[mapeth]

Good Morning everybody,

I have been working on a VoF model with interDyMFoam for a while. It is basicly a mixer with fluid injection from below and an outlet at the side. At the top a gas flow from a gas inlet to an outlet is set. The mixer rotates at 500 rpm. The BCs are inlet vel at liquid inlet, outlet vel at gas outlet (I know thats not ideal), pressure at gas inlet.



I can get this model running, but after some time (ca 0.005 - 0.1 s) it crashes. The velocities start to increase at one random single point without clear indication. I observed that the p equations iterate to eternity just before it happens. What I have tried so far:

1. no BCs => all walls; same problem
2. better initialization; still crashed
3. small Courant no below 0.5
4. improved mesh quality, now completely structured
5. no fluid, just gas; nope
6. various solvers for p eqn and varying their parameters => GAMG, PCG; non of them worked any better, I think PCG is worse

What I need to mention, the location where the calculation blows up is really random. It is never at the same location. Therefore, I think its a numerical issue.


Does anybody have an idea what else I could try to solve this issue Thanks for any help.


Cheers
Mat


P.S.: I forgot to mention. for the inlets/outlets with fixed velocity I used fixedFluxPressure BC