Parallel Multiobjective Evolutionary Algorithms for Waste Solvent Recycling

Abstract

Waste solvents are of great concern to the chemical process industries and to the public, and many technologies have been suggested and implemented in the chemical process industries to reduce waste and associated environmental impacts. As reported in this article, we have developed a novel parallel multiobjective steady-state genetic algorithm (pMSGA) for designing environmentally benign and economically viable processes for waste solvent recycling. This pMSGA can efficiently solve this complex multiobjective design problem and provide accurate and uniform Pareto-optimal (i.e., tradeoff) solutions. In addition, it can approximate a wider range of the Pareto front than other multiobjective genetic algorithms. As a case study, acetic acid recovery from aqueous waste mixtures is investigated under simultaneous maximization of the total profit and minimization of the potential environmental impacts (PEIs). At low acetic acid feed contents (xF = 0.25), many of the Pareto-optimal solutions are economically infeasible and also provided minimal PEI reduction. However, at medium and high feed contents (xF = 0.30 and 0.35), the total profit is very large, and the PEI reduction is significant as well.