[christinasmuda]

Starting from the tutorial mixer2d I tried to do simulations of a different geometry with moving mesh and sliding interfaces. The calculations ran well but unfortunately I get very different flow fields for a coarse and a fine mesh. While the calculation with the coarse mesh results in the expected flow field (first picture below), the simulation with a finer mesh ends up with a wrong solution (second picture). I used the same settings for solver and numerical schemes as in mixer 2d. With adjustable time step Co is < 0.5.

I tried to change some solver settings, numerical schemes and a smaller Co, but I didn't get to a satisfying result with a finer mesh.

Could anyone please give me a hint, which settings I might have to change in order to get the physically right solution?

Thanks a lot,
Christina

coarse mesh:
fine mesh:

[christinasmuda]

Sorry, my first shot with pictures didn't work. Second try:

coarse mesh:
fine mesh:

[christinasmuda]

Again I got a Server Time Out. So here's a link to the pictures: link

[hjasak]

I do not see anything terribly wrong with the solution: it seems the fine mesh is showing transient behaviour (try making a movie) and it would be brave to say this is wrong. Looks to me like you are having moving vortices in some of the cavities...

Could you check the boundary conditions (are you uisng movingWallVelocity?). Also, what happens if you let it run longer?

Hrv

[christinasmuda]

Dear Hrv,

thanks for your quick reply. I thought about it and tried some more simulations - but I'm still not really convinced. I agree, there might be transient vortices arising. But shouldn't they appear periodically at every pin of the rotor? The viscosity is very high (10 Pas), so the flow is completely laminar.

I put two films at this Link. The first one showing the velocity distribution and the second one the pressure distribution.

Looking at the pressure distribution, in my opinion the pressure is supposed to be the same in the whole flow field using an incompressible fluid (maybe minor changes close to the pins due to the rotation). But during some time steps the pressure in one half of the flow field is very different to the other half.

I also tried the same simulations as a steady state simulation with simpleSRFFOAM (fine mesh). It converged without any problem to the same flow field, the coarse mesh with icoDyMFoam showed. Afterwards I built icoSRFFoam from simpleSRFFoam in order to do a transient calculation with moving reference frame. Here the same problem arises as with the icoDyMFoam: the transient simulation shows a "strange" flow field with no periodicity (neither in space, nor in time).

Best regards,
Christina

[deepsterblue]

Looks like spikes in the pressure. Is this animation plotted for every time-step? Or is it at intermittent time intervals? Since you're using icoDyMFoam, I'm assuming that there are topo-changes in the mesh. Do you see the pressure variations immediately after a topology change?
Also, how many PISO correctors/non-ortho correctors are you using?

[christinasmuda]

Thank you very much for your answer. I tried increasing the number of correctors and now it's converging perfectly to the expected flow field.

Thanks for your help,
Christina