Hello everyone,

I am currently investigating flow separation and re-attachment over a rearwards facing step.

To date, I have been using the S-A turbulence model, because of computational efficiency, and in order to get a 1st order approximation of what I am looking for.

My current task is to improve the turbulence model selection. I considered the K-epsilon model, but literature states that it does not accurately simulate the separated region with the reverse flow.

Any suggestions? If I go to a Reynolds Averaged or a Large Eddy simulation, I doubt my computer could handle it !

Actually guys, thoughts on K-Epsilon with Non-Equilibrium wall functions would be appreciated.

you may try v2f model. it can handle this problem better than ke and sa but requires some more comptutational resource than both of them...

V2F - 4 equation, LES and RANS combined, right?

Am I correct in assuming it will take double the time of the 2-equation models, or will it take more?

it is based on rans. accurate tool for seperated flows and will cost slightly larger than a simple k-e model

you can find detailed info and test results and comparisons with k-e at:

http://www.turbulentflow.com/