Wave propagation velocities in integrated PLS based control of a simulated moving bed process

The simulated moving bed (SMB) is a continuous chromatography process for separation of chemical mixtures. It has been used in large-scale industrial production of cheap specialty products like sugar separation, as well as small-scale production of high value products like enantiomer separation in the pharmaceutical industry. Physical models can be employed in control of SMB processes, but it can be difficult to obtain accurate physical parameters. Empirical models do away with the need to obtain parameters, but without physical representation, they may not be suitable for control applications. In this research paper, wave propagation is linked with the empirical model to enhance the physical constraint on designing SMB. However, there is an interaction between the zonal wave velocities and process outputs; thus, control strategies based on the partial least squares (PLS) are proposed in this work. PLS decoupling of the system to achieve pairs of SISO (single input single output) in the subspace is useful because the conventional SISO can be directly applied to the control design. The use of wave propagation allows the application of the constraints in Zones II and III based on the shifting of the wave fronts. Two modes of control are proposed: cycle to cycle and within cycle. The performance of these control strategies is assessed by implementation on a virtual eight-column SMB unit for set-point tracking and disturbance rejection.