采用离散元素法(discrete element method, DEM)进行颗粒系统运动仿真时,其模拟计算量大、计算效率低下,所采用的传统中央处理器(central processing unit, CPU)并行计算模型难以实现较大规模模拟。文章提出了一种基于图形处理单元(grap...采用离散元素法(discrete element method, DEM)进行颗粒系统运动仿真时,其模拟计算量大、计算效率低下,所采用的传统中央处理器(central processing unit, CPU)并行计算模型难以实现较大规模模拟。文章提出了一种基于图形处理单元(graphics processing unit, GPU)和统一计算设备架构(compute unified device architecture, CUDA)的并行计算方法;以球磨机的介质运动仿真为例,利用DEM方法结合CUDA并行计算模型,充分利用GPU众核多线程的计算优势,同时将颗粒属性信息存入GPU的常量存储器,减少信息读取的时滞,将筒体和衬板视为圆柱面和平面,简化了筒体与颗粒的接触判断,实现每个线程处理1个颗粒的相关计算,大幅提高计算速度;对颗粒堆积、筒体内2种尺寸颗粒运动进行仿真,并与基于CPU并行计算的结果进行对比。研究结果表明:在同等价格的硬件条件下,该文的方法可以实现10倍以上的加速比;对于含有复杂几何模型的仿真,如多尺寸颗粒和带衬板筒体的仿真,加速比会减少,但仍然可以实现数倍的加速。展开更多
In order to study particle segregation in the rotating drum,the movement of particles is studied by theoretical deduction and numerical simulation.According to the theoretical deduction,particles near the end wall are...In order to study particle segregation in the rotating drum,the movement of particles is studied by theoretical deduction and numerical simulation.According to the theoretical deduction,particles near the end wall are lifted higher by the friction of the end wall,which leads to small particles gathering in the middle of the drum.The model of particle motion is established based on the discrete element method.It can be shown from the simulation results that the particles accumulated highernear the end w al,which is consistent with the theoretical deduction.In addition,the effects from the aspects of te relative friction between particles,the friction between the end wall a d particle,the drum aspect ratio and the rotation speed were explored by thesimulation.From the simulation results,it can be pointed out that the friction of the eed w a i has a decisive influence on the axial segregation of theparticles,and the other factors merely affect the accumulate forms of particles.展开更多
文摘采用离散元素法(discrete element method, DEM)进行颗粒系统运动仿真时,其模拟计算量大、计算效率低下,所采用的传统中央处理器(central processing unit, CPU)并行计算模型难以实现较大规模模拟。文章提出了一种基于图形处理单元(graphics processing unit, GPU)和统一计算设备架构(compute unified device architecture, CUDA)的并行计算方法;以球磨机的介质运动仿真为例,利用DEM方法结合CUDA并行计算模型,充分利用GPU众核多线程的计算优势,同时将颗粒属性信息存入GPU的常量存储器,减少信息读取的时滞,将筒体和衬板视为圆柱面和平面,简化了筒体与颗粒的接触判断,实现每个线程处理1个颗粒的相关计算,大幅提高计算速度;对颗粒堆积、筒体内2种尺寸颗粒运动进行仿真,并与基于CPU并行计算的结果进行对比。研究结果表明:在同等价格的硬件条件下,该文的方法可以实现10倍以上的加速比;对于含有复杂几何模型的仿真,如多尺寸颗粒和带衬板筒体的仿真,加速比会减少,但仍然可以实现数倍的加速。
基金The National Natural Science Foundation of China(No.51775109)Jiangsu Science and Technology Support Plan Project(No.BE2014142)
文摘In order to study particle segregation in the rotating drum,the movement of particles is studied by theoretical deduction and numerical simulation.According to the theoretical deduction,particles near the end wall are lifted higher by the friction of the end wall,which leads to small particles gathering in the middle of the drum.The model of particle motion is established based on the discrete element method.It can be shown from the simulation results that the particles accumulated highernear the end w al,which is consistent with the theoretical deduction.In addition,the effects from the aspects of te relative friction between particles,the friction between the end wall a d particle,the drum aspect ratio and the rotation speed were explored by thesimulation.From the simulation results,it can be pointed out that the friction of the eed w a i has a decisive influence on the axial segregation of theparticles,and the other factors merely affect the accumulate forms of particles.
