MINLP Model for Synthesis of Paraxylene Separation Processes Based on Crystallization Technology

In this paper we address the development of an MINLP optimization model for the synthesis of paraxylene separation processes based on crystallization technology. A superstructure is proposed that considers different alternative designs and operating conditions for the recovery of a high purity paraxylene product from a feed stream containing other xylene isomers and ethylbenzene. The goal is to maximize the overall annual profitability of the process, including capital and operating cost as well as product revenues. Due to the large size of the MINLP model, as well as the great difficulty in converging the corresponding equations, a solution strategy based on initialization models for the topology and the operating conditions is described to enhance the robustness of the optimization method. Application of the proposed model yields flowsheet designs with improved savings in capital and utilities cost.