At times conjugate heat transfer simulations can diverge and the root cause of divergence is unclear. Often this can be caused by the mesh; meshes with very poor mesh quality tend to diverge. That said, divergence is almost exclusively caused by poor quality cells in the fluid domain; divergence is not caused by bad cells in the solids. This is because conduction heat transfer is extremely stable and linear in nature.
Currently, SimScale’s standard mesher Meshing Log does not report the mesh quality metrics based on fluid vs. solid; the metrics reported are for the entire global domain. For example, in this project I see my Maximum volume ratio is 52.
volumeRatio
Acceptable range: 0 to 100
min: 1
max: 52.11944373558865
average: 1.4139411491717289
Unfortunately, I can’t tell if the high volume ratio is in the solids or fluid. The Mesh Quality viewer can be helpful to find areas of poor mesh, but on very large models it becomes cumbersome to find isolated slivers of poor mesh in a large domain. Instead, I use CAD mode to isolate the fluid domain, mesh it, and check the mesh quality.
The above volume ratio statistics were for the U-Tube Heat Exchanger tutorial model. I isolated one of the fluid sides via CAD mode, meshed it via Incompressible flow, and got the following Vome Ratio metrics:
volumeRatio
Acceptable range: 0 to 100
min: 1
max: 10.000000000000451
average: 1.3657499989289554
Now I can see that the poor mesh quality is in the solids, and therefore the poor mesh quality is likely not triggering the divergence.
Please let me know if you have any questions about my method or approach.
Thanks,
Matt