Identification and predictive control of a simulated moving bed process: Purity control

A model predictive control strategy for a simulated moving bed (SMB) chromatographic process is proposed. For this, the average purities over one switching period of target components in extract and raffinate ports are selected as output variables, while the flow rates in Sections 2 and 3 of the SMB unit are chosen as the input variables. With this arrangement a linear input-output prediction model is identified through subspace identification and used for dynamic control. The realization of this concept is discussed and the implementation on a virtual eight column SMB unit is assessed, in the case of the separation of enantiomers behaving according to the binary bi-Langmuir adsorption isotherm. The identified prediction model is proven to be in good agreement with the first principles model used to simulate the actual SMB process. For typical control objectives encountered in real operation, i.e., disturbance rejection or set -point tracking, it is shown that the proposed controller demonstrates satisfactory control performances in minimizing off-spec products.