I managed to derive and calculate the stiffness matrices for 3-node (flat) and 6-node (curved) triangles. That in itself was a nice challenge and learning experience. However, as expected, the stiffness matrices are fully coupled (displacement in node i leads to a force in node j), so would need to be lumped to defined simple springs for export to CCX. This is not difficult for the three node triangle, but for a general curved 6-node triangle can lead to negative stiffness values. Clearly this is not desirable (if at all possible in CCX).HarryvL wrote: ↑Sun Jan 13, 2019 8:04 amI will derive the elastic stiffness matrix for a 3 and 6 noded isoparametric face element that matches the 4 and 10 noded tetrahedral volume elements generated with GMSH. This can either be used to define SPRING1 elements (by lumping stiffness on the diagonal of the matrix) or in a fully consistent form (in a Calculix UEL.f subroutine).
I conclude that the (my) effort is better spent on implementing the fully coupled/consistent stiffness matrix.
PS: this stiffness matrix can also be used to define an interface element (as I described here: https://forum.freecadweb.org/viewtopic. ... 20#p260275