − mdv ) the mass transfer between phases was dependent on the
saturation temperature, Tsat. The directions and
t
av = 1 − ad magnitudes of the mass transfer rates are described as
follows: evaporation occurs when T > Tsat. The mass of the
(2)
liquid phase in the control volume decreases, but the mass
of the vapour phase increases correspondingly, which
Where ρd is the density of the droplet in kg/m3, αd and αv are
means the mass is transferred from the liquid to the vapour.
the Volume fractions of phase droplet and vapour
The synchronized phase transitions from vapour to droplet
respectively, vdr,d is the drift velocity of phase droplet in m/s,
and from droplet to vapour are shown below.
mvd and mdv are Synchronized phase transitions from
vapour to droplet and droplet to vapour respectively in kg/s. T − Tsat
mdv = d d ,T Tsat
Tsat (7)
The momentum and energy equations of the mixture phase
mvd = 0, T Tsat
are then solved:

( m vm ) Where γ is the relaxation time factor in 1/s, T and Tsat are

+ ( m vm vm )= the temperature and the saturation temperature in K
t (3) respectively.
− p+ ( vm + vm T
)+ mg +F
Similarly, condensation occurs when T < Tsat. The mass of
the liquid phase in the control volume increases, but the
and
mass of the vapour phase decreases correspondingly,
which means the mass is transferred from the vapour to
( m Em )
t
+ ( vm ( m Em + p )) =
(4)
the liquid. The magnitude of the mass transfer is

( keff T ) + ST
mvd =
Tsat − T
,T Tsat
v v
Tsat (8)
Where p is the static pressure in Pa, g is the gravitational mdv = 0, T Tsat
2
acceleration in m/s , F is the external body forces in N,
Em is the volume average energy of mixture in J/m3, keff is Where ρv is the density of the vapour in kg/m3.
the effective heat transfer coefficient in W/(m2.K), ST is the
heat source due to the phase change in W. The coefficient γ needs to be fine-tuned and can be
interpreted as the relaxation time factor that controls the
The shear-stress transport (SST) κ − ϖ model (Menter, strength of the phase transition. The value of γ varies for
1994) was developed to effectively blend the robust and different situations. An excessively large value of γ can
accurate formulation of the κ − ϖ model in the near-wall cause a number of convergence problems, while an overly
region with the free stream independence of the κ − ε values can result in a significant deviation between the
model in the far field region. This is more accurate and interfacial temperature and the saturation temperature.
reliable for this simulation. The value of γ was specified as 0.1 s-1 when the process of
LNG evaporation (Tsat= 110 K) was simulated by Lu
( k)
+
( ku j )= k
+ Gk − Yk (5)
(2016b). In this simulation, γ=0.1 s-1 will also be used.
k
t xj xj xj As long as the mass source term is obtained, the energy
source term can be obtained by taking the latent heat
( )
+
( uj )= Γω + Gω − Yω (6)
during the phase change into account. The latent heat at a
t xj xj xj given temperature of the liquid due to the phase transition
in this model is constant. The energy source term due to
the phase transition denoted as S T , is calculated as
Where k and ω are the turbulence kinetic energy in m2/s2
and the specific dissipation rate in 1/s respectively, Γk and follows:
Γω are the effective diffusivity of k and the effective
diffusivity of ω in Pa.s respectively, Gk and Gω are the ST = hd ( mdv − mvd ) (9)
generation of k in kg/(m.s3) and the generation of ω in

A-234 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

Where hd is the latent heat of phase transformation in J/kg. At the initial period, the tank was filled with 300 K boiling
off gas (BOG) whose pressure is 8000 Pa in the fluid zone
For the cooling down process, the vapour density is (WMT Limited, 2009), and the temperature of all barriers and
affected more by the temperature change rather than by the the hull was 300 K. The non-uniform unstructured
pressure change. Therefore, the incompressible ideal gas tetrahedron mesh was created by GAMBIT software and then
law was applied to compute the density. In this form, the converted to a polyhedron mesh by Fluent, the images of
vapour density only depends on the operating pressure mesh are shown in Figure 2. The total number of mesh cells
instead of on the local gauge pressure. was approximately 2×105, and an initial grid sensitivity study
demonstrated that the grid enables accurate prediction of the
Pop flow and heat transfer parameters.
v = (10)
RT / M W
Table 2. The LNG flow rates at different cooling times
Where Pop is the operation pressure that was set as 8000
Cooling down time LNG flow rate
Pa (WMT Limited, 2009), R is the universal gas constant
(h) (m3/h)
that is 8.31 J/(mol.k), and Mw is the CH4 in kg/mol.
0≤t<2 50
In the inner hull, for the primary barrier and the secondary 2≤t<12 70
barrier, only the energy conversion equations were solved,
which have the following forms:
2.4 TIME-STEP SIZE AND GRID SENSITIVITY
( sh )
= ( ks T) (11) In addition to the base grid described in the previous
t
section, two more grids were generated with a √2
and
T
refinement ratio. One of them was a coarser grid with
140,000 cells, and the other was a finer grid with 280,000
hs = Cp dT (12)
cells. The base time step was set as 18s. For the assessment
Tref of the time-step size dependence, two more time-step sizes
were tested, a larger one, 20s, and a smaller one, 16s. Table
Where ρs is the density of material of hull, primary barrier 3 shows the results of the dependence tests. The results are
and the secondary barrier in kg/m3, ks is conductivity of represented by the accumulated root-mean-square (RMS)
solid material in W/(m.K), hs is the sensible enthalpy of (Lee et al, 2011) values of the temperature differences of
solid material in J/kg, Cp is the specific heat of solid the primary and secondary barriers over the entire
material in J/(kg.K). computation time.

2.3 SIMULATION CONDITION AND MESH 1 T1 -T2 2

ε = ∑N
n=1 √| | (13)
N T2
The spraying nozzle boundaries were set to be velocity
inlets passing through the LNG with variable volumes,
which are shown in Table 2, and the coolant temperature Where ε is accumulated root-mean-square, N and n are the
was 110 K. The outlet was set to be a wall because the count of temperature, T1 and T2 are correlation data of
safety valve could not be opened when the gauge pressure temperature and baseline data of temperature in K
was lower than 25000 Pa (WMT Limited, 2009). The respectively.
interfaces between the fluid and the primary barrier, the
primary barrier, and the secondary barrier, and the Note that the accumulation was done by adding in a
secondary barrier and the hull were set to the coupled discrete sense, i.e., every 5 min. From Table 3 it can be
interface boundary conditions, which allowed for heat flux clearly seen that the differences were quite small, and the
through the interfaces. An atmosphere temperature of 300 solutions were independent of the grids and time-step
K was given to the outside of the hull wall boundary. sizes in the tested range.

Table 3. Time-step size and grid dependence test

Primary barrier Secondary barrier
ϵgrid (Coarse − Medium) 2.10 × 10−2 3.88 × 10−3
ϵgrid (Medium − Fine) 8.87 × 10−3 2.67 × 10−3
ϵtime step size (Large − Medium) 1.61 × 10−3 2.69 × 10−4
ϵtime step size (Medium − Small) 8.27 × 10−4 1.45 × 10−4

©2020: The Royal Institution of Naval Architects A-235

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

(a)principal section of mesh (b)top view of nozzle zone mesh

(c)side view of mesh (d)principal section of nozzle zone mesh

Figure 2: Mesh generation for the computational domain

2.5 EXPERIMENTAL VALIDATION 330

310
A cooling down test of a prototype LNG carrier cooling down numerical
290
test (WMT Limited, 2009), carried out by WMT was simulated
Mean temperature(K)

using the proposed model, and the simulation results were 270 experimental (reference)
compared with the test data. The simulation results were 250
compared with the test data of the prototype LNG carrier for 230
validation, and the comparison of the time histories of the mean
210
temperatures of the primary barrier is shown in Figure 3. It can
be observed that the CFD model slightly underestimated the 190
temperature initially and then overestimated it. However, the 170
gradually reducing trend was clear. 150
130
Considering the complexity of the numerical modeling, -2 0 2 4 6 8 10 12 14
reasonable simplification and difficulty in the measurement,
t (h)
it was agreed that the results were acceptable. Although there
were some discrepancies between the CFD predictions and Figure 3: The comparison of the mean time-varying
the experimental measurements, the CFD model showed an temperature of the primary barrier
acceptable performance. The time-varying temperature was
successfully predicted. This indicates that the proposed To better understand the heat transfer mechanism during
numerical methods can be used to simulate the LNG carrier. the cooling down process, the cooling energy utilization
ratio (ηuse) and the LNG droplet evaporated ratio (ηevap)
were studied:
3. RESULTS AND DISCUSSION
use =Q2 / Q1 = ( Q1 − Q3 ) / Q1 (14)
In this section, the temperature, density, and pressure = m2 / m1 (15)
evap
characteristics of the vapour inside the tank are analysed based
on the model, and the temperature histories of the different Where ηuse is the cooling energy utilization ratio, ηevap is
barriers, as well as the heat transfer characteristics between the LNG droplet evaporated ratio, Q1 is the heat transfer
layers, are shown and discussed. The mean velocity, rate that input to from the current layer in W, Q2 is the heat
temperature, density, pressure of vapour is calculated based on transfer rate that output from the current layer in W, Q3 is
the volume-weighted average of the variable on a 3D location. the residual heat of the current layer in W, m1 is the injected

A-236 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

LNG masses for each hour in kg, m2 is the evaporated utilization ratios, the Nusselt number, and the velocity histories
LNG masses for each hour in kg. of the vapour are shown in Figure 7.

The heat transfer coefficient K (Miana et al, 2016) and the It was found that there were two significantly different
Nusselt number Nu are defined by the following periods for the temperature and density histories of the
expressions. vapour in Figure 4(a). Specifically, the temperature of the
vapour underwent an initial sharp drop and the density of
K = Q /(A T) (16) the vapour increased linearly during the first period (t <
2.3h). Due to the almost constant heat sink, which is
and
caused by the constant LNG spray amount during this
K .L period, the slopes of the temperature vs. time curve stayed
Nu = (17)
at about -91.60K/h. And the slopes of the vapour density
vs. time curve keeps at 0.55kg/m3.h because of the almost
Where K is the heat transfer coefficient in W/(m2.K), Q is constant temperature decrease slope. The vapour average
the net heat transfer rate between two contiguous layers in temperature dropped from 300 K to 215.22 K at 0.5h and
W, A is the heat exchange area in m2, T is the difference 116.80 K at 2h (Figure 5) during this period, and the
of the average temperature for the adjacent layers in K, Nu vapour density increased from 0.70kg/m3 to 0.98kg/m3 at
is the Nusselt number, L is the hydraulic diameter in m, λ 0.5h and 1.80kg/m3 at 2h (Figure 6). The huge temperature
is the fluid thermal conductivity in W/(m.K). differences between vapour and LNG droplets (110K)
leads to a strong heat transfer rate, which tends to result in
As LNG droplets were sprayed into the tank, the LNG flash evaporation of LNG with the same spray amount
droplets flash evaporated due to the violent heat transfer cryogenic coolant droplets injection. Consequently, the
from the hot vapour. The vapour was cooled down by the vapour temperature of the vapour dropped at a constant
LNG droplets, and then the barriers were cooled down slope, and the vapour density increased at a constant slope.
synchronously by the cold vapour. In the second period (t > 2.3h), the vapour temperature and
density are no longer changing.

3.1 VAPOUR COOLING ANALYSIS The gauge pressure history of the vapour inside the tank is
shown in Figure 4(b). It was found that the mean gauge
The vapour changes which include the density, the temperature, pressure declined sharply to -120 Pa from 0 Pa at the initial
and the pressure were analysed and discussed. The time- period. Then the relative pressure increased linearly and
varying mean temperature and the density, as well as pressure reached 0 Pa at the 1.64th hour. At the end of the cooling
histories of the vapour, are shown in Figure 4. The temperature down process, the relative pressure reached 527.85 Pa.
of the primary barrier and the temperature differences of the The relative pressure increased with a constant slope due
primary barrier and the vapour are also included in Figure 4 for to the almost constant LNG volume of LNG injection. The
comparison. The temperature, density and velocity final absolute pressure (8527.85 Pa) of the vapour doesn’t
distributions comparison at a different time in middle sections exceed the outlet (safety valve) opening limit (25000 Pa)
of the tank are shown in Figure 5, Figure 6 and Figure 8 during the whole cooling down process, so the boundary
respectively. The heat transfer rate, the evaporated and cooling condition setting of the outlet in this mode is acceptable.

(a) The histories of temperature, temperature differences, and density (b) Mean pressure histories
Figure 4: The mean temperature, density and pressure histories of the vapour

©2020: The Royal Institution of Naval Architects A-237

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

(a)0.5 h (b)1 h

(c)1.5 h (d)2 h

Figure 5: The temperature distribution comparison in middle sections of the tank

A-238 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

(a)0.5 h (b)1 h

(c)1.5 h (d)2 h

Figure 6: The vapour density distribution comparison in middle sections of the tank

The LNG droplet evaporation ratios experienced three increased to a high level and the Nu number have the
distinct stages, as shown in Figure 7 - namely the fast similar trend as the vapour velocity (Figure 7(b)). After 1h,
cooling stage, the transition cooling stage, and the stable the temperature difference between the LNG droplets and
cooling stage. The fast cooling stage (first stage) lasted the vapour reduces to 45.5 k. Although the LNG droplets
about 1.5 hours. In this stage, the injected LNG droplets still could be fully evaporated, the evaporation and heat
fully evaporated (Figure 7(a)) because the vapour transfer caused vapour velocity and the Nu number are
temperature was sufficiently high. Meanwhile, the vapour reduced (Figure 7(b)). In the fast cooling stage, the heat
velocity caused by LNG evaporation and heat transfer transfer rate between the vapour and the primary barrier
between vapour and the primary barrier is relatively increases rapidly and then keeps at 1.80 × 105 W for
intense (Figure 8(a)), hence the vapour velocity is rapidly about half an hour.

©2020: The Royal Institution of Naval Architects A-239

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

From expression (16) and (17), 7(a)). This stage is defined as the transition cooling stage
because the cooling capacity has degenerated.
L
Nu = Q (18) The third stage is the stable cooling stage, which continues
(A T)
from the 2.5th hour to the end. Before this stage, the vapour
was cooled to near the saturation temperature (Figure 4(a)),
Where L is the characteristic length which is a constant, hence the LNG droplets’ evaporated ratio slowly declined
almost keep constant, A is the characteristic area size to 10%. The vapour velocity and the Nu number stay very
which is a constant and T is the temperature different small (Figure 7(a)) because of the low LNG droplets’
which can be found from Figure 4.(a). So Nu number evaporated ratio, which causes the lower heat transfer rate
behaves similarly to the rate of heat transfer Q, which is between the vapour and LNG droplets.
increase when the velocity increases. In the period, the Nu
number of the heat transfer between vapour and primary There are two heat sources for LNG droplets
barrier have the similar trend as the heat transfer rate, due evaporation. One part comes from the vapour itself and
to the rapidly increasing of the velocity. another is from the primary barrier which is transferred
by vapour. The ratio of heat come from vapour itself in
From 1.5 h to 2.5 h, the injected LNG droplets evaporation the total heat is marked as cooling energy utilization
ratio sharply decreases from 100% to 30% (Figure 7(a)). ratio of the vapour, and the cooling energy utilization
The vapour average velocity and the Nu number are fallen of the vapour itself is shown in Figure 7(a) which is as
to the lowest values (Figure 7(b)) due to the lowest high as 96.63%. The majorities (96.63%) of the heat
temperature different of vapour and LNG droplets, which come from the vapour itself, and only a very small part
causes the heat transfer rate between the vapour and primary of the heat comes from the barrier.
barrier sharply decrease by 53% to 8.51 × 104 W (Figure

(a) The evaporated, cooling utilization ratios and heat (b) The Nu number and the volume average velocity
transfer rate between the barrier and the vapour

Figure 7: The evaporated and cooling utilization ratios, the Nu number and the velocity histories of the vapour

A-240 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

(a)0.5 h (b)1 h

(c)1.5 h (d)2 h
Figure 8: The mixture velocity distribution comparison in middle sections of the tank

3.2 PRIMARY BARRIER COOLING inside the tank, which are the fast heat transfer period, the
CHARACTERISTIC ANALYSIS transition period, and the stable heat transfer period.

The LNG droplets are the final thermal sink of the heat The fast heat transfer stage starts from the beginning and
from the barriers and vapour. The heat of the primary continues for 1.5h. The heat transfer rate and heat transfer
barrier is controlled by the convective heat transfer from coefficients (Figure 10) of vapour and primary barrier are
the vapour inside the tank and the heat conduction from intensely higher than those of the primary and secondary
the secondary barrier. barriers during this period. The volume average velocity and
the Nu number are sharply increased to their maximum values
The time-varying temperature and the temperature (Figure 7), therefore, convection heat transfer dominates the
profiles at a different time of the different layers are shown heat transfer at this stage. Convective heat transfer often
in Figure 9. The heat transfer rate and the heat transfer referred to simply as convection, is the transfer of heat from
coefficients of the different interfaces are shown in Figure one place to another by the movement of fluids, which is more
10 and the cooling energy utilization coefficient is shown efficient than heat conductivity. Therefore, the convective heat
in Figure 11. The heat transfer of the primary barrier has transfer mechanism is a reasonable explanation for this
different characteristics in different periods. It can be intensely higher heat transfer rate.
found in these Figures that:
From 1.5h to the 3.0h, the heat transfer rate from the primary
(1) The temperature of the primary barrier drops with the barrier to vapour inside the tank decreases by 50%, from 1.82
vapour temperature. 105 W to 8.62 104 W, which is defined as the transitory
stage. In this stage, the Nu is gradually decreased, leading to a
(2) There are three significantly different stages of the heat reduction of convective heat transfer. Simultaneously,
transfer rate between the primary barrier and the vapour conductive heat transfer becomes increasingly dominant.

©2020: The Royal Institution of Naval Architects A-241

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

Generally, the convective and conductive heat transfer In the stable cooling stage, the temperature difference
mechanisms have a comparable important position in this stage. between the primary barrier and the vapour becomes
lower with the heat transfer mechanism transition; the heat
After 3h, the heat transfer rate from the primary barrier to the transfer rate sharply declines and the temperature gradient
vapour decreases to the same level as the heat transfer rate of primary barrier consequently decreases. The cooling
from the secondary barrier to the primary barrier. The overall energy utilization coefficient drops to 24.7% during this
heat transfer coefficient (Miana et al, 2016) of the two sides period, which is a decrease of 50%.
of the primary barrier is at the same level, and the heat
transfer is controlled by conductive accordingly. The third stage is the slow cooling stage. In this stage,
the heat transfer efficiency of the two sides of the
(3) Consequently, there are three different stages of the primary barrier is at the same level, and the temperature
temperature changes in the primary barrier, which are the gradient is extremely small. The cooling energy
rapid cooling stage, the stable cooling stage, and the slow utilization coefficient drops to 13.0%. The primary
cooling stage. barrier becomes extremely hard to cool.

In the rapid cooling stage, the heat transfer rate is intensely The heat transfer rate between the primary barrier and the
higher than that of the primary and secondary barriers, vapour drops rapidly from 1.6 105 W to 8 104 W after
from the beginning to 3.25h. Accordingly, the temperature 7.25 h. Accordingly, the injection coolant can be reduced
of the primary barrier drops rapidly. This temperature drop by 50% to economize. The comparison of the LNG flow
is mainly controlled by the temperature of the vapour. The rate between baseline and suggested option is shown in
cooling energy utilization coefficient is higher, and the Table 3. In the suggestion option, 210 m 3 of the coolant
cooling energy utilization coefficient has the lowest value can be saved and the economical ratio is 26.3%.
during this period, 74.3%.

(a) Temperature histories (b) Temperature profiles

Figure 9: The temperature histories and temperature profiles of different layers

(a) Heat transfer rate (b) Heat transfer coefficients

Figure 10: The heat transfer rate and coefficients of different layers of primary barrier cooling

A-242 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

1) The temperature of the secondary barrier drops due to

the drop in the temperature of the primary barrier.
2) The heat transfer mechanisms of the secondary barrier
from the two side layers are conductive, and
consequently, the heat transfer rate (Figure 12(a)) is
mainly controlled by the temperature difference
(Figure 12(b)). The curves of the temperature
difference and the heat transfer rate have the same
tendency.
3) The temperature difference between the secondary
barrier and inner hull becomes higher, and
accordingly, the heat transfer rate from the inner hull
to the secondary barrier becomes higher. The cooling
energy utilization coefficient drops to 61.2% during
the cooling process.
Figure 11: The cooling energy utilization coefficient of the
primary barrier
3.4 TOTAL COOLING ENERGY UTILIZATION
COEFFICIENT
3.3 SECONDARY BARRIER COOLING
CHARACTERISTICS ANALYSIS The total cooling energy utilization coefficients of the
combination of the primary and secondary barriers are
Conductive dominate the heat transfer in the secondary shown in Figure 14. It can be seen there are two different
barrier, and it includes the conductive heat from the stages of the total cooling energy utilization coefficient
primary barrier and the inner hull. The cooling during the cooling process.
characteristics are different from the primary barrier 1) In the high-efficiency cooling stage, from the
because of the different heat transfer mechanism. The beginning to the 4th h, the cooling energy utilization
temperature and temperature difference histories of the coefficient is very high and the lowest cooling energy
different layers associated with the cooling of the utilization coefficient is 94.8%. Up to 94.8% of the
secondary barrier are shown in Figure 12. The heat cooling energy is used to cool the barriers.
transfer rate through the different layers and the cooling 2) In the utilization rate of the decline stage, from 4 th h
energy utilization coefficient associated with the cooling to the end, the cooling energy utilization coefficient
of the secondary barrier are shown in Figure 13. declines The lowest cooling energy utilization
coefficient is 70.7%, and the mean cooling energy
From Figure 12 and 13, the cooling characteristics of the utilization coefficient is 88.7%. During this stage,
secondary barrier are shown as follows. 88.7% of the cooling energy which transfers from
vapour is used to cool the barriers.

(a) Mean temperature (b) Mean temperature difference

Figure 12: The temperature and temperature difference histories of the different layers of cooling for the secondary barrier

©2020: The Royal Institution of Naval Architects A-243

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

(a) Heat transfer rate (b) Cooling energy utilization coefficient

Figure 13: The heat transfer rate of the different layers and cooling energy utilization coefficients of the secondary barrier

almost the same cooling down performance as the baseline,

so it is acceptable and feasible. It saved 315m3 of the LNG
in suggested option and the saving ratio was 39.38%.

Table 4. The LNG flow rate comparison between the

baseline and the suggested option

Cooling down LNG flow rate (m3/h)

time (h) suggested
baseline option
0≤t<2 50 50
2≤t<3 70 70
3≤t<12 70 35
Total 800 485

Figure 14: Cooling energy utilization coefficient of the

two barriers and the vapour Table 5. The RMS of the baseline and suggested option data
RMS
4. OPERATION PROPOSALS TVapour 7.81 × 10−4
The LNG evaporation ratio sharply declined when the 𝜌Vapour 7.86 × 10−4
vapour temperature dropped to near 110K after 2 hours, PVapour 2.95 × 10−1
thus a reduced injected LNG amount was available to save TPrimary barrier 1.02 × 10−2
the LNG. A simulation of the suggested option was TSecondary barrier 1.40 × 10−3
conducted, in which the sprayed LNG was reduced by 50% TInner hull 1.12 × 10−7
from the 3rd hour to the end. The comparisons of the LNG
flow rate between baseline and suggested option are The negative relative pressure could result in an inwards
shown in Table 4. collapsing risk for the invar membranes because the barriers
are located at the outside of the tank but without any support
The vapour pressure and temperature comparisons inside the tank. The pressure oscillation risk, caused by
between baseline and suggested option are shown in negative relative pressure overlying due to multi tanks
Figure 15 and 16, and the RMS is shown in Table 5. The cooling down at the same time, could destroy the vapour
vapour pressure decreased from the 3rd hour due to the header, the LNG header, and the tank. For safe operation
reduction of the LNG flow rate, but the temperature and risk avoided several proposals are shown as follows:
differences between the baseline and suggested option and 1) It is essential to monitor the gauge pressure inside the
the RMS values were very small. The suggested option has tanks carefully, and the emergency response measures

A-244 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

should be predefined and prepared. Cryogenic BOG 5. CONCLUSION

is a possible alternative coolant for the initial period;
another possible alternative involves tuning the A multiphase CFD model was presented to simulate the
injection LNG droplets amounts to avoid negative phase change and heat transfer of the cooling down
relative pressure. Another choice would be to inject process of a single cargo tank. The temperature, density,
BOG from the vapour header during the initial period, and pressure of the vapour inside the tank and the heat
which is able to keep the relative pressure positive. transfer characteristics of the primary and secondary
2) Avoid launching the cooling down process for multi barriers have been comprehensively analysed, and the
tanks at the same time. The middle tank is proposals have been presented.
recommended for the first launch, while the other
tanks work as buffer tanks via the vapour valves that 1) The predicted results show good agreement with the
are kept always opened and connected to the vapour measured data, with a maximum bias of 12.8%.
headers. It is suggested that these tanks be launched 2) The temperature of the vapour drops from 300K to
one by one. The next tank should be launched one 110K and the density of the vapour increases from
hour after the last cooling tank relative pressure 0.70 kg/m3 to 1.89 kg/m3 in 2.3 h. The slopes of the
reaches 0 Pa. This launching rhythm will help to avoid time-varying temperature and density curves change
the pressure oscillation risk. with constant slopes during this period; the slopes are
-48.7 K/h and 0.31 kg/m3.h for the temperature and
the density, respectively.
3) The relative pressure declines sharply to -120 Pa at
the initial period, then the relative pressure increases
with a constant slope and reaches 0 Pa at 2.9 h, and at
the end of the cooling down process, the relative
pressure reaches 527.85 Pa. The maximum absolute
pressure (8527.85 Pa) of the vapour doesn’t exceed
the outlet (safety valve) opening limit (25000 Pa), so
the wall boundary condition setting of the outlet is
acceptable.
4) The cooling process of the primary barrier has three
different periods, which are the rapid cooling period,
stable cooling period, and the slow cooling period,
because of the three significantly different heat
transfer mechanisms.
Figure 15: Comparison of the baseline and optimized
mean pressures of the vapour

(a) Vapour temperature (b) The temperature of the barriers

Figure 16: Comparison of the baseline and suggested option barrier temperatures

©2020: The Royal Institution of Naval Architects A-245

 Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

5) The heat transfer mechanism of the secondary barrier safety using experimentally defined thermal
from the two side layers is conductive. Consequently, properties. Applied Ocean Research, 37, 72-89.
the heat transfer rate is mainly controlled by the 5. CUI, Y. 2001. The liquid cargo technology of
temperature difference. The cooling energy utilization LNG carrier and the thermodynamic study of the
coefficient drops to 61.2% during the cooling process. pre-cooling process. Master, Shanghai Maritime
6) Up to 94.8% of the cooling energy was used for University.
cooling the barriers from 0 to 4h, the cooling energy 6. EIA, U. S. 2016. International Energy Outlook
utilization coefficient dropped to 70.7%, and the 2016. Washington, DC: U.S. EIA.
mean value was 88.7% from 4h to the ending. 7. EKANEM ATTAH, E. & BUCKNALL, R. 2015.
An analysis of the energy efficiency of LNG ships
Based on the calculation results analysis, the proposals for powering options using the EEDI. Ocean
the safety operations and saving costs are shown as Engineering, 110, 62-74.
follows. 8. FULFORD, N. J. & SLATTER, M. D. 1988.
1) Carefully monitor the gauge pressure and the Developments in the safe design of LNG tanks.
emergency response should be predefined. The Cryogenics, 28, 810-817.
cryogenic BOG is an alternative coolant for the initial 9. HARRIS, F. S. 1993. Safety features on LNG
stage, or for tuning the injection LNG amounts. ships. Cryogenics, 33, 772-777.
2) Avoid launching the cooling down process for multi 10. HORVAT, A. 2018. CFD methodology for
tanks at the same time. The middle tank is simulation of LNG spills and rapid phase
recommended for the first launch. The tanks launch transition (RPT). Process Safety and
one by one, this launching rhythm will help to avoid Environmental Protection, 120, 358-369.
the pressure oscillation risk. 11. JIA, S., LV, J. & DENG, Q. 2013. A simulation
3) The injection coolant can be reduced by 50% after 3 study of boil-off gas (BOG) pre-cooling process
hours because of the sharp decrease of the heat in unloading pipelines in an LNG terminal in
transfer rate from the primary barrier to the vapour Zhejiang. Natural Gas Industry, 33, 84-88.
inside the tank. The coolant mounts can be saved by 12. KRIKKIS, R. N. 2018. A thermodynamic and
315 m3, and the economical ratio is 39.38%. heat transfer model for LNG ageing during ship
transportation. Towards an efficient boil-off gas
management. Cryogenics, 92, 76-83.
6. ACKNOWLEDGMENTS 13. LEE, H. B., PARK, B. J., RHEE, S. H., BAE, J.
H., LEE, K. W. & JEONG, W. J. 2011. Liquefied
This work was supported by a Project supported by the natural gas flow in the insulation wall of a cargo
National Science Foundation for Young Scholars of China containment system and its evapouration.
(Grant No.11602222), the Zhejiang basic public welfare Applied thermal engineering, 31, 2605-2615.
research project (Grant No.LY18E090009), Zhoushan 14. LEE, J. H., KIM, Y. J. & HWANG, S. 2015.
City Technology Bureau Project Funding (Grant Computational study of LNG evapouration and
No.2016C41021). heat diffusion through a LNG cargo tank
membrane. Ocean Engineering, 106, 77-86.
15. LEE, W. H. 1980. A Pressure Iteration Scheme for
7. REFERENCES two-phase flow modeling, Washington D C,
Hemisphere Publishing.
1. AL-SHARAFI, A., YILBAS, B. & ALI, H. 2017. 16. LI, P. 1996. On the pre-cooling temperature of
Heat Transfer and Fluid Flow Characteristics in spherical cargo tank for MOSS type LNG Carrier.
a Sessile Droplet on Oil-Impregnated Surface Journal of Shanghai Maritime University, 17, 73-
Under Thermal Disturbance. Journal of Heat 78.
Transfer, 139(9), 1-49. 17. LU, J., XU, S., DENG, J., WU, W., WU, H. &
2. BP. 2017. BP Energy Outlook 2017 edition YANG, Z. 2016a. Numerical prediction of
[Online]. Available: temperature field for cargo containment system
http://www.bp.com/content/dam/bp/pdf/energy- (CCS) of LNG carriers during pre-cooling
economics/energy-outlook-2017/bp-energy- operations. Journal of Natural Gas Science and
outlook-2017.pdf [Accessed 2017]. Engineering, 29, 382-391.
3. CASTILLO, L. & DORAO, C. A. 2013. On the 18. LU, J. S., XU, S., DENG, J. J., WU, W. F., WU,
conceptual design of pre-cooling stage of LNG H. X. & YANG, Z. B. 2016b. Numerical
plants using propane or an ethane/propane prediction of temperature field for cargo
mixture. Energy Conversion and Management, containment system (CCS) of LNG carriers
65, 140-146. during pre-cooling operations. Journal of
4. CHOI, S. W., ROH, J. U., KIM, M. S. & LEE, W. Natural Gas Science and Engineering, 29, 382-
I. 2012. Analysis of two main LNG CCS (cargo 391.
containment system) insulation boxes for leakage

A-246 ©2020: The Royal Institution of Naval Architects

Trans RINA, Vol 162, Part A3, Intl J Maritime Eng, Jul-Sep 2020

19. LU, W. 2012. Study on the pre-cooling process of 33. WILSON, J. J. 1974. An introduction to the
membrane liquid cargo tank of LNG Carriers. marine transportation of bulk LNG and the
Master, Dalian Maritime University. design of LNG carriers. Cryogenics, 14, 115-120.
20. LUO, T. 2011. Study on unsteady heat transfer of 34. WMT LIMITED, U. 2009. Dapeng Moon Cargo
pre-cooling process of Large-Scale LNG Storage Operating Manual. Hong Kong.
Tank. Master, Lanzhou University of Technology. 35. YAN, Y., PFOTENHAUER, J. M., MILLER, F.,
21. MENTER, F. 1994. Two-Equation Eddy-Viscosity NI, Z. & ZHI, X. 2016. Numerical study of heat
Transport Turbulence Model for Engineering transfer characteristics in BOG heat exchanger.
Applications. AIAA Journal, 32. Cryogenics, 80, Part 1, 97-107.
22. MIANA, M., LEGORBURO, R., DíEZ, D. & 36. ZHANG, P. & WU, Z. 2014. Numerical
HWANG, Y. H. 2016. Calculation of Boil-Off simulation of phase change heat transfer of
Rate of Liquefied Natural Gas in Mark III tanks liquid nitrogen based on Fluent. Cryo. ＆
of ship carriers by numerical analysis. Applied Supercond, 42, 26-29.
Thermal Engineering, 93, 279-296. 37. ZHU, K., LI, Y., MA, Y., WANG, L., XIE, F. &
23. PATANKAR, S. V. 1980. Numerical heat transfer WANG, J. 2018a. Experimental study on cool
and fluid flow, London: Hemisphere Publishing down characteristics and thermal stress of
Corporation. cryogenic tank during LN2 filling process.
24. PENG, Q., WU, H., WANG, D. W., HE, Y. J. & Applied Thermal Engineering, 130, 951-961.
CHEN, H. 2019. Numerical simulation of 38. ZHU, K., LI, Y. Z., MA, Y., WANG, J. J., WANG,
aircraft crash on large-scale LNG storage tank. L. & XIE, F. S. 2018b. Influence of filling
Engineering Failure Analysis, 96, 60-79. methods on the cool down performance and
25. QU, Y., NOBA, I., XU, X., PRIVAT, R. & induced thermal stress distribution in cryogenic
JAUBERT, J.-N. 2018. A thermal and tank. Applied Thermal Engineering, 141, 1009-
thermodynamic code for the computation of Boil- 1019.
Off Gas–Industrial applications of LNG carrier.
Cryogenics, 99, 105-113.
26. RAJU, T. B., SENGAR, V. S., JAYARAJ, R. &
KULSHRESTHA, N. 2016. Study of Volatility of
New Ship Building Prices in LNG Shipping.
International Journal of e-Navigation and
Maritime Economy, 5, 61-73.
27. SALEEM, A., FAROOQ, S., KARIMI, I. A. &
BANERJEE, R. 2018a. CFD simulation of a full
scale LNG storage tank. San Diego, the United
States, 1-5 July 2018. Amsterdam: Elsevier.
28. SALEEM, A., FAROOQ, S., KARIMI, I. A. &
BANERJEE, R. 2018b. A CFD simulation study
of boiling mechanism and BOG generation in a
full-scale LNG storage tank. Computers &
Chemical Engineering, 115, 112-120.
29. CHINA CLASSIFICATION SOCIETY. 2014.
《 RULES FOR CONSTRUCTION AND
EQUIPMENT OF SHIPS CARRYING
LIQUEFIED GASES IN BULK 》 . Beijing:
CHINA CLASSIFICATION SOCIETY.
30. VEERSTEG, H. K. & MALALASEKERA, W.
1995. An Introduction to Computational Fluid
Dynamics: The Finite Volume Method. Pearson
Schweiz Ag, 20, 400.
31. WANG, J., LI, Y., WANG, L., ZHU, K., XIE, F.
& LI, C. 2018. Transient modeling of cryogenic
two-phase flow boiling during chill-down
process. Applied Thermal Engineering, 143, 461-
471.
32. WANG, Z., LI, P. & JIN, G. 2010. CFD
simulation for coolant of LNG carrier in pre-
cooling process. Journal of Shanghai Maritime
University, 31, 49-53.

©2020: The Royal Institution of Naval Architects A-247