Multi-Objective Optimization of Dividing Wall Columns and Visualization of the High-Dimensional Results. (2nd November 2020)

• Record Type: Journal Article
• Authors: Ränger, Lena-Marie; von Kurnatowski, Martin; Bortz, Michael; Grützner, Thomas
• Is Part Of: Computers & chemical engineering. Volume 142(2020)
• Publication Date: 2020-11-02
• Subjects: Dividing wall columns -- Simulation -- Optimization -- Multi-objective optimization -- Self-organizing patch plots; Chemical engineering -- Data processing -- Periodicals; 660.0285
• DOI: 10.1016/j.compchemeng.2020.107059
• Languages: English
• ISSNs: 0098-1354

Abstract

Highlights: Multi-objective optimization of dividing wall columns. High-dimensional solution space. Visualization in self-organizing patch plots. Filtration of results to determine operating point. Pareto set reusable for same feed stream composition. Optimization-based design. 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.