Skip to main content
Skip to article control options
Open AccessFull-Length Papers

Research Initiative for Numerical and Experimental Studies on High-Speed Stall of Civil Aircraft

Published Online:


The aerodynamics of transport aircraft at the borders of the flight envelope is characterized by complex interactions and nonlinear, unsteady flow phenomena. The underlying physical mechanisms are not fully understood, and a prediction of aerodynamic properties, transient loads, and aeroelastic behavior is a major challenge that requires the use of sophisticated numerical models and dedicated experiments. A new research initiative investigates the three-dimensional buffet mechanisms at high-speed stall, the interaction of the wing wake with the empennage, and the influence of ultra-high-bypass-ratio-type engines on these phenomena. It comprises unsteady pressure-sensitive paint surface pressure and time-resolved particle image velocimetry flowfield measurements carried out in the cryogenic European Transonic Windtunnel. The experiments are conducted at different Mach numbers and cover a broad Reynolds number regime up to 25×106 using the generic Airbus XRF1 configuration as a reference. This paper explains the rationale of the research initiative, introduces the planned test entries and test conditions, and presents selected results of the first measurement campaign and associated numerical analyses by means of the computational fluid dynamics code TAU, including usage of locally scale resolving simulations. An analysis of numerical and experimental results is presented for a high incidence transonic condition that serves to discuss the topology of wing root flow separation, the evolution of the associated wake, and its interaction with the empennage.


pressure coefficient


Young’s modulus, N/m2


frequency, Hz


Mach number


stagnation pressure, N/m2


Reynolds number


temperature, K


time, s


convective time unit


inflow velocity, m/s


dimensionless wall distance


angle of attack, °


dimensionless spanwise coordinate measured from the aircraft center line

I. Introduction

The economic and safe flight regime of civil aircraft is limited in terms of flight Mach number and angle of attack. At low Mach number and high incidence, massive separation (low-speed stall) occurs at the limit of the flight envelope, while high-speed stall at high Mach numbers is characterized by occurrence of unsteady shock waves and shock-induced separation. Due to these complex flow phenomena, the estimation of the edge of the envelope involves significant uncertainties even when modern computational fluid dynamics (CFD) methods are used during aircraft design. Ultimately, the envelope must nowadays be verified in flight tests. To avoid incorrect sizing of the aircraft, the phenomena at the edges of the flight envelope need to be better understood, and more accurate prediction models need to be developed and specifically validated. Scale resolving methods are hardly validated for this application and need to be extended for higher Reynolds numbers in order to perform high-resolution simulations. Apart from a lack of high-resolution flowfield measurements to study the underlying physical mechanisms, there are few scientifically usable and sufficiently detailed measurements of complete configurations for high, flight-relevant Reynolds numbers at the borders of the flight envelope.

While numerous numerical and experimental studies on the mechanisms of two-dimensional (2D) buffet (e.g., [14]), which is associated with mono-frequent shock oscillations, have already been carried out in the past, there are still gaps in the knowledge on the causes and mechanisms of the buffet phenomenon. Three-dimensional (3D) transonic buffet on swept wings represents an even more complex phenomenon due to geometric effects such as wing sweep and taper, as well as interactions with the 3D separation. Iovnovich and Raveh [5] investigated the influence of sweep angle on the buffet phenomenon on generic straight and swept wings with and without consideration of wing tip effects. They observed that the propagation of pressure disturbances in spanwise direction begins to dominate the flow phenomena with increasing sweep angle. These pressure disturbances originate at the rear foot of the shock near the wing root and propagate toward the wing tip while being amplified. The shock–boundary-layer interaction caused by these disturbances leads to a shock movement, which manifests itself locally in shock oscillations that migrate toward the wing tip in the form of so-called buffet cells. Their oscillation frequency and propagation speed are found to increase with increasing sweep angle. Koike et al. [6] have established a classification of shock oscillation of the Common Research Model (CRM) for three different ranges of the angle of attack. These and earlier works suggest that the transonic buffet at swept wings, contrary to the 2D buffet phenomenon, is associated with broadband pressure fluctuations. Analyses on the shock motion on a swept wing have also been carried out by Dandois [7] based on the results of two wind tunnel campaigns. In the work the 3D buffet phenomenon was characterized exemplary for the AViation Emission Reduction Technologies (AVERT) model, and the observed spanwise motion of the buffet cells was discussed. It was observed that the corresponding pressure oscillations appear at frequencies several times higher than for the 2D scenario. Sugioka et al. [8] supplemented these studies with measurements of the unsteady shock motion on the wing of the CRM at low Reynolds numbers, where they observed two types of shock motion. The first type could be associated with buffet cell propagation in a Strouhal number range between 0.3 and 0.5. The second type, which occurred at higher angles of attack and a Strouhal number range below 0.1, could be attributed to a shock oscillation with large spatial amplitude. Paladini et al. [9] provide an overview of experimental investigations in recent years and evaluate them, deriving case-independent characteristic Strouhal numbers as well as wavelengths and convection speeds of spanwise fluctuations. For flight-relevant Reynolds numbers, time-resolved particle image velocimetry (TR-PIV) measurements of the CRM were carried out in the cryogenic European Transonic Windtunnel (ETW) within the European project ESWIRP [10], which served to validate simulations for low- and high-speed stall conditions (M=0.25 and 0.85 respectively) and to perform buffet and wake flowfield studies.

Sartor and Timme [11] achieved good agreement of predicted and measured statistics of the surface pressure distribution using delayed detached eddy simulation (DDES) on the RBC12 wing–body configuration, where the advantages of the DDES approach compared to unsteady Reynolds-averaged Navier–Stokes (URANS) methods became obvious. Ehrle et al. [12] conducted simulations on the CRM at buffet conditions applying URANS and different hybrid RANS/large eddy simulation (LES) methods, compared the results with the ETW measurements and analyzed the shock dynamics. Here, good agreement with respect to convection velocity and the spanwise propagation of buffet cells with estimations from literature could be observed. The analyses further showed a growth of the wavelength and convection velocity toward the outboard part of the wing.

Crouch et al. [13] performed global stability analyses based on URANS simulations of swept and unswept infinite wings. They postulate that buffet is associated with global flow instability on swept wings, but with a different primary instability mode than on 2D or unswept wings. This mode is stable for the latter and becomes an unsteady and propagating mode with the introduction of sweep. Using a similar approach, also Plante et al. [14] identified a global flow instability and compared the corresponding buffet cells to stall cells, which emerge from a different flow instability in low-speed stall conditions. Nitzsche et al. [15] describe shock buffet as a special case of fluid-mode flutter in the limit of infinitely stiff structure. They define buffet onset as a transition from a stationary stable to a stable unsteady limit-cycle state. Timme [16] conducted global stability analyses on RANS results for the CRM model and has linked buffet cells to an unstable linear eigenmode, supporting the idea that these propagations constitute a global instability that self-sustains shock motion. Masini et al. [17] performed detailed flow physics analyses of unsteady wind tunnel measurements on the RBC12 half-model, considering a wide range of inflow parameters from prebuffet onset conditions with attached flow via buffet onset to buffeting conditions. With respect to shock buffet instability, they identified two phenomena, one low-frequency shock unsteadiness that predominantly propagates pressure disturbances inboard, and broadband higher-frequency outboard-running pressure perturbations along the shock wave and in the separated flow domain downstream of the shock front.

There is also a need for research on the development of the wake downstream of shock-induced flow separation under high-speed stall conditions, the wake interaction with the empennage, and on tail buffet. For the low-speed stall of a transport aircraft, Havas and Rabadan [18] discussed the impact of the reduced downwash on the pressure fluctuations and separation at the tail based on flight test and wind tunnel data and derived a tail buffet model. Illi et al. [19] performed URANS and DDES simulations on the transonic tail buffet and discussed the differences in the resulting pressure spectra and the separation behavior at the horizontal tailplane (HTP) for the A320 ATRA configuration. This work was complemented by URANS calculations of the CRM [20] and highlighted the need for detailed measurements and flow physics studies. Tan et al. [21] investigated the influence of wake turbulence on the HTP of the CRM under low-speed stall conditions. They showed that complex interactions of the large-scale vortices, starting from the kink in the trailing edge of the wing, with the HTP can occur. It is to be expected that the mentioned flow effects and their characteristics will be influenced by future fuel-saving engines with an ultra-high bypass ratio (UHBR) and that further significant transonic interactions will occur, for example, on the lower side of the wing.

Detailed numerical and experimental studies of these high-speed stall phenomena are being carried out within a new research initiative. The main goals of the research initiative are the comprehensive investigation of the unresolved physical issues in the high-speed stall of transport aircraft, the extension and qualification of scale-resolving numerical methods for this application, and the modeling of unsteady flow and nonlinear aerodynamics with reduced order models. The knowledge gain shall be achieved by coordinated numerical and experimental studies of the flow physics in different domains of the flowfield combined with corresponding investigations on a full aircraft configuration. These studies are being carried out on the Airbus XRF1 transport aircraft configuration [22]. Since the examined flow phenomena are Reynolds number dependent and insights up to flight-relevant Reynolds numbers are desired, extensive measurements in the cryogenic ETW are conducted using the XRF1 wind tunnel model provided by Airbus. To investigate interactions of the wing wake with the empennage, a HTP was recently built, instrumented, and integrated into the XRF1 model for the tests within the present initiative. The ETW experiments include elaborate steady and unsteady surface pressure measurements utilizing novel pressure-sensitive paint (PSP) techniques and TR-PIV measurements in relevant flow domains. The present initiative, in a certain way, represents a follow-up to a first TR-PIV campaign, which was carried out within the European ESWIRP project for the scientific use of ETW on the NASA Common Research Model (CRM) [2325].

This paper presents the research objectives and the partners of the research initiative along with their specific contributions; introduces the XRF1 wind tunnel model, the new components, and the planned ETW measurements; and finally presents exemplary results of the first measurement campaign conducted in November 2020 along with hybrid RANS/LES results and flow physics analyses. The remainder of the present paper is organized as follows. Section II describes the structure of the entire research initiative and the research scenarios considered. Section III introduces the numerical model used for the simulation of a high-incidence case. An overview about the XRF1 wind tunnel model with the new components, the wind tunnel, and the unsteady measuring technique including the planned test conditions is given in Sec. IV. The predicted flow separation topology and wake evolution is discussed in Sec. V and compared to wind tunnel data of the first test entry.

II. Structure of the Research Initiative and Research Objectives

The scientific support and exploitation of the ETW measurements is conducted in the frame of the DFG (German Research Foundation) research unit 2895 that is organized in seven subprojects and focuses on detailed numerical studies on 3D buffet phenomena on the upper side of the wing, the development of the wake, and its interaction with the tail plane (cf. Fig. 1). The research unit pursues the goal to study the flow phenomena and their impact on aircraft aerodynamics combining highly resolved simulations focusing on relevant regions of the flowfield with global simulations of the complete aircraft configuration. Different classes of numerical models ranging from RANS solvers in combination with Reynolds-stress turbulence models (RSM) via hybrid RANS/LES methods to wall modeled and wall resolved LES simulations are employed.

The first subproject (TP1, RWTH Aachen University) examines the mechanisms of 3D buffet on the wing suction side and the influence of perturbations generated by the engine using wall resolved and wall modeled zonal LES methods. In TP5 (University of Stuttgart) the impact of the turbulent wake on the HTP aerodynamics is investigated in detail using the wall-modeled LES, whereby the relevant scales of the turbulence entrainment into the HTP boundary layer are fully resolved. TP4 (University of Stuttgart) represents a bridge between TP1 and TP5 and performs hybrid RANS/LES simulations of the full configuration to analyze the influence of wing root separation on the evolution of the wake and its interaction with the HTP. These simulations are performed using scale-resolving methods in the relevant wake regions, which provide the turbulent inflow conditions for the LES of the HTP domain performed in TP5. Beside advancing the development of hybrid RANS/LES methods, the flow in the vicinity of an UHBR nacelle and its impact on shock formations and buffet phenomena on the lower side of the wing are investigated in TP3 (TU Braunschweig, German Aerospace Center [DLR]). These simulations are complemented by RSM-URANS calculations to study the UHBR influence on the unsteady aerodynamics of the wing at the edge of flight envelope with wide parameter variations.

Fig. 1
Fig. 1

Research foci of the DFG-funded research unit 2895.

The numerical studies of the DFG research unit are supplemented by dedicated ETW measurements on the XRF1 wind tunnel model. These tests are financed by Helmholtz Association (HGF) and DLR. In addition to aerodynamic forces and moments, pressure, and deformation measurements, DLR carries out unsteady surface pressure measurements by means of PSP. Further, PIV flowfield measurements with high temporal resolution are performed by DLR supported by TP6 (RWTH Aachen University) of the DFG research unit. Supplementary to the studies on the XRF1 configuration, TP6 will conduct temporally and spatially highly resolved measurements in the Aachen trisonic wind tunnel on a generic, unswept tandem wing configuration, which is numerically studied by TP1/4/5.

The measurement data are evaluated and interpreted synergistically with the numerical results to improve the understanding of the flow physics at edge of envelope conditions. For this purpose, methods for data reduction as well as flowfield analysis tools like proper orthogonal decomposition, dynamic mode decomposition (DMD), or multiresolution DMD are enhanced by TP2 (TU Braunschweig) and applied in all subprojects for flow-physics studies. The synthesis of the database collected in the simulations and the ETW measurements is finally carried out by TP2 and TP7 (TU Munich). For the modeling of unsteady local and global loads in the nonlinear regime, neuro-fuzzy reduced order modeling (ROM) models are utilized in TP7 and trained on the basis of the generated data.

During preliminary RANS-based CFD analyses (compare Sec. III) it quickly became apparent that for the XRF1 configuration buffet onset occurs in the outer half of the wing, and that under these conditions the flow near the wing root remains attached. For buffet onset conditions the wake of the wing passes well below the HTP as evident from the predicted streamlines depicted in Fig. 2. Investigation of the wake development and interaction with the HTP therefore requires different flow conditions than for buffet onset studies. Flow separation near the wing root with the wake directly impinging the HTP is provoked only at a significantly higher incidence.

To address all scientific objectives of the present research unit, a total of four different research scenarios were defined, each characterizing different flow phenomena, aerodynamic requirements, and necessitating different flow conditions:

  • Research scenario 1: Outboard wing upper side buffet
  • Research scenario 2: Inboard wing upper side separation and wake interaction with the HTP
  • Research scenario 3: Inboard wing lower side unsteadiness with UHBR provoked buffet phenomena
  • Research scenario 4: ROM dataset acquisition

Fig. 2
Fig. 2

RANS-predicted wake streamlines at near buffet conditions (scenario 1, M=0.88, α=5°, Re=25×106).

III. Numerical Model

A. XRF1 Meshes

The generation of the meshes used in the research unit for the CFD simulations was a joint effort of the University of Stuttgart and DLR. As basis for the mesh generation, DLR first created a wind-off surface model of the wing–fuselage–vertical tailplane (VTP) combination including flap-track fairings based on the XRF1 wind tunnel construction model. In this model, the new HTP (Sec. IV.B) as well as the wind tunnel support used in the ETW measurements were integrated (see Fig. 3). Due to its influence on the pressure distribution at the fuselage tail and the HTP as well as the spanwise distribution of the shock position on the wing (cf. [2628]), the wind tunnel support is generally taken into account in the future simulations of the research initiative. Based on findings from previous investigations on the CRM configuration [28], the arc sector was not modeled and the modeled support was truncated with a rounded cap at the sector end (cf. Fig. 3). The arc sector holds the model support and rotates in the vertical plane to allow rotation of the model around a fixed point in the fuselage area, the model rotation point. According to [28], the flow separation occurring at the base of the truncated support has negligible impact on the aircraft pressure distribution, and convergence problems were not observed.

Two sets of meshes were created. First, a hybrid mesh was generated for preliminary RANS calculations to define the setting angle and instrumentation of the new HTP (Sec. IV.B) and to identify relevant inflow conditions for the research scenarios before the first ETW entry (Sec. IV.E). Secondly, highly resolved, largely block-structured grids were generated, which are used by the partners of the research group as baseline grids for their scientific simulations and will be described below in more detail.

Fig. 3
Fig. 3

Modeled surface of the XRF1 wind tunnel model including HTP, support, and arc sector.

For preliminary calculations a common grid is generated by means of the commercial mesh generator Centaur by CentaurSoft, using the geometry of the XRF1 wind tunnel model including HTP with and without UHBR flow through nacelle. Static aeroelastic wing deformation is considered, based on the information for similar loading available for the Airbus Simulation Challenge [22] conditions (M=0.86, Re=25×106) and corresponding to the deformed shape at α=5.864°. This approximate deformation was kept constant across all preliminary calculations.

The family of high-quality baseline meshes was generated using the commercial grid generator Pointwise. Adapted meshes have been generated for all Reynolds numbers to be considered in the ETW tests, i.e., Re=3.3/6.7/12.9 and 25 million. A meshing strategy was followed that has already proven successful for high- and low-speed stall wake studies on the CRM configuration [24]. To obtain better control over the extrusion of the boundary-layer blocks, structured grids were introduced to discretize the surface of the wind tunnel model wherever possible. Unstructured mesh areas were only utilized in transitions to domains of different resolution. On the wing a maximum grid size of 0.5% of the local airfoil chord was used in chordwise direction and 0.7% of the mean aerodynamic chord (MAC) in the spanwise direction, with a refinement introduced at the junction to the fuselage (compare Fig. 4). Hexahedral and prism blocks were extruded from the surface grids of fuselage, wing, and empennage. To enable high-quality simulations with RSM, all meshes provided a dimensionless wall distance of the first point of y+<0.4 for their specific range of flow conditions. H-type grids were introduced at the junctions of the fuselage with the wing, the HTP, and the VTP. This allows high resolution of the complex 3D flow in these sensitive domains. The same topology was used to resolve the corner domains of the flap track fairings (see Fig. 4). To ensure high resolution of shock topology and buffet cell motion, highly resolved hexahedral grid domains were introduced above the upper side of the wing with the complete supersonic flow domain being embedded (Fig. 5). The far-field region is discretized by tetrahedra, and the distance from the model to its outer boundary amounts to 50 wingspans. The baseline mesh for Re=25 million contains about 110 million grid cells for the half model. All baseline meshes were generated for wind-off conditions. The measured wing deformations caused by the aerodynamic loads in the wind tunnel are taken into account in the flow simulations through mesh deformation.

These baseline meshes are adapted or extended by the research partners for their specific simulations. For example, a dedicated hexahedral wake block was added for research scenario 2 specific hybrid RANS/LES simulations discussed below. The wake block discretizes the relevant flow domain between the wing and the empennage with quasi-isotropic cells and a high resolution of 0.7% MAC, in order to simulate the development of the turbulent wake in LES mode (Fig. 5). Its extent was chosen with the goal of capturing the entire wing wake at angles of attack of up to 10°. The required height was determined based on steady RANS calculations utilizing the Centaur grid introduced above. This wake block increased the number of cells by about 60 million for the half model. It was utilized in the Automated Zonal Detached Eddy Simulations (AZDESs) described below.

Fig. 4
Fig. 4

Grid topology at the wing–fuselage junction and the flap-track fairing.

Fig. 5
Fig. 5

Hexahedral block above the wing and scenario 2 wake block.

B. Flow Solver and Numerical Setup for Research Scenario 2 Simulations

Within the present study, simulations of the impact of the separated flow at the wing root on the flow over the HTP were carried out and compared with ETW measurement data. The aim is to validate the research scenario 2 inflow conditions chosen in the ETW experiments and to characterize the flow and validate the simulations for this inflow condition. While URANS methods have proven successful and efficient for calculating attached boundary layers up to pressure- or shock-induced flow separation, these methods are not appropriate to investigate the evolution of the turbulent wake and thus its impact on the HTP due to excessive dissipation of turbulent motion. Hence, hybrid RANS/LES models such as the detached eddy simulation (DES) that capture attached boundary layers in URANS mode while switching to scale resolving LES mode in the separated wake domain represent a suitable approach to tackle this problem [11,12,24,25]. Despite various improvements in shielding functions developed over the last years, grid-induced separation from smooth surfaces is still an issue. Too fine grid resolution may cause a collapse of the shielding in the boundary-layer area and lead to an incursion of the LES region into parts of the boundary layer as well as to premature flow separation. Besides a zonal DES (ZDES [29]), the AZDES proposed by Schulte am Hülse [12,24,30] offers stronger user control over the RANS domain and is applied in the present study. The scale resolving behavior in the wake is realized by activating the DDES model outside of predefined RANS areas. The main difference to the ZDES is that the zone separation is based on an evaluation of solution-based quantities from a precursor URANS simulation with different control options for the user.

All simulations in this paper were performed using the finite volume flow solver TAU by DLR [31]. Second-order accuracy in space is achieved by a central differencing scheme for the convective terms. TAU’s matrix-type artificial dissipation is chosen to reduce the numerical dissipation of small-scale structures, with a fourth-order dissipation coefficient of 1/128. The skew symmetric scheme proposed by Kok [32] and a second-order Roe scheme are applied for the discretization of the mean flow fluxes and the turbulence fluxes, respectively. Time integration is realized with an implicit backward Euler scheme. The dual time-stepping scheme is applied for all unsteady simulations and enables second-order accuracy in time. The physical time step size of 1.1×105  s represents 100 time steps per convective time unit tc=MAC/u, and the number of inner iterations was set to 100. The time step relates to a physical frequency of 90.9 kHz and therefore allows the resolution of the buffet phenomenon, which takes place at a much lower frequency. Two turbulence models were employed: the Spalart–Allmaras (SA) model [33], specifically the variant denoted as negative Spalart-Allmaras model [34], which is a reformulation of the “standard” SA model with enhanced numerical robustness, and the SSG/LRR-ω Reynolds stress model [35]. To avoid unphysical side-of-body separation at the wing–fuselage junction the quadratic constitutive relation (QCR) extension [36] was activated for the SA model. The AZDES mode requires a precursor URANS simulation, which was run over 150 convective times related to the MAC in order to get a representative distribution of the turbulent length scale. The Δmax filter, based on the maximum edge length of the grid cells, is used for the hybrid RANS/LES simulations.

For the simulation discussed in Sec. V, the baseline mesh including the refined wake block described above was used neglecting the model support. During the wind tunnel test, the aeroelastic wing deformation was measured in the form of a spanwise bend and twist distribution by means of the stereo pattern tracking system. The static wing deformations as measured for the considered flow condition were imposed on the baseline grid using the mesh deformation algorithm available in the flow solver TAU.

IV. Wind Tunnel Experiments

A. XRF1 Wind Tunnel Model and Airbus FeRiT Test Campaign

Airbus has designed the generic XRF1 transport aircraft configuration as an industrially relevant standard multidisciplinary research test case representing a typical twin-engine widebody transport aircraft [22]. Data from the model are provided to academia for research purposes. To validate CFD methods and for carrying out flow physics studies, Airbus built and instrumented a 137-scale full aircraft wind tunnel model that can withstand the high loads and cryogenic temperatures for flight-relevant conditions in ETW. In 2018, Airbus conducted a first ETW campaign within the “Feature Rich Testing” (FeRiT) approach [22], to generate a broad aerodynamic database for edge of the envelope flight conditions. The XRF1 geometry with clean wings (i.e., cruise configuration), through-flow nacelles, fuselage, belly fairing, and VTP but without HTP was used. Various spoiler and aileron deployment configurations were also examined. The sting entry angle into the model is 5°, which enables the installation of an HTP. The FeRiT edge of the envelope wind tunnel test aimed at generating extensive validation data for CFD simulations of transonic flows, investigating strong 3D shock patterns, nonlinear regions, and separated flow. Besides forces, moments, and deformation measurements, static wing pressure distributions were measured by more than 300 pressure tappings, and unsteady pressure measurements based on 24 Kulites were conducted. A wide Mach number range was covered for Reynolds numbers between 5×106 and 40×106. For Re=25×106, additional steady surface pressure measurements were carried out using PSP for selected configurations. The measured data will be made available to partners of the XRF1 research community step by step in the context of an Airbus initiated multistage simulation challenge. The FeRiT campaign was intended to be a macrolevel test to be complemented by microlevel tests looking at the boundary layer in more detail. A contribution to more detailed flow physics studies is aspired within the present research initiative.

B. New Components of the XRF1 Wind Tunnel Model

To pursue our specific research objectives, the XRF1 wind tunnel model has been supplemented with additional components. To investigate the interaction of the wing wake with the tail plane, a HTP was designed with support from Airbus and manufactured and instrumented by Deharde as part of the present research initiative.

The HTP angle of incidence was selected with the goal of ensuring a realistic distance between center of pressure and aerodynamic center, and in order to avoid strong shock-induced separation on its lower surface. These requirements are driven by flow conditions envisioned for research scenario 2 (cf. Sec. II), which involve interactions of the wing wake with the HTP. This latter goal was aspired because the focus of research scenario 2 is on studying the interactions between the turbulent structures of the wing wake and the flow around the HTP and its boundary layer. To determine a suitable HTP angle, preliminary RANS calculations (Sec. III.A) were performed for the full configuration, in which the HTP angle, the angle of attack, and the Mach number were varied.

Also the HTP instrumentation was specified based on these preliminary RANS calculations for relevant research scenario 2 flow conditions. A total of 50 pressure taps and 10 unsteady pressure sensors (Kulite) were integrated into the HTP. Thereof nine Kulites were implemented in two relevant sections at 60% and 85% of the HTP semispan (see Fig. 6). These spanwise positions were chosen because the HTP inflow in this region was expected to be mainly determined by the wake separating in the root region of the wing rather than by interactions with the fuselage and its boundary layer. Since the section at 60% of the semispan is influenced by the 3D tip flow of the HTP to a lesser degree, five Kulites were placed at 60% on the HTP port side. Three of them (C, D, E) are on the lower side, which is more strongly loaded, and two (H, I) are on the upper side. The rearmost Kulite E on the bottom side is placed at 60% chord, slightly downstream of the shock position expected for research scenario 2 flow conditions. From the second section at 85% of the semispan, two Kulites were implemented on the top of the port side; the corresponding lower-side Kulites are located on the starboard side. One additional Kulite is placed on the bottom side of the port side near the leading edge at 70% of the semispan to get better information about spanwise correlations. The majority of the Kulites were implemented on the port side, because on this side the wing of the XRF1 wind tunnel model is instrumented with three Kulites in the root region of the upper side. This enables a characterization of the unsteady flow from separation at the wing toward the interaction with the HTP also without time-consuming unsteady PSP and TR-PIV measurements.

The foremost Kulites are placed at 16% chord on the top and at 10% chord on the bottom side of the HTP. The position near the leading edge was chosen because pronounced impacts of the unsteady wing wake on HTP pressure fluctuations were expected due to the amplifying effect of the leading edge. To observe the downstream development of the pressure spectrum, a second position at 30% chord was chosen.

Fig. 6
Fig. 6

HTP planform with positions of the pressure taps and Kulites (upper picture: top view; lower picture: bottom view).

The static pressure taps were arranged in two sections at 50% and 70% of the semispan, thus enclosing the Kulite-instrumented section at 60%. As the HTP lower side shows the strongest pressure variations for the research scenario 2 inflow conditions, more pressure taps were allocated on this side than on the upper side. The chordwise distribution was determined on the basis of CFD calculations, whereby refinements were realized in areas with stronger pressure gradients.

To study the impact of future UHBR engines on the buffet mechanism, the wake, and finally the interaction with the empennage, a dedicated UHBR flow through nacelle was designed by DLR. The focus of the design was, on the one hand, to obtain a nacelle that is representative for a modern UHBR turbofan engine in terms of cruise performance, while, on the other hand, triggering high-speed stall phenomena under off design conditions. The nacelle design and instrumentation were particularly adapted to the needs of research scenario 3 (compare Sec. II) while ensuring to meet the requirements for the other scenarios as well. This component has been manufactured and instrumented and will be attached to the XRF1 wind tunnel model in upcoming test entries.

C. European Transonic Windtunnel

The aim of the research initiative is to study buffet phenomena on transport aircraft configurations for different Mach and Reynolds numbers up to the flight relevant regime. To enable such measurements on scaled full-span models, wind tunnels capable of circulating strongly cooled working fluid in order to reduce its kinematic viscosity are required. For the present high-Reynolds-number tests the pressurized cryogenic ETW facility is being used (cf. Fig. 7). ETW is a closed-circuit wind tunnel (Göttingen type) with an electric drive power of up to 50 MW. It can be operated with slotted or solid walls in a Mach number range from 0.15 to 1.35. Increased Reynolds numbers are achieved by a combination of the temperature (110–313 K) and pressure (115–450 kPa). For full-span models, Reynolds numbers up to about 50×106 can be achieved. The dimensions of the test section are 2.4  m×2  m×9  m (width×height×length), and the overall length of the aerodynamic circuit is 142 m. The test gas is nitrogen, which is injected in liquid form by 250 nozzles distributed along four rakes ahead of the compressor. As the values for total temperature and pressure can be varied independently, Mach and Reynolds number effects and also the impact of model deformation can be examined separately. More detailed information about this facility is presented by [37] and on the ETW website (

Fig. 7
Fig. 7

The aerodynamic circuit of ETW (with courtesy of ETW).

D. Advanced Optical Unsteady Measuring Technique

In addition to the balance-based measurement of the aerodynamic coefficients, static pressure distributions by many pressure taps and unsteady pressures by Kulite sensors (cf. [22], Sec. 3) are measured in relevant areas of the XRF1 wing, HTP, and UHBR nacelle. Wing deformation is measured in terms of twist and bend distribution using a stereo pattern tracking technique with markers applied to the wing surface. The experimental data in the present research initiative are augmented by extensive usage of advanced optical measurement techniques to acquire the steady and unsteady surface pressure distributions on wing, HTP, and UHBR nacelle using steady and unsteady PSP.

Compared to conventional pressure measurement techniques, PSP is an optical method that enables pressure measurements over the complete wing surface, rather than measuring at discrete pressure tapping positions. It is based on the principle of deactivation of photochemically excited molecules by oxygen molecules, so-called oxygen quenching. Using suitable light sources to excite the molecules and taking a picture of the emitted fluorescent light, it is possible to determine the pressure on a model surface after adequate calibration. This enables to determine flow phenomena both qualitatively and quantitatively [38]. However, the presence of oxygen molecules in the test gas is a prerequisite. The PSP characteristics depend on the oxygen concentration. When the oxygen concentration is constant (and known), it is correctable by the in situ calibration with the pressure taps or Kulite sensors. This fact was used in the development of so-called cryoPSP measurement technique [39,40]. To enable PSP measurements under cryogenic conditions with nitrogen as test gas, oxygen is added in a controlled manner to achieve a concentration of approx. 1000 ppm oxygen in the nitrogen atmosphere. For cryoPSP an optimized PSP for ETW was investigated by DLR and the University of Hohenheim (Germany), which has a good pressure sensitivity and a negligible temperature sensitivity at cryogenic conditions [40] and shows a thickness of only 5  μm with a surface roughness of better than 0.3  μm [38]. This technique was applied in the first ETW test entry of the present initiative to characterize steady and unsteady surface pressure distribution on wing and HTP.

Further test entries include detailed TR-PIV measurements, covering the flowfield in the vicinity of the oscillating shock front at the wing and the propagation of the wake from wing to HTP, including characterization of the detailed unsteady HTP inflow conditions. DLR developed the cryogenic optical measuring systems in close cooperation with ETW and is in charge for carrying out these measurements. Two different techniques are utilized. On the one hand, a high-speed laser and high-speed cameras are used, which record a temporal resolution of the measured velocity fields with a sampling frequency of 2 kHz. The optical components consist of an LEE LDP-200MQG high-speed laser providing a pulse energy of 10 mJ and a pco.dimax HS 4 CMOS camera [23]. On the other hand, a technique presented by Geisler [41] is being applied that makes use of special charge-coupled device (CCD) camera sensors (so-called FOX camera) to separate images for up to three points in rapid succession (time offset Δt=11730  μs). With such multipulse recordings, it is also possible to directly measure the local accelerations even in flows of high velocities. Furthermore, the multipulse measurements at several different time offsets will be used to reconstruct power spectra of turbulent flows or fluctuations of other flow quantities using a dedicated postprocessing method presented by Schreyer et al. [42]. The placement of the PIV cameras and the light sheets are visualized in Fig. 8.

Fig. 8
Fig. 8

Placement of the PIV cameras and light sheet boxes (courtesy of DLR).

E. ETW Entries and Test Conditions

As outlined in Sec. I, aside from the 3D buffet mechanisms, the interaction of the wing wake with the HTP as well as the influence of an UHBR nacelle on the buffet mechanism and the wake HTP interactions will be investigated. During the first phase of the research initiative, three test campaigns with different XRF1 configurations and test foci were conducted, denoted MK1–MK3 (see Table 1). In MK1 the XRF1 configuration with a clean wing and the new HTP is being examined. MK1 is split into two parts. The first entry MK1a comprised aerodynamic standard measurements as well as static and unsteady PSP surface pressure measurements on wing and HTP. These serve to characterize the clean model at the relevant conditions and to verify the selected buffet and separation-prone inflow conditions before the PIV measurements, which formed MK1b. TR-PIV wake measurements were performed in MK1b for different spanwise positions covering the wake domain between the wing trailing edge and the HTP, while FOX-PIV measurements were conducted to characterize the flowfields in the buffet domain on the upper side of the wing and in the vicinity of the HTP leading edge. In MK2, newly designed and manufactured instrumented UHBR through flow nacelles were installed, and essentially the same flow conditions as in MK1a/b have been considered to identify the impact of the UHBR on the examined flow phenomena. MK3 is dedicated to a more detailed flow survey in the vicinity of the UHBR and the evolving shock system on the lower side of the wing.

For the investigation of the flow-physics phenomena introduced in Sec. I, unsteady measurements need to be carried out with high temporal and spatial resolution. Especially for the TR-PIV measurements, long measurement times are required to sufficiently characterize the flowfield in the buffet domain and the wake development. To examine Reynolds numbers as high as 25×106, measurements under cryogenic conditions are necessary, which is associated with considerable costs due to the immense nitrogen consumption under these flow conditions. To keep the operating costs in an acceptable range, the considered flow conditions relevant for the different research scenarios were therefore carefully selected in advance, based on analyses of preliminary RANS calculations (compare Sec. III.A). The uncertainties in predicting the flow conditions involving noteworthy buffet and shock unsteadiness on the wing and the nature of the edge-of-envelope conditions make precise a priori determination of conditions based on steady RANS results unreliable. Therefore, the a priori RANS calculations were cross-checked with static pressure measurements and Kulite results available from the FeRiT test entry [22]. However, FeRiT was conducted without a HTP and with conventional underwing nacelles, significantly reducing the comparability. The combined analysis of CFD and experiments was nevertheless used as guidance in terms of flow regime definitions.

The preliminary numerical analyses and evaluation of FeRiT data led to the definition of the relevant flow conditions, which take into account the specific requirements of the four scenarios at minimized measurement costs. In the definition of the test matrix the strategy was pursued that both Reynolds and Mach number effects should be investigated at the same dynamic pressure q/E, i.e., at comparable model deformations. To ensure a connection to the FeRiT campaign, Re=25×106 was chosen as upper Reynolds number for the present campaign. Taking into account the constraint that the temperature of the test gas must be sufficiently low to enable PIV measurements using ice particles as tracer, the smallest Reynolds number achievable at the same q/E amounts to Re=12.9×106, which was consequently defined as the lower Reynolds number of the reference measurements. Two principal Mach numbers are envisioned for the bulk of the measurements based on the preliminary CFD analyses: M=0.84 and M=0.90. A third Mach number M=0.87 is included without any PSP/PIV experiments to expand the aerodynamic data base for research scenario 4. The test matrix is additionally expanded to low Reynolds numbers at Re=3.3×106 to provide validation data (but without PSP/PIV) for costly LES-type simulations. Ultimately the flow conditions listed in Table 2 were identified and examined by unsteady PSP measurements during test entry MK1a. The MK1a unsteady PSP measurements were supplemented by steady PSP measurements for further angles of attack and completed by forces and moments measurements as well as steady and unsteady (Kulite) pressure measurements. The MK1b campaign supplemented the PSP data from MK1a by PIV measurements, which were carried out at on a subset of the flow conditions listed in Table 2. This included conditions that were deemed most relevant for the four research scenarios. Finally, the measurements at the same flow conditions have been repeated in MK2 for the XRF1 configuration with attached UHBR through-flow nacelles to identify their impact on buffet and wake HTP interaction.

V. Selected Results from the First ETW Entry and Research Scenario 2 Simulations

A. Procedure of the First ETW Test Entry MK1a

The first ETW entry MK1a took place from November 9 to 20, 2020, and covered the measurements outlined in Table 2. Figure 9 (left) shows the XRF1 wind tunnel model including new HTP with instrumentation during the test preparation, while the right photograph shows the installed model with PSP coating. CAD-CUT dots were used for transition fixing for all measurements at Re=3.3×106 and Re=12.9×106. The dots were applied to the fuselage nose as well as to the wing, HTP, and VTP at 5% chord on both sides to ensure defined conditions for comparison with the numerical calculations, which are usually performed fully turbulent. The height of the dots was selected based on the Reynolds number and in accordance with standard practice for this type of test. For the low Reynolds number of 3.3×106, transition was fixed on the wing surface using dots with a height of 64  μm, whereas dots of 79  μm height were used for HTP and VTP. At Re=12.9×106, the smallest available dots of 38  μm were applied to all surfaces, which resulted in some overtripping. Except for the fuselage nose no fixing was used for the measurements at Re=25×106, as a fully turbulent flow was assumed [43].

The campaign started at the lowest Reynolds number and worked toward the highest. The wind tunnel shakedown runs were followed by runs at Re=3.3×106 at atmospheric conditions. Following these runs the transition fixing was exchanged. Runs at Re=12.9×106 and M=0.84, 0.87, and 0.90 and at q/E=0.4×106 were performed at a temperature of 180 K. Thereafter measurement blocks at a cryogenic temperature of 115 K at Re=12.9×106, q/E=0.2×106 and Re=25×106, q/E=0.4×106 were performed after removing the transition dots. Standard force and static pressure measurement, dynamic pressure measurement, deformation, and accelerometer data were acquired in all of these runs. These data were used for aerodynamic characterization of the XRF1 configuration and to determine maximum achievable angles of attack. In situ data analysis and assessments served to confirm the preselected flow conditions (cf. Sec. IV.E) for buffet onset and the interaction between wing wake and HTP and to adjust them for the PSP measurements if necessary.

Fig. 9
Fig. 9

XRF1 wind tunnel model during preparation and in the test section (with courtesy ETW).

Steady and unsteady PSP measurements at both fin-up and fin-down model orientations were conducted for the two main blocks at q/E=0.4×106 after the polar measurements in order to acquire pressure data on both the upper and lower sides of the wing and the HTP. The wing and HTP unsteady PSP measurements are synchronized, and the unsteady PSP data can be referred to the Kulite data. PSP was acquired at two angles in the linear regime for each Mach number, at one incidence per condition on the lower side of the wing, and at the three incidences for which unsteady PSP was measured.

The procedure of selecting unsteady PSP conditions was continuously adapted in order to maximize the signal length that can be acquired from a given set of conditions with the least possible nitrogen consumption. All unsteady PSP measurements were carried out using 1000 frames per second (fps), with selected conditions attempted at 2000 fps. Unsteady PSP data were acquired only at M=0.84 and M=0.90.

The above procedure was carried out for two main conditions: Re=12.9×106, q/E=0.4×106 and Re=25×106, q/E=0.4×106. The number of unsteady PSP points and the acquisition settings varied slightly due to the ad hoc nature of the final incidence selection. Nevertheless, the required three incidences around buffet-onset were measured via unsteady PSP for both of these conditions, with some additional data acquired at more incidences where time and budget allowed (compare Table 2).

An additional block at Re=12.9×106 and q/E=0.2×106 was used to approach the highest possible incidence by reducing the expected model vibrations. This block did not include the entire procedure described above and was specifically geared at obtaining data at high angle of attack for scenario 2. Some overlap with the buffet conditions at Re=12.9×106 and q/E=0.4×106 exists, enabling cross-comparison. Only fin-down orientation is possible at such high angles of attack due to the sting offset of 5° and the resulting geometrical limitations of the model support system. Therefore PSP was carried out only on the upper sides of the wing and the HTP for this test condition.

The collected measurement data are used to validate the numerical models and for flow physics studies according to the individual research objectives of the partners of the research unit. In the following, selected numerical and experimental results for research scenario 2, i.e., the impact of the flow separation from the wing and the wake on the HTP, are discussed.

B. Results for Wake–HTP Interactions at High Incidence

In the following, selected results from ETW entry MK1a (PSP campaign, XRF1 configuration with HTP but without UHBR) are shown and compared with AZDES simulation results. The focus of the present study is on the plausibility check of the MK1a flow conditions chosen to provide meaningful test data for research scenario 2 as introduced in Sec. II. The procedure was carried out in an analogous manner for the other research scenarios, whereby scenario-specific evaluation criteria were taken into account.

For research scenario 2, a Mach number of M=0.90 and a Reynolds number of 12.9 million have been considered in the ETW test. The measurements were carried out under cryogenic conditions at a temperature of T=115  K. For the anticipated high Reynolds number, the low temperature enables realization of a small dynamic pressure (q/E=0.2×106), which reduces the model loads, yielding an extension of the measurable angle-of-attack range up to occurrence of wing root separation. The tests at this condition were divided into two blocks with the first block including continuous polars to assess the overall aerodynamic characteristics and the maximum model incidence possible, and the second block including the PSP measurements at constant model incidences. At M=0.90 the first shock-induced flow separation can be observed to occur in the outer wing area for the XRF1 wind tunnel model. At the same incidence the flow in the wing root domain is fully attached. With increasing angle of attack, the shock strength and the amplitude of the shock oscillations successively grow and the trailing edge separation expands toward the inner wing area. The dynamics of the shock motion can be seen for an angle-of-attack range of 5–7° in Fig. 10, where the root mean square (RMS) values of the intensity of the emitted light as a measure of the pressure fluctuations are shown as resulting from evaluation of the unsteady PSP measurements. The yellow stripe, which widens continuously as the angle of attack increases, represents the regime of the shock oscillations. The image also shows the locations of the unsteady pressure sensors (Kulites) as black and red dots. The areas where no PSP paint could be applied appear white. At the smallest depicted incidence of α=5°, the strongest shock motion occurs at about 2/3 of the half span, where a kink in the shock front can be observed for higher incidences. With increasing incidence, noticeable shock oscillations can also be observed further inboard. At the same time, the shock front moves toward the leading edge, whereby the innermost area of the shock oscillations remains at the transition to the lambda-shock pattern near the spanwise position of the Yehudi. From this measurement data it is also clear that stronger shock motions, which are associated to flow separation in the wing root area, only occur at the highest angles of attack shown in Fig. 10 (α=6.9° and 7°).

An abrupt decrease of recompression in the measured static pressure distribution (not shown here) downstream of the rear shock leg confirms the onset of flow separation in the wing root domain at an incidence of just below 7°. For further identification of the angle of attack relevant for the study of the wake–HTP interactions, spectra of the pressure fluctuations in the wing root area were evaluated. The unsteady pressure fluctuations were measured with two Kulite sensors in the inboard wing area that are located at a spanwise position of η=26% and axial locations of 70% (KUP10402) and 95% (KUP10401) of the local wing chord, respectively (cf. Fig. 11). For the upstream Kulite 10402 (Fig. 11, left), the amplitude level for α6.5° is low overall but shows two prominent peaks at 50 and at 140 Hz, respectively. The 50 Hz peak can be attributed to some power line hum noise that was observed for some Kulites during the campaign. The second peak at 140 Hz is assumed to be associated with an acoustic phenomenon and structural wing eigenmodes. On the one hand, it can be observed that the peak frequency visible in the pressure spectra scales with the speed of sound, suggesting that it is induced by an acoustic phenomenon, whereas a comprehensive flow quality dataset has recently been acquired at ETW and a detailed assessment is currently ongoing [44]. On the other hand, peaks in this frequency range that correlate to higher wing bending eigenmodes are visible in accelerometer and in Kulite data across most of the flow conditions at different angles of attack, Reynolds numbers, and Mach numbers. For the flow condition depicted in Fig. 11, both effects occur and contribute to the amplitude increase in the 140 Hz range. The remaining spectrum is of broadband nature and represents the turbulence of the attached flow in this domain. At both sensor positions, a sudden amplitude rise can be observed when the angle of attack is increased from 6.5° to 6.9°. While for the upstream Kulite 10402 the abrupt amplitude increase occurs over the entire frequency range, the amplitude level is already higher for the downstream Kulite 10401 (Fig. 11, right) for the smaller angles of attack and the augmentation becomes visible only for frequencies below 300  Hz. The strong increase of the pressure fluctuations for α6.9° suggests the existence of large turbulent structures caused by flow separation in this wing area. The flow separation involves such high and broadband fluctuation amplitudes that the peaks at 50 and 140 Hz observed in the Kulite 10402 spectra become masked.

Fig. 10
Fig. 10

RMS value of the intensity of the emitted light on the wing upper side, XRF1 configuration with HTP, unsteady PSP measurement, M=0.90, Re=12.9×106, q/E=0.2×106.

In this angle-of-attack range, the influence of the wing on the HTP loads also changes qualitatively for the chosen HTP installation angle. The measured pressure distributions in the outboard domain of the HTP show that, from α just below 6°, at least this area of the HTP is again increasingly loaded rather than unloaded with increasing incidence; i.e., the generated down force grows in this area of the HTP with increasing α. This shows a change in wing downwash and induction, which is attributed to the onset of inboard wing separation. Likewise, the pressure fluctuations at the HTP are enhanced, and the amplitude level abruptly increases when α is increased from 6.5° to 6.9°; see Fig. 12 (left). However, the characteristics of the spectra at frequencies above 300 Hz differ from those near the trailing edge of the wing (Kulite 10401 in Fig. 11, left) in terms of angle-of-attack dependence. While the spectrum at the wing reveals an almost α-independent, high level in this frequency range, the spectra at the HTP show a clear angle-of-attack dependence. Below the onset of wing root separation (α6.5°), the amplitudes at the HTP are smaller compared to those near the wing trailing edge and show a growing level with increasing angle of attack. This is attributed to the successively decreasing vertical distance of the turbulent wing wake from the HTP that was observed in RANS calculations, leading to an amplitude increase of the pressure fluctuations imposed on the HTP. At the onset of wing root separation (α6.9°), the amplitude level at the HTP is amplified compared to the wing for higher frequencies, which is attributed to the impingement of the separated wing flow on the HTP. For all angles of attack, the HTP spectra show the peak at 140 Hz, which was already observed in the Kulite spectra of the wing. Figure 12 (right) exemplarily shows the deviation of the instantaneous pressure coefficient from the mean pressure for the lower side of the HTP as resulting from the AZDES simulation for an angle of attack of 7°. The footprint of a shock wave is visible in the outer HTP area, and the yellowish areas show regions with increased unsteadiness. An area of increased unsteadiness can also be observed at the leading edge of the HTP tip, caused by the impingement of turbulent structures in the wake. For the discussed reasons, an angle of attack of α=7° is considered representative for further research scenario 2 studies on the interactions between wing wake and HTP, and the subsequent numerical analyses are conducted for this inflow condition.

Fig. 11
Fig. 11

Measured pressure spectra at the upper side of the inboard wing, XRF1 configuration with HTP, M=0.90, Re=12.9×106, q/E=0.2×106.

To verify the simulations the static pressure distributions measured in ETW and calculated by means of the AZDES with the setup described in Sec. III.B are compared for an inboard wing section and a HTP outboard section in Fig. 13. The measured wing pressure distributions are depicted for two closely adjacent angles of attack. For the higher incidence, the mean pressure distribution at the downstream leg of the lambda shock is clearly smeared, indicating unsteadiness of the shock. At the same time, the recompression downstream of the shock decreases, confirming the onset of flow separation. The AZDES result for α=7° matches the measurement for α=6.78° in terms of cp distribution on upper and lower side including the positions of the upstream and downstream legs of the lambda shock pattern. It should be noted that the attached domain of the boundary layer in this section is still calculated in RANS mode, so that these deviations are probably due to deficiencies of the turbulence model. The right picture shows the pressure distributions for an outer section of the HTP, which is except of a small spanwise offset of about 2 mm almost directly downstream of the wing section shown in the left picture. With the chosen HTP angle and the considered angle of attack, the HTP generates downwash, so that the shock visible in the pressure distribution around midchord relates to the lower side (compare also Fig. 12, right). It appears that the area between the upper and lower side pressure distributions grows when the angle of attack increases from α=6.78° to α=7.05°, confirming higher down force with increasing incidence at this tailplane section caused by the changed induction of the wake after inboard separation sets in. The AZDES result shows deviations from the measurements in the forward part of the upper side and the shock wave pressure gradient, but matches the remaining distribution and the recompression on both HTP sides reasonably well. In the prediction, the pressure distribution in the shock domain is less smeared, suggesting a reduced shock oscillation compared to the measurement. It should be noted that the resulting HTP pressure distribution is very sensitive toward the correct prediction of the complex 3D separation in the wing root region and the accurate calculation of trajectory and development of the wake. Slight deviations can lead to visible differences in the HTP pressure distribution in the direct interaction area. Overall it is supposed that the wake development and the downwash are predicted physically correct and that the AZDES results can be used for the subsequent wake analyses.

Fig. 12
Fig. 12

Measured pressure spectra (Kullite A) at the lower side of the HTP (left) and AZDES result for α=7° (right), XRF1 configuration with HTP, M=0.90, Re=12.9×106, q/E=0.2×106.

The separation topology at the inboard wake and the resulting wake evolution toward the HTP are analyzed in order to better characterize the interaction between wing and HTP for the high incidence case. In Fig. 14, the limiting streamlines as predicted by the AZDES are shown together with the sign of the streamwise skin friction to illustrate the separation topology at the inner wing. Based on works by Legendre and Werle, as well as by Lighthill, a classification of 3D singular points that enables formal analysis of complex separation topologies was introduced by Tobak and Peake [45]. Figure 14 depicts the predicted time-averaged streamlines for the inboard wing domain, where several singular points can be identified. Outboard of the small corner separation at the wing–fuselage junction, a focus point of separation can be observed, which is characterized by convergence of the skin friction lines toward a singular point and represents a foot print of a separating vortex. In the direct vicinity, a saddle point is located, where the flow splits up toward this focus point and a second focus point, which is located further outboard above the most inboard flap track fairing where the legs of the lambda shock pattern merge. A line of separation defined by converging streamlines from up- and downstream directions connects the saddle point and the second focus point. This focus point of separation is fed from a larger area of the upper wing surface. The thick black line illustrates the spanwise location of the pressure distribution discussed and shown in Fig. 13 on the left. The line of separation intersects the section of the pressure tappings at 85% of the local wing chord, which correlates with the observed reduced recompression downstream of the shock that is located at about 80% chord. This confirms shock-induced flow separation in the root region of the wing. Further outward, another separation line and a saddle point follow. Flow separation in this region and the outer wing occurs significantly upstream compared to the wing root separation inboard of the first flap track fairing.

Fig. 13
Fig. 13

Pressure distribution at an inboard wing section (η=23.3%, left) and outboard HTP section (η=70%, right); ETW measurements vs AZDES result (α=7°), XRF1 configuration with HTP, M=0.90, Re=12.9×106, q/E=0.2×106.

As described by Tobak and Peake [45], the lines of separation form the base of so-called dividing surfaces in the flowfield. These dividing surfaces roll up in the sense of rotation of the vortex filaments emanating from the second focus point into the flowfield and constitute a pronounced streamwise vortex in the wake. In the inner region the vertical extent of the wake is considerably smaller than in the outer wing area, hardly any larger-scale turbulent structures are visible, and the wake passes far beneath the HTP, as is depicted in Fig. 15, which shows the distribution of the absolute value of the vorticity distribution in a spanwise plane just upstream of the HTP leading edge. In the transition from the inner, weakly separated flow region to the outer domain with strong separation, a counterclockwise rotational motion in the wake can be observed, originating from the inboard wing separation described above. On its way from the wing to the HTP the vortex moves outboards and breaks up into larger vortical structures. These vortical structures impinge the tip region of the HTP directly and significant interactions are to be expected, as intended for research scenario 2. Figure 15 further shows footprints of longitudinal vortical structures emanating from the flap track fairings that, however, do not impinge the HTP.

Fig. 14
Fig. 14

Pressure distribution and limiting streamlines at the upper side of the inboard wing; AZDES result (α=7°), XRF1 configuration with HTP, M=0.90, Re=12.9×106. The solid lines mark the section for which the pressure distribution is depicted in Fig. 13.

To further visualize the streamwise evolution of the separated flow structures in the wake, Fig. 16 shows the vorticity distribution between wing and HTP. The distributions are shown in two sections that hit the HTP slightly inboard of the tip. The figure confirms that the tip region of the HTP is completely immersed in the viscous wake of the wing that is characterized by large turbulent structures in the spanwise domain, yielding strong interactions with the HTP. Based on these simulation results, the PIV planes considered in MK1b were positioned in the ηHTP=85% section that cover the area from the wing to the HTP to characterize the evolution of vortical structures up to the HTP inflow.

Fig. 15
Fig. 15

Predicted wing wake in a plane just upstream of the HTP leading edge; AZDES result (α=7°), XRF1 configuration with HTP, M=0.90, Re=12.9×106.

Fig. 16
Fig. 16

Vorticity distribution in the wing wake in streamwise cuts through the HTP tip region; AZDES result (α=7°), XRF1 configuration with HTP, M=0.90, Re=12.9×106.

VI. Conclusions

This paper presents a new research initiative on combined numerical and experimental studies of the aircraft high-speed stall phenomenon and 3D buffet mechanisms and outlines its main research objectives. Within this initiative, detailed unsteady PSP pressure distribution and PIV flowfield measurements are conducted using the generic Airbus XRF1 transport aircraft configuration under cryogenic flow conditions. The experiments are carried out in the ETW up to flight-relevant Reynolds numbers. In parallel, simulations are performed with LES and hybrid RANS/LES methods for synergistic flow-physical analysis, supplemented by URANS calculations with dedicated RSM turbulence models. In the combined numerical–experimental studies, four different research scenarios characterizing different flow features are considered, which also include the study of flow separation from the wing root domain and the interaction of the viscous wake with the flow around the HTP. In November 2020, a first PSP campaign on the XRF1 with a new instrumented HTP took place. The present paper discusses analyses of these measurements for a Mach number of M=0.90, Re=12.9×106, and at high incidence of α=7° along with results of a hybrid RANS/LES simulation. For these flow conditions inboard wing separation and pronounced interaction of the wake with the HTP were observed.

It was shown that inboard separation only occurs at very high angles of attack that are considerably above buffet onset and close to the vibration limit of the wind tunnel model at the dynamic pressures realized in the tests. At this high incidence, a 3D separation forms in the inboard domain of the wing, which develops into a streamwise vortical structure in the wake, which impinges the HTP in the outboard area and causes unsteady load fluctuations there. At the same time, the downwash exhibits characteristic changes, reversing the local HTP load gradient, i.e., yielding stronger down force in the outboard domain of the HTP with an increase in the angle of attack. Overall, the present study shows that the flow conditions selected in the first wind tunnel entry provide a good basis for dedicated studies on interactions of the wake downstream of flow separation with the HTP flow.

Recently detailed time-resolved PIV measurements have been conducted for the same configuration and flow conditions considered in the first PSP ETW entry. PIV planes above the wing at different streamwise and spanwise positions in the wake toward the inflow domain of the HTP will be examined. Together with the highly resolved simulations, the test results will serve for detailed flow-physical analysis of the development of the turbulent wake and its interaction with the HTP as well as for flow physics studies of the other research scenarios, like buffet onset studies. In further ETW entries, aerodynamic and unsteady PSP and PIV measurements on the XRF1 configuration with a newly designed, instrumented UHBR flow through nacelle have been be carried out from the end of 2021. These tests and corresponding simulations will serve to identify the UHBR influences on high-speed stall phenomena as well as the wake development and the HTP impact.


The authors gratefully acknowledge the Deutsche Forschungsgemeinschaft DFG (German Research Foundation) for funding this work in the framework of the research unit FOR 2895. The authors would like to thank the Helmholtz Gemeinschaft HGF (Helmholtz Association), Deutsches Zentrum für Luft- und Raumfahrt DLR (German Aerospace Center), for financing the wind tunnel measurements and Airbus for providing the wind tunnel model as well as public support to mature the test methods applied by DLR and ETW. The authors would also like to thank all the scientists involved in the research initiative for the kind cooperation, as well as the colleagues at Airbus and European Transonic Windtunnel, whose invaluable support has made the project possible.


  • [1] Lee B., “Transonic Buffet on a Supercritical Aerofoil,” Aeronautical Journal, Vol. 94, No. 935, 1990, pp. 143–152. CrossrefGoogle Scholar

  • [2] Jacquin L., Molton P., Deck S., Maury B. and Soulevant D., “Experimental Study of Shock Oscillation over a Transonic Supercritical Profile,” AIAA Journal, Vol. 47, No. 9, 2009, pp. 1985–1994. LinkGoogle Scholar

  • [3] Hartmann A., Feldhusen A. and Schröder W., “On the Interaction of Shock Waves and Sound Waves in Transonic Buffet Flow,” Physics of Fluids, Vol. 25, No. 2, 2013, Paper 026101. CrossrefGoogle Scholar

  • [4] Sartor F., Mettot C. and Sipp D., “Stability, Receptivity, and Sensitivity Analyses of Buffeting Transonic Flow over a Profile,” AIAA Journal, Vol. 53, No. 7, 2015, pp. 1980–1993. LinkGoogle Scholar

  • [5] Iovnovich M. and Raveh D. E., “Numerical Study of Shock Buffet on Three-Dimensional Wings,” AIAA Journal, Vol. 53, No. 2, 2015, pp. 449–463. LinkGoogle Scholar

  • [6] Koike S., Ueno M., Nakakita K. and Hashimoto A., “Unsteady Pressure Measurement of Transonic Buffet on NASA Common Research Model,” AIAA Paper 2016-4044, June 2016. LinkGoogle Scholar

  • [7] Dandois J., “Experimental Study of Transonic Buffet Phenomenon on a 3D Swept Wing,” Physics of Fluids, Vol. 28, No. 1, 2016, Paper 016101. CrossrefGoogle Scholar

  • [8] Sugioka Y., Koike S., Nakakita K., Numata D., Nonomura T. and Asai K., “Experimental Analysis of Transonic Buffet on a 3D Swept Wing Using Fast-Response Pressure-Sensitive Paint,” Experiments in Fluids, Vol. 59, No. 6, 2018, pp. 1–20. CrossrefGoogle Scholar

  • [9] Paladini E., Dandois J., Sipp D. and Robinet J.-C., “Analysis and Comparison of Transonic Buffet Phenomenon over Numerical Simulation Several Three-Dimensional Wings,” AIAA Journal, Vol. 57, No. 1, 2019, pp. 379–396. LinkGoogle Scholar

  • [10] Lutz T., Gansel P. P., Godard J.-L., Gorbushin A., Konrath R., Quest J. and Rivers S. M. B., “Going for Experimental and Numerical Unsteady Wake Analyses Combined with Wall Interference Assessment by Using the NASA CRM-model in ETW,” AIAA Paper 2013-0871, Jan. 2013. LinkGoogle Scholar

  • [11] Sartor F. and Timme S., “Delayed Detached–Eddy Simulation of Shock Buffet on Half Wing–Body Configuration,” AIAA Journal, Vol. 55, No. 4, 2017, pp. 1230–1240. LinkGoogle Scholar

  • [12] Ehrle M., Waldmann A., Lutz T. and Krämer E., “Simulation of Transonic Buffet with an Automated Zonal DES Approach,” CEAS Aeronautical Journal, Vol. 11, No. 4, 2020, pp. 1025–1036. CrossrefGoogle Scholar

  • [13] Crouch J. D., Garbaruk A. and Strelets M., “Global Instability in the Onset of Transonic-Wing Buffet,” Journal of Fluid Mechanics, Vol. 881, Oct. 2019, pp. 3–22. CrossrefGoogle Scholar

  • [14] Plante F., Dandois J., Beneddine S., Laurendeau E. and Sipp D., “Link Between Subsonic Stall and Transonic Buffet on Swept and Unswept Wings: From Global Stability Analysis to Nonlinear Dynamics,” Journal of Fluid Mechanics, Vol. 908, Dec. 2020, p. A16. CrossrefGoogle Scholar

  • [15] Nitzsche J., Ringel L. M., Kaiser C. and Hennings H., “Fluid-Mode Flutter in Plane Transonic Flows,” IFASD 2019—International Forum on Aeroelasticity and Structural Dynamics, 2019, Google Scholar

  • [16] Timme S., “Global Instability of Wing Shock-Buffet Onset,” Journal of Fluid Mechanics, Vol. 885, Jan. 2020, p. A37. CrossrefGoogle Scholar

  • [17] Masini L., Timme S. and Peace A. J., “Analysis of a Civil Aircraft Wing Transonic Shock Buffet Experiment,” Journal of Fluid Mechanics, Vol. 884, Dec. 2019, p. 906. Google Scholar

  • [18] Havas J. and Rabadan G., “Prediction of Horizontal Tail Plane Buffeting Loads,” International Forum on Aeroelasticity and Structural Dynamics, IFASD Paper 2009-128, 2009. Google Scholar

  • [19] Illi S. A., Lutz T. and Krämer E., “Transonic Tail Buffet Simulations on the ATRA Research Aircraft,” Computational Flight Testing: Notes on Numerical Fluid Mechanics and Multidisciplinary Design, Vol. 123, edited by Kroll N., Radespiel R., Burg J. and Sϕrensen K., Springer, Berlin, 2013. CrossrefGoogle Scholar

  • [20] Waldmann A., Konrath R., Lutz T. and Krämer E., “Unsteady Wake and Tailplane Loads of the Common Research Model in Low Speed Stall,” New Results in Numerical and Experimental Fluid Mechanics XII, edited by Dillmann A., Heller G., Krämer E., Wagner C., Tropea C. and Jakirlic S., Notes on Numerical Fluid Mechanics and Multidisciplinary Design, Vol. 142, Springer, Cham, Switzerland, 2020. CrossrefGoogle Scholar

  • [21] Tan K. B., Wang P. and Srigrarom S., “Low-Speed Post-Stall Wing Wake Impingement on Horizontal Stabilizer of the Common Research Model,” AIAA Paper 2018-3898, June 2018. LinkGoogle Scholar

  • [22] Mann A., Thompson G. and White P., “Civil Aircraft Wind Tunnel Feature Rich Testing at the Edge of the Envelope,” Proceedings of the 54th 3AF International Conference on Applied Aerodynamics, Association Aéronautique et Astronautique de France, Paper FP68-AERO2019-mann, 2019. Google Scholar

  • [23] Konrath R., Geisler R., Agocs J., Ehlers H., Philipp F. and Quest J., “High-Speed PIV Applied to Wake of NASA CRM Model at High Re-Number Sub- and Transonic Stall Conditions,” CEAS Aeronautical Journal, Vol. 9, No. 2, 2018, pp. 339–346. CrossrefGoogle Scholar

  • [24] Lutz T., Gansel P. P., Waldmann A., Zimmermann D.-M. and Schulte am Hülse S., “Prediction and Measurement of the Common Research Model Wake at Stall Conditions,” Journal of Aircraft, Vol. 53, No. 2, 2016, pp. 501–514. LinkGoogle Scholar

  • [25] Waldmann A., Gansel P. P., Lutz T. and Krämer E., “Unsteady Wake of the NASA Common Research Model in Low-Speed Stall,” Journal of Aircraft, Vol. 53, No. 4, 2016, pp. 1073–1086. LinkGoogle Scholar

  • [26] Rivers M., Hunter C. and Campbell R., “Further Investigation of the Support System Effects and Wing Twist on the NASA Common Research Model,” AIAA Paper 2012-3209, June 2012. LinkGoogle Scholar

  • [27] König B. and Fares E., “Validation of a Transonic Lattice-Boltzmann Method on the NASA Common Research Model,” AIAA Paper 2016-2023, Jan. 2016. LinkGoogle Scholar

  • [28] Waldmann A., Lutz T. and Krämer E., “Wind Tunnel Support System Influence on NASA Common Research Model at Low-Speed Conditions,” Journal of Aircraft, Vol. 55, No. 5, 2018, pp. 1762–1772. LinkGoogle Scholar

  • [29] Deck S., “Zonal-Detached-Eddy Simulation of the Flow Around a High-Lift Configuration,” AIAA Journal, Vol. 43, No. 11, 2005, pp. 2372–2384. LinkGoogle Scholar

  • [30] Schulte am Hülse S., “Simulation of Transonic Buffet on Transport Aircraft Using Hybrid RANS/LES Methods,” Ph.D. Thesis, German Language, Univ. of Stuttgart, Verlag Dr. Hut, Munich, 2016. Google Scholar

  • [31] Schwamborn D., Gerhold T. and Heinrich R., “The DLR TAU-Code: Recent Applications in Research and Industry,” European Conference on Computational Fluid Dynamics ECCOMAS CFD 2006 CONFERENCE, Delft Univ. of Technology, 2006, Google Scholar

  • [32] Kok J., “A High-Order Low-Dispersion Symmetry-Preserving Finite-Volume Method for Compressible Flow on Curvilinear Grids,” Journal of Computational Physics, Vol. 228, No. 18, 2009, pp. 6811–6832. CrossrefGoogle Scholar

  • [33] Spalart P. and Allmaras S., “A One-Equation Turbulence Model for Aerodynamic Flows,” AIAA Paper 1992-0439, Jan. 1992. LinkGoogle Scholar

  • [34] Allmaras S. R. and Johnson F. T., “Modifications and Clarifications for the Implementation of the Spalart-Allmaras Turbulence Model,” Seventh International Conference on Computational Fluid Dynamics (ICCFD7), Paper ICCFD7-1902, June 2012. Google Scholar

  • [35] Cécora R.-D., Radespiel R., Eisfeld B. and Probst A., “Differential Reynolds-Stress Modeling for Aeronautics,” AIAA Journal, Vol. 53, No. 3, 2015, pp. 739–755. LinkGoogle Scholar

  • [36] Spalart P. R., “Strategies for Turbulence Modelling and Simulations,” International Journal of Heat and Fluid Flow, Vol. 21, No. 3, 2000, pp. 252–263. CrossrefGoogle Scholar

  • [37] Hefer G., “ETW—A Facility for High Reynolds Number Testing,” IUTAM Symposium Transsonicum IV. Fluid Mechanics and its Applications, edited by Sobieczky H., Springer, Dordrecht, The Netherlands, 2003, pp. 157–164. Google Scholar

  • [38] Yorita D., Klein C., Henne U., Ondrus V., Beifuss U., Hensch A.-K., Longo R., Guntermann P. and Quest J., “Successful Application of Cryogenic Pressure Sensitive Paint Technique at ETW,” AIAA Paper 2018-1136, Jan. 2018. LinkGoogle Scholar

  • [39] Watkins A. N., Lipford W. E., Leighty B. D., Goodman K. Z., Goad W. K. and Goad L. R., “Results from a Pressure Sensitive Paint Test Conducted at the National Transonic Facility on Test 197: the Common Research Model,” NASA TM 2011-217065, 2011. Google Scholar

  • [40] Yorita D., Klein C., Henne U., Ondruss V., Beifuss U., Hensch A.-K., Guntermann P. and Quest J., “Investigation of a Pressure Sensitive Paint Technique for ETW,” AIAA Paper 2017-0335, Jan. 2017. LinkGoogle Scholar

  • [41] Geisler R., “A Fast Multiple Shutter for Luminescence Lifetime Imaging,” Measurement Science and Technology, Vol. 28, No. 9, 2017, Paper 095403. CrossrefGoogle Scholar

  • [42] Schreyer A.-M., Larchevêque L. and Dupont P., “Method for Spectra Estimation from High-Speed Experimental Data,” AIAA Journal, Vol. 54, No. 2, 2016, pp. 557–568. LinkGoogle Scholar

  • [43] Gross N., “ETW Analytical Approach to Assess the Wing Twist of Pressure Plotted Wind Tunnel Models,” AIAA Paper 2002-0310, Jan. 2002. LinkGoogle Scholar

  • [44] You J., Wright M. and Quix H., “Recent Activities on Flow Quality Assessment at the European Transonic Windtunnel,” ICAS 2022, 33rd International Congress of the Aeronautical Sciences, Paper ICAS2022_0061, 2022. Google Scholar

  • [45] Tobak M. and Peake D. J., “Topology of Three-Dimensional Separated Flows,” Annual Review of Fluid Mechanics, Vol. 14, No. 1, 1982, pp. 61–85. CrossrefGoogle Scholar


Table 1 Wind tunnel entries during the first phase of the research initiative

MK1a 11/2020Clean wing——————————
MK1b 3/2022Clean wing——————
MK2 12/2021, 1/2022UHBR ————
MK3 10/2022UHBR————————

Table 2 MK1a polar measurements and wing upper side unsteady PSP measurements for the four research scenarios

ScenarioMRe,×106α, °Remark
10.8412.93/3.5/4Unsteady PSP
 0.84253/3.5/4/4.5/5Unsteady PSP
 0.9012.92.5/4/5/5.5Unsteady PSP
 0.90252.5/4/5/6Unsteady PSP
20.9012.95/6/6.5/6.9/7Unsteady PSP, reduced q/E
30.84/0.903.3PolarNo PSP, no PIV
 0.8412.9/254PSP, no PIV
 0.9012.9/252PSP, no PIV
40.84/0.87/0.9012.9/25PolarNo PSP, no PIV