解算器
计算科学
计算机科学
计算流体力学
雅可比矩阵与行列式
加速度
软件
物理
可扩展性
流体力学
应用数学
流量(数学)
航程(航空)
多物理
大涡模拟
不确定度量化
超级计算机
切姆金
燃烧
常微分方程
非线性系统
机械
统计物理学
化学过程
微分方程
湍流
流动化学
化学
作者
Zixin Chi,Xin Hui,Kexin Ji,Bosen Wang
出处
期刊:Physics of Fluids
[American Institute of Physics]
日期:2025-12-01
卷期号:37 (12): 125154-
被引量:3
摘要
Reactive flow simulations with detailed chemical kinetics encounter significant computational challenges due to stiff, nonlinear combustion processes. This work presents a high-performance, central processing unit accelerated lightweight chemistry solver for OpenFOAM, an open-source software for computational fluid dynamics, that integrates code-level optimizations—including vectorized reaction rate evaluation, novel Jacobian construction for elementary reactions, and optimized matrix operations—with dynamic load balancing for multi-core scalability. In addition, the solver is fully compatible with the OpenFOAM ecosystem, requiring no external libraries or mechanism-specific code generation, thereby ensuring ease of integration and broad applicability to diverse chemical mechanisms. This solver is rigorously validated based on a hierarchy of test cases, including zero-dimensional autoignition, two-dimensional counterflow diffusion flames, and large eddy simulations of turbulent jet flames. Comprehensive benchmarks against the standard OpenFOAM chemistry solver demonstrate that the proposed solver can achieve up to an order-of-magnitude acceleration across a range of stiff ordinary differential equation algorithms, tolerance settings, and chemistry mechanisms from 10 species to 255 species, meanwhile maintaining solution fidelity. Cross-architecture robustness, superior computational efficiency compared with state-of-the-art solvers (e.g., DLBFoam), and good parallel scalability have also been validated for the proposed solver. This proposed solver presents the capability of significantly extending the practical feasibility of high-fidelity simulations with detailed chemistry.
科研通智能强力驱动
Strongly Powered by AbleSci AI