AFBench: A Large-scale Benchmark for Airfoil Design

🚀 Abstract

Data-driven generative models have emerged as promising approaches towards achieving efficient mechanical inverse design. However, due to prohibitively high cost in time and money, there is still lack of open-source and large-scale benchmarks in this field. It is mainly the case for airfoil inverse design, which requires to generate and edit diverse geometric-qualified and aerodynamic-qualified airfoils following the multimodal instructions,i.e., dragging points and physical parameters.

This paper presents the open-source endeavors in airfoil inverse design,AFBench, including a large-scale dataset with 200 thousand airfoils and high-quality aerodynamic and geometric labels, two novel and practical airfoil inverse design tasks, i.e., conditional generation on multimodal physical parameters, controllable editing, and comprehensive metrics to evaluate various existing airfoil inverse design methods.

Our aim is to establish AFBench as an ecosystem for training and evaluating airfoil inverse design methods, with a specific focus on data-driven controllable inverse design models by multimodal instructions capable of bridging the gap between ideas and execution, the academic research and industrial applications. We have provided baseline models, comprehensive experimental observations, and analysis to accelerate future research. Our baseline model is trained on an RTX 3090 GPU within 16 hours. The codebase, datasets and benchmarks will be available at https://hitcslj.github.io/afbench/.

🔥 Method

In this paper, We propose the use of generative methods for two key tasks in airfoil design: airfoil generation and editing. Additionally, we establish a comprehensive benchmark dataset for evaluating these methods, leveraging the proposed airfoil dataset to provide a standardized evaluation framework.

Automatic Data Engine

In this paper, We propose a comprehensive and diverse airfoil dataset to tackle the challenge of scaling law in Airfoil Generative Design. This dataset encompasses a wide range of airfoil shapes and is accompanied by detailed annotation labels, providing a valuable resource for training and evaluating generative models in this field.

Benchmarking Setup

In this paper, We construct and open-source a codebase that encompasses generative methods in airfoil design, including foundational techniques such as cVAE, cGAN, cVAE-GAN, Bezeir-cGAN, as well as advanced models like PK-VAE and PK-DiT. We provide a user-friendly demo that allows for visualizing and experiencing airfoil design in real-time.

🚀 Airfoil Demo

Our Airfoil Generation and Editing Software. (a) Generating airfoils and editing their physical parameters. (b) Editing the control keypoints of the airfoil.

📌 License

The project is CC BY NC 4.0 (allowing only non-commercial use) and models trained using the dataset should not be used outside of research purposes. The checkpoints are also CC BY NC 4.0 (allowing only non-commercial use).

🔥 Citation

@misc{liu2024afbenchlargescalebenchmarkairfoil,
  title={AFBench: A Large-scale Benchmark for Airfoil Design}, 
  author={Jian Liu and Jianyu Wu and Hairun Xie and Guoqing Zhang and Jing Wang and Wei Liu and Wanli Ouyang and Junjun Jiang and Xianming Liu and Shixiang Tang and Miao Zhang},
  year={2024},
  eprint={2406.18846},
  archivePrefix={arXiv},
  primaryClass={cs.CE},
  url={https://arxiv.org/abs/2406.18846}, 
}