A receding-horizon multi-model approach to the control of SMB separation processes

In this work, a Multi-Model Predictive Control (MMPC) strategy is proposed for SMB processes. The partial differential equation model is reduced using the proper orthogonal decomposition technique, in which the basis functions or modes are updated on-line. The reduced-order model is the core of a receding-horizon control strategy in which the design parameters are the tuning parameters of PI controllers assigning the velocity of the concentration fronts at the extract and raffinate drains. The low-level control structure is established using the wave theory. Simulation tests demonstrate the performance of this MMPC approach.