Subgrid-scale model for large-eddy simulation of isotropic turbulent flows using an artificial neural network

被引:119
作者
Zhou, Zhideng [1 ,2 ]
He, Guowei [1 ,2 ]
Wang, Shizhao [1 ,2 ]
Jin, Guodong [1 ,2 ]
机构
[1] Chinese Acad Sci, Inst Mech, State Key Lab Nonlinear Mech, Beijing 100190, Peoples R China
[2] Univ Chinese Acad Sci, Sch Engn Sci, Beijing 100049, Peoples R China
基金
中国国家自然科学基金;
关键词
Machine learning; Artificial neural network; Subgrid-scale model; Large-eddy simulation; Isotropic turbulent flows; DIRECT NUMERICAL SIMULATIONS; DATA-DRIVEN; FORM UNCERTAINTIES; STRESS TENSOR; INVARIANCE;
D O I
10.1016/j.compfluid.2019.104319
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
An artificial neural network (ANN) is used to establish the relation between the resolved-scale flow field and the subgrid-scale (SGS) stress tensor, to develop a new SGS model for large-eddy simulation (LES) of isotropic turbulent flows. The data required for training and testing of the ANN are provided by performing filtering operations on the flow fields from direct numerical simulations (DNSs) of isotropic turbulent flows. We use the velocity gradient tensor together with filter width as input features and the SGS stress tensor as the output labels for training the ANN. In the a priori test of the trained ANN model, the SGS stress tensors obtained from the ANN model and the DNS data are compared by computing the correlation coefficient and the relative error of the energy transfer rate. The correlation coefficients are mostly larger than 0.9, and the ANN model can accurately predict the energy transfer rate at different Reynolds numbers and filter widths, showing significant improvement over the conventional models, for example the gradient model, the Smagorinsky model and its dynamic version. A real LES using the trained ANN model is performed as the a posteriori validation. The energy spectrum computed by the improved ANN model is compared with several SGS models. The Lagrangian statistics of fluid particle pairs obtained from the improved ANN model almost approach those from the filtered DNS, better than the results from the Smagorinsky model and dynamic Smagorinsky model. (C) 2019 Elsevier Ltd. All rights reserved.
引用
收藏
页数:16
相关论文
共 61 条
[31]   Scale-invariance and turbulence models for large-eddy simulation [J].
Meneveau, C ;
Katz, J .
ANNUAL REVIEW OF FLUID MECHANICS, 2000, 32 (32) :1-32
[32]   A recommended modification to the dynamic two-parameter mixed subgrid scale model for large eddy simulation of wall bounded turbulent flow [J].
Morinishi, Y ;
Vasilyev, OV .
PHYSICS OF FLUIDS, 2001, 13 (11) :3400-3410
[33]   A paradigm for data-driven predictive modeling using field inversion and machine learning [J].
Parish, Eric J. ;
Duraisamy, Karthik .
JOURNAL OF COMPUTATIONAL PHYSICS, 2016, 305 :758-774
[34]  
Pope S.B., 2001, Turbulent Flows, P298, DOI [DOI 10.1017/CBO9780511840531, 10.1017/CBO9780511840531]
[35]   LEARNING REPRESENTATIONS BY BACK-PROPAGATING ERRORS [J].
RUMELHART, DE ;
HINTON, GE ;
WILLIAMS, RJ .
NATURE, 1986, 323 (6088) :533-536
[36]   Neural networks based subgrid scale modeling in large eddy simulations [J].
Sarghini, F ;
de Felice, G ;
Santini, S .
COMPUTERS & FLUIDS, 2003, 32 (01) :97-108
[37]  
Singh A. P, 2015, 53 AIAA AER SCI M KI, P1
[38]   Using field inversion to quantify functional errors in turbulence closures [J].
Singh, Anand Pratap ;
Duraisamy, Karthik .
PHYSICS OF FLUIDS, 2016, 28 (04)
[39]  
Smagorinsky J., 1963, Mon Weather Rev, V91, P99, DOI [10.1175/1520-0493(1963)091<0099:GCEWTP>2.3.CO
[40]  
2, DOI 10.1175/1520-0493(1963)0910099:GCEWTP2.3.CO