Design of experiments for simulation models with stochastic constraints

Abstract

Design of experiments is often used to better understand or improve the output measures of a simulation model. Traditional cuboidal and spherical designs have been effective when the factor input regions are known. However, often in practice, simulation models have additional stochastic constraints causing the input factor space to be restricted and irregularly shaped. This paper looks into several alternative designs in a two stage approach to study these models in the constrained regions. This approach is then applied to a (s,S) inventory simulation.

Robust Engineering for the Design of Resilient Manufacturing Systems

“…Despite the fact that methods for designing experiments, addressed to simulation, have been investigated a long time [29,30], it is observed that are still adopted the simulation models that describe present-day production lines in order to provide a prompt inspection of the system factors. A two-stage sequential approach to design experiments for studying simulation systems which have additional stochastic constraints, so the input factor space is restricted and irregularly shaped suggested by [31]. A multi-objective formulation of the buffer allocation problem in unreliable production lines presented in [32] where the factorial design has been used to build a meta-model for estimating production rate, based on a detailed, DES model.…”