Experiences with analytic and simulation modeling for a factory of the future project at IBM

Abstract

Simulation models have been used extensively for manufacturing systems design analysis. Simulation modeling is a flexible and powerful technique but model development, debugging, modification, and analysis is a time consuming process. Because of this, it is possible to address only a limited number of design issues using simulation models alone.This paper discusses the application of an analysis methodology employing two techniques, queueing network and simulation modeling, to effectively address a variety of issues during the design of a factory of the future at IBM. The queueing network model was used for the initial analysis to quickly reduce a wide range of design alternatives. The simulation model was used to further study alternatives that had been selected using the queueing network analysis. The model included details which were not considered during the initial analysis.Our experience from this project indicates that queueing network models are effective at the initial analysis level. Simulation models should be used only after the alternative choices have been substantially reduced. They can be used, however, to address issues not included previously and to ensure that the assumptions made earlier are still valid. Results from the queueing network models also can be used to design as well as debug and verify simulation models.