Bulletin of the American Physical Society
69th Annual Meeting of the APS Division of Fluid Dynamics
Volume 61, Number 20
Sunday–Tuesday, November 20–22, 2016; Portland, Oregon
Session H29: CFD: LES, DNS, Hybrid RANS/LES |
Hide Abstracts |
Chair: Inanc Senocak, Boise State University Room: F150 |
Monday, November 21, 2016 10:40AM - 10:53AM |
H29.00001: Evaluation of a general hybrid RANS/LES model in smooth wall reattachment sigfried haering, Robert Moser Hybrid RANS/LES modeling approaches often exhibit deficiencies when used for common problems of engineering interest containing flow features such as unsteady smooth-wall separation and reattachment with non-trivial domains and discretization. Often, problem specific modifications and tuning must be employed rendering these models ineffective as generally predictive tools. A new broadly applicable hybrid RANS/LES modeling approach that is being developed to specifically address challenges associated with complex geometries and flows is presented. In general, the approach seeks to a balance between theoretical and actual modeled turbulent kinetic energy provided information from the underlying turbulence model, the resolved turbulence, and the available resolution. Anisotropy in the grid and resolved field are directly integrated into this balance. Here, we examine model performance with the case of a wall-mounted smooth hump of Greenblatt et.al. [1]. Excellent agreement with experimental results is attained while significantly outperforming delayed detached eddy simulation (DDES) for nearly the same computational expense and without any problem-specific modifications. \newline \newline [1] D. Greenblatt, K. B. et. al., ``Experimental investigation of separation control part 1: Baseline and steady suction,'' AIAA Journal, vol. 44, no. 12, pp. 2820--2830, 2006. [Preview Abstract] |
Monday, November 21, 2016 10:53AM - 11:06AM |
H29.00002: A new algebraic transition model based on stress length function Meng-Juan Xiao, Zhen-Su She Transition, as one of the two biggest challenges in turbulence research, is of critical importance for engineering application. For decades, the fundamental research seems to be unable to capture the quantitative details in real transition process. On the other hand, numerous empirical parameters in engineering transition models provide no unified description of the transition under varying physical conditions. Recently, we proposed a symmetry-based approach to canonical wall turbulence based on stress length function, which is here extended to describe the transition via a new algebraic transition model. With a multi-layer analytic form of the stress length function in both the streamwise and wall normal directions, the new model gives rise to accurate description of the mean field and friction coefficient, comparing with both the experimental and DNS results at different inlet conditions. Different types of transition process, such as the transition with varying incoming turbulence intensities or that with blow and suck disturbance, are described by only two or three model parameters, each of which has their own specific physical interpretation. Thus, the model enables one to extract physical information from both experimental and DNS data to reproduce the transition process, which may prelude to a new class of generalized transition model for engineering applications. [Preview Abstract] |
Monday, November 21, 2016 11:06AM - 11:19AM |
H29.00003: Turbulent flow structure response to a varying wall-roughness arrangement: a modelling study Suad Jakirlic, Benjamin Krumbein, Pourya Fooroghi, Franco Magagnato, Bettina Frohnapfel Presently adopted approach to the modelling of rough surfaces relies on introducing an additional drag term in the appropriately `filtered' Navier-Stokes equations, accounting for the form drag and blockage effects, the roughness elements exert on the flow. A non-dimensional drag function D(y) accounting for the shape of roughness elements is introduced. It is evaluated by applying a reference DNS of an open channel flow over a wall characterized by varying arrangement (aligned/staggered) of differently-shaped/sized roughness elements at a bulk Reynolds number Re$=$6500 by Fooroghi et al. (2016, 11th ETMM Symposium; an immersed boundary method is used to resolve the roughness geometry). The prime objective of the present work is to assess the roughness model capability to predict mean velocities and turbulent intensities in conjunction with a recently formulated hybrid LES/RANS (Reynolds-Averaged Navier--Stokes) model (Chang et al., 2014, IJHFF 49), based on the Very Large Eddy Simulation (VLES) concept of Speziale (1998, AIAA J. 36(2)). A seamless transition from RANS to LES is enabled depending on the ratio of the turbulent viscosities associated with the unresolved scales corresponding to the LES cut-off and those related to the turbulent properties of the VLES residual motion. [Preview Abstract] |
Monday, November 21, 2016 11:19AM - 11:32AM |
H29.00004: Detached eddy simulation for turbulent fluid–-structure interaction of moving bodies using the constraint–-based immersed boundary method Nishant Nangia, Amneet P. S. Bhalla, Boyce E. Griffith, Neelesh A. Patankar Flows over bodies of industrial importance often contain both an attached boundary layer region near the structure and a region of massively separated flow near its trailing edge. When simulating these flows with turbulence modeling, the Reynolds--averaged Navier--Stokes (RANS) approach is more efficient in the former, whereas large--eddy simulation (LES) is more accurate in the latter. Detached-eddy simulation (DES), based on the Spalart--Allmaras model, is a hybrid method that switches from RANS mode of solution in attached boundary layers to LES in detached flow regions. Simulations of turbulent flows over moving structures on a body-fitted mesh incur an enormous remeshing cost every time step. The constraint--based immersed boundary (cIB) method eliminates this operation by placing the structure on a Cartesian mesh and enforcing a rigidity constraint as an additional forcing in the Navier--Stokes momentum equation. We outline the formulation and development of a parallel DES--cIB method using adaptive mesh refinement. We show preliminary validation results for flows past stationary bodies with both attached and separated boundary layers along with results for turbulent flows past moving bodies. [Preview Abstract] |
Monday, November 21, 2016 11:32AM - 11:45AM |
H29.00005: A Split Forcing Technique to Reduce Log-layer Mismatch in Wall-modeled Turbulent Channel Flows Rey DeLeon, Inanc Senocak The conventional approach to sustain a flow field in a periodic channel flow seems to be the culprit behind the log-law mismatch problem that has been reported in many studies hybridizing Reynolds-averaged Navier-Stokes (RANS) and large-eddy simulation (LES) techniques, commonly referred to as hybrid RANS-LES. To address this issue, we propose a split-forcing approach that relies only on the conservation of mass principle. We adopt a basic hybrid RANS-LES technique on a coarse mesh with wall-stress boundary conditions to simulate turbulent channel flows at friction Reynolds numbers of 2000 and 5200 and demonstrate good agreement with benchmark data. We also report a duality in velocity scale that is a specific consequence of the split forcing framework applied to hybrid RANS-LES. The first scale is the friction velocity derived from the wall shear stress. The second scale arises in the core LES region, a value different than at the wall. Second-order turbulence statistics agree well with the benchmark data when normalized by the core friction velocity, whereas the friction velocity at the wall remains the appropriate scale for the mean velocity profile. Based on our findings, we suggest reevaluating more sophisticated hybrid RANS-LES approaches within the split-forcing framework. [Preview Abstract] |
Monday, November 21, 2016 11:45AM - 11:58AM |
H29.00006: A comparison study of convective schemes in hybrid RANS-LES calculations Branislav Basara, Zoran Pavlovic Nowadays it is commonly accepted to report on convections schemes in the case of Large Eddy Simulation (LES) calculations. However, in the case of hybrid RANS-LES calculations, the same discussion seems not to be relevant assuming that calculations are anyway performed on the coarser computational meshes and that the amount of unresolved and modelled turbulence impairs the calculation accuracy more than the error of convection schemes used in calculations. Therefore, we want to tackle this issue by using the Partially Averaged Navier--Stokes (PANS) model as the representative hybrid RANS-LES method but the conclusions derived in this work are equally applicable to other models. We will present results by using the central differencing (CD), MINMOD and SMART schemes but also using CD scheme only locally in the area of low unresolved-to-total ratios of kinetic energy (f$_{\mathrm{k}})$. The paper will also show the performance of a step blending function, which depends on the prescribed constant value of the ratio f$_{\mathrm{k}}$ and the performance of a smooth function which directly uses the ratio f$_{\mathrm{k}}$ as the blending value. The results will be presented for the flow around the square cylinder. [Preview Abstract] |
Monday, November 21, 2016 11:58AM - 12:11PM |
H29.00007: Performance investigation of multigrid optimization for DNS-based optimal control problems Cornelia Nita, Stefan Vandewalle, Johan Meyers Optimal control theory in Direct Numerical Simulation (DNS) or Large-Eddy Simulation (LES) of turbulent flow involves large computational cost and memory overhead for the optimization of the controls. In this context, the minimization of the cost functional is typically achieved by employing gradient-based iterative methods such as quasi-Newton, truncated Newton or non-linear conjugate gradient. In the current work, we investigate the multigrid optimization strategy (MGOpt) in order to speed up the convergence of the damped L-BFGS algorithm for DNS-based optimal control problems. The method consists in a hierarchy of optimization problems defined on different representation levels aiming to reduce the computational resources associated with the cost functional improvement on the finest level. We examine the MGOpt efficiency for the optimization of an internal volume force distribution with the goal of reducing the turbulent kinetic energy or increasing the energy extraction in a turbulent wall-bounded flow; problems that are respectively related to drag reduction in boundary layers, or energy extraction in large wind farms. Results indicate that in some cases the multigrid optimization method requires up to a factor two less DNS and adjoint DNS than single-grid damped L-BFGS. [Preview Abstract] |
Monday, November 21, 2016 12:11PM - 12:24PM |
H29.00008: Machine Learning-Assisted Predictions of Turbulent Separated Flows over Airfoils Anand Pratap Singh, Shivaji Medida, Karthik Duraisamy RANS based models are typically found to be lacking in predictive accuracy when applied to complex flows, particularly those involving adverse pressure gradients and flow separation. A modeling paradigm is developed to effectively augment turbulence models by utilizing limited data (such as surface pressures and lift) from physical experiments. The key ingredients of our approach involve Inverse modeling to infer the spatial distribution of model discrepancies, and Neural networks to reconstruct discrepancy information from a large number of inverse problems into corrective model forms. Specifically, we apply the methodology to turbulent flows over airfoils involving flow separation. When the machine learning-generated model forms are embedded within a standard solver setting, we show that much improved predictions can be achieved, even in geometries and flow conditions that were not used in model training. The usage of very limited data (such as the measured lift coefficient) as an input to construct comprehensive model corrections provides a renewed perspective towards the use of vast, but sparse, amounts of available experimental datasets towards the end of developing predictive turbulence models. [Preview Abstract] |
Monday, November 21, 2016 12:24PM - 12:37PM |
H29.00009: A Physics-Informed Machine Learning Framework for RANS-based Predictive Turbulence Modeling Heng Xiao, Jinlong Wu, Jianxun Wang, Julia Ling Numerical models based on the Reynolds-averaged Navier--Stokes (RANS) equations are widely used in turbulent flow simulations in support of engineering design and optimization. In these models, turbulence modeling introduces significant uncertainties in the predictions. In light of the decades-long stagnation encountered by the traditional approach of turbulence model development, data-driven methods have been proposed as a promising alternative. We will present a data-driven, physics-informed machine-learning framework for predictive turbulence modeling based on RANS models. The framework consists of three components: (1) prediction of discrepancies in RANS modeled Reynolds stresses based on machine learning algorithms, (2) propagation of improved Reynolds stresses to quantities of interests with a modified RANS solver, and (3) quantitative, a priori assessment of predictive confidence based on distance metrics in the mean flow feature space. Merits of the proposed framework are demonstrated in a class of flows featuring massive separations. Significant improvements over the baseline RANS predictions are observed. The favorable results suggest that the proposed framework is a promising path toward RANS-based predictive turbulence in the era of big data. (SAND2016-7435 A) [Preview Abstract] |
Monday, November 21, 2016 12:37PM - 12:50PM |
H29.00010: RANS simulations of variable density flows subject to a changing body forces and shocks Rebecca Bertsch, Robert Gore Modeling turbulent mixing in variable density (VD) fluid flows is a key topic of interest in multi-physics applications due to the complex instability characteristics they exhibit. DNS and LES are ideal for studying these types of flows but are computationally expensive. RANS models have developed into accurate and efficient tools to investigate the evolution of turbulence in these complex flow problems and are well validated for prototypical variable density flows such as Rayleigh-Taylor and Richtmyer-Meshkov. However, most lack the ability to accurately capture mix features in VD flows subject to shocks and changing body forces. This talk will present results from a modified RANS model, which substitutes the molecular diffusion term in the species equation with a counter-gradient transport term that is dependent on the turbulent mass flux and species micro-densities. This modification better captures the mix physics across a range of Atwood numbers. Results from the new model will be presented for RM and RT and compared with DNS and experimental 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