Abstract:
The brittleness of rocks has traditionally been measured utilizing mechanical properties, however, it has been established that the mechanical properties of rock masses strongly depend on their petrographic properties, and mineralogical composition plays a critical role in determining the property of rocks to fail. Despite the existence of established brittleness indices based on the mineralogical composition of rocks, the selection of brittle components in rocks remains subjective and can be contradictory. The effects of porosity, stress state, cement type and strength, pre-consolidation factor, grain size, and external mechanical conditions are also not considered in the existing brittleness indices. This study aims to characterize the brittleness indices of platinum-bearing (UG2) reefs on the eastern limb of the Bushveld Igneous Complex using mineralogical analysis. To this aim, microscopic examinations identified different mineralogical characteristics, laboratory analysis of UG2 samples was performed to obtain the mechanical parameters, and additional parameters were obtained from numerical simulations. Innovative brittleness indexes (B1, B2, B3, B4, and B5) are developed utilizing parameters such as textural variables, contact type, contact nature, and packing density. The brittleness indexes are additionally compared with previous important brittleness indices utilizing fieldwork, laboratory analysis, numerical simulation, and predictions (multivariate linear models). Additionally, a similar approach was applied to the mechanical parameters, developing innovative brittleness indexes B1, B2, B4 and B6 using traditional mechanical parameters. The results revealed that contact type and contact nature were the most significant mineralogical predictors of brittleness, with models B2 and B6 demonstrating strong predictive accuracy (R² values of 0.778 and 0.727, respectively) and near-perfect precision (SE = 0.000). Texture and packing variables had a lesser impact on brittleness prediction. While tensile (σt) and compressive strength (σc) emerged as key predictors with high accuracy. Models B2 and B6 were the most reliable, showing near-perfect precision with SE values of 0.000. In contrast, models B1 and B4 were less precise, and B3 lacked statistical significance. The high statistical significance and low SE values of B2 and B6 confirmed their reliability for brittleness prediction. It is recommended that more sophisticated methods be applied as well.
