Advanced control with fundamental and data-based modeling for falling film evaporators

Historical data from a falling film evaporator (FFE) was used to identify both fundamental and data based models without knowledge of the physical dimensions of the plant. The fundamental model showed the highest product dry mass (R) correlation to historical data (0.56) while the data based models struggled on validation data (R of 0.45). PI and LQR (linear quadratic regulator) control were designed with a Kalman estimator successfully used for state generation. The LQR solution provided better disturbance rejection, decreasing the maximum deviation in product dry mass by 60% when subjected to a feed disturbance. Both controllers outperformed operator control while keeping the process at optimal conditions.