Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


On-line optimization and control of simulated moving bed processes


G. Erdem


In recent years Simulated Moving Bed (SMB) technology has become more and more attractive in the areas of pharmaceuticals, fine chemicals and biotechnology in which complex separation tasks are faced on daily bases. There are two striking advantages of SMB technology that make it a powerful alternative. Firstly, it delivers high productivity per unit mass of stationary phase with low solvent consumption compared to conventional chromatographic processes such as batch chromatography. Secondly, it allows a fast and reliable scale-up of separations from analytical chromatography, and therefore, a shorter time to market can be achieved. Thorough understanding of the SMB process has led to optimization tools with different levels of complexity and to new SMB schemes, e.g., VARICOL, PowerFeed and ModiCon, which allow for further improvements in the efficiency of the SMB units. Even though the possible gain from optimization of SMB processes is proven to be significant, the performance of the optimization algorithms is limited by the quality of the available physical data. Unfortunately it is also a fact that the precise characterization of the separation system is a rather difficult task. Moreover, the optimal conditions depend on the operating parameters, i.e., feed concentration/composition, and the physical parameters of the system, i.e., the adsorption isotherm and column properties, which are both subject to change. Therefore, frequent re-characterization of the system and re-optimization of the process are required to account for the changes. Another issue is the high sensitivity of the SMB units to the disturbances, e.g., pump or temperature instabilities, and to above uncertanities that may lead to sub-optimal operation conditions in the best case scenario and off-spec production in general. As a consequence, robust operation of the SMB plants at their economic optimum is still an open issue. The common practice is to operate the SMB units far from the optimal operating conditions to guarantee a certain level of robustness. Development and implementation of robust SMB control schemes has the potential to deliver the full economic potential of the technology. On the other hand, feedback control of SMB process has been regarded as a challenging task for the standard control algorithms not only because of its periodic steady-state, nonlinear, mixed discrete and continuous nature but also due to long delays in exhibiting the effect of disturbances. This thesis proposes a new on-line optimization based SMB control strategy, where a linearized time-varying reduced-order model, that accounts for the mixed discrete and continuous nature of the SMB process, is used to predict the evolution of the plant dynamics. Four internal flow rates, which can be adjusted via external flow rates, are used as the manipulated variables. On-line concentration measurements at the product outlets together with a periodic Kalman filter are used to correct the possible model prediction errors. The SMB control problem is defined as a general constrained dynamic optimization problem to be solved on-line which yields the optimal control inputs allowing for the achievement of process specifications and optimal performance. The developed SMB control scheme has several noteworthy features. Firstly, it only requires the knowledge of the adsorption behavior of the compounds under linear chromatographic conditions, even if the SMB unit operates under nonlinear chromatographic conditions. Hence, it makes the detailed adsorption isotherm measurements redundant. Secondly, it makes use of the average void fraction of the columns in the SMB loop, and therefore, it removes the necessity of detailed characterization of the columns. Hence, it addresses all the above difficulties. The thesis demonstrates the effectiveness of the controller when applied to a virtual SMB plant in the case of systems characterized by both linear and nonlinear isotherms. The performance has been tested thoroughly under extreme model/plant mismatch conditions or large disturbances of various origins. Moreover, the proposed SMB control scheme has been experimentally validated by implementing on a lab-scale eight-column four-section SMB unit used for the high purity separation of the nucleosides (uridine and guanosine). The performance of the controller is proven via several test runs that are designed to challenge its robust performance. This is the first time that a standard experimental SMB plant, i.e., with four sections and two product outlets and used for high purity separation, is successfully optimized and controlled on-line.


Type of Publication:

(03)Ph.D. Thesis

M. Morari

File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhDThesis { Xxx:2004:IFA_2123,
    author={G. Erdem},
    title={{On-line optimization and control of simulated moving bed
Permanent link