Possible improvements in mine valuation due to the use of the gamma-ray fluorescence gold-ore analyser

This paper deals with an experimental programme carried out with a gold analyser in the period June to September 1979 at Loraine Gold Mines. The results were found to be reproducible, with no overall or conditional bias when the measurements were compared with those obtained from chip samples. The standard error of estimation of a 3 m reef exposure, which was the sampling interval, can be reduced by a factor of 2,5 if the complete stretch is scanned, rather than the taking of one chip sample in its centre. Complete statistical and computational procedures were developed for the fitting of the normal-Iognormal model to observed analyser data and for the determination of the most likely gold grade for a particular analyser measurement. (This corection procedure is necessary since a significant number of negative analyser measurements can be observed.) Panel averages of uncorrected scans showed a small conditional bias relative to those of corrected scans. However, uncorrected panel averages can be used for on-the-spot decisions regarding whole panels. For standard calculations of ore reserves, use should be made of corrected scans. Under reasonable assumptions, the use of the detector in the selective mining of low-grade narrow reefs such as the Basal Reef at Loraine Gold Mines, when compared with the conventional chip-sampling method, can result in a saving of as high as R40 million in present-money terms over the life of a project. The difference could be even larger in a highly variable reef like the B Reef at Loraine Gold Mines.