Fri, 12/09/2022 - 11:15
Forums:
Dear OCC Community,
I am currently exploring meshing using OCC and I'd like to know if I reached a dead end.
I am familiar with how tesselation works for OCC when you want to create an STL, for example.
However, is it possible to force the mesher to place Nodes inside using one of the OCC's algorithms?
(Something that you'd require for FEA analysis, although this is not why I want to achieve that.)
I do not want any sample code, but if it is possible could someone point me to the correct class?
If so, then eventually I will post my final working code here to help others :)
Any help is greatly appreciated.
UPDATE: See attachment
Kind regards,
Tamas Balogh
Attachments:
Fri, 12/09/2022 - 13:19
The simplest way is to use the class BRepMesh_IncrementalMesh:
Fri, 12/09/2022 - 15:02
Mikhail, thank you for response!
As I said, I am aware of the incremental mesh method, BUT, that does not mesh the inside of a flat face. Or can it do it somehow? Please see the attachment.
Fri, 12/09/2022 - 17:16
It cannot do it. Even if it could, how would it know in which place and how many nodes to place?
Fri, 12/09/2022 - 17:53
Based on an element size similar to FEA applications. I was hoping that beside linear and angular deflection it is possible to also do a redundancy tesselation inside the surface (even if deflection is zero). Now, if the surface is just a slightly curved, it would place nodes inside so I do not think it would be a big challenge programmatically (I am going to give it a shot at coding it myself unless I find a quicker way to do it).
I think mathematically it is a bigger challenge to mesh a curved face than a planar, I was hoping this extension is already available.
Fri, 12/09/2022 - 18:25
> "I am currently exploring meshing using OCC and I'd like to know if I reached a dead end."
You definitely did. For placing Steiner points inside a region, you would need either a different mesher (NetGen, GMSH or you may also ask OCC to provide you access to their commercial Express Mesh, but's not for free) or go a hard route refining your existing triangulation (split triangles, flip edges, etc.). In quite many cases NetGen is sufficient, although it's slow and sometimes distorts your triangulation.
Fri, 12/09/2022 - 18:30
Thank you very much for the detailed response Sergey! :)
I will close this thread and mark it as SOLVED.
Fri, 12/09/2022 - 18:35
It might be that OCC will clean this up as an ad, but I invite you to our CAD dev community here: http://analysissitus.org/forum/index.php
(To OCC moderators: it should be fine because it's a general-purpose dev forum, not focused on the OCCT library whatsoever.)
Fri, 12/09/2022 - 19:22
Thank you Sergey! I am a keen follower of your great work (if Quaoar is you)! Happy to join! :)