Evaluation of OpenFOAM for Ship Hydrodynamics...Evaluation of OpenFOAM for Ship Hydrodynamics Sung-Eun Kim CFD Group, NSWCCD, U.S.A. June 8, 2007 2 Outline • Motivation • Pilot

Evaluation of OpenFOAM for Ship Hydrodynamics

Sung-Eun KimCFD Group, NSWCCD, U.S.A.

June 8, 2007

2

Outline

• Motivation • Pilot Studies Using OpenFOAM• Summary & Conclusions

3

Motivation• Ideal framework for in-house CFD capability

development• Collaboration and knowledge sharing• Rapid prototyping and technology transfer• Hands-on learning in modern software

engineering applied to CFD

4

Outline

• Motivation • Pilot Studies Using OpenFOAM• Summary & Conclusions

5

Pilot Studies - Objectives• Usability

Learning curve User-friendliness

• ProgrammabilityImpact of language (C++) language barrierTasks for different levels of programming skills

• Extensibility – code architecture & designSolver applicationsClasses and libraries

• Solver performance – benchmarking against commercial and in-house codes

SpeedAccuracyRobustness

6

Pilot Studies Using OpenFOAM

Tasks1. Implementation of a Wilcox’ 1998 k-ω turbulence

modelA new derived class with runtime selectionWall boundary conditions for k and ω (2-layer wall functions)User-interface for model constantsValidation for a body of revolution at incidence

2. RANS solver in a rotating frame of reference1. Source terms2. Transformation of boundary conditions in rotating coordinates3. Validation for an open-water propeller

7

Pilot Development Projects

3. Evaluation of free-surface capability “rasInterFoam”“rasInterFoam” Based on VOF with URANSThe computation with a surface ship is underway.

4. Force/moment monitoring and reporting“forceMoment” has been written and validated

8

Task1 1 - Validation with a BOR

• Body of revolutionReL = 11.7 x 106, L/D = 7.3

• Lift and moment measured for -30° < α < 30°Earlier RANS computations

• Rhee and Sung (2004) • Wilson et al. (2004)

1.3M-cell structured mesh – the same mesh as the one used with FLUENT

• Y+ < 0.7 at wall-adjacent cells• Minimum 15 grid points in BL

QUICK scheme was used for for convection discretization.

9

BOR - Pressure Distribution (ParaFOAM)

10

BOR – Pathlines (FLUENT)

11

BOR - Normal Force Coefficient vs. Angle of Attack

12

BOR -Normal Force Coeff. at α = 18°

13

BOR - Moment Coefficient vs. Angle of Attack

14

Task 2 – Validation for a Marine Propeller

• Computational mesh450K-cell tet. meshCoarse near-wall (wall function) mesh

• FLUENT runs are complete with realizable k-ε and Wilcox’ k-ω model

• Solved in non-inertial (rotating) frame of reference• OpenFOAM computations are in progress

Run into convergence difficulty• Possibly due to strong body-force terms

The solver (sransFoam) and the boundary/initial conditions

15

Task 3 – rasInterFoam ValidationA Surface Ship at Fr = 0.28

• Computational mesh1.8M-cell structured mesh

• FLUENT has been run on the same case.

VOF with modified HRIC

• OpenFOAM computationsMesh translation and problem-setup completedThe computation in progress

16

Conclusions• Usability

User-friendliness – satisfactory• Programmability

A wide range of development (coding) tasks can be accomplished without deep knowledge in C++Learning curve – fairly reasonable with the precompiled solver applicationsC++ knowledge is obviously a great plus and opens up many possibilities.Greatly aided by “learn-by-examples”

• ExtensibilityGreat!

• Solver performance Slightly less robust and less efficient than commercial codesComparable accuracy