Anda di halaman 1dari 6

The 7th Asia-Pacific Workshop on Marine Hydrodymics-APHydro2014 Vladivostok, September 9-13, 2014

Computation of Flow and Resistance around LNG Carrier in Deep and Shallow Water
A. Maimun*, F. A. Adnan, A. Y. Sian, Rahimudin, A. Priyanto, Yasser M. Ahmed
Marine Technology Centre, Universiti Teknologi Malaysia,
81310 UTM Johor Bahru, Johor, Malaysia

Many more methods can be found in the literature to correct shallow
water effects. To mention some; Raven (2012), Jiang (2001), and work
Reynolds Averaged Navier-Stokes (RANS) computations are conducted of Blaauw and van der Knaap (1983). However, it must be noted that
with general purpose solver Fluent to examine the effect of resistance these methods were developed based on limited laboratory or field
and viscous free surface flow of Liquefied Natural Gas (LNG) carrier measurements, which do not work best for all types of vessels in all
hull form. Shear-stress transport k turbulence and multiphase volume types of shallow water parameters. Wave breaking, squat, deformation
of fluid (VOF) free surface employed. The test matrix included depth of free surface in strong shallow water influences often introduce
values of H/d from 1.2 to 5 at several speeds ranging within Fn = complex physical interactions and makes shallow water effect cannot be
0.0403 - 0.2215. The resistance characteristics and wave profile of the predicted with sufficient accuracy. Thus, it is useful to examine the
LNG model also investigated. Model tests were conducted in towing shallow water effect with more than one formula and compare the
tank for validation of the computed results. Overall results agree fairly results based on the type of ship and parameters in restricted water.
well with experimental data, reveals the feasibility of RANS method in
practical prediction of LNG resistance characteristics. Numerical calculations have been less common in the past until now.
The Computational Fluid Dynamics (CFD) offers the opportunities for
KEY WORDS: Resistance; shallow water; computational fluid studies on shallow water hydrodynamic problems which can only be
dynamics (CFD); RANS; free surface. studied by experimental means in the past. In addition, inclusion of
shallow water is relatively simple in CFD compared to model test by
INTRODUCTION simply defining a suitable boundary condition at the bottom of the
computational domain. Exploiting on the powerful computational
The remarkable growth of demand for energy by the year has led to facilities, CFD can be used as a cost-effective alternative for model
increasingly larger liquefied natural gas (LNG) carrier. On the other tests.
hand, increasing dimensions of LNG carrier lead to increasingly
influence of restricted fairway to vessel hydrodynamics, and the vessel In this study, the authors performed a series of calculations of flow
increasingly suffer from the shallow water effects in restricted water. around LNG in different water depths. The resistance and viscous free
surface flow of LNG carrier in calm water are examined. The
Some notable changed regarding ship operated in shallow water which calculations were executed by general purpose commercial RANS
has been observed including increased wave making especially at the solver Fluent V14. The calculated values of LNG resistance can be
forward end of ship, increased width of wake, decreased propulsive divided into the frictional and pressure components. They were
efficiency, and loss of speed. The wave pattern generated by the ship compared and presented in the paper. Model tests were also conducted
hull is significantly affected by the water depth and the ships speed to validate the numerical results. This paper is organized as follows;
relative to the critical wave speed. This affect the flow around the ship First, the mathematical model is presented. This is followed by
significantly, hence significantly affect the resistance of ship. descriptions of the numerical and experimental methods. The
computational results are shown in the results section for analysis, and
It is crucial for a naval architect to be able to assess the influence of discussions. Finally, concluding remarks are made in the conclusion
shallow water on the hydrodynamic performance of the vessel during section.
the design process. Thus, a reliable prediction of the shallow water
effect is of high important. A variety of methods are available in NUMERICAL METHODS
literature on estimation of ship resistance in shallow water.
The hydrodynamics in this study are modeled using RANS equations
Perhaps the most widely practice method for estimation of resistance in employing the Finite volume method (FVM). In this method, water is
shallow water correction could be found from Lackenby (1963). assumed incompressible, thus the volume of water entering
Generally, this shallow water resistance estimation method gives a computational cells in vicinity of ships hull will match with an equal
velocity correction in shallow water relative to deep water velocity and volume of water flowing out, leads to continuity equation. Together
was adopted by ITTC (2012) as recommended standard. To use the with Navier-Stokes equations, conservation of momentum of the flow
correction formula, one would first need to determine the deep water can be defined and flow field around ships hull can be characterized.
resistance of the model to use as input. Despite its popularity, the Free surface and wave elevation are solved based on the Volume of
correction method of Lackenby was actually extended from the work of Fluid method. Formulations and details of the numerical methodologies
Schlichting (which was developed based on warship hull forms). in CFD are well documented in many literatures. Thus, only the main
The 7th Asia-Pacific Workshop on Marine Hydrodymics-APHydro2014 Vladivostok, September 9-13, 2014

features of the methodologies described. The LNG model is attached to the computer controlled planar motion
mechanism at the carriage in bare hull and free to heave and pitch. Two
Governing Equation load cells was used and placed at equal distance from the midship of
model to record the longitudinal force. A glimpse of experimental test
The governing equations applied in this study are instantaneous conducted has been shown in Fig. 1.
conservation of mass (continuity equations) and momentum (RANS) for
incompressible turbulence flow. These equations are expressed in
Cartesian tensor form as follows: a
0 (1)



Where ui is the time averaged velocity components in Cartesian

coordinates xi (i = 1, 2, 3), p, and are the static pressure, fluid
density and fluid viscosity respectively, ij is the Kronecker delta and b
is the Reynolds stress, where a
2 !

Turbulence Model

To close the RANS equations, a classical 2-equation eddy viscosity

model, shear-stress transport (SST) model was used, where

! $%! &! '! (4) c

( ( ) $) &) *) ') (5)

In these equations, $%! and $) represents the generation of turbulence

kinetic energy due to mean velocity gradients and generation of . The
effective diffusivity for the SST k model represented by ! and ) .
The term &! represents the dissipation of turbulence kinetic energy and
term &) represents dissipation of . Cross-diffusion term of the model
is given by *) , and '! and ') are user-defined source terms. The SST
model remained as one of the most accurate and reliable turbulence Fig.1. Captive model test in towing tank: a. Model towed in deep water
models for external hydrodynamics (Fonfach and Guedes Soares, 2010). condition after passing the shallow water platform, b. Model test in deep
water condition, c. Model test in shallow water condition
SIMULATION DESIGN AND ANALYSIS METHODS The LNG model ship was manufactured with a scale ratio of 1:112 and
contains open stern-form with buttock flow line and a wide central skeg.
Experiment The principal dimensions of LNG tanker are presented in Table 1 and
the body plan is shown in Fig. 2.
The experimental results discussed have been obtained by captive
model test in the towing tank at Marine Technology Centre of Universiti Table 1 Principal dimension of the ship and model
Teknologi Malaysia (UTM). This facility, with main dimensions of 120
4 2.5 m3, allows execution of captive model test in deep water
conditions. 20 m shallow water platform was installed in the tank during Principal particulars Full scale Model
the test and water was drained from the tank in order to get to the
desired water height to models draught ratio, resulting into two water Length between perpendicular (L) in meters 266 2.375
depth, the deep water (H/d = 25) and shallow water (h/d = 1.3)
conditions could be investigated in a run.
The 7th Asia-Pacific
Pacific Workshop on Marine Hydrodymics
Hydrodymics-APHydro2014 Vladivostok September 9-13, 2014

Beam (B) in meters 41.6 0.371 near the free surface and surrounding ththe ship hull regions as shown in
Fig. 4 to capture the ship waves.
wave Since wall function is used in this
Draft (d) in meters 11.13 0.099 study, the size of first grid point away from ship hull is refined to y+ =
50 for every Froude number with 20 cells within boundary layer to
Block coefficient (Cb) 0.746 0.746 capture the detailed fluid property. The dimensions of the three grids are
summarized in Table 2.

Table 2 Description of grids

No. Grid points , ,.

Grid 1 (fine) 4,103,026 1

Grid 2 (medium) 1,810,370 1.260

Grid 3 (coarse) 857,626 1.587

Fig. 2. Body plan of the LNG carrier a
Computational Domain and Boundary Conditions

Top (no slip)

2.3L 1.3L
1.3L z
1.6 L
Pressure outlet
Ship hull Bottom a
Pressure inlet (no slip) (no slip)
(no slip)

Fig. 3.. Boundary conditions of the computational domain

For the present study, the domain was made up of seven boundaries:
hull surface, flow pressure inlet, flow pressure outlet, top and bottom as
well as two side walls. Schematic diagram indicating the computational
domain is given in Fig. 3. Location of the hull, tank bottom and two Fig.4.. Computational grid topology: aa. Overview of the grid, b. Frontal
side walls are corresponding
onding to the towing tank with smooth walls and view of the ship with refinement on the surface grid around ship
no-slip condition imposed. Hydrostatic pressure at outlet boundary was
calculated as a function of water volume fraction. Calculations were
performed in model scale with standard wall function used. Computational Setup and Numerical Simulation
Grid Generation The SIMPLE-Consistent algorithm is used to couple the momentum and
continuity equations. Pressure Staggering Option (PRESTO!) scheme is
The grid systems are generated by ICEM CFD with structural grid used for pressure interpolation while other terms are discretized using a
approach. Three geometrical similar grids with different grid spacing in second order upwind method. Volume fraction equations solved by
three directions are generated for grid independent study. The three High Resolution Interface Capturing (HRIC) scheme.
scheme Convergence was
systematically varied grids set were created ated with uniform grid assessed by ensuring the drag converge to steady state and residuals
refinement factor, + , -. , 2 where , -. - and , are the grid
scaled by the initial imbalance of equations drop four oorders of
spacing of two successively refined grids with , corresponding to the magnitude.
grid spacing of finest grid.
Viscous flow field around LNG model at Froude number ranging
rangi from
All the grid generated in this study are such that finer grids distributed 0.04 to 0.22 corresponding to 4 knots
knot to 20 knots in full scale are
computed. The simulation program included five identical series of
The 7th Asia-Pacific Workshop on Marine Hydrodymics-APHydro2014 Vladivostok, September 9-13, 2014

experiments at different H/d ratios, namely H/d = 1.2, 1.3, 1.4, 1.6 and 5
to study the effect of shallow water. 0.075 49
29 = 8 = (7)
log 4@ 2 1
Simulations are run on a shared-memory type machine with 4 2 ' 78
processors (3.6 GHz). The computing time for finest grid (Grid-1) used
4 processors required roughly 30 wall clock hours. It must be emphasize The pressure resistance coefficient, CP is calculated as CT CF.
that the ship model is fixed at even keel for all simulations, whereas
experiments are performed with ship model free in heave and pitch

Table 3 Grid Independent Study

Test matrix
Model In the grid independent study, the solution change between fine and
0.194 0.292 0.403 0.51 0.632
speed (m/s) medium grid 21 is mostly below 1% of fine grid solution S1, whereas
Fn 0.0403 0.0604 0.0835 0.1057 0.1309 the solution change between medium and coarse grid 32 is close to 4%
S1. 21 is smaller at approximate 2.5 times 32. Fig. 5 shows the
Rn (10 ) 0.535 0.801 1.108 1.403 1.738 comparison of total resistance from CFD prediction over different grid
size with experiment measurement. Grids for simulation in others water
depth were generated based on the medium grid size, considering its
H/d = 5.0 accuracy and shorter simulation time in compared to the fine grid.
H/d = 1.6 -
H/d = 1.4 -
H/d = 1.3 - 3.00
H/d = 1.2 -
RT [N]

Model Exp
0.75 0.850 0.9 0.950 1.070 Grid-1 (fine)
speed (m/s) 1.50 Grid-2 (medium)
Fn 0.1554 0.1761 0.1865 0.1968 0.2215 Grid-3 (coarse)
Rn (106) 2.062 2.337 2.475 2.612 2.940 0.12 0.14 0.16 0.18 0.20 0.22

H/d = 5.0 Fig. 5. Comparison between experimental and computational RT over

different grid system
H/d = 1.6 - -
H/d = 1.4 - - Resistance Characteristics
H/d = 1.3 - -
Fig. 6 shows the trend of total resistance from CFD prediction on
H/d = 1.2 - - medium grid system and experiment measurement. The CFD results of
the LNG hull in deep and shallow water (H/d = 1.3) are compared
: Simulation & experiment performed, : simulation performed, -: test against the experimental resistance data. A shallow water effect
not performed. correction was made to the experimental data based on correction
method of Lackenby. In general, the agreement is qualitatively
Resistance Characteristics reasonable, but quantitatively, considerable differences are seen at
shallow water H/d = 1.3 at high Froude number.
The total resistance RT is compose of frictional resistance RF and
pressure resistance RP. In experimental method, frictional resistance is In the deep water conditions, the experimental data shows the lowest RT
defined by ITTC-57 ship model correlation line. Sum of wave and at the lowest speed Fn=0.0835. A further increase in speed produces
hydrostatic terms are defined as residual resistance RR and assumed increase in the RT curve and record the largest RT at highest speed
equivalence to pressure resistance, RP = RT RF. In CFD method, the Fn=0.2215. CFD data show an increase from minimum RT near lowest
frictional resistance RF and pressure resistance RP are computed as speed Fn=0.0403 up to the maximum at the highest speed Fn=0.2072.
Compared to the experimental data, the computational results are under
integral of tangential stresses and normal stresses over the hull. Total
predicted from 6.4% to 11.2% of experimental value.
resistance coefficient CT is expressed by non-dimensionalising RT as

In shallow water condition of H/d = 1.3, the experimental and CFD data
show increase from minimum RT near lowest speed Fn=0.0403 up to the
' 78 maximum at the highest speed Fn=0.2215. Compared to the
2 experimental data, the CFD results are over predicted from 6.5% to
79.78% of experimental value. Differences between the experimental
where S is the wetted hull surface area in still water. The frictional
and CFD data may attribute to the possible error during procedure of
resistance coefficient, CF, in accordance with the ITTC-57 formula is
experiment or numerical simulations and the different in running
defined by: attitudes between CFD and experiment where model is fixed and even
The 7th Asia-Pacific Workshop on Marine Hydrodymics-APHydro2014 Vladivostok, September 9-13, 2014

keel in CFD while physical model in towing tank experience free trim
changes and squat.
CFD, H/d = 1.3 3.00
CFD, H/d = 5.0 (deep water)
Lackenby, H/d = 1.3 (exp) 2.50

CP/CF ratio
Exp, H/d = 1.3 CFD, H/d = 1.2
4.0 2.00
Exp (deep water)
CFD, H/d = 1.3
RT [N]

3.0 1.50
CFD, H/d = 1.4
2.0 CFD, H/d = 1.6
CFD, H/d = 5.0
0.02 0.07 0.12 0.17 0.22 0.27
0.02 0.07 0.12 0.17 0.22 Fn
Fig. 8. Comparison of resistance ratio CP/CF in CFD and experiment
Fig. 6. Comparison of resistance between computational and
experimental results (medium grid) Free Surface

Except the shallow water conditions at high Froude number, overall The study of free surface wave profile around the hull is crucial with
trends are encouraging in that the computational results follow the regard to the study of wave resistance. Fig. 9 shows the computed free
trends of the experimental data well. High validation comparison error surface wave elevation for Fn = 0.1761 and at H/d = 1.2, 1.4, 1.6 and 5.
at the highest Froude number in shallow water may indicates the The zones of high pressure at the bow and the stern regions of hull are
difficulties in resistance prediction due to high wave making resistance. detected wall by the CFD. In the H/d = 5, typical Kelvin wave pattern
More mesh elements around the hull are required, which could not be with symmetric and diverging waves are observed. For the shallower
attained due to limitation in computational resources. water depth conditions, larger amplitude of bow wave appears and
drastic reduction of the stern waves occurs compared to deep water
9.0 conditions as water depth decreases.
CFD, H/d = 1.2
8.0 CFD, H/d = 1.3
CFD, H/d = 1.4 This explained the increase of CP/CF ratio with decrease of water depth
7.0 CFD, H/d = 1.6 that caused by the pressure change due to differential change in wave
CFD, H/d = 5.0 profile. The study clearly shows that the effect of pressure to the ship
6.0 Lackenby, H/d = 1.2
Lackenby, H/d = 1.3 hull becomes more pronounced in shallower water depths.
RT [N]

Lackenby, H/d = 1.4

Lackenby, H/d = 1.6 Shallow water conditions are still a challenge for numerical
4.0 Lackenby, H/d = 5.0
computations. For high Froude numbers, interaction of hull with bottom
3.0 leads to a complex flow and the grid used may still have been somewhat
2.0 coarse. Applying finer mesh in certain critical areas would result in
much longer computing times.

0.02 0.07 0.12 0.17 0.22

Fig. 7. Shallow water resistance from CFD and method of Lackenby

The computed RT at various water depth conditions are summarized in

Fig. 7, showing the shallow water resistance from CFD and method of
Lackenby. Comparing CFD and experimental results, the shallow water
resistance by Lackenby gives a relatively higher values for the total
resistance. This result is also mentioned by Raven (2012), showing the
method of Lackenby gave higher overall resistance estimates.

Fig. 8 shows the resistance ratio CP/CF from CFD calculations. For all
the water depth conditions, the pressure components in the resistance
dominate in current LNG model with increases with higher speed. The
CP/CF in this study shows the minimum at Fn=0.0403, followed by
increase with further increase in speed produces a maximum at
Fn=0.2215, which may attribute to the changes in free surface around
the hull. This demonstrated the pressure on the ship hull is more
pronounced for shallower water depths.
The 7th Asia-Pacific Workshop on Marine Hydrodymics-APHydro2014 Vladivostok, September 9-13, 2014

hydrodynamics of a ship navigating in shallow water. With respect to

the shallow water investigations, the shallower depth gave larger total
resistance (RT). The study shows that the increase of RT for shallower
water depth is due to the changes in amplitudes of the bow and stern
waves. Differences between the CFD and experimental data are
expected due to inconsistency in running attitudes between CFD and
experiment. More studies are required to investigate the effect of finer
grid and covering trim and sinkage in CFD for more practical prediction
of resistance characteristics in future.


The presented work is support by Marine Technology Center (MTC)

and Centre for Information and Communication Technology (CICT) in
Universiti Teknologi Malaysia. The authors would like to express their
sincere gratitude to Ministry of Science, Technology and Innovation
(MOSTI) Malaysia and Malaysia Ministry of Education for the financial
support given to the research.


F. van der Knaap H. Blaauw. Prediction of squat of ships sailing in

restricted water. Delft Hydraulic Laboratory, 1983.
Fonfach, J. M. A., Guedes Soares, C., (2010). Improving the Resistance
of a Series 60 Vessel with a CFD Code. Proceedings of V European
Conference on Computational Fluid Dynamics, Lisbon, Portugal.
H. C. Raven.(2012). Towards a new shallow-water correction method
for ship speed-power relations. Technical report, MARIN.
ITTC. (2012).ITTC Recommended procedures and guidelines. Speed
and power trials, Part 2 Analysis of speed/power trial data.
Lackenby, H. (1963). The effect of shallow water on ship speed.
Shipbuilder 70, ( 672)
Menter, F. R. (1994). Two-Equation Eddy-Viscosity Turbulence
Models for Engineering Applications. AIAA Journal. Vol. 32(8), pp.
1598 -1605.
Muzaferija, S. Peric, M., Sames, P., and Schellin. T. (1998). A Two-
Fluid Navier-Stokes Solver to Simulate Water Entry. Proceedings of
22nd Symposium on Naval Hydrodynamics, Washington, DC. pp. 277
T. Jiang. (2001). A new method for resistance and propulsion prediction
of the ship performance in shallow water. Proceedings of 8th PRADS
Symposium, China.
Van Doormaal, J. P. and Raithby, G. D. (1984). Enhancements of the
SIMPLE Method for Predicting Incompressible Fluid Flows.
Numerical Heat Transfer. Vol. 7, pp. 147 -163.
Volker Bertram. (2000). Practical ship hydrodynamics. London:
Butterworth Heinemann.

Fig. 9. Free surface elevations at Fn = 0.1761 and H/d = 5, 1.6, 1.4, 1.2
(from top to bottom)


Reynolds Averaged Navier-Stokes (RANS) computations with general

purpose solver Fluent to examine the effect of resistance and viscous
free surface flow of Liquefied Natural Gas (LNG) carrier hull form are
presented. The numerical studies in this paper gave a deeper insight into