Extending step-54 to 3D volume constrained by CAD geometry

[Andrey Borzunov]

Hi, all.

I'm trying to create a 3D mesh from CAD file (stanford bunny) to solve a heating equation in this 3D volume.

I tried gmsh, but failed to produce coarse mesh, every try resulted in too fine meshes (700k cells and more).

Than i tried manually create initial-mesh with one cell (because `OpenCASCADE::create_triangulation` fails on 3D case). I used FreeCAD to examine .stp file, selected 8 edges and obtained its vertex coordinates using python console, and than used a code from Step14 Exercise_2_3 to create triangulation.

Finally, i failed to set manifolds correctly.

I have found similar discussions Bruno & Luca, yy.wayne & Wolfgang and wiki page but failed to understand the way i should modify sources (see attachment, it is slightly modified step-54).

Thanks in advance!

Best, Andrey.

 

[Luca Heltai]

I think you are almost there. Once you have created your grid manually, you should be able to set the manifold ids (which you did not do in your case).

tria.set_all_manifold_ids_on_boundary(1);

and then, after creating the manifold with opencascade, you should attach it to the triangulation:

tria.set_manifold(1, manifold_object);

After this, the triangulation is aware of your manifold.

The stp file you provide is not a three dimensional triangulation, but a two dimensional surface embedded on a three-dimensional space, and the opencascade utility won’t create a three-dimensional grid. Moreover, that utility is thought for single nurbs patches (that define internally a rectangular u-v space), and that’s how the grid is generated. In the case of a stanford bunny, the surface is triangulated already, so there is no way to define a unique u-v space and generate a Triangulation<2,3> out of it, and you have to proceed differently.

Ps: your procedure is ok, but remember that if your domain is topologically equivalent to a sphere, you should not start with a cube, but with a hypersphere (in the hypercube, the coarse cell has all faces on the boundary, and this will result after refinement in 8 cells that have three neighboring faces on a smooth boundary, with very high defromations, and give you distorted cells. If you refine instead the coarse mesh of a sphere attaching the manifold of the bunny to the its boundary, every cell only has one face on the boundary, and this will be preserved throughout refinement).

L.

> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en
> ---
> You received this message because you are subscribed to the Google Groups "deal.II User Group" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to dealii+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/dealii/46a757a1-33be-42a6-898c-d025246564c5n%40googlegroups.com.
> <step-54.cc>