FLOW-GLIDE: Global–Local Interleaved Dynamics Estimator for Flow Field Prediction
Source abstract
Abstract
Accurate prediction of the flow field is crucial to evaluating the aerodynamic performance of an aircraft. While traditional computational fluid dynamics (CFD) methods solve the governing equations to capture both global flow structures and localized gradients, they are computationally intensive. Deep learning-based surrogate models offer a promising alternative, yet often struggle to simultaneously model long-range dependencies and near-wall flow gradients with sufficient fidelity. To address this challenge, this paper introduces the Message-passing And Global-attention block (MAG-BLOCK), a graph neural network module that combines local message passing with global self-attention mechanisms to jointly learn fine-scale features and large-scale flow patterns. Building on MAG-BLOCK, we propose FLOW-GLIDE, a cross-architecture deep learning framework that learns a mapping from initial conditions to steady-state flow fields in a latent space. Evaluated on the AirfRANS dataset, FLOW-GLIDE outperforms existing models on key performance metrics. Specifically, it reduces the error in the volumetric flow field by 62% and surface pressure prediction by 82% compared to the state-of-the-art.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Fluid Dynamics and Turbulent Flows · Meteorological Phenomena and Simulations
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
College of Computer Science, Sichuan University, Chengdu 610065, China
National Key Laboratory of Fundamental Algorithms and Models for Engineering Simulation, Sichuan University, Chengdu 610207, China
School of Aeronautics and Astronautics, Sichuan University, Chengdu 610200, China
Funding
the National Key Project of China · the Sichuan Science and Technology Program
References (45)
Show source-provided references
1. Wandel, N., Weinmann, M., and Klein, R. (2020). Learning Incompressible Fluid Dynamics from Scratch–Towards Fast, Differentiable Fluid Models that Generalize. arXiv.
2. Deep learning architecture for sparse and noisy turbulent flow data10.1063/5.0200167 ↗
3. A deep-learning model for predicting spatiotemporal evolution in reactive fluidized bed reactor10.1016/j.renene.2024.120245 ↗
4. Performance and accuracy assessments of an incompressible fluid solver coupled with a deep convolutional neural network10.1017/dce.2022.2 ↗
5. Prediction of aerodynamic flow fields using convolutional neural networks10.1007/s00466-019-01740-0 ↗
6. Convolutional neural networks for fluid flow analysis: Toward effective metamodeling and low dimensionalization10.1007/s00162-021-00580-0 ↗
7. Time-variant prediction of flow over an airfoil using deep neural network10.1063/5.0022222 ↗
8. Iakovlev, V., Heinonen, M., and Lähdesmäki, H. (2020). Learning continuous-time pdes from sparse data with graph neural networks. arXiv.
9. Sanchez-Gonzalez, A., Godwin, J., Pfaff, T., Ying, R., Leskovec, J., and Battaglia, P. (2020, January 13–18). Learning to simulate complex physics with graph networks. Proceedings of the International Conference on Machine Learning, Virtual.
10. A novel graph modeling method for GNN-based hypersonic aircraft flow field reconstruction
11. An implicit GNN solver for Poisson-like problems10.1016/j.camwa.2024.10.036 ↗
12. Unsupervised learning with physics informed graph networks for partial differential equations10.1007/s10489-025-06479-1 ↗
13. Han, X., Gao, H., Pfaff, T., Wang, J.X., and Liu, L.P. (2022). Predicting physics in mesh-reduced space with temporal attention. arXiv.
14. Graph neural networks for laminar flow prediction around random two-dimensional shapes10.1063/5.0064108 ↗
15. Simulation of unsteady flow around bluff bodies using knowledge-enhanced convolutional neural network10.1016/j.jweia.2023.105405 ↗
16. Lee, S., and You, D. (2019). Mechanisms of a convolutional neural network for learning three-dimensional unsteady wake flow. arXiv.
17. Airfoil design and surrogate modeling for performance prediction based on deep learning method10.1063/5.0075784 ↗
18. Pfaff, T., Fortunato, M., Sanchez-Gonzalez, A., and Battaglia, P. (2020, January 30). Learning mesh-based simulation with graph networks. Proceedings of the International Conference on Learning Representations, Addis Ababa, Ethiopia.
19. Predicting unsteady incompressible fluid dynamics with finite volume informed neural network10.1063/5.0197425 ↗
20. Finite-difference-informed graph network for solving steady-state incompressible flows on block-structured grids10.1063/5.0228104 ↗
21. Zhang, M., and Cao, J. (2024, January 28–31). Flow Field Prediction of Variable Geometry using a Novel Deep CNN Model with Dual-Feature of Flow Regions. Proceedings of the 2024 43rd Chinese Control Conference (CCC), Kunming, China.10.23919/CCC63176.2024.10661649 ↗
22. Towards high-accuracy deep learning inference of compressible flows over aerofoils10.1016/j.compfluid.2022.105707 ↗
23. Deep learning-based predictive modeling of transonic flow over an airfoil10.1063/5.0238413 ↗
24. Fortunato, M., Pfaff, T., Wirnsberger, P., Pritzel, A., and Battaglia, P. (2022). Multiscale meshgraphnets. arXiv.
25. Gao, H., and Ji, S. (2019, January 9–15). Graph u-nets. Proceedings of the International Conference on Machine Learning, Long Beach, CA, USA.
26. Attention is all you need
27. Team, G., Riviere, M., Pathak, S., Sessa, P.G., Hardin, C., Bhupatiraju, S., Hussenot, L., Mesnard, T., Shahriari, B., and Ramé, A. (2024). Gemma 2: Improving open language models at a practical size. arXiv.
28. Team, G., Anil, R., Borgeaud, S., Alayrac, J.B., Yu, J., Soricut, R., Schalkwyk, J., Dai, A.M., Hauth, A., and Millican, K. (2023). Gemini: A family of highly capable multimodal models. arXiv.
29. Gu, A., and Dao, T. (2023). Mamba: Linear-time sequence modeling with selective state spaces. arXiv.
30. Wu, H., Luo, H., Wang, H., Wang, J., and Long, M. (2024). Transolver: A fast transformer solver for pdes on general geometries. arXiv.
31. Luo, H., Wu, H., Zhou, H., Xing, L., Di, Y., Wang, J., and Long, M. (2025). Transolver++: An Accurate Neural Solver for PDEs on Million-Scale Geometries. arXiv.
32. Zhou, H., Ma, Y., Wu, H., Wang, H., and Long, M. (2024). Unisolver: PDE-conditional transformers are universal PDE solvers. arXiv.
33. Xiao, L., Zhang, M., and Chang, X. (2024). Learning Airfoil Flow Field Representation via Geometric Attention Neural Field. Appl. Sci., 14.10.3390/app142210685 ↗
34. Universal physics transformers: A framework for efficiently scaling neural operators
35. A deep learning approach for efficiently and accurately evaluating the flow field of supercritical airfoils10.1016/j.compfluid.2019.104393 ↗
36. Understanding the effective receptive field in deep convolutional neural networks
37. Alon, U., and Yahav, E. (2020). On the bottleneck of graph neural networks and its practical implications. arXiv.
38. Chen, D., Lin, Y., Li, W., Li, P., Zhou, J., and Sun, X. (2020, January 7–12). Measuring and relieving the over-smoothing problem for graph neural networks from the topological view. Proceedings of the AAAI Conference on Artificial Intelligence, New York, NY, USA.10.1609/aaai.v34i04.5747 ↗
39. Airfrans: High fidelity computational fluid dynamics dataset for approximating reynolds-averaged navier–stokes solutions
40. Inductive representation learning on large graphs
41. Qi, C.R., Su, H., Mo, K., and Guibas, L.J. (2017, January 21–26). Pointnet: Deep learning on point sets for 3d classification and segmentation. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, USA.
42. Anandkumar, A., Azizzadenesheli, K., Bhattacharya, K., Kovachki, N., Li, Z., Liu, B., and Stuart, A. (2020, January 26–30). Neural operator: Graph kernel network for partial differential equations. Proceedings of the ICLR 2020 Workshop on Integration of Deep Neural Models and Differential Equations, Addis Ababa, Ethiopia.
43. Hao, Z., Wang, Z., Su, H., Ying, C., Dong, Y., Liu, S., Cheng, Z., Song, J., and Zhu, J. (2023, January 23–29). Gnot: A general neural operator transformer for operator learning. Proceedings of the International Conference on Machine Learning, PMLR, Honolulu, HI, USA.
44. Geometry-informed neural operator for large-scale 3d pdes
45. Choose a transformer: Fourier or galerkin