A Coupled Hydromechanical Model for Bed-Separation Water Inrush at Longfeng Coal Mine in China

Abstract

Bed-separation water inrush refers to the sudden water inflow into mine workings from mining-induced roof bed-separation cavities when a hydraulic connection is created, and it is a common and hazardous water disaster in deep coal mines. Previous studies predicting water inrush were based mainly on pure mechanical or seepage models, and ignored the hydromechanical coupling processes. However, pore-pressure changes redistribute effective stress and reshape flow paths and the flow field. Therefore, these models fail to dynamically predict water inrush. In this study, a coupled hydromechanical model incorporating the Mohr–Coulomb model, Biot’s equation, and seepage equations was applied to simulate stress redistribution and water inflow for a karst aquifer in Longfeng Coal Mine in Guizhou Province, China. Compared with the classical mechanical model, the coupled hydromechanical model captures stronger tensile stress variation and sharper vertical displacement discontinuities. The tensile-unloading transition occurs earlier, and the separation belt becomes more continuous in the coupled hydromechanical model. When the working face advanced to 100 m, the stress curve from the coupled hydromechanical model reached about 14.3–14.8 MPa at 90–110 m, whereas it was lower and the peak value occurred later using the classical mechanical model. The predictive performance of the coupled model was validated quantitatively against field observations. The RMS error (RMSE) was 45.94    m 3 / h , the mean absolute percentage error (MAPE) was 12.35%, the coefficient of determination ( R 2 ) was 0.916, and the index of agreement ( d ) was 0.979. When the working face advanced to 500 m, the predicted cumulative inflow reached 2,300    m 3 . This case study demonstrates the potential engineering usefulness of the coupled hydromechanical model and provides a scientific basis for risk evaluation and hazard prevention in complex geological conditions.