湍流模拟壁面距离MPI/OpenMP混合并行计算方法

赵钟; 何磊; 张健; 徐庆新; 张来平

doi:10.7638/kqdlxxb-2017.0199

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

湍流模拟壁面距离MPI/OpenMP混合并行计算方法

研究论文 | 更新时间：2026-05-28

- 湍流模拟壁面距离MPI/OpenMP混合并行计算方法
- MPI/OpenMP hybrid parallel computation of wall distance for turbulence flow simulations
- 空气动力学学报 2019年37卷第6期页码：883-892
- 作者机构：
  
  中国空气动力研究与发展中心计算空气动力研究所,四川,绵阳,621000
- 作者简介：
  
  赵钟(1986-), 男, 四川宜宾人, 助理研究员, 研究方向:非结构网格生成方法及软件开发, 计算流体力学及软件开发.E-mail:bell_cardc@163.com
- 基金信息：
  
  国家重点研发计划（2016YFB0200701）;国家自然科学基金（11532016，91530325）
- DOI：10.7638/kqdlxxb-2017.0199
  中图分类号： V211.3
- 纸质出版：2019
- 稿件说明：
移动端阅览
赵钟, 何磊, 张健, 徐庆新, 张来平. 湍流模拟壁面距离MPI/OpenMP混合并行计算方法[J]. 空气动力学学报, 2019, 37(6): 883-892.

ZHAO Zhong, HE Lei, ZHANG Jian, et al. MPI/OpenMP hybrid parallel computation of wall distance for turbulence flow simulations[J]. 2019, 37(6): 883-892.
赵钟, 何磊, 张健, 徐庆新, 张来平. 湍流模拟壁面距离MPI/OpenMP混合并行计算方法[J]. 空气动力学学报, 2019, 37(6): 883-892. DOI： 10.7638/kqdlxxb-2017.0199.

ZHAO Zhong, HE Lei, ZHANG Jian, et al. MPI/OpenMP hybrid parallel computation of wall distance for turbulence flow simulations[J]. 2019, 37(6): 883-892. DOI： 10.7638/kqdlxxb-2017.0199.

摘要

针对计算流体力学在湍流数值模拟过程中壁面距离计算效率不高的问题，设计了一种基于ADT数据结构搜索的MPI/OpenMP混合并行计算方法，以大幅提高大规模网格壁面距离的计算效率，降低因内存消耗而对网格规模的限制。首先分析了壁面距离计算精度对湍流模拟的重要性，介绍了壁面距离计算的几何基础。随后基于区域分解思想，将计算域划分为不同的子分区，服务器进程收集全局壁面面元网格信息后发送给其他所有进程，各进程根据全局壁面信息，基于ADT数据结构搜索其网格分区内的单元，精确计算壁面距离。为了解决全局壁面信息内存过大的问题，采用MPI/OpenMP混合并行算法，使得各计算节点中仅有一个或少数几个壁面信息备份，这些壁面信息备份被节点内的其他各进程所共享。最后采用大规模网格进行了壁面距离计算测试，网格规模最大达到33.2亿，结果表明，该方法的计算精度和直接搜索法一致，内存耗费下降70%，计算时间减少约1个量级，能满足大规模CFD数值模拟的需求。

Abstract

A parallel wall distance computing approach is proposed for large scale CFD turbulence flow simulations to reduce the wall distance computing time consumption. This method is based on MPI/OpenMP hybrid parallel ADT searching procedure to further improve the computational efficiency and to reduce memory consumption. Firstly

the influence of the accuracy of wall distance computing is analyzed

and the foundational computational geometry algorithm used in this paper is introduced. Then the parallel wall distance computing approach based on the ADT searching procedure is discussed in details. In this method

the total computational field is divided into a number of zones firstly

and then the whole surface faces data on the solid wall is collected and broadcasted to all other processors using MPI communications. The wall distances of spatial cells in each zone are computed based on the ADT searching within each processor. In order to overcome the memory limitation

a MPI/OpenMP hybrid parallel method is proposed

and then only one or a few copies of the surface faces data are created within each computer node other than for each processor. Numerical results show that the present method substantially improves efficiency by one order and reduces the memory consumption by 70%

which indicates that this method can fullfill the demand of large-scale parallel simulations.

关键词

Keywords

references

ZHOU H, ZHANG H X. What is the essence of the so-called century lasting difficult problem in classic physics, the "problem of turbulence"?[J]. Scientia Sinica Physica, Mechanica Astronomica, 2012, 42(1): 1-5. (in Chinese)

TUCKER P G, RUMSEY C L, SPALART P R. Computation of wall distances based on differential equations[R]. AIAA 2004-2232.

TUCKER P G. Assessment of geometric multilevel convergence robustness and a wall distance method for flows with multiple internal boundaries[J]. Applied Mathematical Modeling, 1998, 22(4): 293-311.

LI G N, LI F W, ZHOU Z H. An efficient method for calculating wall distance[J]. Advances in Aeronautical Science and Engineering, 2010, 1(2): 137-142. (in Chinese)

VOS J B, LEYLAND P. NSMB handbook 5.0[R]. IMHEF-DGM-EPFL, 2003.

SPALART P R. Trends in turbulence treatments[R]. AIAA 2000-2306.

WANG G, ZENG Z, YE Z Y. An efficient search algorithm for calculating minimum wall distance of unstructured mesh[J]. Journal of Northwestern Polytechnical University, 2014, 32(4): 511-516. (in Chinese)

ZHAO H Y, HE X Z, LE J L. Recursive box method for wall distance computation[J]. Chinese Journal of Computational Physics, 2008, 25(4): 427-430. (in Chinese)

YANG Y G, CHENG X H, WANG W J. A parallel computation method of wall distance for large-scale flow simulation[J]. Computer Engineering Science, 2016, 38(6): 1086-1090. (in Chinese)

BOGER D A. Efficient method for calculating wall proximity[J]. AIAA Journal, 2001, 9(12): 2404-2406.

SPALART P R, ALLMARAS S R. A one equation turbulence model for aerodynamic flows[R]. AIAA 92-0439.

JOSEPH O R. Computational geometry in C[M]. Cambridge University Press, 1997.

TOMAS M, BEN T. Fast minimum storage ray-triangle intersection[J]. Journal of Graphics Tools, 1997, 2(1): 21-28.

JAVIER B, JAIME P. An alternating digital tree (ADT) algorithm for 3D geometric searching and intersection problems[J]. International Journal for Numerical Methods in Engineering, 1997, 31: 1-17.

HE X, ZHAO Z, ZHANG L P. The research and development of structured-unstructured hybrid CFD software[J]. Transactions of Nanjing University of Aeronautics Astronautics, 2013, 30(s): 116-126.

HE X, ZHANG L P, ZHAO Z. Research of general large scale CFD software architecture and data structure[J]. Acta Aerodynamica Sinica, 2012, 30(5): 557-565. (in Chinese)

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

国家数值风洞风雷软件研发与开源应用进展

国家数值风洞(NNW)软件自动化集成与测试平台设计与研发

类HyTRV升力体阻尼/交叉动导数特性分析

基于Ac-准则的最优输入设计方法

大展弦比飞行器突风响应数值计算