Modeling Damage Evolution in Heterogeneous Granite Using Digital Image-Based Grain-Based Model

A particle-based discrete element model combined with digital image processing to take into account the grain geometry and size distribution is used to investigate the mechanical and microcracking behavior of Beishan granite (BG), a typical crystalline rock potential for nuclear waste disposal in China. First, the procedure to construct a digital image-based grain-based model is introduced. The unsupervised K-means clustering algorithm is used to determine mineral boundaries, and four micro-parameters are assigned according to the Weibull distribution to reflect micro-heterogeneity of rock. Then, the model is calibrated to the macro-properties of BG obtained from laboratory tests, and consistency between model simulation and laboratory test results of both microcracking processes and failure modes is found. It is shown that intragrain cracks are initiated successively in quartz, feldspar, and mica; and their numbers in quartz and feldspar are much more than that in mica. Crack cluster, forking cracks and cataclastic microcracking are typical in quartz, while the “voids” is found in mica. Grain property heterogeneity has a significant impact on micro-cracking behavior and macro-properties of the model, with greater heterogeneity resulting in a lower uniaxial compression strength and elastic modulus, and a more scattered distribution of microcrack inclination. The modeling procedures presented here provide an effective method for investigating mechanical and damage behavior of granites, with specific consideration of heterogeneity involving the distribution of mineral grains and their mechanical properties.