A Robust Model Predictive Controller applied to a Pressure Swing Adsorption Process: An Analysis Based on a Linear Model Mismatch

Abstract The present work studies the control problem of a pressure swing adsorption (PSA) unit using a robust stabilizing infinite-horizon model predictive (RIHMPC) strategy with guarantee of feasibility for realistic mismatch scenarios. The identification of the multi-plant linear models was done based on an operating confidence region. This procedure is based on an optimal point given by an optimization layer, concomitantly with the uncertainty associated with that point. As a case study, it was evaluated the control of a PSA unit for the purification of syngas by porous amino-functionalized titanium terephthalate MIL-125 in the PSA. The results demonstrated that RIHMPC might be an efficient strategy to address the control of cyclic adsorption processes accommodating the intrinsic nonlinearities and uncertainties of these processes.