Optimization of a Manufacturing System

The design of modern manufacturing systems is a complex task. High investments require that the planned system will fulfill the requirements. Moreover, complex interleaving of choices and synchronizations in manufacturing systems may lead to paradoxical behavior. For example, increasing the number of resources (i.e., tokens in a Petri net model) may result in a deadlocked system, and replacing a machine for a faster one can decrease global productivity. Methods and computer tools for the modeling and performance evaluation and optimization of manufacturing systems are therefore important. Manufacturing system design and operation is one of the fields where stochastic discrete event systems are widely used; see the bibliographical notes at the end of the chapter.

Direct optimization methods are not applicable for complex manufacturing systems if a level of detail is necessary in the model that goes beyond the first rough estimation steps in the design. Efficient methods based on a problem description as a linear programming problem are therefore unfortunately not applicable. Models of complex manufacturing systems are nonlinear in principle in addition to that. This is especially the case for selection problems where parameters determine different system layouts, strategies, or machine types. On the other hand, there are problems related to selecting the optimal speed of a transport system, number of transport pallets, and the like. It is not possible to guess the form of the function from a system model directly, which would make simple standard methods (e.g., a gradient Newton search) applicable. These restrictions lead to the application of heuristic search methods for an optimization.