The micro-geometric heterogeneity has a significant effect on the mechanical behavior and failure mode of brittle rocks. To study the influence of the size distribution and preferred orientation of mineral grains, a statistical grain-based discrete element method coupling with the cohesive zone model is proposed on the basis of the universal distinct element code (UDEC) in this study. First, a mixed-mode cohesive model was developed and implemented in UDEC to simulate the nonlinear behavior of grain interfaces. Then, a micro-parameter calibration procedure based on the Plackett-Burman design, central composite design, and particle swarm optimization was established to reproduce the macro-properties of the Lac du Bonnet granite. Finally, the strength characteristics, deformation response, and failure mode of rocks were numerically examined using the proposed model by varying the size distribution and preferred orientation of mineral grains. The relationship between the micro-geometric heterogeneity and the uniaxial compressive strength, the crack initiation stress, the crack damage stress, the elastic modulus, the Poisson's ratio, and the failure mode of rocks are obtained respectively.

微观几何非均质性对脆性岩石的力学行为和破坏模式有重要影响。为了研究矿物晶粒尺寸分布和择优取向的影响，本研究基于通用离散元编码（UDEC），提出了一种基于统计晶粒的离散元方法与内聚力区模型相结合的方法。首先，开发并在UDEC中实现了混合模式内聚模型，以模拟晶粒界面的非线性行为。然后，基于Plackett-Burman设计，中央复合设计和粒子群优化进行微参数校准建立是为了再现Lac du Bonnet花岗岩的宏观特性。最后，通过改变矿物颗粒的大小分布和择优取向，使用提出的模型对岩石的强度特性，变形响应和破坏模式进行了数值检验。分别得到岩石的微观几何非均质性与单轴抗压强度，裂纹萌生应力，裂纹破坏应力，弹性模量，泊松比和破坏模式之间的关系。