학술논문

A Waveform Relaxation Solver for Transient Simulation of Large-Scale Nonlinearly Loaded Shielding Structures
Document Type
Periodical
Source
IEEE Transactions on Electromagnetic Compatibility IEEE Trans. Electromagn. Compat. Electromagnetic Compatibility, IEEE Transactions on. 64(6):2042-2054 Dec, 2022
Subject
Fields, Waves and Electromagnetics
Engineered Materials, Dielectrics and Plasmas
Transient analysis
Electromagnetics
Convergence
Behavioral sciences
SPICE
Mathematical models
Load modeling
Circuit simulation
field circuit coupling
large-scale systems
macromodeling
nonlinear circuits
waveform relaxation
Language
ISSN
0018-9375
1558-187X
Abstract
This article introduces an algorithm for transient simulation of electromagnetic structures loaded by lumped nonlinear devices. The reference application is energy-selective shielding, which adopts clipping devices uniformly spread along shield apertures to achieve a shielding effectiveness that increases with the power of the incident field, thereby blocking high-power interference while allowing low-power communication. Transient simulation of such structures poses a number of challenges, related to their large-scale and low-loss nature. In this work, we propose a waveform relaxation (WR) scheme based on decoupling the linear electromagnetic structure from its nonlinear terminations. In a preprocessing stage, the electromagnetic subsystem is characterized in the frequency domain and converted into a behavioral rational macromodel. Transient simulation is performed by refining estimates of the port signals through iterations. The proposed scheme combines a time partitioning approach with an inexact Newton–Krylov solver. This combination provides fast convergence also in those cases where standard WR schemes fail due to a strong mismatch at the decoupling sections. Numerical results on several test cases of increasing complexity with up to 1024 ports show that the proposed approach proves as reliable as HSPICE in terms of accuracy, with a speedup ranging from one to three orders of magnitude.