Mineral resource estimation using a combination of drilling and IP&Rs data using statistical and cokriging methods

Abstract

The aim of this research is the mineral resource estimation using a combination of drilling and IP-Rs data. Therefore the approach of this paper is to study the correlation of induced polarization (IP) and Electrical resistivity (Rs) data with drilling data in order to grade estimation and mineral resource estimation. Reducing the boreholes number and optimization of the boreholes location is another aim of this research. The Abassabad copper mine located in Miami-Sabzevar mineralization belt northeast Iran was chosen as a case study. Within the borehole locations, geophysical profiles were designed and surveyed. After IP-Rs data inversion, 2D sections were prepared. The 3D block models of IP-Rs were constructed by geostatistical methods. The correlation between IP-Rs and drilling data were examined by statistical and geostatistical methods using regression, multivariate regression analysis, and cokriging. Based on the mentioned methods copper grade was estimated and the 3D block models of Cu grade were constructed. Obtained models were checked and compared with real Cu model compiled according to drilling data which was done after geophysical measurements. Results showed that the regression between IP data and Cu grade was more appropriate with least error. Rs data are not suitable for Cu estimation, due to changing intervals which led to increasing estimation error. Based on the suggestions of this paper, we could reduce the number of boreholes to 30% of the initial number and optimize the boreholes locations.

WITHDRAWN: A new gold grade estimation approach by using support vector machine (SVM) and back propagation neural network (BPNN)- A Case study: Dalli deposit, Iran

“…Over the last decade, the artificial neural networks (ANNs) method has been extensively used in different areas of mining engineering (Doulati Ardejani et al, 2013;Majdi andRezaei, 2013, Jodeiri Shokri et al, 2014;Chao et al, 2018;Mostafaei and Ramazi, 2019b;Shakeri et al, 2020;Lawal and Kwon, 2021;Ramu et al, 2021;Sohrabi et al, 2021). The quality of the suggested models in this technique depends on the relationship between the patterns used in training procedures and the expected patterns (prediction) with the lowest error.…”