A universal sensitivity matrix reduction technique (SMART) to uncover governing thermal transport relationships

Abstract

Sensitivity analysis is a powerful technique to analyze the behaviors of models and experimental data. Most state-of-the-art characterization techniques rely on fitting experimental data to a theoretical multivariate model that can have multiple unknown properties other than the target property. Often, these unknowns cannot be measured separately and are unavoidably fitted for their values simultaneously using a single set of measured data. No rigorous method exists to evaluate their accuracy and elucidate the inherent governing relationships that can change with different measurement conditions. Here, we formulate a systematic approach based on sensitivity analysis to obtain these hidden relationships in a complex multivariate model. Criteria for simultaneously estimating the values of multiple unknowns are also clarified. We demonstrate this approach for a transient thermal transport measurement technique. This approach is versatile and can be applied to most measurement techniques in various disciplines that involve multivariate theoretical models to improve their measurement accuracy and throughput.