Systematic design of chemical reactors with multiple stages via multi-objective optimization approach

The purpose of this study is to extend the approach that was introduced by Hillestad (2010) to handle chemical reactor design problem with multiple stages. Specifically, multi-objective optimization method will be used to generate Pareto optimal solutions that characterize the non-inferior solutions set for the problem. Following the identification of path-dependent design variables, several (possibly conflicting) design objectives will be selected and solutions of the corresponding problem will be generated from multi-objective optimization algorithm. This approach is investigated for two industrially important reactor systems: ethylene oxide and phthalic anhydride synthesis. By using reference-point based multi-objective evolutionary algorithm (R-NSGA-II), Pareto-optimal solutions are successfully generated within the region of user-specified reference points, thus facilitating in the selection of final optimal designs. Apart from the extensive selection of optimal candidate reactor designs, this approach also enables further insights to be obtained regarding the optimal arrangement of the path-dependent design variables along the reactor length.