The first pin cell put in place uses a uniformly-space mesh, needed for responses. Materials are defined at cell center, and hence volume is generally not conserved. In order to better benchmark the methods, a volume-conserving discretization is needed. Another option would be to volume-average cross-sections, but I don't want to do that.
Done. I used a very simple approach. The cylinder is squared, and a half length flange is attached to each face with a thickness I computed that minimizes shape error.
The first pin cell put in place uses a uniformly-space mesh, needed for responses. Materials are defined at cell center, and hence volume is generally not conserved. In order to better benchmark the methods, a volume-conserving discretization is needed. Another option would be to volume-average cross-sections, but I don't want to do that.