[mecarlg]

Hi,

I'm running some 3ddp simulations around the naca aerofoil in the wind tunnel they used originally. for low angles of attack with the Spalart Allmaras model and a fine hex mesh (~4million cells) I get excellent lift prediction, drag is ok and convergence is good after about 3000 iterations.

However, as the angle of attack increases to 14-16 degrees, the convergence is not great - the residuals never level off, there are apparent low frequency errors and the drag monitor shows variation without any real convergence.

Any ideas? I have been working on this for months! Is it just that RANS is just no good when separation occurs? I've checked wall y+ values and my boundary conditions are fine. Any advice on how to tame the residuals would be very much appreciated - relaxing the solution has no effect. Also, k-epsilon models are even worse!

Carlos.