Physics-informed machine learning approach for reconstructing Reynolds stress modeling discrepancies based on DNS data

2016/06/30 by Jian-Xun Wang, Jin-Long Wu, Heng Xiao · 7 citations
Engineering · Physics and Astronomy · #Flow (mathematics) #Fluid Dynamics and Turbulent Flows #Fluid Dynamics and Vibration Analysis #Model Reduction and Neural Networks #Online machine learning #Reynolds number #Reynolds stress #Reynolds-averaged Navier–Stokes equations #Structured prediction #physics.flu-dyn

paper · pdf · doi:10.1103/physrevfluids.2.034603

published as Phys. Rev. Fluids 2, 034603 (2017) · 36 pages, 1 figures

arxiv created 2017/02/27 · openalex created_date 2017/03/16 · openalex publication_date 2017/03/16 · arxiv updated 2017/03/22 · openalex updated_date 2026/08/06

Abstract

We show that the discrepancies in Reynolds-averaged Navier-Stokes (RANS) modeled Reynolds stresses can be explained by mean flow features. A physics-informed machine learning framework is proposed to improve the predictive capabilities of RANS models by leveraging existing direct numerical simulations databases.

Citations

• Two-equation eddy-viscosity turbulence models for engineering applications

Cited by

• DiFVM: A Vectorized Graph-Based Finite Volume Solver for Differentiable CFD on Unstructured Meshes
• Are Discrepancies in RANS Modeled Reynolds Stresses Random?
• Node Assigned physics-informed neural networks for thermal-hydraulic system simulation: CVH/FL module
• Physics-informed machine learning improves detection of head impacts
• Invariance-embedded Machine Learning Sub-grid-scale Stress Models for Meso-scale Hurricane Boundary Layer Flow Simulation I: Model Development and a priori Studies
• Seeing permeability from images: fast prediction with convolutional neural networks
• Machine-learning-based simulation of turbulent flows over periodic hills using a hybrid U-Net and Fourier neural operator framework
• Probabilistic learning on manifolds constrained by nonlinear partial differential equations for small datasets