Machine learning for RANS turbulence modeling of variable property flows

DOI10.1016/J.COMPFLUID.2023.105835MaRDI QIDQ6158535zbMATHFDO

Authors Rafael Diez Sanhueza, Stephan H. H. J. Smit, Jurriaan W. R. Peeters, Rene Pecnik

Publication date 20 June 2023

Published in Computers and Fluids (Search for Journal in Brave)

Copyright license Creative Commons Attribution 4.0 International

Full work available at URL https://arxiv.org/abs/2210.15384

zbMATH Keywords

machine learningturbulence modelingvariable properties

Mathematics Subject Classification ID

Direct numerical and large eddy simulation of turbulence (76F65) Basic methods in fluid mechanics (76M99)

Abstract: This paper presents a machine learning methodology to improve the predictions of traditional RANS turbulence models in channel flows subject to strong variations in their thermophysical properties. The developed formulation contains several improvements over the existing Field Inversion Machine Learning (FIML) frameworks described in the literature, as well as the derivation of a new modelling technique. We first showcase the use of efficient optimization routines to automatize the process of field inversion in the context of CFD, combined with the use of symbolic algebra solvers to generate sparse-efficient algebraic formulas to comply with the discrete adjoint method. The proposed neural network architecture is characterized by the use of an initial layer of logarithmic neurons followed by hyperbolic tangent neurons, which proves numerically stable. The machine learning predictions are then corrected using a novel weighted relaxation factor methodology, that recovers valuable information from otherwise spurious predictions. The study uses the K-fold cross-validation technique, which is beneficial for small datasets. The results show that the machine learning model acts as an excellent non-linear interpolator for DNS cases well-represented in the training set, and that moderate improvement margins are obtained for sparser DNS cases. It is concluded that the developed machine learning methodology corresponds to a valid alternative to improve RANS turbulence models in flows with strong variations in their thermophysical properties without introducing prior modeling assumptions into the system.

Recommendations

• RANS turbulence model development using CFD-driven machine learning
• A turbulent eddy-viscosity surrogate modeling framework for Reynolds-averaged Navier-Stokes simulations
• Data-driven modelling of the Reynolds stress tensor using random forests with invariance
• Quantifying model form uncertainty in Reynolds-averaged turbulence models with Bayesian deep neural networks
• On the Generalizability of Machine-Learning-Assisted Anisotropy Mappings for Predictive Turbulence Modelling

Cites work

• scientific article; zbMATH DE number 4201669 (Why is no real title available?)
• A numerical study of turbulent supersonic isothermal-wall channel flow
• A paradigm for data-driven predictive modeling using field inversion and machine learning
• Approximation by superpositions of a sigmoidal function
• Assessment of performance of turbulence models in predicting supercritical pressure heat transfer in a vertical tube
• Compressible turbulent channel flows: DNS results and modelling
• Machine learning for fluid mechanics
• Neural network modeling for near wall turbulent flow.
• Reynolds averaged turbulence modelling using deep neural networks with embedded invariance
• Scaling and modelling of turbulence in variable property channel flows
• Statistical theory and modeling for turbulent flows.
• The influence of near-wall density and viscosity gradients on turbulence in channel flows
• The turbulent flows of supercritical fluids with heat transfer
• Turbulence attenuation in simultaneously heated and cooled annular flows at supercritical pressure
• Turbulent fluctuations above the buffer layer of wall-bounded flows

Cited in

(8)

• Physics-informed machine learning for surrogate modeling of wind pressure and optimization of pressure sensor placement
• The use of the Reynolds force vector in a physics informed machine learning approach for predictive turbulence modeling
• RANS turbulence model development using CFD-driven machine learning
• Field inversion for data-augmented RANS modelling in turbomachinery flows
• On the improvement of the extrapolation capability of an iterative machine-learning based RANS framework
• Dynamic calibration of differential equations using machine learning, with application to turbulence models
• High Reynolds number airfoil turbulence modeling method based on machine learning technique
• On the Generalizability of Machine-Learning-Assisted Anisotropy Mappings for Predictive Turbulence Modelling

This page was built for publication: Machine learning for RANS turbulence modeling of variable property flows

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6158535)