Clustering mining blocks in presence of geological uncertainty

Abstract

A major trend in mine production planning research is incorporating geological uncertainty in the processes of planning. Many mathematical models and heuristic approaches are proposed to deal with the uncertainty. Although there have been advances in exact methods to solve simpler instances of the mine production scheduling problem more complex instances of the model, especially when incorporating uncertainty, remain intractable and aggregation of blocks can help to decrease solution times. In this paper, we present four variations of the agglomerative hierarchical clustering algorithm, one based on deterministic estimates of properties and three based on possible worlds approach which use Geostatistical realizations to form aggregates with regard to the geological properties and the existing uncertainties. We show, through case studies, that uncertainty-based algorithms can result in aggregates that are less susceptible to uncertainties, and at the same time, the proposed algorithm can produce aggregates that are within a controlled size and have minable shapes.