Date Author Mdress
March 2005 S.L. Toxopeus
Deift UnÍversity of Technology Ship Hydromechanics Laboratory
Mekelweg 2, 26282 CD Deift
TU Deift
DeIftUniversltyot Technology
Calculation of the Flow around the KVLCC2M
Tanker.
by
S.L. Toxopeus
Report No. 1426-P
2005
CFD Workshop "CFDWSO5, National Maritime
Research Institute, Tokyo, Japan, march 9-11,2005
PREPR INTS of
CFD WORKSHOP TOKYO 2005
Editor TAKANORI BINO
March 9-11, 2005
National Maritime Research Institute
Tokyo, Japan
CONTENTS
Part A: lest Cases and Huh! Data
A I. Test Cases
A2. Hull Data
Part B: Computational Methods
Bi. Acronyms and Code Names
Original Questionnaire
Summary of Replies to the Questionnaire
Part C: COmputed Results
CL Global Qúantities,All Test Cases
C2 Local Quantities (Test Case I)
C2.1 Local Quantities (Test Case 1.1)
C2.2 Local Quantities (Test Case 1.2)
C23 Local Quantities (Test Case 1.3)
C2.4 Local Quantities (Test Case 1.4)
Local quantities (Test Case 2)
Local quantities (Test Case 3)
Local quantities (Test Case 4)
Local quantities (Test Case 5)
Part D,: Papers
Luquet, R., Jacquin E., Alessandrini, B. and Guillerm, RE.:
RANSE with Free Surface Computationsaround Fixed and Free
DTMB 5415 Model, in Still Waterand in Waves
Broglia,R., Muscari, R. and Di Mascio, A.:
Computations of Free Surface Turbulent Flows Around Ship Hulls
by a RANS Solver
Rhee, S.H. and Skinner, C:
Unstnctured Grid Based Navier-Stokes Solver for Free-Surface Flow
around Surface Ships
Ptteñden, Ri., Turnock, S.R. and Pashias,C.:
Obli4ue Ship Flow Predictiòns Using Identification of Vortex Centres
to Control Mesh AdaptatiOnHsu, K.L., Chen Y.L, Chau,
Chien, HP. and Kouh,J.S.:
Ship Flow ComputatiOn of DTMB 5415Chien, H.P., Kouh, J.S., Chau, S.W, Hsu KiL; and Chen Y.J.:
Ship FlOw Computation of KVLCC2MÖstman, A.:
Computation of Free-Surface Viscous Flow Around the DTMB 54 F5
and KCS Hull Forms
Deng,G.B. , Guilmineau, E., Queutey, P. and Visonneau, M.:
Ship Flow Sinuilations with the ISIS CFDCode
Chao, K. K.-Y.::Numeric Propulsion Simulátion for the KCS Container Ship
Wilson, R., Carrica,P. and Stern, F.:
RANS Simulation of a Container Ship Using a Single-Phase Level Set Method
with Overset Grids
Carrica,RM.,, Wilson, R.V. and Stern, F:
Linear and Nonlinear Response of Forward Speed Diffìaction
fora Surface COmbatant
Simonsen, C.D and Stern, F.:
RANS Simulation of the Flow around the KVLCC2 Tanker
Regnström, B., Broberg, L.,Östberg, M., Bathfield, N. and Larsson, L.
Drag Prediction forthe KVLCC2M Hull
Gorski, J.J. and COleman, R. M.:
Computations oftheKVLCC2M Tanker under Yawed Conditions
Miller, R., Gorski, J., Wilson, R. and Cárrica, P.:RANS Simulation of a Naval Combatant Usinga Single-Phase
Level Set Method with Overset Grids
Bull, P.W.:
Verification and Validation of KVLCC2M Tanker Flow
Lübke, LO.:
Numerical Simulation of the Flow. aroundrthe PropelledKCS
Cura Hochbaum,A. and Pierzynski, M.:
Flow Simulation for a Combatant in Head Waves
Kim, J. Park. 1.-R. and Van, S.-H.:
RANS Computations for KRISO'Container Ship and VLCC Tanker
using, the WAVIS codeTahara, Y., Wilson,R., and Carrica,P.:
Comparison of Free-Surface Capttiring and Tracking Approaches in Application to Modern Container Ship and. Prognosis for Extension
to Self-Propulsion Simulator
Chou, SK, Chin,
Chang, F.N. and Wu, CH.:
Computations of Flow around Ships with Free Surface or Obliqued
Towing Angle
Starke, B. van der Ploeg, A. and Raven, H.:
Free-surface Viscous Flow Computations for KCS and 5415 Models
Using the PARNASSOS CodeEça, L., Hoekstra, M. and Toxopeus, S.L.:
Calculation of the Flow around the KVLCC2M Tanker
Suzuki, K., Hirakawa, R., OjimaA. and Kamemoto, K.:
Calculation of Viscous Flow around Ship Models by Means of
Advanced Vortex Method
Hirata, N. and Kobayashi, H.:
Ship Flow Computations for KVLCC2M
Hino, T. and Sato, Y.:Ship Flow Computations by Unstructured Navier-Stokes SolverSURF
Kume,K., Hasegawa, J., Tsukada, Y., Fujisawa, J., Fukasawa, R. and Hinatsu, M.:,
Measurements of Hydrodynamic Forces, Surface Pressure and Wake for
Obliquely Towed KVLCC2M ModeL and Uncertainty AnalysisSUMMARY
The flow around the double-body KVLCC2M tanker is:computed ingrids with H-O topology. First the zero-driftangle case is considered Two turbülence models are used: Menter'sone-equationmodel and Kok's
ver-sion ofthek - wmodel A thorough uncertainty anal-ysis based on solutions on 8 grids of various density
is carried out. In the second part of the paper the flow predictions for drift angles 3, 6, 9 and 12 degrees are discussed; Encouraging, results are obtained, but the relatively high level of uncertainty in theiprersure drag evaluation requires further attention.
INTRODUCTION
The calculations for the, KVLCC2M tanker hull
form were performed with the flow solver
PAR-NASSOS (Hoekstra & Eça, 1998), which is based on a finite-difference discretisationof the Reynolds-averaged continuity and momentum equations with fully-collocated variables and discretisation. The equations are solvedwith a coupledprocedure,retain-ing the continuity equation in its original form.
In PARNASSOS several eddy-viscosity
turbu-lence models are available. In a numerical calculation of a ship stern flow, the turbuléncemodel selection is
not only based on the quality ofthe predictions, but
alsoon the numerical robustnessandthe ability to
con-verge the solution, ie. reduce.the iterativeerror tothe desiredvalue.
Mostapplied in PARNASSOSis theone-equation
model proposed by Menter (1997), which leads to a
remarkably robust method andallows convergeüceof the solütion to machine accuracy in almost any case. However, several validation studies have shown that
the predictions of the flow field: in the bilge vOrtex region are not as good as the ones obtained with the k - w model. Therefore, we have also
perfomiedcal-culations for the zero-drift case using the TNT version
of the kw model:(Kok, 1999). The main advantage
of the TNT version when compared with the popular BSL ànd SST versions (Menter, 1994) is the absence of references to the distance to the Wall. The calcula-tions for thedrift cases were conductedusing Menter's turbulence model only.
Calculation. of the Flow around the KVLCC2M Tanker
L; Eça1, M. Hoekstra2 and S.L. Toxopeus2
Instituto Superior Técnico, Portugal
2 Maritime Research
Institute, Netherlands
Deift University of Technology, Netherlands
The Spalart correction to account for the effects
ofstreamwise vorticity, described in Dacles-Marianiet aL (1994), is adopted in both turbulence models. No
attempts have been made to add special features for modelling transition. So the basis turbUlence model acts as the transition model as welL The Reynolds
number is equal to 3945 x 106.
Ail computationswere performed using dimensionless quantities with
LppandU as the reference length and velocityscales.
COMPUTATIONALDOMAIN ANDGRID TOPOLOGY
All calculations described in this papér were
con-ducted for theünappendedthtill form.
Several grid topologies have been tested for the
calculation of the flow aroundthe KVLCC2M double model (Eça & Hoekstra, 2005). Theresúlts presented in this paper wereall obtained onstructuredgrids with H-O topology with someextra grid clusteringcloseto
the propeller plane. For the zero-drift case, a single-block calculation was conducted while for the
non-zerodrificase the domain wasdecomposed into
effec-tively two blocks. The six boundariesofthecomputa
tional domain are the following: the iñlet boundary is
ax = constant plane located upstream of the forward
perpendicúlar;theoútlet boundary is also a transverse plane downstream of the aft perpendicular; the exter-nal boundary isa circular or elliptical cylinder; the re-maining boundariesarethe ship surface, thesymmetiy
planeoftheship and the undisturbedwater surface;
Zero-drift case
Forthe zero-drift case, the inlet boundary was located at 0.25Lppupstreamoftheforward perpendicular. Thé outlet boundary was located at 0.25Lpp downstream
of the aft perpendicular. The radius ofthe cylinder
defining the external boundary was 0.1 8Lpp. Eight geometrically similar gilds have been gen-erated with in-house codes (Eça, Hoekstra & Windt, 2002)for theestimation of the discretisation error. The
variation in the number of grid nodes in the stream-wise, n, normal, n0 and girthstream-wise, nc directions is
Figure 1: Gridat botiñdariesofthecomputational
do-main.
presented in Table 1, which includes also the maxi-mumy obtained at the firstgrid nodeaway from the waIl(y) with Menter's one-equationmodeL Figure 1 givesan impression ofthe(coarsened) point
distribu-tion on the boundaries of the computadistribu-tional domain.
Table 1: Number of gridnodesand y at the first grid node away from thewall.
Non-zero drift cases
The flow around the hull at non-zero drift angles has no port-starboard symmetry and the computational do-main must be extended'to cover the port side as well. Furthermore, alarger dowainisrequiredin order to in-corporate the drift angle. On each side of the dOmain
thegridconsiststof an inner block andan outer block, see Figure 2. The innerblock is the same for all yaw anglesandthe outer blockcan deform toallow forthe
drift angle of the ship Therefore grids for various drift
anglescanbe madeefficiently.
The inner block is generated with a number of
cells similarto the grids as usedfor the zero-drift case.
Based on early calculations (Toxopeus, 2004),. grid clustering at the propeller plane and the bow of the
ship was applied to resolve thegradients of the flow at these locations more accurately.
To incorporate the drift angle of the ship, the in-nerbiock is rotated around the vertical zaxis over the desired yaw angle. Then the outer block is generated
aroundthe innerbiock. Thecell stretching used in the
inner block is automatically appliethto the outer block
as well. lt wasdecidedto have
matchinginterfacesbe-tween the blocks sothattheinner and Outer blocks can be merged. The sizeof the outer blocks is chosen such that the rotated inner block can smoothly be
incorpo-rated in the outer grids This means that increasing drift angles will result in wider domains. The sizeof thedomainisbasedonthe assumption that a solverfor
potential flow isUsedto calculate thevelocities in the inflow and external planes. Before starting the
calcu-lations, theseparateblocks are mergedinto one block forthe portside of the ship andanother block for the
starboard side of the ship.
Figure 2:
Inner and outer blocks (coarsened) at
120 drift angle.
The number ofnodes in the gridsused for the drift casesarepresentedin Table 2, Which includes also the
maximum y+ value for the cells adjacent to the hull
that was obtained during the calculátions. A positive
drift angle ß correspondsto:theflowcomingfrom port
side. Note thatalsora calculation withzero drift angle
was conducted with a grid similar to the grids used
for non-zero drift in order to be able to determine the
relation between the drift angle and integral or local
variablesconsistently.
Table 2: Number of grid nodes andy for drift cases.
Table 3 presents the sizes of the computational domains for the drift case calculations. For
increas-ing drift angles, the computational domain size is
in-creased in order to be able to incorporate the inner
block in theouter deforming mesh. Grid n
n,jn
4-01
561 11357 24
02
481 97 49 0i27 G3 441 89 45 0.26 G4 401 81 4130
G5 361 73 37034
G6 321 65 33 0.37 G7 281 57 29046
G8 241 49 25 0.59ß
n n nodes y24 0° 449 81 45 1.6 x 106 0.32 30 449 95 2 x 453.8x 106 40
6° 449 95 2 x 45 38 X 106 0.55 90449 952x45 38x106
069
12° 449 95 2 x 45 3.8 X 106 0.80Table 3:. Size of'computationaî domain for drift cases.
BOUNDARY CONDITIONS
At the ship surface the no-slip condition is applied di-rectly and the normal pressure derivative is assumed to be zero. The undamped eddyviscosity, the variable in Menter's one-equationmodel, vanishes at a no-slip
wall. With thepresentformúlation' of thek - amodel
Kok and Spekreijse,:2000), alltthe turbulent quantities
are zero at asolidwall.
Symmetry conditions are applied at the
undis-tUrbed Watersürface and on the ship symmetry plane
(for the zero-drift condition). At the inlet boundary,
the velocityprofiles areobtained from apotential flow solution, which also determines the tangential velocity
components and the pressure at the external bound-aiy. At the outlet boundary, streamwise diffusion is
neglected and the stresmwisepressure!derivative isset
equalto zero.
For the drift cases, the lift generated by the hull
form is modelled in theipotential flow solution by
ap-plying a vortex sheet on the symmetry plane of the
ship. At the stem of the ship, the Kutta condition (the flow leaves the trailing edge smoothly) is
ap-plied, which allows thesolution of the unknown
vor-tex strengths on the sheet. Since the only purpose of
the potential flow solutión is to set the boundary
con-ditions for the viscous flow solution at the inlet and
external boundaries, vortex shedding from thethilge of the ship is omitted.
UNCERTAINTY ESTIMATION
We only deal with the discretisation error, assuming
the iterativeandIround-off errors to be negligible. The uncertainty, LI0, of any integral or local floW quantity is estimated witha procedure based on a least squares root version (Eça and Hoekstra, 2002) of the Grid
Con-vergence Index (Gd), proposed by Roache (1998).
Two basic error estimators are involvedin the present
procedure for uncertainty estimation: the
extrapola-tion to grid cell size zerO performed with Richardson extrapolation, ERE, and the maximum difference be-tween the data pointsavailable,AM.
We have collected some experience with several variants of uncertainty estimation procedures (Eça and
Hoekstra, 2004). In the present calculations we have adopted the following options:
o Determine the observed order of accuracy, p,
from the available data.
o For 0:95 p < 2.05, U0 is estimated with the
GCI andthe standard dèviationUj,, of the fit:
=
l'.25ô +U111.o For 0<p< 095,thesame error estimate is made
but is then compared with the value of AM mut tiplied by a factor of safety of 1.25, so that U0 is obtained from:
Lío =min(l.25+Uf,,L25AM).
o Forp
2.05, =max( I .255k + Uf,,,! .25AM),where 5F is also calculated in the least squares root sensewith p
=
2.If monotonicconvergence is not observed, U0
=
3AM.RESULTS 'FOR ZERO DRW1'
Numerical Convergence
In 'the' present calculations we have adopted as con-vergence criterion the redUction of the maximum dif-ference betweenconsecutive iterationsofthethree ve-locity componentsand of the pressure to 10 12, which is equivalent tomachine accuracy.
200
400 600Iteration
800Figure 3: Convergence history on 281 x 57 x 29 grid.
Unfortunately, we were notable to satisI'this cri-teria for the 8 grids With the TNT k - wirnodel. IO the
grids GI, G3 and G5 the convergence stagnates at a
level that does not allow to neglect the iterativeerror andso we have dropped the results obtained inthese 3
grids. The convergencehistories obtained for theG7
grid with the two turbulencemodels are illustrated in
Figure 1 /3 inlet [Lpp]: outlet [Lpp] width H Lpp] depth ELppJ 0°
-073
092
H 0.1818
3° -0.74 0.93 0.42 0.36 6° -0.75 0.94 H 0.49 0.36 90 -0.76 0.95 0.55 0.36 120 -0.76 0.9506l
0:38ResistanceCOefficients
Thepredicted values of total.resistance, C, friction re-sistance,'Cp, and pressure resistance; Cp, arepresented
in Table 4 with the estimated uncertainties. These force components have, been madè non-dimensional
using pU2Sowith S0, the wettedsurfaceat rest.
Table.4: Predicted resistance coefficients and theires-timateduncertainties (a=k-w,.b=Menter). 3.68 3.64 3.6,
k
C.)35
O.8 3.52 3.48o 1.612
0.4 p 4.l.p=2 o k-w U- 0.072 p-0.8 .0 Mentez h/h11.5 o k-w U- 0.344 p-O.4' .0 Meiner 0.5' 1.5. 225
h/b1Figure4: Convergence of the friction and pressure re-sistance'with grid refinement.
RESULTS FOR DRIVf ANGLES 'In both cases, the uncertainty of Cp is clearly
larger thanthe one of CF. The esti'mateduncertainties
are much larger for thek co modebthan forMenter's
one-equation model. As illustratedin Figure 4, theob-served order of accuracy is below I' for the 'solutions
obtainedwith the k wmodel,,whereas the p obtained with Menter'staodel1 is4.7for CF' and 2 forCp.
'The fit to Cpptotted in Figure 4 is'made'with,p2.
2.5
Velodty field at the'pròpeller'plane
Theselection of the turbulence model has a significant effectonìtheprediction of the velocity field at the
pro-peller ;plane The isolines of axial velocity obtained
with thetWo turbulence models and the transverse ve-locity fields are.plottedinfigure 5.
0.2U '02 .04 '004 'O" 'tOO 404 yL
Figure 5: Velocity field
at the propeller plane
(top:axial velocity,bottom: transverse velOcity).
There is a more pronounced 'hook shape' in the k - w solütion 'than in .the prediction with Menter's
model. There are alsodifferences' in the bilge vortex, specially inthe lower, part closetothe symmetry plane.
Figure6: Uncertainty intheaxial'velocity fieldàt he
propeller plane..
An interesting result is the uncertainty of the ax-ial velocity field prediction at the propeller plane The values of U.áre below00i 'for. most ofthe field. How-ever, atthebilge vortex region the maximum values of
U reach levels above 0l with 'the k w predictions
exhibiting the largest values of uncertainty
Numerical Convergence
In the calcúlatioñs of the drift.cases a reduction of the maximum difference in pressure between consecutive
iterations' to 5 x l05wasadoptedas the convergence
criterion. 1nthesecases' there.is noattempt toestimate
the discretisationerror. Therefore, thereis no need to
reduce the'iterative error tomachineaccuracy.
C -ic Cp Cp
x'103' x103 xl 0 xl'03 Xl0 XIO
a 4.454 0.416 3578 0.072 0.876 0.344
IntegraFCoefficlents
In this section, the forces and moments presented are
madenon-dimensional using respectively pU2LppT
and pU2L,T, in accordancewith specificationsfor
the CFD Workshop 2005. Table 5 presents the results
of thecalculatiòns for each drift anglè ¡3 as well as a
comparison between the calculated variables and the
measured ones. CX is the longitudinal force, CI' the
transverseforce and C'N the yawingmomeñt with' re-spect to theorigin of the xyzcoordinatesystem, which
is located at station 10.
The results for zero drift angle. are comparable to the results as presented inTable 4 (C =4.19 x io-i, CF = 3.50 x i- and
Cp = 068x l0-).
Table5: Integral variables.
Except maybe for the results for 3° drift angle,
the. predictions obtained by the' calculations are very promising. In almost all cases the prediction.is within 10% from themeasurements. Noteworthy is the con-.sistent underprediction of the transverse force, while
both the longitudinal force and yawing moment are
predicted quite accurately. Figure 7 presents the
yaw-ing moment as a function ofthe drift angle More
results can be found in the proceedings of the CFD
Workshop 2005.
Compared to the results presented for earlier cal-culations for the KV:LCC2M ata drift angle, see
lOx-opeus (2004), the grid refinement at the bow and
stern has improved the prediction of the longitudinal force The earlier calculations were conducted with
an equidistant grid with 251 nodes in longitudinal
di-rection along the hull surface. For the present calcu-lations, the grid was non-equidistant with 342 nodes along thèhuil surface. The improvement in the pre
diction is mainly caused by the change in the
predic-tion .of the pressure component CXì, siñce the fric-tion component CXF is practically equal for the two
z
C., 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000 -0005 -0.0103 o 3 6 beta .9 12Figure 7: Yawing moment against drift anglà.
(exp: open circles, cfd: solid line)
different grids, see Table 6 and the results for zero drift angle presented in Table 4. The pressure
com-ponent however decreases considerably for the non-equidistant grid. Thisfinally results in a better predic-tion of the longitudinal force for the non-equidistant
grid.
Tàblà 6: CXcomponents fordifferentgrids.
Also for the transverse force CY the difference in pressure component determines the difference in the
total transverse force, see Table 7. Furthermore, the
friction component is an order of magnitude smaller than the prcssúre component and therefore is almost negligible in the total force. However, when
compar-ing the measurements with the results
forthetwodif-ferent grids it is seen that in this case the finer grid
does not lead tota better prediction. A grid refinement study should'be conducted'to verif' whether sufficient
grid nodesin girth-wisedirectionareapplied.
Table 7: CY components for different grids
Side forcedistributlon
To understand the manoeùvrability Of ships and to
be able to generate reliable generic mathematical ma-noeuvringmodels, the longitudinal distribution of the 15 p
cfd
CX.
xl02 CY x102 CN xlO2 CX x'102 CY xl02 CN x'l02 -1.77 -091 -0.76 -1;80 -1.25 -0.71 00 -1.74' 0 0 -1.76 -0.00 -000 3° -1.77 0.91 0.16 -1.78 1.26 0.61 60 -1.79 2.26 1.46 -1.77 2.56 1.39 90 -1.79 4.23 201 -1.73 4.55 1t94 12° -1.78 6.43 2.53 ' -1.75 7.08 2.54 (3 grid CXp CXF CX x'l02 x102 x102 0° equidistant -0.433 -1.455 -1.889 non-equidistant -0.283 -1.458 -1.741 3° equidistant 0:5l9 -1.457 -1.976 non-equidistant -0.299 - 1.468 - 1.767 /3 ECX ECY CN - -3° -2% -27% 8% 00 l% -30 -1% -28% 25% 6° 1% -12% 5% 9° 4% -7% 3% 12° 2% -9% -1%/3 grid CYp CYF CI'
xlO2 xlO2 x102
3° equidistant 0.940 0.049 0.989
sidefòrce is Of interest. Therefore, thepredicted longi-tudinal distribution of the lateral force has been com-pared to the experimental values to determine the
ac-ctiracy ofthepredictions, seeFigure 8.
The comparison shows that although the sideforce according to Table 5 issystematically underpredicted, the predicteddistributionis very close to themeasure-ments and therefore the accuracy of this prediction is
judgedtobe good.
05
0.4 0.3 02 0.1 0 'j oo.o o dV dV o dV dV xp) 6' cId) 6' exp) 12' cfd)12° W;Figúre8: Side:force distribution.
CONCLUSIONS
At zero drift angle, we have performed calculations with two eddy-viscosity models: the one-equation
model proposed by Menter and the TNT version of
the
k - w
model. Grid refinement studies have been performed with bóthmodels to estimate thenumericaluncertainty of the predictions. The results show that
the uncertaintyof the pressure resistancecoefiucient is at least one order of magnitude làrger than the uncer-tainty of the friction resistance coefficient. The level of uncertainty of the selectedflow quantities depends
on the turbulence model choice. The
k w
modelleads to higher levels of uncertainty than the Menter model. However,, the comparison of the predicted ve-locity fleldsat thepropeller planeshows a better
agree-ment of the
k - w
predictions with the experimental results.For the non-zero drift cases, which werealIcon-ducted using Menter's.oneequation model, the com
pliance of the predictedresults With themeasurements
is good and within 10% from the measurements for most cases. Based on a variation of the grid density and grid node spacing it was found howeverthat the
uncertainty in the pressure component of all integral
forces is large. A grid sensitivity study is therefore
recommendèd in orderto verif' the calculated results
0.4 06
REFERENCES
Dacles-Mariani J., ZilliacG.G, Chow J.S., Bradshaw P., "Numerical/experimentalstudyofa Wingtip vortex in the near field", AIAA Journal, Vol. 33, September
l995,pp. 1561-1568.
Eça L., HoekstraM., Windt J., "Practical Grid Genera-tion Toolswith ApplicaGenera-tions to Ship Hydrodynamics", 8' International Conference in Grid Generation in
Computational Field Simulations June 2002 Hawaii,
USA.
Eça L, Hoektra M., "An Es'alùation of Verification Procedures for CFD Applications", 24" Symposium
on Naval Hydrodynamics, July 2002, Fukuoka, Japan. EçaL., Hoekstra M., "A Verification Exercise for Two 2-D Steady Incompressible Turbulent Flows", LIh
Eu-ropean Congress on Computational Methods In Ap-plied Sciences And Engineering ECCOMAS 2004
July 2004, Finland
Eça L., Hoekstra M., "On the influence of grid to-polô' on the accuracy of ship viscous flow
calcula-tions", 5's' Osaka Colloquium on Advanced CFD Ap-plications to Ship Flow and Hull Form Design, 2005, Osaka,Japan.
Hoekstra M., Eça L., "PARNASSOS : An Efficient
Method for Ship Stern Flow Calculation", Third Osaka Colloquium on Advanced CFD Applications to Ship Flow and HulIForm Design, May 1998, pp.331-357, Osaka, Japan.
Kok JC., "Resolving the Dependence on Free-stream
values for the
k - w
Turbulence Model", NLRTP-99295, July 1999, National Aerospace Laboratory,NLR, The Netherlands
Kok LC, Spekreijse SP., "Efficient and Accurateim-plementation of the
k - w
Turbulence Model in the NLR multi-block Navier-Stokes system",NLR-TP-2000- 144,May 2000, National AerospaceLaboratory, NLR, TheNetherlands;
Menter F.R., "Two-Equation Eddy-Viscosity
Turbu-lence Models for Engineering Applications", AIAA
Journal, Vol.32, August1994, pp. 1598-1605. Menter F.R., "Eddy Viscosity Transport Equationsand Their Relation to thek - eModel", Journal of Fluids Engineering, Vol. 119, December l997,pp. 876-884
RoacheP.J, "Verificationand Validation in Computa-tional Scienceand Engineering", HermosaPublishers,
1998.
Toxopeus S.L. "Validation of calculations of the
vis-cous flow around ashipin oblique motion", TheFirst
MARIN-NMRI Workshop, October 2004, pp. 9 1-99.