Hidalgo López, Arturo; Herrera Herbert, Juan
Affiliated Research Center
Universidad Politécnica de Madrid
In underground mining, water inrush is a common hydrogeological hazard and is a deadly killer. Over the recent decades countless water inrush accidents have occurred in the main coal producing countries (China, India, Poland, Russia, etc.) and have killed thousands of miners. Among all the occurred water accidents fault-induced water inrushes account for a large proportion, this brings the urgency of researching the mechanisms and assessing the risk of fault-induced water inrush. This thesis concerns the fault-induced water inrush from three perspectives, the mechanism, the risk assessment and the post-disaster measure. To research the mechanisms of fault-induced water inrush, a computational model of a typical underground stope with a hidden fault was established for quantitatively assessing the magnitude of the stress concentration of the stress fields of the fault-tip. The numerical simulation was performed using the extended finite element method and fracture mechanics, and the simulation results suggested that the stress concentration at fault tip caused by fluid pressure, in-situ stresses and mining-induced stresses plays a key role in inducing fault reactivation and thus further inducing water inrush. To achieve the risk assessment of fault-induced water inrush, two methodologies were introduced into the research, the adaptive neuro-fuzzy inference system (ANFIS) and the rock engineering systems (RES). The former methodology was introduced for predicting the probability of water inrush caused by a specific fault and the later one was introduced for mapping the fault-induced water inrush risk for a whole coalfield. By means of these two methodologies, two quantitative risk assessment models (ANFIS model and RES model) were established and corresponding case studies were also elaborately implemented by using these two established models. The final assessment results showed that the ANFIS model is highly accurate in the prediction of water inrush cause by a specific fault and RES model can clearly get a water inrush safety map for a whole coalfield. For fault-induced water inrushes, the most important post-disaster measure is to quickly recognize the inrush sources, accurately identifying which aquifer or which water body is directly related to the inrush accident is the key step of controlling the accident and reducing casualties and economic losses. In this thesis, BP (back propagation) neural network was proposed to identify the water-inrush sources, according to the case studies conducted in Jiaozuo mine area, the results showed that the proposed method in this thesis is an effective and accurate method in recognizing the water-inrush sources.