Contribution to the update of a mining waste model using blast data (ULiège)

Abstract

This thesis provides a thorough investigation focused on improving the comprehension of spatial organization and chemical composition fluctuations within a dolomite quarry deposit. The study combines sophisticated geostatistical techniques, machine learning algorithms, and geological information to categorize the deposit into separate geometallurgical zones. The clustering techniques of K-means and Gaussian Mixture Model (GMM) were utilized to evaluate the concentrations of Alumina (Al₂O₃), Silica (SiO₂), and Particle Size Distribution (PSD). This analysis provided valuable information on the geological processes and material compositions present in the quarry. This study utilizes the chemical concentrations of alumina (Al₂O₃) and silica (SiO₂) as a proxy to detect and delineate clay zones within a dolomite quarry. The study intends to identify specific locations where clay minerals are likely to be found by utilizing modern geostatistical methods and clustering algorithms to analyse the geographical distribution of these components. This method enables a more precise detection of areas with high clay content, which can have a substantial effect on the quality of extracted dolomite and the overall effectiveness of quarry activities. The analysis effectively distinguished three separate clusters of GMM, where Cluster 1 exhibited a robust link with dolomite-rich Unit D and notable variations in chemical composition, whereas Cluster 2 showed regions undergoing a transition from dolomite to limestone. The results emphasize the spatial arrangement of contaminants in the quarry, namely areas with higher levels of alumina and silica that may necessitate focused extraction or treatment. The findings were further confirmed by utilizing the Leapfrog model, which presented a realistic 3D representation of clay infiltration zones that corresponded with the observed fracture patterns in the field cross validated by drone imagery. The findings of this study have important ramifications for improving dolomite extraction methods, maximizing resource utilization, and advocating for sustainable mining practices. This study also emphasizes the possibility of combining geospatial data with sophisticated clustering methods to enhance the precision and effectiveness of geological models, hence promoting the sustainability and profitability of dolomite quarry operations.