[ericclaro]

I started doing some basic external aerodynamics analysis of a rotating wheel, with the intent of having some reference values for a full-vehicle simulation I'm goingo to make later. I had no problems converging the simulation to reasonable values with a coarser mesh, but when a finer mesh was used, the lift coefficient converged to an absurd value.

I don't know if perhaps I'm using a bad turbulence model (k-e realizable), but i don't think it would interfere so much. Perhaps a transient solution would work better.

The finer mesh is surely a bit exaggerated, I just wanted to make a mesh size comparison, taking into account different computational times and number of iterations needed until convergence was reached.

I will upload a link to the pdf with the investigations made so far, I really appreciate any suggestion on how to improve my solution method. I hope there will be no problem to access the pdf.

here it is: https://www.dropbox.com/s/fac3ni5pls...0Road.pdf?dl=0

[LuckyTran]

I would try to reduce the cell skewness, even 1 cell can cause inaccuracies. Even if numerically it is not unstable that does not guarantee that it will convergence to correct answer.

I noticed your Courant number and under-relaxation factors are very low, my guess is that it is needed to achieve convergence. Once you have obtained your "converged solution", you can try increasing these parameters to see if the solution suddenly blows up, which is a hint that your solution isn't actually converged yet.

[ericclaro]

I do believe that one cell with a skewness higher than 0.9 would cause innacuracies, but from 0.6 to 0.06???

About the relaxation factors and courant number, it would be an interesting attempt to increase them, but how much? And is there a recommended limit value for external flow analysis?

Thanks for your reply!