Design and scheduling of multiproduct batch plants with application to polymer production

In many segments of the chemical industry, emphasis is nowadays placed upon meeting customer needs. This imposes new requirements in terms of various product quality specifications as well as due dates for deliveries, and provides an important incentive for the development of comprehensive procedures for the design and scheduling of multiproduct batch processes. Previous studies fail to address many of the specific issues involved in batch processing, such as distinct due dates, inventory costs, changeover costs, the choice between existing and new equipment, and the many potential alternatives in the synthesis of batch networks. In this thesis, flexible procedures based on the simulated algorithm have been developed to provide a global treatment of scheduling, design, synthesis, and retrofit, with special emphasis on the issues most relevant to batch processes. The flexibility of simulated annealing allows the optimization of complete economic objectives including capital, operating, labor, inventory, and changeover costs as well as penalties for early and late production. The simulated annealing algorithm has been adapted to handle the complex problem structures that arise in design and synthesis. The choice of economic objectives, the effect of uncertainties on the scheduling of flowshops, and the planning of multiplants have been investigated. Optimal scheduling and planning have been incorporated in a design and synthesis procedure to generate processes that are both economical, and capable of meeting a fluctuating demand. This procedure has been extended to retrofit design, to allow the choice between existing and new equipment, and develop a unified treatment of planning, retrofit, and design. Finally, the choice of operating conditions together with design, synthesis, and scheduling has been examined and applied to the batch production of chain growth polymers. The robustness of the optimization procedures has been checked by solving a variety of previously published examples. Significant extensions of these problems have also been developed to explore new issues and the interactions between the different decisions levels. These procedures allow a comprehensive and flexible treatment of a wide class of problems in batch design and scheduling