Studies on parameter estimation and model predictive control of paste thickeners

Paste thickeners have attracted significant interest from mining industry due to its higher dewatering ability as compared to conventional or high rate thickeners. However, the underflow solids concentration, which is an important process variable of thickeners, is often poorly regulated. In this article, a dynamic model based on sedimentation-consolidation theory is adopted and validated using industrial plant data. Based on this model, control studies have been carried out to explore approaches to address a number of difficulties in current industrial operation. An extended Kalman filter is developed to estimate the compressibility parameter of the feed (coal tailing). As a key process parameter, coal tailing compressibility plays a significant role in thickener dynamics, but is time-varying and difficult to measure. Potential improvements of process operation by implementing model predictive control (MPC) are investigated. Simulation studies show that the proposed control can deliver a higher underflow solids concentration and a better regulated underflow removal rate than the existing operation. It is also demonstrated that taking into account the "future" time-varying input constraints in the MPC algorithm can help overcome the current control difficulty caused by co-disposal of coal tailing and coarse reject. (C) 2015 Elsevier Ltd. All rights reserved.