CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CD-adapco (http://www.cfd-online.com/Forums/cd-adapco/)
-   -   Turbulence modelling (http://www.cfd-online.com/Forums/cd-adapco/53869-turbulence-modelling.html)

Panos July 5, 2004 10:26

Turbulence modelling
 
Hi I am currently working on airfoil case and I would like to have a second opinion regarding turbulence modelling. For modelling turbulence shall I use the High Reynolds model since I have very high Reynolds numbers or there is a more adequate model (k-epsilon RNG for example)? My computational domain consists if 23000 cells and the boundary conditions are inlet,outlet and wall for the airfoil. I really appreciate your help.

4xF July 5, 2004 13:55

Re: Turbulence modelling
 
Best 2 equations turbulence model is K-Omega SST (see supplementary release notes of STAR-CD v3.150A on how to set up) for your case. Use a pressure boundary at the outlet and make use of a much finer mesh. To reduce runtime, use the AMG solver.

Panos July 6, 2004 08:20

Re: Turbulence modelling
 
Is it wrong to use oulet boundary condition. I think that if I use pressure the conditions of the flow are unknown, the computational domain is quite large, and this may lead to divergence? Am i right? In addition why should I use a finer mesh since the results I get are grid independent? What is the AMG solver. Again thanks a lot for your help

Peter July 6, 2004 10:52

Re: Turbulence modelling
 
k-w SST needs a fine near wall mesh like TwoLayer or K-eps Low Reynolds models (have a look on the user guide). Therefore you should refine your mesh.

4xF July 6, 2004 14:50

Re: Turbulence modelling
 
Furthermore, what do you mean by mesh independent solution? The mesh independent solution is only reached when you have refined your mesh so many times that the calculated solution does not change anymore. The AMG solver is an Algebraic Multigrid Solver. This has the advantage to be much faster for the solving procedure than the CG solver, especially on mesh with cells with high aspect ratios (like the ones you should have in the boundary layer, near the profile skin).

Panos July 7, 2004 10:04

Re: Turbulence modelling
 
I think I have already a fine mesh. It consists of 28800 cells. I also wanted to know what values I can use for turbulence intensity and turbulence length. again thanks a lot.

Peter July 8, 2004 02:40

Re: Turbulence modelling
 
in order to know, if your mesh is fine enough at the walls, you should check the y+ values. For K-eps y+ from 30 to 100 is required, for Twolayer,LowRe or K-w SST y+ <1


All times are GMT -4. The time now is 12:25.