Resource estimation and the Kunwarara magnesite deposit

The Kunwarara magnesite deposit, in Central Queensland was selected as a study site to evaluate the use of different resource calculation methodologies on tonnages and grades within the deposit. Initial appraisal consisted of classical statistics completed on 6 frequently analysed elements, magnesium, silica, manganese, iron, calcium and aluminium. Sample distribution was skewed, and indicative of mixed sample populations. This was attributed to a combination of factors, the most important being groundwater geochemistry in particular changes in pH. Samples were then split by lithology into sand, silt and clay fractions, and the statistics re-run. Again, the main factor controlling sample distribution was suggested to be groundwater geochemistry. Variography was completed for three dimensional and downhole orientations. Results showed that there were coherent ranges for magnesium, iron, aluminium, and manganese in the sand fraction. The silt lithology displayed better variograms from all elements, with only calcium returning unclear variogram ranges. In the silt fraction, iron and magnesium were not clear, while the remaining elements displayed clear ranges. A feature of all variograms was the "noise" displayed. This was attributed to the mixed sample populations of each element and the sampling technique employed on the minesite. Five resource calculation techniques were tested: inverse distance squared and cubed, ordinary kriging and indicator kriging on both composite and raw data. Tonnage and grade curves produced for each method indicated that there was relatively little difference between the methods in terms of tonnage or grade estimation. Generally, the inverse distance cubed method produced the highest tonnages and grades, while the indicator kriging (raw data) method returned the most conservative grades and tonnages. By evaluating the variance, mean and median returned from each block model, it was determined that indicator kriging was the most appropriate resource calculation method. Using the indicator kriging method on raw assay data, a number of block sizes were evaluated. The 100 x 100 x 3 metre block size showed the minimum sample variance and is of sufficient size to allow for easy mine planning in a 3 million tonne per annum operation