Abstract:The study of partial differential equations (PDE) through the framework of deep learning emerged a few years ago leading to the impressive approximations of simple dynamics. Graph neural networks (GNN) turned out to be very useful in those tasks by allowing the treatment of unstructured data often encountered in the field of numerical resolutions of PDE. However, the resolutions of harder PDE such as Navier-Stokes equations are still a challenging task and most of the work done on the latter concentrate either on simulating the flow around simple geometries or on qualitative results that looks physical for design purpose. In this study, we try to leverage the work done on deep learning for PDE and GNN by proposing an adaptation of a known architecture in order to tackle the task of approximating the solution of the two-dimensional steady-state incompressible Navier-Stokes equations over different airfoil geometries. In addition to that, we test our model not only on its performance over the volume but also on its performance to approximate surface quantities such as the wall shear stress or the isostatic pressure leading to the inference of global coefficients such as the lift and the drag of our airfoil in order to allow design exploration. This work takes place in a longer project that aims to approximate three dimensional steady-state solutions over industrial geometries.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? The paper "Graph Neural Networks for Airfoil Design" aims to use Graph Neural Networks (GNN) to solve the problem of solving the two - dimensional steady - state incompressible Navier - Stokes equations on different airfoil geometries. Specifically, the author attempts to achieve this goal in the following aspects: 1. **Improve the solution accuracy**: - The goal of the paper is not only to obtain a good approximation of different fields within the volume (such as the velocity field, pressure field, turbulent viscosity field), but also to accurately approximate surface quantities, such as wall shear stress and isostatic pressure. - The accurate prediction of these surface quantities is crucial for calculating the lift coefficient and drag coefficient of the airfoil, thus supporting the exploration of airfoil design. 2. **Handle unstructured data**: - Traditional numerical methods (such as the finite element method and the finite volume method) usually need to handle unstructured grid data, which is a challenge in deep learning. GNN can effectively handle this type of unstructured data, making deep - learning methods applicable to numerical simulations in fluid mechanics. 3. **Expand to three - dimensional industrial geometries**: - The work in this paper paves the way for approximate solutions in more complex three - dimensional industrial geometry simulations. By validating the effectiveness of the method on two - dimensional airfoils, the author hopes to be able to apply it to more complex three - dimensional industrial scenarios in the future. 4. **Performance evaluation**: - The author not only evaluates the performance of the model within the volume but also tests its generalization ability on different test sets, including geometric shapes with added noise, a larger range of attack angles, and different aspect ratios. ### Summary In general, this paper attempts to improve the solution accuracy of the two - dimensional airfoil flow field by combining deep - learning and graph - neural - network methods, especially in the prediction of surface quantities. This not only helps in the design optimization of airfoils but also provides a basis for future three - dimensional industrial geometry simulations.

Graph Neural Networks for Airfoil Design

Graph Neural Networks for Aerodynamic Flow Reconstruction from Sparse Sensing

Graph Neural Networks and Differential Equations: A hybrid approach for data assimilation of fluid flows

Enhancing Data-Assimilation in CFD using Graph Neural Networks

A finite element-inspired hypergraph neural network: Application to fluid dynamics simulations

Deep Learning Methods for Reynolds-Averaged Navier-Stokes Simulations of Airfoil Flows

Graph Convolutional Multi-Mesh Autoencoder for Steady Transonic Aircraft Aerodynamics

A complete state-space solution model for inviscid flow around airfoils based on physics-informed neural networks

Graph Neural Networks for Mesh Generation and Adaptation in Structural and Fluid Mechanics

Solving Inverse Problems in Steady State Navier-Stokes Equations using Deep Neural Networks

Aero-Nef: Neural Fields for Rapid Aircraft Aerodynamics Simulations

Physics-Informed Neural Networks for Transonic Flows around an Airfoil

Neural fields for rapid aircraft aerodynamics simulations

Large-scale graph-machine-learning surrogate models for 3D-flowfield prediction in external aerodynamics

Fourier Neural Operator with Boundary Conditions for Efficient Prediction of Steady Airfoil Flows

Aerodynamic Performance Prediction for Wide-Incidence Turbines Using Graph Neural Network Models Driven by Small-Scale Experimental Data

Physics-Informed Machine Learning Using Low-Fidelity Flowfields for Inverse Airfoil Shape Design

Inverse Design of Airfoils Using Convolutional Neural Network and Deep Neural Network

A hybrid numerical methodology coupling Reduced Order Modeling and Graph Neural Networks for non-parametric geometries: applications to structural dynamics problems

Efficient aerodynamic coefficients prediction with a long sequence neural network