A Point-Cloud Deep Learning Framework for Prediction of Fluid Flow Fields on Irregular Geometries

Abstract

We present a novel deep learning framework for flow field predictions in irregular domains when the solution is a function of the geometry of either the domain or objects inside the domain. Grid vertices in a computational fluid dynamics (CFD) domain are viewed as point clouds and used as inputs to a neural network based on the PointNet architecture, which learns an end-to-end mapping between spatial positions and CFD quantities. Using our approach, (i) the network inherits desirable features of unstructured meshes (e.g., fine and coarse point spacing near the object surface and in the far field, respectively), which minimizes network training cost; (ii) object geometry is accurately represented through vertices located on object boundaries, which maintains boundary smoothness and allows the network to detect small changes between geometries; and (iii) no data interpolation is utilized for creating training data; thus accuracy of the CFD data is preserved. None of these features are achievable by extant methods based on projecting scattered CFD data into Cartesian grids and then using regular convolutional neural networks. Incompressible laminar steady flow past a cylinder with various shapes for its cross section is considered. The mass and momentum of predicted fields are conserved. We test the generalizability of our network by predicting the flow around multiple objects as well as an airfoil, even though only single objects and no airfoils are observed during training. The network predicts the flow fields hundreds of times faster than our conventional CFD solver, while maintaining excellent to reasonable accuracy.

References in corpus (21)

• Adam: A Method for Stochastic Optimization
• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Turbulence Modeling in the Age of Data
• Super-resolution reconstruction of turbulent flows with machine learning
• Physics-informed neural networks for inverse problems in nano-optics and metamaterials
• Deep Learning Methods for Reynolds-Averaged Navier-Stokes Simulations of Airfoil Flows
• On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima
• Prediction of Aerodynamic Flow Fields Using Convolutional Neural Networks
• Deep Neural Networks for Data-Driven Turbulence Models
• Sub-grid modelling for two-dimensional turbulence using neural networks
• Accelerating Eulerian Fluid Simulation With Convolutional Networks
• Deep Fluids: A Generative Network for Parameterized Fluid Simulations
• Stress Field Prediction in Cantilevered Structures Using Convolutional Neural Networks
• Training convolutional neural networks to estimate turbulent sub-grid scale reaction rates
• A new prediction method of unsteady wake flow by the hybrid deep neural network
• A neural network approach for the blind deconvolution of turbulent flows
• Assessment of unsteady flow predictions using hybrid deep learning based reduced order models
• An Efficient Deep Learning Technique for the Navier-Stokes Equations: Application to Unsteady Wake Flow Dynamics
• Matrix-free multigrid block-preconditioners for higher order Discontinuous Galerkin discretisations
• A two-dimensional analytical model of vertical water entry for asymmetric bodies with flow separation
• FlowNet3D: Learning Scene Flow in 3D Point Clouds

Cited by in corpus (26)

• Machine Learning in Aerodynamic Shape Optimization
• Reduced-order modeling of advection-dominated systems with recurrent neural networks and convolutional autoencoders
• Convolutional neural network and long short-term memory based reduced order surrogate for minimal turbulent channel flow
• Super-Resolution Analysis via Machine Learning: A Survey for Fluid Flows
• Physics-informed PointNet: A deep learning solver for steady-state incompressible flows and thermal fields on multiple sets of irregular geometries
• Convolutional neural networks for fluid flow analysis: toward effective metamodeling and low-dimensionalization
• Geom-DeepONet: A Point-cloud-based Deep Operator Network for Field Predictions on 3D Parameterized Geometries
• Experimental velocity data estimation for imperfect particle images using machine learning
• Deep reinforcement learning for large-eddy simulation modeling in wall-bounded turbulence
• A Survey on Physics Informed Reinforcement Learning: Review and Open Problems
• Point-Cloud Deep Learning of Porous Media for Permeability Prediction
• Combining physics-based and data-driven models: advancing the frontiers of research with Scientific Machine Learning
• A robust single-pixel particle image velocimetry based on fully convolutional networks with cross-correlation embedded
• Flow Completion Network: Inferring the Fluid Dynamics from Incomplete Flow Information using Graph Neural Networks
• Reconstructing three-dimensional bluff body wake from sectional flow fields with convolutional neural networks
• Physics Informed Neural Networks for Modeling of 3D Flow-Thermal Problems with Sparse Domain Data
• C(NN)FD -- a deep learning framework for turbomachinery CFD analysis
• Multi-Fidelity Machine Learning Applied to Steady Fluid Flows
• Machine learning study to identify collective flow in small and large colliding systems
• A panoramic aerodynamic performance prediction method for turbomachinery cascades using transformer-enhanced neural operator
• Scalar Field Prediction on Meshes Using Interpolated Multi-Resolution Convolutional Neural Networks
• Multi-GPU Approach for Training of Graph ML Models on large CFD Meshes
• Point-DeepONet: Predicting Nonlinear Fields on Non-Parametric Geometries under Variable Load Conditions
• Deep learning modelling of manufacturing and build variations on multi-stage axial compressors aerodynamics
• Boundary-layer transition in the age of data: from a comprehensive dataset to fine-grained prediction
• Spatio-Temporal Uncertainty-Modulated Physics-Informed Neural Networks for Solving Hyperbolic Conservation Laws with Strong Shocks