Elmer Discussion Forum

Fantastic! I did not try loading the 827 with ElmerGUI yet, but I will put in the patch so I don't notice a problem when I do try.

Edit: I did try, and it crashed. Looking at the code, it's not so simple to fix. For example there's a line Which gives 8 for 820, but you can't get 9 from that. I can see how to add facemap9[] which would include the extra face point, but it takes a bit more TODO than I can figure out for the moment. I can always do 1st order, check my mesh with ElmerGUI, then go to second order and feed the mesh to ElmerSolver. For now, that's plenty good enough.

Hi

827 has probably not been used as much as 820. The ordering of the 20 nodes is the same, then then 6 face nodes are ordered in the order of the faces, and the last index is given for the center node. Also 409 is different from 408 just by the last center node.

I think 827 in the logic requires some flag to tell to populate the center node as well.

-Peter

I did this, in addition to the previous change, I could not find any additional changes required. The converted already supported 827

int facenodes = degree * faceedges;
if (code == 827) facenodes = 9;

Mine still crashes, but now you've got me wondering why. I'll chase down this error next Since there are 9 face nodes instead of 8, I would think we should add facemap9[], but maybe that has nothing to do with my problem.

Found my problem! I was compiling just fine, then running the old version because I'd forgotten to move the new ElmerGUI to elmerfem/install/bin/ElmerGUI. Doh! All works fine now - Thanks!

Thanx! I added the fix to "devel" branch. -Peter