颗粒离散元法是一种广泛应用于研究颗粒物料力学行为的数值模拟方法,而计算效率是制约其发展和应用的主要因素之一。通过Pro/E软件建立了料斗模型,利用StreamDEM软件对料斗的颗粒充填过程进行离散元法模拟研究,并对基于CPU和GPU加速算法的运算过程和结果进行对比。结果表明,基于GPU的计算机图形学加速算法可大幅提高颗粒离散元法模拟过程的运算效率。当填充颗粒数量达到13万时,其运算效率比基于CPU的运算效率提高了10倍以上。
Particle discrete element method is a kind of numerical simulation method widely used in the research of granular material mechanics behaviour. Computation efficiency is one of the main factors that restricts its development and application. In this paper, we build a hopper model by using Pro/E software, and use Stream DEM software to study the stimulations of discrete element method in regard to hopper, s particles filling process. We also compare the operation processes and results of CPU-based and GPU-based acceleration algorithms.Results show that the GPU-based computer graphics acceleration algorithm can dramatically improve the computation efficiency of the simulation process of particle discrete element method. When the number of particles to be filled reaches 130 000, its computational efficiency improves over 10 times than that of the CPU-based acceleration algorithm.