Expanding the Acceptance Region for Incoming Materials Via the Usage of a Mid-Course Correction Controller for Wet Granulation

Multivariate Latent Variable Methods are combined with advanced non-linear programming techniques to construct a feed-forward controller for the wet granulation operation. Four types of controllers are contrasted: i)feed-forward using only raw material properties, ii) feed-forward doing mid course corrections using power consumption and temperature measurements from the process, iii) same as ii with the addition of a NIR signal, iv) feed-forward doing mid course corrections using exclusively the NIR signal. Results show that the best performance is obtained when the controller uses the three trajectories available (power, temperature and NIR). These results are then used to illustrate how the acceptance criteria for the raw materials will be affected by the control strategy.