[mqsim]

Hi everyone!

I am looking for advice about turbulence models in OpenFoam. I have tried three turbulence models (k-omega SST, k-epsilon, Spalart-Allmaras) for the same case (study of the wind load on a building). At the beginning, I used the k-omega SST model but results were not enough accurate. So I tried the k-epsilon model but the calculation did not seem to converge. With the Spalart-Allmaras, model results are ridiculous. The stagnation pressure is approximately 50% higher than expected.

The geometry is complex so it is really difficult to create a good mesh using snappyHexMesh. I know that differences between turbulence models could be due to the mesh, but it seems I cannot improve the quality of the mesh with snappyHexMesh. I cannot show the mesh or the geometry, it is for an industrial case.

Regards,

mqsim

[AnasCFD]

Hi mqsim,

You are comparing 3 different turbulence models. k-omega is accurate but requirs yplus ~ 1. The standard k-epsilon is not that accurate but usually it converge easier. On the other hand Spalart-Allmaras is not that accurate since it is a one equation model.

I suggest to check your yplus. Then compare the results obatined with k-omega (with yplus ~1) with that obtained by realizableKE model.

All the best and success.

[mqsim]

Hi anas,

Thank you for your answer. I am not sure to understand what means "yplus". Is it the size of the mesh in the y direction?

mqsim

[AnasCFD]

Please take a look to this wiki

http://www.cfd-online.com/Wiki/Dimen...e_%28y_plus%29

[mqsim]

Thank you Anas! Hopefully it will help me to improve my models.

[Tobi]

I just want to mention that

kEpsilon vs kOmega:

kEpsilon is more accurate far away from walls
kOmega is more accurate at walls

Therefore the kOmega-SST model was introduced which at least blends both models

[mqsim]

Thank you Tobi for your answer, it is really useful to me!