Hello, I am modelling Supersonic Combstionn with Fluent.

I started with a 2d case Coupled Implicit solver, and every thing was pretty fine. The CFL number had to remain under 1 though ( slow convergence, anyway).

Now I try to do the same with a 3d calculation, a,non-reactant flownd it is VERY Slow. I am currently computing a mixing case. What should I do. Is it a problem linked to Couple or Implicit solver? I have like 420 000 nodes for a fine grid and 120 000 nodes for a coarse grid. I use k-epsilon. I refined the coarse grid with gradient adaptation and with the refine grid there is this message at each iteration: " reversed flow in 44 faces on pressure outlet 3 turbulent viscosity limited to viscosity ratio od 1.00000+5 in 2cells"

That did not appear without refinement and it does not appear for the fine grid. Does it slow down the calculations?

Many thanks,

Alex

Fluent recommends SA model for high speed flows ... I have run with that .... if u are interested in turbulent kinetic energy/dissipation rate ... u can converge with SA and then switch to Kepsilon... it is possible to get unrealistic results about turblence production and viscosity ratios ... look only for the quantites u require and match with expts ...

I have faced same problem with turbulence predictions

reverse flow should vanish as the iterations progress