Deep Learning to advance the Eigenspace Perturbation Method for Turbulence Model Uncertainty Quantification

2022/02/11 by Khashayar Nobarani, Nobarani, Khashayar, Seyed Esmaeil Razavi +2
Computer Science · Decision Sciences · Engineering · Mathematics · Physics and Astronomy · #Applied mathematics #Computational Engineering #Computer science #Direct numerical simulation #Eigenvalues and eigenvectors #FOS: Computer and information sciences #Finance #Fluid Dynamics and Turbulent Flows #K-epsilon turbulence model #K-omega turbulence model #Large eddy simulation #Machine Learning (cs.LG) #Mathematics #Mechanics #Model Reduction and Neural Networks #Nuclear Engineering Thermal-Hydraulics #Perturbation (astronomy) #Physics #Probabilistic and Robust Engineering Design #Reynolds number #Reynolds stress #Reynolds stress equation model #Reynolds-averaged Navier–Stokes equations #Statistical physics #Turbulence #Turbulence modeling #and Science (cs.CE) #cs.CE #cs.LG

paper · pdf · doi:10.48550/arxiv.2202.12378

published in arXiv (Cornell University) (Cornell University)

openalex publication_date 2022/02/11 · arxiv created 2022/03/16 · arxiv updated 2022/03/17 · openalex created_date 2022/04/03 · openalex updated_date 2026/08/06

Abstract

The Reynolds Averaged Navier Stokes (RANS) models are the most common form of model in turbulence simulations. They are used to calculate Reynolds stress tensor and give robust results for engineering flows. But RANS model predictions have large error and uncertainty. In past, there has been some work towards using data-driven methods to increase their accuracy. In this work we outline a machine learning approach to aid the use of the Eigenspace Perturbation Method to predict the uncertainty in the turbulence model prediction. We use a trained neural network to predict the discrepancy in the shape of the RANS predicted Reynolds stress ellipsoid. We apply the model to a number of turbulent flows and demonstrate how the approach correctly identifies the regions in which modeling errors occur when compared to direct numerical simulation (DNS), large eddy simulation (LES) or experimental results from previous works.

Related

• Towards Physics Constrained Deep Learning Based Turbulence Model Uncertainty Quantification
• A Deep Learning Approach For Epistemic Uncertainty Quantification Of Turbulent Flow Simulations
• Convolutional Neural Networks For Turbulent Model Uncertainty Quantification
• Multi-fidelity Deep Learning-based methodology for epistemic uncertainty quantification of turbulence models
• Combination of Multi-Fidelity Data Sources For Uncertainty Quantification: A Lightweight CNN Approach