STUDIES IN THE OPTIMAL DESIGN AND SCHEDULING OF BATCH PROCESSING PLANTS

This thesis addresses three aspects of the design of batch processes: design of a new multi-purpose plant, re-design of an existing multi-product plant, and heat integration for existing plants. A novel method that uses a single mixed-integer nonlinear program (MINLP) to represent both the scheduling and design issues for a totally new multi-purpose batch facility is presented. The grouping of products for simultaneous processing must be determined as well as the sizes and number of units in parallel for the equipment to minimize investment cost. By combining the two aspects in the proposed formulation, an efficient solution procedure is obtained whose global optimality can be proven for most cases.; The redesign of existing multi-product batch facilities to accommodate a changing environment also has been examined. The design issues involve maximization of profit by selecting the optimal production goals for each product in addition to the location and size of new equipment. New equipment can be utilized (1) to expand the batch size or (2) to decrease the cycle time. A MINLP model has been developed for this problem. Through the use of exponential transformations and the incorporation of linear under-estimators, a global optimum solution of this formulation is assured using the Outer Approximation Algorithm of Duran and Grossmann (1983). Multi-period operation of the plant to accommodate changes in future demand can be determined with this formulation.; The operation of batch processing plants to maximize heat recovery has been investigated also. At a particular time in a batch process, some tanks with a batch of material may require heating while some other tanks may need cooling. Depending on the processing requirements, the transfer of heat between tanks can present three distinct temperature profiles in time that are similar to cocurrent, countercurrent, and combination cocurrent/countercurrent temperature profiles. For the cocurrent case, a heuristic is given for deciding the sequence of exchanges when multiple hot and/or cold batches are present. This heuristic is shown to lead to the optimal answer in certain instances. For cases in which the desired final temperatures limit some of the exchanges, a mixed-integer linear programming formulation is proposed.