EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Finite-difference-informed graph network for solving steady-state incompressible flows on block-structured grids

Yiye Zou, Tianyu Li, Lin Lu, Jingyu Wang, Shufan Zou, Laiping Zhang, Xiaogang Deng

Physics of Fluids · 01.10.2024 · 36 (10)

DOI: 10.1063/5.0228104 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Physics of Fluids, online erschienen am 04.10.2024. Verfasst von Yiye Zou und 6 weiteren Autor:innen. Die Arbeit wurde bisher 14-mal zitiert (Crossref) und nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National key project of China, Sichuan Science and Technology Program.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Advances in deep learning have enabled physics-informed neural networks to solve partial differential equations. Numerical differentiation using the finite-difference (FD) method is efficient in physics-constrained designs, even in parameterized settings. In traditional computational fluid dynamics (CFD), body-fitted block-structured grids are often employed for complex flow cases when obtaining FD solutions. However, convolution operators in convolutional neural networks for FD are typically limited to single-block grids. To address this issue, graphs and graph networks are used to learn flow representations across multi-block-structured grids. A graph convolution-based FD method (GC-FDM) is proposed to train graph networks in a label-free physics-constrained manner, enabling differentiable FD operations on unstructured graph outputs. To demonstrate model performance from single- to multi-block-structured grids, the parameterized steady incompressible Navier–Stokes equations are solved for a lid-driven cavity flow and the flows around single and double circular cylinder configurations. When compared to a CFD solver under various boundary conditions, the proposed method achieves a relative error in velocity field predictions in the order of 10−3. Furthermore, the proposed method reduces training costs by approximately 20% compared to a physics-informed neural network. To further verify the effectiveness of GC-FDM in multi-block processing, a 30P30N airfoil geometry is considered, and the predicted results are reasonably compared with those given by CFD. Finally, the applicability of GC-FDM to a three-dimensional (3D) case is tested using a 3D cavity geometry.

Förderung

  • National key project of China
  • Sichuan Science and Technology Program

Zitieren

APA

Zou, Y., Li, T., Lu, L., Wang, J., Zou, S., Zhang, L., & Deng, X. (2024). Finite-difference-informed graph network for solving steady-state incompressible flows on block-structured grids. Physics of Fluids, 36(10). https://doi.org/10.1063/5.0228104

Vancouver

Zou Y, Li T, Lu L, Wang J, Zou S, Zhang L, et al. Finite-difference-informed graph network for solving steady-state incompressible flows on block-structured grids. Physics of Fluids. 2024;36(10). doi:10.1063/5.0228104

BibTeX
@article{Zou2024,
  author = {Zou, Yiye and Li, Tianyu and Lu, Lin and Wang, Jingyu and Zou, Shufan and Zhang, Laiping and Deng, Xiaogang},
  title = {Finite-difference-informed graph network for solving steady-state incompressible flows on block-structured grids},
  journal = {Physics of Fluids},
  year = {2024},
  volume = {36},
  number = {10},
  doi = {10.1063/5.0228104}
}
Kostenlos starten →