<p>Precisely controlling mineral grain size and observing microcracks in rocks through laboratory techniques is challenging. A grain-based model (GBM) incorporating moment tensor was developed to simulate the shear failure of crystalline rocks and investigate the effects of grain size heterogeneity for the first time. GBM reveals the initiation and propagation of intergranular and intragranular cracks during shear failure from a microscopic perspective. Moment tensor analysis enables the quantitative evaluation of seismic source mechanisms. A heterogeneity index was introduced to measure the variability caused by the distribution of grain sizes in numerical samples. The shear behavior of the numerical model under varying heterogeneity indices and normal stress conditions was systematically discussed and analyzed. With an increasing heterogeneity index, microcrack distribution becomes increasingly focused around the shear failure area. The rise in normal stress inhibits the development of secondary large-scale fractures while encouraging the expansion of microscopic intragranular cracks. Between normal stresses of 5 and 30 MPa, shear strength tends to decline with rising heterogeneity index, exhibiting minor variations. With higher normal stress, the proportion of high-amplitude acoustic emission events is greater. In shear failure, the number of shear sources exceeds that of tensile and implosive sources. Increasing heterogeneity promotes implosive events while suppressing tensile events. Overall, this study advances the fundamental understanding of how grain heterogeneity governs shear failure processes, while also offering implications for predicting rock mass behavior and improving engineering design strategies.</p>