PanAust’s Phu Kham copper-gold operation (Phu Kham) in Laos is currently limited by throughput in the Semi Autogenous Grinding (SAG) mill. The ability to accurately predict SAG mill throughput (throughput) is therefore critical to accurately forecast copper and gold production. During the fourth quarter of 2016 the proportion of hard unweathered rocks in the mill feed increased, resulting in a period of lower than expected throughput. These low throughput rates prompted a study with the aim of identifying a practical solution for throughput predictions. The study identified an empirical modelling approach based on actual SAG mill throughput rates. The empirical model was then optimised by linear programming to identify the ideal feed blend of unweathered rocks to achieve a maximum SAG mill throughput.
Rock strength was the initial focus of the throughput study at Phu Kham because previous studies had identified a strong link between rock strength and throughput. It was determined that there was insufficient valid data to adequately model but sufficient data to qualitatively characterise rock strength. The data identified that the weathering and lithology had a controlling influence on the rock strength. Complicating throughput predictions were operational improvements (changes to blasting and mill settings) implemented in response to the lower than expected throughput. This created uncertainty with the throughput equations. A self-learning empirical modelling method was developed to predict the throughput by incorporating fundamental rock properties with operational practices and improvements. The method is based on the proportion of mill feed with similar weathering and lithology (a proxy for comminution performance) and the actual mill throughput using the SAG mill as the analytical instrument. The modelling method simultaneously incorporated the sum and interaction of unmodelled influences including blasting, crushing and mill settings. The empirical modelling method worked at Phu Kham because the lithology and weathering has an intuitive and observable controlling influence on SAG mill throughput. The modelling method is described along with a worked example of feed blend optimisation using linear programming.
Carpenter, J and Saunders, B, 2017. Empirical mill throughput modelling and linear programming for blend optimisation at the Phu Kham copper-gold operation, Laos, in Proceedings Tenth International Mining Geology Conference 2017, pp 291–296 (The Australasian Institute of Mining and Metallurgy: Melbourne).