A predictive criterion for 2D/3D DGTD method based on the CFL condition
The zeroth-order parallel-plates mode dominates in most parts of the power-ground plates pairs. Domains with such mode distribution can be solved by a simplified 2D algorithm, while only higher-order modes require 3D computation. For DGTD method, different domains can be decoupled with each other by the numerical flux. To distinguish 2D and 3D domains properly, the field distribution has to be known before it is actually solved. This paradox is walked around by a predictive criterion which roughly predicts tn's field distribution based on tn-1's results. Here we introduce an adaptive criterion based on the CFL condition to realize such prediction. With this predictive criterion, the 2D and 3D domains can be identified properly to ensure an accurate and stable result.
Wending Mai*, Yongpin Chen, Moyu Deng, Jingxian Chen, Jun Hu