Recently, our group has validated experimentally the ‘cycle to cycle' optimizing control scheme for simulated moving beds (SMB) for the separation of achiral and chiral components [1, 2]. The feedback information used for control purposes was measured by online optical detectors, i.e. UV and polarized light detectors, positioned at the extract and raffinate streams. The accuracy of such devices is highly sensitive to experimental factors like impurities in the mixture to be separated or pressure fluctuation in the measuring cell; and has a direct impact on the performance of the controller. This can become a limiting factor when extending the control scheme to multicomponent systems and chiral separations.

In the spirit of widening the spectrum of applicability of the controller and improving the accuracy of the measurements and therefore the performance of the controller, we have extended our control scheme to make use of an existing and reliable monitoring technique, namely HPLC.

This approach has, on one hand, the advantage that the HPLC measurements are straightforward, more accurate than online optical detectors, can handle multicomponent systems and are not greatly affected by impurities. On the other hand, these measurements are less frequent and provide only information about the average concentration of the two species at the outlet streams. Moreover, the analysis time may be in the range of the cycle time, which would introduce a significant time delay in the measurements. Hence, the controller relies on less but more accurate information that is available with a lower frequency, i.e. once every cycle, and can be delayed.

This work presents the design, implementation and validation of the SMB controller in combination with the online HPLC monitoring for a nonlinear chiral separation of the guaifenesin enantiomers in Ethanol using Chiralcel OD as stationary phase. Furthermore, it reports the theoretical modifications and extensions that had to be done to the control scheme to cope with the implemented monitoring system.

The performance of the controller is assessed in various scenarios that cover common situations encountered in SMB practice, such as the separation of racemic mixtures with different overall concentrations, in a range from 0.1g/L (linear) to 15g/L (nonlinear) regime; rejection of disturbances due to pump malfunctioning; setpoint tracking in the case of varying purity specifications or if the feed concentration changes in the course of the operation. The results illustrate that the 'cycle to cycle' controller is able to meet the products' purity specifications and operate the process optimally using minimal information about the system regardless of the disturbances that might take place during the operation.

[1] Grossmann, C., Erdem, G., Morari, M., Amanullah, M., Mazzotti, M., Morbidelli, M., ‘Cycle to cycle' optimizing control of simulated moving beds. AIChE J., vol. 54, pp. 194-208, 2008

[2] Amanullah M., C. Grossmann, M. Mazzotti, M. Morari, M. Morbidelli. “Experimental implementation of automatic ‘cycle to cycle' control of a chiral simulated moving bed separation “, J. of Chromatography A, vol. 1165, no. 1-2, pp. 100-108, 2007.