Reduce-Order Analysis and Circuit-Level Cost Function for the Numerical Optimization of Power Electronics Modules

Abstract

Energy innovation trends such as sustainable grids, and the electrification of the consumer transportation market are accelerating the adoption of medium-voltage (MV) silicon-carbide (SiC) technology. The fast switching times and higher operating temperatures enabled by MV SiC have forced package designers to employ innovative technologies to improve power densities, lower stray parasitics, and increase cooling capability. These new technologies ultimately require new simulation tools to allow for fast, efficient, and accurate computation of the multi-physics phenomena that govern package performance. To that end, this work proposes a multi-physics optimization workflow that utilizes reduced-order 3D package models and behavior circuit models to efficiently quantify the effects of the package on converter performance. A cost function is proposed which utilizes the information to optimize for converter performance, rather than individual package parameters.