Closed krober10nd closed 3 years ago
@jorgensd I also added a delete_low_quality_boundary_cell
step at the very end before exiting generate_mesh
.
The order of the delete boundary cells can be fiddled with. I suspect there's probably an optimal order but as it is in now, it's a significant improvement in minimum cell quality than what it was.
Here is the BP2004 benchmark ran again with the new changes
and here is the disk benchmark (a 40-50 % improvement in minimum cell qualities).
It will be interesting to see how this changes @nschloe benchmarks in meshgen_comparison
.
okay, after rearranging the location of delete_boundary_entities
around the laplacian smooth, min. qualities appear more stable.
for the BP2004 model.
141