Experimental validation of multi-objective optimization techniques for design of vacuum swing adsorption processes

Abstract

The efficacy of multi-objective optimization techniques for designing adsorption-based gas separation processes is validated experimentally for the concentration of CO2 from a mixture of 15% CO2 + 85% N2 using two different vacuum swing adsorption (VSA) cycles. The isotherms of CO2 and N2 on Zeolite 13X pellets were measured using volumetry and their competition was characterized by dynamic column breakthrough experiments. A mathematical model that uses inputs from these experiments was combined with genetic algorithm to obtain the Pareto curves for the multi-objective optimization problem that aims to maximize CO2 purity and recovery. The decision variables corresponding to selected points on the Pareto curve were translated into experiments on a lab-scale two-column VSA rig containing 162 g of adsorbent in each column. The experimental performance indicators, viz., CO2 purity and recovery, and the transients of temperatures, outlet flow and composition showed very good match with the experiments thereby establishing multi-objective optimization approaches as a reliable way to design VSA separations.