Cristian Grossmann

System identification via nuclear norm regularization for simulated moving bed processes from incomplete data sets

The application of nuclear norm regularization to system identification was recently shown to be a useful method for identifying low order linear models. In this paper, we consider nuclear norm regularization for identification of simulated moving bed processes from data sets with missing entries. The missing data problem is of ongoing interest because the need to analyze incomplete data sets arises frequently in diverse fields such as chemistry, psychometrics and satellite imaging. By casting system identification as a convex optimization problem, nuclear norm regularization can be applied to identify the system in one step, i.e., without imputation of the missing data. Our exploratory work compares the proposed method named NucID to the standard techniques N4SID, prediction error minimization, subspace identification and expectation conditional maximization via linear regression and a linearized first principles model. NucID is found to consistently identify systems with missing data within the imposed error tolerance, a task for which the standard methods sometimes fail, and to be particularly effective when the data is missing with patterns, e.g., on multi-rate systems, where it significantly outperforms existing procedures.

Implementation of an automated on-line high-performance liquid chromatography monitoring system for 'cycle to cycle' control of simulated moving beds

In continuous chromatography simulated moving bed (SMB) is a firmly established powerful technique for the separation of fine chemicals and enantiomers. The use of a controller could improve the operation conditions and increase the productivity of an SMB unit. However, the performance of any controller is greatly affected by the reliability and the quality of the feedback information from the plant. Therefore, to overcome the limitations of optical detectors, such as UV and polarimeter, an automated on-line HPLC monitoring system was developed and installed to monitor the product streams. The performance of the system is tested experimentally separating a mixture of guaifenesin enantiomers on Chiralcel OD columns with ethanol as mobile phase in our laboratory SMB unit under both linear and nonlinear chromatographic conditions. The results show that the new monitoring system provides precise and accurate data about the concentration of the components in the two product streams. Moreover, they prove that despite disturbances a combination of the controller and the new on-line monitoring system allows to fulfill the product specifications and to improve the performance of the process in terms of feed throughput and solvent consumption.

OPTIMIZING CONTROL OF VARIABLE CYCLE TIME SIMULATED MOVING BEDS

Abstract Simulated moving bed (SMB) is a cost-efficient chromatographic technique employed for difficult separation tasks. SMB processes exhibit complex dynamics, e.g., its hybrid nature due to the inlet/outlet port switches with nonlinearities and delays. The novel feature presented in this work is the extension of the ‘cycle to cycle’ control concept to make use as manipulated variables, of the four sectional flow rates and the switch time. Its effectiveness is demonstrated through two case studies; the startup of a SMB unit under plant-model mismatch and the rejection of a pump disturbance.

On-line optimizing control of the intermittent simulated moving bed process

The I-SMB process is one of the modifications to the standard SMB process that has been demonstrated both theoretically and experimentally to exhibit rather competitive performance (Katsuo and Mazzotti in J Chromatogr A 1217:1354, 2010a, 3067, 2010b; Katsuo et al. in J Chromatogr A 1218:9345, 2011). This work aims at showing that also the I-SMB process can be controlled and optimized by using the optimizing on-line controller developed at ETH Zurich for the standard SMB process (Erdem et al. in Ind Eng Chem Res 43:405, 2004a, 3895, 2004b; Grossmann et al. in Adsorption 14:423, 2008, AIChE J 54:1942008). This is achieved by using a virtual I-SMB unit based on a detailed model of the process; past experience with the on-line controller shows that the controller’s behavior on a virtual platform is essentially the same as in laboratory experiments.

Multi-rate optimizing control of simulated moving beds

This paper presents an optimizing control scheme for simulated moving beds (SMB) that allows multi-rate (MR) sampled measurements to be incorporated into the control and estimation problem in a clear and transparent manner. This is particularly relevant for chiral separations where online monitoring requires the combination of various analytical techniques that may operate on widely varying time scales. An MR periodic linear time-varying (PLTV) model is derived for the SMB process. The cyclic nature of the process is exploited by formulating the MR PLTV model within a repetitive model predictive control framework. Simulation results for a chiral separation are presented. The proposed multi-rate controller is able to deliver increased productivity while respecting the process and product specifications.