Evaluation and Development of Digital Image Correlation Approach to Predict Coal Pillar Failure

Abstract

Underground Coal mining is showing an increasing trend to address the technical, societal and environmental challenges faced by surface mining operations. Conventionally coal structure behavior is predicted from laboratory investigation using contact­based approaches. Scientific pillar design has potential to reduce the blocked pillar dimensions. There have been many attempts to study the pillar behavior under loading. Though there exists many investigations to correlate stress and strain under loading, yet pillar failures continue to occur. It is observed that pre­existing cracks in the coal specimen dilate substantially before failure occurs under loading. Digital image correlation (DIC), a non­contact based approach is being used successfully to predict material weakness in many industrial applications. The technique is being used to predict crack propagation in geomaterials with limited success. The undertaken investigation is an attempt to establish correlation of crack development and propagation with the weakening of coal pillar through DIC approach. Conventional measurement systems as strain gauges and Linear variable Differential Transducer (LVDT) and DIC based analyses were evaluated for heterogeneous specimen as coal. The experiments were conducted for more than 270 carefully prepared coal specimen sourced from varying depths of 30 m, 75 m and 100 m. DIC based analysis involved more than 20,000 images taken through an image acquisition system. A detailed characterization i.e. physico­mechanical and microstructural analysis of those coal specimen were carried out at beginning. Development of experimental setup for DIC is a challenge. DIC approach involves taking images of the surface under loading and comparing those at different stages with that of the un­deformed surfaces. The investigation involved more than Speckle pattern is an important step prior to performing DIC analysis with improved accuracy. In this work, two types of speckle patterns were evaluated. A novel method of applying speckle pattern used in this research work was by generating it numerically and applying onto the specimen by silk printing method. The DIC processing parameters as subset size and step sizes were chosen after careful consideration. A comparative study between two types of speckle patterns were evaluated using FEM analysis and a subset size of 99 x 99 and step size of 8 pixels were determined. The DIC approach was found to be very sensitive to small strains than that by conventional approaches irrespective of depths of occurrence. The knowledge of crack threshold stresses are required to realize the failure process of any material. In this investigation, the failure process of coal has been analysed by crack initiation stress, crack damage stress, and peak stress. The crack initiation and

crack damage stresses were determined using three popular methods using volumetric strain, crack volumetric strain and lateral strain. The rate of change in the effective stored strain energy was used to analyze and develop a precursor failure indicator for coal specimens under uniaxial loading. An algorithm has also been developed to find the precursor failure indicator for coal specimens. A comparative study between codes using DaVis and open source code NCorr has been carried out. NCorr was observed to be comparable favourably with that by DaVis with an error of about 4.25 %. The applicability of DIC under cyclic loading in coal specimens exhibited threshold values for fatigue failure strength at 77.10 %,
79.95 % and 66.79 % of uniaxial strength of respective coal specimens at 30 m, 75 m and 100 m depths respectively. DIC analysis was more reliable and exhibited sensitive behavior of coal under axial loading. DIC is a better method of deformation measurement approach.