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 D5: CFD II: LES I |
Hide Abstracts |
Chair: Satbir Singh, Carnegie Mellon University Room: 327 |
Sunday, November 24, 2013 2:15PM - 2:28PM |
D5.00001: Direct Simulations of Breaking Ocean Waves with Data Assimilation James Rottman, Douglas Dommermuth, Lucas Rhymes An algorithm is developed to assimilate ocean wave data into the Numerical Flow Analysis (NFA) code. NFA is a Cartesian-based implicit LES code with Volume of Fluid (VOF) interface capturing. The assimilation of data into NFA permits the investigation of higher bandwidths than is possible using either High-Order Spectral (HOS) methods or field measurements. NFA models wave breaking that cannot be modeled using HOS. Direct simulations of wave breaking allow a more detailed analysis of ocean-wave physics than is possible using remote sensing of the ocean surface. Examples of simulated ocean surface waves exhibiting wave breaking, based on the JONSWAP or Pierson-Moskowitz ocean wave spectra, will be shown. Future plans include performing data assimilation of larger patches of the ocean surface with higher resolution to investigate the statistics and underlying structure of breaking waves. [Preview Abstract] |
Sunday, November 24, 2013 2:28PM - 2:41PM |
D5.00002: Large-eddy simulation of boundary layer flow on a non-uniform grid using explicit filtering and reconstruction Lauren Goodfriend, Fotini Katopodes Chow, Marcos Vanella, Elias Balaras Many realistic flows, such as the urban boundary layer, are too expensive to simulate directly. Large-eddy simulation (LES) and adaptive mesh refinement (AMR) reduce the computational cost of turbulence modeling by restricting resolved length scales, but combining these techniques generates additional errors. The grid refinement interfaces in AMR grids can reflect resolved energy and create interpolation errors. This study investigates the use of explicit filtering and reconstruction to mitigate grid interface errors in LES of a pressure gradient forced boundary layer. The domain is split in the streamwise direction into two equally sized structured grids, one fine and one coarse, with periodic boundaries in the streamwise and spanwise directions. This simple test case allows observation of the effects of the grid interfaces. Explicit filtering is found to reduce accumulation of resolved energy at the fine-to-coarse interface and improve the shape of coherent structures, compared to basic LES. Reconstruction of the subfilter velocity is shown to further the improvements of explicit filtering. These results inform the use of LES on block-structured non-uniform grids, such as nested grids in local atmospheric models or on more complex Cartesian AMR grids. [Preview Abstract] |
Sunday, November 24, 2013 2:41PM - 2:54PM |
D5.00003: Large Eddy Simulations of Turbulent Reacting Flows in an Opposed-Piston Two Stroke Engine Shalabh Srivastava, Harold Schock, Farhad Jaberi The two-phase filtered mass density function (FMDF) subgrid-scale model has been used for large eddy simulation (LES) of turbulent spray combustion in a generic single cylinder, opposed-piston, two-stroke engine configuration. The LES/FMDF is implemented via an efficient, hybrid numerical method in which the filtered compressible Navier-Stokes equations are solved with a high-order, multi-block, compact differencing scheme, and the spray and FMDF are implemented with stochastic Lagrangian methods. The reliability and consistency of the numerical methods are established for the engine configuration by comparing the Eulerian and Lagrangian components of the LES/FMDF. The effects of various operating conditions like boost pressure, heat transfer model, fuel spray temperature, nozzle diameter, injection pressure, and injector configuration on the flow field, heat loss and the evolution of spray and combustion are studied. [Preview Abstract] |
Sunday, November 24, 2013 2:54PM - 3:07PM |
D5.00004: Scalable FDF Simulation of Reacting Flows Patrick H. Pisciuneri, S. Levent Yilmaz, Peter A. Strakey, Mehdi B. Nik, Peyman Givi The ``irregularly portioned Lagrangian Monte Carlo-finite difference'' (IPLMCFD) methodology is developed for efficient large eddy simulation (LES) via the filtered density function (FDF) subgrid scale closure. This methodology is particularly suited for simulation of chemically reacting flows and offers efficient utilization of thousands of processors. Various aspects of the scalability are presented for the LES of several premixed and non-premixed turbulent flames at low and high speeds. This method paves the way for petascale LES/FDF. [Preview Abstract] |
Sunday, November 24, 2013 3:07PM - 3:20PM |
D5.00005: Evaluation of the Partially-Averaged Navier-Stokes method for high Mach flows Branislav Basara The performance of the variable-resolution Partially-Averaged Navier-Stokes (PANS) method has been well documented for predictions of separated and wall bounded flows but not for the high speed flows. This will be demonstrated in the present work. Furthermore, some of the latest variants of the PANS models will be discussed as well. For the present study, we use the PANS $\zeta $-f model which is based on the near-wall RANS $\zeta $-f model. This RANS model is a variant of the $v^{2}-f$ model, the difference being that a transport equation for the wall-normal velocity scale ratio $\zeta $ is included rather than one for the velocity scale. Additionally, an elliptic relaxation equation for $f$, which is a parameter closely related to the pressure strain redistribution term, is solved. It is expected that such PANS variant could accurately predict a flow considered here. A test case is a transonic flow over a channel bump. It has been used as a benchmark for turbulence modelling in shock-wave/boundary-layer interactions. Measurements, but also previous RANS calculations, are used as a reference point to the present calculations. [Preview Abstract] |
Sunday, November 24, 2013 3:20PM - 3:33PM |
D5.00006: Large-eddy simulations of impinging jets at high Reynolds numbers Wen Wu, Ugo Piomelli We have performed large-eddy simulations of an impinging jet with embedded azimuthal vortices. We used a hybrid approach in which the near-wall layer is modelled using the RANS equations with the Spalart-Allmaras model, while away from the wall Lagrangian-averaged dynamic eddy-viscosity modelled LES is used. This method allowed us to reach Reynolds numbers that would be prohibitively expensive for wall-resolving LES. First, we compared the results of the hybrid calculation with a wall-resolved one at moderate Reynolds number, $Re=66,000$ (based on jet diameter and velocity). The mean velocity and Reynolds stresses were in good agreement between the simulations, and, in particular, the generation of secondary vorticity at the wall and its liftup were captured well. The simulation cost was reduced by 86\%. We then carried out simulations at $Re=266,000$ and 1.3 million. The effect of Reynolds number on vortex development will be discussed. [Preview Abstract] |
Sunday, November 24, 2013 3:33PM - 3:46PM |
D5.00007: Impact of Model Fidelity on Jet Impingement Simulations Benjamin Reibman, Mike Benson, David Helmer, Gregory Rodebaugh Turbulence modeling in the RANS framework has difficulty properly capturing free jet impingement, as both the shear layer of the jet upon entrance to the free stream and the impingement onto the flat plate are complex turbulent situations. In an effort to assess the accuracy of different turbulence modeling approaches and quantify fidelity/computational cost tradeoffs, a case study was conducted using the experimental data of Cooper et al 1993. This study consisted of both RANS and LES modeling. RANS analysis was conducted over a sweep of case conditions (z/D $=$ 2, 6 and Re $=$ 23000, 70000) leveraging multiple commercial solvers and a variety of common two-equation turbulence models. A grid refinement study was conducted with both structured and unstructured meshes to determine grid and solver dependence. The predictive capability of DDES in addition to WALE and dynamic one-equations k LES SGS closures for the z/D $=$ 2 Re $=$ 23000 case were compared against RANS and experimental data. The experimental data consisting of Nusselt numbers over radial diameters, pipe flow bulk velocity and wall jet velocity were the principal means of determining accuracy of the various models. This study will provide an assessment of the impact of the use of lower fidelity models in future analysis involving more complex geometries. This will provide guidance on the required fidelity at each stage of the design process in multiple fields utilizing impinging jets. [Preview Abstract] |
Sunday, November 24, 2013 3:46PM - 3:59PM |
D5.00008: Simulations of a Normal Shock Train in a Constant Area Duct Using Wall-Modeled LES Zachary Vane, Ivan Bermejo-Moreno, Sanjiva Lele Large-Eddy Simulations (LES) of a turbulent boundary layer interacting with a normal shock train in a constant area duct (STCAD) are performed using an unstructured solver. Comparisons between wall-modeled LES (WMLES) and wall-resolved LES (WRLES) calculations of a spanwise-periodic flow at M$=$1.61 and Re $=$ 16,200 are used to evaluate an equilibrium wall-model's ability to replicate the wall-resolved results. The WMLES approach is then used for simulations at the flow conditions (M$=$1.61, Re$=$162,000) of the Carroll {\&} Dutton STCAD experiments where traditional WRLES was inaccessible. Spanwise-periodic WMLES calculations were unable to duplicate the experimental wall pressure and Laser Doppler Velocimetry data obtained along the spanwise center plane of the duct. Investigations of the full, low aspect ratio duct geometry were then performed using WMLES. Comparisons with experimental data provide an assessment of the wall-model's ability to simulate realistic, high Reynolds number, non-equilibrium flows. However, the lack of information with respect to the sidewall boundary layers from the experiment led to a WMLES parameter study of the effects of boundary layer confinement on the shock train. Initial results suggest that the tunnel blockage due to the boundary layer displacement thickness determines many of the STCAD's characteristics. A possible improvement to the wall-model through the inclusion of previously omitted non-equilibrium terms is currently being pursued. [Preview Abstract] |
Sunday, November 24, 2013 3:59PM - 4:12PM |
D5.00009: CFD predictions of confined turbulent swirling flows in a microscale multi-inlet vortex reactor Michael Olsen, Yanxiang Shi, Gianluca Iaccarino, Rodney Fox Turbulent swirling flows have proven to be an efficient way of promoting mixing for chemical reactions. Accordingly, a multi-inlet vortex reactor was designed for use in the synthesis of nanoparticles. LES and RANS simulations have been applied to understanding the underlying fluid dynamics in this reactor. Both simulations are performed with the open-source platform, OpenFOAM, and the validity of the chosen models are verified against $\mu$-PIV data. For the LES framework, the simplest and yet the most commonly chosen model, Smagorinsky-Lilly model, is used. For the RANS simulations, however, the $k$--$\varepsilon$ model is not capable of capturing the swirling motions. Instead, the four-equation $v^2$--$f$ model is formulated to account for the velocity fluctuations perpendicular to the streamlines and is therefore chosen in this work. Comparisons of the simulation results with the experimental data show both approaches accurately predict the mean velocity fields. Considering the computational cost, the RANS with the $v^2$--$f$ model is recommended for obtaining statistical quantities whereas the LES simulations are more suitable for understanding transient flow behaviors. Based on the validation, the velocity field as well as the turbulence field is also analyzed. [Preview Abstract] |
Sunday, November 24, 2013 4:12PM - 4:25PM |
D5.00010: Large Eddy Simulation for round jet in cross-flow using Local Mesh Refinement Mehtap Cevheri, Thorsten Stoesser The aim of this research is the simulation of near field multi-phase plumes in cross-flows to understand the physical processes of oil spill in Gulf of Mexico. Since this is a multi-phase and multi-scale problem, a local mesh refinement (LMR) technique has been coupled to the multi-grid method to solve the unsteady, incompressible Navier-Stokes problem on a Cartesian grid with staggered variable arrangement. Wall-Adapting Local Eddy Viscosity (WALE) subgrid model has been used to simulate the turbulent flow. In this current study, the verification of the developed code will be presented before the simulation of multi-phase plumes. The accuracy of local mesh refinement and the subgrid model are presented with two test cases: moderate Reynolds number turbulent channel flow and a round turbulent jet into a laminar cross-flow. For the first test case, turbulence statistics for the fully developed turbulent flow are compared with the DNS data. For the second test case, a simulation with a 3.3 velocity ratio and 6930 jet Reynolds number is tested and compared with the experimental and other computational data. [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