首页 | 本学科首页   官方微博 | 高级检索  
     

一种全堆芯精确到每个通道的子通道并行模拟方法
引用本文:王先梦,赵民富,吕玉凤,蔡银宇,储根深,卢旭,王昭顺,郭苏萱,周志锋,胡长军,杨文.一种全堆芯精确到每个通道的子通道并行模拟方法[J].原子能科学技术,2020,54(6):1108-1117.
作者姓名:王先梦  赵民富  吕玉凤  蔡银宇  储根深  卢旭  王昭顺  郭苏萱  周志锋  胡长军  杨文
作者单位:北京科技大学 计算机与通信工程学院,北京100083;中国原子能科学研究院 反应堆工程技术研究部,北京102413
摘    要:为实现全堆芯精确到每个通道的并行子通道模拟,本文提出一种基于子通道的并行任务划分和进程映射方法,可对全堆芯或单个组件进行计算任务划分,计算任务和进程的映射可灵活进行。该方法可根据计算机(群)的核数确定恰当的全堆芯子通道的任务划分方式,从而使全堆芯热工水力模拟可在单机、小型集群到超级计算机等不同环境运行。在天河二号超级计算机上进行全堆芯157组件、精确到每个真实流道、轴向划分为125层的稳态模拟,可使用4~6 280核实现。使用4核时需约22 h,使用6 280核时需470 s。引入混合编程实现方式后,使用6 280核完成模拟需397 s。


Parallel Strategy for Full-core and Real-channel-resolved Thermal-hydraulic Subchannel Simulation
WANG Xianmeng,ZHAO Minfu,LYU Yufeng,CAI Yinyu,CHU Genshen,LU Xu,WANG Zhaoshun,GUO Suxuan,ZHOU Zhifeng,HU Changjun,YANG Wen.Parallel Strategy for Full-core and Real-channel-resolved Thermal-hydraulic Subchannel Simulation[J].Atomic Energy Science and Technology,2020,54(6):1108-1117.
Authors:WANG Xianmeng  ZHAO Minfu  LYU Yufeng  CAI Yinyu  CHU Genshen  LU Xu  WANG Zhaoshun  GUO Suxuan  ZHOU Zhifeng  HU Changjun  YANG Wen
Affiliation:School of Computer and Communication Engineering, University of Science and Technology Beijing, Beijing 100083, China; Division of Reactor Engineering Technology Research, China Institute of Atomic Energy, Beijing 102413, China
Abstract:In order to implement full-core and real-channel-resolved parallel subchannel simulations, a subchannel-based parallel partitioning and mapping method was presented. The method is able to partition either a full-core or a single assembly, and it maps the computation tasks and processes flexibly. The method chooses the appropriate partitioning method according to the core number of the computers/clusters used, which enables the full-core thermal-hydraulic simulation to be carried out on various computing environments ranging from personal computers to supercomputers. A full-core and real-channel-resolved steady simulation, which analyzes 157 assemblies with each subchannel divided into 125 layers axially, is carried out on the Tianhe Ⅱ supercomputer using 4-6 280 cores. The running time is about 22 h using 4 cores and the running time is 470 s using 6 280 cores. By utilizing hybrid parallel computing, the running time on 6 280 cores is reduced to 397 s.
Keywords:thermal-hydraulics  subchannel simulation  parallel computation  parallel partitioning  MPI  OpenMP  
点击此处可从《原子能科学技术》浏览原始摘要信息
点击此处可从《原子能科学技术》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号