Skip to main content
main-content
Top

About this book

This book gathers the proceedings of the Seventh Symposium on Hybrid RANS-LES Methods, which was held on September 17-19 in Berlin, Germany. The different chapters, written by leading experts, reports on the most recent developments in flow physics modelling, and gives a special emphasis to industrially relevant applications of hybrid RANS-LES methods and other turbulence-resolving modelling approaches. The book addresses academic researchers, graduate students, industrial engineers, as well as industrial R&D managers and consultants dealing with turbulence modelling, simulation and measurement, and with multidisciplinary applications of computational fluid dynamics (CFD), such as flow control, aero-acoustics, aero-elasticity and CFD-based multidisciplinary optimization. It discusses in particular advanced hybrid RANS-LES methods. Further topics include wall-modelled Large Eddy Simulation (WMLES) methods, embedded LES, Lattice-Bolzman methods and turbulence-resolving applications and a comparison of the LES methods with both hybrid RANS-LES and URANS methods. Overall, the book provides readers with a snapshot on the state-of-the-art in CFD and turbulence modelling, with a special focus to hybrid RANS-LES methods and their industrial applications.

Table of Contents

Frontmatter

Keynotes

Frontmatter

Performance of Under-Resolved, Model-Free LBM Simulations in Turbulent Shear Flows

The paper outlines the predictive capabilities of lattice Boltzmann methods (LBM) in turbulent shear flows. Attention is devoted to a specific collision operator which relaxes the distribution functions in cumulant space. The study highlights the benefits of a carefully defined discrete collision operator by scrutinizing the numerical stability and the predictive accuracy for a wide scope of resolutions—ranging from DNS to RANS—when no ad hoc turbulence closure is employed. Examples included are concerned with two frequently computed fundamental flows, i.e. Taylor-Green vortex and channel flows. Results reveal a fair accuracy and a remarkably small resolution dependence for the investigated cumulant collision operator, which is quite the contrary for other collision models.

Martin Gehrke, Amir Banari, Thomas Rung

Development of Advanced Flow Diagnostic Techniques to Characterize Aircraft Icing Phenomena

Aircraft icing phenomena involve in complicated interactions among multiphase flows (i.e., gaseous airflow, super-cooled liquid water droplets/surface water film flows, and ice accreting solid surface) coupled with unsteady heat transfer (i.e., the release of the latent heat of fusion) and dynamic phase changing (i.e., solidification) processes. In the present study, the research progress made in our recent efforts to develop advanced flow diagnostic techniques to characterize the complex multiphase flows pertinent to aircraft icing phenomena is presented. A novel lifetime-based molecular tagging thermometry (MTT) is introduced to achieve simultaneous measurements of droplet size (in terms of volume, height, contact area and the contact angle of the droplet) and temporally-and-spatially-resolved temperature distributions within micro-sized, icing water droplets to quantify the unsteady heat transfer and phase changing process pertinent to the ice formation and accretion processes as water droplets impinge onto frozen cold aircraft wings. A novel structure-light-based digital image projection (DIP) technique is also introduced to achieve quantitative measurements of the droplet/film thickness distributions to quantify the dynamic surface water runback process pertinent to glaze ice accretion process over an airfoil/wing surface. The quantitative measurements are very helpful to elucidate the underlying physics pertinent aircraft icing phenomena as well as to develop more effective and robust anti-/de-icing strategies for aircraft icing mitigation.

Hui Hu

Hierarchical Zonal Industrial Turbulence and Geometry Modelling Framework

In a hierarchical fashion both the handling of turbulence and geometry are considered. The latter is necessary to help more economically deal with the increasingly coupled nature of many aerodynamic problems and also the drive towards considering ever increasing levels of geometrical complexity. The hybridization of RANS (Reynolds Averaged Navier-Stokes) and LES (Large Eddy Simulation), in various forms is explored. In relation to this, a taxonomy is presented. These aspects have all been presented here with a focus more on turbomachinery. The design needs for future engines is explored and it is discussed how these need to be met in a very integrated way, encompassing global installations all the way to avionics systems. However, it is believed that the applicability of these ideas goes beyond turbomachinery and is relevant to many other industrial applications. It is expected that the combination of these ideas will allow engineers to appropriately perform eddy resolving simulations in systems where there is significant aerodynamic coupling and a high level of geometrical complexity. The proposed unified framework could be exploited all the way through initial fast preliminary design to final numerical test involving various bespoke combinations of hierarchical components.

P. G. Tucker

Turbomachinery Research and Design: The Role of DNS and LES in Industry

The role of high-fidelity CFD in industry is rapidly evolving due to the growth of computational power. Direct and large eddy simulations of realistic turbomachinery flows are now possible to analyze not only fundamental problems, but also to investigate compressor and turbine design spaces. Nevertheless, it is practically impossible to replace conventional Reynolds averaged models with scale resolving simulations in the framework of industrial design iterations. Along these lines, this paper describes how scale resolving simulations can have a direct impact on both the design and the design tools of turbomachinery components. The presented results prove how high-fidelity simulations can explain unsteady loss generation, and how advanced post-processing indicates performance top offenders. When coupled with machine-learning, scale-resolved simulations are also capable of improving the accuracy Reynolds averaged models routinely used in design work.

Vittorio Michelassi

Scale-Resolving Modelling

Frontmatter

Hybrid RANS/LES of an Adverse Pressure Gradient Turbulent Boundary Layer Using an Elliptic Blending Reynolds Stress Model and Anisotropic Linear Forcing

Scale-resolving simulations of turbulent boundary layers (TBLs) still imply high computational costs which limit their applicability to realistic industry problems. A typical strategy to overcome this issue is a restriction of the scale-resolving simulation to a computational sub-domain. Here, we investigate the suitability of a hybrid RANS/LES based on an Elliptic-Blending Reynolds Stress Model (EB-RSM) RANS and turbulence synthetization in the LES sub-domain with Anisotropic Linear Forcing (ALF) to an adverse pressure gradient (APG) TBL case that was experimentally investigated by Hu and Herr. We compare different application strategies and show that the hybrid RANS/LES is capable to consistently predict the measured mean velocity profiles, stresses and wall pressure spectra with a coarser mesh and time-step choice than typically recommended for wall-resolved LES. An open point of research remains the quality of the ALF target fields, from the EB-RSM, in the APG region.

Lars Erbig, Sylvain Lardeau

A Rapid and Low Noise RANS-to-WMLES Condition in Curvilinear Compressible ZDES Simulations

The rapid and low-noise strategy of Deck et al. [9] for the RANS-to-WMLES switch compatible with compressible flow solvers on curvilinear grids is presented. It can be used both as an inflow condition or as an embedded resolved turbulence injection and combines Zonal Detached Eddy Simulation, Dynamic Forcing and Zonal Immersed Boundary Conditions (for roughness elements) approaches. The relaxation length is close to 7 boundary layer thicknesses on coarse grids and the feasibility on a 3-element high-lift airfoil is demonstrated. On a flat plate, no spurious acoustic footprint of the inflow is visible in the wall pressure spectra, of which the low-frequency part is obtained. The intermittent nature of wall turbulence is captured. The hybrid RANS/LES context makes the computational effort affordable for industrial applications, e.g. aeroacoustic studies.

Nicolas Renard, Sébastien Deck, Pierre-Élie Weiss

Assessment and Comparison of a Recent Kinematic Sensitive Subgrid Length Scale in Hybrid RANS-LES

A recent kinematic sensitive subgrid length scale, $$\varDelta _{lsq}$$ , initially developed for LES applications, is now considered for DES. Even though it is presented as a subgrid length scale, instead of a grey area mitigation (GAM) technique, this initial study shows how it could also be a good and natural approach for addressing this well-known DES shortcoming. In this paper, the $$\varDelta _{lsq}$$ has been compared with a well-known kinematic sensitive length scale, $${\tilde{\varDelta }_{\omega }}$$ . It includes a mesh resilience and a shear layer delay study with a Decaying Homogeneous Isotropic Turbulence configuration and two Backward Facing Step configurations, respectively. Encouraging results have been obtained, indicating $$\varDelta _{lsq}$$ as a subgrid length scale to be considered.

A. Pont-Vílchez, F. X. Trias, A. Revell, A. Oliva

Development of Alternative Shielding Functions for Detached-Eddy Simulations

This paper presents recent developments in finding alternative shielding functions in the framework of Delayed Detached-Eddy Simulation (DDES). The weaknesses of the standard shielding function are elaborated for the turbulent flow over a flat plate and an axisymmetric adverse pressure gradient flow. In both cases a small filter width compared to the boundary layer height caused a degeneration of the shielding function and led to severe model stress depletion. To overcome the strong grid dependency of the standard shielding two alternative shielding functions are proposed. The first determines the boundary layer edge by integrating the vorticity in the wall-normal direction, while separated flow is identified based on a comparative analysis of the individual vorticity components. For the second switching function the boundary layer edge is estimated by evaluating a localized formulation of the Bernoulli equation. The shielding disintegrates under resolved turbulent content by a sensor that includes the $$\sigma $$ -velocity gradient operator. The novel shieldings are verified for basic canonical test cases. Compared to DDES, a superior protection of attached boundary layers could be demonstrated.

Pascal Weihing, Johannes Letzgus, Thorsten Lutz, Ewald Krämer

The Grey-Area Improved -DDES Approach: Formulation Review and Application to Complex Test Cases

The paper presents a formulation review and recent results for a non-zonal enhanced DES variant called $$\sigma $$ -DDES [9], which seeks to address the issue of delayed transition from RANS to LES in free shear layers (also known as the grey-area problem) frequently experienced for standard delayed DES [13]. The two-part approach is based on modifying the LES behaviour of DES (the new model now behaves similar to the $$\sigma $$ LES model of Nicoud et al. [10]) as well as using a more elaborate LES filter width formulation. Combining both means delivers an enhanced behaviour in the crucial early shear layer region. First, the derivation of the approach based on two popular RANS models (i.e. Spalart-Allmaras and Menter SST) is discussed. Subsequently, recent OpenFOAM results from different test cases are presented, such as pressure-induced separation from a 2D hump, a generic car flow and aeroacoustic results for a rudimentary landing gear (RLG). The $$\sigma $$ -DDES variant is seen to deliver a more consistent flow prediction with varying grid resolution and topology independent of the applied RANS background model, and could also be established as a viable turbulence modelling approach for aeroacoustics prediction at low Mach numbers.

Marian Fuchs, Charles Mockett, Jörn Sesterhenn, Frank Thiele

Reynolds-Constrained Large-Eddy Simulation: Sensitivity to Constraint and SGS Models

Reynolds-constrained large-eddy simulation (CLES) method proves to have advantage over traditional large-eddy simulation (LES) approach in both attached and separated turbulent flows, but its sensitivity to the constraint model and base subgrid-scale (SGS) model remains unclear. Here, a comparative study is carried out to clarify the level of dependence of CLES method upon the constraint and base models. Specifically, the Baldwin-Lomax (BL) model, Spalart-Allmaras (SA) model, and Menter’s Shear Stress Transport (SST) model are used for the Reynolds constraint models, while Smagorinsky model (SM), Wall-adapting local eddy-viscosity (WALE) model, and Dynamic Smagorinsky model (DSM) for SGS models. The compressible flow past a circular cylinder is simulated at Reynolds number $$2\times {10^5}$$ and Mach number 0.75, respectively. It is manifested that the CLES method is sensitive to the constraint models, but less sensitive to the base SGS models, which provides a guideline for further optimization of the CLES method.

Rui Wang, Zuoli Xiao

On PANS-ζ-f Model Assessment by Reference to Car Aerodynamics

The present work discusses the predictive capabilities of the PANS model of turbulence (Partially-Averaged Navier Stokes; Basara et al. [4], representing a hybrid RANS/LES (Reynolds-Averaged Navier-Stokes/Large Eddy Simulation) modelling scheme, by means of simulating the flow past different car configurations including also overtaking maneuver cases. The unresolved residual turbulence is modelled by an appropriately adapted RANS-ζ-f formulation (proposed originally by Hanjalic et al. [9]). The investigated car configurations include a 40% down-scaled BMW model [17] as well as the so-called “DrivAer” car model [10]. As outcome of an intensive computational campaign by employing the PANS-ζ-f model formulation detailed mean flow and turbulence fields are obtained illustrating the model’s predictive capabilities in capturing unsteady features and corresponding time-averaged flow properties in a wide range of car configurations considered.

S. Jakirlic, L. Kutej, B. Basara, C. Tropea

Modelling-Related Numerical Aspects

Frontmatter

Adaptive Mesh Refinement with an Automatic Hybrid RANS/LES Strategy and Overset Grids

A simulation strategy combining an octree-AMR and an HRL turbulence model with overset grids is investigated in FLUSEPA™, the Finite-Volume solver developed by ArianeGroup. The turbulence model used is a $$k- \varepsilon $$ with a Delayed Detached Eddy Simulation [14] treatment for the destruction term of k. First, we study the influence of the refinement criteria on the pressure fluctuations in the recirculation area of a cylindrical backward facing step. Using an initially coarse grid, we compare the results with those from a fixed fine grid calculation. Then, the methodology is applied to the numerical simulation of the buffeting of an Ariane 5 model.

Alexandre Limare, Houman Borouchaki, Pierre Brenner

Investigation of Structured and Unstructured Grid Topology and Resolution Dependence for Scale-Resolving Simulations of Axisymmetric Detaching-Reattaching Shear Layers

The sensitivity to grid changes of a detaching-reattaching shear layer flow over a generic space launch vehicle geometry is investigated with Improved Delayed Detached Eddy Simulation (IDDES). First and second order statistical moments of the flow are compared and show good agreement with experimental and numerical data from literature. A sufficient axial resolution in the initial shear layer region is found to be crucial for an accurate capturing of the spectral content of the flow. Results on a grid with prismatic elements compare well to those on grids with hexahedral elements. A finer circumferential resolution changes the flow field drastically and leads to a merging of the first and second recirculation region, also affecting the pressure field and spectral features. Additionally, an absolute and a directional grid sensors are used that are found to be well suited to support anisotropic grid refinement.

Jan-Erik Schumann, Volker Hannemann, Klaus Hannemann

Advanced Numerical Strategy for the Prediction of Unsteady Flow Aerodynamics Around Complex Geometries

The present work focuses on an advanced numerical methodology named ZIBC standing for Zonal Immersed Boundary Conditions and enabling to account for realistic configurations at high Reynolds numbers. This coupling between a modelling method with a high level of maturity regarding the prediction of turbulent separated flows namely the Zonal Detached Eddy Simulation (ZDES) and Immersed Boundary Conditions (IBC) is detailed. Such a numerical strategy is applied to complex configurations dealing with external and internal aerodynamics namely a full space launcher configuration in the transonic regime and a supersonic air inlet in the subcritical regime before the buzz phenomenon occurs.

Pierre-Élie Weiss, Sébastien Deck

Comparative Assessment of Synthetic Turbulence Methods in an Unstructured Compressible Flow Solver

Three different synthetic turbulence methods are assessed in hybrid RANS/LES simulations with the unstructured compressible flow solver DLR-TAU. Fluctuations computed with either the Synthetic-Eddy Method, its divergence-free version, or the Synthetic-Turbulence Generator are injected via momentum sources into the flow field. In a flat plate flow, the latter method yields minimal deviations from reference data when combined with suited volume forcing, while the induced noise is not larger than in the other methods. In a mixing co-flow, all approaches yield decent predictions of the flow development apart from a slightly too high mixing rate.

Axel Probst, Philip Ströer

Wing/Airfoil Flows

Frontmatter

Transition Effect on the Vertical Flow Past the VFE-2 Delta Wing with Rounded Leading Edge

The influence of transition on the flow pattern and pressure distribution on VFE-2 was investigated by implementing the three-equation k-ω-γ transition model into DDES method (DDES-Tr). The DDES based on full turbulence (DDES-FT) is involved for comparison. For the attached flow, the vortex structure and flow pattern are largely affected by the transition exists on the leading-edge. Both the primary vortex and inboard vortex are captured by DDES-Tr, while the inboard vortex disappears in the DDES-FT, indicating the importance of transition in simulating the attached flow on VFE-2.

Wenyao Cui, Jian Liu, Zhixiang Xiao

Simulation of a Three-Dimensional Wing with Laminar Separation in Large-Scale Freestream Turbulence

The process of laminar separation is very sensitive to freestream turbulence. In the present work the influence of freestream turbulence with different length scales on the flow separation on a three-dimensional wing is investigated. In order to predict the separation it is necessary to resolve the turbulence. Since it appears not affordable to fully resolve the turbulence in this moderately complex three-dimensional configuration the hybrid model Delayed DES has been applied. The results show that smaller scales directly affect the flow separation by impinging the boundary layer while larger scales have an indirect impact when they change the local angle of attack.

Eike Tangermann, Markus Klein

Hybrid RANS/LES Investigation of the Interaction of a Longitudinal Vortex with an Inclined Airfoil

The interaction of a generic longitudinal wake vortex with an HGR-01 single-element airfoil is simulated with hybrid RANS/LES and compared with reference RANS results. The influence of different grid types in the vortex transport region, either structured hexahedral grid cells or unstructured tetrahedral cells, is shown to be insignificant. Moreover, two modeling approaches of the RANS and LES regions around the airfoil are compared: Firstly, the turbulence in the entire boundary layer is modeled with the RANS approach. Secondly, the turbulence in the boundary layer is partly resolved with the wall-modeled LES approach using synthetic turbulence at the RANS/LES interface. It is found that in the first setup the interaction of the vortex with the boundary layer is weaker than in the second setup with wall-modeled LES, however, the mean flow is comparable in both cases.

S. Probst, T. Landa, T. Knopp, R. Radespiel

Performance of a Modified DDES for the Near Stall Flow Past a NACA0015 Airfoil

A modification of DDES with adaptive coefficient CDES (DDES-AC) is proposed to deal with the delay transition from RANS to LES in the stall flows over a NACA0015 airfoil. The coefficient CDES is adaptive with the flow patterns, quasi-2D shear layer or 3D full developed separation, which helps to reduce the eddy viscosity in the separated shear layer. The performance of DDES-AC is validated by computing the flows over a NACA005 airfoil with mild trailing edge separation and during dynamic stall. It is found that the “grey area” in the original DDES is exacerbated in the simulation of dynamic stall. The DDES-AC is effective in accelerating the transition from RANS to LES and alleviating the “grey area” to some extent.

Jian Liu, Wenqing Zhu, Zhixiang Xiao

DDES and OES Simulations of a Morphing Airbus A320 Wing and Flap in Different Scales at High Reynolds

The present study concerns the use of unsteady numerical simulations by means of Navier Stokes Multi Block (NSMB) solver including both high order schemes and turbulence resolving methods. Firstly, this work attempts to highlight the role of the morphing applied to the supercritical Airbus A320 wing and flap in the trailing-edge for a Reduced Scale (RS) prototype at the clean position, this morphing includes a slight deformation of the trailing edge with a selected frequency and amplitude, which has an impact on the flow near the trailing edge and specially in the wake structures. This solution can transform the 3-dimensional chaotic flow into a 2-dimensional one by enhancing coherence of 2D structures rows of von Kármán vortices. In Addition, the highlift A320 wing-flap at the take-off position in Large-Scale (LS) configuration have been studied using advanced hybrid models DDES, the Organised Eddy Simulation OES and SST for the RANS regions as well as LES Smagorinsky model.

A. Marouf, N. Simiriotis, J. B. Tô, Y. Bmegaptche, Y. Hoarau, M. Braza

ZDES and URANS Simulations of 3D Transonic Buffet Over Infinite Swept Wings

This paper presents a study of transonic buffet over three-dimensional infinite swept wings. These configurations consist of the extrusion of an ONERA OALT25 airfoil with periodic boundary conditions in the spanwise direction. Unsteady Reynolds Averaged Navier-Stokes (URANS) simulations are first performed to assess the effect of the sweep angle. Spanwise flow structures are observed on these essentially 2D or 2.5D configurations, forming what has been named buffet cells. The 3D buffet frequency is correlated with the wavelength of these cells and the sweep angle. Then, Zonal Detached Eddy Simulations (ZDES) are carried out over small span and large span wings to provide numerical validation to the URANS simulations. These simulations are used to assess the capacity of hybrid RANS-LES methods to model the 3D effects involved in transonic buffet. At last, these simulations will provide insights into the complex physics involved in transonic buffet.

Fédéric Plante, Julien Dandois, Éric Laurendeau

An Automated Zonal Detached Eddy Simulation Method for Transonic Buffet

We present simulations with the Automated Zonal DES (AZDES) approach on the supercritical OAT15A airfoil and the Common Research Model (CRM) airplane configuration. Comparing URANS and AZDES simulations the shock prediction capabilities of URANS can be preserved. Turbulent structures in the wake are resolved in LES mode, which enables investigation of the propagation of turbulence in the wake. AZDES was shown to provide consistent results across different grids for the OAT15A airfoil. Validation using experimental data for the CRM showed good agreement in terms of wing pressure distributions.

Maximilian C. Ehrle, Andreas Waldmann, Thorsten Lutz, Ewald Krämer

Application of DDES to Iced Airfoil in Stanford University Unstructured (SU2)

This paper presents the investigation of the turbulent flow around Gates Learjet Corporation-305 airfoil with a leading edge horn-shape glaze ice using Delayed Detached Eddy Simulation (DDES) based on the Spalart–Allmaras turbulence model. The DDES algorithm implemented within the Stanford University Unstructured (SU2) solver was used for all the simulations. Numerical results of this validation effort were compared with experimental data, showing the increase of the prediction accuracy added with High Resolution (HR)-SLAU2 numerical scheme with the Shear-Layer Adapted (SLA) sub-grid scale (SGS) length.

Eduardo S. Molina, Daniel M. Silva, Andy P. Broeren, Marcello Righi, Juan J. Alonso

Aero-acoustic Analysis

Frontmatter

Industrial Prediction of Jet-Flap Interaction Noise with Advanced Hybrid RANS-LES Methods

Improvements to an industrial process for the simulation of jet-flap interaction noise are evaluated for a single-stream jet and a coaxial jet installed under a wing and flap. Prediction of the strong installation effect agrees well with measurements in a blind comparison. Alongside an advanced DES model with “grey-area” improvements, the importance of software infrastructure aspects such as meshing, numerics and process automation is demonstrated.

C. Mockett, M. Fuchs, T. Knacke, F. Kramer, U. Michel, M. Steger, F. Thiele

Other Aero- and Hydrodynamic Applications

Frontmatter

Assessment of Delayed Detached-Eddy Simulation of Dynamic Stall on a Rotor

High-fidelity unsteady Reynolds–averaged Navier–Stokes (URANS) and Menter-SST delayed detached-eddy simulations (DDES) of dynamic stall on a rotor with cyclic pitch control are presented and compared to experimental surface pressures and particle-image-velocimetry (PIV) data. Before the dynamic-stall event, the DDES suffers from modeled-stress depletion (MSD) leading to grid-induced separation (GIS) due to a breakdown of the boundary-layer shielding function $$f_d$$ . Combined with the “grey-area” problem, this leads to severe erroneous load peaks. After dynamic stall, flow is completely separated and only DDES shows realistic small-scale, incoherent vortical structures. Two approaches are investigated to eliminate MSD/GIS: Firstly, increasing the empirical constant $$C_{d1}$$ of the $$f_d$$ function to 30 basically eliminates GIS. Secondly, a non-local, grid-independent vorticity-integrated algebraic DES, which replaces the $$f_d$$ function, is introduced that provides robust boundary-layer shielding and enables the LES mode in case of massive flow separation.

Johannes Letzgus, Pascal Weihing, Manuel Keßler, Ewald Krämer

Scale-Adaptive Simulation (SAS) of Dynamic Stall on a Wind Turbine

Scale-adaptive simulation (SAS) approach is employed to investigate the complex dynamic stall phenomena occurring on a wind turbine blade. The results are compared with the more popular less computationally-expensive unsteady Reynolds-averaged Navier-Stokes (URANS) approach where the latter is validated using three sets of experimental data. The comparison reveals that the two approaches have similar predictions of the instant of the formation/bursting/shedding of the laminar separation bubble (LSB) and dynamic stall vortex (DSV), the size of the LSB and aerodynamic loads during the upstroke. This is while the two approaches exhibit dissimilar predictions of the trailing-edge vortex characteristics, its interaction with the DSV, number of secondary vortices and aerodynamic loads during the downstroke.

Abdolrahim Rezaeiha, Hamid Montazeri, Bert Blocken

Application of SST-Based SLA-DDES Formulation to Turbomachinery Flows

In turbomachinery CFD simulations, Reynolds-Averaged Navier-Stokes (RANS) based approaches tend to under-predict turbulent mixing, which is of particular importance in the endwall region. Detached Eddy Simulation (DES), solving Reynolds-averaged equations near the wall and partially resolving turbulent content elsewhere, bypasses the difficulties in statistic modeling. Previous work using Delayed-DES (DDES) in a linear compressor cascade [6] showed improved loss profile prediction owing to better representation of flow structures. The observation was that certain grid refinement near separated shear layer is necessary for resolving unsteadiness and further more, to reveal a bimodal behavior of the separation. Recent development in DES, the Shear-Layer-Adapted (SLA) modification [5], is designed to accelerate the growth of resolved turbulence downstream of the RANS-to-LES interface. In the current work, the impact of SLA modification on the DES simulation of linear compressor cascade flow is investigated. We incorporated the SLA modification into the SST-DDES [7] model and quantified the benefit of it in turbomachinery applications.

Guoping Xia, Zifei Yin, Gorazd Medic

Hybrid RANS/LES of an Isolated Engine Nacelle with Crosswind Using an Unstructured CFD Solver

The present contribution focuses on the high-fidelity scale-resolving simulation of an isolated engine nacelle subjected to strong crosswind. The work, carried out with the DLR TAU code, shows shortcomings of a steady RANS approach in predicting total pressure losses for the transonic partially-separated intake flow and proves the higher accuracy of advanced hybrid RANS/LES methods. In particular, an IDDES approach is combined with a hybrid numerical scheme that assures low-dissipation and low-dispersion errors in the focus area and numerical stability in the surrounding regions (hybrid LD2 scheme). The results are validated by means of theoretical turbulence spectra and experimental integral data.

Marco Burnazzi, Axel Probst, Mathias Steger

Numerical Investigation of the Flow Around a Simplified Ground Vehicles Using Hybrid RANS/LES Method

The aim of this work is to fine-tuned a numerical procedure to predict the external aerodynamics flow around a real car. The hybrid RANS/LES method, called DDES, has been selected based on previous work and comparison between many turbulence approaches. First, the numerical procedure has been developed on the $$25^{\circ }$$ Ahmed body. Based on this simplified geometry, we were able to define the numerical setup and grid in order to reproduce very accurately the flow around this shape. The second part of the work was to transpose this methodology on a more realistic vehicle, with some geometrical simplifications. At this preliminary stage of the study, the results are encouraging.

F. Delassaux, V. Herbert, I. Mortazavi, C. Ribes

Hybrid RANS/LES Simulations of Aerodynamic Flows Around Superstructures of Ships

The paper reports on a combined computational and experimental study on the aerodynamics of the superstructure of a container vessel. The study aims to assess the potential of add-on components to reduce the aerodynamic drag and validate the capabilities of hybrid RANS/LES approaches against traditional URANS methods. Results indicate the superior predictive performance of hybrid approaches and highlight the potential of aerodynamic add-on devices in the deckhouse region.

Rupert Angerbauer, Thomas Rung

Hybrid RANS-LES Simulation of a Tall Building in a Complex Urban Area

The aerodynamic performance of a building becomes increasingly important as its height protrudes further into the atmospheric boundary layer. For tall buildings, a thorough understanding of the aerodynamics is essential to make informed structural design decisions. However, this is a challenging case for numerical simulation as it involves accurate representation of both the bluff body aerodynamics and the incoming turbulent flow field over a large range of scales. Hybrid RANS-LES methods provide an attractive option for practical cases due to lower mesh requirements close to the wall of complex geometries. In this study, Detached Eddy Simulation has been used to simulate the flow around both an isolated standardised building as well as a tall building in a complex urban environment. Results show a good comparison to wind tunnel data, suggesting that the method could be a valuable tool in the design of tall buildings under wind loading.

Joshua Millar, Rhys Wayland, Joshua Holgate

Computational Analysis of the Flow Around a Surface Combatant at Static Drift

This paper presents a local and global computational study of the flow around the US Navy frigate DTMB 5415 at $$10^\circ $$ static drift configuration. A thorough validation study comparing isotropic $$k-\omega $$ SST, non linear anisotropic EARSM statistical closures and the unsteady hybrid RANS-LES DES-SST model is conducted. This validation study includes detailed global and local comparisons with IIHR (Iowa Institute of Hydraulic Research) remarkable TPIV experiments. These validations are conducted on a very fine unstructured grid, which includes local boxes of refinement in the cores of the two main vortices, for a total number of 163M cells.

M. Visonneau, E. Guilmineau, G. Rubino

Experimental/Numerical Study of Turbulent Wake in Adverse Pressure Gradient

The paper presents a bilateral German-Russian project launched in 2017 and aimed at investigation of turbulent wakes in the presence of Adverse Pressure Gradient (APG). Such wakes are a common feature of high-lift wing flows near the maximum lift conditions (take-off and landing), when the wake of the main wing is subjected to APG created by flaps. This type of flow is known to be poorly predicted by available RANS models. Hence, an ultimate goal of the project is their improvement based on a detailed experimental dataset and on results of high-fidelity turbulence resolving simulations providing relevant second moment closure terms not accessible by measurements. After a brief overview of the experimental and numerical parts of the project, the paper focuses on the first zonal RANS-IDDES computations of a wake of the flat plate in APG created by a plane diffuser. These computations performed in the initial stage of the project (before obtaining experimental data) are aimed at evaluating the capability of this approach to ensure the required accuracy with reasonable computational resources. Results of the simulations conducted on 3 grids (18, 30, and 50 million cells) support the credibility of the approach and suggest that it ensures not only virtually grid-independent prediction of the mean flow characteristics of the wake but also the dissipation-rate which is a key quantity in the context of improvement of the Reynolds Stress Transport RANS models. This is achieved, despite a relatively large grid step in the wake region (about 75 Kolmogorov length scales), thanks to computing this quantity based on the balance of the separate terms of the Reynolds stress transport equations.

E. Guseva, M. Shur, M. Strelets, A. Travin, W. Breitenstein, R. Radespiel, P. Scholz, M. Burnazzi, T. Knopp
Additional information

Premium Partner

    Image Credits