Towards a better understanding of the epoxy-polymerization process using multi-objective evolutionary computation

摘要: The epoxy-polymerization process can be better understood by investigating the underlying optimization problem involving a number of conflicting objectives and more than 20 decision parameters. A combination minimization or maximization objectives, such as average molecular weight, polydispersity index reaction time, are considered in this paper. first two related to properties polymer, whereas third objective is productivity polymerization process. variables addition quantities various reactants, e.g. amount for bisphenol-A (a monomer), sodium hydroxide epichlorohydrin at different time steps (modeled semi-batch operation), satisfaction all species balance equations treated constraints. multi-objective evolutionary algorithm (the elitist non-dominated sorting genetic NSGA-II) used obtain set solutions single simulation run. results show substantial improvement (with about 300% productivity) over benchmark condition (reported performing one-time reactants beginning batch process). Importantly, study brings out salient aspect using an approach solving. availability multiple optimal trade-off allows engineer have information Changes distribution polymer course observed among Pareto-optimal identified explained purpose. Such provide important operating conditions corresponding trade-offs which otherwise difficult obtain. systematic starting from two-objective problems capture essential features interesting finally solving three-objective associated with discovering interactions should motivate further studies on other chemical problems. Overall, paper demonstrates how fundamental principles systematically reliably find optimum complex operations.