Optimize Simulation

pd_bbe · June 10, 2021, 2:26pm

Hello,
the Simulation Iam doing is very time intense. Are there general tips to reduce the simulation time?

I already tried to coarse the mesh, but if i do this, the Forces getting very high and the simulation is not working. On the other hand I tried to reduce the iterations, but I want to have see that the process is converting.

Is it just normal or is my general setup too fine? Maybe you have some tips for me, that would be very nice!

Philipp

ggiraldof · June 10, 2021, 2:35pm

Hi Phillip!

You have already tried the main ways to reduce computation time, but I agree with you that you can not go further this way.

One thing you can try is to switch from ‘Automatic’ to ‘Manual’ relaxation in ‘Numerics’. High relaxation factors can sometimes speed up the convergence, but be cautious, because they can also lead to divergence.

dschroeder · June 11, 2021, 10:52am

Hey @pd_bbe

i have some general tips for you.

Your meshes seem to be over 20 million cells. This is WAY too high for a half car simulation with only a diffuser. This is your main problem with the long simulation times. As your geometry stands, you should be around 5-8 million cells.

The main way to reduce this cell count is to reduce the size of your region refinement box or reduce the fineness of these cells. One good method is to switch to a “distance region refinement” which increases mesh fineness only around your geometry and doesnt create a box.

Also you could reduce the fineness of the bounding box as well as finer cells are only needed closer to the geometry

You could also play with your feature refinement angle to try and reduce the unecessary cell fineness on the mono however the main cause is the region refinements.

CFD meshes are very sensitive to bad cells… which are almost always related to problems in CAD geometry. Only a few bad cells can cause divergence or have the mesh fail completely. I would add a radius between mono & diffuser and also a higher surface refinement to the two wing elements to avoid mesh degredation. Checking the mesh thuroughly helps a ton in saving simulation time.

Finally, the meshing log will tell you if simulation should be attempted. A high volume ratio means there are bad cells somewhere. (should be under 100) Also you are fairly close to the 90deg limit of Non-ortho cells. The mesh qaultiy feature is EXTREMELY USEFUL in finding bad cells. Check it out.

Let me know if you have any more questions. The diffuser looks awesome by the way.

Viel Gluck!

Dan