用于甲烷水合物数值模拟的高效求解器

doi:10.11871/jfdc.issn.2096-742X.2026.05.008

数据与计算发展前沿 ›› 2026, Vol. 8 ›› Issue (5): 96-107.

CSTR: 32002.14.jfdc.CN10-1649/TP.2026.05.008

doi: 10.11871/jfdc.issn.2096-742X.2026.05.008

• 专刊：“东方”超级计算系统，助力青年科学家科技创新 • 上一篇下一篇

用于甲烷水合物数值模拟的高效求解器

吕秋楠¹(),徐然²,林安城³,王武³,李刚^1,^*(),李小森¹

¹ 中国科学院广州能源研究所，广东广州 510640
² 中国科学技术大学，安徽合肥 230026
³ 中国科学院计算机网络信息中心，北京 100083

收稿日期:2026-06-25 出版日期:2026-10-20 发布日期:2026-10-20
通讯作者: ^*李刚（E-mail: ligang@ms.giec.ac.cn）
作者简介:吕秋楠，中国科学院广州能源研究所，研究员，博士生导师，主要研究方向为天然气水合物开采技术和水合物技术应用研究等。
本文承担工作为论文撰写及实验设计与实施。
LYU Qiunan is a researcher and Ph.D. Supervisor at the Guangzhou Institute of Energy Conversion, Chinese Academy of Sciences. Her research interests include natural gas hydrate production technologies and hydrate-based applications.
In this paper, she is responsible for the manuscript writing, experimental design and implementation. E-mail: lvqn@ms.giec.ac.cn|李刚，中国科学院广州能源研究所，研究员，博士生导师，主要研究方向为多孔介质中天然气水合物开采理论与关键技术，基于水合物法的海洋/冻土沉积物二氧化碳封存等。
本文承担工作为FISH模拟器模型开发与优化设计。
LI Gang is a researcher and Ph.D. Supervisor at the Guangzhou Institute of Energy Conversion, Chinese Academy of Sciences. His research interests include theory and key technologies of natural gas hydrate production from porous media, as well as in hydrate-based carbon dioxide storage in marine and permafrost deposits.
In this paper, he is responsible for the development and optimization design of the FISH simulator model. E-mail: ligang@ms.giec.ac.cn
基金资助:
国家重点研发计划(2024YFC2814705)

Efficient Solvers for Numerical Simulation of Methane Hydrate

LYU Qiunan¹(),XU Ran²,LIN Ancheng³,WANG Wu³,LI Gang^1,^*(),LI Xiaosen¹

¹ Guangzhou Institute of Energy Conversion, Chinese Academy of Sciences, Guangzhou, Guangdong 510640, China
² University of Science and Technology of China, Hefei, Anhui 230026, China
³ Computer Network Information Center, Chinese Academy of Sciences, Beijing 100083, China

Received:2026-06-25 Online:2026-10-20 Published:2026-10-20

摘要/Abstract

摘要：

【目的】全隐式数值模拟中稀疏线性求解器的选型与优化具有重要的意义。【方法】本文通过对比LAPACK和MUMPS两种直接法求解器，rocALUTION和HYPRE两种迭代法求解器在不同网格规模下的计算性能与适用性，并将计算结果与实验结果进行对比验证，分析求解器选型的影响。【结果】求解器选择对计算速率具有决定性影响，对于数千至数十万网格的隐式求解问题，基于MUMPS的直接法优于迭代法求解器HYPRE；对于百万以上网格规模的模拟，HYPRE展现出更好的求解效率和可扩展性。【结论】对于数万网格以下的非线性方程隐式求解问题，基于MUMPS的直接法优于迭代法。而对于百万网格规模的模拟，则需转向并行直接法或预处理迭代法。本研究为隐式数值模拟中稀疏线性求解器的选择与性能优化提供了定量依据。

关键词: 稀疏线性系统, FISH模拟器, 求解器, MUMPS, HYPRE

Abstract:

[Objective] The selection and optimization of sparse linear solvers are of significant importance in fully implicit numerical simulations. [Methods] The computational performance and applicability of two direct solvers (LAPACK and MUMPS) and two iterative solvers (rocALUTION and HYPRE) were evaluated and compared across different grid scales to analyze the influence of solver selection, with the numerical results validated against experimental data. [Results] Solver selection shows a decisive influence on computational speed for the implicit solution of nonlinear equations. For small to medium-scale problems with grid numbers ranging from thousands to hundreds of thousands, the MUMPS-based direct solver outperforms the iterative solver HYPRE. For large-scale problems with one million or more grid cells, the iterative solver HYPRE demonstrates higher efficiency and better scalability. [Conclusions] For implicit solutions of nonlinear equations on grid scales up to tens of thousands of cells, the MUMPS-based direct solver outperforms iterative solvers. For million-grid-scale simulations, parallel direct solvers or preconditioned iterative methods are required. This study provides a quantitative basis for selecting and optimizing sparse linear solvers in implicit numerical simulations.

Key words: sparse linear systems, FISH simulator, solver, MUMPS, HYPRE

吕秋楠, 徐然, 林安城, 王武, 李刚, 李小森. 用于甲烷水合物数值模拟的高效求解器[J]. 数据与计算发展前沿, 2026, 8(5): 96-107.

LYU Qiunan, XU Ran, LIN Ancheng, WANG Wu, LI Gang, LI Xiaosen. Efficient Solvers for Numerical Simulation of Methane Hydrate[J]. Frontiers of Data and Computing, 2026, 8(5): 96-107, https://cstr.cn/32002.14.jfdc.CN10-1649/TP.2026.05.008.

图/表 10

图1

图2

表1

表2

图3

图4

表3

表4

图5

图6

参考文献 18

[1]	TIONG M, PENG WZ, LIU Q, et al. Natural Gas Hydrate Exploitation: A Comprehensive Review of Structural Properties, Technical Progress and Environmental Challenges[J]. Gas Science and Engineering, 2025, 144: 205769. https://doi.org/10.1016/J.JGSCE.2025.205769.
[2]	王令颁, 孙漪霏, 卜禹豪, 等. 大尺度扇柱形反应釜内甲烷水合物降压开采规律研究[J]. 化工学报, 2025, 76(6): 2958-2973. https://www.ciesc.cn/site/content/4308.html.
[3]	孙贤, 柳军, 王晓辉, 等. 含下伏气的第一类天然气水合物藏开发实验与模拟研究进展[J]. 化工进展, 2024, 43(4): 2091-2103. https://doi.org/10.16085/j.issn.1000-6613.2023-0676.
[4]	孙嘉鑫, 赵洪宝, 曹鑫鑫, 等. 南海荔湾区域水合物水平井降压开采模拟研究[J]. 科学技术与工程, 2021, 21(24): 10246-10256. http://www.stae.com.cn/jsygc/article/abstract/2101804?st=search.
[5]	ANDERSON BJ, KURIHARA M, WHITE MD, et al. Regional Long-Term Production Modeling from A Single Well Test, Mount Elbert Gas Hydrate Stratigraphic Test Well, Alaska North Slope[J]. Marine and Petroleum Geology, 2011, 28(2): 493-501. https://doi.org/10.1016/j.marpetgeo.2010.01.015.
[6]	ZATSEPINA OY, POOLADI-DARVISH M. Storage of CO₂ as Hydrate in Depleted Gas Reservoirs[J]. SPE Reservoir Evaluation & Engineering, 2012, 15(1): 98-108. https://doi.org/10.2118/137313-PA.
[7]	ANDERSON B, HANCOCK S, WILSON S, et al. Formation Pressure Testing at the Mount Elbert Gas Hydrate Stratigraphic Test Well, Alaska North Slope: Operational Summary, History Matching, And Interpretations[J]. Marine and Petroleum Geology, 2011, 28(2): 478-492.https://doi.org/10.1016/j.marpetgeo.2010.02.012.
[8]	余路, 李贤, 崔国栋, 等. 启动压力对南海北部水合物藏开发动态的影响[J]. 地学前缘, 2025, 32(2): 178-194. https://doi.org/10.13745/j.esf.sf.2024.00.00.
[9]	LI G, LI XS, LV QN, et al. Full Implicit Simulator of Hydrate (Fish) and Analysis on Hydrate Dissociation in Porous Media in the Cubic Hydrate Simulator[J]. Energy, 2023, 280: 128191. https://doi.org/10.1016/j.energy.2023.128191.
[10]	WAN YZ, WU NY, CHEN Q, et al. Coupled Thermal-Hydrodynamic-Mechanical-Chemical Numerical Simulation for Gas Production from Hydrate-Bearing Sediments Based on Hybrid Finite Volume and Finite Element Method[J]. Computers and Geotechnics, 2022, 145: 104692. https://doi.org/10.1016/j.compgeo.2022.104692.
[11]	ZHANG ZB, LI YX, LI SD, et al. Optimization of the Natural Gas Hydrate Hot Water Injection Production Method: Insights from Numerical and Phase Equilibrium Analysis[J]. Applied Energy, 2024, 361: 122963. https://doi.org/10.1016/j.apenergy.2024.122963.
[12]	LI G, LV QN, LI XS, et al. Analysis of Methane Hydrate Dissociation Experiment in a Pilot-Scale Hydrate Simulator by a Full Implicit Simulator of Hydrate[J]. Industrial & Engineering Chemistry Research, 2023, 62(19): 7704-7720. https://doi.org/10.1021/acs.iecr.3c00597.
[13]	LI B, WEI WN, LI G, et al. Experimental and Theoretical Investigation on Decomposition of Hydrate-bearing Layer by Full Implicit Simulator of Hydrate[J]. International Journal of Heat and Mass Transfer, 2025, 245: 127001. https://doi.org/10.1016/J.IJHEATMASSTRANSFER.2025.127001.
[14]	LI G, LI XS, LV QN, et al. Long-Term Analysis of the 2020 Gas Hydrate Production Test in the South China Sea by Full Implicit Simulator of Hydrate[J]. Energy & Fuels, 2023, 37(5): 3785-3798. https://doi.org/10.1021/acs.energyfuels.2c04355.
[15]	LI G, ZHAO ZC, LI XS, et al. Control Factors for Gas Recovery in the Second Methane Hydrate Field Test in the Nankai Trough[J]. Energy & Fuels, 2023, 37(18): 13985-13996. https://doi.org/10.1021/acs.energyfuels.3c02695.
[16]	ZHANG JC, MORIDIS GJ, BLASINGAME TA. Effect of Geomechanics and of Grid Discretization on The Predictions of Production from Natural Hydrate Deposits and of The Associated Geomechanical System Response[J]. Gas Science and Engineering, 2023, 112: 204942. https://doi.org/10.1016/j.jgsce.2023.204942.
[17]	AMESTOY P R, DUFF I S, L'EXCELLENT J Y, et al. A Fully Asynchronous Multifrontal Solver Using Distributed Dynamic Scheduling[J]. Siam Journal on Matrix Analysis and Applications, 2001, 23 (1): 15-41. https://doi.org/10.1137/S0895479899358194.
[18]	Henson V E, Yang U M. BoomerAMG: A Parallel Algebraic Multigrid Solver and Preconditioner[J]. Applied Numerical Mathematics, 2002, 41 (1): 155-177. https://doi.org/10.1016/S0168-9274(01)00115-5.

参数	数值
反应器有效容积（L）	88.3
气体组分（-）	99.9% CH₄
石英砂孔隙度（%）	52.17
注入气体总量（mol）	181.4
注水量（kg）	30.97
生成结束最终温度（℃）	11.7
初始气体饱和度S_G（%）	27.80
初始水饱和度S_A（%）	44.94
初始水合物饱和度S_H（%）	27.26

操作系统	Linux x86-64 Centos 7
编译器	compiler/intel/2017.5.239
MPI	mpi/hpcx/2.7.4/intel-2017.5.239
计算节点异构加速卡	海光一号CPU（32核心）+128 G内存 4×DCU（Z100，16 GB显存）
LAPACK	Intel MKL
MUMPS	MUMPS_5.4.1+Metis-4.0.3
rocALUTION	rocm/rocm-3.9.1/rocalution
HYPRE	hypre-v2.32+Intel MKL

模拟算例	总耗时（s）	构建矩阵占比（%）	求解方程组占比（%）
RUN 1-725	22,130	33.39	41.65
RUN 2-725	1,272	14.80	83.29
RUN 3-725	33,648	11.39	88.26
RUN 4-725	11,910	2.97	96.31

模拟算例	总耗时（s）	构建矩阵占比（%）	求解方程组占比（%）
RUN 2-725	1，272	14.80	83.29
RUN 2-1，450	4，472	13.61	85.19
RUN 2-2，900	10，371	14.68	84.34
RUN 2-11，600	104，276	13.29	86.02

用于甲烷水合物数值模拟的高效求解器

Efficient Solvers for Numerical Simulation of Methane Hydrate

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 18

相关文章 1

编辑推荐

Metrics

本文评价