A mineral company's resource models are a measure of its foundational assets that provide the basis for forward looking statements of corporate value and cash-flow estimates. Accuracy of the estimation process underpins corporate legitimacy. Importantly, local improvements in estimation process can translate into improvements in mine planning, and ultimately better-informed investment decisions. Traditionally, resource models use estimation parameters that are based on statistical patterns and spatial variability within a geologically informed volume constraint ('the domain'). The variogram, block size analysis and determination of search parameters are assessed from the data within the geologically delineated domain. The set of parameters so determined are then applied to every estimation block within the domain, and the block model is then provided to the mine planner for optimisation. The mine planning optimisation process responds to each block grades. The focus of the mine planning process is to minimise ore loss and mining dilution and so provide the best possible opportunity for the orebody and its value to be realised. However, overly smooth grade models restrict a mine planner's ability to achieve the best outcome for the project and for the asset owners. Despite the estimation of every block in a resource model being conducted independently of every other block in the model, Resource Geologists continue to generalise parameters across a domain of blocks. This paper challenges the global parameter approach, and instead seeks a more locally contextual set of parameters. This challenge is in keeping with innovations across industries and around the globe that seek real time bespoke responsiveness built on big data, machine learning and artificial intelligence.
展开▼
