摘要
格子Boltzmann方法 LBM(Lattice Boltzmann Method)已成为计算流体力学中模拟不可压缩流的有效工具。结合D2Q9多松弛时间格子Boltzmann方法和大涡模拟方法模拟复杂流体流动。为了模拟高雷诺数下流场中出现的湍流,提高网格的精度,提出格子Boltzmann方法的并行算法。算法中包括格子类型的判断、区域分解和数据交换策略。考虑到集群的负载均衡,给出区域分解时子区域的详细划分策略。数值结果表明,算法具有良好的可扩展性,效率可以在128个核心上达到86. 5%。
The lattice Bohzmann method has become an efficient approach for simulating incompressible flow in computational fluid dynamics. In this paper, we combined D2Q9 multi-relaxation-time lattice Bohzmann method with large eddy simulation to simulate complex fluid flow. In order to simulate the turbulent flows in the flow field at high Reynolds numbers, we improved the accuracy of the grid and presented a parallel algorithm for the lattice Bohzmann method. The algorithm included the judgments of lattice style, domain decomposition and data exchange strategy. Considering load balancing on large scale cluster, we offered partition strategy of subdomain in detail in domain decomposition. The numerical results indicate that the presented algorithm has considerable scalability and the efficiency can reach 86.5% on 128 cores.
作者
徐磊
宋安平
刘智翔
张武
Xu Lei;Song Anping;Liu Zhixiang;Zhang Wu(School of Computer Engineering and Science,Shanghai University,Shanghai 200444,Chin;College of Information Technology,Shanghai Ocean University,Shanghai 201306,China)
出处
《计算机应用与软件》
北大核心
2018年第10期226-231,共6页
Computer Applications and Software
基金
国家自然科学基金重大研究计划重点支持项目(91630206)
