Multi-objective genetic algorithms for chemical engineering applications

Abstract

The current environmental and social awareness leads optimization researchers to be more and more concerned with multi-objective optimization problems (MOOPs). Evolutionary methods and particularly genetic algorithms are commonly used in chemical engineering, where the problems generally involve complex models embedded in an outer optimization loop. This paper first presents two multi-objective genetic algorithms to tackle continuous and mixed-integer chemical engineering problems. The algorithms are then illustrated by classical chemical engineering benchmark problems often used in the literature for mono-objective optimization studies: three bi-objective ones (ammonia synthesis reactor, alkylation plant, natural gas transportation network), a structural mixed-integer design problem and three multi-objective problems (Williams-Otto process, new product development in the pharmaceutical industry and economic and environmental study of the HDA process). The results are compared with the ones reported in the literature, and the analysis highlights the efficiency of the proposed algorithms either in the continuous case or in the mixed-integer one.