Two-layer K-E model for a diffuser
 

Hi

I am modelling a axisymmetric diffuser with a two layer Turbulence model in STAR-CD. in the near wall boundary region the mesh is denser and y+ has to be between 30 and 100. Do i have to refine the mesh using cell refine command or is there any other way of doing this. I used the norris model for the near wall boundary. Where can i find any information on this. Can anyone help please.

Thank You

Saleem

Re: Two-layer K-E model for a diffuser
 

I am not sure of your question. If your y+ values are between 30 and 100 you should probably be ok. There is a paper by Gianluca Iaccarino, Prediction of turbulent flow in a diffuser with commercial CFD codes (can be found on the internet). It should be noted that one of the important factors is if there is seperation off the wall of the diffuser. For example, in calculations I have done on jetpumps, there was no seperation and k-epsilon models were able to predict the pressure drops fairly well. According to the paper above, that won't happen if you do have seperation and the v2-f model by Durbin may be necessary. Tom

Re: Two-layer K-E model for a diffuser
 

Many thanks. Actually i want to know for the refined mesh near the wall boundary i have to use Cell refine command or there is any other command. i used cell refine command to have a denser grid at the wall.

Can you please tell me the right way of doing this.

Thanks

Regards

saleem

Re: Two-layer K-E model for a diffuser
 

I believe the user guide recommends that for a two-layer model the y+ at the wall should be approximately 1.

crefine & cmrefine are two useful commands for refining the mesh. The on-line help desribes their usage.

Re: Two-layer K-E model for a diffuser
 

hi,

I can say from my experience that Mentars SST model is suitable for any diffuser flow and Transonic flow than any other model.U know it works (SST) basically on combination of K-omega (near wall sublayer) and K-Eps far from wall. U know in diffuser there is sudden change in area because of which K-eps(low or High Re) could not predict the recirculation (Pk=0) but K-omega predicts well.

so dont use K-eps model and recent v2f also go for SST which is best

Ragards

Alex