槽流拟颗粒模型的并行算法

Parallel algorithm of pseudo-particle simulation for channel flow

将流体处理为离散粒子,应用拟颗粒硬球模型来研究槽流中的流动现象,与分子动力学模拟的算法类似,是研究槽流机理的一种行之有效的方法。为了作大规模的模拟,本文采用区域分解算法和消息传递编程模型技术,将该模型串行程序并行化,应用一维划分、单相传递的方法简化了并行算法,采用轮换搜索法来避免硬球碰撞次序对结果的影响。在可扩展的机群系统上用实例计算,通过与串行程序的对比,验证了并行程序的正确性,表明本文设计的并行算法取得了较高的并行计算效率。

The pseudo-particle model （PPM） is an effective approach used in high-resolution simulation of particle fluid systems. PPM simulation is a suitable method for turbulent flow research that can be implemented with an algorithm similar to molecular dynamics （MD） simulation. To avoid the influence of particles＇ collision sequences, the external forces using Rotation Searching Method are computed at each time step. The spatial decomposition method is used in the parallel algorithm because of the great number of hard sphere particles in channel flow systems. In this manner classical method used in the parallel codes is predigested with One-Dimensional Partition and One-Way Transmission. The parallel implementation will be helpful to make use of Hard Sphere method in large-scale simulations of channel flow.