SuperWing: a comprehensive transonic wing dataset for data-driven aerodynamic design

2025-12-16

AI Summary

To address the scarcity of large-scale, high-diversity benchmark datasets for aerodynamic prediction of three-dimensional wings, this work introduces the first large-scale swept-wing aerodynamic dataset covering the transonic flight envelope—comprising 4,239 parametric airfoils and 28,856 RANS flow-field solutions. We propose a novel spanwise-varying cross-section–twist–dihedral joint parameterization method that balances conciseness and expressiveness, departing from conventional baseline-perturbation paradigms. Leveraging this dataset, we design a Transformer-based surface flow-field prediction model, achieving a mean drag coefficient error of 2.5 counts on held-out test samples. Crucially, we demonstrate, for the first time, zero-shot transferability to complex configurations—including DLR-F6 and NASA Common Research Model—without fine-tuning, thereby significantly enhancing model generality and physical fidelity.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Application Domains: Mobility, Driving & Flight
• Natural Language Processing: (Large) Language Models

Application Category

• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Machine-learning surrogate models have shown promise in accelerating aerodynamic design, yet progress toward generalizable predictors for three-dimensional wings has been limited by the scarcity and restricted diversity of existing datasets. Here, we present SuperWing, a comprehensive open dataset of transonic swept-wing aerodynamics comprising 4,239 parameterized wing geometries and 28,856 Reynolds-averaged Navier-Stokes flow field solutions. The wing shapes in the dataset are generated using a simplified yet expressive geometry parameterization that incorporates spanwise variations in airfoil shape, twist, and dihedral, allowing for an enhanced diversity without relying on perturbations of a baseline wing. All shapes are simulated under a broad range of Mach numbers and angles of attack covering the typical flight envelope. To demonstrate the dataset's utility, we benchmark two state-of-the-art Transformers that accurately predict surface flow and achieve a 2.5 drag-count error on held-out samples. Models pretrained on SuperWing further exhibit strong zero-shot generalization to complex benchmark wings such as DLR-F6 and NASA CRM, underscoring the dataset's diversity and potential for practical usage.

Problem

• Develops a comprehensive dataset for 3D transonic wing aerodynamics
• Addresses limited diversity in existing datasets for machine learning models
• Enables accurate flow prediction and generalization to complex wing designs

Innovation

• Comprehensive transonic wing dataset with 4,239 geometries
• Simplified geometry parameterization for enhanced shape diversity
• Transformer models achieve accurate flow prediction and generalization