Particle Image Velocimetry at a Generic Space Launcher Model at Mach 5.9

Transcription

1 Particle Image Velocimetry at a Generic Space Launcher Model at Mach 5.9 Marcus Casper 1,*, Sören Stephan 1, Peter Scholz 1, Rolf Radespiel 1 1: Institute of Fluid Mechanics, Technische Universität Braunschweig, Braunschweig, Germany * correspondent author: m.casper@tu-bs.de Abstract This contribution discusses particle image velocimetry measurements at a generic space launcher model with the focus on liquid tracer particles at a Mach number of M=5.9. The test facility, the Hypersonic Ludwig Tube Braunschweig, is a blow down type wind tunnel which allows unit Reynolds numbers between Re= m -1 and Re= m -1 at a Mach number of M=5.9. Due to the conditions inside the storage tube, a pressure of up to 30bar and a temperature of up to 623K, the material of the tracer particles have to be chosen carefully. The presented work includes a qualification of the atomizer ATM210 with two different oils, Emery 3004 and Plantfluid. The results show that the quality of the tracer particles at the outlet of the atomizer complies with the requirements. During the wind tunnel measurements the effects of the aerosol facility operating parameters, wind tunnel parameters, and laser energy to the tracer particles were investigated. Both oils were tested and the oil Plantfluid was found to be more resistant to higher temperatures. Additionally, the number of particle images increased at higher laser energy. However, the resulting velocity vector fields show that the quantity of tracer particles inside the boundary layer, the wake, and the recirculation area behind the generic space launcher model was insufficient to measure turbulent stresses. A mean velocity vector field, based on 160 double images, could be computed. Based on numerical flow simulations and particle path calculations the problems with the tracer particles were analyzed. Measurements inside the recirculation area, using local seeding, was also tested. 1. Introduction Particle image velocimetry is a promising and established non-intrusive measurement technique. Due to the fact that the velocity is measured indirectly by tracer particles these particles must comply with important requirements. Among the optical properties of the tracer particles, one important requirement is the relaxation time which could be influenced by the size as well as the density of the tracer particle. There are basically two kinds of tracer particles, solid [4, 6, 8, 10, 11] or liquid [7], which can be used in hypersonic test facilities. Among other things the choice of a solid or liquid material depends on the conditions inside the wind tunnel. Hypersonic test facilities can have temperatures higher than 1500K [4] and, thus, solid materials like titanium dioxide or alumina oxide are used [5]. These materials are available as powders. The resulting tracer particles are usually agglomerated which increases the tracer particle size [8]. If needed, the agglomerate can be broken down to primary particle size and, finally, the nano particles can be prevented to reagglomerate [10]. Based on solid nano tracer particles the measured velocities are in good agreement with theory [5, 10]. However, in hypersonic test facilities solid tracer particles used to seed the wind tunnel flow can cause problems. The tracer particles pollute and harm moving parts of the wind tunnel, such as valves and pumps. These problems can be solved using liquid, or rather oil, based tracer particles. Additionally, oil based tracer particles like Emery 3004 have a reduced mass compared to solid tracer particles of same size. Thus, the relaxation time is reduced and this will enhance the resulting velocity vector field. However, a new technical problem appears with the storage tank of a blow down facility. For example hypersonic Mach 6 Ludwieg tubes operate with stagnation pressures of around 30 bars and total temperatures between 400K and 800K. At these conditions an oil-air mixture could result in an explosion and, thus, the aerosol has to be handled with care [1]

2 2. Experimental Setup Wind tunnel The measurements were conducted in the Hypersonic Ludwig Tube Braunschweig (HLB) which is a blow down type wind tunnel as seen in figure 1. The high-pressure section consists of 17m long storage tube including a 14m non-heated part and a 3m heated part. Inside this section a pressure of up to 30bar and a temperature of up to 623K are possible. The low-pressure section consists of a laval-nozzle, a test section, and a 6m³ vacuum tank. The high- and low-pressure sections are separated by a fast acting pneumatic valve. Inside the test section unit Reynolds numbers between Re= m -1 and Re= m -1 at a Mach number of M=5.9 are possible. The measurement time per wind tunnel run is approximately 80ms. Detailed information about the test facility are described in [2]. The particle image velocimetry (PIV) measurements presented in this work were conducted with a pressure of 15.7bar and a temperature of 418K inside the storage tube, if not noted otherwise. The resulting unit Reynolds number is Re= m -1. Storage tube Heater Fast acting valve Laval-nozzle Test section Diffusor Vacuum tank Aerosol HP In Out Aerosol LP ATM 210 Aerosol tank Figure 1 The Hypersonic Ludwig Tube Braunschweig (HLB) and the aerosol facility Partcile image velocimetry setup The PIV setup is presented in figure 2a. The laser light source was a Nd:YAG double pulse laser system with an energy of 150mJ per pulse. The laser light sheet with a thickness of less than 1mm was focused by one pair of lenses consisting of a plano-concave lens with a focal length of -50mm and a plano-convex lens with a focal length of 75mm. A cylindrical lens with a focal length of mm forms the light sheet. To observe the effect of an increased laser energy this cylindrical lens was exchanged by a lens with a focal length -25mm and then -50mm. The camera Imager Pro X 11M with a Tamron SP AF 180mm F/3.5 Di LD[IF] Macro 1:1 recorded the PIV images. The camera aperture was set to 3.5 to maximize the brightness of the particle images. The complete setup was mounted on a scaffold which had no direct contact to the wind tunnel and was placed on small rubber mats. This prevented misalignments of the light sheet due to vibrations caused by the wind tunnel. The atomizer ATM 210 (Topas) was used to generate oil based tracer particles. It allows pressures up to 10bar at the outlet. The particle size distribution of DEHS based aerosols has a peak at 200nm using a pressure difference between inlet and outlet of approximately 5bar. Two oils, the standard oil Emery 3004 (Cognis Corporation) and the oil Plantfluid (Bechem), were tested during the measurements. The aerosol for every run was pumped (figure 1, Aerosol LP) into a aerosol tank with a volume of 10litres. To inject the aerosol into the storage tube the aerosol tank was pressurized (figure 1, In) and connected to the storage tube (figure 1, Aerosol HP). Finally, the aerosol tank was depressurized (figure 1, Out)

3 Light sheet optics a) b) Laser Calibration grid Test section U Camera Corundum Figure 2 The experimental setup (a) and the generic space launcher model with calibration grid (b) Generic space launcher wind tunnel model The 144mm long front part of the generic space launcher model consists of a spherical nose with a radius of 10mm and a cone with an angle of 36. The rear part is a 328.6mm long cylinder with a diameter of 108mm. The wind tunnel model is made of Plexiglas and painted in dull black. Transition is fixed with a ring of corundum grain (400µm) glued on the surface of the conical part 98mm behind the nose. A tail sting with a diameter of 43mm connected the wind tunnel model with the diffuser downstream of the test section. Data eavluation To calculate the velocity vectors of the PIV measurements the software Davis 7.2 was used. The recorded raw images were preprocessed applying a shift correction and a filter (subtract sliding minimum). Areas with laser light reflections were masked. Finally, the velocity vectors were calculated. The tracer particle images of the PIV measurements were counted by a program based on Matlab R2008a. The program analyzed an area of x 44.53mm² above the wind tunnel model. Before the counting was initialized, every individual image was prepared to account for the differences in background brightness, and variations in the brightness and size of the tracer particle images. First, the non-uniform background brightness in the image was reduced by using two filters, a wiener 2 lowpass-filter to remove noise and a medfilt 2 median-filter to remove the tracer particle images. The resulting background image was used to remove unfocused bright areas from the non-filtered image. The new image with homogeneous background had some noise and, thus, it was also filtered with the wiener 2 lowpass-filter. This prepared image was converted into a binary black white image. The level to differ between black and white was a mean value of all local minima in this image with an additional offset. This level ensures that no noise was interpreted as tracer particle image. The local minima should not exceed this level and the violations were counted. The maximum violations were 5.3% of the registered tracer particle images. The tracer particles paths were also simulated in the present work. For this purpose, numerical flow simulations based on the Reynolds-averaged Navier-Stokes equations were obtained from RWTH Aachen [3]. These computations used the same flow geometry and the same Reynolds and Mach number. The simulation data was then input to Tecplot that allows to compute tracer particle paths for given particle data

4 16th Int Symp on Applications of Laser Techniques to Fluid Mechanics 3. Results The manufacturer of the aerosol generator ATM210, Topas, reports a particle size distribution with a peak at 200nm atomizing the oil DEHS with a pressure difference of approximately 5bar between the inlet and the outlet. As different oils were used in the present work the resulting particle size distribution was analyzed here. Both oils, Emery 3004 and Plantfluid, were atomized using a pressure of 5 bar at the inlet. Two oil temperatures, 18 C and 50 C, were investigated. This varied the viscosity of the oil which could influence the particle size distribution. Note that the oil Plantfluid has an increased viscosity compared to the oil Emery 3004 and, thus, the oil Plantfluid was heated to 50 C during the PIV measurements. The particle size distribution was measured by a Fast Mobility Particle Sizer (FMPS, TSI Inc.) which detects particles sizes between 5.6 nm and 560nm. It should be noticed that particle sizes larger than 560nm are detected as smaller particles. The results are presented in figure 3. a) b) Figure 3 Particle size distribution absolute (a) and normalized (b) of the oils Emery 3004 and Plantfluid Figure 3a shows that the peak of the particle size distribution is at 200nm independent of the oil and the oil temperature but it seems that the number of particles changes. Thus, the particle size distribution is normalized by the number of particles detected between 50nm and 560nm (Emery 3004: particles at 18 C and particles at 50 C; Plantfluid: particles at 18 C and at 50 C). The normalized particle distribution in figure 3b indicates that more smaller particles and less larger particles are produced with the oil Plantfluid compared to the oil Emery The influence of the temperature seems to be small. Obviously, with respect to the particle size distribution there exists no reason to exclude any of the oils from the PIV measurements. More than 85% of the detected particles are smaller than 300nm. Before a wind tunnel run was initialized a certain volume filled with tracer particles was injected into the storage tube. Figure 4a shows the theoretical saturation of oil inside the storage tube based on a fixed amount of oil per run. The amount of oil within the storage tube increases with every aerosol injection until the aerosol losses due to a wind tunnel run are equal to the amount of injected oil. Figure 4b presents the detected tracer particle images of the observed start-up phases of seeding the storage tube. The figure includes our standard injection procedure using two injections of aerosol which was produced by the atomizer during a runtime of 9 minutes. Additionally, the figure includes three start-up procedures with a different first injection. The storage tube pressure was initially set to 6bar and the aerosol was directly pumped into the storage tube for 20 minutes. -4-

5 However, the resulting first runs in figure 4b are identical. The observed start-up procedure indicates that only a partial amount of the injected tracer particles are visible. After three wind tunnel runs the number of detected tracer particles is approximately constant. This differs from the theoretical saturation. Possibly a part of the tracer particles evaporated or were lost by wall contact during the time between two runs of about 20min. An aerosol inlet closer to the fast acting valve could increase the number of tracer particles but this causes other problems [1]. a) b) Figure 4 Theoretical saturation of oil (a) and (b) detected tracer particle images inside the wind tunnel using the oil Emery 3004 During the measurements the injection procedure of the aerosol was optimized. The atomizer always filled up the aerosol tank first. Then, the aerosol tank was pressurized to a higher level than the storage tube, in order to transfer the particles into the storage tube. Figure 5a presents the detected tracer particle images based on a pressure of 25bar and 18bar within the aerosol tank. The figure clearly shows that a lower pressure results in more detected tracer particles. This indicates that tracer particles collided with the pipe walls at higher aerosol tank pressures causing particle losses. a) b) Figure 5 Influence of (a) the pressure inside the aerosol tank using the oil Emery 3004 and (b) the oil - 5 -

6 In a next step the oil was changed from Emery 3004 to Plantfluid. Plantfluid was first tested at the same conditions inside the wind tunnel and the aerosol facility. The detected number of Plantfluid and Emery 3004 based tracer particle images is identical as seen in figure 5b. This result corresponds to the measured particle size distributions in figure 3. a) b) Figure 6 Influence of (a) the laser energy using the oil Emery 3004 and (b) the temperature of the storage tube The influence of the laser energy was analysed by decreasing the length of the light sheet. This was realized with two different cylindrical lenses. The standard cylindrical lens had a focal length of mm to illuminate the complete rear part of the generic space launcher model. To increase the laser energy by a factor of 4 a cylindrical lens with a focal length of -50mm was used. The result is presented in figure 6a. It shows that more tracer particles are detected with an increased laser energy. Thus, there were small tracer particles which did not scattered enough light to be detected. However, the increased laser energy also caused a problem. The reflections on the surface of the wind tunnel model were increased. This impaired measurements inside the boundary layer. Several approaches exist for reducing the reflections of the present set up and these should be explored in future work. The influence of the temperature in the heated part of the storage tube is displayed in figure 6b. The number of Emery 3004 based tracer particles decreases at an increased temperature of 450K. The number of Plantfluid based tracer particle images at 450K is similar to the value of Emery based tracer particle images at 418K. The results of figure 5b and figure 3 indicate that there are no losses of Plantfluid tracer particles due to the increased temperature. This allows higher temperatures inside the storage tube which is useful to avoid condensation in the test section. Qualitative data of raw images of Emery 3004 (a) and Plantfluid (b) based tracer particles are presented in figure 7, to supplement the quantitative results of figure 6b. The raw images are recorded at a temperature of 450K inside the storage tube. The raw images confirm the count result of the detected tracer particle images. Figure 7b also shows the low seeded area behind the generic space launcher model. The number of small tracer particles is too low, even though the tracer particles are more resistant to higher temperatures. Thus, it should be clarified if the procedure to transfer the aerosol into the storage tube influences particle size distribution

7 a) 1 1 b) Figure 7 Raw images with (a) Emery 3004 and (b) Plantfluid based tracer particles at a temperature of 450K inside the storage tube The mean velocity vector field in figure 8a is based on 160 double images which were measured with the oil Emery The resolution is 1.02 x 1.02 mm² and every third velocity vector is shown. The coloured contour represents the velocity. The mean velocity vectors in the complete upper area 2-7 -

8 16th Int Symp on Applications of Laser Techniques to Fluid Mechanics are available whereas the mean velocity vectors in the wake are partly unknown. The basic problem is that the single velocity vector fields have more blank spaces in the boundary layer and the wake due to less tracer particles as seen in figure 7b. a) b) Figure 8 Mean velocity vector field (a) and number of velocity vectors (b). Tracer particles: Emery 3004 Figure 8b visualizes the problem. The figure presents the number of velocity vectors normalized by the number of double images. In the lower boundary layer every mean velocity vector is based on less than 5% of the possible velocity vectors. This value increases to 50% 5mm above the surface and the isoline of this value in the wake is horizontal. 1 1 Figure 9 Raw image with Emery 3004 (main flow) and TiO2 (recirculation area) based tracer particles -8-

9 Based on the results obtained so far we have to conclude that oil based seeding of the wind tunnel free stream does not allow for measuring the wake region. One solution could be local seeding but the flow should not be influenced by any active device needed to transport the tracer particles into this area. In one approach the tracer particles are placed in the area before the wind tunnel run starts. This approach is tested by two wind tunnel runs with some non-prepared and hence agglomerated titanium dioxide powder. The test section was opened and the powder was positioned behind the generic space launcher model on the tail sting. Then, the test section was closed, evacuated, and one wind tunnel run was done. One example is presented in figure 9. The wake of the generic space launcher model is filled with titanium dioxide particles. Such a method of local seeding will possibly allow to measure turbulent stresses in this area. Additionally, only a small quantity of powder is needed which should not harm the wind tunnel. These first results demonstrate that local seeding of the recirculation area is feasible. However, PIV data evaluation parameters must be adjusted to obtain resolution of both the outer flow field and details of the wake flow. We note that local seeding of the wake does not resolve the lack of tracer particles in the forebody boundary layer and, hence, within parts of the wake shear layer. In the following analysis we therefore assess the inhomogeneous tracer particle distribution observed along the model forebody. This is performed by using a numerical simulation of the generic space launcher model at the same Reynolds- and Mach number [3]. The numerical simulation allows to analyze various sources of particle image inhomogeneities as observed in the measurements. Here we assume that the tracer particles are homogeneously mixed inside the storage tube. In ideal conditions the tracer particles react to changes of the flow immediately and, thus, the tracer particles will be distributed as the flow density. In a hypersonic Mach 6 blow down facility one observes three major density variations: 1.) drop of the density about the nozzle, 2.) density increase across shock waves, and 3.) density drop across boundary layers. To visualize the ideal tracer particle distribution around the generic space launcher model figure 10a shows the density close to the surface at the beginning and the end of the cylindrical part. The presented density distributions in z-direction may be regarded as an upper bound of possible seeding based on homogeneously distributed tracer particles inside the storage tube. a) b) Figure10 Density ratio (a) and comparison of the density ratio with the number of velocity vectors (b) at x=- 10mm An indicator for the observed tracer particle distribution can be derived from the counted velocity vectors as presented in figure 8b. With respect to all double images and the used PIV data evaluation method a large number of tracer particles at one position will more often result in a - 9 -

10 velocity vector compared to a position with less tracer particles. In order to make the flow density and the velocity vector count comparable, both are normalized with their uniform values outside the boundary layer edge. The result is presented in figure 10b. Comparing the ideal and the observed gradients show that the trends, an increase of the ratio from the surface to the outer flow, are similar. The difference between both lines could be a result of the relaxation time of the tracer particles. At the end of the conical part and the begin of the cylindrical part non-ideal tracer particles are shifted away from the wall due to the relaxation time. This could reduce the number of the tracer particles close to the surface at the trailing edge. a) b) Figure 11 Streamlines and particles paths at (a) the nose and (b) the end of the conical part The paths of ideal tracer particles are the streamlines as presented in figure 11. Figure 11a shows a streamline which starts in front of the generic space launcher model and figure 11b shows a streamline which starts at the end of the conical part. Based on the assumption that a continuum flow exists tracer particles with given mass and locally calculated drag coefficient C D =var. can be simulated and compared with streamlines. The simulated tracer particles start at the same position as the streamlines. Figure 11a presents the simulated tracer particles with a diameter of 200nm and 1000nm at the nose. The deviations from the streamlines are insignificant independent of the tracer particle diameter. The influence of the drag coefficient is checked for the tracer particles with the diameter of 200nm. The drag coefficient is set to the fixed values C D =[1, 10, 100, 1000]. The deviation between the particle path with fixed and locally calculated drag coefficient decreases with a higher drag coefficient. This indicates that the locally calculated drag coefficient, which changes due to the changes of the tracer particle velocity, has a mean value in the order of Figure 11b shows the begin of the cylindrical part. Again the sensitivity of the tracer particle path due to the drag coefficient is displayed. The drag coefficient depends on the flow regime and must be calculated with respect to the Reynolds and Mach number as well as the Knudsen number [9]. The authors [9] recommend the formula 0. [ ( k Re) ] ξ( Kn)C C D = k (1) Re because it can be used in rarefaction regimes over broad ranges of Reynolds number (Re 200), Mach number (M 1) and Knudsen number (continuum to free molecular). The formula differs to the formula which is used by Tecplot to calculate the drag coefficient. Based on a tracer

11 particle path with locally calculated drag coefficient the velocity difference between the tracer particle and surrounding flow can be estimated. These estimates were used at the end of the conical part and the beginning of the cylindrical part for evaluations of equation (1) which resulted in drag coefficient around C D =3. Thus, the particle paths based on C D =1 and C D =10 in figure 11b are possibly more realistic representations of real tracer particle behaviors. We note that the tracer particle with a diameter of 200nm and a drag coefficient of C D =1 results in displacement of 0.6mm versus the streamline at the trailing edge of the generic space launcher model. This value increases to 1.6mm based on a tracer particle with a diameter of 1000nm and a drag coefficient of C D =1. Thus, it seems that both the flow density close to the model surface and relaxation time of tracer particles contribute to the observed decrease of tracer particles close to the wall. 4. Conclusions and future work The present work shows that it is possible to use oil based tracer particles in a blow down type wind tunnel with heated storage tube. An alternative oil, Plantfluid, is presented which is more resistant to higher temperatures. The number of detected tracer particles can be increased by using low pressure differences to transfer the tracer particles from the aerosol tank to the storage tube and by using high levels of laser energy for PIV measurements. The results also show that there are still problems to seed the level sensor compressed air oil reservoir and pressure reducer Figure 12 Sketch of an integrated atomizer nozzle tracer particles ATM210 nozzle flow with tracer particles independent of the used oil. In hypersonic flows the seeding density decreases inside the boundary layer compared to the outer flow. This problem can only be solved by additional tracer particles injected inside the storage tube or close to the surface of the generic space launcher model. The tracer particle path simulation shows that oil based tracer particles with a diameter of 200nm have only small deviations from the streamlines, as the relaxation time is short. The atomizer ATM210 is well suited to produce the needed tracer particles but the effect to the tracer particles during the procedure to inject the particles into the storage tube of the wind tunnel is still unknown. Thus, the particle size distribution in the storage tube should be measured as well as to check the possibility of droplet growth. One approach to simplify the aerosol transport into the storage tube is to integrate the atomizer nozzle into the storage tube. Figure 12 displays that this can be accomplished with only a few modifications of flange and support tubing. The problem, that there are no tracer particles inside the recirculation area, must be solved with local seeding. The tests show that injected tracer particles remain in the recirculation area even though the tracer particles were injected prior to a run. If titanium dioxide particles will be used the particles should be prepared [10]

12 Acknowledgment The work was funded by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) within the framework Sonderforschungsbereich Transregio 40 (Technological foundations for the design of thermally and mechanically highly loaded components of future space transportation systems). The authors want to thank I. Kirsch and E. Uhde (Fraunhofer Wilhelm- Klauditz-Institut, WKI) for providing the equipment for the measurements of the particle size distribution as well as V. Statnikov for providing the numerical Simulation. References [1] Casper M, Stephan S, Windte J, Scholz P, Radespiel R, Scharnowski S, Kähler C J (2012) Hypersonic PIV in a Ludwieg Tube Wind Tunnel at Mach 5.9, 28th AIAA Aerodynamic Measurement Technology, Ground Testing, and Flight Testing Conference, New Orleans, Louisiana, USA, June [2] Estorf M,Wolf T, and Radespiel R (2005) Experimental and numerical investigations on the operation of the Hypersonic Ludwieg Tube Braunschweig, Proceedings of the 5th European Symposium on Aerothermodynamics for Space Vehicles., ESA SP-563: [3] Glatzer C, Meiss J-H, Meinke M, Schröder W (2011) Numerical Investigation of the Near Wake of Generic Space Launcher Systems at Transonic and Hypersonic Flows, 4th European Conference for Aerospace Science (EuCASS), St. Petersburg, Russia [4] Havermann M, Haertig J, Rey C, George A (2002) Application of particle image velocimetry to high speed supersonic flows in a shock tunnel, 11th international symposium on application of laser technology to fluid mechanics, Lisbon, Portugal [5] Havermann M, Haertig J, Rey C, and George A (2008) PIV Measurements in Shock Tunnels and Shock Tubes, Particle Image Velocimetry, Topics in Applied Physics, Volume 112/2008, , DOI: / _23 [6] Humphreys W M, Rallo R A, Hunter W W, and Bartram S M (1993) Application of Particle Velocimetry to Mach 6 Flows, 5th International Conference of Laser Anemometry, Advances and Applications, Veldhoven, The Netherlands, Aug. [7] Lang N (2000) Investigations of the Supersonic Flow Field around a delta wing using Particle- Image-Velocimetry, Proceedings of the 10th International Symposium on Applications of Laser Techniques to Fluid Mechanics, Lisbon, Portugal [8] Schrijer, F F J, Scarano F, Oudheusden B W van (2006) Application of PIV in a Mach 7 double-ramp flow, Experiments in Fluids 41: [9] Tedeschi G, Gouin H, Elena M (1999) Motion of tracer particles in supersonic flows, Experiments in Fluids, Volume 26, Number 4, , DOI: /s [10] Wernet J H, Wernet M P (1994) Stabilized Alumina/Ethanol Colloidal Dispersion for Seeding High Temperature Air Flows, Proceedings of the ASME Symposium on Laser Anemometry: Advances and Applications, Lake Tahoe, NV, June [11] Ünalmis Ö H, Hou Y X, Bueno P C, Clemens N T, and Dolling D S (2000) PIV Investigations of Role of Boundary Layer Velocity Fluctuations in Unsteady Shock-Induced Separation, 21st AIAA Aerodynamic Technology and Ground Testing Conference, Denver, CO, Juni