Semiconductor manufacturing is one of the most complex production processes. This is in fact a high-tech environment characterised by huge fixed costs for the required machinery, equipment and tooling. In order to achieve a satisfactory level of return on shareholders' equity, wafer fabrication facilities (fabs) are usually run to obtain short cycle times, low work-in-process inventory levels, and high equipment utilisation rates. In modern wafer fabs, an Automated Material Handling System (AMHS) is mandatory in order to satisfy ergonomic constraints and productivity objectives. Although the AMHS has a very important role in wafer fabs, it has been little studied in the literature from an operational point of view (i.e. vehicle positioning and dispatching strategies). Existing works consider the AMHS as an isolated system and do not study the impact of the variability generated in the rest of the fab. Because of the complexity of the wafer fabrication process, traditional methodologies have used discrete-event simulation approaches.The objective of this dissertation is to propose a formal methodology for the analysis and optimisation of the AMHS in order to improve the fab productivity metrics. We first propose a hybrid model of the fab using the Process-Interaction Approach and coloured timed Petri nets. These modelling formalisms aid us to characterise the physical production system, its control system, and the production flow process, in order to better identify the components of the simulation model. Then, various strategies for the operation of the AMHS are compared. This study is performed by coupling classical Operational Research-based techniques (i.e. integer programming, heuristics) with the simulation model of the fab. The problem of vehicle positioning is approached from a tactical level so as to identify the optimal parameters of the simulation model. This last allows us to study the behaviour of the transport system and facilitates the implementation of operation vehicle control policies. Final application of these strategies is done on a fab of the new 300mm wafer generation.
