Reconstruction of RANS Model and Cross-Validation of Flow Field Based on Tensor Basis Neural Network

2019 ◽  
Author(s):  
Xudong Song ◽  
Zhen Zhang ◽  
Yiwei Wang ◽  
Shuran Ye ◽  
Chenguang Huang
Keyword(s):  
Neural Network ◽  
Flow Field ◽  
Reynolds Stress ◽  
Prediction Accuracy ◽  
Eddy Viscosity ◽  
Viscosity Model ◽  
Strain Rate Tensor ◽  
Rans Model ◽  
Eddy Viscosity Model ◽  
Tensor Basis

Abstract The solution of the Reynolds-averaged Navier-Stokes (RANS) equation has been widely used in engineering problems. However, this model does not provide satisfactory prediction accuracy. Because the widely used eddy viscosity model assumes a linear relationship between the Reynolds stress and the average strain rate tensor and these linear models cannot capture the anisotropic characteristics of the actual flow. In this paper, two kinds of flow field structures of two-dimensional cylindrical flow and circular tube jet are calculated by using the RANS model. Secondly, in order to improve the prediction accuracy of the RANS model, the Reynolds stress of the RANS model is reconstructed by the tensor basis neural network algorithm based on nonlinear eddy viscosity model. Finally, the model trained by neural network is cross-validated, and compare the cross-test results with the traditional RANS k-eps model. The results show that the multi-layer neural network method has achieved good results in turbulence model reconstruction.

Subgrid-scale structure and fluxes of turbulence underneath a surface wave

2019 ◽  
Vol 878 ◽  
pp. 768-795
Author(s):  
Kuanyu Chen ◽  
Minping Wan ◽  
Lian-Ping Wang ◽  
Shiyi Chen
Keyword(s):  
Strain Rate ◽  
Eddy Viscosity ◽  
Isotropic Turbulence ◽  
Wave Theory ◽  
Viscosity Model ◽  
Strain Rate Tensor ◽  
Linear Wave ◽  
Subgrid Scale ◽  
Eddy Viscosity Model ◽  
Wave Induced

In this study, the behaviours of subgrid-scale (SGS) turbulence are investigated with direct numerical simulations when an isotropic turbulence is brought to interact with imposed rapid waves. A partition of the velocity field is used to decompose the SGS stress into three parts, namely, the turbulent part $\unicode[STIX]{x1D749}^{T}$, the wave-induced part $\unicode[STIX]{x1D749}^{W}$ and the cross-interaction part $\unicode[STIX]{x1D749}^{C}$. Under strong wave straining, $\unicode[STIX]{x1D749}^{T}$ is found to follow the Kolmogorov scaling $\unicode[STIX]{x1D6E5}_{c}^{2/3}$, where $\unicode[STIX]{x1D6E5}_{c}$ is the filter width. Based on the linear Airy wave theory, $\unicode[STIX]{x1D749}^{W}$ and the filtered strain-rate tensor due to the wave motion, $\tilde{\unicode[STIX]{x1D64E}}^{W}$, are found to have different phases, posing a difficulty in applying the usual eddy-viscosity model. On the other hand, $\unicode[STIX]{x1D749}^{T}$ and the filtered strain-rate tensor due to the turbulent motion, $\tilde{\unicode[STIX]{x1D64E}}^{T}$, are only weakly wave-phase-dependent and could be well related by an eddy-viscosity model. The linear wave theory is also used to describe the vertical distributions of SGS statistics driven by the wave-induced motion. The predictions are in good agreement with the direct numerical simulation results. The budget equation for the turbulent SGS kinetic energy shows that the transport terms related to turbulence are important near the free surface and they compensate the imbalance between the energy flux and the SGS energy dissipation.

Modeling Separation-Induced Transition Using a Non-Linear Three Equation Turbulence Model and a Reynolds Stress Turbulence Model

2010 ◽  
Author(s):  
Zinon Vlahostergios ◽  
Kyros Yakinthos
Keyword(s):  
Transport Equation ◽  
Reynolds Stress ◽  
Turbulence Model ◽  
Eddy Viscosity ◽  
Turbulence Models ◽  
Viscosity Model ◽  
Transitional Flows ◽  
Eddy Viscosity Model ◽  
Non Linear ◽  
Stress Models

This paper presents an effort to model separation-induced transition on a flat plate with a semi-circular leading edge, by using two advanced turbulence models, the three equation non-linear model k-ε-A2 of Craft et al. [16] and the Reynolds-stress model of Craft [13]. The mechanism of the transition is governed by the different inlet velocity and turbulence intensity conditions, which lead to different recirculation bubbles and different transition onset points for each case. The use of advanced turbulence models in predicting the development of transitional flows has shown, in past studies, good perspectives. The k-ε-A2 model uses an additional transport equation for the A2 Reynolds stress invariant and it is an improvement of Craft et al. [12] non-linear eddy viscosity model. The use of the third transport equation gives improved results in the prediction of the longitudinal Reynolds stress distributions and especially, in flows where transitional phenomena may occur. Although this model is a pure eddy-viscosity model, it borrows many aspects from the more complex Reynolds-stress models. On the other hand, the use of an advanced Reynolds-stress turbulence model, such as the one of Craft [13], can predict many complex flows and there are indications that it can be applied to transitional flows also, since the crucial terms of Reynolds stress generation are computed exactly and normal stress anisotropy is resolved. The model of Craft [13], overcomes the drawbacks of the common used Reynolds-stress models regarding the computation of wall-normal distances and vectors in order to account for wall proximity effects. Instead of these quantities, it employs “normalized turbulence lengthscale gradients” which give the ability to identify the presence of strong inhomogeneity in a flow development, in an easier way. The final results of both turbulence models showed acceptable agreement with the experimental data. In this work it is shown that there is a good potential to model separation-induced transitional flows, with advanced turbulence modeling without any additional use of ad-hoc modifications or additional equations, based on various transition models.

Full Field Algebraic Anisotropic Eddy Viscosity Model for the Film Cooling Flows

2012 ◽  
Author(s):  
Xueying Li ◽  
Jing Ren ◽  
Hongde Jiang
Keyword(s):  
Flow Field ◽  
Film Cooling ◽  
Eddy Viscosity ◽  
Turbulence Models ◽  
Viscosity Model ◽  
Wall Region ◽  
Mean Flow ◽  
Experiment Data ◽  
Anisotropic Models ◽  
Eddy Viscosity Model

The algebraic anisotropic eddy viscosity model proposed by the authors is further developed to make it suitable to the full flow field in order to focus not only in the near wall region but also in the main flow field. The three anisotropic eddy viscosity ratios for u′v′, u′w′, v′w′ are determined from the eddy viscosity hypothesis and algebraic Reynolds stress transport equations and expressed in Cartesian coordinate system. This model is applied to four isotropic two-equation turbulence models to make them anisotropic. These anisotropic models are validated with the experiment data from Sinha et al.[1]. Thorough tests are performed with all these isotropic and anisotropic turbulence models for film cooling on a flatplate with different blowing ratios. Detailed analyses of computational simulations are presented. The predicted adiabatic film cooling effectiveness and mean flow field show that the algebraic anisotropic eddy-viscosity turbulence models agree better with the experiment data. Among the four anisotropic models, the anisotropic models based on the realizable k-ε and RNG k-ε models stand out as the most promising models for flatplate film cooling predictions. It’s a big advantage of this model that it deals with the whole flow field and can be combined with different turbulence models.

scholarly journals Reynolds averaged turbulence modelling using deep neural networks with embedded invariance

2016 ◽  
Vol 807 ◽  
pp. 155-166 ◽  
Author(s):  
Julia Ling ◽  
Andrew Kurzawski ◽  
Jeremy Templeton
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Reynolds Stress ◽  
Network Architecture ◽  
Eddy Viscosity ◽  
Deep Neural Networks ◽  
Test Cases ◽  
Neural Network Architecture ◽  
Stress Anisotropy ◽  
Anisotropy Tensor

There exists significant demand for improved Reynolds-averaged Navier–Stokes (RANS) turbulence models that are informed by and can represent a richer set of turbulence physics. This paper presents a method of using deep neural networks to learn a model for the Reynolds stress anisotropy tensor from high-fidelity simulation data. A novel neural network architecture is proposed which uses a multiplicative layer with an invariant tensor basis to embed Galilean invariance into the predicted anisotropy tensor. It is demonstrated that this neural network architecture provides improved prediction accuracy compared with a generic neural network architecture that does not embed this invariance property. The Reynolds stress anisotropy predictions of this invariant neural network are propagated through to the velocity field for two test cases. For both test cases, significant improvement versus baseline RANS linear eddy viscosity and nonlinear eddy viscosity models is demonstrated.

scholarly journals On the extension of the eddy viscosity model to compressible flows

2014 ◽  
Vol 26 (4) ◽  
pp. 041702 ◽  
Author(s):  
M. Germano ◽  
A. Abbà ◽  
R. Arina ◽  
L. Bonaventura
Keyword(s):  
Compressible Flows ◽  
Eddy Viscosity ◽  
Viscosity Model ◽  
Eddy Viscosity Model
Sign in / Sign up

Export Citation Format

Share Document