Percolation framework and monte carlo techniques for improved probabilistic design of variability in products and systems

Abstract

Variability is an inherent feature of any manufactured product or system that can be caused by process inhomogeneity, environmental perturbations, operating stress-induced degradations or unexpected human interference during fabrication and use. The design of a product or system should account for and model the probabilistic nature of these variations that have an impact on its yield, reliability and robustness. This study is intended to present a combination of a theory based on the concept of "percolation" and an algorithm based on Monte Carlo that can serve as a key model-based design tool to quantify the variability in performance/lifetime/ material properties/time-based events and identify the possible root cause(s) for the variance so that the design process could be refined to improve and optimize the homogeneity of the population of devices that would be manufactured on a large scale from the evolving product design stage.