The Implementation and Efficiency Analysis of Parallel Mesh Mapping based on the Multi-physics Coupling Framework
-
摘要: 反应堆精细化物理热工耦合计算可以更准确地模拟堆芯行为，但现有分析程序对不同物理场进行计算时，采用不同的离散格式和网格划分，从而导致各个物理场之间离散变量的传递需要复杂网格映射关系，特别是全堆芯精细化建模，其大规模网格映射将影响耦合系统的求解精度与效率。本文基于自主研发的多物理耦合框架MORE，以及集成于MORE的热工水力子通道软件CORTH、蒙卡程序RMC，采用区域分解并行网格映射的方法，实现了全堆芯精细网格的物理热工耦合计算，百万级的结构化网格与非结构化网格映射，20个核并行映射时间最少为8 s，最高并行映射效率为10个核并行所达到的77.96%，提升了耦合计算效率。Abstract: Refined physical-thermal coupling calculations of reactors can simulate the core behavior more accurately. However, existing analysis programs adopt different discrete formats and mesh divisions when calculating different physical fields, resulting in a complex mesh mapping relationship for the transfer of discrete variables between physical fields. Especially for the refined whole-core modeling, the large-scale mesh mapping will affect the accuracy and efficiency of the coupled system solution. In this paper, based on the self-developed multi-physics coupling framework MORE, the thermal-hydraulic sub-channel software CORTH integrated in MORE, and the Monka program RMC, the whole core refined mesh of the physical-thermal coupling calculation was realized by the method of area decomposition parallel mesh mapping. The mapping time between the million-level structured mesh and unstructured mesh can be reduced to 8 s on 20 cpus, and maximum parallel mapping efficiency reaches 77.96% on 10 cpus, which improves the coupling calculation efficiency.
-
图 1 RMC/CORTH并行计算流程示意图
L—源组所使用的核数；N—RMC所使用的核数；M—CORTH所使用的核数
Figure 1. Parallel Computing Flow Chart of RMC/CORTH
图 2 MORE界面上搭建RMC/CORTH耦合计算流程
Figure 2. Building RMC/CORTH Coupling Calculation Process on MORE
图 6 JRR-3单组件第8个燃耗步轴向功率因子对比图以及热流密度分布图
Figure 6. Axial Normalized Power Factor Comparison Diagram and Heat Flux Distribution Diagram in the 8th Burn-up Step of JRR-3 Single Assembly
图 7 JRR-3单组件第8个燃耗步冷却剂密度、温度分布对比图
Figure 7. Comparison Diagram of Coolant Density and Temperature Distribution in the 8th Burn-up Step of JRR-3 Single Assembly
图 8 JRR-3单组件第8个燃耗步的包壳温度、燃料温度分布对比图
Figure 8. Comparison Diagram of Cladding Temperature and Fuel Temperature Distribution in the 8th Burn-up Step of JRR-3 Single Assembly
表 1 测试算例及其网格数说明
Table 1. Description of Test Example and its Grid Number
算例名 RMC CORTH X方向 Y方向 Z方向 网格总数 子通道数目 燃料板数目 轴向节块数 网格总数 JRR-3单组件 16 102 25 40800 21 20 25 1025 JRR-3全堆 112 714 25 1999200 777 740 25 37925 IAEA 10 MW MTR全堆 64 585 20 748800 432 414 20 16920 RSG-GAS全堆 128 856 20 2191360 1144 1092 20 44720 
下载: 导出CSV
表 2 JRR-3全堆算例并行映射时间及效率
Table 2. Parallel Mapping Time and Efficiency of JRR-3 Full Core
程序名 CORTH RMC 网格类型 非结构网格 结构网格 网格单元数 37925 1999200 映射关系 CORTH→RMC RMC→CORTH 核数 1核 5核 10核 15核 20核 1核 5核 10核 15核 20核 3次并行映射平均时间/s 148.33 40.00 22.67 19.00 18.67 148.00 65.00 34.67 21.00 22.00 并行映射效率/% 100.00 74.17 65.43 52.05 39.72 100.00 45.54 42.69 46.98 33.64
下载: 导出CSV
表 3 IAEA全堆算例并行映射时间及效率
Table 3. Parallel Mapping Time and Efficiency of IAEA Full Core
程序名 CORTH RMC 网格类型 非结构网格 结构网格 网格单元数 16920 748800 映射关系 CORTH→RMC RMC→CORTH 核数 1核 5核 10核 15核 20核 1核 5核 10核 15核 20核 3次并行映射平均时间/s 88.33 22.67 11.33 9 8 88.33 34.33 19 15 12 并行映射效率/% 100 77.93 77.96 65.43 55.21 100 51.46 46.49 39.26 36.80
下载: 导出CSV
表 4 RGS-GAS全堆算例并行映射时间及效率
Table 4. Parallel Mapping Time and Efficiency of RGS-GAS Full Core
程序名 CORTH RMC 网格类型 非结构网格 结构网格 网格单元数 44720 2191360 映射关系 CORTH→RMC RMC→CORTH 核数 1核 5核 10核 15核 20核 1核 5核 10核 15核 20核 3次并行映射平均时间/s 136.67 38.67 22.67 19 19 137 45.67 25 19.33 15 并行映射效率/% 100 70.69 60.29 47.95 35.97 100.00 60.00 54.80 47.25 45.67
下载: 导出CSV
-
[1] TURINSKY P J, KOTHE D B. Modeling and simulation challenges pursued by the Consortium for Advanced Simulation of Light Water Reactors (CASL)[J]. Journal of Computational Physics, 2016, 313: 367-376. doi: 10.1016/j.jcp.2016.02.043 [2] STANEK C R. Overview of DOE-NE NEAMS program: A-UR-19-22247[R]. Los Alamos: Los Alamos National Lab, 2019. [3] CHAULIAC C, ARAGONÉS J M, BESTION D, et al. NURESIM – A European simulation platform for nuclear reactor safety: Multi-scale and multi-physics calculations, sensitivity and uncertainty analysis[J]. Nuclear Engineering and Design, 2011, 241(9): 3416-3426. doi: 10.1016/j.nucengdes.2010.09.040 [4] JIMENEZ G, HERRERO J J, GOMMLICH A, et al. Boron dilution transient simulation analyses in a PWR with neutronics/thermal-hydraulics coupled codes in the NURISP project[J]. Annals of Nuclear Energy, 2015, 84: 86-97. doi: 10.1016/j.anucene.2014.11.002 [5] CHANARON B. Overview of the NURESAFE European Project[J]. Nuclear Engineering and Design, 2017, 321: 1-7. doi: 10.1016/j.nucengdes.2017.09.001 [6] 佘顶,王侃,余纲林. 堆用蒙卡程序燃耗计算功能开发[J]. 核动力工程,2012, 33(3): 1-5,11. [7] 梁金刚,王侃,余纲林,等. 基于RMC的计数器数据分解方法研究[J]. 核动力工程,2014, 35(4): 142-146. doi: 10.13832/j.jnpe.2014.04.0142 [8] 潘俊杰,曾未,强胜龙,等. 基于耦合框架的多通道程序网格与接口模块开发[J]. 原子能科学技术,2022, 56(S1): 32-41. [9] 崔显涛,强胜龙,匡邓晖,等. 基于最近邻域算法的蒙卡负载均衡实现[J]. 核动力工程,2021, 42(S2): 37-40. -
计量
- 文章访问数: 317
- HTML全文浏览量: 41
- PDF下载量: 69
- 被引次数: 0
