ABSTRACT The characteristics of geological faults in an opencast mine have always been a precursor for several operational and geotechnical challenges. To assess how the spatial distribution and intersection of faults affect the stability of slopes, the present study estimated the fault fractal dimensions (Db) and fault influence factor (FIF) of eight large opencast coal mines in India and correlated them with the challenges faced by mine authorities through observation‐based fault characterisation and mapping of bench scale instabilities. The results of Db and FIF showed that the Kistram, Indaram, SRP OCP, RGOC‐II, and JVR OCP mines were classified as relatively complex, and the Gevra, RGOC‐III, and PKOC mines exhibited moderate complexity. Furthermore, a logistic regression model was developed to estimate the likelihood of fault‐induced slope failure utilising the FIF and Db values. Due to limited failure datasets, leading to class imbalance, a Synthetic Minority Over‐Sampling Technique (SMOTE) was applied to generate a synthetic dataset for model development. Sensitivity analysis was performed across probability cutoffs from p > 0.2 to 0.8 to define the threshold values of FIF and Db for failure. The initial field observations indicated that slope failures occur when FIF ≥ 0.5 and Db ≥ 1.42. Using the logistic regression model, these thresholds were refined to FIF ≥ 0.44 and Db ≥ 1.42, the minimum values at which the probability of failure consistently exceeds 65%. These findings underscore the importance of FIF and Db in predicting slope instability and demonstrate the effectiveness of SMOTE in mitigating class imbalance.
Asif et al. (Tue,) studied this question.