| The selection of manufacturing system configurations, that include arrangement of machines, equipment selection and assignment of operations, has a significant impact on their performance especially when considering the new paradigm, namely Reconfigurable Manufacturing Systems (RMS). The objective of RMS is to provide the capacity and functionality needed when needed with the least amount of reconfiguration effort. The use of stochastic analysis and rules-guided planning for the anticipated reconfiguration process in the optimal selection of multiple-aspect RMS configurations, capable of producing multiple-part types simultaneously, achieves the above RMS objective.; In order to achieve the goal of this work, a new "RMS Configuration Selection Approach" was developed. It consists of two stages; the first deals with the selection of near-optimal alternative configurations for each possible demand scenario over the considered configuration periods. It uses a novel constraint satisfaction procedure and powerful meta-heuristics, Genetic Algorithms (GAs) and Tabu Search (TS), for optimizing capital cost and system availability. The second stage utilizes GAs and TS to determine the alternatives, from those produced in the first stage, that would optimize the degree of transition smoothness over the planning horizon. A reconfiguration smoothness (RS) metric is introduced to provide a relative measure of the effort, time and cost required to reconfigure the system. It performs a stochastic analysis of the level of reconfiguration smoothness across all the configuration periods in the planning horizon according to the anticipated demand scenarios. Reconfiguration planning rules are introduced to guide the development of execution plans for system-level reconfiguration, and accordingly reduce the physical effort of reconfiguring the system. The developed approach was demonstrated and validated using a case study. Analysis of different cases of availability considerations was performed. Results using GAs versus TS were consistent for most of the optimization models developed.; This research work enhances the existing knowledge with regards to performance evaluation and configuration selection of manufacturing systems. This work also supports management in selecting RMS configurations at the beginning of each configuration period. |
