Recent Advances on Machine Learning for Computational Fluid Dynamics: A Survey

Haixin Wang,Yadi Cao,Zijie Huang,Yuxuan Liu,Peiyan Hu,Xiao Luo,Zezheng Song,Wanjia Zhao,Jilin Liu,Jinan Sun,Shikun Zhang,Long Wei,Yue Wang,Tailin Wu,Zhi-Ming Ma,Yizhou Sun

2024-08-22

Abstract:This paper explores the recent advancements in enhancing Computational Fluid Dynamics (CFD) tasks through Machine Learning (ML) techniques. We begin by introducing fundamental concepts, traditional methods, and benchmark datasets, then examine the various roles ML plays in improving CFD. The literature systematically reviews papers in recent five years and introduces a novel classification for forward modeling: Data-driven Surrogates, Physics-Informed Surrogates, and ML-assisted Numerical Solutions. Furthermore, we also review the latest ML methods in inverse design and control, offering a novel classification and providing an in-depth discussion. Then we highlight real-world applications of ML for CFD in critical scientific and engineering disciplines, including aerodynamics, combustion, atmosphere & ocean science, biology fluid, plasma, symbolic regression, and reduced order modeling. Besides, we identify key challenges and advocate for future research directions to address these challenges, such as multi-scale representation, physical knowledge encoding, scientific foundation model and automatic scientific discovery. This review serves as a guide for the rapidly expanding ML for CFD community, aiming to inspire insights for future advancements. We draw the conclusion that ML is poised to significantly transform CFD research by enhancing simulation accuracy, reducing computational time, and enabling more complex analyses of fluid dynamics. The paper resources can be viewed at <a class="link-external link-https" href="https://github.com/WillDreamer/Awesome-AI4CFD" rel="external noopener nofollow">this https URL</a>.

Machine Learning

What problem does this paper attempt to address?

### Problems Addressed by the Paper

This paper aims to explore the latest advancements in machine learning (ML) techniques within computational fluid dynamics (CFD) and systematically review the research work in this field over the past 5 years. Specifically, this review paper attempts to address the following key issues:

1. **Enhancing CFD Tasks**:
- Explore how machine learning techniques can improve the performance of CFD, including enhancing simulation accuracy, reducing computation time, and achieving more complex fluid dynamics analyses.

2. **Classification and Methodology**:
- Propose a novel classification system that categorizes the applications of machine learning in forward modeling into three types: Data-driven Surrogates, Physics-Informed Surrogates, and ML-assisted Numerical Solutions. Additionally, it provides a classified discussion on inverse design and control problems.

3. **Practical Applications**:
- Showcase practical application cases of machine learning in CFD, covering several key scientific and engineering fields such as aerodynamics, combustion, atmospheric and ocean sciences, biofluids, and plasma.

4. **Challenges and Future Directions**:
- Identify the key challenges currently faced in CFD research, such as multi-scale representation, encoding physical knowledge, foundational scientific models, and automated scientific discovery, and propose corresponding future research directions.

Through these efforts, the paper aims to provide guidance for the rapidly developing ML and CFD community, inspiring future innovation and development.