The influence of variability models for selected geological parameters on the resource base and economic efficiency measures - Example of coking coal deposit

Abstract In the research we examined the influence of a few geostatistical models (interpolators and kriging models) on the resource base and economic efficiency measures, on the example of coking coal deposits. The research was carried out in scenario-based mode, with 4 alternative analytic models prepared. The assessment criteria for fitted variability models based on deviations of selected technical and economic indicators and parameters, such as: volume of reserves, price of coking coal, production cost, EBITDA, NPV, IRR, mining height, coal yield, total waste volume, mine lifetime were examined. The variability of seam structural parameters and coking coal quality parameters was modelled with use of interpolators based on finite element method (FEM), inverse distance weighting method (Inverse), as well as triangulation coupled with extrapolation (Planar). Also model C was built separately, where ordinary kriging was used to assess empirical variograms and theoretical models of quality and quantity parameters. A dedicated discount model was used for generating the resultant economic and financial indicators. Monte Carlo simulation (bootstrap method) was applied with empirical copulas as binding functions. It was demonstrated in the course of the research that the selection of geostatistical model is of significant importance for the assessment of geological, technical and economic potential of the analysed deposit. It was shown that for the mean size of resource base, in the order of 44.6 million Mg (Ref. Model), the differences in estimated volume of reserves may reach 18%. For relatively insignificant differences in operating costs and coal price distributions, the highest differences in totalled EBITDA amounted to some 2.054 billion PLN (18%). NPV differences were calculated at maximum of 16%, whereas the IRR deviations reached 9%. Depending on the interpolator used, the mining height may differ relatively by maximum of 18%, mean waste volume by 17%, mean coal yield by 7%, while mine lifetime by maximum of 10%. Relative differences may reach a few dozen percent (in certain cases of other technical and economic parameters). Based on this data, we formed the conclusion that these differences are significant. Also the assessment of Spearman's rank correlation coefficients may provide insufficient information about correlation structure, what had been verified with use of empirical copulas.