Multi-Objective Optimization of Dividing Wall Columns and Visualization of the High-Dimensional Results

Abstract Multi-objective optimization of distillation configurations is a difficult problem that can lead to significantly improved process designs. The identification of improvements depends on a reliable and precise calculation of Pareto points, and the ability to visualize them in potentially high-dimensional spaces. This paper presents a methodology for comparing optimal solutions for different distillation configurations. Representative Pareto points are calculated and visualized in self-organizing patch plots. A subsequent filtration of the solutions is possible showing only those that satisfy specific requirements. The procedure is illustrated with a ternary feed stream (benzene, toluene and p-xylene), that is purified in a dividing wall column. The data filtration is explained with a case study. The methodology is applicable to any distillation configuration. This means that the operating ranges for different distillation options can be compared as well as the optimal configuration and its operating point chosen, without performing additional optimization computations.