Search

International audience; This article presents a data-based methodology to build Reynolds-Averaged Navier-Stokes (RANS) wall models for aerodynamic simulations at low Mach numbers. Like classical approaches, the model is based on nondimensional local quantities derived from the wall friction velocity u τ , the wall viscosity µ w , and the wall density ρ w. A fully-connected neural network approximates the relation u + = f (y + , p +). We consider reference data (obtained with RANS simulations based on fine meshes up to the wall) of attached turbulent flows at various Reynolds numbers over different geometries of bumps, covering a range of wall pressure gradients. After training the neural networks on a subset of the reference data, the paper assesses their ability to accurately recover data for unseen conditions on meshes that have been trimmed from the wall up to an interface height where the learned wall law is applied. The network's interpolation and extrapolation capabilities are quantified and carefully examined. Overall, when tested within its interpolation and extrapolation capabilities, the neural network model shows good robustness and accuracy. The global error on the skin friction coefficient is a few percent and behaves consistently over all the considered test cases.

International audience; This communication describes how datasets for the 1st challenge on Lagrangian Particle Tracking (LPT) and Data Assimilation (DA), held in 2020 and organized within the UE-funded H2020 project HOMER, have been generated. The physical situation, a turbulent wall-bounded flow in the wake of a cylinder, has been simulated using LES with the ONERA HPC solver FASTS. The free-stream and momentum thickness Reynolds numbers were respectively of 665,000 and 4,150; the boundary layer thickness upstream of the cylinder was roughly of 60 mm, with a cylinder diameter of 10 mm, located with a gap of 10 mm from the wall. Virtual particles have been seeded and propagated in the flow using the LES velocity, within the simulation. A virtual experimental domain has been considered downstream of the cylinder, for which a virtual four-camera setup has been used, leading to LPT images. Several acquisition types (two-pulse, four-pulse and time-resolved) and a range of seeding densities have been considered. Depending on the acquisition type, either the particles positions only, or also their velocities and acceleration, have been requested to participants to the challenge. For the DA challenge, particle trajectories have been provided as input, while the requested output consisted in the velocity, velocity gradient and pressure on a prescribed 3D mesh embedded in the experimental domain. Further work is presently underway to complete the website hosting the datasets, towards a full benchmark portal with automatic evaluation, conceived to be progressively completed with other test cases in the long run.; Cette communication présente les principes de la génération des jeux de données formant la base du 1er challenge sur le suivi de particules lagrangien (LPT) et l'assimilation de données (DA), qui s'est tenu en 2020 et était organisé dans le cadre du projet UE H2020 HOME. La situation physique, celle d'un écoulement turbulent pariétal dans le sillage d'un cylindre, a été simulée à l'aide du code HPC FASTS de l'ONERA. L'écoulement était caractérisé par un nombre de Reynolds infini de 665000, et un nombre de Reynolds basé sur l'épaisseur de quantité de mouvement de 4150. L'épaisseur de couche limite était d'environ 60 mm, le diamètre du cylindre étant de 10 mm, placé à 10 mm de hauteur par rapport à la paroi. Des particules virtuelles ont été générées puis propagées à l'aide des vitesses de la LES, de façon intégrée dans la simulation. Un domaine expérimental virtuel a été considéré à l'aval du cylindre, visualisé par un système de quatre caméras virtuelles, menant aux imagers LPT. Différents types d'acquisition (deux pulses, quatre pulses et résolu en temps), et une gamme de différentes densités d'ensemencement ont été considérées. Selon le type d'acquisition, les participants au challenge devaient fournir les positions des particules uniquement, ou bien également leurs vitesses et accélérations. Pour le challenge DA, les trajectoires des particules ont été fournies en entrée, la sortie consistant en la vitesse, le gradient de vitesse et la pression de l'écoulement sur une grille 3D contenue dans le domaine expérimental. Un travail est actuellement en cours pour étendre les capacités du site web utilisé pour télécharger les données, en le rendant capable de fournir une évaluation automatique des données uploadées ainsi qu'un classement des performances ; il sera conçu pour pouvoir être étendu régulièrement à de nouveaux jeux de données.

In this article, we present an immersed boundary method (IBM) for the simulation of compressible flows encountered in aerodynamics. The immersed boundary methods allow the mesh not to conform to obstacles, whose influence is taken into account by modifying the governing equations locally (either by a source term within the equation or by imposing the flow variables or fluxes locally, similarly to a boundary condition).

International audience; In this article, we present an immersed boundary method (IBM) for the simulation of compressible flows of complex geometries encountered in aerodynamics. The Immersed Boundary Methods allow the mesh not to conform to obstacles, whose influence is taken into account by modifying the governing equations locally (either by a source term within the equation or by imposing the flow variables or fluxes locally, similarly to a boundary condition). A main feature of the approach we propose is that it relies on structured Cartesian grids in combination with a dedicated HPC Cartesian solver, taking advantage of their low memory and CPU time requirements but also the automation of the mesh generation and adaptation. Turbulent flow simulations are performed with Reynolds-Averaged Navier-Stokes equations or with Large-Eddy Simulation approach, in combination with a wall function at high Reynolds number, in order to mitigate the cell count resulting from the isotropic nature of Cartesian cells. The objective of this paper is to demonstrate that this automatic workflow is fast and robust and enables to get quantitative aerodynamics results on geometrically complex configurations. Results obtained are in good agreement with classical body-fitted approaches but with a significant reduction of the time of the whole process, that is a day for RANS simulations, including the mesh generation.; Dans cet article, nous présentons une méthode de frontières immergées (IBM) pour des simulations d'écoulements compressibles pour évaluer l'aérodynamique de géométries complexes. Les méthodes de frontières immergées autorisent le maillage à ne pas être conforme aux obstacles et leur présence est alors prise en compte par une modification local des équations (soit par un terme source dans les équations ou par imposition en certains points des variables de l'écoulement ou par imposition du flux en certaines faces). Une caractéristique essentielle de notre approche réside dans le fait qu'elle s'appuie sur des grilles cartésiennes adaptatives structurées. Associée à un solveur HPC cartésien dédié, cette méthode tire parti à la fois de l'automatisation de la méthode de génération et d'adaptation du maillage et du faible coût en terme de temps CPU et de mémoire de la méthode. Par conséquent, nous avons mis en oeuvre une approche rapide et efficace, qui, à partir de la définition des obstacles sous forme discrétisée, permet d'effectuer des simulations CFD autour de géométries complexes en moins d'une journée. Les simulations d'écoulements turbulents sont effectuées par une approche RANS (équations de Navier-Stokes moyennées) ou par une simulation des grandes échelles (LES), associées à un modèle de paroi à grand nombre de Reynolds pour limiter le nombre de points nécessaire, du fait de l'isotropie des mailles cartésiennes.

A large-eddy simulation of laminar transonic buffet on an airfoil at a Mach number $M=0.735$, an angle of attack $\unicode[STIX]{x1D6FC}=4^{\circ }$, a Reynolds number $Re_{c}=3\times 10^{6}$ has been carried out. The boundary layer is laminar up to the shock foot and laminar/turbulent transition occurs in the separation bubble at the shock foot. Contrary to the turbulent case for which wall pressure spectra are characterised by well-marked peaks at low frequencies ($St=f\cdot c/U_{\infty }\simeq 0.06{-}0.07$, where $St$ is the Strouhal number, $f$ the shock oscillation frequency, $c$ the chord length and $U_{\infty }$ the free-stream velocity), in the laminar case, there are also well-marked peaks but at a much higher frequency ($St=1.2$). The shock oscillation amplitude is also lower: 6 % of chord and limited to the shock foot area in the laminar case instead of 20 % with a whole shock oscillation and intermittent boundary layer separation and reattachment in the turbulent case. The analysis of the phase-averaged fields allowed linking of the frequency of the laminar transonic buffet to a separation bubble breathing phenomenon associated with a vortex shedding mechanism. These vortices are convected at $U_{c}/U_{\infty }\simeq 0.4$ (where $U_{c}$ is the convection velocity). The main finding of the present paper is that the higher frequency of the shock oscillation in the laminar regime is due to a different mechanism than in the turbulent one: laminar transonic buffet is due to a separation bubble breathing phenomenon occurring at the shock foot.

International audience; This paper sums up some recent validations of an immersed boundary method for compressible flow simulations. It has been already shown that this method is able to provide accurate results without meshing effort around more or less complex geometries. Here, the authors focus on several simple test cases to assess the application range of the method. From subsonic to hypersonic flows, from steady to large-eddy unsteady simulations, in 2D or 3D, results are compared to classical simulations with body-fitted meshes and experiment. Moreover, the use of an efficient Navier-Stokes solver for all the cases is part of the benefit of the approach to provide a CFD solution in a reduced time.; Ce paapier présente les validations récentes d'une méthode de frontières immergées pour des simulations CFD d'écoulement compressible. Il a été déjà montré que cette méthode est capable de fournir des résultats précis sans effort de maillage autour de géométries plus ou moins complexes. Ici, les auteurs s'intéressent à plusieurs cas-test simples afin de démontrer la gamme d'applications de cette méthode. Pour des écoulements allant du subsonique à l'hypersonique, pour des simulations stationnaires ou des simulations des grandes échelles instationnaires, en 2D ou en 3D, les résultats sont comparés à des simulations classiques avec des maillages conformes à la paroi et aux données expérimentales. De plus, l'utilisation d'un code Navier-Stokes efficace pour toutes ces configurations est un ingrédient important au bénéfice de la méthode pour fournir une solution CFD en un temps réduit.

International audience; The scattering of an acoustic wave by a 3D spatially-developing plane turbulent mixing layer is investigated by means of a Large Eddy Simulation (LES). First, a plane turbulent mixing layer of initial Reynolds number Reω, 0 ≈ 1200 and convective Mach number Mc ≈ 0.12 is computed and its characteristics are validated by comparisons to experimental and numerical studies. Then, an acoustic source is introduced in the computational domain and the LES of both the turbulent mixing layer and the acoustic field is carried out, allowing the direct computation of the scattered pressure field. Computational Aero Acoustics (CAA) methods have been implemented to minimize spurious pressure fluctuations created while the turbulent structures leave the computational domain. The side-lobes of the scattered pressure power spectral density are recovered and a parametric study, involving the convection velocity of the large turbulent structures, the amplitude and the tonal frequency of the source is carried out. The part of scattered energy, the frequency shift of the side-lobes and the spectra decrease are found to be correctly estimated. Analysis of directivity at the side-lobes frequencies show a progressive re-distribution of the acoustic energy, as the acoustic wave propagates through the turbulent shear layer.

An improvement of the efficiency of implicit schemes based on Newton-like methods for the simulation of turbulent flows by compressible LES or DNS is proposed. It hinges on a zonal Self-Adaptive Newton method (hereafter denoted SAN), capable of taking advantage of Newton convergence rate heterogeneities in multi-scale flow configurations due to a strong spatial variation of the mesh resolution, such as transitional or turbulent flows controlled by small actuators or passive devices. Thanks to a predictor of the local Newton convergence rate, SAN provides computational savings by allocating resources in regions where they are most needed. The consistency with explicit time integration and the efficiency of the method are checked in three test cases: – The standard test-case of 2-D linear advection of a vortex, on three different two-block grids. – Transition to 3-D turbulence on the lee-side of an airfoil at high angle of attack, which features a challenging laminar separation bubble with a turbulent reattachment. – A passively-controlled turbulent transonic cavity flow, for which the CPU time is reduced by a factor of 10 with respect to the baseline algorithm, illustrates the interest of the proposed algorithm.

Large Eddy Simulation of the bubble bursting process over a NACA-0012 airfoil at $Re_{c}=10^{5}$ indicates that the flow at a fixed angle of attack below the critical stall value exhibits a short (with respect to Gaster’s criteria, Gaster, Number CP-4 in AGARD, 1966) Laminar Separation Bubble (LSB) at the leading edge of the airfoil. The airfoil is smoothly pitched-up through the static stall angle to reproduce the bursting process of the short LSB that initiates a leading edge stall typical of low Reynolds number airfoil. The temporal evolution of characteristic length scales is monitored during the transient flow. Particular attention is paid to the characteristic time involved during the growth and bursting of the LSB. A recent empirical bursting criterion is used to analyse the LES results.

A methodology to mix Chimera and Immersed Boundary approaches in the same CFD computation is presented in order to deal with complex geometries. Where accuracy is required in a vicinity of an obstacle, a body-fitted mesh is used to accurately simulate the boundary layer. Otherwise, an immersed boundary method (IBM) is applied to deal with this obstacle. Each body-fitted mesh is generated independently in an overset grid fashion. A set of Cartesian grids that are adapted to the spatial resolution of body-fitted grids and immersed boundaries is generated automatically and overlaps the body-fitted grids. In this paper, we describe the overall procedure to mix the overset grid and immersed boundary approaches, then we detail how the present IBM enables to deal with geometrically complex configurations. Finally, preliminary results of the simulation of a turbulent flow past two cylinders in tandem are presented.

The present study consists in an analysis of the DNS database of a flow overcoming a transitional separation induced by an adverse pressure gradient on a flat plate under a curved upper wall. This study takes place in the context of improving RANS models for the simulation of the stall phenomenon for rotor blades applications. To mimic the real flow mechanisms, the flow characteristics are chosen to be typical of the leading edge of the OA209 airfoil at an incidence just below stall occurence: 15° of incidence, R c ∞ = 1.8 × 10 6 and M ∞ = 0.16 . The budgets of the turbulent kinetic energy transport equations have been computed by the DNS. Their evolution is analysed from the separation up to the downstream turbulent flow. Comparisons with a URANS k–ω Wilcox computation are presented to illustrate the application of the DNS simulation.

The objective of this current study is to investigate the course of events leading to stall just before its occurrence. The stall mechanisms are very sensitive to the transition that the boundary layer undergoes near the leading edge of the profile by a Laminar Separation Bubble (LSB). To provide helpful insights into this complex flow, different LES of the flow around an airfoil near stall have been achieved. Attention has been given to the transition mechanism in the LSB. In particular, the results are successfully compared to the linear stability theory. Furthermore, a zonal Reynolds averaged Navier–Stokes/large eddy simulation (RANS/LES) hybrid method has been employed for the same flow configuration to resolve more accurately the transitional flow than with the RANS approach. The analysis of the results highlights the strong impact of the LSB structure on the downstream boundary layer.

The objective of the current study is to examine the course of events leading to stall just before its occurrence. The stall mechanisms are very sensitive to the transition that the boundary layer undergoes near the leading edge of the profile by a so-called laminar separation bubble (LSB). In order to provide helpful insights into this complex flow, a zonal Reynolds-averaged Navier–Stokes (RANS)/large-eddy simulation (LES) simulation of the flow around an airfoil near stall has been achieved and its results are presented and analyzed in this paper. LSB has already been numerically studied by direct numerical simulation (DNS) or LES, but for a flat plate with an adverse pressure gradient only. We intend, in this paper, to achieve a detailed analysis of the transition process by a LSB in more realistic conditions. The comparison with a linear instability analysis has shown that the numerical instability mechanism in the LSB provides the expected frequency of the perturbations. Furthermore, the right order of magnitude for the turbulence intensities at the reattachment point is found.

The scattering of an acoustic wave by a fully turbulent mixing layer is investigated by means of a Large Eddy Simulation. The turbulent characteristics of the mixing layer have been validated by comparison to experimental and numerical studies. Computational Aero Acoustics (CAA) methods have been used to minimize spurious pression fluctuations created while the turbulent structures leave the computational domain. Then, an acoustic source has been placed inside a uniform flow and the acoustic propagation has been validated. Finally, both the turbulent mixing layer and the acoustic field are calculated and the scattering phenomenon is investigated. The ’double-humped’ structures of the far field power spectral density is recovered and a parametrical study, involving the tonal frequency of the source is carried out. The comparison to previous experimental works have been enabled thanks to the definition of a relevant scattering parameter. The spectra are in good agreement with experimental works on acoustic scattering. The frequency shift of the sidelobes and the spectra’s decrease are well predicted and in a good agreement with the ones found in the literature.

A large eddy simulation (LES) of a turbulent flow past a 70° sweep angle delta wing is performed and compared with wind tunnel experiments. The angle of attack and the Reynolds number based on the root chord are equal to 27° and 1.6×10 6 , respectively. Due to the high value of the Reynolds number and the three-dimensional geometry, the mesh resolution usually required by LES cannot be reached. Therefore a local mesh refinement technique based on semi-structured grids is proposed, whereas different wall functions are assessed in this paper. The goal is to evaluate if these techniques are sufficient to provide an accurate solution of such flow on available supercomputers. An implicit Miles model is retained for the subgrid scale (SGS) modelling because the resolution is too coarse to take advantage of more sophisticated SGS models. The solution sensitivity to grid refinement in the streamwise and wall normal direction is investigated.

Large-eddy simulation ofcompressible Navier-Stokes equations is used to study flows where a laminar boundary-layer separation is followed by a turbulent reattachment. The aim of the present work is to predict and describe the transition process and its interaction with the wake dynamics for a subsonic blade turbine configuration. Indeed, a better knowledge of this mechanism can help to improve the accuracy of a Reynolds-averaged Navier-Stokes turbulence model for such a flow case. High-resolution large-eddy-simulation-type computations have been carried out for the T106 low-pressure blade turbine at inlet Mach number of 0.1 and chord Reynolds number of 1.6 x 10 5 based on the exit isentropic velocity. The simulated mean and turbulent quantities compare well with the available experimental data. The primary two-dimensional instability that originates from the free shear in the bubble is unstable via the Kelvin-Helmholtz mechanism. Then, the three-dimensional motions spread on the boundary layer, leading to full breakdown to turbulence after the reattachment point

Large-eddy simulations of the flow over a deep cavity are performed. The computations reproduce identically all the parameters of the experiment by Forestier and co-workers [J. Fluid Mech. (to be published)], including the high Reynolds number ReL=8.6×105. Spectra show an accurate prediction of the peak levels of the fundamental frequency and its first harmonics. Results are also analyzed both in terms of Reynolds and phase averages, the procedure used to compute phase averages being identical to the one used during the experiment. Agreement with the experimental data is found to be excellent. The expansion rate of the shear layer is accurately described, and the temporal physics of the flow, including the dynamics of the coherent structures, is fully recovered. By comparison with an auxiliary computation wherein the wind-tunnel upper wall is not taken into account, the cavity is found to oscillate in a flow-acoustic resonance mode. New values for the γ constant of Rossiter’s model are then proposed for a...

A large eddy simulation (LES) of a turbulent flow past an airfoil near stall at a chord Reynolds number of 2.1 x 10 6 is performed and compared with wind-tunnel experiments. This configuration still constitutes a challenging test case for Reynolds-averaged Navier-Stokes (RANS) simulation and LES as a result of the complexity of the suction side boundary layer: an adverse pressure gradient creates successively a laminar separation bubble, a turbulent reattachment, and a turbulent separation near the trailing edge. To handle this high-Reynolds-number flow with LES on available supercomputers, a local mesh-refinement technique and a discretization of the convective fluxes are developed in a block-structured finite volume code to reduce the total number of grid points and the numerical dissipation acting on the small scales, respectively. Influence of subgrid scale modeling (SGS) is assessed through the comparisons of explicit selective mixed scale model (SMSM) and implicit monotone-integrated LES model results. Moreover, the solution sensitiveness to grid refinement and spanwise extent is investigated

An algorithm for the simulation of unsteady, viscous, stratified compressible flows, which remains valid at all speeds, is presented. The method is second-order accurate in both space and time and is independent of the Mach number. In order to remove the stiffness of the numerical problem due to the large disparity between the flow speed and the acoustic wave speed at low Mach number, an approximate Newton method, based on artificial compressibility, is proposed. Additionally, a modified advection upstream splitting method (AUSM +) scheme is used, which permits accurate computations of both compressible and incompressible flows. A detailed description of the method and an efficiency comparison with other approximate Newton methods described in the literature are given. Furthermore, it is shown that the accuracy of the algorithm is not dependent on the Mach number through the computations of various benchmark test cases. Copyright (C) 2000 John Wiley & Sons, Ltd.

The paper presents hybrid LES/RANS computations of turbulent flows with a segregated approach. This approach employs strict steady RANS and strict LES in pre-defined regions of the computational domain coupling the solution between them by specifically designed interfaces. The latter can be obtained by enhancement of the interfaces needed for domain decomposition in any block structured code. The paper covers inflow and outflow conditions of the LES subdomain when linked to a RANS domain as well as tangential coupling. Compressible as well as incompressible simulations are reported.

In the context of the AEROCAV French Research Program, Large-Eddy Simulations were conducted to compute the turbulent structures and the noise generated by the flow over realistic cavities, similar to those present on air craft fuselages and wings. The radiated acoustic field was computed using a Kirchhoff integ ral method. Two geometrical configurations are considered, both with cylindrical shapes but with different Depth/Diameter aspect ratio (H/D = 1 and H/D = 0.5 ) embedded in a subsonic flow ( Mach = 0.2 ) and for a Reynolds number based on the length of the cavity equal to 4.6·10 5 . Dedicated experimental aerodynamic and aeroacoustic measurements databases are used to validate the numerical computations. Experimental and numerical flows are similar except that the boundary layer upstream the cavity is turbulent in the experiments and laminar for the CFD. The aerodynamic fields in the two sets of results are in good agreement, whereas the acoustic field is largely influenced by the install ation effects. These effects are analyzed through computations based on Boundary Element Method and CAA and then corrections factors are provided. The acoustic emission for the H/D = 1 case is characterized by a discrete tonal mode, as a superposition of structur al and Rossiter resonance phenomena, while the H/D = 0.5 case presents a directive and highly dissymmetrica l broadband noise emission. A numerical process is also proposed to c ontrol this dissymmetry.

A high-resolved Large Eddy Simulation (LES) of the flow around an airfoil near stall has been achieved. We have observed that the laminar boundary layer undergoes a quick transition to turbulence in a Laminar Separation Bubble (LSB) close to the leading edge of the airfoil profile. The flow structures in this transitional flow region have been analysed and the transition mechanism seems accurately resolved. Hence, a reliable statistical data base of the transitional flow has been obtained. Then, the ability of the RANS method to reproduce the LES results, using a rough transition model, is analysed.

The purpose of this paper is to investigate the acoustical field generated by the flow over a cylindrical cavity on a flat plane, similar to those located on wings and fuselage of transport aircraft, using a hybrid numerical method. Firstly, a CFD computation was carried out, using a Large-Eddy Simulation method, to compute the turbulent structures and the noise generating mechanisms. Secondly, the near-field results were injected in a Kirchhoff integral method code, in order to compute the far-field noise radiation. The retained case for this study is represented by a cavity with a unitary depth/diameter ratio, in the presence of a subsonic flow with a Reynolds number based on the diameter of the cavity equal to 4.6x10 5 . As inflow condition, a fluctuation-free velocity profile was considered. Computational results were compared to experimental data measured in an anechoic wind tunnel for the same geometry configuration, but in the presence of a turbulent boundary layer. As expected, a symmetrical flow pattern was observed joined with a fluid resonant mechanism. The computation shows that the acoustical field is less influenced by the inflow condition and results are in good agreement with the experimental measurements despite a slight under-estimation of the main tone amplitude.

An algorithm is proposed for the simulation of unsteady viscous stratified compressible flows. The advantage of the method is its capability to deal with a broad range of subsonic Mach numbers, including nearly incompressible flows, with a single modelling, based on the fully compressible Navier-Stokes equations. The method is second-order accurate both in space and time. To remove the stiffness of the numerical problem due to the large disparity between the flow and the acoustic wave speeds at low Mach number, an approximate Newton method, based on artificial compressibility, is used. After a detailed description of the method, the accuracy of the algorithm is checked by computing compressible and incompressible benchmark test cases. Results of flows over a backward facing step with or without stratification effects for a Reynolds number of 800 are compared with the steady solution of incompressible methods, The computation of a compressible natural convection allows to evaluate the accuracy of the method for flow including significant compressibility effects, whereas numerical results of a Poiseuille-Benard channel flow for a Reynolds number of 10 demonstrate the efficiency of the present algorithm to compute unsteady hows. (C) 2000 Elsevier Science Ltd. All rights reserved.

A 30 dB reduction of the peak pressure tone and a reduction by 6 dB of the background pressure found in an experiment of high-subsonic cavity flow controlled by a spanwise rod are retrieved numerically. The injection of deterministic upstream fluctuations in the large-eddy simulation (LES) domain is found to be of crucial importance, in contrast with the baseflow case. Reduction of the vortex impingement onto the aft edge of the cavity is confirmed, together with reduction of mass flow rate breathing through the grazing plane. Visual evidence of merging between the Kelvin–Helmholtz-type vortices shed downstream of the fore edge of the cavity and the von Karman vortices shed behind the cylinder is provided. Shocklets downstream of the cylinder are also observed.

Institute for Technical Chemistry and Polymer Chemistry, University ofKarlsruhe, 76128 Karlsruhe, GermanySummary. The paper proposes two methods to couple an LES zone with a down-stream RANS zone. Both employ eddy viscosity models. One is based on a filtering–enrichment procedure, the other on an intermediate damping zone. Simulations ofdeveloped plane channel flow at

An improvement of an implicit scheme for Large-Eddy Simulation is proposed in this paper. This local optimization (LO) of the convergence is based on the reduction of the high computational eort per time step inherent to the use of implicit time stepping methods. Indeed, the LO is a local adpatation of the iteration number in the inner process. Moreover an implicitation of the boundary condition between two domains where the iteration number is dieren t, is also used. This procedure is validated on two test cases. The rst one is a 2D laminar vortex advection which allows us to check the potential of this LO. On a transitional boundary layer around an airfoil near stall at a chord Reynolds number of 2:1 10 6 , it is shown that this procedure can be used without damaging the accuracy.

The use of a database as unsteady ino w condition for turbulent boundary layers of complex o w is evaluated. This technique is employed to achieve a full zonal coupling between a Reynolds average Navier Stokes simulation (RANS) and a large-eddy simulation (LES). The database is rst extracted from a plane of a compressible turbulent at plate boundary layer and contains all conservative variables. Scaling functions are developed to use this database as ino w conditions for a at plate at dieren t Reynolds and Mach numbers. The inuence on the solution of several parameters, such as the database duration or the time-step between two samples of the database, is discussed. Some results are compared with those given by the recycling rescaling method of Lund, Wu and Squires. 9

The present study evaluates a new method to reconstruct the Reynolds Average NavierStokes (RANS) eddy viscosity corresponding to a flow eld stemming from a Large Eddy Simulation (LES) computation. Within the framework of zonal RANS/LES coupling, the goal is to obtain a more precise RANS eddy viscosity at the interface between LES zones when the RANS zone is located downstream the LES zone. This technique is rst performed on a simple flow test case, a compressible channel flow at Re = 180. Then the technique is carried out for a realistic flow test case, which corresponds to the flow around an airfoil at Re =2 :1 10 6 .

The aim of the present work is to predict and describe the boundary layer transition process influenced by an incoming upstream turbulent wake generated by a moving cylinder for a subsonic blade turbine configuration. High-resolution Large-Eddy Simulation type computations have been carried out for the T106 low-pressure blade turbine at Mach number of 0.1 and chord based Reynolds number of 1.6 105 and have been compared with experiments.

This paper describes a new step in the development of a Computational AeroAcoustics (CAA) process whose general long term objective is the numerical prediction of the aerodynamic sound radiated by the airframe of large aircraft at approach, and especially the noise generated by deployed high lift devices such as slats and flaps. The proposed 3-step hybrid process combines CFD (Computational Fluid Dynamics) techniques and acoustic numerical methods, each one being adapted to a particular domain in which specific physical fluid mechanisms are simulated solving an adequate set of equations. In a first step, the nearfield unsteady flow is computed via a compressible three-dimensional LES (Large Eddy Simulation). In a second step, LES-computed perturbations are injected at the inner boundary of a larger domain in which the outward propagation of small perturbations over a non-uniform mean flow is simulated using LEE (linearized Euler equations). In the third and last step, the acoustic field radiated at the external boundary of the LEE domain becomes the entry data of a Kirchhoff integration which provides the noise radiated in the far field. The critical point of the process is the coupling, via an interface, of the LES with the LEE. This process has been carefully studied using analytical fields, an acoustic point source monopole and a convected Eulerian vortex. It has been found that the correct injection of such fields requires severe conditions in terms of space resolution, conditions which are especially difficult to meet for purely vortical fields. In a former study, the LES of the unsteady flow around a NACA0012 airfoil has formed the basis of numerical noise predictions using acoustic integral methods. In the present paper, the same LES is used as a basis for the 3-step CAA process. First results revealed the generation of non-physical noise at the boundary interface where the airfoil's turbulent wake is injected in the Euler domain. Additional tests based on the injection of an analytical vortex suggest that this problem was most probably caused by the under-resolution of the injected vortical structures. This difficulty was not solved, but by-passed by using a LES/LEE interface which did not intercept the airfoil's wake. The final result integrates the three components, including the nearfield LES, the midfield noise propagation using LEE and the farfield noise radiation using the Kirchhoff integral. __________________________________________________ Copyright © 2002 by ONERA. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission INTRODUCTION The general context of this paper is the numerical prediction of the aerodynamic noise generated by the high lift devices HLD, slats and flaps of large airliners, an important contributor to the total radiated airframe noise, especially in approach configuration. It is commonly admitted that the design of new low-noise HLD concepts incorporating specific noise reduction devices, although still relying on necessary experiments, will take growing advantage of the numerical simulation in terms of lower costs and shorter delays, especially considering the spectacular continuing progress of Computational AeroAcoustics (CAA) methods. The problem of the numerical simulation of HLD noise is still beyond the capabilities of complete Direct Numerical Simulation (DNS), so hybrid methods are used in most practical cases. Figure 1 sketches the possible numerical strategies, showing how the nearfield turbulent flow and the farfield noise are computed separately. The idea is to divide the physical space into several domains, in which specific physical mechanisms are simulated using the most adequate set of equations with the cheapest discretization strategy. Computational Fluid Dynamics (CFD) techniques are used to simulate the nearfield flow which contains the aerodynamic noise sources. Available techniques include steady ReynoldsAveraged Navier-Stokes (RANS) computations, in conjonction with stochastic models of the wavenumber-frequency spectrum of the turbulence [1-3], unsteady RANS methods [4-5], and Large Eddy Simulation (LES) [6-9]. This local flow solution has to be coupled to an acoustic numerical technique for the prediction of farfield noise. The most practical formulations are the integral methods such as Lighthill's analogy [7] [10] (including the Ffowcs WilliamsHawkings (FW-H) equation [4, 5, 11, 12]), the Boundary Element Method (BEM) [13] and the Kirchhoff integral. In a former study, the compressible LES of the unsteady flow around a symmetrical NACA0012 airfoil with a blunted trailing edge has formed the basis of airfoil aerodynamic noise predictions. A detailed analysis of the nearfield unsteady flow showed that the local aeroacoustic characteristics were correctly simulated, including the local acoustic field. This suggested to define a control surface around the airfoil, on which the acoustic nature of the pressure field was established. The pressure field and normal derivative on this surface where used to compute the farfield noise via a 3D Kirchhoff method. In a second step, another noise prediction based on the Ffowcs WilliamsHawkings equation was performed using the same LES data. 8th AIAA/CEAS Aeroacoustics Conference & Exhibit Fire 17-19 June 2002, Breckenridge, Colorado AIAA 2002-2573 Copyright © 2002 by the author(s). Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

A Large Eddy Simulation (LES) of a turbulent flow past a 70° sweep angle delta wing is performed and compared with windtunnel experiments. The angle of attack and the Reynolds number based on the root chord are equal to 27° and 1.6 × 106, respectively. Due to the high value of the Reynolds number and the three-dimensional geometry, the mesh resolution usually required by LES cannot be reached. Therefore a local mesh refinement technique, based on semi-structured grids, and some wall functions are assessed in this paper. The goal is to evaluate if these techniques are sufficient to provide accurate solution of such flow on available supercomputers. An implicit Miles model is retained for the Subgrid Scale (SGS) modelling, because the resolution is too coarse to take advantage of more sophisticated SGS models. The solution sensitiveness to grid refinement in the streamwise and wall normal direction is investigated.

A Large Eddy Simulation (LES) of a turbulent flow over the A-airfoil near stall at a chord Reynolds number of 2.1 x 10 is performed and compared with windtunnel experiments. This configuration still constitutes a challenging test case for Reynolds Average Navier-Stokes (RANS) simulation and LES due to the complexity of the suction side boundary layer: an adverse pressure gradient creates successively a laminar separation bubble, a turbulent reattachment and a turbulent separation near the trailing edge. To handle this high Reynolds number flow with LES on available supercomputer, a local mesh refinement technique and a discretization of the convective fluxes are developed in a block-structured finite volume code to reduce the total number of grid points and the numerical dissipation acting on the small scales, respectively. Influence of Subgrid Scale Modelling (SGS) is assessed through the comparisons of explicit Selective Mixed Scale Model (SMSM) and implicit MiLES model results. Moreover the solution sensitivity to grid refinement and spanwise extent is investigated. With the use of the largest grid (7.2 x 10 cells) and SMSM model, the computed mean and fluctuating velocity profiles compares favourably with experimental measurements, which constitutes to the authors knowledge the first satisfying LES of this complex flow.

A Large Eddy Simulation (LES) of a turbulent flow over the A-airfoil near stall at a chord Reynolds number of 2.1 × 106 is performed and compared to windtunnel experiments. This configuration still constitutes a challenging test case for Reynolds Averaged Navier-Stokes (RANS) simulation and LES due to the complexity of the suction side boundary layer: an adverse pressure gradient creates successively a laminar separation bubble, a turbulent reattachment and a separation near the trailing edge. To handle this high Reynolds number flow with LES on available supercomputers, a local mesh refinement technology and a low dissipative scheme are developed. It appears that the computed mean and fluctuating velocity profiles compare reasonably well with experimental measurements.

The aim of the present work is to predict and describe the transition process and its interaction with the wake dynamics for a subsonic blade turbine configuration. High-resolution Large-Eddy Simulation type computations have been carried out for the T106 low-pressure blade turbine at Mach number of 0.1 and chord Reynolds number of 1.6 105 and have been compared with experiments.