1. Combining the strengths of UMIST and The Victoria University of Manchester Aspects of Transitional flow for External Applications A review presented by Clare Turner

2. Combining the strengths of UMIST and The Victoria University of Manchester Presentation Outline Review of T3 flat plate tests Direction for transition prediction Current and future work Conclusions from the flat plate tests

3. Combining the strengths of UMIST and The Victoria University of Manchester T3 Flat Plate Tests

4. Combining the strengths of UMIST and The Victoria University of Manchester Original Simulation Set-up Domain size taken from S. R. Likki : S = 0.05m L = 3.3m H = 0.08m Inlet conditions: k = 0.0536 m 2 /s 2 ε = 1.35 m 2 /s 3 U = 5.08 m/s y+ ≈ 1

5. Combining the strengths of UMIST and The Victoria University of Manchester Initial Results

6. Combining the strengths of UMIST and The Victoria University of Manchester Anomalous Results After Craft et al.:

7. Combining the strengths of UMIST and The Victoria University of Manchester Possible Areas for Error Differencing scheme Grid density y+ value Residual tolerance Domain size Boundary conditions Cell distribution Near wall treatment

8. Combining the strengths of UMIST and The Victoria University of Manchester Differencing Scheme

9. Combining the strengths of UMIST and The Victoria University of Manchester Number of Streamwise Cells

10. Combining the strengths of UMIST and The Victoria University of Manchester Y+ Value

11. Combining the strengths of UMIST and The Victoria University of Manchester Number of cross-stream cells Grids have 23, 34 and 39 cells in the viscous layer respectively

12. Combining the strengths of UMIST and The Victoria University of Manchester Domain Height

13. Combining the strengths of UMIST and The Victoria University of Manchester Cell Distribution Several meshes generated to compare results and convergence; largest cell distribution: More efficient to concentrate cells at the leading edge 39390

14. Combining the strengths of UMIST and The Victoria University of Manchester Stream-wise Growth-rate

15. Combining the strengths of UMIST and The Victoria University of Manchester Convergence To the left: growthrate=1; to the right growthrate = 1.012

16. Combining the strengths of UMIST and The Victoria University of Manchester Low Reynolds Number Models Available in Star-CD: “Standard” where

17. Combining the strengths of UMIST and The Victoria University of Manchester Low Reynolds Number Models Available in Star-CD: Suga’s where

18. Combining the strengths of UMIST and The Victoria University of Manchester Near Wall Treatment (1)

19. Combining the strengths of UMIST and The Victoria University of Manchester Near Wall Treatment (2)

20. Combining the strengths of UMIST and The Victoria University of Manchester Determination of Inlet Values Dissipation rate originally tailored to FSTI curve

21. Combining the strengths of UMIST and The Victoria University of Manchester Re-calibration of Inlet Values Free-stream taken to be at the edge of boundary layer Experimentally free-stream is at approximately 3 delta Turbulence intensities calculated with new definition

22. Combining the strengths of UMIST and The Victoria University of Manchester Effect of Chosen Free-stream boundary Error was not in the calculation of the free-stream

23. Combining the strengths of UMIST and The Victoria University of Manchester Inlet Values in Literature Inlet k and epsilon values differ to those in literature Differences may arise due to choice of U_inlet 3 different inlets are tested with U = 5 m/s: 1)Using a correlation for lengthscale from work of Savill and interpolating FSTI at inlet 2)As above but with a higher FSTI 3)As 2) but with a lengthscale used by Chen et al.

24. Combining the strengths of UMIST and The Victoria University of Manchester Choice of Inlet Velocity Original velocity set to 5.08 Velocity at leading edge should be approx 5 m/s Turbulence model does not replicate acceleration at the leading edge

25. Combining the strengths of UMIST and The Victoria University of Manchester Effect of Chosen Inlet Values (1)

26. Combining the strengths of UMIST and The Victoria University of Manchester Effect of Chosen Inlet Values (2)

27. Combining the strengths of UMIST and The Victoria University of Manchester T3 Conclusions Largest contributing factor is the low Reynolds number treatment Sufficient work has been done on the grid to only require small alterations for future tests Non-linear eddy-viscosity models improve transition onset prediction and running time is not increased dramatically but no alterations can improve transition length prediction.

28. Combining the strengths of UMIST and The Victoria University of Manchester Approaches for a Physical Solution rather than Correlation Based Models 3 possible approaches: 1)Adaptation of low Reynolds number RANS models, eg. higher order of closure, tailored to specific application. 2)An intermittency model derived using PDFs 3)A model using the concept of laminar kinetic energy

29. Combining the strengths of UMIST and The Victoria University of Manchester 1) Low Reynolds Number RANS Models Advantages: a)Large amount of in-house knowledge b)Have models to develop, low risk Disadvantages: a)Nothing new to contribute b)Low Re models do not accurately represent the transition process c)Higher orders of closure increase computational cost

30. Combining the strengths of UMIST and The Victoria University of Manchester 2) Physical Intermittency Model Advantages: a)New approach b)Should be able to predict all modes of transition Disadvantages: a)No literature to refer to - risky b)Have little expertise in PDFs c)Little or no time would remain to apply to rear wing

31. Combining the strengths of UMIST and The Victoria University of Manchester 3) Concept of Laminar Kinetic Energy Advantages: a)Relatively new giving areas for improvement b)Does not rely on diffusion dominated transition c)Only requires one additional equation Disadvantages: a)Walters and Leylek model (2002) gives poor reaction to large pressure gradients b)Determination of the effective length-scale is only based on distance from the wall

32. Combining the strengths of UMIST and The Victoria University of Manchester Walters and Leylek Model (2003) This is the starting point for 2k model development

33. Combining the strengths of UMIST and The Victoria University of Manchester Walters and Leylek Model (2003) Walters and Leylek assume large scales contribute to the laminar kinetic energy and small scales to the turbulent kinetic energy. The cut – off point is the effective length-scale : Sveningsson uses a different definition:

34. Combining the strengths of UMIST and The Victoria University of Manchester Effective Length-scale Along the Plate At 400 mm (transitional region) : δ 99 = 18.5 mm

35. Combining the strengths of UMIST and The Victoria University of Manchester Effective Length-scale Along the Plate At 800 mm (turbulent region) : δ 99 = 24.1 mm

36. Combining the strengths of UMIST and The Victoria University of Manchester Current Work Adjusting individual terms of Wilcox’s transport equations for turbulent kinetic energy and specific dissipation rate to those of Walters and Leylek Testing code by inserting fully turbulent values from completed simulations Aiming to... Use code with Saturne assuming laminar kinetic energy = 0 Create new scalar variable K L to complete Walters and Leylek Model

37. Combining the strengths of UMIST and The Victoria University of Manchester Any Questions?

38. Combining the strengths of UMIST and The Victoria University of Manchester Walters and Leylek Model 2002