Is there anyway to mesh this shape into first order hex elements? I have tried all the various approaches I can think of with no success. This is a section of larger part that will be tied together and run in Openradioss
this is running in WINE but otherwise everything works normally
Perhaps it’s modeled differently or has some redundant edges. Can you share the file ? Here’s mine:
Revolved mesh.pmx (140.0 KB)
Thanks, your part works perfectly so we know it isn’t the software, there must be a geometry issue although I sure can’t see it. I will dig into that.
thanks
Luke
Sometimes revolution doesn’t work for specific larger angles. Sweep should work (and allows the Quasi-structured Quad algorithm for nice regular meshes), but has some requirements regarding the side surfaces and their edges (listed in the documentation).
1 Like
Revolve only works for revolved geometries in which the curvature of all revolved edges shares a common revolve axis.
Sweep should be able to mesh the model in more cases.
1 Like
Theoretically, the geometry shown here should meet this requirement (at least my version does - it was modeled via sketch revolution), but maybe the OP’s model has some small misalignment.
1 Like
Hi there,
I want to use this topic to ask, is it possible to do a full 360 revolution of a hex mesh? Often there are examples of tubes / profiles with revolutions < 180 deg.
If the part is split into two solids in Freecad I can’t get matching nodes for compelx parts. Is there a trick to slice the part properly in Freecad?
Thanks and best,
Stoli
Gmsh doesn’t support 360 deg revolutions, but you could revolve 180 deg and use mirroring and then merge coincident nodes:
If you use Slice apart and then compounding in PrePoMax, the mesh should be continuous, but only transfinite hex meshing can be used then (needs subvolumes with 5/6 faces having 3/4 edges each).
3 Likes
Great, thanks for the fast reply. I was thinking of rotating the part but couldn’t get it done properly with the “pattern”-option. The “mirror” did actually the trick!
Best,
Stoli
The mirror function should work better in general, since on the mirror plane, the nodes are always coincident.
1 Like