(197a) Integrated scheduling and dynamic optimization for semi-batch polymerization processes

In this work, we propose an integrated optimization approach for an industrial semi-batch polymerization process. The method introduces a discrete time formulation for the simultaneous optimization of process scheduling and operation decisions. The resource task network (RTN) representation is applied to model the interconnection between units and production sequences. Polymerization reactors are described by first-principles models addressing heat/mass balances, reaction kinetics, etc.

Our modeling framework accommodates general complications in scheduling and unit operation, such as customer orders, transfer policy and requirements on product quality and process safety. In the mathematical formulation, the scheduling and unit operation equations are linked with the task history state variables in the state space RTN model. A tailored generalized Benders decomposition (GBD) algorithm is applied to efficiently solve the resulting large nonconvex mixed-integer nonlinear program by exploring the particular model structure.

The studied polymerization process has two parallel semi-batch reactors for ring-opening polymerization, continuous storage tanks, and purification units. The two polymerization reactors share cooling utility from the same source, and the utility price depends on the consumption rate. The optimization objective is to design the process schedule and reactor control policies simultaneously to maximize the overall process profit. The case study results suggest improvements in plant profitability for the integrated approach, in contrast to the conventional approach, where recipes of the polymerization tasks are individually optimized and the interactions among process units are overlooked.