The goal of this research is to devise ranking and selection (R&S) techniques for use with manufacturing simulations (by relaxing the normality and independence assumptions required in classical statistical analyses). Two lines of research are : (1) determining which of two stationary simulated systems has the largest mean and (2) consideration of a nonparametric (R&S) procedure for finding the best of k simulated systems. Work will yield a collection of new methodologies, combining older R&S methods with newer variance-estimation techniques that are more tolerant of autocorrelation in the data. Results will allow analysts more flexibility.
