Iron Grinding Advanced Control Implementation to Improve Mineral Throughput and Size Classification

摘要

For a continuous and effective operation of any process, it is required a reliable control system which also minimizes human intervention. Therefore, CAP Mineria decided to update the control system of the iron grinding area from the Huasco's Pellets Plant which has been controlled by traditional PID controllers for several years. The challenges have been to design and implement a control strategy that maximizes the mineral processing, reduce energy consumption and stabilized size classification. The iron grinding control system includes from the ore bin which feeds the ball mill to the hydro cyclones cluster that classifies the mineral. Besides, the quality of the processed ore is measured by laboratory samples, so it is not an on-line variable to include in the control system, therefore hydro cyclones' pressure and density are the main variables to stabilize size classification. To better meet these demands, the advanced control system was devised with two layers. The lower layer objective is to stabilize the process based on model predictive control tools, chosen by its good stability, convergence and delay handling. The second layer is a supervisory control that aims to optimize the process targets, making decision to increase performance and automatically adapt its performance to changing ore and/or process conditions. After some months of operation, preliminary results show an increase in throughput up to 3.9% and a decrease in variability of the size classification main estimator up to 60%. Additionally, the energy consumption has decreased about 18%. This type of advanced control strategy is appropriated to replace outdated reactive control strategies in all process with lags or long dead times and high variability. A control system that can adapt itself to changing process conditions, anticipating disturbances and incorporating process behaviour knowledge represents a further step towards autonomous operated plants.