DeepFlame: A deep learning empowered open-source platform for reacting flow simulations

Mao，Runze1,2; Lin，Minqi1; Zhang，Yan3,4; Zhang，Tianhan2,5; Xu，Zhi Qin John6,7; Chen，Zhi X.1,2

2023-10-01

10.1016/j.cpc.2023.108842

Computer Physics Communications   Impact Factor & Quartile

0010-4655

1879-2944

Recent developments in deep learning have brought many inspirations for the scientific computing community and it is perceived as a promising method in accelerating the computationally demanding reacting flow simulations. In this work, we introduce DeepFlame, an open-source C++ platform with the capabilities of utilising machine learning algorithms and offline-trained models to solve for reactive flows. We combine the individual strengths of the computational fluid dynamics library OpenFOAM, machine learning framework Torch, and chemical kinetics program Cantera. The complexity of cross-library function and data interfacing (the core of DeepFlame) is minimised to achieve a simple and clear workflow for code maintenance, extension and upgrading. As a demonstration, we apply our recent work on deep learning for predicting chemical kinetics (Zhang et al., 2022 [8]) to highlight the potential of machine learning in accelerating reacting flow simulation. A thorough code validation is conducted via a broad range of canonical cases to assess its accuracy and efficiency. The results demonstrate that the convection-diffusion-reaction algorithms implemented in DeepFlame are robust and accurate for both steady-state and transient processes. In addition, a number of methods aiming to further improve the computational efficiency, e.g. dynamic load balancing and adaptive mesh refinement, are explored. Their performances are also evaluated and reported. With the deep learning method implemented in this work, a speed-up of two orders of magnitude is achieved in a simple hydrogen ignition case when performed on a medium-end graphics processing unit (GPU). Further gain in computational efficiency is expected for hydrocarbon and other complex fuels. A similar level of acceleration is obtained on an AI-specific chip – deep computing unit (DCU), highlighting the potential of DeepFlame in leveraging the next-generation computing architecture and hardware. Program summary: Program Title: DeepFlame CPC Library link to program files: https://doi.org/10.17632/3pg9xmypp3.1 Developer's repository link: https://github.com/deepmodeling/deepflame-dev Licensing provisions: GPLv3 Programming language: C++ Nature of problem: Solving chemically reacting flows with direct (quasi-direct) simulation methods is usually troubled by the following problems: 1. as the widely-used computational fluid dynamics (CFD) toolbox, OpenFOAM features limited ODE solvers for chemistry and oversimplified transport models, yielding non-negligible errors in simulation results; 2. the chemical source term evaluation is the most computationally expensive and usually accounts for more than 80% of total computing time. Solution method: An open-source platform bringing together the individual strengths of OpenFOAM, Cantera and PyTorch libraries is built in this study. In the present implementation, CVODE solvers, detailed transport models and deep learning algorithms are all adopted to assist the simulation of reacting flow. Note that here machine learning is introduced in combination with heterogeneous computing to accelerate the most demanding solving procedure for chemical source term evaluation.

Chemical kinetics Compressible reacting flow Computational fluid dynamic High performance computing Machine learning

SCI

English

Others

National Natural Science Foundation of China[52276096];National Natural Science Foundation of China[92270203];

Computer Science ; Physics

Computer Science, Interdisciplinary Applications ; Physics, Mathematical

WOS:001048967000001

ELSEVIER

PHYSICS

2-s2.0-85165471884

Scopus

1.State Key Laboratory of Turbulence and Complex Systems,College of Engineering,Peking University,Beijing,100871,China
2.AI for Science Institute (AISI),Beijing,100080,China
3.CAEP Software Center for High Performance Numerical Simulation,Beijing,100088,China
4.Institute of Applied Physics and Computational Mathematics,Beijing,100088,China
5.Department of Mechanics and Aerospace Engineering,SUSTech,Shenzhen,518055,China
6.Institute of Natural Sciences,School of Mathematical Sciences,Shanghai Jiao Tong University,Shanghai,200240,China
7.MOE-LSC,Qing Yuan Research Institute,Shanghai Jiao Tong University,Shanghai,200240,China

Mao，Runze,Lin，Minqi,Zhang，Yan,et al. DeepFlame: A deep learning empowered open-source platform for reacting flow simulations[J]. Computer Physics Communications,2023,291.

Mao，Runze,Lin，Minqi,Zhang，Yan,Zhang，Tianhan,Xu，Zhi Qin John,&Chen，Zhi X..(2023).DeepFlame: A deep learning empowered open-source platform for reacting flow simulations.Computer Physics Communications,291.

Mao，Runze,et al."DeepFlame: A deep learning empowered open-source platform for reacting flow simulations".Computer Physics Communications 291(2023).