How to tune up interfaceCompression and virtualMass in multiphaseEulerFoam?
 

OK, another brain-teaser for openfoam (2.1.x) multi-phase aficionados.

Here is my problem: I'm solving flow of multiphase mixture of tiny air bubbles in water (air:water = 1:9) flowing through a straight tube (diameter 26mm) at 50l/min (= 1.57 m/s). It is possible to solve the case with interFoam, but due to various reasons I need also to use the multiphaseEulerFoam (mpEF). I'm setting up the solver according to tutorial cases, i.e.
- standard boundary conditions (inlet+outlet: fixedFluxPressure and totalPressure for p, surfaceNormalFixedValue and inletOutler for U, walls: zeroGradient for p, fixedValue for U),
- 1st order Euler in time,
- vanLeer and limitedLinear in space,
- GAMG as solver for p,
- smoothSolver for U.
- PIMPLE (without outer correctors => no relaxation possible), nCorrectors 3, nAlphaSubCycles 3,
- laminar flow.

As I need to investigate motion and clustering of bubbles, I have to turn on the interfaceCompression (= interface tracking). Then the solver runs almost without problems and gives reasonable results.
Now comes the problem - as the virtual mass is also important in such cases (or it should be if I get the concept right), I've tried to turn it on (with a standard value of 0.5, which comes from theory). Now the solution start to diverge after ~ 0.01 s of flow even with Co=0.3. I've tried to play with various pimple settings (incresing number of outer correctors...), different meshes etc., but nothing helps.

Do you have any idea, how to put the interfaceCompression and virtualMass together?

I have also had problems with virtual mass--although in my case it is only seen when using MRF zones. In my case, though, virtual mass is not so important (liquid--liquid) so I have just turned it off (set coefficient to 0). I think Henry was working on fixing this and I can check.
-Kent

FYI, you might submit a bug report on it if you want--it is a known issue, but a bug report might help it get fixed sooner--especially since it is not a priority for me personally and I have been the 'customer' for the development of mpEF.

Thanks for the reply - I will post a bug report. So far, I run computations with virtualMass turned off. It is interesting however that even so the results correspond to the experimental data.

Hi Wardle

I want to model bubbly flow,can I USE multiPhaseEulerFoam?

thank you