颗粒物料在自然界和工业生产中广泛存在。当作用在颗粒上的外荷载超过某个临界值时，颗粒会发生破碎。颗粒破碎显然会影响整个床层的压降、热-力性能等特性。因此，准确地预测颗粒的破碎，对于控制和优化相关涉及颗粒物质操作具有重要的意义。实验中，颗粒的破碎强度通常是通过单轴压缩等实验得到。然而实际颗粒物料中某个颗粒可能会同时和多个颗粒发生接触作用，且基础构型随颗粒而变化。这就导致一个重要的问题：单轴压缩得到的颗粒强度是否适用于多点压缩状态，有没有一个唯一的普适性的颗粒破碎准则？针对这一问题，本论文通过开展三维离散单元法数值模拟颗粒破碎行为，展开了相关研究。论文的第二章对经典的Hertz接触作用模型进行了修正。基于采用的离散单元法数值模型模拟得到的试样宏观杨氏模量、拉伸强度和压缩强度同时能与实验结果吻合，说明本论文所采用的离散单元法数值模型的合理性和可靠性。论文第三章以方形试样为对象，通过开展拉伸和压缩模拟，讨论了离散单元法模拟中固体键设置对模拟结果的影响。模拟结果表明，对于给定的材料，只存在一种拉伸/剪切临界强度组合，使得模拟得到的试样拉伸强度和压缩强度同时能与实验结果吻合。拉伸下试样的破坏呈现一种“成核”效应，固体键的断裂主要源自于拉伸破坏；压缩加载下破坏模式则呈现“聚并”响应，固体键的破坏主要源自于剪切破坏。在所考察的参数范围内，压缩加载下试样的宏观断裂强度对固体键强度的分布形式和分布宽度都不敏感，而拉伸加载下固体键强度对试样强度的影响取决于其具体分布形式。论文第四章采用球压缩和壁面压缩两种方式，模拟研究了球形颗粒的破碎行为，在此基础上考察了文献中常见的几种颗粒破碎准则（最大接触力、最大主应力、平均主应力、最大拉应力、八面体剪切应力）的适用性。模拟结果表明，球压缩条件下，只有最大接触力不随加载点数目而变化；但是对于给定的加载点数目，不同加载构型得到的最大接触力波动较大。单轴、双轴和三轴壁面加载结果表明，在给定的加载构型下，颗粒破碎时对应的最大接触力波动较小，其数值与加载点数目相关。这些模拟结果说明，当采用这些准则判断颗粒破碎时，理论上必须要考虑加载点数目以及加载点空间排列型式的影响。论文第五章针对本论文研究工作进行了总结，并对下一步的研究进行了展望。;Granular materials are commonly encountered in nature and a variety of industrial processes. Granular particles will break when the external load acting on it is larger than its critical value. The fragmentation of particles will obviously influence the pressure drop across the whole assembly and also the thermal-mechanical properties of the system. Thus, accurately predicting the breakage of particles is obviously useful to the better controlment and optimization of any equipment related to the handling of granular material. Experimentally, the failure strength of granular particle is usually obtained through performing uniaxial compressive test. Nevertheless, inside real granular assembly, a given particle might interact with several neighbor particles at the same time, and the spatial arrangement of contacting neighbor particles varies from particle to particle. It thus raises an important question: can the results obtained from uniaxial compressive test be used to other loading condition, i.e., is there a unique particle breakage criterion that can be applied to any loading condition? Targeting at this question, in this work the breaking behavior of granular particle is investigated by performing three-dimensional Discrete Element Method (DEM) simulations.In chapter two, the DEM contact model used in this work is introduced. And the possible influences of loading rate and also the size of sample on the predicted variation of stress are discussed. Based on the simulation results, proper loading rate and sample size is then identified. In chapter three, the possible influences of bond property setting are investigated by modeling the mechanical responses of rectangular samples under tensile and compressive loadings. It is found that, for given brittle material, there only exist one tensile/shear bond strength setting that can ensure both the predicted tensile strength and compressive strength are in quantitatively agreement with experimental data. The simulation results indicate that the failure of samples under tensile loading is dictated by the nucleation of crack, whereas for compressive loading it is linked to the coalescence of cracks. By monitoring the time sequence of bond breakage and its failure mode, it is found that for tensile loading the dominate failure mode of bond is tensile fracture, whereas for compressive loading it is shear fracture. For the conditions considered in this work, the simulation results show that the influence of the distribution of bond strength on the compressive fracture strength of sample is minor, whereas for tensile fracture strength, it depends on the type of distribution.In chapter four, the crushing behavior of spherical aggregating under different loading configurations is investigated by performing both particle loading tests and wall loading tests. The simulation results are then used to evaluate the applicability of five particle breakage criteria reported in literature: major principal stress criterion, maximum tensile stress criterion, mean principal stress criterion, the octahedral shear stress criterion and maximum contact force criterion. The simulation results indicate that only the maximum contact force at crushing keeps nearly constant with the number variation of loading contact. Nevertheless, for given loading contact number, the obtained maximum contact forces at crushing show clear dispersion. The uniaxial, biaxial and triaxial wall loading tests demonstrate that the dispersion of the maximum contact force is minor for given loading configuration, and the maximum contact force does vary with loading contact number. All these simulation results suggest that, when using the above mentioned criteria to evaluate the breakage of particle, the influences of the number and also the spatial arrangement of loading contacts should be taken into account. In chapter five, a brief summary of this work is presented and the possible future works are addressed..