Water-inrush Assessment Using a GIS-based Bayesian Network for the 12-2 Coal Seam of the Kailuan Donghuantuo Coal Mine in China

摘要

The Donghuantuo coal mine is geologically unusual, with 60 normal faults, 18 reverse faults, and 1 syncline. The coal seam floor is highly fractured and the fractures act as conduits for groundwater, which flows from the Ordina limestone aquifer into the no. 12 coal seam. From 2005 to 2010, there were 7 water-inrush disasters through the floor of this coal seam. The largest water-inrush event exceeded 63 m~3/min; there are five points where the water-inrush continues to exceed 1.0 m~3/min. Comprehensive modeling of the probability of water-inrush through the floor is required to reduce the likelihood and severity of such events. The water-inrush situation was assessed using a GIS-based Bayesian network (BN). In the developed BN-GIS model, the geometry of the coal mine working face was incorporated in suitable detail and resolution. The results of the modeling compared well with field water-inrush observations. Based on documented water-inrush events, the accuracy of the fit of the model data is 83.4 %, and the probability of making an incorrect prediction is less than 0.5, which means that using this method could significantly enhance coal production at the mine.