Calm Water Resistance Prediction of A Container Ship Using Reynolds Averaged Navier Stokes Based Solver

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 194 (2017) 25 – 30

10th International Conference on Marine Technology, MARTEC 2016

Calm Water Resistance Prediction of a Container Ship using


Reynolds averaged Navier-Stokes based Solver
Hafizul Islama , Md. Mashiur Rahamanb , Hiromichi Akimotoc , M. Rafiqul Islamb
a CENTEC, Instituto Superior Tecnico, Tecnico Lisboa, Portugal
b Department of Naval Architecture and Marine Engineering, BUET,Dhaka-1000,Bangladesh
c Center for the Advancement of Research and Education Exchange Network in Asia (CAREN), Graduate School of Engineering, Osaka

University, 2-1 Yamadaoka, Suita, Osaka, Japan 565-0871

Abstract
Maneuverability and resistance prediction with suitable accuracy is essential for optimum ship design and propulsion power pre-
diction. This paper aims at providing some of the resistance and maneuverability characteristics of a container ship model, MOERI
KCS in calm water using a computational fluid dynamics solver named Ship Motion. The solver is based on the Reynolds average
Navier-Stokes method (RaNS) and solves overset-structured grid using the Finite Volume Method (FVM). This paper comprises
the numerical results of calm water test for the KCS model with available experimental results. The calm water test results include
the total drag coefficient, average sinkage, and trim data. Visualization data for pressure distribution on the hull surface and free
water surface have also been included. The paper concludes that the present solver has the capability to predict the resistance and
maneuverability characteristics of the container ship with reasonable accuracy utilizing minimum computational resources.

©c 2017
2017TheThe Authors.
Authors. Published
Published by Elsevier
by Elsevier Ltd.is an open access article under the CC BY-NC-ND license
Ltd. This
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 10th International Conference on Marine Technology.
Peer-review under responsibility of the organizing committee of the 10th International Conference on Marine Technology.
Keywords: container ship; CFD; calm water; RaNS; resistance
1. Introduction
Prediction of ships maneuverability characteristics and resistance is essential at early phase of ship design to un-
derstand its design efficiency. Although model tests are most popular for such predictions, in recent years, CFD has
gained high popularity for such predictions because of its high efficiency and economics. CFD is very efficient be-
cause it allows exploration of new engineering design frontiers in a very economical way. Furthermore, over the years,
the accuracy of CFD prediction has improved significantly. Thus, more and more research is being done on how to
predict the stability and maneuvering capabilities of a ship in the design phase using CFD.
Maneuverability prediction in the design phase is nothing new for the shipping industries. According to reports of
20th 24th IITC[1–3], the first symposium on Naval Maneuverability was held in Washington DC in 1960. However,
back then work was mostly experiment based. Numerical prediction for maneuverability using the boundary layer
method was first introduced in 1980. However, it failed to predict the pressure distribution in the stern part properly.
This had led to the development of various RaNS models throughout the 1980s. In 1990, it was found that RaNS

∗ Corresponding author. Tel.: +88-02-9673605 ; fax: +88-02-8613046.


E-mail address: mashiurrahaman@name.buet.ac.bd

1877-7058 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 10th International Conference on Marine Technology.
doi:10.1016/j.proeng.2017.08.112
26 Hafizul Islam et al. / Procedia Engineering 194 (2017) 25 – 30

models showed high promise but failed to predict detail shape of velocity contours in central part of wake. Later in
1993, Deng et al.[3] revealed that the problem was in inadequate turbulence modeling. The 1994 Tokyo workshop[3]
brought a breakthrough in free surface calculation using RaNS. For The Gothenburg 2000 workshop[4], three new
ship models- KCS, KVLCC and DTMB 5415 were introduced with experimental results for validation. The self-
propulsion system included CFD models were also introduced in this workshop. In The SIMMAN 2008 workshop[1],
benchmarking was conducted for both EFD based and CFD based methods. The Gothenburg 2010 workshop[4]
discussed global and local flow variables, grid dependency, and turbulence modeling.
Over the years, CFD in maneuverability prediction has gone through and is still going through continuous im-
provement. However, with increase in prediction accuracy, computational time has increased as well, despite the fact
that, computing capacity has increased significantly over the years. This is mainly because of our tendency to sim-
ulate exact real life conditions. The requirement of high computational resources has driven us to high performance
computers and reduced the economy of CFD. The solver used for this paper uses the RaNS equation and the finite
volume method with combined use of the Baldwin-Lomax and Dynamic Sub-grid scale turbulence model. This paper
contains validation data for calm water test results of a KCS model using the solver.

2. Method

Detailed numerical modeling of the solver has been discussed by several other previous publications by Orihara[5],
Akimoto[6], Kim[7] and Islam[8]. Therefore, only a short description has been provided in this paper.
2.1. Numerical Modeling
The developed solver, SHIP Motion, uses the Reynolds averaged Navier-Stokes (RaNS) equation and the continu-
ity equation as the governing equations and can solve motions with 6 degrees of freedom (DOF). Interface modeling
is done using the marker density function method. For turbulence modeling, it uses the Baldwin-Lomax and dynamic
sub-grid model. The weighted average of the two models is used for local turbulence viscosity. Physical values are
defined in the staggered manner in control volume, i.e., pressure is defined at the volume center, whereas velocity
components are defined at the face centers of the primary hexahedral control volume. Tri-linear interpolation of flow
variables is done between two domains.
2.2. Discretization Method
SHIP Motion uses the finite volume method (FVM) for special discretization. The numerical discretization is
performed using central difference and upwind scheme. SHIP Motion uses 3rd order upwind scheme for differencing
of advection and 2nd order central differencing for other discretization in space. Temporal discretization is performed
by Adams-Bashforth 2nd order explicit method.
2.3. Meshing
In this research, an overset structured single block mesh system is used for simulation. The coarse rectangular
outer mesh with high resolution around the free surface is used to capture the free surface deformation. The fine O-H
type inner mesh around the hull surface is used for capturing the flow properties around the hull surface. x-z plane
symmetry condition is used in head wave and calm water condition. Fig.1 below shows the mesh applied.

3. Ship model and particulars


The model ship used for this research is the KRISO Container ship (KCS). It is a 3600TEU capacity container
ship designed by KRISO (formerly MOERI) for research purpose. The ship is a very popular test model, as many
experimental and CFD test results are open to public and have been discussed in many workshops and conferences like
Gothenburg, Tokyo and SIMMAN workshops. Fig.2 shows the body plan of KCS model, and the ship specifications
are provided in Table 1. The mesh used in the simulation is non-dimensional.

4. Computational Procedure

The simple solver with zero-equation Baldwin-Lomax turbulence model and the structured mesh around the bare
hull, make the present simulation light and fast. However, the OpenMP memory sharing model used in the solver
Hafizul Islam et al. / Procedia Engineering 194 (2017) 25 – 30 27

Fig. 1: Overset structured mesh (inner mesh and outer mesh)

Fig. 2: Body Plan of KCS model


Table 1: Specification of KCS

Particulars Unit KCS ship (full scale)


Length between perpendiculars, Lpp [m] 230.00
Breadth, B [m] 32.20
Depth,D [m] 19.00
Draft,T [m] 10.80
Wetted Surface Area,S [m2 ] 9530.00
Displacement Volume,V [m3 ] 52030.00
LCB from midship(fwd+), LCB [%] -1.48
Kyy [%] 0.25Lpp

limits it to distributing computation only in CPU cores of the same node, but not in multi nodes. The simulations
were performed in a single node of an Intel(R) Core i7 CPU with 8 cores, clock speed 2.27 GHz and 4 GB of physical
memory. The average time step used was 1.5 × 10-4 non-dimensional time and for simulating each non-dimensional
time, the required physical time was about 3 hours per case. All the simulations were run up to 7 non-dimensional
times for attaining stable results.

5. Results and Discussions

Calm water resistance prediction is the estimation of drag force for a ship while moving forward in calm water.
Ship’s drag resistance is summation of frictional resistance and pressure resistance. Frictional resistance arises from
the hull surface friction and pressure resistance is mainly the wave making resistance encountered by ship during
its forward motion. When ship moves forward, the bulbous bow create waves on the free surface, thus energy is
transferred from ship to free surface and this loss is termed as wave making resistance. Thus, higher resolution at the
bow and stern sections and also at the free surface near these sections are essential to attain a good prediction. As
for mesh distribution, the emphasis is on capturing the boundary layer near the hull surface. To validate the present
28 Hafizul Islam et al. / Procedia Engineering 194 (2017) 25 – 30

solver, calm water simulation has been performed for KCS model for different Froude and Reynolds numbers. The
mesh resolution for simulation has been selected from previous experience with the solver and thus mesh dependency
study has not been shown here. However, mesh dependency of the solver for different ship models has been shown in
several publications by Ock[9] and Islam[8,10]. Thus, it wasnt repeated here. The mesh resolution used for simulation
has shown in Table 2.
Table 2: Mesh configuration for simulation of KCS in calm water

Specification Domain KCS ship


Degrees of freedom — 2DOF (Heave and Pitch)
Number of grid point (ni × n j × nk) inner 185 × 27 × 106
outer 192 × 45 × 63
Computational region or domain size (×Lpp ) inner (half domain) 1.65L × 0.30L(R)
outer (half domain) 3.8L × 1L × 1L
Minimum grid spacing in longitudinal direction inner 0.0012
outer 0.0187
Minimum grid spacing in radial/lateral direction inner 0.0008
outer 0.0198
Minimum grid spacing in girth-wise/vertical direction inner 0.0008
outer 0.00068

Table 3: Comparison of EFD and CFD results of total drag for KCS

Fn Rn C T × 10-3 (CFD) C T × 10-3 (EFD) Deviation (%)

0.108 5.23 × 1006 3.580 3.796 5.69


0.152 7.33 × 1006 3.500 3.641 3.87
0.195 9.42 × 1006 3.330 3.475 4.17
0.227 1.10 × 1007 3.368 3.467 2.86
0.260 1.26 × 1007 3.830 3.711 -3.21
0.282 1.36 × 1007 4.490 4.501 0.24

Table 4: Comparison of EFD and CFD results of sinkage and trim for KCS

Fn Rn σ × 10-2 [m] (CFD) σ × 10-2 [m] (EFD) Deviation (%) τ0 [deg.] (CFD) τ0 [deg.] (EFD) Deviation (%)

0.108 5.23 × 1006 -0.0989 -0.0900 -9.89 -0.02 -0.017 -17.65


0.152 7.33 × 1006 -0.1012 -0.2750 63.20 -0.0423 -0.053 20.19
0.195 9.42 × 1006 -0.1030 -0.5990 82.20 -0.08 -0.097 17.53
0.227 1.10 × 1007 -0.1081 -0.9440 88.55 -0.11 -0.127 13.39
0.260 1.26 × 1007 -0.1104 -1.3940 92.08 -0.13 -0.169 23.08
0.282 1.36 × 1007 -0.1150 -1.7020 93.24 -0.1035 -0.159 34.91

Total six cases were simulated for different Froude and Reynolds number and for validation; results were compared
with the experimental data presented in the Tokyo 2015 workshop[11]. The simulation data has been processed to
predict total drag coefficient, sinkage and trim values of the ship. The results have been shown in Table 3 and 4.The
results have also been shown in Figs.3 and 4 to make it easier to comprehend.
It can be seen from Fig.3 that the prediction of total drag are in good agreement with experimental data. The
deviation margin decreases with increase in Froude number, Fn . However, at design speed, the deviation becomes
negative. Generally, at design speed, ships perform optimally and the flow pattern around hull form changes slightly.
Since a comparatively lower mesh resolution has been used for the presented cases, the solver was unable to capture
these changes properly and ended up over predicting the value. Also from Fig.4 for sinkage result, although the
Hafizul Islam et al. / Procedia Engineering 194 (2017) 25 – 30 29

Fig. 3: Total drag co-efficient for KCS model at various Froude no., Fn

a b

Fig. 4: Sinkage and Trim Prediction for KCS in Head Waves using Ship Motion (a) Sinkage ; (b) Trim

deviation is high in terms of percentage the actual difference is only around 2 centimeter but for the trim, results are
well predicted as well.

Fig. 5: Pressure distribution on the hull surface and free surface in calm water motion of KCS at Fn =0.26

As mentioned before in this section, during calm water motion, maximum resistance is encountered by the bow
front of the ship. Pressure distribution on hull surface during calm water motion is shown in Fig.5 to illustrate this
30 Hafizul Islam et al. / Procedia Engineering 194 (2017) 25 – 30

phenomenon. Pressure distribution on water surface is shown. In the shown simulation result, the Froude number is
0.26 and Reynolds number is 1.26 × 107 . Another Fig.A.6 for pressure distribution is shown in Appendix.

6. conclusion

Calm water simulation results for a container model, KCS, has been shown in the paper. The paper presented
resistance, sinkage and trim results for different Froude numbers and showed comparison with experimental data. The
average deviation in case of resistance prediction is found to be around 2%, which is very reasonable considering the
used mesh resolution of around 1 million. The sinkage result shows a significant percentage of deviation, however,
in actual, the deviation value is in centimeter scale. This deviation mainly comes from the mesh resolution and the
prediction can be improved by applying a higher mesh resolution. However, considering the actual value of deviation,
it may not be economical to do so. As for trim result, the average deviation is around 17%. In motion prediction,
deviation with experimental data increases with increase in Froude number. This is mainly because of the increase in
turbulence with increasing Reynolds and Froude number. Overall, it may be concluded that the solver SHIP Motion
is efficient and economical in predicting ship resistance and maneuverability characteristics. It predicts ship resistance
and motion with acceptable accuracy, within reasonable time, using minimum computational resources.

Appendix A.

Fig. A.6: Pressure distribution on the hull surface in calm water motion of KCS

References
[1] Summary of proceedings of SIMMAN 2008 Workshop, 2008.
[2] Larsson, Lars, Stern, Frederick and Bertram, Volker, ”Benchmarking of Computational Fluid Dynamics for Ship Flows: The Gothenburg 2000
Workshop 2003”, Journal of Ship Research, Vol. 47, pp. 63-81.
[3] Proceedings of CFD Workshop Tokyo 1994: An International Workshop for Improvement of Hull Form, 22-24 March, 1994, Japan
[4] Lars Larsson, Frederic Stern and Michel Visonneau, ”CFD in ship hydrodynamics: results of the Gothenburg 2010 workshop”, In L.Eca et
al. (eds.) MARINE 2011, IV International Conference on Computational Methods in Marine Engineering, Computational Methods in Applied
Sciences, 2011.
[5] Orihara, Hideo. Development and Application of CFD simulation technology for the performance estimation of ship in waves. Department of
Environmental and Ocean Engineering, School of Engineering, The University of Tokyo. 2005. Ph.D. dissertation.
[6] Akimoto, Hiromichi and Miyata, Hideaki, ”Finite-volume simulation method to predict the performance of sailing boat”, Journal of Marine
Science and Technology, Vol. 7, 2002, pp. 31-42.
[7] Kim Hyuncheol, Akimoto Hiromichi and Islam Hafizul, ”Estimation of the hydrodynamic derivatives by RaNS simulation”, Ocean Engineering,
Vol. 108, 2015, pp. 129-139.
[8] Islam, Hafizul. Prediction of ship resistance in oblique waves using RaNS based solver. Master’s Thesis, Division of Ocean Systems Engineer-
ing. s.l. : MS Thesis, Division of Ocean Systems Engineering, KAIST, 2015.
[9] Ock, Yu Bin. numerical Simulations of Added Resistance around Ships in Regular Head Waves using Overset Grids. Department of Naval
Architecture and Ocean Engineering, Pusan National University. 2014. Master’s thesis.
[10] Islam, Hafizul and Akimoto, Hiromichi, ”Prediction of ship resistance in Head Waves Using RaNS based solver”, International Conference on
Mechanical Engineering (ICME), BUET, Dhaka, December 2015.
[11] Tokyo 2015 Workshop. [Online] 2015. http://www.t2015.nmri.go.jp/.

You might also like