[WilhelmCFD]

Hello.

I am working on a model of a wet scrubber with heat recovery in ScFlow. I simulate droplets with a Lagrangian approach and solve the fluid flow with a Eularian approach. Since there are large temperature differences in the scrubber, the fluid is solved as compressible. The simulation runs reasonably well but I seem to have a problem sometimes when the droplets get close to a wall. Unreasonable temperature fluctuations starts at single volume elements which lead to divergence and that the simulation crashes (Generally due to negative pressure).

The problem seem to be resolved if I reduce the length of time steps enough but since that leads to increased computation time I would prefer to not reduce it too much (it seems relatively stable at 0.01 ms).

Do anyone have any suggestions on how to mitigate this problem without reaching too short time steps.

Thankful for all input.

[Guvennergiz]

Dear Wilhelm,

I think 0.01 ms is normal due to complexity of this simulation. You can also pick Courant Number (should be <1) as type of time-step in Cradle with initial time-step definition.

I hope that you have solved this problem.

[WilhelmCFD]

Hello.

Thank you for your reply!

Yes, I have now settled on keeping the Courant number below 0.95 which gives me time steps of about 0.03 ms.

However, after some tinkering I realized that the largest problem was probably my Mesh. I had used CAD-files from the manufacturers of the investigated scrubber which were not optimal for simulations. I had tried to use Cradle software to adjust them, but in the end I had to learn the basics of AutoCAD and make new parts which works better for my purpose.

I still have some way to go to get proper results, but I feel that I'm well on my way now at least.

[Guvennergiz]

Hi Wilhelm,

So good to hear it. Also if you use Cradle 2021 maybe you can try Voxel Fitting mesher for fine meshing on bad CAD geometry. It can be solve your CAD cleaning problems.

Please let me know if you try it.

Cradle_VoxelFitting.PNG

[WilhelmCFD]

As of now, I'm working in the 2020 version so this feature does not seem to be available. I will update if I get access and try it.

[CFDfan]

Originally Posted by WilhelmCFD

Hello.

Unreasonable temperature fluctuations starts at single volume elements which lead to divergence and that the simulation crashes (Generally due to negative pressure).

You could "control" the divergence if in the analysis control you enable "reduce the gradient" (above/below a reasonable threshold) of the parameter that diverges.

[Guvennergiz]

Originally Posted by CFDfan

You could "control" the divergence if in the analysis control you enable "reduce the gradient" (above/below a reasonable threshold) of the parameter that diverges.

How can we do that in scFlow? Any screenshot for the exact location of relevant settings?