Int. J. Production Economics 85 (2003) 11–31 M A MA-S: An introduction to a methodological approach for the simulation of distributed industrial systems S. Galland a, *, F. Grimaud c , P. Beaune b , J.P. Campagne c a Systems and Transportation Laboratory, University of Technology of Belfort-Montb ! eliard, Belfort 90010, France b Multi-Agent Systems Department, Higher National School of Mines of Saint- ! Etienne, Saint- ! Etienne 42023, France c Scientifical Methods for Industrial Management Department, Higher National School of Mines of Saint- ! Etienne, Saint- ! Etienne 42023, France Abstract We are located in the context of the industrial system simulation, which is complex and distributed in operational, informational and decisional terms. In this article, we present the problems of the simulation of such systems. We propose a methodological approach based on the systemic and on the multi-agent concepts. It authorizes the data- processing and modeling distributions. r 2003 Elsevier Science B.V. All rights reserved. Keywords: Industrial system simulation; Multi-agent systems; Methodology; Distribution; Decisional processes 1. Introduction The simulation is a tool adapted to the modern industrial problems. It permits to support and study the dynamic behavior of industrial systems. Even if the simulation is a powerful tool, some problems always exist. We concentrate our work on four kinds of them. First, we assume that the simulation tools are still seldom packaged with a methodology, which allows an easier modeling of an industrial system. The second problem, that we want to tackle, is the poor support of the component modeling or of the modular modeling. It is still difficult to reuse the existing models (or a part of them) without substantial changes or a complete rewriting. The third problem is the strong relationship between the physical, the informational and the decisional aspects of an industrial system. For example, if we want to simulate a system with a pulled-flow management policy, and just after with a pushed-flow policy, we must completely rewrite the simulation model, even if only the decisional part changed. Why must the rest of the model be rewritten? Finally, the last problem is about the difficult to model the new industrial organizations, such as the enterprise consortiums or the virtual enterprises. To solve these different problems, we propose a methodo- logical approach: M A MA-S 1 (Galland, 2001). It offers a modeling framework that is independent of any software platform (simulation tool or multi- agent system). In this paper, we present the major ARTICLE IN PRESS *Corresponding author. Tel.: +33-3-84583418; fax: +33-3- 84-583342. E-mail address: stephane.galland@utbm.fr (S. Galland). 1 Multi-agent methodological approach for the simulation of industrial systems. 0925-5273/03/$-see front matter r 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0925-5273(03)00083-5