APPLICATION OF FRACTAL ANALYSIS TO TRADITIONAL TOP-CUT IN RESOURCE ESTIMATION OF GOLD DEPOSITS

Abstract

The accuracy of correct resource estimation is crucial in integrity of a feasibility study as well as the daily running of mining activities. The success of a mine operation depends on the correctness of resource estimation. In order to certify an acceptable level of confidence in resource modeling, it is necessary to construct a high-quality 3D block model, where in each block; the grade of interest is reported. However, each deposit has an exclusive structure and subsurface geology, thus, it is extremely complex to predict and produce such an exact model. In addition, the presence of extremely high values in the dataset negatively affects resource estimation and produces biased estimates. This problem is especially noticeable in gold deposits with a long-tailed distribution. There is a method in the industry to solve this problem so-called capping, where sample grades that exceed the provided specific threshold (cutting value) will be downgraded to a specific threshold. To reduce the impact of outliers to final resource estimation, a new Top-Cut technique was proposed. In this method, the grade variable is spitted into three parts: the truncated grade, an indicator and residual. The pros and cons of this method are that we can choose the value of the threshold manually. In practice, this value is selected by using various statistical techniques and sometimes it is quite objective. On one hand, this increases the variability of the method, but from the other hand, we do not have a specific rule for choosing the best and optimum threshold. Therefore, the use of fractal analysis can be useful, since the fractal model allows determining several populations in a dataset that can be used as individual thresholds. Moreover, to determine the best threshold, the ratio of indicator variogram analysis is used. This work utilizes a borehole dataset from a Kazakhstani gold deposit, where Traditional Top-Cut and Fractal Top-Cut modeling approaches are applied to model the Gold in this deposit. In order of comparison, conventional capping method is also applied. The results showed that Fractal Top-Cut outperforms the Traditional Top-Cut and capping technique in better resource modeling and produces more accurate results with better average grade above different cut-off grades and less mean error. However, the drawback is that the method does not produce a variance map.