Learning Forum

[Lucy]

Hi!

I'm relatively new in CFD. I'm trying to model the multiphase flow through a thin (5mm) porous internal using a porous media model. The simulation converges nicely, but then suddenly crashes. Looking at the gas volume fraction contour I can see that in some regions the fluid flows freely (the porous media model does not create any resistance), but in another, gas gets stuck under the object and there is no flow through it. This creates a kind of "dead zone" after the object. I think that causes deconvergence, but I don't know how to solve this problem. I couldn't find any similar problems anywhere. I tried different meshing strategies including local sizing in porous zone regions. The time step is small (0.001). The mesh there is very fine (max.1 mm cells) The quality of mesh is good. The model is as simple as possible.

Does anyone has an idea what can cause such issue? I truly appreciate any ideas.

[Rob]

Please can you post some images? 

[Lucy]

[Lucy]

[Lucy]

Thank you Rob for such a quick answer. More or less it looks like this. I have a similar problem with different meshes so I think that it could be more set up. Generally, the gas phase does not cross the porous media model in some regions.

[Rob]

Do you have any interface zones? Also, can you zoom into the porous region and show the mesh? The checks will tell you the cell quality, but if you've got a huge jump in cell size between zones it will cause problems. 

Please can you plot volume fraction? 

[Lucy]

Here is the volume fraction with the mesh. The min. orthogonal quality is 0.35. I did not define any interfaces in the fluent. 

[Rob]

The cells are OK but the mesh isn't. One of the requirements is aspect ratio and another is growth rate.  I suspect you've got a change in flow (ie a gradient) in the left-right direction and it's then causing the solver to fail. 

Have a look at porous jump (it may not be suitable here) and also some of the meshing videos/tutorials to see what's been done there. As an aside, for multiphase I tend to avoid poly-hexcore and just use a poly mesh to avoid the jump in cell size in the free stream. That's fine for aero models where they can refine wakes ahead of time but not so good for multiphase models where we generally aren't as sure where stuff is going to go. 

[Lucy]

Hi Rob! Indeed polyhedral mesh seems to provide more stable simulations. Thanks for this recommendation. I also decreased the growth rate which solved the problem that I described above. However, now I see that there is another issue with no difference in a gas fraction before and after the porous media model. As porous zone causes resistance I expect to see such a difference. The model converges nicely now and I don't see any discrepancies that would cause such inaccuracy in results.

Do you have any advice regarding this one? Thanks a lot for helping!

[Lucy]

Thank you for your valuable feedback. I will look into those videos but as far as I know porous jump unfortunately won't be suitable for my case. I'll also change the mesh to poly, decrease the growth rate and let you know how it works. Thanks again for your help! 

[Rob]

What are you seeing now? Why would gas fraction be different? 

[Lucy]

Thanks for the quick reply. Generally, gas fraction during simulation is the highest in the porous zone and higher above the porous zone than below. As the simulation runs, gas fractions after and before the porous zone become equal.

I expect gas fraction to be different based on other research and as the porous media in the model is supposed to cause a physical obstacle hindering the flow of gas. 

[Rob]

If you set phase specific resistance coefficients and the gas has somewhere to go along/away from the porous media then, yes it will. Otherwise the gas has to go somewhere! 

[Lucy]

Hi Rob! As you said, I decreased the growth rate and changed the mesh to polyhedral. In the beginning, everything was fine and I thought that it solved the problem. However, then I lengthened the simulation time and the error with gas phases stocking before the porous zone came back. The growth rate is only 1.05 and the aspect ratio is 8,42.

Furthermore, I am getting the upward flow near the walls and downward in the centrum of geometry, while from experimental data it should be opposite. Before the porous zone, the flow is correct.  

Do you think that it's still the mesh? If I decrease the growth rate even further then I'm getting mesh over a million. Now, I'm using 250k. 

[Rob]

Please can you post some images, phase, pressure and velocity would be helpful. Depending on volume fraction etc you can get liquids collecting and then dropping down in a porous zone in the experiment and gas then diverts up due to zone pressure effects. If the porous zone is very thin with low resolution in the CFD model you may not see that. 

I assume gravity is pointing in the correct direction?