Experimental investigation of flow control devices for the reduction of transonic buffeting on rocket afterbodies F.F.J. Schrijer 1, A. Sciacchitano 1, F. Scarano 1 1: Faculty of Aerospace Engineering, Delft University of Technology, Delft, The Netherlands, f.f.j.schrijer@tudelft.nl Abstract In the optimization of launch vehicles, the phenomenon of transonic buffeting plays an important role since it generates high structural loads on the launcher propulsion system (nozzle). The cause of transonic buffeting is the reattachment of the shear layer on the nozzle. This investigation is aimed at studying the flow field around a rocket after-body in transonic conditions with emphasis on the buffeting phenomenon. Furthermore, the flow around chevron type flow control devices has been investigated. In the experiments a blow down wind tunnel was used that generates a Mach 0.7 transonic flow. The wind tunnel model features a cone-cylinder fore-body. At the base a cylindrical after-body is placed to simulate the presence of the nozzle and the exhaust flow. Velocity measurements have been performed by means of planar PIV and statistical and time resolved data are acquired at 500 Hz and 10 khz respectively. From the statistical analysis it is found that the addition of the chevron devices causes a downstream shift of the reattachment location. Furthermore it was found that the maximum backflow velocity in the separated region is increased when compared to a clean configuration. The influence of the chevron devices on the rms velocity fluctuations is minimal. From the time resolved measurements it is observed that for the clean configuration, the region where back flow occurs grows and shrinks in an oscillatory manner. When the chevron devices are added it is found that in one case, where the chevrons are relatively large, the behavior remains and the back-flow magnitude is increased. For the smaller chevrons it is observed that the oscillating behavior is reduced. From Fourier analysis a characteristic frequency around 450 Hz is found for the freshly separated shear layer. When moving further downstream it is found that this peak is reduced. 1. Introduction In the aerodynamic design of launch systems, flow separation plays an important role. It not only increases the drag of the vehicle but it also causes unsteady loads on structural components located in separated flow regions. An important example is the flow separating from the base, upon separation a shear layer is formed which tends to oscillate and generates low frequency wallpressure fluctuations in the reattachment region. When these flow-generated frequencies approach the frequencies of the structural modes, buffeting occurs and the structural integrity is no longer guaranteed. Generally, the propulsion system (nozzle) is located in this region, which is already subject to high pressure and thermal loads and therefore very sensitive to any additional unsteady loading. Therefore, to further optimize future launch systems, knowledge is needed on how these unsteady loads are being generated and, at the same time, how to reduce them by applying flow control devices. - 1 -
Fig. 1: Schematic of the transonic buffeting problem (photo Ariane V: ESA) Several investigations have already been performed to investigate transonic buffeting. In the experiments of Deprés et al (2004), it was found that the ratio between the nozzle length (L) and the body diameter (D) has an important effect on the unsteady flow features on the base. For L/D = 0.6 it was found that the flow field is governed by the large-scale vortices in the wake oscilating with a typical Strouhal number of St D 0.2. Furthermore it was observed that for these small L/D shear layer reattachment occurs on the jet. For L/D = 1.2 the shear layer was found to reattach on the afterbody near the end of the nozzle. For this configuration a dominant frequency was found for St D 0.6. This frequency was related to the convection of turbulent eddies in the separated shear layer. Furthermore a small peak still existed for St D 0.2. In the work of Weiss at al (2009) CFD computations are presented using a zonal detached eddy simulation (ZDES) which agree well with the measurements presented by Deprés, however the peak at St D 0.6 was not observed. In general there are two approaches that can be follow in order to reduce the transonic buffeting problem. The most simple is by adding a skirt to the base of the vehicle so that the separation location is moved further downstream so that it no longer impinges on the nozzle surface. However, this has the result that the shear layer impinges on the hot exhaust flow coming from the nozzle. This causes the hot gasses to be ingested into the separated region and transported back to the base of the rocket, which may lead to structural problems related to the thermal loading of the base itself. An alternative approach is by adding flow control devices that modify the frequency content of the shear layer, effectively reducing the energy contained within the frequencies that resonate with the structural modes. In the current investigation, chevrons are used that generate stream-wise vortices which interact with the large span-wise vortices that are already present in the shear layer, causing the large span-wise vortices to be broken down into small-scale vortices, see Figure 2. - 2 -
Fig. 2: Chevron configurations on a supersonic splitter plate (iso-density surfaces), Dietiker and Hoffman (2007) Meliga and Reijasse (2007) performed experiments using the model of Deprés with a chevron configuration added to the base and no significant effect on the wall pressure frequencies was measured. However in these experiments only a single chevron configuration was investigated. In the current investigation, experiments have been performed on a generic rocket-like wind tunnel model using particle image velocimetry. Different chevron geometries are tested and the effect on the flow field is investigated in terms of flow statistics and flow field frequency content. 2. Experimental apparatus 2.1 Flow facility The flow facility used in the experiments is the TST27 blow down wind tunnel of Delft University of Technology having a test section of 27 28 cm 2, see Figure 3. By using a contoured flexible nozzle which allows to continuous variation of the throat section, the facility can be operated in a Mach range spanning from 0.5 to 4.2, with a total pressure range from a minimum of P t = 1.5 bar at Mach 0.5 to a maximum of 20 bar at Mach 4.2. To achieve subsonic/transonic flow conditions a choke-system is used downstream of the test section, which allows the Mach number to be set with an accuracy of 0.01. In the current experiments, the free stream Mach number is M = 0.7 with a total pressure of P t = 1.9 bar. The total temperature is equal to T t = 285 K, resulting in a Reynolds number of Re D = 1.3 10 6 where D is the model diameter. Fig. 3: Schematic of the TST27 blow down wind tunnel - 3 -
2.2 Wind tunnel model The experiments are performed on the FESTIP (Future European Space Technology Investigation Program) wind tunnel model, see Figure 4. In previous investigations it has been used for supersonic flow measurements including a cold flow exhaust jet (van Oudheusden and Scarano, 2008). The model is supported by a sting in the bottom, which also accommodates the supply air for the exhaust jet. Fig. 4: The FESTIP wind tunnel model The total length of the main body is 187 mm and the diameter at the base is 50 mm. A 0.15 mm diameter tripping wire is placed at fore-body (40% of the cone length) to ensure a turbulent boundary layer at separation. The current investigation has been performed without exhaust jet; instead a 127 mm long and 16.7 mm diameter cylinder after-body has been fitted in order to simulate the nozzle and jet. The investigation focuses on chevron type flow control devices. Different configurations were tested; see Figure 5 for the exact dimensions. Fig. 5: Chevron dimensions [mm] - 4 -
2.3 Illumination, recording and seeding A Quantronix Darwin Duo Nd-YLF double pulse laser provides the illumination for the PIV experiments. The laser is inserted into the test section by means of a periscope probe that also holds the beam forming optics. In the experiments the laser sheet thickness was set to 1.5 mm. The images are recorded by a Photron FastCAM SA1.1 CMOS camera. In order to obtain flow statistics, 1000 PIV image pairs were recorded at 500 Hz with a 35 mm Nikkor objective at f # = 2 and an active sensor size of 1024 400 pixels. For the time resolved measurements, image pairs were recorded at 10 khz using a 105 mm Nikkor objective at f # = 2.8. In this case the active sensor size was 1024 288 pixels. The field of view is 93 36 mm 2 for the 500 Hz measurements, see Figure 6, and 106 30 mm 2 for the measurements performed at 10 khz with a similar location of the field of view. In all cases the images are processed having a final window size of 21 21 pixels with a 75% overlap. Fig. 6: Field of view Fig. 7: Laser sheet location For each chevron configuration, two measurements were performed, each with a different orientation of the laser light sheet with respect to the chevrons in order to check for azimuthal variations in the velocity field. The chevron modules were fitted to the wind tunnel model so that they could be rotated. Measurements were performed with a plane that was aligned with a chevron peak (laser on peak, Figure 7-left) and with a chevron valley (laser on valley, Figure 7-right). The seeding material for the experiments performed at 500 Hz was DEHS. For the measurements performed at 10 khz, TiO 2 was used to increase the amount of light that is scattered by the particles. For specific information regarding the seeding particles see Ragni et al (2010). 3. Results 3.1 Flow statistics As a reference situation, the clean configuration was taken without any flow control devices. In Figure 8, the mean velocity field over this configuration is given. The separated shear layer is clearly resolved as well as the recirculation region in the back. The mean shear layer reattaches at x = 54 mm, where x = 0 mm is the location of the base of the model. The separated length with respect to the base diameter now becomes L sep /D = 1.1, which similar to the values given by Deprés and Reijasse (2004) and Weiss et al. (2009). In the separated region maximum reverse flow velocities up to -59 m/s are measured. - 5 -
Fig. 8: Mean velocity field for the clean configuration Rms velocity fluctuations are shown in Figure 9. The fluctuations in the horizontal u-direction were found to be stronger that the fluctuations in v-direction. The maximum fluctuations in u-direction are measured just upstream of the reattachment location and are u = 54 m/s (u /U = 0.24). The maximum fluctuations in v-direction are found downstream of reattachment and are v = 43 m/s (v /U = 0.19). Fig. 9: Rms velocity fluctuations for the clean configuration, u (left) and v-component (right) For the chevron devices a and f, the flow field was found to vary in azimuthal direction while for all other configurations the average flow field does not depend on peak or valley location. For all chevron devices, the reattachment point is shifted further downstream compared to the clean configuration due to the presence of the element. For configurations a and f the reattachment point for the peak location is observed further downstream compared to the valley location while for the other devices there is no distinction between both cases. In Figure 10 the results are shown for the chevron device g. The mean velocity field is given for the u-component as well as its rms fluctuations. In the region where the measurement region is obscured by the chevrons, the data is blanked. The results for this configuration are typical for all configurations that produce an axisymmetric flow field. The reattachment point for chevron g is located at x = 56 mm and the maximum back-flow velocity in the separated region is -68 m/s. This is an increase of 15% with respect to the clean configuration. In general it was found that the maximum back-flow velocity increases with the chevron length up to a maxim back-flow velocity of -72 m/s for configuration d. The maximum velocity fluctuations are u = 57 m/s (u /U = 0.27) and v = 47 m/s (v /U = 0.23), which are slightly higher than in the clean case. The largest - 6 -
maximum u and v are found for configuration e, which are u = 66 m/s and v = 48 m/s. Fig. 10: Chevron g (laser on peak); horizontal mean velocity (left) and rms values (right) Figures 11 and 12 show the results for the chevron device a. As already mentioned before, the addition of the chevrons to the base causes the reattachment point to move downstream. When the measurement plane is aligned with the chevron peak (Figure 11) the reattachment point is located at 62 mm, while for the valley case (Figure 12) the reattachment point is more upstream at 58 mm. In both cases the maximum back-flow observed in the separated region is -73 m/s, which is an increase of 24% compared to the clean configuration. The maximum velocity fluctuations are comparable to configuration g and are 57 m/s (u /U = 0.28) for the u-component and 47 m/s (v /U = 0.21) for the v-component. Fig. 11: Chevron a (laser on peak); horizontal mean velocity (left) and rms values (right) Fig. 12: Chevron a (laser on valley); horizontal mean velocity (left) and rms values (right) 3.2 Time resolved results In the time resolved measurements, configurations a and g are studied in more detail, as well as the clean configuration as a reference. In Figure 13 a sequence of snapshots (subsampled with a factor 5) is given that shows the typical dynamics of the flow for the clean configuration. In the images the region that contains backflow is contoured with a black outline. At t = t 0 the region containing backflow is relatively small, however as time progresses the backflow region grows in size. Then, around t = t 0 + 2.5 ms the region has its - 7 -
maximum size, after which its size decreases again. t = t 0 t = t 0 + 3.0 ms t = t 0 + 0.5 ms t = t 0 + 3.5 ms t = t 0 + 1.0 ms t = t 0 + 4.0 ms t = t 0 + 1.5 ms t = t 0 + 4.5 ms t = t 0 + 2.0 ms t = t 0 + 5.0 ms t = t 0 + 2.5 ms t = t 0 + 5.5 ms Fig. 13: Sequence of snapshots showing the flow dynamics on the clean configuration; the black contour indicates the back-flow region For configuration a (laser on valley) a similar behavior is observed, see Figure 14. Only the magnitude of the back flow velocity has increase, which was also already observed from the flow statistics. - 8 -
t = t 0 t = t 0 + 3.0 ms t = t 0 + 0.5 ms t = t 0 + 3.5 ms t = t 0 + 1.0 ms t = t 0 + 4.0 ms t = t 0 + 1.5 ms t = t 0 + 4.5 ms t = t 0 + 2.0 ms t = t 0 + 5.0 ms t = t 0 + 2.5 ms t = t 0 + 5.5 ms Fig. 14: Sequence of snapshots showing the flow dynamics on the configuration a (laser on valley); the black contour indicates the back-flow region For configuration g (laser on valley), see Figure 15, the situation is somewhat different. In this case the growing and shrinking of the back flow regions is not that evident. It seems that this chevron configuration prevents the oscillating behavior of the back flow region. - 9 -
t = t 0 t = t 0 + 3.0 ms t = t 0 + 0.5 ms t = t 0 + 3.5 ms t = t 0 + 1.0 ms t = t 0 + 4.0 ms t = t 0 + 1.5 ms t = t 0 + 4.5 ms t = t 0 + 2.0 ms t = t 0 + 5.0 ms t = t 0 + 2.5 ms t = t 0 + 5.5 ms Fig. 15: Sequence of snapshots showing the flow dynamics on the configuration g (laser on valley); the black contour indicates the back-flow region Furthermore a fourier analysis was performed in order to obtain the dominant frequencies in the flow field. In the analysis three different regions are investigated, first a region containing the freshly separated shear layer, a region halfway between separation and reattachment and a region near reattachment, see Figure 16. Each region contains 21 11 points, and the obtained spectra are averaged. The power spectra as depicted in Figures 17 to 19 show the spectral power normalized with respect to the total energy present in the signal. - 10 -
Fig. 16: Definition of the regions for the power spectrum analysis For the clean configuration (Figure 17-left) a peak in the spectrum is found at 412 Hz, which corresponds to a Strouhal number of St D = 0.1. Moving further downstream (Figure 17 middle and right), it can be seen that the peak at 412 Hz is decreased although it is still there. Fig. 17: Power spectra for the clean configuration, region a (left), b (middle) and c (right) For configuration a, near separation (Figure 18-left) also a peak is found at 412 Hz. However for a more downstream location (Figure 18-middle), the peak disappears and a maximum is found around 235 Hz although not as pronounced. The same is observed when moving even further downstream, in this case a maximum is found at 137 Hz. Fig. 18: Power spectra for configuration a laser on valley, region a (left), b (middle) and c (right) For configuration g near separation a maximum is found at 490 Hz (see Figure 19-left), which is slightly higher compared the clean case and configuration a. Moving downstream to location b a peak is found at a higher frequency (784 Hz) and in location c two maxima are found for 510 Hz and 1 khz. - 11 -
Fig. 19: Power spectra for configuration g laser on valley, region a (left), b (middle) and c (right) 4. Conclusions Chevron-type flow control devices have been studied for the reduction of buffeting in the transonic flow regime. It has been found that although the chevron devices are of three-dimensional nature, the majority of the chevron configurations result in a statistically axisymmetric flow field. The addition of the devices to a clean base increases the rms fluctuations slightly and larger backflow velocities are measured. From the time resolved measurements it is observed that the clean configuration features a flow where the area in which back flow occurs exhibits an oscillating behavior where it grows and shrinks. When the type a chevrons (larger chevrons) are added to the base, the oscillating behavior remains and the magnitude of back flow increases. For configuration g (smaller chevrons) it is observed that the oscillating behavior is reduced. The characteristic frequencies in the flow field near the shear layer are examined using Fourier analysis. For the clean configuration and configuration a a peak in the power spectrum is found at 412 Hz which corresponds to a Strouhal number of St D = 0.1. Moving further downstream, this peak becomes less pronounced. For configuration g a peak is found at a slightly higher frequency 490 Hz, which also decreases when moving further downstream. References D. Deprés, P. Reijasse and J. Dussauge (2004) Analysis of unsteadiness in afterbofy transonic flows, AIAA journal, Vol. 42, No. 12 J. Dietiker and K. Hoffman (2007) Numerical investigation of turbulent shear layers in jet exhaust flows, 37th AIAA Fluid Dynamics conference and Exhibit, Miami, FL, USA, Paper number AIAA 2007-3856 P. Meliga and P. Reijasse (2007) Reduction of unsteadiness in the flow past a three dimensional afterbody, 2nd European conference for aerospace sciences B. van Oudheusden and F. Scarano (2008) PIV investigation of supersonic base-flow-plume interaction, In: A. Schroeder and C. Willert (Eds), Particle Image Velocimetry, pp 465-474. D. Ragni, F. Schrijer, B. van Oudheusden and F. Scarano (2010) Particle tracer response across shocks measured by PIV, Experiments in Fluids, DOI: 10.1007/s00348-010-0892-2 P. Weiss, S. Deck, J. Robinet and P. Sagaut (2009) On the dynamics of axisymmetric turbulent separating/reattaching flows, Physics of Fluids, Vol. 21-12 -