Dear all,

I want to simulate a turbulent boundary layer on a zero pressure gradient flat plate and compare the evolution of the numerical results with the theoritical ones.

At the entrance, I use a profile of a turbulent boundary layer I generated using the laws of coles.

I would like to know the best tubulence model to use for this study, and also the size of the mesh to use. In generale, books give the size of the first cell (y+ or more depending on the wall functions) but not the ratio of the growing mesh. Besides, I don't really know the size of the mesh in the flow direction I should use.

If someone knows a bit on this subject I would be glad to read his advices !!

Thanks all for you help through this site !

Vince

Dear Vince,

There is no hard and fast rule for mesh generation for CFD problems, only that a few basic guidelines need to be followed. In the case you want to simulate, considering a one unit long flat plate, you can have a domain of 0.2*1.2, where 0.2 units is the height normal to the plate. The length of the domain is kept 1.2, so that for the first 0.2 units you can have a slip wall b.c. and the remaining one unit of the flat plate you can have no-slip b.c. The inlet velocity can be uniform or prescribed as you have mentioned. Around 100-150 points in the x-direction is sufficient. The grid must be stretched in the y-direction with a first spacing close to the wall with y+ close to 1 (if no wall functions are being employed), so you can use this with a y+ calculator available on net to find the first spacing close to the wall, for the Re you want to simulate. Typically this is around 1e-5, and the stretching can be around 1.15 to 1.2. A fully structured grid with these guidelines coupled with a SA or k-2 model will serve your purpose. Theses models are better known for the performance for aerodynamic applications, so you can as well consider them if you have plans to simulate complex turbulent flows. As far as your problem is concerned a typical check on the correctness of the solution is to compare the skin friction distribution with the available empirical correlation. You can take a look at Phillipe Geuzaine's PhD thesis in this regard(it must be there on net !!!)

Hope this helps

Regards,

Ganesh

Thank you Ganesh for your answer !

I tried to compare the evolution of the velocity profile and the way it fits with the log law, comparing the evolution to the skin friction distribution must be a relevant study as well.

I'm going to find Phillipe Geuzaine's PhD thesis.

Thanks again

Vince