Grade control problems dilution and geostatistics: choosing the required quality and number of samples for grade control

"Geo-statistics was born from mining problems: grades not being what was expected; more tons of waste than expected; in other words, dilution. Geo-statisticians now claim to have solved many of these problems, but the solutions seem to have a hard time being picked up by practitionners. One of the reasons may be the jargon which geo-statisticians are using. The reason for jargon is that obtaining solutions required a strict formulation in a statistical framework. Practicing mining engineers and geologists are not used to such a framework, and they are not familiar with the jargon. In this paper, the authors re-examine dilution in operating mines and answer two extremely practical problems, using practical language as much as possible. Can dilution be reduced by improving the quality of blast hole samples? How many samples are needed to obtain a good ore-waste discrimination, i.e. to reduce dilution to acceptable levels? Citing real situations, this paper will try to show how fairly simple concepts can be used to answer both problems. It will try to present the solut ions using a minimum of jargon, even if the theoretical solution could be obtained in a few precisely worded paragraphs.The two problems considered are the following: the first one was encountered in the Panguna deposit of Bougainville Copper Limited, P.N .G., but it occurs in every open pit operation where selective mining is required. As the cut-off is lowered, more and more mining occurs in marginal areas. Grade control is mostly based on the grades obtained from blast holes, hence, it is essential that the quality of the blast hole samples be as good as possible in order to reduce dilution, waste being sent to the mill and ore being sent to the waste dump.The second example stemmed from the grade control technique used in many gold open-pit operations in Western Australia: channels are cut on lines 5 m apart and assays are taken every metre. The pattern in the same regardless of rock type, whether one is in the oxidized zone or not. The number of samples to be treated is enormous, hence, the question arises how to find an acceptable sampling grid, and whether changing it can reduce dilution at an acceptable cost.A solution will be given to each problem separately, although in fact it could be seen that the formal framework is the same in both situations."