turbulence models
 

Hi all,

I was just wondering if it is possible to use different turbulence models in different regions of a given domain. I guess this would require some kind of blending in the interface regions.

Thank you

Some turbulence models already have in-built mechanisms that blends from one turbulence model to another: different models are used in the near-wall and the free flow region. Thats the case for the k-omega SST model or DES type models (well in fact that's even the case for a standard k-epsilon model).

If you do not want a wall region blending, you can use the ELES (embedded LES) capability in Fluent 13 where you can explicitly specify a flow zone that will be solved using a LES model, whereas a RANS model will be used in all the other areas.

Finally, you can implement any custom blending turbulence model you wish by coding UDFs.

Thank you tschump. What I had in mind was a slightly different doubt. Say we have a domain which has two cavities connected by a small duct. The flow structure is very different inside each cavity. We know that in the first cavity a model A behaves well, and in the second one another model B behaves better. In my case A and B would RANS models but not necessarily solving the same equations, say mixing k-e with k-w, or 1-equation with 2-equation models.

Could we apply the model A in the first cavity, and the model B in the second cavity, allowing if necessary a certain region for blending both models? WOuld it be possible to define two subdomains, and use each turbulence model in each subdomain? I am not very sure what information would be shared between subdomains for this case if solving different equations.

Thank you

In that case, if the calculation is steady, you may do one separate calculation for each cavity, each one having its own turbulence model. You can easily couple the two calculation either by UDF or by using profiles.

Thank you tschump.

Not very sure how to couple the two cavities. Do you mean running first one cavity, take then the profiles at the interface and use them as boundary conditions for the second cavity, run then the second cavity, and copy back boundary conditions at the interface for first cavity and keep iterating until convergence? This could be a quite large, tedious process. Or maybe you mean that FLUENT can run the two cavities simultaneously, passing information across at the interface.

Thank you

Ｈｉ　ａｌｌ，
ａｎｙｂｏｄｙ　ｃａｎ　ｔｅｌｌ　ｍｅ，　ｗｈａｔ　ｉｓ　ｂｅｔｔｅｒ　ｖｉｓｃｏｕｓ　ｍｏｄｅｌ　 ａｎｄ　ｈｏｗ　ｉｓ　ｍｏｄｅｌ　ｃｏｎｓｔａｎｔｓ　ｆｏｒ　ｃａｌｃｕｌａｔｉｏｎ　ｓｈｉｐ　ｒｅ ｓｉｓｔａｎｃｅ，
Ｔｈａｎｋ　ｙｏｕ　ｖｅｒｙ　ｍｕｃｈ