Experimental Studies of Transitional Boundary Layer Shock Wave Interactions

Transcription

1 44th AIAA Aerospace Sciences Meeting and Exhibit 9-12 January 2006, Reno, Nevada AIAA Experimental Studies of Transitional Boundary Layer Shock Wave Interactions Z.R. Murphree *, J. Jagodzinski, * E.S. Hood, Jr., * N.T. Clemens, + and D.S. Dolling Center for Aeromechanics Research Department of Aerospace Engineering and Engineering Mechanics, University of Texas at Austin Austin, Texas Shock wave boundary layer interactions generated by a cylinder on a flat plate were visualized with a particular emphasis on transitional interactions. Surface-streakline (kerosene-lampblack), schlieren, and low-repetition rate planar laser scattering were used to visualize the flowfield. Furthermore, high-speed (10 khz) planar laser scattering in a streamwise-spanwise plane (plan view) was used to visualize the time-varying structure of the separated flow. The shapes and scales of the transitional interactions are compared to those of laminar and turbulent interactions. It was observed that transitional interactions are highly variable in their structure, but they appear to be a composite of the other two types with laminar scaling along the plate centerline and turbulent scaling in the outboard region. The plan view planar imaging shows that even the most laminar interactions exhibit separated flow regions that are transitional or turbulent. Furthermore, the fluctuations in the size of the separated flow for more laminar-like transitional interactions were significantly larger than for turbulent interactions. Nomenclature d = cylinder diameter X cyl = distance from plate leading edge to cylinder leading edge X sep = distance from plate leading edge to primary separation line along plate centerline X trans = distance from plate leading edge to transition along plate centerline λ sep = distance from cylinder leading edge to primary separation line along plate centerline I. Introduction hock wave/boundary layer interactions (SWBLI), often accompanied by separation, are a ubiquitous Sphenomenon in high-speed flight. Most of the work done in this field over the past 50 years has been in fullydeveloped turbulent flows because most practical applications were at transonic and low supersonic speeds at altitudes where Reynolds numbers are large and turbulent flow the norm. An understanding of these turbulent interactions is very important for vehicle design because the interactions result in very high unsteady thermal and acoustic loads that can result in diminished component performance and material failure. In contrast, transitional shock wave/boundary layer interactions, in which the incoming boundary layer is in a transitional state, or in which transition is induced within the interaction itself, appear to be even more unsteady and could have greater adverse effects, and yet have received little attention. This lack of attention has stemmed from both the lack of critical applications and from the formidable challenges that the study of such flows poses to both experiment and computation. However, high Mach number air-breathing propulsion systems are of increasing interest to the U S Air Force, and the inlets of such systems will have extremely complex shock/boundary layer interactions with significant regions of transitional flow. These transitional interactions will have a powerful influence on the local inlet flow properties and on the uniformity, the quality, and the steadiness of the flow entering the combustor. It is fair to say that our current understanding of transitional interactions is extremely poor. The capabilities of modern * Graduate Research Assistant, Member + Professor, Associate Fellow Professor, Fellow 1 Copyright 2006 by Z.R. Murphree, J. Jagodzinski, E.S. Hood, Jr., N.T. Clemens, and D.S. Dolling. Published by the, Inc., with permission.

2 non-intrusive instrumentation, especially particle image velocimetry (PIV) and planar laser scattering (PLS), now offer an opportunity to initiate studies of transitional interactions and build the knowledge base. II. Experimental Program and Techniques The experimental program was carried out in the high-reynolds-number Mach 5 blow-down wind tunnel located at the Wind Tunnel Laboratories at the Pickle Research Campus of the University of Texas at Austin. The facility uses a bottle field of eight tanks that has a combined storage volume of 140 ft 3 and can be pressurized to a maximum of 2550 psia. The stagnation temperature and pressure were 653º R and 363 psia. Downstream of the 2-D Mach 5 nozzle, the flow enters a 27 in long constant-area test section which is 7 in high and 6 in wide. The interactions for this study were generated by a circular cylinder on a flat plate. The cylinder could be translated streamwise along the plate, and the state of the boundary layer at a given position would dictate the type of interaction (laminar, transitional, or turbulent). Three different plates were used, all with smooth surface finishes and tapered leading and trailing edges. The first, used for kerosene-lampblack surface flow visualization, was an 18 inch long flat brass plate that was bolted to the tunnel walls. The second surface was a 10 inch long brass plate designed to provide optical access to study the interactions. This plate was mounted in the tunnel with a supporting strut, but the strut presented so much blockage to the flow that the cylinder had to be very small in order for the tunnel to start. The third surface was also a 10 inch long plate, but was mounted without a supporting strut in order to reduce the blockage so that larger cylinders (up to a half inch in diameter) could be used to generate larger interactions. The third plate also did away with the threaded holes for attaching the cylinder in order to reduce flow disturbances. In this case the cylinder was held in place with the compressive force of a screw-jack. Experimental techniques employed to date have included: a) surface-streakline visualization using a kerosenelampblack mixture, used to determine quantitatively the mean separation line, b) schlieren imaging, and c) PLS. The schlieren setup consisted of a high-speed camera, neutral density filters, a razor blade for a knife-edge and either a Helium-Neon laser (Spectra-Physics Model controlled by a Spectra-Physics Laser Exciter Model 212-1) or a flash-lamp (EG&G Electro Optics model LS flash-lamp, 2-microsecond duration). The choice depended whether overall flow characteristics or local image quality were of primary concern. Figure 1 shows the arrangement. Images were acquired using a KODAK EKTAPRO HS Motion Analyzer, Model 4540 mx or a Cohu The cameras were fitted with either a Computar MC Zoom Macro 1:3.5~5.3 lens for the laser schlieren or a Nikon ED AF Micro Nikkor 200 mm 1:4D lens paired with the flash-lamp. The EKTAPRO camera processor could record full frames at 30, 60, 125, 250, 500, 750, 1,125, 2,250 and 4,500 frames per second, whereas the Cohu operated at 30 frames per second. PLS images were taken using two different setups enabling plan views (i.e. streamwise-spanwise view) of the flow. The flow was seeded with a finely atomized ethanol fog, which was then illuminated with laser light. The scattering from the condensed fog was then captured by a high-speed 1kx1k CMOS array (Photron FASTCAM-ultima APX) with a 50 mm 1:1.4 lens. In both setups the laser sheet was brought in parallel to the surface of the flat plate through a side window. In the first setup the camera was imaging the sheet through the opposite side window and was therefore positioned at an oblique angle as shown in Fig. 2. Due to the Figure 1. Schematic diagram of flash-lamp schlieren setup. viewing angle, these images exhibit some geometric distortion and blurring at the edges. The light was provided by a flashlamp-pumped frequency-doubled Nd:YAG laser (Spectra-Physics PIV 400) operating at 10 Hz and around 150 millijoules per pulse. In the second setup the camera was positioned normal to the plate surface. For these tests the light was provided by a diode-pumped Nd:YLF laser (Coherent Evolution 90) operating at 10 khz and 9 millijoules per pulse. At this high framing rate the maximum resolution of the camera was 512x256. 2

3 CMOS camera M =5 cylinder flat plate laser sheet (a) (b) Figure 2. Schematic diagram of PLS setup. (a) Overall assembly and (b) inside the test section III. Results In the first phase of the study, surface streak line visualizations were obtained for several cylinder locations relative to the plate leading edge. Three examples, which highlight three flow regimes, are shown in Fig. 3. At each of these stations, and others, results were repeatable. The pattern shown in Fig 3a is the classic image seen in numerous studies in which the incoming boundary layer is fully turbulent. The separation line is symmetric about the plate-cylinder center line, with separation on center line occurring about two cylinder diameters upstream of the cylinder leading edge. In Fig 3b, with the cylinder shifted 1 in further upstream (4 in downstream of the plate leading edge) the flow structure and scales change significantly. The separation scale is significantly larger (on center line the primary separation line now is about 5 cylinder diameters upstream of the cylinder leading edge) and the flow structure is no longer symmetric about the cylinder center line. There is some indication of multiple separation lines. The significant variation in spanwise structure is also evident in the plan view PLS images and in the conventional side-view schlieren images which will be shown later. At this stage it is not clear if the incoming boundary layer is transitional, if transition occurs in the separated shear layer, elsewhere in the interaction, or some combination of the latter. Which ever it is, it is probable that the location and process will not be symmetric across the span and will result in an asymmetric flow structure. It is possible that disturbances originating at the plate leading edge trigger early transition, or waves from the tunnel floor or sidewalls hasten or delay transition at some locations but not at others. In any event, the global behavior is repeatable, since the essential features of the image remain the same from run to run. (a) (b) (c) Figure 3. Surface streakline visualization (kerosene-lampblack) of cylinder-induced interactions in a Mach 5 flow. (a) Fully turbulent : X cyl = 5 in, (b) transitional : X cyl = 4 in, and (c) laminar : X cyl = 2.75 in. 3

4 When the cylinder is shifted 1.25 in further upstream (X cyl = 2.75 in) the interaction becomes larger still, and nominally symmetric about the cylinder center line, indicative of a laminar-like interaction. The maximum primary separation distance is slightly larger than in Fig. 3b. Again, it is difficult to characterize the flowfield as being laminar or transitional since it may well be a mix of these two flows. With respect to primary separation location, Young et al. (1968) reported that in laminar flow at Mach 3 the center line separation distance varied from 4.1 diameters to 6.8 diameters as the flow Reynolds number was increased by a factor of about 20, indicating a Reynolds number dependence. 1 Smaller values were observed at Mach 5. At Mach 5.5, Hung and Clauss (1980) reported values of 9 to 12 diameters for laminar flow. 2 However, as was pointed out by Özcan and Holt, they assumed that separation occurred where the surface temperature or static pressure first begins to rise above the undisturbed value and this assumption will overpredict the separated flow scale significantly since the distance from the initial rise in pressure to the separation location can be several cylinder diameters. 3 The Özcan and Holt study noted that for laminar cases with a height to diameter ratio greater than about 4 (essentially a semi-infinite cylinder) the maximum primary separation distance was typically between about 6.6 and 7.6 diameters upstream of the cylinder. In Fig. 3c it is likely that the incoming undisturbed boundary layer is on the verge of transition, or is perhaps laminar, and transition occurs early in the separated shear layer. Considering that the location of primary separation on center line decreases from about 6-8 diameters to about two diameters as the flow goes from entirely laminar to fully turbulent, a large range of scales are possible depending on where and how transition occurs in the disturbed flowfield. The normalized separation distances, l sep /d, from the surface flow visualization of the current study agree well with the results of Kaufman et al., as shown in Fig In the figure, X sep and X trans are the locations of separation and the end of transition, respectively, measured from the plate leading edge. The aforementioned large change in mean interaction scale as it changes from laminar to transitional to turbulent is evident. The reduction of l sep /d from the laminar/transitional values to the fully turbulent values occurred at the same positions relative to transition even though the largest laminar/transitional extent of separation was about 10% higher than in Kaufman et al. 4 Comparisons with the data of Özcan and Holt 3 and Young et al. 1 suggest that this difference in l sep /d is reasonable. In order to make the comparison shown in Fig. 4 it was necessary to employ a correlation function developed by Ramesh and Tannehill 5. This correlation predicts the Reynolds number for the onset and end of transition (between which is the transition band ). The value for the end of transition, which was calculated to be 4.1 in, was used to normalize the x-axis of Fig. 4. Figure 4. Comparison of data from current study with that of Kaufman et al. 4 Planar laser scattering images for similar cases to those discussed above are consistent with surface streakline images and bring out more details of these complex flows. Figure 5a shows the case that presumably corresponds to a laminar interaction similar to Fig. 3c. In this case the cylinder diameter is slightly smaller (0.375 in compared to 0.5 in and is positioned about 0.7 in closer to the plate leading edge). Recall that the light sheet is parallel to and 0.1 in above the plate surface and the camera views the scattering from an oblique angle and so some distortion and blurriness are present at the edges of the images. Light and dark regions correspond to regions of high and low fog density, respectively. Low velocity regions are rendered as black owing to droplet evaporation in the relatively warm fluid. Figure 5a reveals a separation shock that stands off from the cylinder by about 3.6 diameters. Note that the laser sheet is elevated off the plate and so the origin of the shock is farther upstream. We should emphasize that although we assume this shock is a separation shock, we see no evidence of separated flow immediately downstream 4

5 of it. It is possible that separated flow is not observed because the laser sheet location was too high to see it. This issue will be explored in future work. We observe that the separation shock is relatively steady and its shape along the centerline of the plate is almost straight. The latter observation is consistent with many of the surface streakline patterns that we have taken in laminar interactions. In Fig. 5b the cylinder is at the same location relative to the plate leading edge but a trip strip (a strip of tape) has been attached to the plate surface just downstream of the leading edge. The interaction now has a mixed structure with the flow upstream and to the left of the cylinder (looking upstream) exhibiting a laminar-like character whereas the flow on the right exhibits a turbulent-like character with separation much closer to the cylinder. On the left hand side, downstream of the laminar-like separation location dark turbulent streaks/spots are evident in the image. We believe that a turbulent separation shock may exist on the left side that is similar to that on the right, but it cannot be seen due to the blur in the image. We note that this flow is much more unsteady than the previous case as the laminar shock is highly variable in its shape and the region of turbulent flow exhibits large variations in its overall scale. cylinder M =5 plate leading edge "laminar" separation shock 2.1" (a) "turbulent" separation shock M =5 boundary layer trip strip "laminar" separation shock (b) Figure 5. Planar laser fog scattering images of cylinder-induced interactions. The in cylinder is 2.1 in from the leading edge. The view shown is a plan view that exhibits some distortion because it was taken at an oblique angle. (a) untripped boundary layer, (b) tripped boundary layer. 5

6 Figure 6 shows the case when the cylinder is moved about 1.9 in further downstream. The image now becomes essentially symmetric with respect to the centerline of the plate (consistent with the surface streakline visualizations). The dark streaks in the image are associated with increased temperatures associated with what is presumably the separated flow. On center line these dark streaks first appear about 2 cylinder diameters upstream of the cylinder, consistent with the location of turbulent separation as inferred from other techniques. Interestingly, there is no separation shock that can be identified in the turbulent cases. It may be that the laser sheet is too close to the wall to effectively see the separation shock. Note that the same flow structure is observed in Fig. 6 whether the trip strip is in place or not. M = 5 t u r b u l e n t s e p a r a t e d f l o w Figure 6. Planar laser for scattering images for a case where the cylinder is farther downstream, thus generating a turbulent interaction. X cyl = 3.95 in. When the flow was imaged at a higher rate (10 khz vs. 10 Hz), some very interesting phenomena were observed. The problem with imaging the separation shock again presented itself, so the state of the boundary layer could not be inferred from λ sep. However, the location of transition could be found by imaging the boundary layer development of the tripped flow with the cylinder removed. Figure 7 shows representative images of the boundary layer development on the plate. The laser sheet is 0.04 in from the plate surface, and the flow is left to right. This PLS technique seems to be very effective for imaging this development; laminar, transitional, and turbulent flows are clearly distinguishable and the cylinder could be placed accordingly to generate a certain type of interaction. (a) (b) (c) Figure 7. Development of boundary layer on flat plate without interaction. (a) laminar, (b) transitional, and (c) turbulent. Flow is from left to right. 6

7 The changes in scale of the separated flow in both the streamwise and spanwise directions in a transitional interaction are quite substantial. The shape of the interaction is also highly variable and very sensitive to the state of the incoming boundary layer. Figure 8 shows four non-sequential images of such an interaction taken at 10 khz. These particular images were chosen to illustrate the unsteadiness of the interaction. This particular interaction is occurring towards the laminar edge of the transition band. The bulge a quarter of the way up the frame remains fairly constant and looks to be caused by a roughness element upstream either on the plate surface or leading edge. It is also interesting that the transitional interactions do not seem to be time resolved at 10 khz, but the fully turbulent interactions do. This interaction extends farther upstream for a given spanwise location than in the more turbulent cases. Figure 8. Non-sequential images of a more laminar transitional interaction. Flow is from left to right. Cylinder and dark area to right of white line added in processing. When the interaction occurs farther into the transitional band, the separated flow begins to look more like the combined laminar/turbulent interaction, with laminar scaling close to the centerline and turbulent scaling off-center, as seen in Fig. 9. The interactions are more swept and become more steady, relatively speaking. Figure 9. Non-sequential images of turbulent transitional interaction. Flow is from left to right. Cylinder and dark area to right of white line added in processing. A fully turbulent interaction is shown in Fig. 10. The scales of this interaction are smaller in both dimensions than those previous, and both the scales and the shape remain relatively constant from frame to frame. While the turbulent case is much more unsteady than a purely laminar interaction is, the data would suggest that the 7

8 unsteadiness is at a maximum near the laminar end of transition. This may seem counter-intuitive, but another flow feature was observed that might be offer some insight into this trend. Figure 10. Sequential images of turbulent interaction imaged at 10 khz. Flow is from left to right. Cylinder and dark area to right of white line added in processing. When the cylinder was placed in tripped flow that appeared essentially laminar there were intermittent bursts of turbulence extending up to 4.5 diameters upstream and occurring as much as 3.5 diameters off of the centerline, as shown in Fig. 8. These bursts tend to have a consistent general shape, that of a V pointing downstream, and bear a striking resemblance to Emmons turbulent spots that appear in the transition process 6. The light band immediately upstream of these structures is characteristic of a shock. These spots were found to occur in intermittent groupings of three or four spots, but within each grouping the spots were temporally very close together, with as little as 200 microseconds separating them. While no definitive conclusions can be drawn yet, it is possible that the highly unsteady nature of the transitional interaction could be closely related to the formation of these spots. Figure 9 shows sample side-view schlieren images of a transitional interaction. Since these images are the result of integration of the light beam across the width of the test section care is needed in their interpretation. The blurred inclined wave in the upper half of the image and the clearly defined vertical wave immediately upstream of the Figure 8. Turbulent spot formation in interaction region. Flow is from left to right. Cylinder and dark area to right of white line added in processing. d = in, X cyl = 5.9. Images are not sequential. 8

9 cylinder are easily identified as the rippling plate leading edge shock wave and the steady cylinder bow shock, respectively. There appear to be several shock waves that coalesce near the triple point, with the upstream wave appearing weaker. The upstream wave is likely the laminar separation shock, whereas the downstream waves are likely turbulent separation shocks. The plan-view PLS images and surface streakline images show that the flow structure is highly non-uniform in the spanwise direction and so these waves are probably generated at different spanwise locations. Cylinder leading-edge shock Plate (a) Leading-edge shock Flow laminar separation shock turbulent separation shocks Cylinder bow shock (a) (b) Figure 9. Spark schlieren photographs of a transitional interaction. (a) Schematic showing field of view, (b) and (c) are sample images. IV. Conclusion The plan view PLS images have proven to be very effective in visualizing the global flowfield structure of laminar, transitional and turbulent interactions. The laminar interactions are characterized by a single, relatively steady, separation shock, which exhibits a large radius of curvature. The turbulent case is more symmetrically distributed and more three-dimensional. The transitional interactions exhibit a laminar separation shock near the cylinder centerline but turbulent separation shocks in the outboard regions. This dual-shock behavior leads to a separation shock structure that exhibits an inflection point. The PLS images demonstrate that the flow structure of the transitional interactions is highly spanwise non-uniform and therefore care should be taken when interpreting spatially integrated techniques such as schlieren or shadowgraphs. A significant piece of the explanation for the highly unsteady nature of the transitional interaction may well lie in the formation of turbulent spots and their behavior within the interaction itself. Future work should focus on these spots both independently and in combination with the SWBLI. Additionally, future work will have to address the PLS imaging of the primary separation shock. Acknowledgments This work was sponsored (in part) by the Air Force Office of Scientific Research, USAF, under grant/contract number FA The views and conclusions contained herein are those of the authors and should not be 9

10 interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the Air Force Office of Scientific Research or the U.S. Government. The authors would like to thank Dr. Bharath Ganipathisubramani and Pablo Bueno for their help with the experimental setup and execution and Edward J. Zihlman and Frank Wise for their technical expertise. The authors would also like to thank machinists Travis Crooks, David Gray and Joe Edgar References 1 Young, F.L., Kaufman, L.G., and Korkegi, R.H., Experimental Investigation of the Interactions Between Blunt Fin Shock Waves and Adjacent Boundary Layers at Mach Numbers 3 and 5, ARL , Dec Hung, F.T. and Clauss J.M., Three-Dimensional Protuberance Interference Heating in High-Speed Flow, AIAA , AIAA 18 th Aerospace Sciences Meeting, Jan Özcan O. and Holt M. (1984), Supersonic Separated Flow past a Cylindrical Obstacle on a Flat Plate, AIAA Journal, Vol. 22 No. 5, pp Kaufman, L. G., II, Korkegi R. H. and Morton, L. C., Shock Impingement Caused By Boundary Layer Separation Ahead of Blunt Fins, ARL , Aerospace Research Laboratories, Wright-Patterson Air Force Base, Ohio, Aug Ramesh, M. D. and Tannehill J. C., Correlations To Predict Transition In Two-Dimensional Supersonic Flows, 33 rd AIAA Fluid Dynamics Conference and Exhibit, June Cantwell, B.J., Coles, D. and Dimotakis, P., Structure and entrainment in the plane of symmetry of a turbulent spot, Journal of Fluid Mechanics, Vol. 87,