(136d) Optimization Based Predictive Control of Simulated Moving Bed Process Using Subspace Identification

Optimization based dynamic control using an identified model is designed and implemented on a simulated moving bed (SMB) process. A linear output prediction model is obtained by the method of subspace identification and used for the predictive control. The controller is designed for optimizing the production cost while maintaining the specified product purities. For all of these, the purities of the target components in extract and raffinate streams averaged over one switching period, the reciprocal productivity, and the solvent consumption are selected as output variables, while the flow rates in section 1 to section 4 are chosen as the manipulated variables. The realization of this concept is discussed and assessed by a simulation study as well as by an experimental work. For simulation, the first principles model for the separation of enantiomers of 1-1'-bi-2 naphthol on a 3,5-dinitrobenzoyl phenylglycine bonded to a silica gel stationary phase is considered as the actual SMB process. For experimental implementation, a laboratory scale SMB is applied to the separation of two nucleosides, uridine and guanosine, on the reversed phase SOURCE^TM 30 RPC stationary phase. The feedback information consists of the concentration profiles of extract and raffinate measured online using UV detectors. The identified prediction model is proven to be in good agreement with the first principles model in the simulation study and with the laboratory scale SMB for experimental work. For typical control objectives encountered in actual operation, i.e., disturbance rejection and set-point tracking, it is demonstrated that the proposed controller exhibits excellent performances both in the simulation study as well as in the experimental work, hence it is an effective tool for optimization based dynamic control of SMB processes.

Keywords: Simulated moving beds; optimization based control, subspace identification, model predictive control, experimental implementation