An efficient sample selection methodology for a geometallurgy study utilizing statistical analysis techniques

A geometallurgy study aims to link metallurgy and geology to reduce technical risk and enhance the economic performance of a mineral-processing plant. It does so by accounting for variability in a deposit to develop cash-flow models with variable throughput rates. High-quality sample selection for metallurgical test work that is representative of the deposit is an essential component of a geometallurgy study, but the large multidimensional data set makes sample selection a daunting task, as classifying the data set while respecting its heterogeneity is difficult. This paper presents a streamlined approach for sample selection, using statistical analysis techniques in Python. It cuts down time to select samples from around 1,200 s per drillhole to about 60 s for data classification and from 12 h to 8 h for handpicking samples from the classified data set, translating to cost savings. The cumulative sum method and k-means clustering method are used in the methodology to elegantly classify the data and select representative samples. The effectiveness of the methodology is demonstrated by presenting data from a prefeasibility study of a copper-iron mine in which 40 samples were selected for flotation test work.