Thermal Enhancement of Silicon Carbide (SiC) Power Electronics and Laser Bars: Statistical Design Optimization of a Liquid-Cooled Power Electronic Heat Sink.

摘要

The objective of this project was to investigate the utilization of a design of experiments (DOE) methodology as an efficient means to realize an optimized configuration for a single phase, liquid-cooled, pin fin, heat sink for power electronic thermal management applications. The process being utilized in this effort is a computational fluid dynamics (CFD) software solution of a series of quarter fraction, two-level (2n-2) DOE runs over defined ranges of input factors, variables, or treatments. The software used on this project is the Navier-Stokes CFD solver Fluent(trademark) package marketed by ANSYS, Inc., which is a finite volume discretization PDE solution technique. The output response variables of interest are computationally determined for each of the factorial run conditions and used in all subsequent statistical analysis. The ultimate goal is to ascertain whether a generic design algorithm for pin fin heat sinks can be realized using subject matter (SM) expertise coupled with a focused DOE methodology, thereby simplifying thermal management design for optimal heat transfer performance that can lead to acceptable power device junction temperatures and reliable operation.