Challenges of Quality Management in Sampling and Measurement of Geometallurgical Variables

The prime objective of geometallurgy is to improve the profitability of mines through the use of spatial models of rock properties that have a significant impact on value. Although a key property is the grade of the component of interest, because it directly influences the revenue from saleable material, increasingly attention is being focused on other attributes. These 'non-grade' variables may have a less direct, but equally profound impact on value, either by reducing potential revenue (eg lowering recovery, deleterious elements, etc) or increasing mining or treatment costs (eg comminution, grindability, throughput, reagent use, etc). Collectively, these variables have been described as 'geometallurgical'. There are three main areas of activity required to realise the value of geometallurgy: 1. acquisition of appropriate data through sampling and measurement, 2. building spatial models of geometallurgical variables, and 3. using these models to optimise value. Each of these is built on the foundation of the preceding step. In building reliable spatial models, it is critical that the amount of error introduced to data by the sampling and measurement processes is understood and quantified, in order that the inherent variability of the component of interest can also be properly understood. Management of the quality of grade variables is now applied (with different levels of diligence and success) across nearly all sectors of the mining industry, not least because of the necessity to adhere to Codes of practice (eg JORC, 2004). The necessity for quality management and at least the broad framework of what to do in practice is reasonably understood by many relevant professionals. There are also a number of tools, including software products, statistical analyses and presentation formats, available for those managing grade data. It is also abundantly clear, that in many cases people have little understanding of why they are doing particular quality control activities, or how to sensibly interpret the results. The same is not true for many geometallurgical variables. In almost all instances the authors have seen, the magnitude of sampling and analytical/measurement error variance embedded in geometallurgical test data is untested and unknown. This is partly because until relatively recently, metallurgical data has mainly been collected for the purpose of process and mill design, with few, expensive tests carried out on 'average' samples of the orebody. Increasingly however, there is a move towards collection of much larger data sets from spatially distributed sampling, and thus a growing necessity to understand the errors inherent in both sampling and analysis of these types of variables, and how to manage them. In order to understand how the quality management practices applied to grade variables can be extended to geometallurgical variables, it is essential to understand the theoretical foundation that supports these practices. This requires an analysis of the concepts that underpin the 'principles of additivity' (Matheron, 1963), the 'theory of sampling' (Gy, 1982, Pitard, 1993), and 'statistical process control' (Shewhart, 1930). Relevant concepts are then integrated to form a foundation to develop methods to improve the quality management of geometallurgical variables. Application of this framework will significantly increase the integrity of geometallurgical data, the reliability of geometallurgical models and ultimately the value derived from implementing geometallurgy. This is an ABSTRACT ONLY no paper was prepared for this presentation.