Flow structure and unsteadiness in the supersonic wake of a generic space launcher

Transcription

1 Sonderforschungsbereich/Transregio 4 Annual Report Flow structure and unsteadiness in the supersonic wake of a generic space launcher By Anne-Marie Schreyer, Sören Stephan AND Rolf Radespiel Institute of Fluid Mechanics, Technische Universität Braunschweig Hermann-Blenk-Str. 37, 3818 Braunschweig The wake flow of a generic axisymmetric space-launcher model with and without propulsive jet is investigated. Measurements are performed at Mach 2.9 and a Reynolds number Re D = based on model diameter D, and the nozzle exit velocity of the jet is at Mach 2.5. Velocity measurements in the wake flow by means of Particle Image Velocimetry and mean and unsteady wall-pressure measurements on the main-body base are performed simultaneously. This way, the evolution of the wake flow was observed along with its spectral content. In this paper, the mean flow topology and turbulence behavior of the wake is described based on PIV measurements, and the influence of the afterexpanding jet plume is discussed. For the case without propulsive jet, a large separated zone is forming downstream of the main body shoulder and the flow is reattaching downstream on the nozzle fairing. Under the influence of the jet plume, the separated region is increased and reattachment does not occur anymore. The flow is displaced away from the wall. The jet plume appears to have a stabilizing effect on the wake flow. The development of the shear layer and the magnitude of the turbulent intensities are damped, which we showed with the axial and radial components of the turbulent velocity fluctuations, as well the Reynolds shear stress evolution along the launcher afterbody. 1. Introduction The afterbody geometry of a classical space launcher is characterized by an abrupt step decrease in diameter at the junction between main body and rocket engine, which causes the separation of the turbulent boundary layer on the main body at the shoulder of the launcher base. A large recirculation region forms downstream of the step and a turbulent shear layer develops. This separation-dominated flow field is highly unstable and induces strong wall-pressure oscillations, which can excite structural vibrations detrimental to the launcher [1]. A better understanding of the flow field is thus crucial to lay the groundwork for minimizations of those detrimental effects and efficient launcher design. The conditions and topology of the wake flow vary tremendously along the flight trajectory of the launcher. To a large part this is due to the influence of the propulsive jet, which becomes more and more underexpanded with altitude. The resulting afterexpanding jet plume has a strong displacement effect on the outer flow. This enlarges the recirculation region and causes an increase in pressure at the main-body base [2]. Hence depending on the stage in the flight trajectory, the dimensions of the afterbody, and the interaction with the jet plume the flow may or may not reattach on the outer wall of the engine nozzle, causing different mean and turbulent flow topologies. For this reason, experimental and numerical studies covering a wide range of flow

2 94 A.-M. Schreyer, S. Stephan & R. Radespiel Mach numbers have been performed. In the transonic range, Deprés et al. [1], Deck & Thorigny [3], and Weiss et al. [4] have investigated cases both with and without propulsive jet. In supersonic flow, such cases were studied by Bannink et al. [5], Bourdon & Dutton [6], Janssen & Dutton [7], and Stephan et al. [8], for example. Saile et al. [9, 1] and Statnikov et al. [2] studied hypersonic cases. These studies describe the mean flow topology and survey base pressure fluctuations including dominant frequencies or observe the formation of turbulent structures in the wake. In supersonic flow, which is relevant for space launchers at higher altitudes, the propulsive jet is usually underexpanded at the nozzle exit, leading to an afterexpanding jet plume that has a strong displacement effect on the outer flow. Therefore, a large recirculation zone forms around the nozzle, causing an increase in pressure at the mainbody base [2]. This coincides with an increased base-pressure fluctuation level, so that structural vibrations are excited that may be critical for the launcher. If the flow is not reattaching on the nozzle fairing, also thermal loads can become a problem: hot gases from the propulsive jet may be convected upstream in the separation zone, potentially harming the structure. The chosen test case is thus critical in respect to structural loads and therefore of special interest. It is crucial to identify the sources of such potentially detrimental base pressure fluctuations and eventually understand the governing mechanisms in the turbulent wake flow field. Recently, progress has been made in the source identification of pressure fluctuations at specific characteristic frequencies. Janssen & Dutton [7] experimentally investigated a supersonic base flow at Mach2.46, and described wall-pressure fluctuations at a dominant frequency of a Strouhal number of St D =.1 (based on main-body diameter D), which they attributed to large turbulent structures in the separated region. These structures in the developing shear layer ware altered under the influence of an afterexpanding jet plume [2, 7]. Deprés et al. [1] investigated several generic launcher configurations in transonic flow experimentally and found that the shedding of large-scale structures from the separation bubble affects the wall-pressure fluctuations. The shedding is governed by interactions between the structures formed in the shear layer and the wall in the reattachment region. The characteristic frequency for this phenomenon is around St D =.2, and has been observed in the signal for both reattaching and non-reattaching flow, as well as for cases with and without supersonic propulsive jet. These findings were later confirmed by Deck & Thorigny [3] with their numerical approach. Statnikov et al. [2] performed a sparsity promoting dynamic mode decomposition combined with statistical analysis on a zonal RANS/LES simulation of a generic launcher model at Mach 6, and identified the radial flapping motion of the shear layer as the source of a dominant peak at St D =.85, and the swinging motion of the shear layer contributes to the signal with a dominant peak at St D =.6. With this study, we intend to contribute to the understanding of the flow structure and governing mechanisms by providing detailed surveys of the mean and turbulent flow topologies in the supersonic wake of a generic space launcher model, as well as the influence of a moderately underexpanded propulsive jet on the wake flow. Our approach is to survey the turbulence structure in the developing shear layer together with its "footprint" on the afterbody. To achieve this, we measure the mean and turbulent velocity fields in the wake flow simultaneously with the time-resolved pressure fluctuations at the main body base. We study the wake flow of an axisymmetric generic space-launcher

3 Flow structure and unsteadiness in the supersonic wake 95 FIGURE 1. Sketch of the Hypersonic Ludwieg Tube Braunschweig (HLB) in supersonic configuration. model for two flow cases: a baseline case without propulsive jet and a jet-simulation case with a cold air jet. Experimental investigations are performed in the Hypersonic Ludwieg tube Braunschweig in supersonic configuration [11] at a flow Mach number of 2.9 and a Reynolds number Re D = based on model diameter D. This corresponds to a simulated trajectory point at 25 km altitude. Pressure fluctuations at the main body base are measured with Kulite pressure transducers, while the velocity field in a plane covering the development region of the shear layer in the wake is measured with Particle Image Velocimetry (PIV). This manuscript is structured as follows: The experimental setup is presented in section 2 and includes presentations of the experimental facility (section 2.1), the model (section 2.1), the PIV setup (section 2.2), and the investigated test cases (section 2.3). The experimental results of this study are shown in section 3 subdivided into a discussion of the overall flow topology (section 3.1) and the turbulence behavior (3.2) and the meaning and implications of these results is discussed. Conclusions follow in section Experimental facility and setup 2.1. Wind tunnel, jet simulation facility, and experimental model The experimental investigations were performed in the Hypersonic Ludwieg Tube Braunschweig (HLB) in its supersonic configuration (see Figure 1). In this variant of the original Mach 6 tunnel, the nozzle is replaced by a tandem nozzle configuration with a second nozzle and settling chamber, resulting in a Mach 2.9 supersonic wind tunnel with effective measurement times of approximately 6 ms and a unit Reynolds number range of Re = /m. For details on this configuration of the HLB facility see Wu et al. [11]. The experimental conditions for our present study were chosen to agree with the SFB TR4 supersonic test case (see for example Stephan et al. [8]) and are summarized in Table 1. The launcher model consists of a nose cone and two cylindrical bodies representing the main body and nozzle fairing, respectively. The geometry and dimensions of this generic model of a space launcher are shown in Figure 2. The model is installed in the wind tunnel with a sword-shaped strut support, fixed in a window opening in the upper test-chamber wall. Inlets are fitted into the top and bottom windows (used as a

4 96 A.-M. Schreyer, S. Stephan & R. Radespiel M, U, m/s Re D, p SP, bar T RK, K p t,rk, bar ±.25% 285 ±3.5% 1.52 TABLE 1. Experimental conditions: freestream Mach number M, mean freestream velocity U in the streamwise direction, Reynolds numberre D based on the model diameter D, pressurep SP in the storage tube, and temperaturet RK and pressure p RK in the settling chamber. FIGURE 2. Sketch of the generic space launcher model with dimensions in mm. Figure from Statnikov et al. [12]. cable outlet) in order to get a smooth inner surface in the test section and thus minimize disturbances. To investigate the influence of a propulsive jet on the launcher wake flow in the HLB facility, the jet simulation facility (TSA) described by Stephan et al. [8], is integrated in the wind tunnel model. This facility to simulate afterbody flows in the HLB facility works according to the same principle as the HLB itself, with a long heated storage tube that can be pressurized up to 14bar and heated up to9k outside of the wind tunnel, and a tandem nozzle consisting of two nozzles and an intermediate settling chamber (see Figure 3). The second nozzle represents the nozzle of the launcher model and was designed in the shape of an axisymmetric Truncated Ideal Nozzle (TIC) with a mean exit Mach number of 2.5, and a nozzle exit inner diameter of d e = 42 mm. The length axis of the launcher model with the TSA is aligned with the HLB center line. A wide range of conditions (pressure, temperature), as well as different gases (air and helium) can be applied in order to achieve similarity with realistic conditions as far as flow displacement by the shape of the plume and entrainment into the jet plume go [8]. For the present measurement campaign, cold air was used as working gas for the simulation of the moderately under-expanded propulsive jet (p e /p = 5.7). These conditions are by no means close to simulating realistic launcher propulsion conditions, but they contribute to the qualitative understanding of the influence of a propulsive jet on the wake flow, as well as to create a comparative data set. The jet simulation parameters are summarized in Table 2. Particle Image Velocimetry (PIV) measurements are performed in a vertical plane on the opposite side of the strut (θ = 18, bottom side of the launcher model) to reduce influences of the flow disturbances induced by the strut support as much as possible.

5 Flow structure and unsteadiness in the supersonic wake 97 FIGURE 3. Schematic of the jet simulation facility. Figure from Stephan et al. [8]. M e, p t,sc, bar T t,e, K p e/p, (u max u )/u max, TABLE 2. Jet simulation parameters: nozzle exit Mach number M e, jet total pressure p t,sc, total temperature T t,e at nozzle exit, nozzle to freestream pressure ratio p e/p, and freestream to maximum velocity ratio (u max u )/u max with u max = [(2κ)/(κ 1)(R T t)/m Mol ].5. Three Kulite XCS-93 pressure sensors with a pressure range of.35 bar absolute are flush-mounted into the main body base at a radial position of r/d =.42 and at different angles (θ = 18,19, and 24 ) to capture 3D effects to a small extent. The measurement locations are chosen based on previous studies on the same model geometry [8, 9] Particle Image Velocimetry setup A Litron Nano T18-15 PIV double-pulse Nd-YAG laser with a pulse energy of 15mJ was used to illuminate the area of interest, namely the wake-flow region. The lightsheet had a thickness of 1mm. The field of view (FoV) was illuminated from the downstream direction in order to minimize reflections. This means that the laser-optical setup was placed downstream of the vacuum tank, which was equipped with a quartz-glass window. Due to the long optical path, special care was taken in the process of aligning the laser sheet with the plane and location of the intended field of view. Two LaVision Imager Pro X 11M cameras with pix CCD chips and Tamron SP AF 18mm F3.5 macro lenses were used to record the PIV images. The cameras were mounted on a stall built around the wind tunnel test section, decoupled from the wind tunnel vibrations. Oil droplets of a temperature resistant lubricant oil (Plantfluid) were used to seed the outer flow, yielding a mean droplet size just below 2nm. The seeding droplets were introduced into the wind-tunnel storage tube prior to each wind-tunnel run. Note that the jet flow was not seeded in this study. The PIV system was discussed in detail by Casper et al. [13].

6 98 A.-M. Schreyer, S. Stephan & R. Radespiel (a) (b) FIGURE 4. (a) Locations and sizes of the FoV in the two PIV measurement campaigns in respect to the generic launcher model. (b) Definition of the local coordinate system Investigated test cases A preliminary measurement campaign (campaign I) was performed, where the wake flow behind the launcher model was only studied without propulsive jet. In the main PIV campaign (campaign II), both the basic wake flow and the influence of a propulsive jet on the structure of the flow field were investigated. The area of interest for all cases is the region along the nozzle outer surface downstream of the launcher model base, covering the development region of the shear layer along with the separation and reattachment of the flow on the afterbody. For the different measurement campaigns, the fields of view differ in size, as visualized in Figure 4. For the preliminary measurement campaign, the field of view (FoV) had a size of mm D.9D and a spatial resolution of δx = 27.5pix/mm. The temporal delay between the images of an image pair is τ =.48 µs, leading to an average particle displacement of x max = 8 pix in the freestream and a discretisation velocity, i.e. the velocity corresponding to the displacement of one pixel, of u d = 76m/s. These properties were chosen to obtain a large field of view to gain an overview of the mean flow topology, as well as to demonstrate the suitability of the PIV system for measurements in this complex separation-dominated flow field. The resolution is not sufficient for investigations of the turbulent quantities and structures. Assuming that the turbulence intensity is approximately u rms = 2 to 3 % of U (based on a preliminary numerical study), a discretization velocity of u d = 76 m/s would mean a particle displacement of only.2pix. This cannot be spatially resolved; severe peak-locking effects will occur, effectively locking the turbulent intensities to zero. In order to keep errors due to peak locking small, a much larger average particle displacement has to be aimed for. The conditions for the main measurement campaign were determined by applying the model of Angele & Muhammad-Klingmann [14], which states that the discretisation velocity u d should be u d 2 u rms in order to reduce the peak-locking error in the corresponding velocity rms-fluctuations u rms to approximately 1%. The resulting conditions are summarized in Table 3, along with the conditions of the first campaign. The required conditions can be achieved by increasing the temporal delay between the two frames, as well as substantially increasing the spatial resolution, and thus reducing the size of the field of view. Since we still wanted to be able to observe as much of the near wake-flow field as possible, a second camera of the same type was added,

7 Flow structure and unsteadiness in the supersonic wake 99 δx, pix/mm x max, pix τ, µs u d, m/s FoV, mm 2 ÑÔ Ò Á D.9D ÑÔ Ò ÁÁ D.5D TABLE 3. PIV conditions for the two measurement campaigns: calibration δx, maximum streamwise particle displacement x max, temporal delay between two images τ, discretisation velocity u d, and resulting size of the field of view. FIGURE 5. Schlieren-optic visualizations (averaged images) of the afterbody flow. Top image: baseline case, bottom image: with propulsive jet. so that the area of interest was split up into two fields of view (FoVs II a and II b, see Figure 4). Care was taken to obtain the same conditions in both FoV. 3. Results and discussion 3.1. Mean-flow topology To get an overview of the organization of the mean flow field, Schlieren optic visualizations for the cases with and without propulsive jet are presented in the bottom and top images in Figure 5, respectively. The presented images were averaged from 4 instantaneous images and previously shown by Stephan et al. [8]. As expected, the turbulent boundary layer on the main body separates at the shoulder of the main body due to the step decrease in diameter in both cases. The expansion wave developing from the shoulder (origin of the x-y-coordinate system marked in Figure 5 (top)) can be observed. In the baseline case without jet flow, the recompression shock, which indicates the beginning of the reattachment process of the flow on the outer surface of the afterbody [1] is clearly visible. This shock evolves, since the shear layer is bent towards the afterbody and eventually impinges on the surface, which causes another deflection of the flow away from the surface. Apart from the directly visible effects of the afterexpanding jet plume the barrel shock and the jet expansion fan the most prominent influence of the propulsive jet on the flow organization can be observed from the behavior of the reattachment shock. In the jet case, it is located slightly further away

8 1 A.-M. Schreyer, S. Stephan & R. Radespiel from the nozzle surface and also further downstream compared with the baseline case, visualizing the displacement effect on the outer flow (compare the two images in Figure 5). Furthermore, the shock only develops fully quite far away from the surface, indicating that the flow does not reattach in this case, although the shear layer is still deflected towards the surface and a gradual but incomplete deflection away from the surface follows further downstream. To take a closer look on the mean-flow topologies, the mean velocity fields U in the axial x-direction (normalized with the mean incoming flow velocity U ) are shown in Figures 6 (a) and (b) for the baseline and jet cases, respectively. Streamtraces are superimposed (in white) onto the velocity contours. For x/d <, the locations on the launcher main body, a classical turbulent boundary layer can be observed for both cases. From the normalized mean velocity fieldsv/u in the radial y-direction shown in Figure 7, it can be verified that the wall-normal component is indeed zero as expected. Starting from the shoulder of the launcher main-body (x/d = ), a shear layer starts to develop and a large separation zone is forming in the corner. Under the influence of the propulsive jet, this separated zone becomes more prominent and increases in size (compare Figures 6 (a) and (b)). From the streamtraces it is clearly visible that the flow is deflected towards the launcher afterbody. This has been previously observed by Statnikov et al. [2] in their numerical study of a a hypersonic launcher wake flow. In the baseline case, the flow reattaches to the outer surface of the afterbody around the streamwise location x/d.77, which cannot be observed under the influence of the propulsive jet. The displacement effect of the jet plume can be seen in the U/U -contours, but even more so in the V/U -contours shown in Figure 7. Also, the resulting deflection towards the wall is weaker in the wake under the influence of the jet plume (compare the sizes and magnitudes of the zones of negative radial velocity in Figures 7 (a) and (b), respectively). The direct influence of the jet plume can also be seen in the V/U -contours for locations x/d > 1.1 in form of positive velocities in the radial direction (Figure 7) Turbulence behavior in the flow field The turbulence behavior, especially the development of the shear layer that is starting to form at the main-body shoulder, will be discussed in the following paragraphs based on the normalized turbulent fluctuations of the velocity components in the axial and radial directions, as well as the Reynolds shear stresses. The contour plots of the normalized turbulent fluctuation component v rms /U in the radial direction presented in Figure 8 give a good overview on the location of the shear layer in both cases of the studied wake flow field. The deflection towards the wall is clearly visible both in the baseline case and the jet case (Figures 8 (a) and (b), respectively). Up to a streamwise distance of approximately x/d =.6 from the shoulder (x/d =.7 in the case with propulsive jet), the shear layer is distinctly recognizable from the local maximum in turbulence intensity. In the baseline wake, a second weaker maximum further away from the surface starts to develop around location x/d =.6 and indicates the formation and location of the reattachment shock (see Figure 8 (a)). In general, the extent of the shear layer in radial direction (i.e. the width of the zone of maximum turbulence intensity) is larger for the case without propulsive jet. Another prominent feature is the difference in turbulent intensity between the two

9 Flow structure and unsteadiness in the supersonic wake U/U x/d (a) U/U x/d (b) F IGURE 6. Contours of the normalized mean velocity component U/U in the main flow direction from PIV measurements. (a) baseline case, (b) with propulsive jet. investigated cases. The overall intensity level is much lower in the flow field influenced by the propulsive jet, and the maximum values are lower as in the baseline case as well. It appears that the propulsive jet stabilizes the wake flow field. In order to look into that in more detail, fluctuation profiles were extracted for several locations between.5 x/d 1.2 along the afterbody. For the radial component, this is shown in Figure 9, and for the axial component in Figure 1. Both are normalized with U. In both fluctuating velocity components, the respective maximum values are higher for the baseline case. For vrms /U, a turbulent intensity of 15% is achieved in the baseline case, compared to 11% in the case influenced by the jet plume. For urms /U, the effect is smaller, but can still be observed (see Figures 9 and 1, respectively). In the wake that is uninfluenced by a propulsive jet (see Figure 9 (a)), the maximum of the radial component of the turbulent velocity fluctuations increases continuously until a distance from the base of x/d =.4. This is expected for shear-layer instability. At the same time, the shear layer moves from the shoulder towards the outer surface of the nozzle fairing, interacts with the wall and the boundary layer, and finally gets reflected. The maximum representing the shear layer starts to decrease and then vanishes. From a streamwise distance of x/d =.8 onward, the distribution of the local intensity maximum close to the wall (.25) approaches that of a disturbed turbulent boundary layer that is starting to relax. Equilibrium conditions cannot be reached until the nozzle exit due to the small length of the afterbody. Further away from the wall, a second intensity maximum develops from x/d.6 and shifts away from the wall with increasing streamwise distance. This represents the reattachment shock.

10 12 A.-M. Schreyer, S. Stephan & R. Radespiel V/U x/d (a) V/U x/d (b) FIGURE 7. Contours of the normalized mean velocity componentv/u in the radial direction from PIV measurements. (a) baseline case, (b) with propulsive jet. Under the influence of the jet plume (see Figure 9 (b)), the increase of turbulent intensity in the developing shear layer seems damped (no continuous increase can be observed). The maximum corresponding to the shear layer stays recognizable for a longer streamwise distance (x/d =.8), and no outer maximum corresponding to a reattachment shock develops properly. This again confirms that reattachment on the afterbody surface does not occur in this case. Also the displacement effect on the outer flow can be observed clearly when comparing Figures 9 (a) and (b): the location of the shear layer (indicated by the local maximum in the turbulence intensity profile) is further away from the surface under the influence of the jet plume for corresponding locations in the streamwise direction. The same qualitative behavior can be observed in the axial velocity fluctuations profiles shown in Figures 1 (a) and (b), and even more clearly in the normalized profiles of the Reynolds shear stresses compared in Figure 11. Deprés et al. [1] made a similar observation in transonic wake flow. Deprés et al. [1] performed pressure measurements on a number of different afterbody configurations at M =.85, and found that a supersonic propulsive jet interacts with the recirculation region on the nozzle fairing and modifies the flow topology. They observed that the jet stabilizes the near-wake region, leading to a more axisymmetric external flow field [1]. Their explanation for this effect was, that since the propulsive jet is located in the center of the wake, it obstructs the development of large-scale vortical structures. Some information on the dynamic wake flow behavior can be gained from the power spectral densities (PSD) of the wall-pressure fluctuations measured at the base of the launchermodel main body. For the cases without and with propulsive jet, the PSD (in Pa 2 ) is

11 Flow structure and unsteadiness in the supersonic wake v rms /U x/d (a) v rms /U x/d (b) FIGURE 8. Contours of the normalized turbulent fluctuation componentv rms/u in the radial direction from PIV measurements. (a) baseline case, (b) with propulsive jet x/d=.5 x/d=.2 x/d=.4 x/d=.6 x/d=.8 x/d=1. x/d= x/d=.5 x/d=.2 x/d=.4 x/d=.6 x/d=.8 x/d=1. x/d= v rms /U v rms /U (a) (b) FIGURE 9. Profiles of the turbulent fluctuations v rms of the radial velocity component for several locations along the launcher afterbody. v rms is normalized with the mean incoming flow velocity U. plotted versus the Strouhal number St D based on main-body diameter D, i.e. St D = f D/U in Figure 12 (a) and (b), respectively. As mentioned previously, the pressure sensors were located at a radius of r/d =.42 at three different angular positions (Θ = 18, 19, and 24 ). The PSD are shown both from respective single readings (black, red, and green lines) and averaged from 38 readings (black lines) for each sensor. The main information from these pressure spectra can be gained by observing the behavior

12 14 A.-M. Schreyer, S. Stephan & R. Radespiel.2.2 x/d=.5 x/d=.2 x/d=.4 x/d=.6 x/d=.8 x/d=1. x/d= x/d=.5 x/d=.2 x/d=.4 x/d=.6 x/d=.8 x/d=1. x/d= urms/u urms/u (a) (b) F IGURE 1. Profiles of the turbulent fluctuations urms of the axial velocity component for several locations along the launcher afterbody. urms is normalized with the mean incoming flow velocity U. F IGURE 11. Profiles of the Reynolds shear stresses u v normalized with the mean incoming 2 flow velocity U for several locations along the launcher afterbody. of the intensity peak for StD =.25 that can be observed in the baseline case. This peak has previously observed by numerous researchers in different configurations [1, 2, 6, 15, 16], and is generally attributed to the vortex shedding from the separation bubble. It is governed by the interaction between those turbulent structures and the wall in the reattachment zone [1]. The fact that this peak cannot be observed in the propulsivejet case (see Figure 12 (b) ) does therefore again back up by the finding that the flow does not reattach in this case. Another interesting point is that the peak disappears in azimuthal direction, which is possibly related with helical vortex structures, as Deprés et al. [1] and Deck & Thorigny [3] have discussed. These coherent anti-symmetrical fluctuations are relevant for the buffet problem. 4. Conclusions The flow field in the wake of a generic axisymmetric space-launcher model was investigated, and the influence of an afterexpanding propulsive jet was described in de-

13 Flow structure and unsteadiness in the supersonic wake 15 (a) (b) FIGURE 12. Wall pressure spectra at the launcher base from single measurements (colored lines) and averaged from 38 readings (black lines). (a) baseline case, (b) jet case. tail. Measurements were performed in the Hypersonic Ludwie Tube Braunschweig at a Mach number of 2.9 and a Reynolds number Re D = based on model diameter D. The nozzle exit velocity of the cold air jet was at Mach 2.5. Velocity measurements in the wake flow by means of Particle Image Velocimetry and mean and unsteady wall-pressure measurements on the main-body base are performed simultaneously. This way, the evolution of the wake flow was observed along with its spectral content. The mean flow topology and turbulence behavior of the wake was described based on mean and turbulent velocities from Particle Image Velocimetry measurements. The wall-pressure fluctuations measured on the main-body base were used to gain spectral information on the footprint of the wake on the launcher base. A large separated zone forms downstream of the step decrease in diameter at the interface between launcher main body and nozzle fairing. In the baseline case, the flow is reattaching downstream on the nozzle fairing. Under the influence of the jet plume, the separated region increased in size and reattachment does not occur anymore. The flow is displaced away from the wall. The propulsive jet appears to have a stabilizing effect on the wake flow. The development of the shear layer and the magnitude of the turbulent intensities are damped, which we showed with the axial and radial components of the turbulent velocity fluctuations, as well the Reynolds shear stress evaluation along the launcher afterbody. This effect is probably caused by the fact that the jet is located in the center of the wake and obstructs the growth of larger vortical structures in the shear layer. Due to the great need for means to stabilize and reduce the reattachment length to reduce dynamic loads on the launcher, we currently investigate the influence of a convoluted trailing edge (lobes) on the wake of a generic space-launcher model in supersonic flow (see Figure 13), and analyze the techniques potential to control the separation length and dynamics. Such passive vortex generators are of particular interest [18], both due to promising results on two-dimensional backward-facing step geometries [17, 19], and their simplicity and robustness [18]. Kähler et al. [17, 19] investigated different lobe geometries on a 2D backward-facing step in transonic flow. The lobes introduce counterrotating streamwise vortices that increase turbulent mixing and create a highly threedimensional flow field with a spanwise component [17]. Behind lobe peaks, the separation is locally rather strong, but the overall separation length is significantly reduced, and the dynamic motion of the reattachment location is stabilized [17, 19]. The induced

14 16 A.-M. Schreyer, S. Stephan & R. Radespiel FIGURE 13. Left: Sketch of the axisymmetric space-launcher model with convoluted trailing edge and strut support (dimensions in mm). Right: Rotatable lobe ring with measurement locations L1 and L2. vortices should be of similar size as coherent structures in the wake [17], and larger lobes have a stronger effect. In our current study, we investigate the effects of such lobes on the wake topology of an axisymmetric model and analyze the interaction with the same afterexpanding cold air- jet plume that was described above. Experimental investigations are performed at a free-stream Mach number of M = 2.9 and a Reynolds number based on the model diameter of Re D = Square lobes with a height of.2 step heights h and a deflection angle of 18 were installed at the main-body base (see Figure 13). Kulite XCQ-93 pressure sensors with a range of.35 bar absolute are flush-mounted into the base at a radial position of r/d =.42 and at different angles (θ = 18, 19, and 24 ), and further Kulite LE-62 sensors are installed on the nozzle fairing at x/d =.77 (reattachment location in baseline case) and x/d =.31, the presumed reattachment location for the controlled case [17, 19]. The dynamic behavior and spectral content of the wake is analyzed based on these unsteady wall-pressure signals. In addition, we carry out schlieren flow visualizations to study the effect of the lobes on the mean flow topology, in particular the reattachment length. The wake with and without jet plume is investigated downstream of the lobe peak and valley positions (see Figure 13 right). Acknowledgments Financial support has been provided by the German Research Foundation (Deutsche Forschungsgemeinschaft DFG) in the framework of the Sonderforschungsbereich Transregio 4. The authors also acknowledge Sven Puelm for his help during the PIV measurement campaign. References [1] Deprés, D., Reijasse, P., and Dussauge, J.-P. (24). Analysis of Unsteadiness in Afterbody Transonic Flows. AIAA Journal, 42(12), [2] Statnikov, V., Sayadi, T., Meinke, M., Schmid, P., and Schröder, W. (215). Analysis of pressure perturbation sources on a generic space launcher after-body in supersonic flow using zonal RANS/LES and dynamic mode decomposition. Phys. Fluids 27(1), 1613, DOI: 1.163/ [3] Deck, S. and Thorigny, P.(27). Unsteadiness of an axisymmetric separatingreattaching flow: Numerical investigation. Phys. Fluids 19(6), 6513, DOI: 1.163/

15 Flow structure and unsteadiness in the supersonic wake 17 [4] Weiss, P.-É., Deck, S., Robinet, J.-C., and Sagaut, P. (29). On the dynamics of axisymmetric turbulent separating/reattaching flows. Phys. Fluids 21(7), 7513, DOI: 1.163/ [5] Bannink, W. J., Houtman, E. M., and Bakker, P. G. (1998). Base flow / underexpanded exhaust plume interaction in a supersonic external flow. 8th AIAA International Space Planes and Hypersonic Systems and Technologies Conference, AIAA , DOI: / [6] Bourdon, C. J. and Dutton, J. C. (1999). Planar visualizations of large-scale turbulent structures in axisymmetric supersonic separated flows. Phys. Fluids 11, 21, DOI: 1.163/ [7] Janssen, J. R. and Dutton, J. C. (24). Time-Series Analysis of Supersonic Base-Pressure Fluctuations. AIAA Journal, 42(3), [8] Stephan, S., Wu, J., and Radespiel, R. (215). Propulsive jet influence on generic launcher base flow. CEAS Space Journal, 7(4), [9] Saile, D., Guelhan, A., Henckels, A., Glatzer, C., Statnikov, V., and Meinke, M. (213). Investigations on the Turbulent Wake of a Generic Space Launcher Geometry in the Hypersonic Flow Regime. EUCASS Progress in Flight Physics, 5, [1] Saile, D. and Guelhan, A. (214). Plume-Induced Effects on the Near-Wake Region of a Generic Space Launcher Geometry. 32nd AIAA Applied Aerodynamics Conference, AIAA Aviation, American Institute of Aeronautics and Astronautics, AIAA , DOI: / [11] Wu, J., Zamre, P., and Radespiel, R. (215). Disturbance characterization and flow quality improvement in a tandem nozzle Mach 3 wind tunnel. Experiments in Fluids, 56(1), 2, DOI: 1.17/s [12] Statnikov, V., Stephan, S., Pausch, K., Meinke, M., Radespiel, R., and Schröder, W. (216). Experimental and numerical investigations of the turbulent wake flow of a generic space launcher at M=3 and M=6. CEAS Space Journal, 8, (2), , DOI: 1.17/s x. [13] Casper, M., Stephan, S., Scholz, P., and Radespiel, R. (214). Qualification of oil-based tracer particles for heated Ludwieg tubes. Experiments in Fluids, 55(6), 1753, DOI: 1.17/s [14] Angele, K. P. and Muhammad-Klingmann, B. (25). A simple model for the effect of peak-locking on the accuracy of boundary layer turbulence statistics in digital PIV. Experiments in Fluids, 38(3), [15] Piponniau, S., Dussauge, J.-P., Debiève, J.-F., and Dupont, P. (29). A simple model for low-frequency unsteadiness in shock-induced separation. Journal of Fluid Mechanics, 629, 87 18, DOI:1.117/S [16] Smits, A. J., Hayakawa, K., and Muck, K.-C. (1983). Constant Temperature Hotwire Anemometer Practice in Supersonic Flows: Part 1: The Normal Wire. Experiments in Fluids, 1, [17] Bolgar, I., Scharnowski, S., Kähler, C. J. (215). Control of the Reattachment Length of a Transonic 2D Backward-Facing Step Flow. Int. conf. on jets, wakes and separated flows, Stockholm, Sweden, [18] Pearcey, H. H. (1961). Shock-induced separation and its prevention by design and boundary layer control. In: G. V. Lachmann (Hg.): Boundary Layer and Flow Control: Its Principles and Application, Part IV:, Pergamon Press, Oxford, [19] Scharnowski, S., Bolgar, I., Kähler, C. J. (215). Control of the recirculation re-

16 18 A.-M. Schreyer, S. Stephan & R. Radespiel gion of a transonic backward-facing step flow using circular lobes. 9th International Symposium on Turbulence and Shear Flow Phenomena (TSFP 9), Melbourne, Australia.