Control and diagnosis of simulated moving bed chromatography
A Simulated moving bed (SMB) process that uses simulated counter-current movements of the solid and liquid phases was introduced in 1960s. The conventional four-zone SMB process forms a closed ring with several packed-bed columns. This ring is divided into four zones by two inlets (feed and desorbent) and two outlets (extract and raffinate). The positions of four ports are shifted to the same direction of the liquid phase flow to simulate the counter-current flow of the solid phase. A feed mixture continuously enters into the process through the feed port, and the solute-free solvent, usually the same solvent composition as feed, enters into the process through the desorbent port. The mixture components are separated in the chromatographic columns and collected at two outlets (raffinate: less-retained components, extract: more-retained components). Because of the periodic repetition of port switching, the SMB process does not reach the steady-state but cyclic steady-state (CSS). The cyclic repetition of the outlet profiles causes difficult monitoring of the process output. The most common method is collecting process outputs for one or several port switching intervals and analyzing them with external analysis unit, which provides delayed process output.
During the last years, several powerful control concepts for SMB processes have been developed. Most of them were based on model predictive control, i.e. repetitive dynamic process optimization, which is computationally challenging if a rigorous nonlinear process model based on partial differential equations is applied. To overcome this problem, a nonlinear mixing cell model was applied to estimate the migration of internal concentration profiles in conventional four-zone SMB chromatography. The number of cells assigned in a chromatographic column was dramatically reduced by introducing a 'progressive but fictitious dispersion' to counteract numerical dispersion, which corrupts the numerical solution for a small number of mixing cells. By means of this approach, rather steep concentration fronts (shock wave) can be described with a very small number of mixing cells. Based on this model, a model predictive controller was developed to dynamically find the optimal operating conditions that satisfy product purity and recycle stream concentration constraints. In a simulation study, the proposed controller successfully estimated the process states by estimating isotherm parameters and dispersion coefficients, and the operating conditions were optimized ‘switch by switch’. Furthermore, in an experimental validation, the controller was successfully applied to a pilot-scale SMB unit for the separation of bicalutamide racemates, which contains complex system void volumes and provides delayed feedback information.
Simulated Moving Bed (SMB) processes are currently operated successfully in various scales in the petrochemical and sugar industries. The basic concept and the Sorbex™ processes were developed by UOP. Many of the currently running units are well adjusted to their specific separation tasks.
An operating industrial unit initially installed in the 1980’s is reevaluated. Since its start up the unit has been successful in continuously splitting a rather constant feed stream into two product streams. The original design was based significantly on large empirical experience without detailed knowledge regarding the underlying competitive adsorption isotherms. More recently alternative feed streams characterized by composition fluctuations became of interest. A required adjustment of the settings of the unit addressing these new conditions was found to be not straight forward. First attempts caused a reduction of important performance criteria.
Therefore a diagnostic approach is used to increase the range and flexibility of operation of the industrial unit.
The approach first uses the well-established equilibrium theory as a diagnostic tool to classify available process data from the past, in particular the purities observed at the two outlets. With this and the assumption of competitive Langmuir adsorption isotherms a first estimation of isotherm parameters and the location of the separation region can be made. These first estimates allowed the identification of critical regions of the internal flow rates. In these regions selected rationally planned perturbation experiments can be performed using the industrial plant within tight limits. Systematic variations of the regeneration streams provided improved understanding of the process and allow for refinement of the parameters and the thermodynamically confined region.
In a second line of investigations the equilibrium theory is used again to identify and evaluate the size and shape of another region defined by the specific restrictions of the unit as installed, e.g. accessible ranges of the pumps and valves installed, the maximal pressure drop, limitations due to various productivity goals. A promising range of operating parameters for this SMB unit is then identified by the overlap of the two regions estimated.
Combining the vast empirical knowledge and the additionally acquired information the diagnostic approach described allowed increasing the range and flexibility of operation of the industrial unit within its given limitations.
Lee.J. W. (2021). Double-Layer Simulated Moving Bed Chromatography for Ternary Separations: Serialized Layer Configurations, Industrial & Engineering Chemistry Research, 60, 8911-8926.
Lee, J. W., Kienle, A., & Seidel-Morgenstern, A. (2020). On-line Optimization of Four-Zone Simulated Moving Bed Chromatography using an Equilibrium-Dispersion Model: I. Simulation Study. Chemical Engineering Science, 225: 115810.
Lee, J. W., Kienle, A., & Seidel-Morgenstern, A. (2020). On-line optimization of four-zone simulated moving bed chromatography using an Equilibrium-Dispersion Model: II. Experimental validation. Chemical Engineering Science, 226: 115808.
Lee, J. W. (2020). Expanding Simulated Moving Bed Chromatography into Ternary Separations in Analogy to Dividing Wall Column Distillation, Industrial & Engineering Chemistry Research, 59, 9619-9628.
Suvarov, P., Lee, J. W., Vande Wouwer, A., Seidel-Morgenstern, A., & Kienle, A. (2019). Online estimation of optimal operating conditions for simulated moving bed chromatographic processes. Journal of Chromatography A, 1602, 266-272.
Lee, J.W., Seidel-Morgenstern, A., (2018). “Model Predictive Control of Simulated Moving Bed Chromatography for Binary and Pseudo-Binary Separations: Simulation Study” IFAC-PaperOnLine, 51-18, 530-535.
Suvarov, P., Wouver, A.V., Lee, J.W., Seidel-Morgenstern, A., Kienle, A., (2016). “Control of incomplete separation in simulated moving bed chromatographic processes” IFAC-PaperOnLine, 49-7, 153–158.
Gas chromatographic enantioseparation of chiral inhalation anaesthetics
This project, done in collaboration with Otto von Guericke University, is devoted to provide pure enantiomers of fluorinated volatile anaesthetics isoflurane and desflurane using gas chromatography. For that purpose a cyclodextrin based selector developed in earlier work was immobilized on porous glass beads. The research procedure includes simulation of the process using thermodynamic data for the applied stationary phase. The goal of the simulation study is to improve the productivity and analyse the performance of larger columns applying simplifying scale-up rules. Additionally, the development of the process comprises a capture step for isolating the pure components and storing them after the separation in non-selective columns. The last step involves experimental evaluation of the predictions and final production of pure enantiomers.
Mutavdžin, I. , Munkelt, T. , Enke, D. and Seidel-Morgenstern, A., (2019). Gas chromatographic enantioseparation of fluorinated anesthetics: Single column performance and scale‐up estimation. Chem. Eng. Technol, 42, 241 - 251.