Optimization of Simulated Moving Bed Process Applying An Optimal Control Concept: Separation of Racemic Bicalutamide Mixture

Wednesday, October 19, 2011
Exhibit Hall B (Minneapolis Convention Center)
Ju Weon Lee1, Kerstin Larson2, Robert Arnell2 and Andreas Seidel-Morgenstern1, (1)Max-Planck-Institute for Dynamics of Complex Technical Systems, Magdeburg, Germany, (2)Pharmaceutical Development, AstraZeneca, Södertälje, Sweden

Simulated moving bed (SMB) is a popular separation technique in fine chemical and pharmaceutical industries. The basic concept of SMB is separating two components, which have different migration velocities in a chromatographic packed bed, with simulated counter-current of adsorbent and desorbent. The standard 4-zone SMB process consists of four zones divided by two inlet and two outlet streams, so that there are four design parameters normally described by the ratio of desorbent to adsorbent flows in each zone with one fixed operating condition. The optimal control of these four design parameters was performed applying a proportional-derivative (PD) controller to satisfy product requirements and operation constraints [1]. The optimal control can obtain unique operating conditions which satisfy the objective of each zone and these objectives were changed in the second optimization stage to achieve the global goal, the maximum ratio of productivity to desorbent consumption. All simulation work was performed to separate racemic mixture of bicalutamide [2]. The isotherm and rate parameters used in the simulations were measured with chiral stationary phase (CHIRALPAK AD, 20mm, 2.5cm I.D., 9.7cm length, Chiral Technologies Europe, France) and pure methanol mobile phase. The target product requirements are 99.5 % of purity and yield or 93% of purity and yield and the global goal is to maximize the ratio of productivity to desorbent consumption. Four to eight columns were used to change column configurations. In the case of low purity and yield requirements (93% of purity and yield), the 1/1/1/1 configuration obtains the maximum performance. The 1/2/1/1 and 2/2/1/1 configurations obtain better performance than others in the case of high purity and yield requirements (99.5% of purity and yield).


[1]   Karl J. Åström and Tore Hägglund, “PID Controllers: Theory, Design, and Tuning”, 2nd Edition, ISA, 1995.

[2]   Henning Kaemmerer, Matthew J. Jones, Heike Lorenz, and Andreas Seidel-Morgenstern, “Selective crystallisation of a chiral compound-forming system—Solvent screening, SLE determination and process design”, Fluid Phase Equilibria, 296 (2010) 192-205.

Extended Abstract: File Not Uploaded