Anda di halaman 1dari 2

Adsorption Properties, Mathematical Modelling, Optimization, Control and Supervision of Simulated Moving Bed (SMB) processes

Conventional batch chromatography is relatively inefficient in terms of adsorbent and solvent consumption and significant benefits can be achieved by performing separation of high-added value products, such as enantiomers produced in the pharmaceutical industry, with a simulated moving bed (SMB) process (Lehoucq et al., 2000). The SMB process allows a counter-current movement of the liquid and the solid to be achieved in order to increase the exchange capabilities between both phases. In practice, there is no real solid movement but a "simulated" counter-current which is achieved by periodically switching the inlet and outlet valves in the direction of the liquid flow.

However, the transfer of the SMB technology, used industrially for hydrocarbons and sugars separation, to the separation of fine chemicals is not immediate. Indeed, the conditions and requirements (product quantities and purities, characteristics of the phases, interactions ...) are very different. The main issues are the selection of optimal operating conditions and process control, problems which require the development of mathematical models of the process. Two modelling approaches are commonly applied to SMB processes. The first one is called TMB (true moving bed) and assumes an equivalent counter-current movement of the solid phase (see Figure 1). The second, more rigorous approach, called SMB, considers the system as an arrangement of static chromatographic columns and takes the discrete nature of the solid movement into account. In the literature, it is generally agreed that the TMB model represents the average behaviour of the SMB and that the correspondence between TMB and the average SMB behaviour becomes better and better when the number of columns in the SMB increases. The TMB model brings significant reduction of the computational complexity (Haag et al., 2001) and can be used for a first analysis in design, optimization and control. However, the TMB model, as opposed to the SMB model, does not show the cyclic steady-state of the process and introduces modelling inaccuracies. The objectives of our research are as follows:

derive simplified system representations which capture the main observed phenomena, while keeping model complexity and computational costs at the lowest (Grosfils et al., 2007); carefully study system identifiability from batch and SMB experiments and propose systematic identification procedures (including error estimates) (Grosfils et al., 2007, under review); optimize the process operating conditions using various degrees of freedoms (feed flow rate, feed concentration, switching period) (Garcia et al., 2006); design robust control structures to maintain the plant close the optimal operating conditions despite process disturbances;

develop fault detection and isolation algorithms to face process evolution (for instance the ageing of the adsorbent phase) and failures. This research stream has been and is being supported by the following projects:

MOVIDA (2002-2006; in collaboration with the Control Department at ULB and CREGI-FUCAM) supported by the Walloon Region (DGTRE - Initiative 4). A pilot plant (NOVASEP) could be set-up in 2007 thanks to the financial support of the Belgian National Fund for Scientific Research (FNRS), the Free University of Brussels (ULB) and FPMs in the framework of a collaboration between the Control Departments of FPMs and ULB and the Thermodynamics Department at FPMs.

Anda mungkin juga menyukai