1. 1 * 2 conf. paper papers presented by Y. Addad (UoM) and R. Howard (EDF) April 26 – 1 May, 2009 Hammamet, Tunisia * Int. Symp on Convective Heat and Mass Transfer in Sustainable Energy, April 26 – 1 May, 2009 Hammamet, Tunisia * France-Japan joint Seminar on Thermal fatigue, 5th to 6th October 2009, Tokyo, Japan STURM4-progress meeting, Clamart 13 Oct 2009 EDF – U o Man collaborations on Unstructured meshes for DNS and LES

2. THE EFFECT OF DIFFERENTS GRIDS AND DIFFERENT LES MODELS ON TURBULENCE STATISTICS IN CHANNEL FLOWS Richard Howard and Yacine Addad International Symposium on Convective Heat and Mass Transfer in Sustainable Energy April 26 – 1 May, 2009 Yasmine Hammamet, Tunisia

3. 3 DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS Stretched uniform hybrid Box Reynolds (u tau) 180 Code Saturne http///www.code-saturne.org Collocalised, Cell centered, 2nd order central differences

4. 4 Production Dissipation Viscous Pressure Turbulent Subgrid model contributions DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS

5. 5 Second derivatives Gradients of turbulent viscosity Products of velocity gradients DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS

6. 6 DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS Stretched

7. 7 DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS Stretched Term I (second derivatives) Term II (gradients of turbulent viscosity) Term III (products of gradients) WALE Smagorinsky

8. 8 DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS Uniform

9. 9 DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS Hybrid

10. 10 DIFFERENTS GRIDS AND DIFFERENT LES MODELS IN CHANNEL FLOWS The dominant term is TermIII (products of velocity gradients which includes a kind subgrid dissipation term) The WALE model dissipates the resolved fluctuations in v and w more than the other models. The hybrid grid show numerical noise particularly at the channel centreline. The sloping faces of the non structured grid act to transfer the local mean velocity field between each velocity component.

11. Thermal mixing in T-Junction: Hexa Grid, STAR V4 LES Grid Cells= 2.56 M. LES RUNS: SGS Model: Smago. Pr SGS =0.9 (Default). URANS RUNS: Non-linear k-  model Suga et al. 2006. Numerical schemes: 2 nd order in space. 2 nd order in time. 3.1D h 3D c 13D c

12. 12 - URANS model: - Fails to capture the complex features of the flow. - Most information needed for thermal fatigue studies is lost. - Not able to capture high frequency events. - Can this be improved or other approaches tested (example DES) ?? - LES with Unstructured grids and Professional Software: - Second order accuracy seems OK. - Mesh is extremely important, adapted to (1/10) large eddy scale ? => need unstructured mesh - Work in progress: How about tetrahedral and polyhedral cells, different SGS models?? Conclusions and future work Acknowledgements: UK research council project “Keeping the Nuclear Option Open” (KNOO) FR ANR-CIS STURM4