A hybrid ADI-FDTD subgridding scheme for efficient electromagnetic computation: Research Articles

Subgridding has been a challenge in FDTD modelling. While it can significantly decrease memory requirements by using coarse and dense grids or meshes wherever they are needed, a small time step must normally be applied to the dense or fine mesh due to the Courant–Friedrich–Levy (CFL) stability condition. In this paper, a technique that combines FDTD and ADI-FDTD methods for subgridding is proposed to circumvent the problem. The solution domain is divided into coarse grid regions and fine subgridded regions whenever necessary. The conventional FDTD is then applied to the coarse grid regions, while the ADI-FDTD is used in the finely subgridded regions. In comparison with subgridding schemes using solely the conventional FDTD, the hybrid method allows the use of a much larger time step and therefore reduces the CPU time. In comparison between the subgridding scheme and pure ADI-FDTD schemes, the hybrid method minimizes the use of the memory because the conventional FDTD algorithm is applied to the coarse grid region. Numerical examples are given to validate these advantages. Copyright © 2004 John Wiley & Sons, Ltd.