Maritime Technology and Engineering 3 – Guedes Soares & Santos (Eds)

© 2016 Taylor & Francis Group, London, ISBN 978-1-138-03000-8

CFD assessment of Ropax hull resistance with various

initial drafts and trim angles

J. Labanti, H. Islam & C. Guedes Soares

Centre for Marine Technology and Ocean Engineering (CENTEC), Instituto Superior Técnico,
Universidade de Lisboa, Lisbon, Portugal

ABSTRACT: A Ropax hull resistance has been studied with OpenFOAM considering different Froude num-
bers, trim angles and drafts. The simulations have been set in calm water, model scale, using a multiphase domain
implemented with a volume-of-fluid method and a SST k-ω turbulence model. The first simulation has been
run even keel at design draft, Fr = 0.2, both with the hull fixed in its original position and with the hull free to
sink and trim. Comparison of results with another RaNS code has been also performed. Next, systematic data
collection has been done with the hull fixed in various particular positions, at Fr = 0.20. Similar simulations at
higher Fr numbers have been run; in these cases a coarser mesh has been used in order to achieve a less request
of computational power and a more stable behavior.

1 INTRODUCTION formulas; simulations at Fr = 0.2 have shown conver-

gence whereas the ones at higher Fr numbers have
In order to predict the ship’s resistance and, as a con- shown problems, producing wave form the inlet. This
sequence, the ship’s consumption in an early stage of problem is also reported in Ciortan et al (2007) and has
design, CFD has become an increasingly reliable and been partially avoided using different solver, changing
powerful tool. Several studies about using this tool for pressure boundary conditions at the outlet and insert-
ship hull optimization have been done, as described by ing coarser mesh zones near the sides and at the outlet
Percival et al (2001), Campana et al (2006) and Peri like pointed by Yang et al (2011); however, further
et al (2001). investigations in this field are required.
Furthermore, ship resistance is not only related to Even though there aren’t towing tank data as bench-
the shape of the hull, but also depends on dynamic mark, comparison with a well validated RaNS code
effects like sinkage and trim developed during navi- has been done and data from numerical simula-
gation. It follows the importance of predicting prop- tions reported here represent nevertheless interesting
erly these motions: a possible basic CFD setup to insights about trim and draft effects.
study the problem is explained by Yang et al (2000) Firstly, an even keel resistance evaluation has been
whereas Carrica et al (2011) go further implementing performed to test the software; insights by Maki
a self-propelled model free to sink and trim. Similar (2011), Maki et al (2013) and Ahmed et al (2009)
investigations have been done by Wortley (2013) on a have been followed. Then, after an initial comparison
container ship and by Tarbiat et al (2014) on the Wigley between the even keel resistance calculated with the
hull, and they have been the references for the work hull fixed in its original position and the hull free to
presented here. sink and trim, likewise Hajivand and Mousavizadegan
Particularly, the Ropax object of these paper is a (2015), the present paper shows the results of the resis-
kind of ship where resistance shifts considerably due tance calculation at design draft with different initial
to dynamic effects combined with loading conditions trims, Fr = 0.2, using a static mesh.
as shown by data recorded on operating ship by Galli The considered trims are weighted average of the
et al (2014). most common angles a Ropax can reach due to
In order to reproduce numerically these situations, dynamic effects combined with loading conditions.
OpenFOAM has been used. It is a C++ open source Trim angles are included in a range from 0.4◦ by
software where a Reynolds Average Navier Stokes set stern to 0.2◦ by bow, step 0.2◦ . Furthermore, the
of equations has been implemented in the finite vol- resistance’s dependence from draft has been studied.
ume scheme, plus the two equations of the SST k-ω Simulations even keel with different drafts have been
turbulence model to close numerically the problem. run: design draft, minimum draft allowed and interme-
ITTC guidelines (2011) and Ciortan et al (2012) have diate draft between the former two values have been
been followed for the model setup. Turbulence param- considered. At higher Fr numbers the same investi-
eters have been calculated using the most common gations of trim and draft showed problems and the

325
Table 1. Ship data, model scale.

Ship scale Model scale

Lpp , m 90 3.600
Bwl , m 17.82 0.713
Tdesign, m 4 0.168
D, m 14.8 0.592 Figure 1. Reference system.
∇, m3 3725 0.238
S, m2 1743 2.788
Cb 0.55 0.55 scale simulations have been considered a reasonable
Awl, m2 1311 2.097 approximation.
xG, m −3.5 −0.14 All forces and moments are referred to a system
GML, m 170 10.152 parallel to the reference one but centered in G; that’s
irrelevant for the resistance calculation but it counts
for the moment around y-axis.
setup able to avoid them does not produce reliable
results. 3 MODEL SETUP
The final purpose of this paper is describing how
this open source software can be used in an early design The paper mainly focuses on OpenFOAM to predict
stage, obtaining a rough idea of the ship behavior in trim effect on ship resistance. Given that no experi-
terms of relationship between resistance, pitch angles mental data was available for proper validations, the
and drafts with no license costs and less request of initial results have been compared to simulation results
computational power. produced by another RaNS solver, SHIP_Motion. The
mathematical model of the solver has been elaborately
discussed by Orihara et al. (2003), and Kim et el.
2 SHIP DATA AND REFERENCE SYSTEM (2015) in their works, respectively. The solver has been
validated for several ship models previously.
The Ropax object of the simulation is a monohull
whose main data are reported in Table 1.
3.1 OpenFOAM
It is fitted with skeg and has no bulbous bow. It also
has two four-bladed counter-rotating (inwards) pro- The domain size for OpenFOAM simulation has been
pellers and it’s equipped with two Becker flap spade set following ITTC (2011) guidelines and the work by
rudders. Shi et al (2012): the inlet has been placed one ship
For the resistance computation only bare hull and length windward the bow, the outlet five Lpp down-
skeg were considered. Furthermore, only few data of stream the stern; each lateral boundary is two ship
this ship were available; the hydrostatic features such lengths away from the ship’s symmetry plane.
as the position of the centre of buoyancy (B) and The bottom has been placed at a sufficient distance
the wet surface have been calculated from the IGES from the free surface in order to avoid interactions
model, whereas the centre of gravity position (G) is with the generated wave pattern and reproduce a deep
unknown. Due to lack of data, G has been set in the water condition: one ship length is enough to achieve
same longitudinal position as B in order to have equi- it. The atmosphere has been placed at 0.5 L above the
librium at pitch. Ship is upright in every simulation. free surface.
G height has been set at D/2, but the vertical posi- Mesh has been generated using snappyHexMesh
tion of G is useful only when dealing with dynamic utility, which creates automatically a “body fitted”
mesh to evaluate the constraint in pitch (C55 in lin- hexahedral mesh from an STL surface, following
earized equation of motion). C55 is function of GML, suggestion by Jackson (2012).
but KML is much more higher than KG, so it’s been A mesh refinement across the free surface has
considered a reasonable approximation. been performed in order to have at least 40 cells per
The origin of the reference system used in Open- expected wavelength. The height of the first cell inside
FOAM to construct the domain and calculate the G the expected boundary layer has been estimated for
and B position is placed as follows: y+ = 1; the max cell height has been calculated instead
for y+ = 50 and corresponds approximatively to the
– Ox: PP1/2, x positive forward
beginning of the logarithmic region.
– Oy: symmetry plane, y positive portside
In order to damp the outgoing waves and, hope-
– Oz: waterplane, z positive upward
fully, prevent making waves from the inlet a sponge
Simulations have been run in model scale (1:25) layer zone should have been necessary as pointed by
to reduce the required computational power. This, Yang et al (2011) and Hu et al (2015): however, only a
according to Jin et al (2016) leads to consistent errors coarser mesh near the sides of the block has been suffi-
when calculating hydrodynamic manoeuvering coef- cient instead of proper “numerical beaches” based on
ficients (10% to 30%) whilst scale effects lead to relaxation method. Thanks to this artificial damping,
errors below 8% for surge force; for this reason model incoming waves have been avoided for Fr = 0.2.

326
 Table 3. Physical constants.

Water Air

ρ, kg/m3 1000 1
ν, m2 /s 1.09e−06 1.48e−05

Table 4. Boundary conditions.

Figure 2. Hull mesh. Inlet Outlet Atmosphere Hull

Table 2. Mesh quality. U FV ZG PIOV MWV

p_rgh FFP ZG TP FFP
Fr 0.20 0.25/0.30 α.water FV ZG IO ZG
k FV ZG IO kqRWF
Points 1.4 mln 800 k nut FV ZG ZG nutkRWF
Faces 4 mln 226 k omega FV IO IO omegaWF
Cells 1.3 mln 223 k
Hexahedra 1.25 mln 730 k
Prisms 2k 700
Polyhedra 58 k 36 k
k is the turbulent kinetic energy per unit mass; δ is
the height of the boundary layer; l is the turbulence
length scale, describing the size of the large energy-
containing eddies in a turbulent flow; Cµ is an empir-
This problem is still present for higher Fr numbers
ical constant; ω is the turbulence specific dissipation
considered here, so a different set up has been imple-
rate; and ν T is the turbulent kinetic eddy viscosity.
mented. The pressure boundary condition at the outlet
The density and the kinematic viscosity of the fluids
has been changed in order to force the wave damping,
used in simulations are reported in Table 3.
and interFoam instead of LTSInterFoam has been used
due to its more stable behaviour. This set up has less 3.1.2 Boundary conditions
physical sense and the results seem to be less reliable. The control volume represents a deep water condition,
A coarser mesh has been used to save computational so the two lateral sides and the bottom are symmetry
time. plane type faces; no additional information is required
for this kind of boundary condition. Inlet, outlet and
3.1.1 Turbulence parameters
atmosphere are patch faces with specific boundary
Turbulence has been modelled with a Reynolds-
condition for each one, whilst hull is a wall type
averaged stress (RAS) SST k-ω two equation model.
one. Boundary condition setting is based on Wortley
The parameters have been calculated following the
(2013), Mordhorst (201), Yang et al (2011). For the
most common guidelines as follows:
Fr = 0.2 case boundary conditions are as follow:
Where:
– FV is fixedValue, specified by the user
– ZG is zeroGradient
– FFP is fixedFluxPressure, that adjusts the pressure
gradient such that the flux on the boundary is that
one specified by the velocity boundary condition
– PIOV is pressureInletOutletVelocity, that applies
zero-gradient for outflow, whilst inflow velocity is
the patch-face normal component of the internal-
cell value.
– TP is totalPressure, calculated as static pressure ref-
erence plus the dynamic component due to velocity
– IO is inletOutlet, that provides a zero-gradient
outflow condition for a fixed value inflow
– MWV is movingWallVelocity. This condition has
been placed for dynamic mesh, but it also works
with static mesh (in this case the velocity is simply
zero)
– WF is wallFunction, and each turbulence parameter
has its own.
where I is the turbulence intensity, analytically defined
as the ratio between the root-mean-square of the tur- For the higher Fr numbers, p_rgh outlet bound-
bulent velocity fluctuations and the mean velocity; ary condition (zeroGradient) has been replaced with

327
Table 5. Internal fields initial values. Table 6. Simulation setup used in SHIP_Motion.

IF type IF value unit Fr 0.20

U, m/s 5.9427
U uniform −U m/s Re 4.907e08
p_rgh uniform 0 Kg/ms2 DOF Heave and pitch
α.water uniform (setFields) 0/1 −
k uniform (2) m2 /s2
nut uniform (7) m2 /s Table 7. Mesh resolution used in SHIP_Motion.
omega uniform (6) 1/s
Inner mesh Outer Mesh

Length in x-direction 1.85 L (171) 3.8 L (192)

(n◦ of cells)
Length in y-direction 0.30 L (27) 1 L (45)
(n◦ of cells)
Length in z-direction 0.30 L (66) 1.2 L (63)
(n◦ of cells)

*L = Ship length.

Figure 3. Overset mesh structure for SHIP_Motion.

fixedValue (0) in order to force the damping of the

Kelvin pattern.

3.2 SHIP_Motion model

For running simulation using SHIP_Motion, an over-
set structured mesh system has been used (Figure 3). Figure 4. Residuals checking, Fr = 0.2, LTSInterFoam.
The coarse rectangular outer mesh with high resolu-
tion around the free surface has been used to capture
possible to enable the simulation to reach steady-state
the free surface deformation. The fine O-H type inner
quickly. LTSInterFoam first maximizes the time-step
mesh around the hull surface has been used for cap-
according to the local Courant number, then processes
turing the flow properties around the hull surface.
the time-step field by smoothing the variation in time
The mesh has been generated using commercial mesh
step across the domain to prevent instability due to
generation tool Pointwise.
large conservation errors caused by sudden changes
SHIP_Motion being a dedicated solver for ship
in time step. This approach works very well for low
hydrodynamic simulation requires limited input data
Fr whereas for higher Fr numbers contributes to the
for running simulation. Thus, most of its boundary
problem explained in section 3 and leads to divergence.
conditions and solver modeling are pre-defined. The
In the static hull case, the expected dynamic sinkage
solver just requires basic geometric information of the
and trim have been calculated from the global z-force
ship model and the applied Reynolds and Froude num-
and y-moment acting on hull with following formulas,
ber. All simulations are ran in non-dimensional scale.
under the hypothesis that the ship’s sides are vertical
The simulation conditions and mesh resolution are in
and displacements are small:
Table 6 and Table 7 respectively.

4 SOLVERS

4.1 OpenFOAM
All considered solvers (interFoam, LTSinterFoam and
interDyMFoam) work with the PIMPLE algorithm.
The first simulation has been run even keel both with where Fz is the total force in z direction around the hull;
a static hull and with a dynamic hull able to translate My total moment around y-axis; Iy is the moment of
along z-axis and rotate around y-axis. For the static inertia of the waterplane around y-axis.Trim is positive
hull LTSInterFoam has been used, where LTS means by bow.
local-time stepping: it manipulates the time step for For the dynamic mesh case interDyMFoam has been
each individual cell in the mesh, making it as high as used. The area interested by the mesh movement has

328
 Table 8. Results comparison between the two solvers.

Static Dynamic Dynamic

mesh mesh mesh
LTSInterFoam interDyMFoam SHIP_Motion

Fr 0.20 0.20 0.20

U, m/s 1.1885 1.1885 5.9427
Time step Fixed (1e−3) Adjustable Fixed (1e−4)
Cells 1328903 1328903 849042
Iterations 15000 65000 50000
CT 3.631e−03 3.634e−03 3.370e−03
Sinkage, % 0.60 1.5 2.16
Figure 5. Resistance time history, Fr = 0.2, LTSInterFoam. Trim, deg 0.104 0.109 0.12

been incorporated, Baldwin-Lomax and Dynamic sub-

grid scale model. Wall function has been used to
reduce dependency on mesh for capturing boundary
layer properties. Parallel processing is by the shared
memory model of OpenMP.
A Marker and Cell (MAC) type pressure solu-
tion algorithm has been employed. The pressure is
obtained by solving the Poisson equations using the
SOR method and velocity components are gained by
correcting the velocity predictor with the implicitly
Figure 6. Trim time history, Fr = 0.2, interDyMFoam. evaluated pressure. In the overlapping grid system,
inner domain moves according to floating body’s
been set as a vertical cylinder of one Lpp radius cen- equation of motion and outer domain represents free
tred in the origin of the reference system. Hull’s centre surface. Grid points located at the overlapping region
of mass, mass moments of inertia and constraints have exchange information through interpolation to update
also been specified in the relative dynamic mesh dic- both the domains at every time step.
tionary. Dynamic trim and sinkage have been checked The solver runs simulation in non-dimensional
run-time from the log of the simulation. For both the scale and the results are later converted to dimen-
solvers the maximum time step has been set as func- sional scale. For calm water simulation, only half-hull
tion of the flow velocity (0.001·L/U) and the maximum has been considered, but following results presented
Courant number has been set as 0.9. here are referred to the full hull for consistence with
The simulations have shown convergence that OpenFOAM data.
has been checked run-time with the x-force and
the dynamic pressure residuals histories (Figure 4).
To achieve convergence, about 800 CPU hours are 5 RESULTS AT FR = 0.20
required, equal around 25 seconds of simulation;
more time is needed in case of interDyMFoam, as The first simulation has been run at design draft,
seen in Figure 6, due to oscillations around dynamic even keel and Fr = 0.2 to compare the behavior of
equilibrium position. the two solvers of OpenFOAM and the other RaNS
code. Resistance and sinkage are presented here in a
non-dimensional form as follows:
4.2 SHIP_Motion
The governing equation for the mathematical model of
SHIP_Motion is the Reynolds averaged Navier-Stokes
(RaNS) equation and continuity equation. Two sets of
coordinate system, body fixed and earth fixed, have
been used. The spatial discretization is by Finite Vol-
ume Method (FVM). 3rd order upwind differencing
has been used for advection, whereas, discretization The resistance calculated by each solver are compa-
in space is by 2nd order central difference scheme. rable, but as seen in the figures above LTSInterFoam
Definition of flow variables are in staggered manner. reach convergence quickly: 15000 iterations against
Free surface capturing is by Marker density method, the 65000 required by interDyMFoam. This happens
where, 3rd order upwind scheme performs space dif- because the dynamic mesh, during its run, fluctuates
ferencing and 2nd order Adams-Bashforth solves time around its equilibrium position, as shown both from the
differencing. Two type of turbulence models have x-force (Figure 6) and trim (Figure 7). At the end of the

329
Figure 7. Trim time history, Fr = 0.2, interDyMFoam. Figure 8. CT against initial trim.

Table 9. CT as function of Fr and initial trim. Table 10. CT as function of Fr and draft.

Initial trim deg Fr = 0.20 Draft m Fr = 0.20

−0.4 3.905e−03 0.12 (min) 4.018e−03

−0.2 3.742e−03 0.144 (mid) 3.924e−03
0.0 3.631e−03 0.168 (design) 3.631e−03
0.2 3.750e−03

Table 11. Wet surface and longitudinal position of G for

runs, the CT difference between the two approaches is every draft.
about 0.08% and the difference with SHIP_Motion is
below 7%. However, the static mesh approach doesn’t Draft m S m2 xG m
allow the proper evaluation of the expected dynamic
effects due to the flow: sinkage calculated form the 0.12 (min) 2.134 −0.060
0.144 (mid) 2.440 −0.097
global z-force in the static mesh method is less than the
0.168 (design) 2.788 −0.140
half of the one obtained by the dynamic mesh approach
as well as from SHIP_Motion.
Anyhow, CFD in general is not very accurate in pre-
dicting sinkage, as it is highly mesh dependent. Thus,
although there is difference between predictions of dif-
ferent solvers, the results give a rough idea regarding
the sinkage of the ship.
Trim obtained by the two OpenFOAM solvers are
very close, with the one obtained by SHIP_motion
being slightly lower with a difference of 10%. Over-
all, since, results from both the OpenFOAM solvers
and SHIP_motion are in good agreement, it may be
concluded that the results predicted here are reliable.

5.1 Trim effects Figure 9. CT against initial draft.

The initial trim angles have been chosen as reasonable
angles, the ship can have due to its loading conditions
and dynamic effects during navigation. For each con- 5.2 Draft effects
dition the wet surface has been considered the same as The minimum draft allowed corresponds to the com-
the even keel one. plete immersion of the propeller disc; the intermediate
A second order dependence of the resistance from draft is an average between this and the design one,
the trim has been observed, with the minimum value considered as the maximum allowed.
for the even keel situation. Polynomial fit has been Remember that in this case, against the trim one, the
constructed for CT dependence from trim, as shown wet surface changes significantly for every draft so the
in Eq. (12). S used to obtain the non-dimensional form changes for
each case.
In following plot drafts are reported as fraction of
design draft (considered as maximum).
A second order dependence of the resistance from
where θ is trim in degrees, positive by bow. the draft has been observed, with the minimum value

330
 Table 12. Setup comparison, static mesh.

Static mesh Static mesh

LTSInterFoam interFoam

Fr 0.20 0.20
U, m/s 1.1885 1.1885
Time step Fixed (1e−3) Adjustable
Total cells 1328903 733974
P_rgh BC (outlet) zeroGradient fixedValue (0)
Iterations 15000 8000
Figure 10. Wave pattern that appears at higher Fr numbers
CT 3.631e−03 4.462e−03
(amplified for illustration purposes).
Sinkage, % 0.60 2.02
Trim, deg 0.104 0.075
for the design condition. Polynomial fit has been con-
structed for CT dependence from draft, as shown in
Eq. (13).
Table 13. Setup comparison, dyamic mesh.

Dynamic mesh Dynamic mesh

interDyMFoam interDyMFoam

Fr 0.20 0.20
where T is a fraction of design draft (0 is design draft, U, m/s 1.1885 1.1885
−0.5 is half of design draft). Time step Adjustable Adjustable
Total cells 1328903 733974
P_rgh BC (outlet) zeroGradient fixedValue (0)
Iterations 65000 10000
6 RESULTS AT HIGHER FR NUMBERS CT 3.634e−03 6.331e−03
Sinkage, % 1.45 1.88
As pointed before, the model set up for Fr = 0.20 does Trim, deg −0.109 −0.073
not work for higher Fr numbers. This is supposed to
happen due to the fact that the wave pattern not dis-
sipates completely before the outlet, reflecting and
producing waves from the inlet.A much longer domain
would be required, but this implies also a substantial Table 13, so even if simulations have shown conver-
increase of cells number that at this stage is being gence results haven’t been considered reliable. Further
avoided. investigation in this field is required.
Simulation without the ship has been performed in
order to check if wave from the inlet are related only to
the Kelvin pattern or they’re caused by other reasons. 7 CONCLUSIONS
It’s been checked that, at higher fluid velocity, waves
are present as well (Figure 10), so the problem is sup- Bare hull resistance of a ropax has been calculated
posed to be related to the outlet boundary conditions with numerical simulation run in OpenFOAM. Var-
or the solution schemes. ious Fr numbers, trim angles and drafts have been
The problem has been avoided forcing a dynamic considered. Due to lack of experimental data, initial
pressure boundary condition at the outlet, setting it to comparison with another well validated RaNS code
zero. That is equal to say that all the flow perturbations has been performed in order to check the goodness of
are exhausted when they reach the outlet, and it is not the results.
physically true. However, this avoids the creation of As first step, has been observed that the even keel
the wave pattern of Figure 10 and allows simulations resistance calculation in OpenFOAM is more reliable
reach convergence. if done with a dynamic mesh able to sink and trim.
The implicit incorrectness of this model set up leads Using interDyMFoam handling a dynamic mesh is rec-
to results not very reliable. This model set up has ommended for resistance evaluation at design point
been used to repeat runs at Fr = 0.2 and the results and, as well, in design steps that require more preci-
have been compared to those shown in section 5. Due sion. However, much time and computational power
to the implicit incorrectness of this new model set is needed to deal with dynamic mesh, so in an early
up, a coarse mesh has been used in order to save stage a static mesh approach can be used to evaluate
computational time. hull resistance as well as insights regarding the depen-
Refinement across the free surface has been per- dence of resistance by fixed trim angles and drafts.
formed, but at a lower level than that one performed The LTSInterFoam solver is able to reach convergence
before; features of this new mesh are reported in quickly on a mid-refined mesh, so it’s a useful tool for
Table 2. Comparison shows significant differences this kind of investigations. Influence of trim and draft
between the two approaches, as pointed in Table 12 and at Fr = 0.2 has been found, evaluating also a regression

331
just as pointed in sections 5.1 and 5.2. The same analy- ITTC, 2011. Pratical guidelines for ship CFD applications.
sis at higher Fr number have not been run successfully In:Recommended procedure and Guidelines.
and further investigation will be necessary. Jackson, A., 2012. A comprehensive tour of snappyHexMesh.
7th OpenFOAM workshop, Darmstadt.
Jin, Y., Duffy, J., Chai, S., Chin, C., Bose, N., 2016. URANS
study of scale effects on hydrodynamic maneuvering coef-
ACKNOWLEDGMENTS ficients of KVLCC2. Ocean Engineering 118: 93–106.
Kim, H., Akimoto, H., Islam, H., 2015. Estimation of
The work is a part of the SHOPERA (Energy Efficient the hydrodynamic derivatives by RaNS simulation of
Safe SHip OPERAtion) collaborative project which is planar motion mechanism test. Ocean Engineering 108:
co-funded by the Research DG of the European Com- 129–139.
Löhner, R., Yang, C., Oñeto, E., Idelssohn, S., 1999. An
mission within the RTD activities of the FP7 Thematic
unstructured grid-based, parallel free surface solver.
Priority Transport/ FP7-SST-2013-RTD-1. Applied Numerical Mathematics 31: 274–293.
Maki, K., 2011. Ship resistance simulation with Open-
FOAM. 6th OpenFOAM workshop. The Pennsylvania
REFERENCES State University, USA.
Maki, K., Broglia, R., Doctors, L.J., Di Mascio, A., 2013.
Ahmed, Y. and Guedes Soares, C. Simulation of Free Surface Numerical investigation of the components of calm-water
Flow around a VLCC Hull using Viscous and Poten- resistance of a surface-effect ship. Ocean Engineering 72:
tial Flow Methods. Ocean Engineering. 2009; 36(9–10): 375–385.
691–696. Mordhorst, C.J., 2011. Investigation of open-source CFD
Campana, E.F., Peri, D., Tahara, Y., Stern, F., 2006. Shape software on shipyards. Thesis, Chalmers University of
optimization in ship hydrodynamics using computational technology.
fluid dynamics. Computer methods in applied mechanics Orihara, H., Miyata, H., 2003. Evaluation of added resistance
and engineering 196: 634–651. in regular incident waves by computational fluid dynam-
Carrica, P.M., Fu, H., Stern, F., 2011. Computations of ics motion simulation using an overlapping grid system.
self-propulsion free to sink and trim and of motions in Journal of Marine Science and Technol (2003) 8: 47–60.
head waves of the KRISO Container Ship (KCS) model. Prever, R., Grabert, R., 2004. Improving fuel efficiency in
Applied Ocean Research 33: 309–320. Ro-Pax design. RoRo 2004 Exhibition and Conference –
Ciortan, C., Wanderley, J., Guedes Soares, C., 2007.Turbulent the International RoRo Event from Ship to Shore,
free-surface flow around a Wigley hull using the slightly Goeteborg.
compressible flow formulation. Ocean Engineering 34: Shi, A., Wu, M., Yang, B., Wang, X., Wang, Z., 2012. Resis-
1383–1392. tance calculation and motions simulation for free surface
Ciortan, C., Wanderley, J.B.V., Guedes Soares, C., 2012. Free ship based on CFD. Procedia Engineering 31: 68–74.
surface flow around a ship model using an interface- Percival, S., Hendrix, D., Noblesse, F., 2001. Hydrodynamic
capturing method. Ocean Engineering 44: 57–67. optimization of ship hull forms. Applied Ocean Research
Galli, A.M., Gualeni, P., Stranieri, G., Qualich, S., Cusano, 23: 337–355.
G., 2014. Monitoring and analysis of the performance data Peri, M., Rossetti, M., Campana, E.F., 2001. Design opti-
of a RO-PAX ship in the perspective of energy efficiency. mization of ship hulls via cfd techniques. Journal of ship
Polish Maritime Research 4(84) Vol. 21: 18–26. research 45: 140–149.
Haack, T., Krüger, S., Vorhölter, H., 2009. Optimization of Tarbiat, S., Lavrov., A., Guedes Soares, C., 2014. Numerical
a fast monohull with CFD-methods. 10th International simulation of the free surface turbulent flow of a Wigley
Conference on Fast Sea Transportation (FAST), Athens. hull with trim and drift angle. Guedes Soares, C. & Santos
Hajivand, A., Mousavizadegan, S.H., 2015. Virtual maneu- T.A. (Eds.). Maritime Technology and Engineering. Taylor
vering test in CFD media in presence of free surface. Int. & Francis Group, UK: 1009–1018.
J. Nav. Ocean. Eng. 7: 540–558. Wooliscroft, M.O., Maki, K.J., 2016. A fast-running CFD
Higuera, P., Lara, J.L., Losada, I.J., 2013. Realistic wave formulation for unsteady ship maneuvering performance
generation and active wave absorption for Navier-Stokes prediction. Ocean Engineering 117: 154–162.
models application to OpenFOAM. Coastal Engineering Wortley, S., 2013. CFD analysis of container ship sinkage,
71: 102–118. trim and resistance. Thesis, Curtain University.
Hu, Z., Tang, W., Xue, H., Zhang, X., Guo, J., 2015. Numeri- Yang, C., Löhner, R., Noblesse, F., Huang, T.T., 2000. Calcu-
cal simulations using conserved wave absorption applied lation of ship sinkage and trim using unstructured grids.
to Navier-Stokes equation model. Coastal engineering European Congress on Computational Methods inApplied
99: 15–25. Sciences and Engineering (ECCOMAS 2000), Barcelona.
Islam, H., Akimoto, H., 2015. Prediction of ship resistance in Yang, C., Huang, F., Wang, L., 2011. Numerical simulations
Head Waves Using RaNS based solver. 13th International of highly nonlinear steady and unsteady free surface flows.
Conference on Mechanical Engineering (ICME), BUET. Journal of Hydrodynamics 23(6): 683–696.

332