Presenting a new holistic robust approach for predicting geo-failures in complex underground engineering projects

The exploration of geo-resources in complex geological provinces poses significant challenges, often resulting in severe geological disasters with economic losses and fatalities. Fractured rocks, pre-existing fractures, and excavation-induced fractures contribute to the complexity of these disasters. Despite numerous studies, understanding the coupling mechanism of induced seismicity and geological deformations in such complex mining environments remains unclear. This study introduces the "Acousto-Frac Model," a novel, cost-effective, and robust geophysical approach designed to comprehensively understand experimental microseismicity and reveal such a mechanism. Traditionally, physical models and numerical simulations have been employed for dynamic disaster prediction in underground coalmines. However, these methods are often neither cost-effective nor robust. The Acousto-Frac Model offers an innovative methodology for mapping induced fracture networks through Acoustic Emission (AE) experiments conducted on coal and rock samples. This approach tracks each AE event, constructs networks of induced fractures, identifies geological lineaments, and predicts zones prone to failure/disasters. To implement the model, AE and rock mechanics testing systems were utilized to conduct experiments on coal and rock samples under uniaxial loading. The proposed model successfully identified weak zones, predicting general deformation propagation directions. Moreover, the 3D crack growth theory and the criterion for microcrack density were employed to analyze the fracture transformation process, ranging from small-scale microfractures to large-scale microfractures and from local deformation to complete damage for the coal and rock samples subjected to uniaxial loading. The study further leverages Single Link Cluster (SLC) simulations and b-value theory to characterize the spatiotemporal response of microearthquakes, including b-value, spatial correlation length (), and information entropy (H). Notably, the results indicated that at the onset of initial loading (15%), the spatial correlation length () exhibited an upward trend, while the b-value remained comparatively stable. These parameters showed a significant change trend before the buckling failure of coal and rock samples, suggesting that, in combination with the proposed model, spatial correlation length (), b-value, and information entropy (H) provide a new and robust method for complete deformation evaluation and the prediction of geo-material failure. This innovative method, while a panacea for imaging the entire fracturing phenomenon, provides insights with widespread implications for academic researchers and industry practitioners. It serves as a valuable tool for predicting geological failures in global underground engineering excavations, offering a comprehensive and cost-effective solution for the mapping of induced seismicity and geological deformations in underground mines.