Abstract
This article addresses the synthesis and optimization of crystallization processes for p-xylene recovery for systems with feed streams of high concentration, a case that arises in hybrid designs where the first step is commonly performed by adsorption. A novel superstructure and its corresponding mixed-integer nonlinear programming (MINLP) model are proposed. The distinct feature of this superstructure is the capability to generate optimum or near optimum flow sheets for a wide range of specifications of p-xylene compositions in the feed stream of the process. To cope with the complexity of the MINLP model, a two-level decomposition approach, consisting of the solution of an aggregated model and a detailed model, is proposed. The results obtained show good performance of the decomposition strategy, and the optimal flow sheets and p-xylene recoveries are in agreement with the results reported in patents. (c) 2008 American Institute of Chemical Engineers AIChE J, 55: 354-373, 2009