[gbettega]

Hello forum,
I did a run using a hydrid grid (prismatic layers at walls, and a fully tetrahedral mesh at the fluid domain interior). The mesh engine is custom.
SU2 tells me, about mesh quality:

+--------------------------------------------------------------+
| Mesh Quality Metric| Minimum| Maximum|
+--------------------------------------------------------------+
| Orthogonality Angle (deg.)| 51.2934| 89.7623|
| CV Face Area Aspect Ratio| 1.75245| 655.572|
| CV Sub-Volume Ratio| 1.01781| 5.38083e+08|
+--------------------------------------------------------------+

The last parameter has a very high maximum value: 5.3E8.
The solver starts and exits with residual divergence after a few iterations.

Question:

supposing the volume mesh is well defined (no intersecting elements, ...), in your experience could these very streched elements lead to this divergence, in a few iterations?

Thanks a lot
Giovanni

[bigfootedrockmidget]

It's difficult to say, you might be able to see where the problem originates by saving the residual information and checking the locations of high residuals. An aspect ratio of 1000 might be OK if it is in the direction of the flow and there is nothing special going on there. The sub-volume ratio is pretty high, but again it is difficult to say if this is the root cause. You can remesh and aim for a lower ratio and see if that helps. Since SU2 does not have a lot of 'tricks' up it's sleeve to keep the solver from diverging when the solution is locally misbehaving, it is better to keep the almighty gods of CFD happy by offering it your best meshes.