Comment on 'A simple way to model curved metal boundaries in FDTD algorithm avoiding staircase approximation' [and reply]

摘要

For the original article see ibid., vol. 5, no. 8, p. 267-9 (1995). The triangular subdivision of mesh cells as described in the aforementioned article were first published (to the commenter's knowledge) over 15 years ago in the context of a frequency-domain method. This improvement has been proposed for time-domain computations with particle beams and used in numerous publications for two-dimensional, 2.5-dimensional, and three-dimensional computations for static fields, eddy current fields, high-frequency fields and, last but not least, for time-domain fields Furthermore, the triangular subdivision is not only applicable for metallic boundaries, but also for any type of material boundary. In reply the authors point out that the originality of their work does not consist of the use of triangular cells, but resides essentially in the following points: 1) The general formulations for the simulation of slanted walls have been specialized to a very simple and efficient formula for metallic walls laying across the cell diagonals; 2) a specific example has been used to compare the formula with the conventional staircase approximation to quantify the consequent improvement in accuracy; and 3) the main novelty of their article consists of the combination of the triangular subdivision with a suitable graded mesh such that the mesh nodes lie on the metal boundaries.