Direct Prediction of Steady-State Flow Fields in Meshed Domain with Graph Networks

2021/05/06 by Lukas Harsch, Harsch, Lukas, Stefan Riedelbauch +1 · 1 citation
Computer Science · Physics and Astronomy · #Artificial Intelligence (cs.AI) #Computer Graphics and Visualization Techniques #Data Visualization and Analytics #FOS: Computer and information sciences #FOS: Physical sciences #Fluid Dynamics (physics.flu-dyn) #Model Reduction and Neural Networks #cs.AI #physics.flu-dyn

paper · pdf · doi:10.48550/arxiv.2105.02575

9 pages, 4 figures

arxiv created 2021/05/06 · openalex publication_date 2021/05/06 · arxiv updated 2021/05/07 · openalex created_date 2021/05/10 · openalex updated_date 2026/07/28

Abstract

We propose a model to directly predict the steady-state flow field for a given geometry setup. The setup is an Eulerian representation of the fluid flow as a meshed domain. We introduce a graph network architecture to process the mesh-space simulation as a graph. The benefit of our model is a strong understanding of the global physical system, while being able to explore the local structure. This is essential to perform direct prediction and is thus superior to other existing methods.

Citations

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Learning to Simulate Complex Physics with Graph Networks

Cited by

• PLAID: A Unified Data Model for Machine Learning on Heterogeneous Physics Simulations

Related

• Enhancing Graph U-Nets for Mesh-Agnostic Spatio-Temporal Flow Prediction
• Predicting Transonic Flowfields in Non-Homogeneous Unstructured Grids Using Autoencoder Graph Convolutional Networks
• Multiscale Graph Neural Network for Turbulent Flow-Thermal Prediction Around a Complex-Shaped Pin-Fin
• Learning the Dynamics of Physical Systems from Sparse Observations with Finite Element Networks
• Generalization capabilities of MeshGraphNets to unseen geometries for fluid dynamics