Multi-objective optimization is used in many chemical engineering fields that have conflict objective functions. Prevaporation is an effective process for removing trace or minor amount of the component of diluting solutions. This process is used for dehydration of alcohols containing small amounts of water. In this process membrane flux and separation factor have conflict with each other. So a multi-objective optimization approach can be used for optimization of the process. In this paper, in first stage a neural network based model was developed for preparation conditions for polyetherimide membrane in isopropanol prevaporation. Four major variables involved in the membrane preparation procedure, including polymer concentration, additive content, solvent evaporation temperature and time was considered. Membrane flux and separation factor were considered as objective functions. Elitist Non-dominated sorting genetic algorithm with jumping gene and altruistic adaptation (Alt-NSGA-aJG) was applied for simultaneous maximization of flux and separation factor. Pareto optimal solutions for membrane preparation conditions and effect of decision variables (four preparation variables) on Pareto front were investigated.
