Calculix Suggestion

Hi,

I had a quad dominate surface mesh that CCX failed to expand. It took 40mins before it got to that point in the solve. I’m wondering if it would be possible to have the shell expansion as a separate process that also gave some mesh metrics like element shape quality. This way the mesh could be created ahead of time and reused. Right now CCX recreates the same mesh over and over when doing additional solves. Having the expansion be first would make the process like a traditional solve. Hope that makes sense.

Anthony

1 Like

Could it be that the mesh fails at small radius (smaller than thickness)? You could check this in CAD assigning thikness to the surface, and if the CAD operation fails, then is very prone to fail the mesh expansion also.

hi sergio,

the all tet version of the same mesh ccx can expand. i only get errors with the quad dominate version. i can run both meshes as shell models in a different program. if i try to extrude either mesh in a different program it generates errors. so not entirely sure what’s going wrong. however, this model brought to my attention the usefulness of having the ccx mesh expansion process be a separate thing up front. one, to catch problems sooner. two, to be able to re-use the mesh and not have to expand it every solve. i changed the mesh that took 40min to generate an error and got it down to around 7-8 mins. i’m not so concerned about the mesh error. this was just a test and using the all tet mesh isn’t a problem.

In my experience second order quads with nodes following geometry are poorely generated very often in small radius areas, and cannot be extruded/expanded. With *NO ANALYSIS card the mesh is not expanded?

thanks for the tips sergio. in netgen i don’t see an option to have the midside nodes follow the geometry. so i’m not sure what it’s doing there. i tried the *NO ANALYSIS option. unfortunately, that wasn’t very helpful. ccx seems to go through all the same steps and take a lot of time. it does stop at the solve. if the mesh fails nothing is shown. if the mesh succeeds i can’t view it very well in the program i’m using. so it would be better to just go through the full solve right now.

i think it is possible to generate midiside nodes in netgen,
but you can generate midside nodes with cgx, after meshing with netgen,
please check:
B.10 How to change the order of elements . . . . . . . . . . . . . . . . 146