Mineral exploration is a risky business due to the inherent geological and economic uncertainties that accompany the discovery and exploitation of mineral deposits. Consequently, the ability to quantify the risk in terms of the economic consequences of exploration can lead to improved decision-making and more profitable exploration. Quantifying economic risk in exploration requires an understanding of the economic value of exploration. This in turn requires information about possible numbers and sizes of undiscovered deposits, in addition to the economics associated with the discovery, delineation and mining of these undiscovered deposits. A method for quantifying exploration risk based on these requirements is detailed here and demonstrated via a four part case study, which examines the economic risk associated with exploration for undiscovered porphyry Cu and VHMS deposits in the Yarrol region (SE Queensland) and their contained copper and gold resources. The first part of the case study outlines the study rationale and background, and presents descriptive geological models for eastern Australian porphyry Cu and VHMS deposits. The deposit models were then used to delineate permissive tracts for each deposit type. In the second part, prospectivity modelling was carried out on a sub-block basis using a multivariate Bayesian weighted probabilistic neural network (PNN). The PNN was used to integrate geological data and estimate the conditional probability that a sub-block belonged to one of twelve classes, defined by a set of mineral deposit types. Prior to modelling, the geological data used in the PNN underwent preparation that involved quality control and conversion to a set of geological themes which accentuate characteristics associated with mineralisation in the Yarrol region. To compensate for inherent imprecision in the data, fuzzy set algorithms were used to sample the themes prior to PNN modelling. The algorithms were designed to handle the line, polygon and raster data using fuzzy membership functions and fuzzy entropy. The final PNN input vector contained 50 fuzzy variables, describing lithology, structure and geophysics. Because the PNN is an intrinsic classifier, which uses the training samples as nodes in the network, independent validation with classified samples not used in training is required to appropriately gauge performance. To overcome the issue of excluding classified samples from training, multiple PNN tests were carried out using randomly selected 70:30, training to validation, splits of the classified samples. In the case study 100 PNN tests were carried out and independent validation recorded a mean of 84 % correct classifications, with 4 % standard deviation. After each PNN test, unclassified subblocks covering the case study region were classified by the trained PNN and the results used to map the prospectivity of the region. The prospectivity maps highlighted both geological features and areas prospective for the different mineral classes. A total of 26 areas were modelled as being prospective for porphyry Cu and VHMS deposits of which 6 corresponded to the deposits used in training. In the third part of the case study, estimates of the undiscovered mineral resources in the Yarrol region were modelled using a Monte Carlo simulation. Simulation involved generating grade and tonnage models for eastern Australian porphyry Cu and VHMS deposits, and estimating the numbers of undiscovered deposits in the region using the probabilities generated by prospectivity modelling. Models and estimates were converted to cumulative distribution functions using histogram smoothing and normal score transformation. This approach differs from the documented proportional method and the benefits of using the new approach, such as better fitting and more representative grade and tonnage models, are shown. Results of the mineral resource simulation found that the simulated Cu resources in undiscovered VHMS deposits were much smaller than those estimated in undiscovered porphyry Cu deposits (i.e. Kt rather than Mt) while the probabilities for VMHS resources were an order of magnitude higher. For example, there was a 30 % chance of 100 Kt or more Cu in a VHMS deposit compared to a 2 % chance of 1 Mt or more Cu in an undiscovered porphyry Cu deposit. However, in the case of undiscovered Au resources the VHMS deposits had similar resource sizes to the porphyry Cu deposits but the probabilities were much higher for VHMS deposits (e.g. there was a 2 % chance of 5 Moz of Au, or more, in a porphyry Cu deposit and a 15 % chance of 5 Moz of Au, or more, in a VHMS deposit). These results reflected the high probability (i.e. 92 % chance) for zero undiscovered porphyry deposits and much higher Au grades in the VHMS deposits. In the final part of the case study the mid-term value (mine value) and the long-term value of exploration were estimated in terms of NPV and RoR. The mid-term values were calculated using mining cost models based on mining capacity calculated via Taylor’s rule. A new approach using variable mine capacity and scheduled development costs is presented and demonstrated. Results for the mid-term values indicate that economic risk was lower for resources in VHMS deposits. For example, the expected NPV1 for an undiscovered porphyry Cu deposit was $ 27 million compared to $ 53 million for a VHMS deposit and the probabilities of these occurring were 0.03 (a 3 % chance) and 0.2 (a 20 % chance), respectively. However, a risk assessment using utilities for the mid-term NPV indicated that a risk taker would favour porphyry Cu exploration because of the possibility for extremely large porphyry deposits. The long term value of exploration was estimated using an exploration model that randomly sampled both a discovery cost model (which mapped the cost of exploration to the probability of a discovery) and the simulated undiscovered resources (which act as possible targets for exploration). A range of exploration budgets were examined for both deposit types. Results demonstrated that smaller budgets had higher risk because they resulted in longer discovery times which reduce profits and increased the chance of missed economic opportunities due to Gambler’s Ruin (e.g. a budget of $ 2 million/year had a 50 % of Gambler’s Ruin after 6 years, whereas a $ 6 million/year budget had only a 3 % chance). It was also shown that economic returns levelled out as the budget increased because the increased chance of making a discovery is balanced out by higher exploration costs and lower discounted returns for the discovered deposit. Small and large exploration budgets were examined in detail and it was shown that in both cases exploration for VHMS deposits has less risk. For example1, a $ 2 million/year exploration budget targeting VHMS deposits had a 13 % chance of returning a NPV of $ 50 million, or greater, whereas the same budget targeting porphyry Cu deposits had only a 3 % chance of returning the same NPV, or greater. The RoR results displayed a similar outcome. The case study concluded with a sensitivity study examining the effects that variation in Cu price, Au price, exchange rate, mean exploration cost, development costs and operating costs has on the long-term economic values for both deposits types. In the case of porphyry Cu deposits both the NPV and RoR were highly sensitive to variations in Cu price and to a lesser extent operating and development costs. The long-term value of VHMS exploration was highly sensitive to the Cu price, Au price and the exchange rate. (the latter related to commodity prices and mining costs). It is recommended that future prospectivity studies could include geochemical and alteration data to assist in modelling the location of an undiscovered deposit. The use of historical exploration data (e.g. drill hole density maps) in the prospectivity modelling process would also help to identify areas that either lack detailed exploration or have been extensively explored. The use of more localised mining costs models, incorporating real Australian mine development and production costs is also recommended. Probably the most important recommendation deals with improvements to the exploration model such as the use of more accurate exploration costs and staged exploration processes. It is also suggested that other economic risk analysis techniques (i.e. portfolio analyses, capital asset fixing etc.) could be used to model the exploration process.
| Identifer | oai:union.ndltd.org:ADTP/252431 |
| Creators | Hedger, Darren Charles |
| Source Sets | Australiasian Digital Theses Program |
| Detected Language | English |
Page generated in 0.0022 seconds