Hello,
I have a blade volume to fill with hexaedron mesh elements.
I prepared the geometry to get 6 faces (trimming the leading edge), but I only got tetraedron (even with combine option enabled).
Please explain what is wrong. Here is zip of the pmx project file and initial step solid geometry file.
wing_tip_solid.zip (2.9 MB)
Transfinite meshing can generate hex elements only for volumes with 5 or 6 faces (having 3 or 4 edges each). Here you enabled the “Allow pyramid elements” option so it used them or just tetras. Pyramid elements consist of either collapsed wedges or collapsed hexahedrons (there is a setting for that) because CalculiX doesn’t have pyramid elements yet. Without this option, the algorithm won’t work.
If you want to get hex elements, you will need to ensure that this part meets the criteria for the number of faces and their edges. You could try partitioning (splitting) the volume this way.
This small face is the problem here:
Without it you can even sweep mesh this part:
blade mod.pmx (2.2 MB)
Many thanks. I prepared the solid geometry for having 6 faces, but I miss the tiny internal one. Good catch.
Could you please explain how the sweep works in this case (for kind of ‘freeform’ volume to be filled) ?
And do you think it would preferable compare to transfinite meshing (in terms of elements density/quality) : I was for example wondering how to control the number of elements layers through the thickness, using the transfinite meshing technique.
Sweep works in a similar way to the equivalent CAD operation - it sweeps the mesh from the selected source surface(s) along the side surfaces up to the target surface. It has some limitations mentioned in the documentation and here: A new developer/preview PrePoMax version v2.2.3 has been released - #2 by Matej
So it’s similar to extruded meshing, but supports also the Quasi-structured Quad algorithm (for more regular quad meshes on the source side) and curved path (extrusion is done along a straight line).
Transfinite meshing can provide very good regular meshes, but its rules often make it necessary to do a lot of preparation (volume partitioning) in CAD software. Thus, it’s often easier to use sweep meshing (no volume partitioning, but there are rather strict rules for side surfaces too). In this particular case, both techniques can provide pretty much the same mesh. That’s also because sweep meshing uses transfinite algorithm for the source surface meshes in the case of 3- and 4-sided surfaces by default.
Regarding the amount of control you have, sweep meshing lets you adjust the source surface meshing algorithm, but (unlike extrusion) doesn’t provide controls for the number of elements in the sweep direction. You could try using the Local Mesh Size for that, though. Transfinite meshing is fully automatic.
This thread shows an interesting case study: Problem with sweep mesh on a spring
Thanks for your feedback. I was able to remove the tiny face directly in prepomax, using the defeature tool. Great.
Could you please tell me now if it is possible to extract shell elements from sweeped volumic elements (selecting volume element face, and propagate selection using angle tolerance) ? I would like to create shell mesh for upper and lower faces of the blade (sandwich structure).
Yes, that’s what I did too. This OCC tool is really helpful.
Not directly (there’s no such tool, only the opposite is possible with Thicken Shell Mesh), but you could use e.g. this workflow:
Here it’s offset just for visualization (using the Exploded view tool).
Great. Thanks a lot ! very valuable.
And definitely great software.
Sorry, just a last question about same topic: I made a modification on my solid, but it is not conform (seems to have more than 3 or 4 faces). The geometry analysis (small faces) does not highlight anything wrong this time.
Could you please tell what is wrong with it ?
Thanks for your help again.
wing_solid_all_simplified3.zip (275.8 KB)
Thank you.
My bad : I shared the wrong stp file. Here is the good one (with no small faces), revision 4.
wing_solid_all_simplified4.zip (263.3 KB)
Oh yes, thanks. I made error between different stp files (so many testings…).
Sorry for this, and may thanks again for your help and quick reply.
