Unsupervised deep learning for super-resolution reconstruction of turbulence

Kim, Hyojin; Kim, Junhyuk; Won, Sungjin; Lee, Changhoon

doi:10.1017/jfm.2020.1028

Cited by 159 publications

(98 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In addition, machine learning methods have been employed for super-resolution reconstruction of turbulent flows (Maulik & San 2017;Fukami, Fukagata & Taira 2019;Kim & Lee 2020;Liu et al 2020;Yuan et al 2020;Kim et al 2021). Such reconstruction can be used to improve predictions based on coarse wall measurements in wall-bounded turbulence (Güemes et al 2021;Vinuesa & Brunton 2021).…”

Section: Introductionmentioning

confidence: 99%

Kinetic-energy-flux-constrained model using an artificial neural network for large-eddy simulation of compressible wall-bounded turbulence

Yuan

et al. 2021

J. Fluid Mech.

View full text Add to dashboard Cite

Kinetic energy flux (KEF) is an important physical quantity that characterizes cascades of kinetic energy in turbulent flows. In large-eddy simulation (LES), it is crucial for the subgrid-scale (SGS) model to accurately predict the KEF in turbulence. In this paper, we propose a new eddy-viscosity SGS model constrained by the properly modelled KEF for LES of compressible wall-bounded turbulence. The new methodology has the advantages of both accurate prediction of the KEF and strong numerical stability in LES. We can obtain an approximate KEF by the tensor-diffusivity model, which has a high correlation with the real value. Then, using the artificial neural network method, the local ratios between the real KEF and the approximate KEF are accurately modelled. Consequently, the SGS model can be improved by the product of that ratio and the approximate KEF. In LES of compressible turbulent channel flow, the new model can accurately predict mean velocity profile, turbulence intensities, Reynolds stress, temperature–velocity correlation, etc. Additionally, for the case of a compressible flat-plate boundary layer, the new model can accurately predict some key quantities, including the onset of transitions and transition peaks, the skin-friction coefficient, the mean velocity in the turbulence region, etc., and it can also predict the energy backscatters in turbulence. Furthermore, the proposed model also shows more advantages for coarser grids.

show abstract

Section: Introductionmentioning

confidence: 99%

Kinetic-energy-flux-constrained model using an artificial neural network for large-eddy simulation of compressible wall-bounded turbulence

Yuan

et al. 2021

J. Fluid Mech.

View full text Add to dashboard Cite

show abstract

“…Figure 1: Schematic illustration of GAN architecture (adapted from Kim et al (2021)). The novel element is the use for training and testing of gappy high-resolution fields provided directly by binned PTV data.…”

Section: Methodology 21 Piv-tailored Ganmentioning

confidence: 99%

GANs-based PIV resolution enhancement without the need of high-resolution input

Güemes¹,

Vila²,

Discetti³

2021

ISPIV21

View full text Add to dashboard Cite

A data-driven approach to reconstruct high-resolution flow fields is presented. The method is based on exploiting the recent advances of SRGANs (Super-Resolution Generative Adversarial Networks) to enhance the resolution of Particle Image Velocimetry (PIV). The proposed approach exploits the availability of incomplete projections on high-resolution fields using the same set of images processed by standard PIV. Such incomplete projection is made available by sparse particle-based measurements such as super-resolution particle tracking velocimetry. Consequently, in contrast to other works, the method does not need a dual set of low/high-resolution images, and can be applied directly on a single set of raw images for training and estimation. This data-enhanced particle approach is assessed employing two datasets generated from direct numerical simulations: a fluidic pinball and a turbulent channel flow. The results prove that this data-driven method is able to enhance the resolution of PIV measurements even in complex flows without the need of a separate high-resolution experiment for training.

show abstract

“…While LES may be practiced in isolation from specific concerns of a consistent framework, a specific definition of that which an LES aspires to accurately reproduce is required for advanced techniques such as data-driven closure (Sarghini, de Felice & Santini 2003;Moreau, Teytaud & Bertoglio 2006;Gamahara & Hattori 2017;Vollant, Balarac & Corre 2017;Wang et al 2018;Beck, Flad & Munz 2019;Cheng et al 2019;Yang et al 2019;Zhou et al 2019;Sirignano, MacArt & Freund 2020;Xie, Wang & Weinan 2020a;Xie, Yuan & Wang 2020b;Yuan, Xie & Wang 2020;Bode et al 2021;Duraisamy 2021;Freund & Ferrante 2021;Park & Choi 2021;Portwood et al 2021;Prakash, Jansen & Evans 2021;Stoffer et al 2021;Wang et al 2021) and super-resolution enrichment (Domaradzki & Loh 1999;Scotti & Meneveau 1999;Stolz & Adams 1999;Milano & Koumoutsakos 2002;Leonard 2016;Ghate & Lele 2017;Maulik & San 2017;Bassenne et al 2019;Wang, Zhao & Ihme 2019;Ghate & Lele 2020;Liu et al 2020;Kim et al 2021). For example, without a clear definition of what an LES solution should represent, one cannot train a neural network to serve as a sub-grid closure in a robust way.…”

Section: Introductionmentioning

confidence: 99%

A physics-inspired alternative to spatial filtering for large-eddy simulations of turbulent flows

Johnson

2022

J. Fluid Mech.

View full text Add to dashboard Cite

Large-eddy simulations (LES) are widely used for computing high Reynolds number turbulent flows. Spatial filtering theory for LES is not without its shortcomings, including how to define filtering for wall-bounded flows, commutation errors for non-uniform filters and extensibility to flows with additional complexity, such as multiphase flows. In this paper, the theory for LES is reimagined using a coarsening procedure that imitates nature. This physics-inspired coarsening approach is equivalent to Gaussian filtering for single-phase wall-free flows but opens up new insights for both physical understanding and modelling even in that simple case. For example, an alternative to the Germano identity is introduced and used to define a dynamic procedure without the need for a test filter. Non-uniform resolution can be represented in this framework without commutation errors, and the divergence-free condition is retained for incompressible flows. Potential extensions of the theory to more complex physics such as multiphase flows are briefly discussed.

show abstract

Unsupervised deep learning for super-resolution reconstruction of turbulence

Abstract: Abstract

Cited by 159 publications

References 55 publications

Kinetic-energy-flux-constrained model using an artificial neural network for large-eddy simulation of compressible wall-bounded turbulence

Kinetic-energy-flux-constrained model using an artificial neural network for large-eddy simulation of compressible wall-bounded turbulence

GANs-based PIV resolution enhancement without the need of high-resolution input

A physics-inspired alternative to spatial filtering for large-eddy simulations of turbulent flows

Contact Info

Product

Resources

About