Predicting Fracture Network Development in Crystalline Rocks

Abstract The geometric properties of fractures influence whether they propagate, arrest, or coalesce with other fractures. Thus, quantifying the relationship between fracture network characteristics may help predict development, and perhaps precursors to catastrophic failure. To constrain predictability characteristics, we deform eight one centimeter tall rock cores under triaxial compression while acquiring in situ X-ray tomograms. tomograms reveal precise measurements above spatial resolution 6.5 µm. We develop machine learning models value each characteristic using excluding macroscopic stress strain imposed on rock. development more accurately experiments performed granite monzonite, than marble. Fracture be predictable these igneous rocks because their microstructure is mechanically homogeneous marble, producing systematic that not strongly impeded by grain contacts cleavage planes. varying performance suggest volume, length, aperture are most orientation least predictable. Orientation does correlate as suggested idea evolves increasing differential thus length. This difference observed expected length highlights mechanical heterogeneities local perturbations growth link, coalesce.