ASSESSMENT OF THE INFLUENCE OF MECHANICAL PROPERTIES AND MINERALOGICAL COMPOSITION ON FRAGMENTATION OF SELECTED GRANITIC ROCKS

In this study, a Proportional-Integral-Derivative (PID) control system was designed and implemented on a DC This study analysed the effect of geo-mechanical properties, modal composition of minerals, and powder factor on fragment size distribution (X80) in granitic rocks to optimise blasting operations. Using a data set that included UCS, Young's modulus (E), Poisson’s ratio (V), Quartz (Qz), K-Feldspar (K-F), Plagioclase (Pc), Mica (M), and powder factor (PF), a Random Forest model was applied to model non-linearities and predict X80. The sensitivity analysis and regression techniques were applied to evaluate the significance of each parameter. Results indicate that primary predictors were PF and Qz, with PF having a significantly negative association with X80. Mica and plagioclase were more dependent on fragmentation, with sensitivity scores of 0.2808 and 0.1797, respectively, for mineralogical characteristics. The Random Forest model had R2 and RMSE values of 0.8753 and 3.46, respectively. However, E and V geo-mechanical properties exerted minimal effects on X80. Despite the model being relatively strong, the small sample size and multicollinearity among features in the study posed challenges. This study highlights the importance of both operational and compositional factors while predicting and optimising the rock fragmentation process. Future studies should expand the dataset, explore advanced machine learning techniques, and validate findings across diverse geological settings to enhance reliability and applicability.