I ran into an issue trying to hook the stressRefine to another finite element code, which is written in c. Global “wrapper” functions which and be called from c are required. I am busy creating these, and after testing them they should be available in a few days and I’ll add them to the source. I’ll also provide the same functions callable from Fortran (using standard Fortran-to-c conventions).
It also became apparent that while the PFEA course has tried to explain conceptually what is needed to hook an external FEA code to the stressRefine library, detailed instructions are needed. I’ll provide them at the same time I provide the wrapper functions.
Another issue is that some FEA codes make use of iterative solvers. My experience in the past has shown that preconditioners for iterative solvers that work well with conventional elements may not lead to good convergence with higher-order p elements, because they result in smaller (fewer equations) but denser stiffness matrices. It is ok to try them with p elements but the iterative solver may not perform as well. Direct sparse solvers work fine with p elements.