Artificial neural network-based spatial gradient models for large-eddy simulation of turbulence

Chemical kinetics modeling within the numerical simulations of turbulent reactive flows is a critical issue. In this study, first a large eddy simulation of a non-premixed planar methane jet flame with artificial neural network -based chemical kinetics is performed. To consider the effects of turbulence on the evolution of the thermo-chemical state space, the artificial neural network training patterns are obtained by solving the unsteady reaction-diffusion equations of linear eddy mixing sub-grid combustion model. The optimization procedure for selecting the optimum neural network architecture is discussed in detail. A flow field analysis and a comparison between a non-reactive jet with the same velocity boundary conditions with the flame jet are performed. It is shown that the non-reactive jet is more turbulent and vortical structures of the non-reactive jet occur closer to the jet exit compared to the flame jet. Second, in a first attempt to quantify the influence of different chemical kinetics methods, large eddy simulation computations are performed with other methods used to represent the chemical kinetics such as direct integration and look-up table. A comparative study is performed between the averaged, instantaneous and fluctuation profiles of the scalar field and flame structures obtained by each method. Although all large eddy simulation results are very close to each other, the main discrepancy is observed for CO mass fractions obtained by artificial neural network and look-up table-based chemical kinetics simulations. It is also shown that the averaged scalar field and the conditional averaged values of the flame are overestimated by look-up table based chemical kinetics large eddy simulation computations. Moreover, the potential savings of artificial neural network in computational costs in comparison with other chemistry approaches are illustrated.

Download Full-text

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

Computers & Fluids ◽

10.1016/j.compfluid.2019.104319 ◽

2019 ◽

Vol 195 ◽

pp. 104319 ◽

Cited By ~ 9

Author(s):

Zhideng Zhou ◽

Guowei He ◽

Shizhao Wang ◽

Guodong Jin

Keyword(s):

Neural Network ◽

Artificial Neural Network ◽

Large Eddy Simulation ◽

Turbulent Flows ◽

Scale Model ◽

Subgrid Scale ◽

Eddy Simulation ◽

Large Eddy ◽

Subgrid Scale Model ◽

Artificial Neural

Download Full-text

Analysis and optimization of multi-inlet gas cyclones using large eddy simulation and artificial neural network

Powder Technology ◽

10.1016/j.powtec.2017.02.004 ◽

2017 ◽

Vol 311 ◽

pp. 465-483 ◽

Cited By ~ 39

Author(s):

Lakhbir Singh Brar ◽

Khairy Elsayed

Keyword(s):

Neural Network ◽

Artificial Neural Network ◽

Large Eddy Simulation ◽

Eddy Simulation ◽

Large Eddy ◽

Artificial Neural

Download Full-text

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

Journal of Fluid Mechanics ◽

10.1017/jfm.2021.1012 ◽

2021 ◽

Vol 932 ◽

Author(s):

Changping Yu ◽

Zelong Yuan ◽

Han Qi ◽

Jianchun Wang ◽

Xinliang Li ◽

...

Keyword(s):

Neural Network ◽

Artificial Neural Network ◽

Kinetic Energy ◽

Large Eddy Simulation ◽

Energy Flux ◽

Mean Velocity ◽

New Model ◽

Eddy Simulation ◽

Large Eddy ◽

Wall Bounded Turbulence

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.

Download Full-text