Because Detran will likely never use unstructured meshes, a way to allow for the study of discretization schemes on triangles and tetrahedrals is to introduce discretizations such that the Cartesian mesh cell solution is either (1) the average of the sub-cell solutions or (2) represented in space (multiple unknowns) by the sub-cell solutions.
Option 1 would keep cell-averaged angular moments and accept the typical single flux per surface. Fluxes in each subcell would be averaged, and fluxes leaving the far subcell would be transferred.
Because Detran will likely never use unstructured meshes, a way to allow for the study of discretization schemes on triangles and tetrahedrals is to introduce discretizations such that the Cartesian mesh cell solution is either (1) the average of the sub-cell solutions or (2) represented in space (multiple unknowns) by the sub-cell solutions.
Option 1 would keep cell-averaged angular moments and accept the typical single flux per surface. Fluxes in each subcell would be averaged, and fluxes leaving the far subcell would be transferred.