2025/03/07 by Claudia Elijas-Parra, Alexis Cartwright‐Taylor, Ian Main +3 · 1 voice
Engineering · Earth and Planetary Sciences · #Rock Mechanics and Modeling #Hydrocarbon exploration and reservoir analysis #Seismic Imaging and Inversion Techniques
paper · doi:10.1016/j.jsg.2025.105404
openalex publication_date 2025/03/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/05/25
High resolution, time-resolved X-ray imaging of deformation processes has revealed unprecedented detail on the microstructural and mechanical processes involved in system-scale failure during laboratory experiments. This has highlighted the role of the evolving crack population in controlling localisation. However, in porous rocks, accurate segmentation of evolving crack populations from the rock matrix and pre-existing pores is challenging because of the narrow aperture of cracks. Here we develop and test a new global statistical method of distinguishing between these three phases in tomographic slices, obtained from an X-ray transparent triaxial compression experiment on Clashach sandstone at 20 MPa effective pressure. Compared with a watershed-based method, the new method is less sensitive to artefacts of sampling at the voxel scale ( l ≥40 μm ). Furthermore, the segmented crack length distributions have a greater dynamic range, yielding exponents (1≤ a ≤3) above 150 μm that are in the same range as those observed in field outcrop, while anisotropy in the crack orientation distribution is better represented. Using this new method, we observe the spontaneous localisation of cracks into a shear band, whose orientations evolve from a random distribution to one more favourably oriented for synthetic shear and tensile fracturing. The observed shear band orientation (∼60° dip) is intermediate between a directed percolation model (50°) and the mode of the crack orientation distribution in the shear band (66°). • New method to separate microcracks from pores and rock matrix in x-ray images. • Computationally inexpensive and more precise than locally adaptive methods. • More realistic and better resolved distributions of crack orientations and lengths. • Shear band orientation may be partly, not wholly, controlled by directed percolation.