Slow very localized divergence
 

I am running an external aero case with simpleFoam and kOmegaSST (although my problem seems to be unrelated to the turbulence model, I see the same behavior if I disable turbulence). The solution looks very good... I am getting the same drag coefficient that I got in Star-CCM+ with a similar mesh and settings.

However, there are always a few cells, like on the order of 2-10, that will slowly get more absurd velocity and pressure values. I am having to stop the run every 50 or so iterations to delete these cells to avoid a FPE. Every time I delete the cells and restart, a whole new set of cells will start behaving similarly and I have to repeat the process. I have tried slightly different mesh densities on this geometry and I get the same behavior. All highly skewed cells are deleted before I even begin the run. I have switched to very conservative convergence criteria and enabled cell limiting.

This is about the forth geometry I have run with similar settings, but the first to have this problem. Is this a common problem to run into with OpenFOAM? If not, any advice on how to proceed would be great.

This sort of problem is often caused by unphysical boundary conditions. Can you share your case directory? If the mesh is too big, strip off everything except for the mesh description or the CAD geometry.

Patricia