[alex_c]

Hi everybody,

I'm trying to simulate axisymmetric flow in a pipe with contraction (ERCOFTAC database, Case 58). I've used simpleFOAM, and got very nice results with RNG k-epsilon. I've tried some other models offered in simpleFOAM, and with RSTM I had some problems. Here they are:

1) When I use RSTM I have to specify Reynolds stress tensor (R) at the inlet, am I right? Should it be isotropic, or not? What about the outlet - is it pressure outlet?

2) I've tried with isotropic and non-isotropic R at the inlet, but after about 500 time-steps (iterations) the solution singularity is written in my shell :-). I suppose it's because thr time-step continuity error is increasing during iterations. Any solution?

Thaks in advance,
Aleksandar.

[hjasak]

Heya Aleksandar,

Doing axi-symmetric flows with RSTM is known to be more demanding form the point of view of convergence and stability than the k-epsilon turbulence variants. I would advise running a turbulent viscosity model first, using the application called R which will calculate the Reynolds stress and starting from that as the initial condition. You may need to change the boundary conditions on R before restarting - I don't think they get picked up.

There are no limitations on what your R should look like at the inlet. It would be nice if it were in equilibrium with the flow + satisfying thr Schwartz inequality but the code will run anyway.

I would guess you will need to lower the under-relaxation factors to get the code behaving in a stable manner, especially if you are starting from uniform R field. You can control the continuity error by tightening the pressure solver tolerance - a combination of the above should be sufficient to get the things working.

Enjoy,

Hrv