[cutter]

Hi,

I'm observing nonphysical results when using twoPhaseEulerFoam on non-orthogonal meshes. I'm trying to simulate a long tank with rigid lid and aeration (usage of a simplified geometry in order to track down the problem):
geometry.jpg

The inflow of air is supposed to introduce large vortices and lead to mixing of the fluid. These results can be observed when using complete orthogonal hexaedral meshes:
orthogonal_mesh.jpg orthogonal_U.water.jpg

Whenever I try to use somewhat skew or non-orthogonal meshes the whole simulation either blows up or yields wrong results (wrong direction of vortices!). There seems to be an artificial source of momentum:
non-orthogonal_mesh.jpg non-orthogonal_U.water.jpg
I already tried to use non-orthogonal correction loops and different numerical schemes. This issue probably relates to the following bug report: http://www.openfoam.org/mantisbt/view.php?id=1379

Please find the two sample cases illustrating the problem within this Dropbox shared folder: https://www.dropbox.com/sh/eqoaca7b9...fPLSKSiha?dl=0

None of the tutorial cases concerning air-water simulations seems to use non-orthogonal meshes or non-trivial geometries (except mixerVessel2D). Are there any example cases that match my use case closely?

Any other comments regarding case setup and boundary conditions are welcome too! I will happily provide further infos that may be required to shed some light on this issue.

Thanks
Cutter

[sharonyue]

I would like to say this this is related to mesh, I happened to run across this problem long long time ago, this is still not solved, The only solution for me is to use hex mesh.

I downloaded your case, there is no files under 0 file. n which openfoam edition r u using?

And maybe u can just try this simple 2D case. pls see the mesh attached. Anyway, from my experience using twoPhaseEulerFoam, this solver is sensitive to mesh.

[cutter]

Hi,

thanks for the feedback. I will have a look at the 2D mesh.

The boundary condition files are located in the init directory. Please copy them into the 0 directory before running the solver. See run_simulation.sh for more infos! The usage of the init dir allows to recover the boundary condition files at a later point in time. They sometimes get modified by tools like setFields or screwed up by the user. This way there is always a working backup. Some people use a folder called 0.org for the same reason.

I did the tests with a recent version of OF 2.3.x.

Sorry for the short reply - I am not in the office this week. More infos next week...

Cutter

[sharonyue]

Hello!

Have u ever solved this problem?

Best,

[cutter]

Hi,

sorry for this very late response, I've been busy with some other projects in the meantime.

The problem has (at least partially) been fixed with the patches in OpenFOAM-dev (see https://github.com/OpenFOAM/OpenFOAM-dev) during the first months of 2015 (there've been quite a lot regarding twoPhaseEulerFoam directly).

The solver is now capable to produce plausible results for the simple test case I provided in the first post of this thread and meshes similar to the 2d samples of post #2. It is also able to run much more complicated cases using irregular hex meshes produced with cartesianMesh (part of the cfMesh grid generation suite). However it constantly crashes with a similar case which differs in geometry size and some load parameters only.

Many thanks to the OF team for the ongoing development effort!

Cutter

[Mehrez]

Hello Foamers,

Thank you for this thread and for the great job you are doing.
I am facing the same stability issues using multiphaseEulerFoam (OpenFoam 2.3.0) with a mesh of 99% hex elements.
Is this problem solved or how can I run stable simulations using this solver (or twoPhaseEulerFoam) ?
Thank you for your answer.
Best,
Mhrz

[masoudsh]

Hi Cutter

I use twophaseEulerFoam openfoam 2.4,I have problem in my case
Alpha became negative,and epsilon increases and then deltaT is reduced very much
Do you khnow how can I fix that?



Best Regards

Masoud

[cutter]

Hi,

please check OpenFOAM 3.0.1, there've been many important changes in the 3.x.x branch that solved some of my own issues.

Also check the latest dev versions (https://github.com/OpenFOAM/OpenFOAM-dev), there's currently quite a lot development regarding the twoPhaseEulerFoam solver.

Cutter