Coreform Cubit 2022.9 RC (yesterdays build) I’m seeing some “overcompensation” regarding the spline fitting, even with the creasing option you taught me yesterday. The below image shows the triangulation facets in blue/red (lets call that the true geometry), the blue thin lines are the edges of the hexes in thee mesh geometry used as the basis for spline generation, and finally in white we can see the IGA elements as viewed in paraview (2d slices in the same location)
The bleeding through of the spline into the gap I’m surprised at since I would’ve thought it shouldnt bleed through the linear side of the element. Any suggestions?
If I’m understanding your image correctly, the reason is that we don’t support “non-trivial” creasing on volumes, yet. A “trivial crease” would be the case where the U-spline topology requires a crease (e.g., nearby extraordinary point, is the “free vertex” of a hex). Here’s an example of a swept volume mesh that exhibits the issue I believe you’re seeing:
The “free vertex” are the four sharp corners on top and bottom.
If I change the surface mesh scheme (in this case to polyhedron) I can modify the layout of the mesh so that I have a “free vertex” on those sharp vertices.
So I can use polyhedral surface mesh, which I presumably then do a 3d poly mesh from? Could you attach an example? Would an alternative to be webcut across the inflexion point?
The polyhedral surface mesh would be swept. You are correct though that a webcut across the inflection point would also do the trick – and has the benefit of being more repeatable/robust.
