| This dissertation demonstrates the control of particle size distribution (PSD) in semi-batch emulsion polymerization. The system investigated is vinyl acetate-butyl acrylate co-polymerization with non-ionic surfactants.; An experimental study of the feasibility and methods of producing multi-modal distributions was performed, and the best control strategy was identified.; The next aspect of the research was to develop a detailed first principle model for the process of the evolution of the PSD in semi-batch emulsion co-polymerization. The model is cast in a population balance framework, accounting for the nucleation, growth and coagulation phenomena. The population balance equations—which are hyperbolic partial differential equations—and the associated complete model equations can be solved using standard techniques, as was demonstrated in our study. However, besides being high-dimensional, the system is also characterized by a large stiffness due to the wide difference in the time constants that characterize the nucleation, growth and coagulation processes, rendering the solution computationally very expensive and time consuming. This served as an impetus to develop better numerical solution techniques for the model. An efficient solution technique has been developed, that employs process understanding to re-cast the solution in a hierarchical framework, with the nucleation, growth and coagulation phenomena considered individually to update the PSD. The improved solution technique brings on-line feedback control within the feasibility realm.; The final aspect of the research was to perform optimization and control studies on the PSD. Optimization studies were performed, to design recipe for semi-batch operation that attain target PSDs in the latex. The discontinuities in the process preclude the application of traditional gradient-based optimization techniques to this problem. Thus, direct optimization techniques such as a genetic algorithm were employed to solve this optimization problem. The tracking of a trajectory of PSD was transformed into a multi-objective optimization problem, again in a hierarchical framework in terms of the nucleation and growth rates, and this was solved using a multi-objective extension of genetic algorithm (Non-dominated Sorting Genetic Algorithm). The ability to account for model-mismatch using the on-line measurements was examined using different estimation techniques. Also (off-line) feedback studies were performed which provide encouraging results about the optimization and control of the PSD in semi-batch emulsion polymerization. The studies reveal the potential for different methodologies of feedback control, specifically in-batch (on-line) feedback control and batch-to-batch feedback control. (Abstract shortened by UMI.)... |
