Computations of equilibrium and non-equilibrium turbulent channel flows using a nested-LES approach

2016 ◽  
Vol 793 ◽  
pp. 709-748 ◽  
Author(s):  
Yifeng Tang ◽  
Rayhaneh Akhavan

A new nested-LES approach for computation of high Reynolds number, equilibrium, and non-equilibrium, wall-bounded turbulent flows is presented. The method couples coarse-resolution LES in the full computational domain with fine-resolution LES in a minimal flow unit to retain the accuracy of well-resolved LES throughout the computational domain, including in the near-wall region, while significantly reducing the computational cost. The two domains are coupled by renormalizing the instantaneous velocity fields in each domain dynamically during the course of the simulation to match the wall-normal profiles of single-time, ensemble-averaged kinetic energies of the components of ‘mean’ and fluctuating velocities in the inner layer to those of the minimal flow unit, and in the outer layer to those of the full domain. This simple renormalization procedure is shown to correct the energy spectra and wall shear stresses in both domains, thus leading to accurate turbulence statistics. The nested-LES approach has been applied to computation of equilibrium turbulent channel flow at $Re_{{\it\tau}}\approx 1000$, 2000, 5000, 10 000, and non-equilibrium, strained turbulent channel flow at $Re_{{\it\tau}}\approx 2000$. In both flows, nested-LES predicts the skin friction coefficient, first- and higher-order turbulence statistics, spectra and structure of the flow in agreement with available DNS and experimental data. Nested-LES can be applied to any flow with at least one direction of local or global homogeneity, while reducing the required number of grid points from $O(Re_{{\it\tau}}^{2})$ of conventional LES to $O(\log Re_{{\it\tau}})$ or $O(Re_{{\it\tau}})$ in flows with two or one locally or globally homogeneous directions, respectively.

2018 ◽  
Vol 15 (2) ◽  
pp. 75-89
Author(s):  
Muhammad Saiful Islam Mallik ◽  
Md. Ashraf Uddin

A large eddy simulation (LES) of a plane turbulent channel flow is performed at a Reynolds number Re? = 590 based on the channel half width, ? and wall shear velocity, u? by approximating the near wall region using differential equation wall model (DEWM). The simulation is performed in a computational domain of 2?? x 2? x ??. The computational domain is discretized by staggered grid system with 32 x 30 x 32 grid points. In this domain the governing equations of LES are discretized spatially by second order finite difference formulation, and for temporal discretization the third order low-storage Runge-Kutta method is used. Essential turbulence statistics of the computed flow field based on this LES approach are calculated and compared with the available Direct Numerical Simulation (DNS) and LES data where no wall model was used. Comparing the results throughout the calculation domain we have found that the LES results based on DEWM show closer agreement with the DNS data, especially at the near wall region. That is, the LES approach based on DEWM can capture the effects of near wall structures more accurately. Flow structures in the computed flow field in the 3D turbulent channel have also been discussed and compared with LES data using no wall model.


Author(s):  
Saeed Jafari ◽  
Mohammad Rahnama

Generalized Lattice Boltzmann Equation (GLBE) was used for computation of turbulent channel flow for which Large Eddy Simulation (LES) was employed as a turbulence model. The subgrid-Scale turbulence effects were simulated through a Shear-Improved Smagorinsky Model (SISM) which is capable of predicting turbulent near wall region accurately without any wall function. Computations were done for a relatively coarse grid with shear Reynolds number of 180 in a parallelized code. Good numerical stability was observed for this computational framework. Results of mean velocity distribution across the channel showed good correspondence with Direct Numerical Simulation (DNS) data. Negligible discrepancies were observed for computed turbulent statistics between present computations and those reported from DNS. Three-dimensional instantaneous vorticity contours showed complex vortical structures appeared in such flow geometries. It is concluded that such framework is capable of predicting accurate results for turbulent channel flow without adding significant complication and computational cost to the standard Smagorinsky model.


1996 ◽  
Vol 310 ◽  
pp. 269-292 ◽  
Author(s):  
Hugh M. Blackburn ◽  
Nagi N. Mansour ◽  
Brian J. Cantwell

An investigation of topological features of the velocity gradient field of turbulent channel flow has been carried out using results from a direct numerical simulation for which the Reynolds number based on the channel half-width and the centreline velocity was 7860. Plots of the joint probability density functions of the invariants of the rate of strain and velocity gradient tensors indicated that away from the wall region, the fine-scale motions in the flow have many characteristics in common with a variety of other turbulent and transitional flows: the intermediate principal strain rate tended to be positive at sites of high viscous dissipation of kinetic energy, while the invariants of the velocity gradient tensor showed that a preference existed for stable focus/stretching and unstable node/saddle/saddle topologies. Visualization of regions in the flow with stable focus/stretching topologies revealed arrays of discrete downstream-leaning flow structures which originated near the wall and penetrated into the outer region of the flow. In all regions of the flow, there was a strong preference for the vorticity to be aligned with the intermediate principal strain rate direction, with the effect increasing near the walls in response to boundary conditions.


Author(s):  
Atsushi Nagamachi ◽  
Takahiro Tsukahara

Abstract We tested Artificial Neural Networks (ANNs) to predict a fully-developed turbulent channel flow of a viscoelastic fluid in preparation for elucidating flow phenomenon and solving the difficulty in DNS (Direct Numerical Simulation) due to numerical instability of the viscoelastic fluid. Two kinds of ANNs (multi-layer perceptron (MLP) and U-Net) were trained using DNS data to predict conformation stress from given instantaneous field. The MLP showed accurate predictions and predictions got better with z-score normalization. ANN predicted accurately in near-wall region having coherent structures. In addition, we demonstrated that ANN get the nonlinear relationship between velocity gradient and viscoelastic stress partially.


Author(s):  
Dongmei Zhou ◽  
Kenneth S. Ball

This paper has two objectives, (1) to examine the effects of spatial resolution, (2) to examine the effects of computational box size, upon turbulence statistics and the amount of drag reduction with and without the control scheme of wall oscillation. Direct numerical simulation (DNS) of the fully developed turbulent channel flow was performed at Reynolds number of 200 based on the wall-shear velocity and the channel half-width by using spectral methods. For the first objective, four different grids were applied to the same computational domain and the biggest impact was observed on the logarithmic law of mean velocity profiles and on the amount of drag reduction with 28.3% for the coarsest mesh and 35.4% for the finest mesh. Other turbulence features such as RMS velocity fluctuations, RMS vorticity fluctuations, and bursting events were either overpredicted or underpredicted through coarse grids. For the second objective, two different minimal channels and one natural full channel were studied and 3% drag reduction difference was observed between the smallest minimal channel of 39.1% and the natural full channel of 36.2%. In the near-wall region, however, the minimal channel flow did not exhibit significant difference in the mean velocity profiles and other lower-order statistics. Finally, from this systematical study, it showed that the accuracy of DNS depends more on the spanwise resolution, and it also confirmed that a minimal channel model is able to catch key structures of turbulence in the near-wall region but is much less expensive.


Author(s):  
Boris Arcen ◽  
Anne Tanie`re ◽  
Benoiˆt Oesterle´

The importance of using the lift force and wall-corrections of the drag coefficient for modeling the motion of solid particles in a fully-developed channel flow is investigated by means of direct numerical simulation (DNS). The turbulent channel flow is computed at a Reynolds number based on the wall-shear velocity and channel half-width of 185. Contrary to most of the numerical simulations, we consider in the present study a lift force formulation that accounts for the weak and strong shear as well as for the wall effects (hereinafter referred to as optimum lift force), and the wall-corrections of the drag force. The DNS results show that the optimum lift force and the wall-corrections of the drag together have little influence on most of the statistics (particle concentration, mean velocities, and mean relative and drift velocities), even in the near wall region.


1992 ◽  
Vol 114 (3) ◽  
pp. 598-606 ◽  
Author(s):  
N. Kasagi ◽  
Y. Tomita ◽  
A. Kuroda

A direct numerical simulation (DNS) of the fully developed thermal field in a two-dimensional turbulent channel flow of air was carried out. The isoflux condition was imposed on the two walls so that the local mean temperature increased linearly in the streamwise direction. With any buoyancy effect neglected, temperature was considered as a passive scalar. The computation was executed on 1,589,248 grid points by using a spectral method. The statistics obtained were root-mean-square temperature fluctuations, turbulent heat fluxes, turbulent Prandtl number, and dissipation time scales. They agreed fairly well with existing experimental and numerical simulation data. Each term in the budget equations of temperature variance, its dissipation rate, and turbulent heat fluxes was also calculated. It was found that the temperature fluctuation θ′ was closely correlated with the streamwise velocity fluctuation u′, particularly in the near-wall region. Hence, the distribution of budget terms for the streamwise and wall-normal heat fluxes, u′θ′ and v′θ′, were very similar to those for the two Reynolds stress components, u′u′ and u′v′.


2008 ◽  
Vol 609 ◽  
pp. 349-375 ◽  
Author(s):  
E. COSTA-PATRY ◽  
L. MYDLARSKI

The interaction of two passive scalars (both temperature in air) emitted from concentrated line sources in fully developed high-aspect-ratio turbulent channel flow is studied. The thermal fields are measured using cold-wire thermometry in a flow with a Reynolds number (Uh/ν) of 10200.The transverse total root-mean-square (RMS) temperature profiles are a function of the separation distance between the line sources (d/h), their average wall-normal position (ysav/h), and the downstream location (x/h), measured relative to the line sources. Similarly, profiles of the non-dimensional form of the scalar covariance, the correlation coefficient (ρ), are a function of the same parameters and quantify the mixing of the two scalars.The transverse profiles of the correlation coefficient are generally largest at the edges of the thermal plume and smallest in its core. When the line sources are not symmetrically located about the channel centreline, the minimum in the correlation coefficient transverse profiles drifts towards the (closer) channel wall. For source locations that are equidistant from the channel centreline, the minimum correlation coefficient occurs at the centreline, due to the underlying symmetry of this geometry. The initial downstream evolution of the correlation coefficient depends significantly on d/h, similar to that in homogeneous turbulence. However, there is always a dependence on ysav/h, which increases in importance as both the downstream distance is increased and the wall is approached. Lastly, the correlation coefficient profiles tend towards positive values in the limit of large downstream distances (relative to the source separation), though further measurements farther downstream are required to confirm the exact value(s) of their asymptotic limit(s).Spectral analysis of the cospectra and coherency spectra indicates that the large scales evolve more rapidly than the small ones. Furthermore, the fast evolution of the large scales was most evident when the sources were located close to the wall. This presumably derives from the large-scale nature of turbulence production, which is strong in the near-wall region.


Sign in / Sign up

Export Citation Format

Share Document