Dilution and geostatistics

"The usual complaint of mines is that grade does not match exploration expectations. Quite often, more tons are mined than expected but at a lower grade. This can be seen as dilution. Geostatistics can help predict which dilution can be expected both in terms of grade and tonnage for different mining methods.The usual vocabulary of geostatistics speaks of block distribution; block variance, size effect, information effect, all terms which say very little to the practitioner. In f act, all this can be rephrased in terms of dilution .In this paper, the following is presented - an example which shows how reserves change with the size of selection units and number of samples taken, and tables which show the dilution that are bound to happen when dealing with the commonly encountered log-normal distribution. Finally, the paper shows how grade tonnage computations should be made in the case of an arbitrary distribution. IntroductionGeostatistics progresses more and more rapidly if one judge s by the number and quality of papers published in Mathematical Geology or at APCOM meetings. More and more case studies describing successful applications are published (David et al., 1986)(Kwa and Mousset-Jones, 1987). It is estimated that more than 270/0 of the gold mines around the world use geostatistics. There is still, however, a lot of resistance from the remaining 73%. Geostatistics has more formulae and equations than it needs to reach year 2000. We will be well into the third millenium before all the models available now are actually used. The progress may come in the form of finding better ways to present known statistical results and getting away from the jargon of geostatisticians. One example may be in trying to explain how we can answer the most common plea of the mining industry: production grade is not what was expected in the feasibility study."