[JaZo]

Hello everybody

I´m quite new to the field of multiphase-simulation and would be thankful for any helpful advice. Here is the problem I´m working on:

Water runs through a swirl apparatus in order to marginalize and separate particles from the flow.
A single-phase simulation has been done first with k-epsilon and SST later with RSM which seems to match the measurement result better (pressure difference between inlet and outlet has been evaluated in the lab) but not accurate enough.
In the middle of the flow, right after the apparatus is an area of cavitation. Because of that I want to do a multiphase-simulation, first without any particle injection.

I checked the different CFX-tutorials dealing with cavitation and tried to setup my case similar to them. First doing a single-phase simulation, taking that as an initial guess for the twophase-simulation with cavitation turned off and then finally turn on the cavitation-model.

Boundary Conditions:
Water at 25C
Water Vapour at 25C
Inlet: Normal Speed ~ 0.5 m/s
Outlet: Opening, Entrainment with rel. pressure of 6000 Pa

So, here are my major concerns:

1. Is the RSM SSG Model the way to go? I read different things about that, on the one hand it seems that for multiphase flow it´s more difficult to get converged results, on the other hand the results from the lab try to make me believe it´s better suited for this problem.

2. How do I get a correct physical timescale? I looked at the shortest cell length divided it by a mean velocity and took a third of that value, like the cfx-modelling guide suggests. Is that a correct approach or do I misinterpret something here? I chose 1.2e-5 s

3. What´s wrong with the vapour fraction? The simulation seems to converge but when postprocessing I get max vapour fraction of something like 1.6e-15?


Thanks to everybody for reading through this wall of text, any helpful response or advice would be very much appreciated.

Regards
Jan

[ghorrocks]

If you are running a cavitation model then do not do an RSM model. Cavitation is hard to converge, and RSM is hard to converge, so if you run the two together you have no hope. Use an SST model, possibly with curvature correction with a cavitation model to give you a chance of convergence.

The physical timescale is what ever is required to make it converge. The comments about fluid tiem scales is just a starting point, you can adjust it from there. See the discussion on http://www.cfd-online.com/Wiki/Ansys...gence_criteria

Based on your maximum vapour fraction (1e-15) it sounds like you are not triggering cavitation. Are you sure you are pulling pressures low enough to generate cavitation in the simulation?

[JaZo]

Thanks Glenn I will give the SST-Model with curvature correction a try.

There is a large region where the pressure is lower than the saturation pressure of 3574 Pa or even negative so I guess cavitation should be triggered.