Geostatistical analysis and simulation of a metallurgical parameter towards improved mining efficiency and confidence in the geological model of a kimberlite

Abstract

The construction of geological models of kimberlites during an evaluation project typically suffers due to a lack of outcrop and dependency on visual core-logging criteria. A case study is presented showing how the inclusion of a metallurgical parameter (% Dense Media Separator, or DMS yield) obtained from large diameter drilling can be used to: • enhance geological definition in a kimberlite, • corroborate and support the identification of geological units, and • contribute towards understanding the volcanology at the time of emplacement. DMS yield represents the ratio of the mass of wet concentrate/mass of wet head-feed to the sampling plant and is a relative value. The %DMS yield data exhibited spatial structure within the three lobes which allowed the construction of variograms for unit M/PK, the dominant kimberlite type in the South lobe of the AK06 kimberlite. This kimberlite type revealed an extreme range of %DMS yield values which will present a challenge to the recovery of diamonds. It was therefore essential for mine planning purposes that zones of high %DMS yield were accurately defined and quantified. Ordinary Kriging was applied to obtain the “best linear unbiased estimate” of %DMS yield at a local block scale. Conditional simulations using the Turning Bands and Sequential Gaussian methods were generated to quantify the variance of %DMS yield and the potential uncertainty. Indicator kriging was applied to the kimberlite to obtain the probability of intersecting %DMS yields above a particular cut-off (20%) which the main treatment plant design could not accommodate. A possible reason for the high %DMS yield values in the kimberlite was proposed and the location of problematic zones illustrated in 3-D space. This study represents pro-active use of geology to investigate resource risk, delineate problem areas in advance and develop a geo-metallurgical model.