Comparison of NSGA-III with NSGA-II for multi objective optimization of adiabatic styrene reactor

Nature is continuously evolving itself by the different mechanisms so that every aspect of it becomes optimized for different objectives around it. Evolutionary algorithms are developed by mimicking these different mechanisms of evolution. Genetic algorithm is one such evolutionary algorithm. The family of non-dominated sorting genetic algorithms (NSGA) due to its simplicity and efficiency is the most widely used method for solving industrial multi-objective optimization problems. The consideration of elitism in NSGA-II has made it much computationally efficient than NSGA. Recently developed NSGA-III is reported to be more efficient for many-objective (more than two) optimization problems. In this work, the efficacy of NSGA-III is compared with NSGA-II for a three-objective optimization problem of a Styrene reactor. Productivity, Yield and Selectivity of styrene is considered as the three objective functions for the adiabatic styrene reactor. Pareto optimal sets are obtained from both NSGA-II and NSGA-III. The results obtained from NSGA-III shows to provide a more diverse range of optimal operating conditions than NSGA-II. Copyright (c) 2022 Elsevier Ltd. All rights reserved. Selection and peer-review under responsibility of the scientific committee of the International Chemical Engineering Conference 2021 (100 Glorious Years of Chemical Engineering & Technology)