Using Geostatistics to Predict the Characteristics of Washed Coal

Stringent controls now placed on the quality of coal make it important to be able to predict not only the tonnage and the grade of the coal in situ, but also the recovery factory and average grade after the coal has been washed. The established techniques of linear geostatistics (ordinary kriging) can be used to estimate the total tonnage and the grade of the coal in situ, but more sophisticated techniques are required for predicting the characteristics of washed coal. Two approaches are currently being investigated at Fontainebleau. One involves a parametric model of the washability curves and disjunctive kriging. This is described in Armstrong (1980(a) and (b)). The other is similar to the service variable approach used for estimating recoverable uranium reserves. The method is described in this paper.