A study on model predictive control in paste thickeners with rake torque constraint

Abstract

As a separation process, a paste thickener produces underflow with a high solids concentration. Such underflow leads to a high rake torque which could cause serious operational problems such as underflow blockage, pumping problems and potential donut formation. In this work, a model predictive control approach has been developed to control the underflow solids concentration subject to operational constraints. State observers including linear and extended Kalman filters are studied to determine a cost-effective approach to estimating the solids concentration profile in the paste thickener, which is important for thickener control. A rake torque model is validated with industrial plant data. By utilising the monotonic property of the rake torque model, a linear model predictive control (MPC) approach is developed to deal with the nonlinear constrained control problem. Using an industrial paste thickener as a case study, simulation results demonstrates that the proposed control approach, including an extended Kalman filter, can effectively regulate the underflow solids concentration within permissible operating windows, preventing rake lifting and pumping problems.