Regression metamodeling for the design of automated manufacturing system composed of parallel machines sharing a material handling resource

This study concerns a particular type of automated manufacturing system, composed of parallel machines sharing a material handling resource often encountered in practice. For the company, with which we work, these expensive systems are frequently the bottleneck of the global plant, so that designers are interested in continuously improving them. In this respect, the important factors and the consequences on the average system utilization, if certain factors change, are pointed out and quantified, through regression metamodeling of a stochastic simulation model. The metamodel developed provides the decision-makers and the designers with useful insight into these systems. For example, it highlights that the material handling resource speed and acceleration value were quite not as influential on the performance measure as expected by the experts.