Optimizing fracture toughness estimation for rock structures: A soft computing approach with GWO and IWO algorithms

The importance of mode I fracture toughness (KIc) K Ic ) lies in its role in studying rock fractures, finding application across diverse sectors like mining, tunneling, and surface construction. Given the time-consuming, costly, and arduous nature of determining fracture toughness through lab and field tests, employing indirect techniques is advised. Soft computing's indirect methods excel at modeling intricate nonlinear connections with optimal efficiency and accuracy, even when geological parameters introduce uncertainty. The innovative aspect of this research lies in leveraging Grey Wolf Optimization (GWO) and Invasive Weed Optimization (IWO) algorithms to model 88 Chevron Notched Brazilian Disc (CCNBD) specimens, designated by ISRM, considering uncertainties. The input parameters encompass uniaxial tensile strength (6t), t ), initial crack length (alpha 0), 0 ), disc specimen radius (R), R ), crack length (alpha B), B ), disc thickness (B), B ), and final crack length (alpha 1). 1 ). The research proceeds by employing 80% of the dataset (70 data points) for constructing models via the algorithms, reserving the remaining 20% (18 data points) for model validation. The outcomes, evaluated using statistical metrics such as mean square error (MSE), squared correlation coefficient (R2), mean absolute error (MAE), and root mean square error (RMSE) underscore the exceptional accuracy and performance of GWO and IWO algorithms in predicting and estimating rock structure's KIc. Ic . Moreover, this paper conducts a sensitivity analysis on K I c estimation's input parameters using @RISK software. Findings reveal that among the input parameters, 6t t exerts the most substantial influence on the model's output. Consequently, even a minor alteration in 6t t yields significant fluctuations in the relationship's output.