Dynamic real-time optimization to mitigate critical state effects in expert-controlled SAG mills

Abstract

Semi-autogenous (SAG) mills are critical components in mineral processing, widely used for size reduction to liberate valuable minerals in ore. These mills account for a significant portion of energy consumption and operational costs in mineral concentrator plants, making their optimization a key factor for improving overall process efficiency. Traditional expert control systems used in SAG mill operations lack predictive capabilities, which can result in critical operational states requiring a halt in the process and significantly affecting productivity.

This work introduces a dynamic real-time optimization (D-RTO) strategy to mitigate the impact of critical states, such as overload or high pressure, in SAG mills controlled by expert systems. The proposed supervisory layer dynamically adjusts the limits of controlled variables based on predictive capabilities provided by a digital twin. Simulation results demonstrate significant performance improvements, including reductions in total tonnage loss of up to 57%, decreases in critical state duration by up to 33%, and increases in average throughput by up to 3.5%. These results underline the effectiveness of the D-RTO strategy, particularly under dynamic and variable conditions. Future research will expand the framework to include power consumption considerations for comprehensive economic optimization.