Isolating the Factors That Govern Fracture Development in Rocks Throughout Dynamic In Situ X‐Ray Tomography Experiments

• Published in: Geophysical research letters 2019-10, Vol.46 (20), p.11127-11135
• Main Authors: McBeck, Jessica, Kandula, Neelima, Aiken, John M., Cordonnier, Benoît, Renard, François
• Format: Article
• Language: English
• Subjects: Artificial intelligence; Clustering; Compression; Crack propagation; Crystalline rocks; Distance; Experiments; Failure analysis; fracture growth; Fracture mechanics; Growth; Learning algorithms; logistic regression; Machine learning; Microtomography; Orientation; Regression analysis; Rock; Rocks; Stress distribution; Thickness; Tomography; triaxial compression; X‐ray tomography
• ISSN: 0094-8276; 1944-8007
• DOI: 10.1029/2019GL084613

Centuries of work have highlighted the importance of several characteristics on fracture propagation. However, the relative importance of each characteristic on the likelihood of propagation remains elusive. We rank this importance by performing dynamic X‐ray microtomography experiments that provide unique access to characteristics of evolving fracture networks as rocks are triaxially compressed toward failure. We employed a machine learning technique based on logistic regression analysis to predict whether or not a fracture grows from 14 fracture geometry and network characteristics identified throughout four experiments on crystalline rocks in which thousands of fractures propagated. The characteristics that best predict fracture growth are the length, thickness, volume, and orientation of fractures with respect to the external stress field and the distance to the closest neighboring fracture. Growing fractures tend to be more clustered, shorter, thinner, volumetrically smaller, and dipping closer to 30–60° from the maximum compression direction than closing fractures. Plain Language Summary What controls fracture growth in rocks? Previous work highlights the importance of many characteristics on the likelihood of fracture growth but has not ranked the importance of these characteristics. We use triaxial compression experiments of rocks and machine learning to predict fracture growth using measures of the fracture network clustering and fracture size, shape, and orientation. The characteristics that are the best predictors of fracture growth are the fracture length, thickness, volume, and orientation and distance to the nearest fracture. Fractures that grow (increase in volume) during each increase in differential stress are shorter, thinner, smaller in volume, and more clustered than fractures that close (decrease in volume). Key Points We predict fracture growth in four X‐ray tomography experiments using the machine learning method of logistic regression The best predictors of growth are the fracture size, length, aperture, and orientation and fracture network clustering Growing fractures are smaller in volume, shorter, thinner, more obliquely dipping from σ1, and more clustered than closing fractures