Hi, I am modelling a cavity (L/D=5) with flow speed M=0.85. It's a 2-d very fine quadrilateral mesh so I do not think I should need to refine further to solve my problem. I am able to get a converged 1st order unsteady solution but when I switch to 2nd order discretization, my solution blows up (specifically, continuity residual). I am using a segregated solver. I have tried decreasing under-relaxation factors (down to 0.3!) and time steps significantly (down to 1e-8!) and iterating, and still no luck.

FYI, I am using Fluent 6.2.16. Turbulence model is k-omega SST with transitional flow, compressibility effects and viscous heating options all turned on.

Any suggestions out there?? Thanks in advance, Regina

Yes!use the coupled solver.Luca