混合粒群在振动作用下的离析分层为矿物颗粒的筛分分级提供了实现基础，颗粒形状特征及粒级结构是影响离析速率及结果的重要因素。基于离散元法数值模拟探究了单个(非)球形大颗粒在小颗粒群中离析动力学特性，研究了颗粒形状对其离析速率的影响规律，分析了二元均匀混合粒群中形状与粒度之间的交互作用，揭示了多元混合粒群中离析速率及转动取向分布规律。试验结果表明：较强振动作用下，球形大颗粒离析速率随粒度比的增加呈现先增加后降低的非单调变化趋势，而非球形大颗粒离析速率显著高于球形大颗粒，主要原因是其颗粒底端延伸至孔隙率波动较大的深床层。此外，非球形大颗粒与球形大颗粒的离析速率差值随着振幅的增加同样呈现先升高后降低的变化趋势；当二元混合颗粒群中存在对流时，其离析速率明显高于无对流颗粒群的离析速率；二元混合料群中非球形大颗粒的离析速率与转动取向呈现“类高斯”分布特征，且非球形颗粒的形状与粒度对最终离析结果的影响存在竞争关系，导致球形大颗粒−非球形小颗粒床在小粒度比、低球形度条件下出现异常的“反巴西果”现象；此外，颗粒形状对离析速率的影响显著性高于粒级结构，增加中间粒级颗粒会减缓粒群的离析过程，而拓宽粒度分布范围同样也会造成离析速率下降。

The segregation of granules under vibration is the premise of minerals screening, and particle shape and size are important factors affecting the segregation rate and results. Based on the Discrete Element Method simulation, the dynamics of a single (non-)spherical large particle segregation in the sea of small particles were investigated, the influence of particle shape on its segregation rate was studied. The interaction effect between the shape and the size in a binary uniform mixed particle bed was analyzed, and the distributions of segregation rates as well as the orientations in a multi-component particle group were revealed. The results show that under a strong vibration, the segregation rate of spherical large particles showed a non-monotonic trend of first increasing and then decreasing with the increase of particle size ratio. And the segregation rate of non-spherical large particles is significantly higher than that of spherical large particles, mainly because the particle's bottom end extends to a deeper bed where large porosity fluctuation occurs. In addition, the segregation rate difference between non-spherical large particles and spherical large particles also shows a trend of first increasing and then decreasing with the increase of amplitude. When there is convection in the binary mixed particle group, the segregation rate is significantly higher than that of the non-convective particle group. The segregation rates and rotation orientations of non-spherical particles in the binary bed show a “Gaussian-like” distribution, and there is a competition effect between shape and size, which leads to the abnormal “Reverse-Brazil nut” phenomenon when a small size ratio and low sphericity are given. In addition, the influence of particle shape on the segregation rate is significantly higher than that of particle size structure. Adding intermediate-size particles slows down the segregation process while widening the size distribution range also reduces the segregation rate.