Estimating the Reserve of Lake- Siah Iron Ore Body by Geostatistical and Artificial Network Methods

Investments and progress of mining projects depend on the quantity and quality of mineral resources and reserves; therefore, it is significant to know the reliability of the ore deposit estimation based on various methods. This research has investigated the role of geostatistics and artificial neural networks in modeling, estimating the concentration of exploration blocks, and estimating mining reserves. Spatial distribution modeling of iron values has been performed using three methods of ordinary kriging, Gaussian simulation, and artificial neural networks. To construct the block model of ore deposit, 29 exploratory boreholes with an average depth of 142.75 meters and a total length of 4139.9 meters were used. After drawing the variograms, the search ellipse was calculated, and a 3D model was obtained to estimate the concentration by ordinary kriging and Gaussian sequential simulation methods. Also, modeling and estimating the concentration was done by the artificial neural network method. Results showed that the artificial neural network technique has high validity. Also, due to its ease of use and no need for extra variographic calculations can be an appropriate alternative to geostatistical and simulation approaches. Finally, based on different cut-offs, the concentration-volume curve was drawn. The results show that this ore body has 439 million tons of ore with an average concentration of 42 % per 20% cut-off concentration.