Sergei Sabanov
Associate Professor
Nazarbayev University
Dr. Sergei Sabanov started his carrier in 1992 working in the underground oil shale mines in Estonia before he received his PhD grade in mining engineering in 2008. Dr. Sabanov was executing multiple projects and teaching at a time, tracking technical progress, and reporting on both to the university and to the mining companies, and regulatory and technical review boards. Dr. Sabanov industry experience with reserves estimates, mine ventilation, mine optimisation, mine design and production scheduling, blasting operations, material handling, mine services and maintenance, mine automation, mine closure, sustainable mining, and mining risk management enabled him to understand detailed requirements of mining companies in mining specialists. Dr. Sabanov is responsible for implementation of students’ theoretical knowledge to industrial practices, optimising study programmes to real life work of mining projects. For teaching, he uses a variety of innovative teaching methods, technologies stimulated engagement in class and demonstrated a comprehensive expertise. Dr. Sabanov also using an operational state-of-the-art research laboratory on ventilation and a vast number computer software.
AI + Geostatistics for the Analysis of Mine Tailings Storage Facilities
This paper presents a novel approach to modelling mine tailings storage facilities through the integration of geostatistical data expansion and artificial intelligence methods. The study is based on a polymetallic tailings facility in Eastern Kazakhstan, characterised by high spatial heterogeneity and the absence of a well-defined geological structure. To improve data density and continuity, the original dataset from 90 boreholes was augmented using ordinary kriging, resulting in the generation of a regular three-dimensional grid.
The results demonstrate that the use of an expanded dataset in combination with a Gaussian Mixture Model (GMM) significantly improves clustering performance compared to traditional approaches based solely on borehole data. In this case, a substantial increase in spatial coherence of clusters was achieved, with Moran’s I rising from 0.29 to 0.52, alongside improved cluster compactness, reflected by an increase in the Silhouette Index from 0.41 to 0.59.
As a result, three stable geochemical clusters were identified: a zone enriched in base metals (Cu, Pb, Zn), a zone with elevated concentrations of precious metals (Au, Ag), and a zone characterised by low to intermediate grades. The spatial distribution of these clusters shows high continuity and is suitable for block modelling applications.
The findings confirm that the proposed approach enables the delineation of geologically and operationally meaningful domains, which can be directly applied to selective tailings reprocessing strategies and to improving the efficiency of valuable component recovery.
