Planning, control and management of multicellular manufacturing systems by production authorization cards (PAC) system

The object of this research is to study the issue of proper coordination in a multicellular manufacturing system. In 1992, Buzacott and Shanthikumar introduced the Production Authorization Cards (PAC) system, a new coordination scheme for material and information flow control in multicellular manufacturing systems. In this thesis we describe both a simulation model and optimization method for the PAC control and show their application in a number of settings.;The thesis first describes and reviews the literature for common manufacturing control policies and then focuses on the description of the PAC concept. Since the PAC system forms a basis for the research, a complete presentation is essential for understanding of the research issues.;The development of a simulation model of a multi-cell multi-product manufacturing system coordinated by Production Authorization Cards was an essential part of this research. This model allows for the performance evaluation of the PAC controlled manufacturing system. It permits the analysis of systems with assembly aspects, batching and complex routing. Execution speed and the ability to be called as a subroutine in an overall optimization are key features.;The PAC optimization algorithm uses the simulation model for the evaluation of the cost function to be optimized. The optimization algorithm is based on hill climbing methods similar to Hooke and Jeeves. It is a search consisting of a sequence of exploration steps about a base set of PAC parameters, which, if successful, are followed by pattern moves. The PAC optimization model can, at best, be guaranteed to produce local optima. A second algorithm, based on random search principles, has been developed to verify how well the first method performs.;The use of the simulation model and the PAC optimization algorithm are illustrated with 5 different examples of manufacturing configurations, ranging from a very simple one to a quite complex one. The optimization algorithm is quite feasible for systems with a small number of final and intermediate products (less than 20). Numerical results for nine different coordination policies (PAC-general, PTO, PTO-variant with...