Bulletin of the American Physical Society
66th Annual Meeting of the APS Division of Fluid Dynamics
Volume 58, Number 18
Sunday–Tuesday, November 24–26, 2013; Pittsburgh, Pennsylvania
Session G21: Turbulence: Simulations III - DNS and LES II |
Hide Abstracts |
Chair: Robert Moser, University of Texas Room: 316 |
Monday, November 25, 2013 8:00AM - 8:13AM |
G21.00001: Progress and opportunities in direct numerical simulations at the next higher resolution P.K. Yeung, K.R. Sreenivasan In recent years, many researchers in the turbulence community have been able to exploit the steady advancement of computing power to advance our understanding of turbulence, including new parameter ranges and the effects of coupling with other physical processes. However it is remarkable that, the ``record'' grid resolution of $4096^3$, first achieved just over 10 years ago (Kaneda {\em et al.}, {\em Phys. Fluids} 2003) still stands in the literature of the field. In this talk, we will present preliminary results from an $8192^3$ simulation of turbulence on a periodic domain, carried out using 262144 CPU cores on the {\em Blue Waters} supercomputer under the NSF Track 1 Petascale Resource Allocations program. Since a simulation at this magnitude is still extremely expensive, and the resources required are not easily secured, very careful planning and very aggressive efforts at algorithmic enhancement are necessary (which we will also briefly discuss). This new simulation is expected to allow us to probe deeply into fundamental questions such as intermittency at the highest Reynolds numbers and the best possible resolution of the small scales at the current limit of computing power available. [Preview Abstract] |
Monday, November 25, 2013 8:13AM - 8:26AM |
G21.00002: Backward tracking for the study of turbulent dispersion in direct numerical simulations over a range of Reynolds numbers D. Buaria, P.K. Yeung, B.L. Sawford The dispersive character of turbulence is well known, and readily observed through, for example, the increase with time of the mean separation between fluid particle pairs in a Lagrangian framework. Usually, in both direct numerical simulations (DNS) and laboratory experiments, a population of fluid particles is tracked forward in time from specified initial conditions. However, from a modeling perspective, it is more important to track the particles backwards, which would help address questions about the dynamical origins of a patch of contaminant material, or a highly convoluted multi-particle cluster. In this talk we present numerical results on backward statistics obtained by sampling particle pairs of desired separation at the final time of a relatively long DNS run. The calculation essentially involves processing large datasets consisting of the complete time history of position, velocity and velocity gradients along the trajectories. Promising results on both forward and backward dispersion from up to 16 million particles have been obtained over time intervals spanning the ballistic, inertial, and diffusive ranges. This approach will allow us to study backward dispersion and relate Lagrangian studies to scalar mixing, at Taylor-scale Reynolds numbers up to 1000. [Preview Abstract] |
Monday, November 25, 2013 8:26AM - 8:39AM |
G21.00003: A Web-Services accessible database for channel flow turbulence at $Re_\tau$=1000 J. Graham, K. Kanov, E. Givelberg, R. Burns, G. Eyink, A. Szalay, C. Meneveau, M.K. Lee, N. Malaya, R.D. Moser In this presentation we describe a new public database archiving a DNS data set of the space-time evolution of fully developed channel flow at $Re_\tau = 1000$. The database will contain data from a DNS of channel flow with domain-size of $8\pi \times 2 \times 3\pi$, at the resolution of 2048$\times$512$\times$1536, with 2048 time-frames of velocity and pressure fields spanning about a flow-through time scale. After simulation, the data are ingested into the database cluster using a space-filling Morton-curve to index the computational space uniformly, and also to organize data partition and distribution. The database system allows users access and to process the data remotely through an interface based on the Web-Service model. Users are thus able to perform numerical experiments on the high-resolution DNS data using least capable desktop computers. Test calculations are performed to illustrate the usage of the system and to verify the correctness of the data. Construction of the database also involves developments of MPI-DB, a new tool to facilitate coupling of parallel simulations and databases. [Preview Abstract] |
Monday, November 25, 2013 8:39AM - 8:52AM |
G21.00004: Estimating Uncertainties in Statistics Computed from DNS Nicholas Malaya, Todd Oliver, Rhys Ulerich, Robert Moser Rigorous assessment of uncertainty is crucial to the utility of DNS results. Uncertainties in the computed statistics arise from two sources: finite sampling and the discretization of the Navier-Stokes equations. Due to the presence of non-trivial sampling error, standard techniques for estimating discretization error (such as Richardson Extrapolation) fail or are unreliable. This talk provides a systematic and unified approach for estimating these errors. First, a sampling error estimator that accounts for correlation in the input data is developed. Then, this sampling error estimate is used as an input to a probabilistic extension of Richardson extrapolation in order to characterize the discretization error. These techniques are used to investigate the sampling and discretization errors in the DNS of a wall-bounded turbulent flow at $Re_{\tau}$ = 180. We will show a well-resolved DNS simulation which, for the centerline velocity, possesses 0.02\% sampling error and discretization errors of 0.003\%. These results imply that standard resolution heuristics for DNS accurately predict required grid sizes. [Preview Abstract] |
Monday, November 25, 2013 8:52AM - 9:05AM |
G21.00005: Direct numerical simulation for incompressible channel flow at $Re_\tau = 5200$ Myoungkyu Lee, Nicholas Malaya, Robert D. Moser High-resolution direct numerical simulation (DNS) of wall-bounded canonical channel flow at $Re_\tau = 5200$ is performed. The computational domain is $8\pi\delta \times 2\delta \times 3\pi\delta$ with $10240 \times 1536 \times 7680$ grid points in streamwise($x$), wall-normal direction($y$), and spanwise($z$) directions, respectively. Fourier spectral method($x$, and $z$) and B-splines($y$) are used for the the computation of derivatives. In this presentation we demonstrate that the simulation exhibits several features of high Reynolds number wall-bounded turbulence. The value of von K\'{a}rm\'{a}n constant appears to be $\kappa$ = 0.384 in the region of $y^+$=300 $\sim$ $y$=$0.2\delta$ where the mean velocity profile shows logarithmic variation. Also, distinct inner($\lambda _x^+$= 800, $\lambda _z^+$= 120) and outer($\lambda _x$= 8$\delta$, $\lambda _z$= $\delta$) peaks in one-dimensional premultiplied spectra of the velocity variance are observed. Finally, the $k_x^{-1}$ region is observed in the range of $y^+$= $120\sim 150$ and $k_x$= $6 \sim 10$. [Preview Abstract] |
Monday, November 25, 2013 9:05AM - 9:18AM |
G21.00006: LES of radial wall jets over smooth and rough surfaces Rayhaneh Banyassady, Ugo Piomelli Large-eddy simulations of turbulent radial wall jets were conducted over both smooth and rough surfaces ($5 |
Monday, November 25, 2013 9:18AM - 9:31AM |
G21.00007: Large-eddy simulation of turbulent dispersion from a localized source in a build-up environment Bing-Chen Wang, Mohammad Saeedi Turbulent dispersion from a continuous ground-level point-source within an array of 16x16 cubes has been simulated using wall-modelling large-eddy simulations. The major challenges associated with this problem involve obtaining a deeper understanding of the interaction of the dynamically evolving flow structures with the complex boundary conditions, coupling of the momentum and scalar transport processes, and a high Reynolds number tested for an modeled urban atmospheric boundary layer (Re$=$12,005 based on the free stream velocity and obstacle height). A fully-parallelized in-house computer code was used for performing the simulation. An advanced dynamic nonlinear model (DNM) and dynamic full linear eddy diffusivity model (DFLTDM) have been used for closure of the filtered momentum and scalar transport equations, respectively. A non-equilibrium thin boundary-layer wall model is applied to all solid surfaces. Inlet boundary conditions based on solid grids have also been investigated in order to generate high turbulence levels typical for an approaching urban atmospheric boundary-layer flow. The predicted results for the flow and concentration field have been thoroughly validated against a set of high-quality water-channel measurement data. [Preview Abstract] |
Monday, November 25, 2013 9:31AM - 9:44AM |
G21.00008: Analysis of effective eddy viscosity in DNS results of stratified turbulence Sina Khani, Michael L. Waite In order to perform large-eddy simulation (LES) of stratified turbulence, subgrid-scale (SGS) models are necessary to represent the effects of small scales on large scale motions. Since the inertial subrange of stratified turbulence is anisotropic for scales larger than the Ozmidov scale, isotropic SGS models do not seem to be the proper approach for LES of stratified turbulence. In this talk, direct numerical simulations of decaying stratified turbulence are analyzed to investigate the effective eddy viscosity in the presence of stratification. The results are studied under different Reynolds numbers, stratifications, and test cutoffs. It is shown that the presence of stratification causes a non-local horizontal energy transfer between large and small horizontal scales, which is not seen for the unstratified case. These results suggest the idea of using different eddy viscosities in the horizontal and vertical directions for scales larger than the Ozmidov scale, as is often done in large-scale atmospheric models, in which stratification is important. Overall, isotropic SGS models should be modified for use in LES of stratified turbulence. [Preview Abstract] |
Monday, November 25, 2013 9:44AM - 9:57AM |
G21.00009: ABSTRACT WITHDRAWN |
Monday, November 25, 2013 9:57AM - 10:10AM |
G21.00010: The $k^{-2}$ spectrum in decaying magnetohydrodynamic turbulence Vassilios Dallas, Alexandros Alexakis We investigate the origins of the $k^{-2}$ spectrum in a decaying Taylor-Green magnetohydrodynamic flow with zero large scale magnetic flux. So far a possible candidate for this scaling exponent has been the weak turbulence phenomenology. From our numerical simulations, we observe that current sheets in this flow are formed in regions of magnetic discontinuities. Based on this observation and by studying the influence of the current sheets on the energy spectrum, using a filtering technique, we demonstrate that magnetic discontinuities are responsible for the $-2$ power law scaling of the energy spectra in this flow. We also show that initial strong correlations between the velocity and the magnetic field exhibit a $k^{-2}$ spectrum at the peak of dissipation of decaying MHD turbulence. Therefore, the presence of a clear $k^{-2}$ spectrum due to strong current sheets could imply lack of universality in decaying MHD turbulence. However, our highest resolution simulations ($2048^3$) indicate that we have classes of universality at moderate Reynolds numbers and possibly a universal power law in the high Reynolds number limit. [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700