Process Design and Multiobjective Optimization of Reflux-Integrated Two-Stage Pressure Swing Adsorption for CO ₂ Capture

Rapidly rising atmospheric CO2 causes environmental issues, necessitating the urgent development of carbon capture technologies. The traditional adsorption process for carbon capture has limited separation performance, while simply cascaded two-stage processes still suffer from low recovery, with insufficiently in-depth research on the mechanism of variables affecting performance indicators. A reflux-integrated two-stage vacuum pressure swing adsorption process was proposed aiming to enrich carbon dioxide from dry flue gas, achieving a purity of 85.73% and a recovery of 90.15%. To fully exploit the performance of the process, an artificial neural network surrogate model was developed, which was employed to solve optimization problems for six decision variables. Pareto analysis of these variables was conducted, with explanations provided for the reasons. Optimization achieved a purity of 96.57% and a recovery of 95.01%. Furthermore, energy consumption decreased by 18.8%, while productivity increased from 0.186 to 0.193 kgCO2/h/kgads.