The ore extraction from the nature has always been a difficult and challenging task to achieve, yet it is still tempting enough for many mining companies around the globe because of its high profits and contributions to the countries’ economy. Mining is a capital intensive industry where the costs tend to vary depending on the characteristics of the individual deposits, engineering and economic feasibility of different mining methods, processing plant and equipment requirements, infrastructure needs and future rehabilitation requirements. All these capital expenditures are referred to as sunk costs that cannot be recovered by sale or transfer of the corresponding asset. Thus, the major aim in mining industry is always to extract the maximum ore tonnage from the ground at minimum cost.
Given the mining of the laterite-type bauxite deposits, the resource models are generally constructed based on the geological and chemical composition of the core samples obtained from the sparsely spaced boreholes drilled on a regular grid. The size of the drilling grid is chosen such that the drilling costs are minimised, while the grade variation can still be reliably estimated throughout the area. However, because of the weathering process involved in the formation of the laterite-type bauxite ore, the base of the bauxite unit tends to be highly variable, which complicates the fixed productive capacity in the short term mine plan. The accurate volume estimates are also difficult to derive using the traditional interpolation techniques, because of the fact that the borehole spacing of an economically viable drilling programme is often greater than the spatial frequency of the variations in profile thickness. One solution to this mining problem is the saturation (in-fill) drilling that can detect the peaks and troughs between the adjacent sparsely spaced boreholes. This practice may however significantly increase the drilling costs to the extent that the mining of bauxite ore may no longer be feasible. In addition, even the saturation drilling grid may sometimes not be dense enough to detect the short-scale lateral variability of the bauxite/ironstone interface.
However, the geophysical surveys, which can provide continuous data acquisition and much larger areal coverage, might be an alternative to the costly saturation drilling programme at the laterite-type bauxite deposits. Provided that there is sufficient difference in physical and electrical properties of the bauxite and underlying ironstone units within the laterite profile, an appropriate geophysical survey, if properly designed, can image the bauxite/ironstone interface at the high resolutions and relatively low costs in comparison to the saturation drilling. The fine-scale geo-referenced data acquired by a geophysical method can then be augmented with the borehole data obtained from the sparsely spaced discrete locations through the geostatistical algorithms, resulting in more detailed surfaces representing the lateral variability of the bauxite/ironstone interface. In this way, the uncertainty associated with the lower contact of the bauxite unit can be alleviated and as a result the more representative resource models can be constructed.
Considering the bauxite mining process, the quantity and the chemical composition of the bauxite ore being sent to the processing plant are of crucial importance in terms of the consumption of the chemical substance (caustic soda) being used to produce alumina (Al2O3) from the bauxite ore. For instance, at the Weipa bauxite mine in Australia, every 1% of ironstone (ferricrete) included in the excavated product causes a 0.2% increase in reactive silica which increases the Bayer processing cost by about 80 cents per tonne. At most lateritic bauxite deposits, the mining of the bauxite ore is performed by means of front-end-loaders, as the pisolitic bauxite is of free flow characteristics. As the exploration boreholes are sparsely spaced, the bauxite/ironstone interface specified by the resource model is more likely to be different from the one that is featured by the actual front-end loader regions. As a result, the ironstone dilution and the bauxite ore loss that are likely to occur at the time of mining cannot be quantified and linked to the mine plan. If representative front-end-loader regions are created based on the predicted bauxite/ironstone interface, the ironstone dilution and bauxite ore loss that occur during mining activity can be quantified.
In this thesis, the lateral variability of the horizon surfaces representing the bauxite/ironstone interface at two selected mine areas, namely the Oak and Kumbur at the Weipa open cut bauxite mine in Australia were predicted using the sparsely spaced borehole data on a regular gird of 76.2 x 76.2 m and the fine-scale high-resolution ground penetrating radar (GPR) data through various statistical and geostatistical algorithms. The predicted surface representing the bauxite/ironstone interface was compared with the surface estimated from the sparsely spaced borehole data alone and with the actual mine floor. The cross-sections clearly indicated that inclusion of the GPR data in the prediction process substantially improved the overall estimation quality. The front-end-loader region-based mining strategies were proposed to account for the quantity of the ironstone dilution and bauxite ore loss occurring at the time of mining at the Oak and Kumbur mine areas. It was found out that the bauxite ore tonnage that was calculated using the surfaces representing the “no ironstone dilution” mining strategy was very close to the one calculated from the actual case. The spatial uncertainty associated with the in-situ bauxite ore tonnage was also assessed by the univariate and multivariate conditional simulation algorithms.