scholarly journals Machine learning-augmented turbulence modeling for RANS simulations of massively separated flows

2021 ◽  
Vol 6 (6) ◽  
Author(s):  
Pedro Stefanin Volpiani ◽  
Morten Meyer ◽  
Lucas Franceschini ◽  
Julien Dandois ◽  
Florent Renac ◽  
...  
Keyword(s):  
Machine Learning ◽  
Turbulence Modeling ◽  
Separated Flows ◽  
Massively Separated Flows ◽  
Rans Simulations

Comparative assessment of SAS and DES turbulence modeling for massively separated flows

2015 ◽  
Vol 32 (1) ◽  
pp. 12-21 ◽  
Author(s):  
Weilin Zheng ◽  
Chao Yan ◽  
Hongkang Liu ◽  
Dahai Luo
Keyword(s):  
Turbulence Modeling ◽  
Comparative Assessment ◽  
Separated Flows ◽  
Massively Separated Flows

Machine learning methods for turbulence modeling in subsonic flows around airfoils

2019 ◽  
Vol 31 (1) ◽  
pp. 015105 ◽  
Author(s):  
Linyang Zhu ◽  
Weiwei Zhang ◽  
Jiaqing Kou ◽  
Yilang Liu
Keyword(s):  
Machine Learning ◽  
Turbulence Modeling ◽  
Learning Methods ◽  
Subsonic Flows ◽  
Machine Learning Methods

scholarly journals Stochastic turbulence modeling in RANS simulations via multilevel Monte Carlo

2020 ◽  
Vol 201 ◽  
pp. 104420
Author(s):  
Prashant Kumar ◽  
Martin Schmelzer ◽  
Richard P. Dwight
Keyword(s):  
Monte Carlo ◽  
Turbulence Modeling ◽  
Multilevel Monte Carlo ◽  
Rans Simulations

scholarly journals Cartographing dynamic stall with machine learning

2020 ◽  
Vol 5 (2) ◽  
pp. 819-838
Author(s):  
Matthew Lennie ◽  
Johannes Steenbuck ◽  
Bernd R. Noack ◽  
Christian Oliver Paschereit
Keyword(s):  
Machine Learning ◽  
Dynamic Time Warping ◽  
Data Science ◽  
Leading Edge ◽  
Separated Flows ◽  
Dynamic Stall ◽  
Energy Yield ◽  
Time Warping ◽  
Coherent Vortex ◽  
Dynamic Time

Abstract. Once stall has set in, lift collapses, drag increases and then both of these forces will fluctuate strongly. The result is higher fatigue loads and lower energy yield. In dynamic stall, separation first develops from the trailing edge up the leading edge. Eventually the shear layer rolls up, and then a coherent vortex forms and then sheds downstream with its low-pressure core causing a lift overshoot and moment drop. When 50+ experimental cycles of lift or pressure values are averaged, this process appears clear and coherent in flow visualizations. Unfortunately, stall is not one clean process but a broad collection of processes. This means that the analysis of separated flows should be able to detect outliers and analyze cycle-to-cycle variations. Modern data science and machine learning can be used to treat separated flows. In this study, a clustering method based on dynamic time warping is used to find different shedding behaviors. This method captures the fact that secondary and tertiary vorticity vary strongly, and in static stall with surging flow the flow can occasionally reattach. A convolutional neural network was used to extract dynamic stall vorticity convection speeds and phases from pressure data. Finally, bootstrapping was used to provide best practices regarding the number of experimental repetitions required to ensure experimental convergence.

Assessment of Large-Eddy Simulation of Internal Separated Flow

2009 ◽  
Vol 131 (7) ◽  
Author(s):  
Marco Hahn ◽  
Dimitris Drikakis
Keyword(s):  
Separated Flow ◽  
Separated Flows ◽  
Scalar Conservation Laws ◽  
Third Order ◽  
Eddy Simulation ◽  
Massively Separated Flows ◽  
Large Eddy ◽  
Stress Profiles ◽  
Scalar Conservation ◽  
Fifth Order

This paper presents a systematic numerical investigation of different implicit large-eddy simulations (LESs) for massively separated flows. Three numerical schemes, a third-order accurate monotonic upwind scheme for scalar conservation laws (MUSCL) scheme, a fifth-order accurate MUSCL scheme, and a ninth-order accurate weighted essentially non-oscillatory (WENO) method, are tested in the context of separation from a gently curved surface. The case considered here is a simple wall-bounded flow that consists of a channel with a hill-type curvature on the lower wall. The separation and reattachment locations, velocity, and Reynolds stress profiles are presented and compared against solutions from classical LES simulations.

Modeling the pressure strain correlation in turbulent flows using deep neural networks

2021 ◽  
pp. 095440622110429
Author(s):  
Jyoti P Panda ◽  
Hari V Warrior
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Turbulent Flows ◽  
Reynolds Stress ◽  
Channel Flow ◽  
Turbulence Modeling ◽  
Critical Role ◽  
Turbulent Channel Flow ◽  
Correlation Models ◽  
Turbulence Data

The pressure strain correlation plays a critical role in the Reynolds stress transport modeling. Accurate modeling of the pressure strain correlation leads to the proper prediction of turbulence stresses and subsequently the other terms of engineering interest. However, classical pressure strain correlation models are often unreliable for complex turbulent flows. Machine learning–based models have shown promise in turbulence modeling, but their application has been largely restricted to eddy viscosity–based models. In this article, we outline a rationale for the preferential application of machine learning and turbulence data to develop models at the level of Reynolds stress modeling. As an illustration, we develop data-driven models for the pressure strain correlation for turbulent channel flow using neural networks. The input features of the neural networks are chosen using physics-based rationale. The networks are trained with the high-resolution DNS data of turbulent channel flow at different friction Reynolds numbers (Reλ). The testing of the models is performed for unknown flow statistics at other Reλ and also for turbulent plane Couette flows. Based on the results presented in this article, the proposed machine learning framework exhibits considerable promise and may be utilized for the development of accurate Reynolds stress models for flow prediction.

Sign in / Sign up

Export Citation Format

Share Document