Stochastic LOD-FDTD method for two-dimensional electromagnetic uncertainty problems

Abstract

Purpose

Random media uncertainties exhibit a significant impact on the properties of electromagnetic fields that usually deterministic models tend to neglect. As a result, these models fail to quantify the variation in the calculated electromagnetic fields, leading to inaccurate outcomes. This paper aims to introduce an unconditionally stable finite-difference time-domain (FDTD) method for assessing two-dimensional random media uncertainties in one simulation.

Design/methodology/approach

The proposed technique is an extension of the stochastic FDTD (S-FDTD) scheme, which approximates the variance of a given field component using the Delta method. Specifically in this paper, the Delta method is applied to the locally one-dimensional (LOD) FDTD scheme (hence named S-LOD-FDTD), to achieve unconditional stability. The validity of this algorithm is tested by solving two-dimensional random media problems and comparing the results with other methods, such as the Monte-Carlo (MC) and the S-FDTD techniques.

Findings

This paper provides numerical results that prove the unconditional stability of the S-LOD-FDTD technique. Also, the comparison with the MC and the S-FDTD methods shows that reliable outcomes can be extracted even with larger time steps, thus making this technique more efficient than the other two aforementioned schemes.

Research limitations/implications

The S-LOD-FDTD method requires the proper quantification of various correlation coefficients between the calculated fields and the electrical parameters, to achieve reliable results. This cannot be known beforehand and the only known way to calculate them is to run a fraction of MC simulations.

Originality/value

This paper introduces a new unconditional stable technique for measuring material uncertainties in one realization.

Citation

Salis, C., Kantartzis, N. and Zygiridis, T. (2017), "Stochastic LOD-FDTD method for two-dimensional electromagnetic uncertainty problems", COMPEL - The international journal for computation and mathematics in electrical and electronic engineering, Vol. 36 No. 5, pp. 1442-1456. https://doi.org/10.1108/COMPEL-02-2017-0087