Segmentation of mine overburden dump particles from images using Mask R CNN.

The stability of mine overburden dumps is crucial for the efficient operation of mining industries. The size distribution of particles affects the shear strength of dump slopes. Identification of dump particles from images is challenging as they vary in size, shape, color, granularity, and texture. In this paper, a unique way of identifying the particles from dump images using Artificial Intelligence is presented that can be used to determine the particle size distribution of dump. Mask R CNN with ResNet50 plus an FPN as a backbone network which is the current state of the art for instance segmentation has been implemented to segment the particles from dump images at detailed pixel level and to obtain their boundary. Experimental results showed promising results to delineate the particles and obtain masks over them. Our model has achieved a training accuracy of 97.2% for the dataset containing 31,505 particles. The model predicted the areas of dump particles with a mean percentage error of 0.39% and a standard deviation of 0.25 when compared to the ground truth values. The calculation of coordinates of the detected boundaries using the model significantly reduces the time and effort that are generally put in rock mechanics laboratories.

Development of one dimensional geomechanical model for a tight gas reservoir.

Estimating rock-mechanical, petrophysical properties and pre-production stress state is essential for effective reservoir planning, development, and optimal exploitation. This paper attempts to construct a comprehensive one-dimensional mechanical earth model (1D MEM) of the Mandapeta gas reservoir of Krishna Godavari (KG) basin, India. The methodology comprises a detailed stepwise process from processing and analysis of raw log data, calibration of log-derived dynamic properties with static ones using regression models developed from tested core samples, and final rock mechanical property estimation. Pore pressure profiles have been estimated and calibrated with the Repeat formation tester (RFT) data for every thirty-five wells. Overburden and horizontal stresses have also been evaluated and calibrated using data from the Leak-off Tests (LOT) or Extended Leak-off Tests (XLOT). A menu-driven program is developed using PYTHON code for visualization and on-time revision of 1D MEM. The resulting comprehensive 1D MEM predicts and establishes the rock-mechanical properties, pore pressure, and in-situ stress values of the basin. Besides its use in planning future wells, development of the field, and yielding insight into the various well challenges, it can also be used to develop a 3D MEM of the reservoir.

Optical strain based pre-failure indication in failure process of geomaterials.

Multilevel extended digital image correlation (X-DIC) technique based on finite element method (FEM) is applied for measuring deformation of geomaterials under uni-axial loading condition. The concept of Smooth Particle Hydrodynamics (SPH) is introduced for smoothing computed displacements as well as for calculating strain tensors at every nodal point of FEM mesh. Cumulative effective strain estimated from strain tensors is found to be a well suited parameter to identify the change in stress-strain behaviour in experimented samples. Further analysis suggests that onset of microcrack development and yielding in samples can also be identified using this parameter. Based on these findings, an indicator is developed for determining onset of both microcrack development and yielding in geomaterials. This indicator is found to be related to volumetric strains and may also signify dilation of the sample. The potential of the developed indicator is tested by conducting four experimental works with concrete and rock samples.

Optical strain for monitoring of concrete failure mechanism with discontinuity.

Finite-element-based digital image correlation (FEM-DIC) is one of the most widely used noncontact techniques in the field of experimental mechanics for measurement of deformation/strain. In this paper, the FEM-DIC method is refined by introducing the concept of multilevel extended digital image correlation (X-DIC), which also can capture deformation across discontinuity planes if they exist in images. Using regular and enhanced displacements at each node, strain tensors are estimated by applying the concept of smooth particle hydrodynamics (SPH). Numerical works are carried out to check the accuracy level of the developed algorithm by considering discrete discontinuity on the surface of a sample. Work is further extended to determine displacements and strains developed at the surface of several cubical concrete samples under uniaxial loading conditions. The tests are conducted until fractures are developed in the post-failure region. Using the concept of cumulative effective strain, a parameter is identified, which can be used as a precursor in the object failure process.