[Variation of the solution as the time step varies. RANS/Channel/Moser]
 

Dear Foamers,


While testing the differents two equations eddy viscosity models with pisoFoam (modified) on a channel I experienced difficulties.

With the k-w models, the converged mean profile and turbulent kinetic energy are time steps dependent. According to this, I would expect better results as I reduce the time step, or it is the reverse. As the time step is reduced the wall shear stress decreases and therefore the pressure drop I apply (to maintain a constant mass flux over my stream-wise periodic length). Result is that the Re_tau is way to low... Target Re_tau is 395 as in Moser simulations.

http://img842.imageshack.us/img842/9412/plota.th.png

The corresponding tau_wall and Re_tau is given in the table below (as a function of the time step):

http://img573.imageshack.us/img573/5...donline.th.png



Uploaded with ImageShack.us

Does anyone experienced such results with the k-w and k-w SST models?

My mesh is refined in the near wall region and I am using omegaWallfunction to provide proper boundary conditions to omega.



PS: I am not observing the same behavior with "resolved" (without wallFunctions) LaunderSharmaLowKE. I therefore suspect the trouble from the boundary condition of omega or the numerics...

I'm using OpenFoam 1.6.x

any insight? :) I am really confused now...

Is it a good idea to increase to bulk velocity in order to obtain the right wall shear stress? I guess not....