Dynamic Probabilistic Mineral Resources Classification

Abstract:

Mineral resources classifications based on grade confidence criteria are becoming increasingly common in the mining industry. These criteria aim to measure the confidence level of the expected grade on panels of volumes equivalent to production volumes relevant for medium and long-term planning. Thus, monthly or quarterly production panels that show a high probability (usually 90% or more) of having a grade within a certain variability (usually ±15%) of the expected panel grade are typically used to support the Measured Resources category. Similar metrics evaluated over yearly panels are used to inform the Inferred Resources category. The confidence metrics are usually obtained from the post-processing of geostatistical simulations.

As Mineral Resources Classification occurs before mine planning and scheduling, the final shape of the production volumes is unknown. Consequently, the panels are assumed to be rectangular parallelepipeds of uniform size. This practice can result in the following inconvenient outcomes:

  1. Classification based on these regular volumes results in abrupt category boundaries that may introduce artifacts into the mine design.
  2. Depending on their location, these panels can contain different quantities of minerals. As a result, some may exceed the production targets while others may fall short.
  3. Large panels may contain different domains with varying degrees of geological variability. Inter-domain variability affects the overall grade confidence in the whole panel, potentially leading to the category upgrading of high variability domains but downgrading of high continuity domains.

To address these issues, the authors propose:

  1. Factoring geological and grade uncertainty in the Mineral Resources classification by the application of geostatistical simulation for modelling the contacts between mineralisation domains and the grades within them.
  2. The use of dynamic panels with changing sizes and moving centres.
  3. The application of the confidence metrics for each domain within the panels separately.
  4. The combination of geometric and probabilistic classification criteria.

As demonstrated in a case study from a porphyry deposit, this strategy results in smoother boundaries between Mineral Resource categories that honour the production targets and different geological and grade confidence levels within the domains present in the panels.


Click here to view all presentations at this event.