(60c) Dynamic Modeling and Optimization of Ring Opening Polymerization: Modeling Challenges and Computation Improvements

Recent work has focused on modeling and optimizing semi-batch reactors for ring opening polymerization. Dynamic models have been developed, using first principles (mass balance, heat balance, reaction kinetics, etc.) one based on population balances and another using the method of moments. These models have been validated with historical plant data, matching reactor pressure profiles. The moment model has the added benefit of increased computational efficiency. When using these models to optimize the reactor temperature and monomer feed rate, reductions in polymerization time are seen. Because of the promise from initial product recipes, these simulations and optimizations should be applied to many similar polymerizations.

A reformulation procedure has already been applied to the above system, to separate the fast dynamic model from the differential equation, modeling the fast kinetics algebraically in their quasi-steady state form. However, further challenges with a large, nonlinear system make it difficult for simulations and optimizations to converge. The most realistic system inputs must be used, and the best initial conditions for the optimization are critical. From the chemistry of polymerization, heuristics were developed to determine these initial conditions. Here we present rules of thumb, and general assumptions that enable good initial conditions to be calculated, given a realistic polymerization system. With these informed initial guesses, improved computation time is seen for these optimizations.