CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Siemens (https://www.cfd-online.com/Forums/siemens/)
-   -   crack at interface of hexahedron & tetrahedron mes (https://www.cfd-online.com/Forums/siemens/52870-crack-interface-hexahedron-tetrahedron-mes.html)

KS April 1, 2002 16:52

crack at interface of hexahedron & tetrahedron mes
 
Hello,

The geometry was meshed by using hexahedron meshes while a small part was meshed with tetrahedron meshes. At the interface of these two types of cells, cell faces match nicely. vmerge merged all multiple vertices. However there is a crack region all over the interface. I think this shouldn't be there. Anybody has any idea about this issue? Thanks in advance for any response.

KS

cjtune April 1, 2002 21:43

Re: crack at interface of hexahedron & tetrahedron
 
I think you need some kinda coupling, either arbitrary or integral. I use ICEM-CFD for creating hybrid meshes and at the interfaces beween the hexa and tetras, ICEM always creates pyramid elements for 1-to-1 connectivity. 1-to-1 as in cell surface, not just vertices. Note that you need at least 2 tets to cover any one side of a hex (2 triangles = 1 square). I can't comment on Pro*AM as I haven't used it for anything other than trimmed cell meshes.

cjtune April 2, 2002 04:04

Re: crack at interface of hexahedron & tetrahedron
 
You need a 1-to-1 match not only with the vertices but also with the surfaces. Since you can only cover on side of a hex (square) with two tets (triangles), I suggest you try arbitrary or integral coupling. If you want perfect matching, you need pyramids as intermediate cells between the hexs and the tets. I use ICEM-CFD for hybrid meshes and this is the way it matches hex and tet domains. Not sure about how Pro*AM handles hybrids meshes...

cjtune April 2, 2002 04:07

Sorry, internet connection lag... (n/t)
 

steve April 2, 2002 13:35

Re: crack at interface of hexahedron & tetrahedron
 
You need to add cpmatches at the hex/tet interface. If they match exactly (ie 2 triangles to each quad face) then you can you integral cpmatches. If not then you need the arbitraries.


All times are GMT -4. The time now is 05:52.