Inferential Control Strategy for Continuous Crystallizers

The main characteristics of a crystal population from a crystallizer to be controlled are, in order of importance, the amount of fine crystals and the average size of the overall crystal product. Therefore a suitable control strategy may be based, firstly, on a feedback control loop of the magma density of fines and, secondly, on a correction of the set-point of the fines magma density in order to obtain a more convenient value of the average crystal size. The on-line measurement of the magma density can be provided by quite simple instruments, like as turbidimeters based on laser-light diffraction, whereas the on-line measurements of the average crystal size is a difficult and expansive task. Thus the estimation of this latter variable by means of a Projection to Latent Structures (PLS) algorithm is proposed in this work, in the frame of an inferential cascade control strategy. The investigated system has been a cooling crystallizer of potassium sulphate from aqueous solutions at bench pilot scale. The feasibility of the proposed control startegy has been checked by using a dynamic model of the experimental apparatus.