Hello, I work on an air flow around a square-based cylinder, I have to study a large range of Reynolds number (from 1 to 100 000), which turbulence model should I use in CFX ? I already tried k-epsilon bit it didn't give results expected. Thanks a lot

I think u should choose all of them to show the diffrences among them! How about it and have a good work!

Hi Paul,

That range of Re goes from stokes flow (very laminar) to highly turbulent. No turbulence model is going to be able to give sensible results for such a wide range.

Obviously the laminar model will be the one to choose up until the turbulence starts. In the turbulent region it will depend on the results you want (eg lift/drag or more detail than that). As Rogerio says best to try them for yourself and find out.

Glenn Horrocks

Hi, thanks for your response. My main issue concerns the turbulent region. With Re=3000, I have tried several models (Shear Stress Transport, k-epsilon..) and I have no vortex shedding! Consequently, I was wondering which model would be more relevant. Paul

Hi,

To get accurate vortex shedding you also have to have a good mesh, get the timestep right, have a 2nd order spacial and temporal discretisation scheme, sufficient convergence etc etc. Are all these things under control? Only then should you think about the turbulence model - and also keep in mind the different turbulence models require different meshes. It is not a matter of simply flicking over to a new turbulence model.

Glenn Horrocks