颗粒物质是由大量离散的、直径在微米到毫米量级的宏观颗粒组成的聚集物，具有丰富的非平衡现象与行为，如颗粒聚集、颗粒分离、巴西果效应和反巴西果效应等。最近的实验研究发现自由下落的颗粒流会发生断裂聚集现象。到目前为止，大多数关于自由下落颗粒流断裂聚集现象的研究集中于不同材料、不同的表面特性、空气等因素对它的影响。而对于各参数对自由下落颗粒流行为的影响，特别是自由下落颗粒流产生断裂聚集现象的物理机制及临界行为还缺乏系统的研究。 本文主要采用分子动力学方法对自由下落颗粒流的行为进行模拟研究，发现除内聚力外，耗散也是自由下落颗粒流发生断裂聚集行为的主要原因，还发现了团聚系数与内聚力之间的幂律关系、与耗散常数之间的临界饱和行为。 在第一章中，简要介绍了颗粒物质及其聚集特性，分析了发生颗粒聚集现象的原因。在第二章中，通过文献综述对自由下落颗粒流发生断裂聚集现象的原因进行分析，得出其主要原因是颗粒间微弱的内聚力。第三章介绍了本文研究所采用的分子动力学模拟方法。 在第四章中，首先用分子动力学方法制备了颗粒在漏斗容器内堆积的初始状态，然后模拟得到当内聚力不同时自由下落颗粒流分别表现出断裂聚集和散开的行为，验证了Royer等人的实验结果。将团聚系数定义为颗粒流中粒子总数除以聚团总数，此系数可以描述颗粒流的行为。团聚系数越大表明颗粒流的断裂聚集行为越明显，团聚系数趋于1说明颗粒流散开。随后通过改变模拟参数研究了内聚力、耗散常数和切向摩擦系数对自由下落颗粒流行为的影响。发现团聚系数随内聚力的增大普遍以幂律形式增大，幂指数在0.52到0.59之间；团聚系数随耗散常数的增大在临界点处发生临界行为，然后以幂指数0.741247143±0.11442指数增大，最终趋于饱和，并且此临界饱和行为是普遍的，说明耗散也是自由下落颗粒流发生断裂聚集现象的主要原因；改变切向摩擦系数时团聚系数基本不发生变化，说明切向摩擦系数不影响自由下落颗粒流的行为。本文还应用MATLAB程序制作出自由下落颗粒流演化的动画，最后讨论了自由下落颗粒流中聚团大小的分布情况。

Granular materials consist of huge numbers of discrete and macroscopic particles with diameters ranging from microns to millimeters. Granular materials show a wealth of non-equilibrium phenomena, such as clustering, segregation, the Brazil nut effect, the reverse Brazil nut effect, and so on. Recent studies show that free-falling granular streams can commonly break up into clusters. Moreover, most of the current studies focus on the effect of grain asperities, surface features and surrounding air. However, the behavior of free-falling streams under different parameters, especially, the physical mechanism and the critical behavior of rupture-clustering phenomenon, are far from clear, and a systematic study is required.In this thesis, the behavior of free-falling granular streams is studied by molecular dynamics simulation. Our results reveal that: (1). the inelasticity (damping), besides the cohesive force, is also the key mechanism for the rupture-clustering phenomenon. (2). The clustering quotient has a power-law relationship with the cohesive force among particles. (3). A critical damping constant exists for the emergence of the rupture-clustering transition, and the clustering quotient saturates when the damping constant becomeslarge enough.In Chapter 1, we give a brief introduction to granular material, the clustering phenomena and studies of its mechanism. In Chapter 2, we point out that the cluster formation is driven by minute, nanoNewton cohesive forces among particles. In Chapter 3, we introduce the molecular dynamics used in our simulation programmes.In Chapter 4, we firstly give the initial condition which is a pile of 20,000 grains in a funnel for the free-falling granular stream simulation. Then two opposite behaviors, namely the rupture-clustering and the drift-apart phenomena of the streams, are observed when the cohesive force is changed. This phenomena have been observed experimentally by Royer and his colleagues. We define a clustering quotient to characterize the degree of clustering of free-falling granular streams. When grains in the stream drift completely apart, the clustering quotient equals 1, whereas the stream ruptures into clusters the clustering quotient is larger than 1.The effects of the cohesive force, the damping constant and the tangential damping on the behavior of granular streams are investigated. Our results reveal that: (1). The clustering quotient has a universal power-law relation with the cohesive force among particles, with the exponent ranging from 0.52 to 0.59. (2). Clustering quotient exhibits a critical transition behavior when damping constant is larger than a critical value, and the quotient saturates as the damping constant becomes large enough. This critical transition of clustering quotient is also a universal behavior, with a critical exponent 0.741247143±0.11442. This result reveals that the damping is also crucial for the behavior of free-falling granular streams. (3). The clustering quotient remains almost unchanged when the tangential damping varies from E-5 to E5, implying that tangential damping plays a neglectable role in rupture-clustering phenomenon.We also create movies of our simulation results by using MATLAB to give a visualized observation of the rupture-clustering phenomenon of free-falling granular streams. In the last part of Chapter 4, we discuss the size distribution of clusters under different parameters.