Quantifying model form uncertainty in Reynolds-averaged turbulence models with Bayesian deep neural networks

DOI10.1016/j.jcp.2019.01.021zbMath1451.76059arXiv1807.02901OpenAlexW2864723431WikidataQ128482728 ScholiaQ128482728MaRDI QIDQ2214626

Nicholas Geneva, Nicholas Zabaras

Publication date: 10 December 2020

Published in: Journal of Computational Physics (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1807.02901

zbMATH Keywords

turbulenceBayesianuncertainty quantificationdeep neural networksmodel form uncertaintyReynolds-averaged Navier-Stokes equations (RANS)

Mathematics Subject Classification ID

Artificial neural networks and deep learning (68T07) Statistical turbulence modeling (76F55) Stochastic analysis applied to problems in fluid mechanics (76M35) Statistical solutions of Navier-Stokes and related equations (76D06) Neural nets and related approaches to inference from stochastic processes (62M45)

Related Items

Solving inverse problems using conditional invertible neural networks ⋮ Data-driven discovery of coarse-grained equations ⋮ Uncertainty quantification for data-driven turbulence modelling with Mondrian forests ⋮ State estimation with limited sensors -- a deep learning based approach ⋮ Data-driven modelling of nonlinear spatio-temporal fluid flows using a deep convolutional generative adversarial network ⋮ Feature importance in neural networks as a means of interpretation for data-driven turbulence models ⋮ Parameter estimation with the Markov chain Monte Carlo method aided by evolutionary neural networks in a water hammer model ⋮ On the improvement of the extrapolation capability of an iterative machine-learning based RANS framework ⋮ Numerical Analysis for Convergence of a Sample-Wise Backpropagation Method for Training Stochastic Neural Networks ⋮ Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data ⋮ Modeling the dynamics of PDE systems with physics-constrained deep auto-regressive networks ⋮ A turbulent eddy-viscosity surrogate modeling framework for Reynolds-averaged Navier-Stokes simulations ⋮ Enforcing Imprecise Constraints on Generative Adversarial Networks for Emulating Physical Systems ⋮ Reynolds-averaged Navier–Stokes equations with explicit data-driven Reynolds stress closure can be ill-conditioned ⋮ Mosaic flows: a transferable deep learning framework for solving PDEs on unseen domains ⋮ A Bayesian multiscale deep learning framework for flows in random media

Uses Software

• OpenFOAM
• Turbulence Modeling
• Adam
• PRMLT
• Spalart-Allmaras

Cites Work

• Unnamed Item
• Unnamed Item
• Quantifying and reducing model-form uncertainties in Reynolds-averaged Navier-Stokes simulations: a data-driven, physics-informed Bayesian approach
• Machine learning strategies for systems with invariance properties
• Solution of the implicitly discretized fluid flow equations by operator- splitting
• A methodology to evaluate statistical errors in DNS data of plane channel flows
• Bayesian deep convolutional encoder-decoder networks for surrogate modeling and uncertainty quantification
• Neural network modeling for near wall turbulent flow.
• Reynolds number dependence of mean flow structure in square duct turbulence
• Predictions of Channel and Boundary-Layer Flows with a Low-Reynolds-Number Turbulence Model
• A more general effective-viscosity hypothesis
• Simulation of vortex shedding past a square cylinder with different turbulence models
• Is the steady viscous incompressible two‐dimensional flow over a backward‐facing step at Re = 800 stable?
• 10.1162/15324430152748236
• High-Lift Aerodynamic Computations with One- and Two-Equation Turbulence Models
• Turbulent Flows
• Reynolds averaged turbulence modelling using deep neural networks with embedded invariance