Evaluation of multi-objective optimization methods applied to ternary dividing-wall columns

Multi-objective optimization algorithms are widely employed in the optimization of Dividing-Wall Column (DWC). However, due to the inherent complexity of DWC design problems, the choice of different optimization algorithms significantly influences the results. This study aims to investigate the applicability of various multiobjective optimization algorithms in the design of ternary DWCs. Three multi-objective algorithms including multi-objective genetic algorithm (MOGA), multi-objective differential evolution with tabu list algorithm (MODE-TL), and multi-objective particle swarm algorithm (MOPSO) are used to solve the optimization problem of six configurations of dividing-wall columns including direct dividing-wall column, indirect dividing-wall column, Petlyuk dividing-wall column, and three configurations of liquid-only transfer dividing-wall column (LTS), and the optimization results are compared and analyzed. The results show that MOPSO has the worst performance and is relatively less applicable to dividing-wall columns. The MOGA has great global exploration ability but is sensitive to structure complicity. The MODE-TL algorithm can make balance between exploration and exploitation and is less sensitive to structure complicity and the number of variables. Finally, according to the comprehensive performance of the three algorithms, we conclude that MODE-TL has the highest applicability in ternary dividing-wall column optimization.