Development of an energy consumption benchmarking model for platinum group metal comminution circuits

Abstract

The mining sector plays a pivotal role in bolstering South Africa's GDP. Nonetheless, it has encountered numerous obstacles of late, most notably a marked surge in energy costs. The extraction of platinum group metals (PGMs) is particularly energy-intensive, relying heavily on comminution circuits for processing and electric furnaces for smelting. Yet, electric furnaces must operate at full power to reach the necessary temperatures, complicating efforts to enhance energy efficiency. Nevertheless, comminution circuits possess various adjustable variables that offer opportunities for optimising energy use. The primary challenge lies in enhancing the circuits' energy efficiency without compromising production goals or altering the size of the product particles. Hence, this study aimed to develop an energy benchmarking model for comminution circuits at PGM concentrators, employing metrics and data analysis to evaluate the performance of these circuits. Two energy benchmarking models were developed: an average benchmark model and a frontier benchmark model. Single or multivariable regression functions can be used to predict the effect of given variables on the energy consumption of the comminution circuits. These functions were developed for the critical milling circuit components, including pre-milling, primary milling, and secondary milling. The above measures resulted in an average benchmark model that constructed through ordinary least squares regression. In contrast, the resulting frontier benchmark model employs corrected ordinary least squares and stochastic frontier analysis derived from the average model. These benchmarks facilitate the creation of a streamlined scoring system, assigning a performance score to each milling circuit relative to others. This approach enables a performance ranking and ongoing monitoring of the circuit's efficiency. The scoring system indicated that the comminution circuits performed between 98-114% of their expected performance, and 73-127% of the frontier performance. Furthermore, the developed score-based method ranked the performance of the circuits with an 92% accuracy, when compared to the current intensity benchmarking. However, the score-based method aims to address the shortcomings of the current intensity benchmarking method. By spotlighting inefficient sections and technologies within circuits, it narrows down the areas requiring scrutiny for inefficiencies. This targeted approach allows for strategic enhancements to the circuits, resulting in significant energy savings, lowered carbon emissions, and diminished energy intensity.