Geotechnics

Factor of Safety (FOS) is a significant index to measure the stability condition of a rock slope in mining or civil engineering. In this paper, we evaluate and compare four different machine learning models, Gaussian Process Regressor (GPR), Support Vector Regressor (SVR), Random Forest (RF), and a hybrid genetic algorithm–multi-layer perceptron (GA-MLP), using two separate real-world datasets. The two separate datasets used in this study are from a previously conducted study on highway excavation with rock cutting in China, and another one in a mining site in Peru, with five geotechnical properties used as inputs, including slope height, slope angle, unit weight, cohesion, and friction angle. The two separate datasets were separated into training, validation, and testing datasets. The testing dataset of the models is unseen data used to assess model performance in an unbiased manner. The result shows that the SVR had the highest prediction accuracy, followed by GPR for the mining dataset, and GPR had the highest performance among all the models for the highway excavation dataset. From the boxplot, we can see that SVR, while having the highest predictive accuracy, has a larger variance in prediction compared to GPR for the mining dataset.

Underground coal gasification (UCG) is an emerging energy technology that involves the in situ conversion of coal into syngas through controlled combustion within a subsurface excavation. The geomechanical processes associated with UCG can lead to significant overburden caving and surface subsidence, posing risks to surface infrastructure and groundwater systems. To accurately predict the size of overburden caving zones and associated surface subsidence, a prediction model was developed based on simulation results using discrete element method (DEM) numerical models. The main purpose of developing such a model is to establish a systematic and computationally efficient method for the rapid prediction of the height of overburden caving and its associated surface subsidence induced by underground excavation. The model is broadly applicable to different types of underground excavations, and UCG is used in this study as a representative application scenario to demonstrate the relevance and performance of the model. Sensitivity analysis indicates that excavation span, tensile strength, and burial depth are the primary controls on the height of the caving zone within the ranges of parameters investigated. Rock density is retained as a secondary background parameter to represent gravitational loading and its contribution to the in situ stress level. The derived model was validated using published numerical, experimental, and field measurement data, showing good agreement within practical ranges. To further demonstrate the application of the model developed, the predicted caving geometries were incorporated into finite element method (FEM) models to simulate surface subsidence under different geological conditions. The results highlight that the arch structure formed by overburden caving can help redistribute stresses and thereby reduce surface deformation. The proposed model provides a practical, parameter-driven tool to assist in underground excavation design, environmental risk evaluation, and ground stability management.

To accurately analyze the time-dependent stability of large-span tunnels traversing the F2 fault fracture zone, this study focused on the deep-buried red-bed mudstone of the Jishan Tunnel. Rock cores were retrieved from the critical Grade V surrounding rock section (depth 370 m). Uniaxial and triaxial compression tests were conducted to determine basic mechanical parameters. Through step-loading creep tests, the creep characteristics were analyzed, and a long-term strength of 19.2 MPa was identified. Analysis revealed that the deformation aligns well with the stress-dependent Burgers model, where parameters evolve with stress level. Using the Levenberg–Marquardt algorithm, the variable model parameters were derived. Finally, three-dimensional creep parameters were obtained for numerical validation. Engineering recommendations for support timing and yielding mechanisms are proposed to mitigate rheological risks in fault-affected zones.