Mechanical responses of grain-based models considering different crystallographic spatial distributions to simulate heterogeneous rocks under loading

Abstract

Grain-based models (GBM) are widely used to reproduce the heterogeneous characteristics of natural rocks at the grain size level. To incorporate the spatial distribution of the constituent minerals into the GBM, the digital image processing (DIP) technique and the random distribution of property (RDP) method are frequently adopted. However, an analysis of the differences, advantages and disadvantages of the named two popular methods in simulating the mechanical response of rocks under loading has so far not been undertaken. This study considers Brazilian and uniaxial compression tests using DIP and RDP models, showing distinct differences in stress-deformation behavior (especially tensile strength) and final failure patterns. Fracture initiation , accumulation, propagation, and the corresponding stress-deformation behavior of polycrystalline rocks are sensitive to the crystallographic spatial distribution and the micro-strength parameters of mineral contacts. The clustering of mineral grains tends to induce more significant local stresses while the stress fluctuations in RDP models are more widely distributed and isolated, leading to different microcracking behavior furtherly affected by the superimposed effect of heterogeneous contact properties. Therefore, it is essential to use DIP methods to model heterogeneous rocks if the grain-scale mechanical behavior is of interest. If a heterogeneous model is used to study the overall macro-mechanical behavior of rocks, using the RDP method may be appropriate with the advantage of reduced effort for model set-up.