数据与计算发展前沿 ›› 2026, Vol. 8 ›› Issue (5): 84-95.

CSTR: 32002.14.jfdc.CN10-1649/TP.2026.05.007

doi: 10.11871/jfdc.issn.2096-742X.2026.05.007

• 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

面向国产DCU的黑油模拟器高效求解关键方法研究

吴学凇(),王宇鹏*(),曹建文*(),卢亚楠   

  1. 中国科学院软件研究所， 并行软件与计算科学实验室， 北京 100190
  • 收稿日期:2026-06-30 出版日期:2026-10-20 发布日期:2026-10-20
  • 通讯作者: *王宇鹏（E-mail: yupeng@iscas.ac.cn）；曹建文（E-mail: jianwen@iscas.ac.cn）
  • 作者简介:王宇鹏，中国科学院软件研究所，助理研究员，主要研究方向为高性能科学计算、并行算法优化与国产异构平台适配。
    本文承担工作为划分与通信实验、DCU 求解实现、算例验证和论文修改。
    WANG Yupeng is a corresponding author and assistant researcher at ISCAS. His interests include scientific computing, parallel algorithms optimization, and adaptation to domestic heterogeneous platforms.
    In this paper, he is responsible for partitioning and communication experiments, DCU implementation, validation, and revision. E-mail: yupeng@iscas.ac.cn|王宇鹏，中国科学院软件研究所，助理研究员，主要研究方向为高性能科学计算、并行算法优化与国产异构平台适配。
    本文承担工作为划分与通信实验、DCU 求解实现、算例验证和论文修改。
    WANG Yupeng is a corresponding author and assistant researcher at ISCAS. His interests include scientific computing, parallel algorithms optimization, and adaptation to domestic heterogeneous platforms.
    In this paper, he is responsible for partitioning and communication experiments, DCU implementation, validation, and revision. E-mail: yupeng@iscas.ac.cn|曹建文，中国科学院软件研究所，研究员，博士生导师，主要研究方向为并行算法、高性能计算与非线性求解。
    本文承担工作为研究指导、技术把关、实验论证和论文修改。
    CAO Jianwen is a corresponding author, researcher, and doctoral supervisor at ISCAS. His interests include parallel algorithms, high-performance computing, and nonlinear solvers.
    In this paper, he is responsible for supervision, technical review, experimental validation, and revision. E-mail: jianwen@iscas.ac.cn
  • 基金资助:
    国家重点研发计划(2023YFB3002901);中国科学院软件研究所基础研究项目(ISCAS-JCZD-202402)

Key Methods for Efficient Black-Oil Simulation on Domestic DCU

WU Xuesong(),WANG Yupeng*(),CAO Jianwen*(),LU Yanan   

  1. Lab of Parallel Software and Computational Science, Institute of Software, Chinese Academy of Sciences, Beijing 100190, China
  • Received:2026-06-30 Online:2026-10-20 Published:2026-10-20

RichHTML

0

PDF

5

摘要：

【目的】面向含千口井、千万至亿级网格的复杂角点黑油模型，研究多井工况下保持流动拓扑和近井耦合的并行划分方法，以及面向国产深度计算单元（Deep Computing Unit，DCU）的全隐式异构求解机制。【方法】构造包含笛卡尔面连接、尖灭处理（PINCH）/非邻接连接（Non-Neighboring Connection，NNC）、断层侧向搭接和井射孔内聚边的物理连通图，生成井亲和分块。求解侧围绕约束压力残差（Constrained Pressure Residual，CPR）预条件框架组织 DCU 端细网格稀疏操作、重叠区交换、油藏侧装配、收敛归约和矩阵填充。【结果】以千万级网格、千口井油田算例的64分块为例，井亲和分块在井事件步将全局归约（Allreduce）同步通信较结构化划分降低约34%~43%，内聚边方法的累计Allreduce与MPI 等待时间最低；在亿级网格400分块算例上，固定400路分解和相同MPI 布局时，DCU设备侧执行相对单线程CPU进程路径取得约6.6倍全程加速，稳态步均值约5.66倍，注水事件步约3~10 倍。【结论】面向真实流动拓扑和井射孔耦合的井亲和分块可在负载均衡基本保持时降低井事件全局同步通信，但会增加子域包围盒和点对点（Point-to-Point，P2P）通信代价；面向 DCU的异构求解在完整报告步层面取得稳定收益，但逐单元数值对拍、强非线性事件步、首步预热和更大规模完整时程仍需进一步验证。

关键词: 黑油模型, 油藏数值模拟, 井亲和, 国产DCU, 异构计算

Abstract:

[Objective] This paper studies scalable black-oil simulation on domestic deep computing units (DCUs) for complex real corner-point models with about one thousand wells and tens to hundreds of millions of cells, focusing on partitioning that preserves flow topology and near-well coupling for multi-well operations, as well as heterogeneous fully implicit solution on domestic DCUs. [Methods] A physical-connectivity graph containing Cartesian-face, pinchout/non-neighboring-connection (NNC), fault-juxtaposition and well-cohesion edges is constructed to generate well-affinity partitions. Around the constrained-pressure-residual (CPR) preconditioning framework, fine-grid sparse operations, halo exchange, reservoir-side assembly, convergence reduction, and matrix filling are organized for DCU execution. [Results] On a ten-million-cell, thousand-well field case partitioned into 64 subdomains, well-affinity partitioning reduces Allreduce synchronization in well-event steps by about 34%-43% relative to structured partitioning, and the cohesion-edge variant gives the lowest cumulative Allreduce and MPI wait time. On a hundred-million-cell case partitioned into 400 subdomains, under a fixed 400-way decomposition and identical MPI layout, DCU-side execution achieves about 6.6×full-run speedup over a single-threaded CPU-rank path; steady-step execution reaches about 5.66×speedup and water-injection event steps reach about 3-10×speedup. [Conclusions] Well-affinity partitioning based on real flow topology and perforation coupling reduces global synchronization in well-event-dominated operation while increasing bounding-box and point-to-point communication costs. The DCU-oriented heterogeneous solve shows stable gains across complete report steps, while cell-wise numerical cross-checks, strongly nonlinear event steps, first-step warm-up cost, and complete full-schedule runs at larger scales require further validation.

Key words: black-oil model, reservoir simulation, well affinity, domestic DCU, heterogeneous computing