Degasification system selection for US longwall mines using an expert classification system

Abstract

Methane emissions from the active face areas and from the fractured formations overlying the mined coalbed can affect safety and productivity in longwall mines. Since ventilation alone may not be sufficient to control the methane levels on a longwall operation, gob vent boreholes (GVB), horizontal and vertical drainage boreholes, and their combinations are drilled and used as supplementary methane control measures in many mines. However, in most cases, the types of degasification wellbores chosen are decided based on previous experiences without analyzing the different factors that may affect this decision. This study describes the development of an expert classification system used as a decision tool. It was built using a multilayer perceptron (MLP) type artificial neural network (ANN) structure. The ANN was trained using different geographical locations, longwall operation parameters, and coalbed characteristics as input and was tested to classify the output into four different selections, which are actual degasification designs that US longwall mines utilize. The ANN network selected no degasification, GVB, horizontal and GVB, and horizontal, vertical and GVB options with high accuracy. The results suggest that the model can be used as a decision tool for degasification system selection using site- and mine-specific conditions. Such a model can also be used as a screening tool to decide which degasification design should be investigated in detail with more complex numerical techniques.