hi, i tried to mesh of parts below. Look not too complex models and expected one of approach by sweep or thicken shell mesh will works. Unfortunately, even extrude mesh by thicken shell in FE model did generate expected results and shown hexahedral is broke up at some areas. Attached input files, thank you for looking into the problems.
I see that you imported a solid and hollowed it (converted to a shell by removing volume). Then you meshed the shell part and created an element set. Did you use the top or bottom surface of the shell part for that ? When I do it, convert it to part, and try to use the Thicken Shell Mesh tool in v2.4.6 dev, I get “Index was outside the bounds of the array” error.
I’ve seen a similar effect, and using first-order elements removed it.
I just turn off the mid node position on geometry and re mesh and it works ok (with shell elements), if you want a solid mesh you need to remove all the upper and lateral faces of the model in CAD and leave only one side, and use a thicken shell mesh definition.
basically, it’s a solid model i want to sweep (file attached). I do convert to shell feature and thickening shell in FE models since shell part not separated after conversion.
The whole procedure, including elset creation and conversion to part, and then using the Thicken Shell Mesh too ? Or just meshing the hollowed original part ? I keep getting the “Index outside…” error.
thanks for example files, theoretically both approach will yield the same result but thicken shell from Geometry of CAD shown expected results and consistent (linear, quadratic, offset)
If you use the midsurface, it works with both versions of the tool - the one in the Geometry tab (added as Mesh Setup Item) and the one in the FE Model tab (Model → Tools → Thicken Shell Mesh).
can it be specific in version being used? it seems not in latest development versions (v2.4.6) and mesh-based extrusion becomes more problematic in case of no offset.
I have looked into it and improved the solution. I could not follow all the problems in the conversion, but there are two ways in which PrePoMax computes normals, which are used to expand the mehs. If a geometry part of the mesh exists, the normals are computed from geometry (precise). If the geometry part does not exist (due to some features such as Create part from element set, Merge parts, …), the normals are computed from the mesh. The second procedure was too sensitive to small changes in positions.
Btw. are there any important differences between the two Thicken Shell Mesh tools other than the level (tab) where they can be accessed and the fact that the Mesh Setup Item one is more persistent and can be easily added/removed/disabled as it stays in the tree ?
first i did not expect this problem is fixed also in latest stable versions, many thanks it’s working expected in mesh result when i do test without offset and negative or positive value offset.