A distributed time stepped simulation approach for analysis and comparison of shop floor control architectures

Abstract

Different shop floor control architectures (SFCAs) are trying to resolve control problem of Flexible Manufacturing Systems (FMSs) by proposing different paradigms. Hierarchical SFCAs enable a layered hierarchy and try to overcome the complexity of the system by aggregating the problem. Heterarchical approaches see the system as a whole that consists of autonomous and cooperative agents. To develop efficient applications of SFCAs, simulation is an appropriate approach which lets a limitless configuration and test possibility. But the methodology of the simulation application is very important and must be developed with considering the system at hand. When modeling a system composed of entities, which interchange messages, consideration of messaging structure becomes essential. In this respect, Parallel and Distributed Simulation (P/DS) is the most appropriate methodology for these systems. In this thesis, simulation applications for both hierarchical and heterarchical SFCAs are built with a distributed and parallel simulation approach. Thus a test bed for comparing the performances of different architectures and for developing new approaches to decision making, information sharing, or communication within each architecture is obtained. In developing P/DS applications, existing real-time SFCA applications in Bo˘gazi歔ci University Flexible Automation and Intelligent Manufacturing (BUFAIM) laboratory are taken as basis. Applications are developed in a modular and object oriented fashion, to enable easy progression from simulation to real-time control and vice versa. With running each distributed module on a different computer and keeping the messaging real it is aimed to catch the dynamics of the communication structures which are the defining characteristics of the SFCAs.