Manufacturing systems belong to the class of distributed discrete event systems. Their size requires distributing the software to control them on architecture of several industrial computers connected by networks. In this context, it becomes crucial to be able to evaluate the impact of a specific architecture on the manufacturing systems services both in terms of performance and quality. The performance of the underlying network can notably affect the productivity of the system. In traditional methodology proposed in literature, this aspect is not taken into account in the design stage. Thus, modeling such systems is important to verify some properties at that stage. In this thesis, we propose a component-based modeling approach with High Level Petri nets based method for modeling some network protocols in order to evaluate the manufacturing systems as being distributed systems. The selection of Petri nets is justified by their expression power with regard to the modeling of distributed and concurrent systems. Component-based approach can decrease modeling complexity and encourages genericity, modularity and reusability of ready-to-use components. This allows building new models easily and reducing the systems development cost. Moreover, this can help in better managing services and protocols and to easily change/modify a system element. Finally, this modeling enables us to evaluate discrete event systems by means of centralized simulations
