Friends, Romans, and Countrymen,

I try to simulate the transient flow after a valve opens. So essentially I have two regions with different initial pressures (i.e. the Riemann problem but with a 3d-geometry). Everything works OK when the pressure ratio between these two areas is small (<= 1.3 or so).

But what I want to do is to have one region at almost vacuum (1e-6 bar) and the other at up to 2.5 bar. I tried several things such as changing the fluid models, turbulence models, advection scheme, or time step control parameters to their most stable settings, but to little effect. When I look at the crash files I see that the temperature in some low pressure regions is in the order 1e8 just before the solver (understandably) crashes. Curiously, the density and pressure fields look reasonable, i.e., they show a shock and an expansion wave. The high temperatures are observed *ahead* of the shock wave, which is unphysical.

All residuals are still reasonable - probably because the density is so small there that the product \rho*C_p*T is still small such that the residual of the energy equation is also still small.

Decreasing the time step to ridicuosly small values (1e-8s) did not help either so I feel that there must be a better solution for this particular problem. Has anybody made similar experiences or can offer advice?

Thanks for your input.

Stefan

OK,

I ran the same geometry but with initial and boundary conditions such that the whole domain is at the same pressure and temperature, i.e., there should be no flow and no heat transfer. The solver still crashes after a number of time steps. Is there a problem with CFX and very low pressures?I set the reference pressure to zero and the fluid pressure (and the pressure given at the inlet and outlet) is 1e-6 bar.

Stefan