For hex mesh what should be the convergence control relaxation factors. The default settings are too slow to converge. I think sI read somewhere that second order upwinding would be better. But I don't have the book with me right now. Any suggestions for fast convergence schemes for natural convection? Thanks.

We always increase underrelaxations for energy to 0.98 and body force to 0.85 or higher. These drive the solution in buoyancy and are most critical in natural convection problems.