A predictive modelling approach towards developing an economic block model for an open-pit diamond mine

Abstract

At Kao Mine, the current value to pit estimates are restricting the identification of localised variations in size frequency distribution (SFD), large-stone recovery potential, and value distribution. This study addresses this limitation by developing a block-level, elevation-based forecasting framework that integrates mining, geological and sales data. The research focuses on three key business metrics, namely dollar value per tonne ($/tonne), stones per hundred tonnes (SPHT) and carats per hundred tonnes (CPHT). Initial analyses included descriptive statistics, correlation and regression to evaluate relationships between elevation, grade and value. Building on these insights, Empirical Bayesian Kriging in three dimensions (EBK 3D) was applied in ArcGIS Pro to generate spatially resolved forecasts across alphanumeric mining blocks and elevations. The results demonstrate that 3D geostatistical modelling significantly improves zone-specific forecasting compared to conventional approaches, particularly when supported by robust historical and continuous datasets. Even in cases of limited data, such as for large diamonds (≥10.80 carats), the model shows strong potential to identify zones with higher probabilities of high-value stone recovery. However, model reliability is enhanced when integrated with geological interpretation, mining history and continuous data updates.This study contributes to both academic and practical domains by advancing diamond resource evaluation and enabling more granular, data-driven mine planning. The proposed framework has the potential to improve economic decision-making in open-pit operations, particularly for marginal diamond mines.