Reconstruction of laminated shale specimens through advanced physics-informed image processing

Shale is distinguished from other rocks by its laminated and fissile nature, as it consists of many thin layers and easily splits into thin pieces along the bedding planes. A common approach in a geomechanical simulation of shale is to simplify its complex structure by representing bedding planes as continuous, straight, and equidistant. This approach is acceptable for approximating the general behavior of shale and it is numerically efficient, but also limits our understanding of the true mechanical response of shale to mining-induced stress changes. This paper discussed the limitations of these assumptions and reconstructed laminated shale specimens through image processing to capture bedding planes and through numerical simulation to calibrate their micro-properties. Specifically, this paper used an advanced physics-informed image processing method to extract the coordinates, number, spacing, length, and frequency of the bedding planes, forming a comprehensive geometry database. Using this database, a laminated shale specimen modeling (S1) was developed using UDEC software. A systematic modeling calibration procedure is proposed and applied to calibrate the model by obtaining a final set of calibrated micro-properties, which are verified through comparative analyses of shale specimens with and without flaws, as well as specimens with extracted and parallel bedding planes, along with a reasonableness analysis of predicted bedding planes using a bedding plane geometry database. Furthermore, two more shale specimens (S2, S3) were reconstructed, which incorporate extracted bedding planes and calibrated micro-properties. The results show that the three reconstructed shale specimens achieve stress-strain behaviors and crack propagation that match those observed in the laboratory. This paper bridges the gap between the shale complex structure and geomechanical simulations by introducing an advanced physics-informed image processing approach for an accurate representation of bedding planes and micro-properties. • A systematic modeling procedure in numerical simulation is proposed to calibrate micro-properties. • Laminated shale specimens are reconstructed through image-processing. • This paper bridges the gap between the shale complex structure and geomechanical simulations.