Model predictive control and optimization of vacuum pressure swing adsorption for carbon dioxide capture

This study presents a control strategy of carbon dioxide capture from flue gas by Vacuum Pressure Swing Adsorption (VPSA). The key objective of this work is to control and optimize the purity and recovery of the adsorption product (carbon dioxide) by manipulating the time duration of the VPSA cycle steps. Based on a single fixed-bed 6-step VPSA process simulation using 5A zeolite as adsorbent for CO
2
capture from gas mixture with 15% CO
2
and 85% N
2
(resembling post-combustion flue gases of power stations), a two-input/two-output control model was obtained, in which Feed step time and Purge step time are taken as manipulative variables and CO2 Purity and Recovery are the controlled variables. For control purpose, a multi-input/multi-output model predictive control (MIMO-MPC) system is proposed to handle the inherent non-linear nature and discontinuous operation of the VPSA process. The MIMO-MPC control scheme is tested and showed good results in terms of system stability and fast tracking of the set-points. Finally, PSO algorithm is applied to maximize the purity and recovery by finding the optimal duration time of feed step and purge step, the results shows that with such VPSA process, over 98% of CO
2
is recovered with purity of 59%, at t
feed
= 202s and t
purge
= 114s.