Turbulence modelling of problem aerospace flows

Unsteady Reynolds averaged Navier–Stokes (URANS) and detached eddy simulation (DES) related approaches are considered for high angle of attack NACA0012 airfoil, wing–flap, generic tilt‐rotor airfoil and double‐delta geometry flows. These are all found to be problem flows for URANS models. For DES fifth‐order upwinding is found too dissipative and the use of, for high speed flows, instability prone centred differencing essential. An existing hybrid ILES–RANS modelling approach, intended for flexible geometry, relatively high numerical dissipation codes is tested along with differential wall distance algorithms. The former gives promising results. The standard turbulence modelling approaches are found to give perhaps a surprising results variation. Results suggest that for the problem flows, the explicit algebraic stress and Menter shear stress transport (SST) URANS models are more accurate than the economical Spalart–Allmaras (SA). However, the explicit algebraic stress model (EASM) in its k–ε form is impractically expensive to converge. Here, SA predictions lack a rotation correction term and this is likely to improve these results. Copyright © 2005 John Wiley & Sons, Ltd.     

Unsteady CFD simulations of a pump in part load conditions using scale-adaptive simulation

The scope of this work is to demonstrate the applicability of an eddy resolving turbulence model in a turbomachinery configuration. The model combines the Large Eddy Simulation (LES) and the Reynolds Averaged Navier Stokes (RANS) approach. The point of interest of the present investigation is the unsteady rotating stall phenomenon occurring at low part load conditions. Since RANS turbulence models often fail to predict separation correctly, a LES like model is expected to give superior results. In this investigation the scale-adaptive simulation (SAS) model is used. This model avoids the grid dependence appearing in the Detached Eddy Simulation (DES) modelling strategy. The simulations are validated with transient measurement data. The present results demonstrate, that both models are able to predict the major stall frequency at part load. Results are similar for URANS and SAS, with advantages in predicting minor stall frequencies for the turbulence resolving model.     

Hybrid URANS/LES Turbulence Simulation of Vortex Shedding Behind a Triangular Flameholder

In this study a detached eddy simulation (DES) model, which belongs to the group of hybrid URANS/LES turbulence models, is used for the simulation of vortex shedding behind a triangular obstacle. In the near wall region or in regions where the grid resolution is not sufficiently fine to resolve smaller structures, the two-equation RANS shear-stress transport (SST) model is used. In the other regions with higher grid resolution a LES model, which uses a transport equation for the turbulent subgrid energy, is applied. The DES model is first investigated for two standard test cases, namely decaying homogeneous isotropic turbulence and the backward facing step, respectively. For the decaying homogeneous isotropic turbulence test case the evolution of the energy spectra in wavenumber space for different times are studied for both the DES and a Smagorinsky type LES model. Different grid resolutions are analyzed with a special emphasis on the modeling constant connecting the filter length scale to the grid size. The results are compared to experimental data. The backward facing step test case is used to study the model behavior for a case with a transition region between a RANS modeling approach close to the wall and LES based modeling in the intense shear flow region. The final application is the simulation of the vortex shedding behind a triangular obstacle. First, the influence of the inlet condition formulation is studied in detail as they can have a significant influence especially for LES based models. Detailed comparisons between simulation and experiment for the flow structure past the obstacle and statistical quantities such as the shedding frequency are shown. Finally the additional temporal and spatial information provided by the DES model is used to show the predicted anisotropy of turbulence.     

Towards a Unified Turbulence Simulation Approach for Wall-Bounded Flows

A hybrid Reynolds-averaged Navier–Stokes/Large-Eddy Simulation (RANS/LES) methodology has received considerable attention in recent years, especially in its application to wall-bounded flows at high-Reynolds numbers. In the conventional zonal hybrid approach, eddy-viscosity-type RANS and subgrid scale models are applied in the RANS and LES zones, respectively. In contrast, the non-zonal hybrid approach uses only a generalized turbulence model, which provides a unified simulation approach that spans the continuous spectrum of modeling/simulation schemes from RANS to LES. A particular realization of the non-zonal approach, known as partially resolved numerical simulation (PRNS), uses a generalized turbulence model obtained from a rescaling of a conventional RANS model through the introduction of a resolution control function F _R , where F _R is used to characterize the degree of modeling required to represent the unresolved scales of turbulent motion. A new generalized functional form for F _R in PRNS is proposed in this study, and its performance is compared with unsteady RANS (URANS) and LES computations for attached and separated wall-bounded turbulent flows. It is demonstrated that PRNS behaves similarly to LES, but outperforms URANS in general.     

Non-zonal detached eddy simulation coupled with a steady RANS solver in the wall region

Xiao and Jenny (2012) proposed an interesting hybrid LES/RANS method in which they use two solvers and solve the RANS and LES equations in the entire computational domain. In the present work this method is simplified and used as a hybrid RANS-LES method, a wall-modeled LES. The two solvers are employed in the entire domain. Near the walls, the flow is governed by the steady RANS solver; drift terms are added to the DES equations to ensure that the time-averaged DES fields agree with the steady RANS field. Away from the walls, the flow is governed by the DES solver; in this region, the RANS field is set to the time-averaged LES field. The disadvantage of traditional DES models is that the RANS models in the near-wall region – which originally were developed and tuned for steady RANS – are used as URANS models where a large part of the turbulence is resolved. In the present method – where steady RANS is used in the near-wall region – the RANS turbulence models are used in a context for which they were developed. In standard DES methods, the near-wall accuracy can be degraded by the unsteady agitation coming from the LES region. It may in the present method be worth while to use an accurate, advanced RANS model. The EARSM model is used in the steady RANS solver. The new method is called NZ S-DES . It is found to substantially improve the predicting capability of the standard DES. A great advantage of the new model is that it is insensitive to the location of the RANS-LES interface.     

Detached eddy simulation of flow around two wall-mounted cubes in tandem

We investigate the performance of unsteady Reynolds-averaged Navier–Stokes (URANS) computation and various versions of detached eddy simulation (DES) in resolving coherent structures in turbulent flow around two cubes mounted in tandem on a flat plate at Reynolds number (Re) of 22,000 and for a thin incoming boundary layer. Calculations are carried out using four different coherent structure resolving turbulence models: (1) URANS with the Spalart–Allmaras model; (2) the standard DES [Spalart, P.R., Jou, W.H., Strelets, M., Allmaras, S.R., 1997. Comments on the feasibility of LES for wings, and on a hybrid RANS/LES approach. In: Liu, C., Liu, Z., (Eds.), Advances in DNS/LES. Greyden Press, Columbus, OH]; (3) the Delayed DES (DDES); and (4) the DES with a low-Re modification (DES-LR) [Spalart, P., Deck, S., Shur, M., Squires, K., Strelets, M., Travin, A., 2006. A new version of detached eddy simulation, resistant to ambiguous grid densities. Theor. Comput. Fluid Dyn. 20 (3), 181–195]. The grid sensitivity of the computed solutions is examined by carrying out simulations on two successively refined grids. The computed results for all cases are compared with the experimental measurements of Martinuzzi and Havel [Martinuzzi, R., Havel, B., 2000. Turbulent flow around two interfering surface-mounted cubic obstacles in tandem arrangement. ASME J. Fluids Eng. 122, 24–31] for two different cube spacings. All turbulence models reproduce essentially identical separation of the approach thin boundary layer and yield an unsteady horseshoe vortex system consisting of multiple vortices in the leading edge region of the upstream cube. Significant discrepancies between the URANS and all DES solutions are observed, however, in other regions of interest such as the shear layers emanating from the cubes, the inter-cube gap and the downstream wake. Regardless of the grid refinement, URANS fails to capture key features of the mean flow, including the second horseshoe vortex in the upstream junction and recirculating flow on the top surface of the downstream cube for the large cube spacing, and underestimates significantly turbulence statistics in most regions of the flow for both cases. On the coarse mesh, all three DES approaches appear to yield very similar results and fail to reproduce the second horseshoe vortex. The standard DES and DDES solutions obtained on the fine meshes are essentially identical and both suffer from premature switching to unresolved DNS, due to the mis-interpretation of grid refinement as wall proximity, which leads to spurious vortices in the inter-cube region. Numerical solutions show that the low-Re modification (DES-LR) is critical prerequisite in DES on the ambiguously fine – not fine enough for full LES – mesh to prevent excessive nonlinear drop of the subgrid eddy viscosity in low cell-Re regions like in the inter-obstacle gap. Mean flow quantities and turbulence statistics obtained with DES-LR on the fine mesh are in good overall agreement with the measurements in most regions of interest for both cases.     

用基于M-SST模型的DES数值模拟喷流流场

脱体涡数值模拟方法(dettached eddy simulation,DES)是把雷诺平均Navier-Stokes方程(RANS)方法及大涡模拟方法(LES)结合起来模拟有脱体涡的湍流流场的数值模拟方法,其主要思想是在物面附近解雷诺平均Navier-Stokes方程、在其他区域采用Smagorinski大涡模拟方法。本文在剪切应力传输(SST)湍流模型的基础上用DES及混合非结构网格数值模拟具有横向喷流的湍流流场,算法采用Osher逆风格式,利用该套程序(包括网格生成及算法),对导弹在不同马赫数下的喷流流场进行了数值模拟,并与同时开展的实验研究的结果进行了对比,结果表明用该方法处理这类问题是较准确的。     

URANS prediction of flow fluctuations in rod bundle with split-type spacer grid

Turbulent flow in a rod bundle with split-type spacer grid has been studied using Unsteady Reynolds-Averaged Navier–Stokes (URANS) approach. In the previous studies of turbulent flow in rod bundles URANS (as well as steady-state RANS) simulations predicted mean velocity profiles fairly well. However, they severely underpredicted velocity fluctuations, which is investigated in the present study. Our simulations were performed with the Shear Stress Transport (SST) turbulence model and automatic wall-treatment using OpenFOAM, an open-source CFD code. Results of URANS simulations are compared with the measurements of the MATiS-H experiment, which was performed at Korean Atomic Energy Research Institute (KAERI) in 2011–2012.The URANS predictions of velocity fluctuations have been improved by appropriately summing up fluctuations resolved by the basic URANS model and non-resolved fluctuations, which were modelled with the turbulence model. This treatment of turbulent fluctuations, which are directly measured in high-quality experiments, allows more detailed evaluation of various URANS turbulence models. It was found out that the best agreement is achieved when resolved and modelled fluctuations are assumed to be uncorrelated, which indicates that the large-scale structures in this particular flow are distinct in the spectral space from the rest of turbulence. Turbulent flow in the MATiS-H experiment was reproduced by numerous authors using different approaches and our results are among the most accurate.     

On the Application of LES to Seal Geometries

Five Large Eddy Simulation (LES) and hybrid RANS-NLES (Reynolds-Averaged Navier-Stokes-Numerical-LES) methods are used to simulate flow through a labyrinth seal geometry and are contrasted with RANS solutions. Results show that LES and RANS-NLES is capable of accurately predicting flow behaviour of two seal flows with a scatter of less than 5 %. RANS solutions show the potential to perform poorly for the turbulence models tested. LES and hybrid RANS-NLES are found to be consistent and in agreement with measurements, providing a flexible numerical platform for design investigations. It also allows greater flow physics insights.     

Investigation of the sensitivity of turbulent closures and coupling of hybrid RANS‐LES models for predicting flow fields with separation and reattachment

Hybrid models have found widespread applications for simulation of wall‐bounded flows at high Reynolds numbers. Typically, these models employ Reynolds‐averaged Navier–Stokes (RANS) and large eddy simulation (LES) in the near‐body and off‐body regions, respectively. A number of coupling strategies between the RANS and LES regions have been proposed, tested, and applied in the literature with varying degree of success. Linear eddy‐viscosity models (LEVM) are often used for the closure of turbulent stress tensor in RANS and LES regions. LEVM incorrectly predicts the anisotropy of Reynolds normal stress at the RANS‐LES interface region. To overcome this issue, use of non‐linear eddy‐viscosity models (NLEVM) have started receiving attention. In this study, a generic non‐linear blended modeling framework for performing hybrid simulations is proposed. Flow over the periodic hills is used as the test case for model evaluation. This case is chosen due to complex flow physics with simplified geometry. Analysis of the simulations suggests that the non‐linear hybrid models show a better performance than linear hybrid models. It is also observed that the non‐linear closures are less sensitive to the RANS‐LES coupling and grid resolution. Copyright © 2016 John Wiley & Sons, Ltd.     

Applications of URANS on predicting unsteady turbulent separated flows

Accurate prediction of unsteady separated turbulent flows remains one of the toughest tasks and a practi cal challenge for turbulence modeling. In this paper, a 2D flow past a circular cylinder at Reynolds number 3,900 is numerically investigated by using the technique of unsteady RANS （URANS）. Some typical linear and nonlinear eddy viscosity turbulence models （LEVM and NLEVM） and a quadratic explicit algebraic stress model （EASM） are evaluated. Numerical results have shown that a high-performance cubic NLEVM, such as CLS, are superior to the others in simulating turbulent separated flows with unsteady vortex shedding.     

Experiences with two detached eddy simulation approaches through use of a high-order finite volume solver

In the present study, two advanced detached eddy simulation (DES) approaches, shear-layer-adapted delayed DES and zonal DES in mode II, which are known to help transition from RANS to LES mode, are employed in various flow problems in conjunction with a high-order finite volume solver. The numerical scheme, being only applicable on structured grids, has low-dissipation and low-dispersion features. Such features benefit mostly in the LES mode, minimizing the interference of numerical diffusion with subgrid eddy viscosity. First, corresponding subgrid models are validated via decaying homogeneous turbulence benchmark case. Then, a channel flow problem is chosen to examine these models in attached flow situations. Finally, flow around an airfoil at low Reynolds number is solved using the shear-layer-adapted delayed DES approach only, in an aim to obtain trailing-edge noise spectrum at an observer location. Despite some log-layer mismatch over turbulent boundary layers, which is typical of most DES methods, the combined application of high-resolution numerical method and advanced DES approaches, which are implemented on a stabilized Spalart-Allmaras turbulence model, shows merit in resolution of turbulence in regions of interest.     

Simulation of wing-body junction flows with hybrid RANS/LES methods

In this paper, flows past two wing-body junctions, the Rood at zero angle of attack and NASA TN D-712 at 12.5° angle of attack, are investigated with two Reynolds-Averaged Navier-Stokes (RANS) and large eddy simulation (LES) hybrid methods. One is detached eddy simulation (DES) and the other is delayed-DES, both are based on a weakly nonlinear two-equation k–ω model. While the RANS method can predict the mean flow behaviours reasonably accurately, its performance for the turbulent kinetic energy and shear stress, as compared with available experimental data, is not satisfactory. DES, through introducing a length scale in the dissipation terms of the turbulent kinetic energy equation, delivers flow separation, a vortex or the onset of vortex breakdown too early. DDES, with its delayed effect, shows a great improvement in flow structures and turbulence characteristics, and agrees well with measurements.     

A New Version of Detached-eddy Simulation, Resistant to Ambiguous Grid Densities

Detached-eddy simulation (DES) is well understood in thin boundary layers, with the turbulence model in its Reynolds-averaged Navier–Stokes (RANS) mode and flattened grid cells, and in regions of massive separation, with the turbulence model in its large-eddy simulation (LES) mode and grid cells close to isotropic. However its initial formulation, denoted DES97 from here on, can exhibit an incorrect behavior in thick boundary layers and shallow separation regions. This behavior begins when the grid spacing parallel to the wall Δ_∥ becomes less than the boundary-layer thickness δ, either through grid refinement or boundary-layer thickening. The grid spacing is then fine enough for the DES length scale to follow the LES branch (and therefore lower the eddy viscosity below the RANS level), but resolved Reynolds stresses deriving from velocity fluctuations (“LES content”) have not replaced the modeled Reynolds stresses. LES content may be lacking because the resolution is not fine enough to fully support it, and/or because of delays in its generation by instabilities. The depleted stresses reduce the skin friction, which can lead to premature separation.For some research studies in small domains, Δ_∥ is made much smaller than δ, and LES content is generated intentionally. However for natural DES applications in useful domains, it is preferable to over-ride the DES limiter and maintain RANS behavior in boundary layers, independent of Δ_∥ relative to δ. For this purpose, a new version of the technique – referred to as DDES, for Delayed DES – is presented which is based on a simple modification to DES97, similar to one proposed by Menter and Kuntz for the shear–stress transport (SST) model, but applicable to other models. Tests in boundary layers, on a single and a multi-element airfoil, a cylinder, and a backward-facing step demonstrate that RANS function is indeed maintained in thick boundary layers, without preventing LES function after massive separation. The new formulation better fulfills the intent of DES. Two other issues are discussed: the use of DES as a wall model in LES of attached flows, in which the known log-layer mismatch is not resolved by DDES; and a correction that is helpful at low cell Reynolds numbers.     

Application of the SAS turbulence model to predict the unsteady flow field behaviour in a forward centrifugal fan

In the current study, the unsteady flow in a centrifugal fan is carried out using Computational Fluid Dynamics calculation based on the Scale Adaptive Simulation (SAS) approach to model the turbulence phenomenon. The SAS concept is based on the introduction of the von Karman length scale into the turbulence scale equation. The information provided by the von Karman length scale allows SAS models to dynamically adjust to resolved structures in an Unsteady Reynolds-Averaged Navier–Stokes (URANS) simulation, which results in a Large Eddy Simulation-like behaviour in unsteady regions of the flow field. At the same time, the model provides standard RANS capabilities in stable flow regions. The introduction of the von Karman length scale is based on the reformulation of Rottas's equation for the integral length scale. To validate the numerical results, the overall performances of the fan and the wall pressure fluctuations computed upon the volute casing surface are compared with the unsteady measured data.     

Numerical simulations of non‐equilibrium turbulent boundary layer flowing over a bump

Large‐eddy simulation (LES) and Reynolds‐averaged Navier–Stokes simulation (RANS) with different turbulence models (including the standard k?ε, the standard k?ω, the shear stress transport k?ω (SST k?ω), and Spalart–Allmaras (S–A) turbulence models) have been employed to compute the turbulent flow of a two‐dimensional turbulent boundary layer over an unswept bump. The predictions of the simulations were compared with available experimental measurements in the literature. The comparisons of the LES and the SST k?ω model including the mean flow and turbulence stresses are in satisfied agreements with the available measurements. Although the flow experiences a strong adverse pressure gradient along the rear surface, the boundary layer is unique in that intermittent detachment occurring near the wall. The numerical results indicate that the boundary layer is not followed by mean‐flow separation or incipient separation as shown from the numerical results. The resolved turbulent shear stress is in a reasonable agreement with the experimental data, though the computational result of LES shows that its peak is overpredicted near the trailing edge of the bump, while the other used turbulence models, except the standard k?ε, underpredicts it. Analysis of the numerical results from LES confirms the experimental data, in which the existence of internal layers over the bump surface upstream of the summit and along the downstream flat plate. It also demonstrates that the quasi‐step increase in skin friction is due to perturbations in pressure gradient. The surface curvature enhances the near‐wall shear production of turbulent stresses, and is responsible for the formation of the internal layers. The aim of the present work is to examine the response and prediction capability of LES with the dynamic eddy viscosity model as a sub‐grid scale to the complex turbulence structure with the presence of streamline curvature generated by a bumpy surface. Aiming to reduce the computational costs with focus on the mean behavior of the non‐equilibrium turbulent boundary layer of flow over the bump surface, the present investigation also explains the best capability of one of the used RANS turbulence models to capture the driving mechanism for the surprisingly rapid return to equilibrium over the trailing flat plate found in the measurements. Copyright © 2010 John Wiley & Sons, Ltd.     

Hybrid RANS/LES computations of plane impinging jets with DES and PANS models

The qualities of a DES (Detached Eddy Simulation) and a PANS (Partially-Averaged Navier–Stokes) hybrid RANS/LES model, both based on the k–ω RANS turbulence model of Wilcox (2008, “Formulation of the k–ω turbulence model revisited” AIAA J., 46: 2823–2838), are analysed for simulation of plane impinging jets at a high nozzle-plate distance (H/B = 10, Re = 13,500; H is nozzle-plate distance, B is slot width; Reynolds number based on slot width and maximum velocity at nozzle exit) and a low nozzle-plate distance (H/B = 4, Re = 20,000). The mean velocity field, fluctuating velocity components, Reynolds stresses and skin friction at the impingement plate are compared with experimental data and LES (Large Eddy Simulation) results. The k–ω DES model is a double substitution type, following Davidson and Peng (2003, “Hybrid LES–RANS modelling: a one-equation SGS model combined with a k–ω model for predicting recirculating flows” Int. J. Numer. Meth. Fluids, 43: 1003–1018). This means that the turbulent length scale is replaced by the grid size in the destruction term of the k-equation and in the eddy viscosity formula. The k–ω PANS model is derived following Girimaji (2006, “Partially-Averaged Navier–Stokes model for turbulence: a Reynolds-Averaged Navier–Stokes to Direct Numerical Simulation bridging method” J. Appl. Mech., 73: 413–421). The turbulent length scale in the PANS model is constructed from the total turbulent kinetic energy and the sub-filter dissipation rate. Both hybrid models change between RANS (Reynolds-Averaged Navier–Stokes) and LES based on the cube root of the cell volume. The hybrid techniques, in contrast to RANS, are able to reproduce the turbulent flow dynamics in the shear layers of the impacting jet. The change from RANS to LES is much slower however for the PANS model than for the DES model on fine enough grids. This delays the break-up process of the vortices generated in the shear layers with as a consequence that the DES model produces better results than the PANS model.     

End-to-end differentiable learning of turbulence models from indirect observations

The emerging push of the differentiable programming paradigm in scientific computing is conducive to training deep learning turbulence models using indirect observations. This paper demonstrates the viability of this approach and presents an end-to-end differentiable framework for training deep neural networks to learn eddy viscosity models from indirect observations derived from the velocity and pressure fields. The framework consists of a Reynolds-averaged Navier–Stokes(RANS) solver and a neuralnetwork-represented turbulence model, each accompanied by its derivative computations. For computing the sensitivities of the indirect observations to the Reynolds stress field, we use the continuous adjoint equations for the RANS equations, while the gradient of the neural network is obtained via its built-in automatic differentiation capability. We demonstrate the ability of this approach to learn the true underlying turbulence closure when one exists by training models using synthetic velocity data from linear and nonlinear closures. We also train a linear eddy viscosity model using synthetic velocity measurements from direct numerical simulations of the Navier–Stokes equations for which no true underlying linear closure exists. The trained deep-neural-network turbulence model showed predictive capability on similar flows.