Thermal Simulation

[Anonymous]

Getting errors with this simulation.  Can Anyone help me?

http://a360.co/2DeO8GN

[henderh]

Hi Big_Mak,

I was able to get the solve to complete without errors (attached).
After using the Interferences command in the Simplify workspace, it was discovered that the 'baseplate' body was duplicated.  After removing that (and the other small components that were made invisible in the BMCH-000017 sub-assembly) we still couldn't solve without errors.  Finally, I created a new Thermal study, then copy / pasted the loads from the original study, and it solved without errors.

We'll look into why the original study didn't solve after the intereferences were removed (ARRO-9323).

Thank you for bringing this to our attention,

[Anonymous]

I was able to get it to solve in the cloud. Locally I was having issues.

Thanks for looking into this.

[mcmillr]

Hi Big_Mak, I wanted to add a few more comments about this issue.

First of all, I noticed you have many 'hidden' components. Are you aware that we actually mesh and solve hidden components?

I suspect the reason you needed to change your mesh settings was because your model would not mesh with the defaults. The meshing issues are actually happening on the 'hidden' components, so if you didn't intend for these components to be included in Simulation, you should just go to Simplify and delete all of them. This will NOT affect your original Design just your Sim studies.

Then you can change your mesh settings back to the defaults, and the Simulation will solve just fine and run a lot faster. That is the reason that Hugh was able to solve once he created a new Study, the mesh settings were reverted to defaults.

The solve error that was happening on your original Study with the fine grained mesh setting is due to some distorted tetrahedrons being sent to the solver. I consider this a bug, we should fail to mesh, send good mesh to the solver, or the solver should correct for (slightly) bad elements if this is possible. We should never fail in the solver due to bad mesh geometry.

As to why it succeeded when it solved in the cloud, it could be that the mesher generated slightly different output on the cloud compute worker, and this was just enough difference to avoid the solver detecting 'bad' mesh elements.