Non-uniform grid finite-difference time-domain method for the simulation of electromagnetic distributions

摘要

To solve the memory requirement problem of the FDTD method some previous studies presented subgridding methods that employ two different grid sizes to divide the calculation space. With small structures or on the boundary of different materials a small grid size is used, otherwise a large grid size is used. There are two types of subgridding: the mesh refinement algorithm (MRA) and the variable step size method (VSSM). The MRA makes two runs resulting in a longer computing time. The VSSM varies the step from area to area; this limits its applicability. Both methods use a coarse grid as the boundary of the refinement grid; because the coarse grid is not accurate this affected the refinement result. As a development of the VSSM, the article presents a non-uniform method that can freely use any grid size in the computation area only that the grids are integer times each other. The non-uniform method uses the adjacent grid as the boundary of a smaller grid size, this increases the calculation accuracy. A simulation study of a human thigh section in microwave hyperthermia was examined by this method.