I am fairly new in CFD. I wrote 2D cell-centered finite volume NS solver employing Roe FDS and Baldwin-Lomax model. However, I have covergence problems. Some research engineers compute eddy viscosity in every 5-6 iterations. Do you think it works well for better convergence? Thanks in advance...

Keanu.

Hi there,

I believe that a better way to smooth-out convergence problems is to include relaxation, rather than skip to compute the eddy-viscosity. Thus, compute the eddy-viscosity each iteration however update it as: mu_t_new=alpha*mu_t_new+(1-alpha)*mu_t_old This scheme can be applied to a number of variables. Another way to test is to compute a simple non-turbulent flow, and exclude the Baldin-Lomax model. There might be some problems with the momentum equation rather than the turbulence equation,

Cheers Jonas