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.


Automatic optimizing control of simulated moving bed processes


S. Jermann

Master Thesis, FS 09

Simulated Moving Bed (SMB) chromatography is a cost-efficient separation technique that offers high productivity and low solvent consumption. Besides the cost effectiveness, SMB allows for fast and reliable scale-up of separations from analytical chromatography and thus a shorter time to market can be achieved. Therefore SMB has gained importance in the areas of pharmaceuticals, fine chemicals and biotechnology over the past decade. Modeling, design and optimization of SMB units are regarded to be well established, however, long term robust/optimized operation is still a challenging task. This problem has been addressed in the past and a ’cycle to cycle’ SMB control concept, based on repetitive model prediction control has been developed. This controller made use of a simplified first principle model to predict the future evolution of the plant on basis of its estimated current state. This model required only the determination of the Henry constants and the overall void fraction of the columns constituting the SMB plant. Although being based on very few information, the controller was able to meet the products’ purity specification both under linear and nonlinear chromatographic conditions. This was proven in the past experimentally for the resolution of racemic guaifenesin. The present master thesis is based on these previous results and is divided into two parts: The first part consisted of a model comparison based on simulations whereas the second part dealt with the experimental implementation of the control concept. In the first part of this thesis, the established simplified first principle model was compared to an identified model using a virtual SMB plant. The identified model was derived by nuclear norm identification (NucID), a novel method that aims at identifying the lowest order model that accurately describes the system dynamics. It has been shown that NucID yielded a 7th order model whose impulse response was basically equivalent to the one of the simplified first principle model, which was of 11th order. Furthermore it was shown that the closed-loop performance of both models were very similar, however being the performance of NucID slightly superior. The second part of the thesis dealt with the experimental implementation of the control scheme to separate racemic Troeger’s Base on Chiralpak AD. In order to solve this task, a new controller, based on simplified first principles, had to be designed from scratch and an existing lab-scale SMB unit had to be adapted to this specific separation. Afterwards correct operation of the plant was verified in several open-loop experiment and finally four closed-loop experiments both under linear and non-linear chromatographic conditions were run. The results show that the controller performed equally well with respect to the one previously implemented for the guaifenesin separation. Hence this thesis has proven that the control concept is simple enough to be implemented quickly and reliably for new separation tasks.


Type of Publication:

(12)Diploma/Master Thesis

M. Morari

No Files for download available.
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2009:IFA_3440
Permanent link