1. CFD for Aerodynamics of Fast Ships Volker Bertram

2. 2 Scenario - what, why, etc. Aerodynamic flow around a ship superstructure is important in many ways: Exhaust dispersal Ventilation of occupied spaces Wind forces, especially for maneuvering Special operational conditions - helicopter landing, etc. Used to make design & operational decisions

3. 3 Wind tunnel proven tool This information now predominantly from wind tunnel tests Wind tunnel proven tool to provide useful information about the airflow.

4. 4 Aero CFD: An alternative now! Aerodynamics CFD effective in other engineering fields: aerospace automotive civil engineering Advantages: All information available at any time More precise control over what is viewed More details are possible Full scale (but still idealized...) Non-intrusive

5. 5 CFD for ship aerodynamics now a topic Problem difficult due to various factors: Grid generation very difficult Large grid cell count Complex physics Recent progress addresses these issues: Unstructured, more robust solvers Improved, automatic grid generation tools Advanced numerical modeling techniques Affordable parallel computing

6. 6 Several applications in last years DMI Sirehna Marintek NRL JJMA Stanford KRISO... (?) Sirehna JJMA NRL

7. 7 Several applications also at HSVA

8. 8 Tools and Methods Used Typical geometry imported from IGES format Unstructured, tetrahedral grids generated using ICEM- CFD, cell count of up to 5 million cells Calculations with Comet on a parallel PC cluster

9. 9 Geometric modeling of all superstructures impossible

10. 10 Baffle Elements Tested model global effect of replacing filigree structures by semi-permeable cell boundaries  22 000 cells6 000 cells cell count for a 2D case

11. 11 Baffle elements disappointed geometric model Assorted baffle parameters k and |v| for mast

12. 12 Simple block does the trick geometric modelSimple block k and |v| for mast

13. 13 Application to fast ferry Superfast VI, HDW, 29 kn IGES file from yard too detailed: several weeks work to downstrip

14. 14 Grid: 680,000 cells per symmetry half

15. 15 Model tests performed at IFS wind tunnel Physical model (1:150) in wind tunnel

16. 16 Local refined grid reduces discretization errors Exhaust concentration

17. 17 CFD Similar agreement for wind from abaft Experiment

18. 18 They cannot be fulfilled all at the same time! Some similarity laws always violated ratio of velocities geometric similarity ratio of mass flux Reynolds number of the inflow Reynolds number of the jet Froude number of the jet ratio of momentum flux

19. 19 Parameter studies exhaust gas temperature 300°C inviscous computation model test parameters full scale Rn

20. 20 stream lines (0°) pressure distribution (30°) turbulent kinetic energy k (0°) Visualisation of different quantities

21. 21 Forces OK for small-to-medium angles Drag Side force Roll moment Differences for large oblique angles attributed to flow separation insufficiently captured by turbulence model

22. 22 Application to fast SES AGNES 200, French SES, 40 kn First step: Create IGES description

23. 23 Grid topology allows easy re-gridding Inner cylinder in outer block Matching every 5° 2.9 million cells

24. 24 Pressures change with angle of attack 180°170° 150°

25. 25 Vortex formation behind superstructure streamlines  =180°

26. 26 Strongly 3-d flow streamlines  =180° 5 cm higher

27. 27 Features similar for 170° streamlines  =170°

28. 28 Less complex “foil” flow for for 150° streamlines  =150° Streamlines return to original direction further downstream Flow follows low-pressure side of SES

29. 29 Flow strongly 3-dimensional Virtual Reality may help understanding the flow

30. 30 Virtual Reality comes in many shapes Sources: VRL, Univ. Of Michigan; VRCA RWTH Aachen cave head-gear PC Poor man’s VR suffices! VRML

31. 31 What is VRML? 3D file interchange format 3D analogue to HTML ISO standard VRML = Virtual Reality Modeling Language

32. 32 VRL, Univ. of Michigan INSEAN INSEAN+TUHH TUHH Steps creating a CFD VRML model Study experience of others

33. 33 Direct export: 43000 polygons 2810 KByte Steps creating a CFD VRML model Study experience of others Export geometry data from RANSE solver Downsize geometry

34. 34 Direct export: 43000 polygons 2810 KByte After merging: 900 polygons 130 KByte Steps creating a CFD VRML model Study experience of others Export geometry data from RANSE solver Downsize geometry

35. 35 Steps creating a CFD VRML model … Build VRML geometry model Process and downsize flow data Add flow data to VRML model Add interaction to VRML model Interactive high-lighting Interactive selection

36. 36 Pressures use colour VRML interpolation Work continues: Refine algorithm to downsize model

37. 37 Conclusions CFD offers more insight than wind tunnel Further work required for validation Wind tunnel may be too optimistic for smoke tracing VRML suitable for post-processing