Coupling optimization and statistical analysis with simulation models

Abstract

Simulation optimization has become commonplace in commercial simulation tools, but automated statistical analysis of the impacts of varying input parameters is much less common. In this paper we explore how both optimization and statistical analysis can be coupled with simulation models to provide key insights for decision makers. A manufacturing example is provided to illustrate the results of multi-objective optimization and post-optimization statistical analysis of the simulation runs. We demonstrate how automated statistical analysis can provide analysts with valuable information on variable sensitivities and good and bad regions of the decision trade space.