A prototype tool for modeling manufacturing systems using discrete event simulation is described. The report claims new results in three areas: representation of decision-making functions using intelligent agents, modeling of hierarchical organization (decision-making) structures, and aggregating information within the simulation model for decision making. The claims are reasonably well founded, with the exceptions noted below. The paper is well written and includes 41 references. The prototype runs on a Symbolics machine and is probably not portable to general user environments.
This report has several significant shortcomings. First, much of the prior work on structures for modeling and representation of manufacturing systems is not cited and was not used as a base. An example of significant prior work that would have provided a formal basis is Engelke et al.[1]. Second, prior work in representing the information structures for manufacturing systems is not included, and the authors therefore invented their own structured view of manufacturing. They give no proof that their view is complete or unambiguous. An example of prior work that could have been used is the Open System Architecture for CIM[2]. Third, the prototype was constructed as a standalone system. There was no integration with the actual manufacturing system being modeled. The existing logic (already in the form of computer programs) could have been used to eliminate a large knowledge acquisition and representation task. Fourth, the authors do not mention the treatment of the statistical issues associated with a discrete event simulation model, such as initial transient, replications, characterization of data for input distributions, and the stochastic nature of human decisions.
The authors claim early in the paper that “Traditional simulation tools do not support the notion of hierarchical modeling.” Several well-known tools do support hierarchical modeling, such as RESQ/ME [3].
In my opinion, the prototype is not usable as a modeling tool for manufacturing systems in its current state. The report does provide a status report on this research. It will be of interest only to other researchers on parallel missions.