On Discrete-Event Simulation and Integration in the Manufacturing System Development Process

DES is seldom used in the manufacturing system development process, instead it is usually used to cure problems in existent systems. This has the effect that the simulation study alone is considered being the cost driver for the analysis of the manufacturing system. It is argued that this is not a entirely correct view since the analysis has to be performed anyway, and the cost directly related to the simulation study is mainly in the model realization phase. It is concluded that it is preferred if the simulation study life cycle coincides with the corresponding manufacturing system's life cycle to increase the usability of the simulation model and to increase efficiency in the simulation study process. A model is supplied to be used for management and engineering process improvements and for improvements of the organizational issues to support simulation activities. By institutionalizing and utilizing well defined processes the conceived complexity related to DES is considered to be reduced over time.

Cost is highly correlated to the time consumed in a simulation study. The presented methodology tries to reduce time consumption and lead-time in the simulation study by: (i)~reducing redundant work, (ii)~reducing rework, and (iii)~moving labor intensive activities forward in time.

To reduce the time to collect and analyze input data a framework is provided that aims at delivering high granularity input data without dependencies. The input data collection framework is designed to provide data for operation and analysis of the manufacturing system in several domains.

To reduce the model realization time two approaches are presented. The first approach supplies a set of modules that enables parameterized models of automated subassembly systems. The second approach builds and runs the simulation model based on a copy of an MRP database, i.e. there is no manual intervention required to build the simulation model. The approach is designed to forecast the performance of an entire enterprise. Since the model is generated from a database, the approach is highly scalable. Furthermore, the maintenance of the simulation model is reduced considerably.