Application of fluorescent particles for particle tracking velocimetry in wind tunnels

Transcription

1 Application of fluorescent particles for particle tracking velocimetry in wind tunnels Tamara Guimarães 1,*, K. Todd Lowe 1 1: Dept. of Mechanical Engineering, Virginia Polytechnic Institute and State University, United States of America 2: Dept. of Aerospace and Ocean Engineering, Virginia Polytechnic Institute and State University, United States of America * Correspondent author: tgbucalo@vt.edu Keywords: PTV, PIV, Fluorescence ABSTRACT Laser flare from surfaces is one of the most common issues encountered in near surface measurements using optical techniques. This work presents the application of fluorescent particles for particle tracking velocimetry for near surface measurements in wind tunnels. By using polystyrene latex spheres doped with Kiton Red 620 (KR620) fluorescent dye, the flare issue is largely mitigated by optically filtering the incident scattered light. These particles have been well characterized in past work; however, previous velocimetry has been conducted using purely particle image velocimetry, demonstrating velocity measurements within 100 μm from the surface. The use of a particle tracking velocimetry approach provides the opportunity for measurements as close as 33 μm of the surface with the same imaging parameters. Results for laminar boundary layer flow over a flat plate illustrate that PTV with fluorescent particles for near surface measurements in wind tunnels is a valuable approach for resolving very near surface flow. 1. Introduction Despite the several advantages for making near wall measurements using optical techniques, the application of particle image velocimetry (PIV) and particle tracking velocimetry (PTV) near surfaces (Dracos 1996, Kähler et al. 2012) is still a challenge. Several methods have been developed to address the issue of laser flare, one of the most successful being based on fluorescent paints (e.g., Cadel et al. 2015) or, as in the case for the current work, fluorescent particles that absorb the laser light and emit in a different wavelength. As such, the intense surface-scattered laser flare may be rejected by a long pass filter, while the fluorescent emission from the particles remains in the received signal. This method has been proven efficient in water flows (e.g. Tauro et al. 2012),

2 but particles doped with safe fluorescent dyes have only recently been reported (Petrosky et al. 2015). In this work, a particle tracking velocimetry approach was combined with the fluorescent particle method of Petrosky et al. (2015) for measurements in a laminar boundary layer flow over a flat plate. Kiton Red (KR620)-doped polystyrene latex (PSL) particles of 0.87 μm mean diameter small enough for a wide range of wind tunnel applications were used for flow seeding, with imaging of only the fluorescent signal. Particles were resolved at distances as close as 33 μm to the plate surface, offering an approach for improved velocimetry in the near surface region. 2. Methods 2.1 Fluorescent Particles Polystyrene latex particles have been extensively and successfully used for PIV applications (Adrian and Westerweel, 2010). New manufacturing techniques have been applied by NASA Langley to ensure uniform size distribution of particles in the order of 1 μm (Tiemsin et al, 2012). In this work, PSL particles were doped with Kiton Red (KR620) for fluorescence, yielding particles with a mean diameter of 0.87 μm and standard deviation of 0.3 μm. The toxicity of KR620 is very low, making it a good candidate for use in large, open air applications. For comparison, two other common fluorescent dyes, Rhodamine B and Rhodamine 6G, are over five and twelve times more toxic than KR620, respectively, and are also considered carcinogenic (Adrian, 2011). Flat plate Fig. 1 Sequence of images highlighting the movement of one individual particle at a distance of approximately 60 μm from the surface. Flow is from bottom to top for ease of movement observation. The peak fluorescence emission of the excited particles occurred within the wavelength range of nm, when illuminated by a 527 nm double pulsed-laser. With that, any long-pass filter for wavelengths above 530 nm and below 580 nm would be appropriate for filtering the laser incident light and transmitting the fluorescent light. One of the main challenges still to be resolved with this technique is that the intensity of the fluorescent light is orders of magnitude lower than

3 the Mie scattered light, as shown in Fig. 2. The lower particle density obtained, however, makes this technique a good candidate for PTV (Scarano and Riethmuller, 2000), as can be seen in Fig. 1, where one particle is highlighted in a sequence of six images taken at 2.5 khz. 2.2 Experimental Setup Two different setups were used for this experiment. The experimental setup seen in Figs. 3 and 4 was used to capture images of the Mie-scattered and fluorescent light from the particles simultaneously for comparison. Results of an analysis using PIV are shown in Petrosky et al. (2015) and a side by side comparison of the light intensity for each case is presented in Fig. 2. Two Photron SA1.1 Fastcam high-speed cameras with a 1024 x 1024 pixel resolution and 12-bit intensity digitization were used for recording both a filtered and unfiltered image. The cameras were positioned above and below the airflow and perpendicular to the laser sheet, for two-dimensional (2D) PIV, Fig. 3. Camera 1 was mounted on a sliding rail, as seen in Fig. 4, which was attached to a piece of aluminum extrusion and secured to the optical table. Camera 2 was placed on a 3-axis traverse and 3-axis camera mount so that it could be positioned to image the same particles in the flow as camera 1. Fig. 2 Image of Kiton Red (KR620) doped polystyrene latex (PSL) particles with fluorescent filtering (Left) and Mie scattering (Right). Color was added for visualization purposes (Petrosky et al. 2015) The distance from the lens edges to the laser sheet plane for both cameras was approximately 5 inches. Two Sigma 105 mm f/2.8 EX DG macro lenses were used with the cameras to obtain a close-up image of the flow. The camera field of view was 29.6 x 29.6 mm for the two 2 camera testing and 30.7 x 30.7 mm for the one camera setup shown in Fig. 5. 2

4 18th International Symposium on the Application of Laser and Imaging Techniques to Fluid Mechanics LISBON PORTUGAL JULY 4 7, 2016 Fig. 3 Schematics of the 2-camera experimental setup at the Newport News Shipbuilding AOE Instructional Lab at Virginia Tech. Fig. 4 Details of the nozzle and optical system used (Left); Setup with the 2 cameras and the exit of the 6 cm nozzle (Center); 3-axis traverse where camera 2 was mounted (Right).

5 For fluorescence imaging, an Omega Optical 560 nm long pass filter, model 560HLP, was attached to the lens, rejecting Mie-scattered light from the particles while transmitting particleemitted fluorescent light to be captured by the camera. A 527 nm dual-head Nd:YLF laser (Photonics Model DM30) was used at approximately 23 mj/pulse to illuminate the flow and was synced with the camera by LaVision DaVis software, recording at 2.5 khz for time-resolved images. Finally, a cylindrical lens with focal length of f = -20 mm was used to form a thin laser sheet at the nozzle exit in the orientation depicted in Fig. 4. The laser sheet was approximately 1.25 mm thick and 3.5 cm wide in the measurement plane at the nozzle exit. The cameras imaged a region of flow about 8 cm from the nozzle exit. The KR620-doped PSL particles characterized previously were atomized using two Air-o- Swiss 7146 ultrasonic humidifiers. Seed was introduced well upstream of the wind tunnel nozzle into the blower inlet, where it mixed with the air at room temperature and flowed through a nozzle of 6 cm exit diameter. The airflow at the exit for PIV tests was about 4.5 m/s, corresponding to a Reynolds number per meter of approximately 280,000 m. Before each test, the KR620 particle -1 solution was mixed in equal proportions with distilled water and sonicated for fifteen minutes in an L&R Quantrex 90H ultrasonic disruptor to prevent particle agglomeration. The mixture was then removed from the disruptor and placed immediately into the vaporizer. Fig. 5 (a) Parallel flat plate orientation; (b) Rotated plate orientation and (c) Plate dimensions. For the second part of the experiment, the setup remained the same as seen in Fig. 3 except for the removal of camera 2. In these tests, the near-surface flow over a blunt leading edge flat aluminum plate was measured. In the first test setup, seen in Fig. 5 (a), the plate was oriented

6 parallel to the flow exiting the nozzle and perpendicular to the incident laser sheet. In the second test, seen in Fig. 5 (b), the plate was oriented 45 to the incident laser sheet but still parallel to the airflow. The dimensions of the flat plate are shown in Fig. 5 (c). For these single-camera tests, a set of 2000 double-frame fluorescent images was obtained with the filter over the lens. Then, the filter was quickly removed and another set of 2000 double-frame images was recorded of the Miescattered light from the seed particles. Fig. 6 Top: Flat plate camera images for: Fluorescent f/2.8 (Left), Mie f/2.8 (Center), and Mie f/22 (Right). Bottom: Contour plots covering the area within 5 mm from the plate surface for: Fluorescent f/2.8 (Left), Mie f/2.8 (Center), and Mie f/22 (Right). (Petrosky et al. 2015) The PIV processing of the image sets has been presented in previous work and an example result for velocity contours is provided in Fig. 6 (Petrosky et al. 2015). As noted in prior work, the first mean velocity point was resolved to 100 μm, only limited by the particle density and interrogation window size for the experiment. Despite the PIV resolution limitations found, examination of the measurement set reveals that particle images are resolvable closer to the surface than 100 μm. In Fig. 1 an example sequence highlighted the tracking of one particle at a distance of around 60 μm from the surface. Other sequences reveal particles even closer, and some appear to be on the plate surface.

7 2.3 Particle Tracking Velocimetry Post Processing A particle tracking velocimetry approach was used to resolve the particles within 2 mm of the plate surface. This technique is recommended for cases where the seeding density is not high enough to apply particle image velocimetry. In the approach used here, double frame images were analyzed and particles were identified based on their light intensity. The displacement of the individual particle is determined and information about the spatial resolution and the time delay between frames was used to calculate its local velocity. For this imaging setup, the pixel size was 30 μm and the time delay between frames was 45 μs, leading to a particle velocity of m/s for a particle displacement of one pixel between frames. Centroid interpolation (Raffel and Willert, 2007) was used to calculate the displacement of individual particles between frames. Given this sub-pixel resolution, it was possible to resolve particles as close as 33 μm from the plate surface and velocities of less than 0.2 m/s. Results are presented in the following section. 3. Results PTV was used to analyze the velocity components of particles in two distinct images of the image data set taken at 2.5 khz. The images are presented in full in Fig. 7, where both frames of each image are superimposed. Particles in the first frame are shown as blue and particles in the second frame are orange, making it easier to visualize the displacement of the particles. The first image, Image A, was chosen because it presents a somewhat uniform distribution of particles in the region of interest, up to 2 mm above the plate, shown in detail in Fig. 8. It is also possible to observe significant pixel intensity counts near the plate surface, likely due to fluorescent particles that were stuck on the plate surface. Particles with zero pixel displacement were not considered in this analysis due to this ambiguity. The second image to be analyzed, Image B, is also presented in Fig. 7 (bottom). This image generally shows fewer particles in the near surface region (shown in Fig. 9), but it was possible to resolve more particles close to the surface, at distances below 100 μm. There were also fewer particles stuck to the surface of the plate, so it was possible to identify a few particles at a distance of 33 to 35 μm from the plate. With the PTV approach, it was possible to track particles and calculate the local velocity for each one of them, shown as vectors in Figs. 8 and 9. As mentioned above, Image A had more particles in general, but not as many in the near surface region. Image B shows several vectors with very low velocity magnitudes, which are located close to the plate surface, as can be seen in Fig. 9.

8 Fig. 7 Two instantaneous images were selected for the analysis. Image A (top) was taken at the beginning of the data set. Several particles can be observed on the plate. The tip of the plate is located at x = and y = -7.4, approximately. Image B (bottom) was taken almost at the end of the data set. Frames 1 and 2 of each image were superimposed. Particles in frame 1 are represented in blue and particles in frame 2 are represented in orange. Time delay between frames is 45 μs.

9 y [mm] x [mm] 2 m/s Fig. 8 Near surface region of Image A (top) and vectors calculated through particle tracking velocimetry (bottom). Leading edge of flat plate is located at x, y = 0, 0. Blue reference vector indicates a velocity of 2 m/s. y [mm] m/s x [mm] Fig. 9 Near surface region of Image B (top) and vectors calculated through particle tracking velocimetry (bottom). Leading edge of flat plate is located at x, y = 0, 0. Blue reference vector indicates a velocity of 2 m/s. The particle velocity/position data from both images were compiled as in ensembles and presented as scatter plots in Fig. 10. It is expected that the flow close to the surface will have a constant velocity gradient, providing a method for estimating the actual positions of the nearest wall particles. A least-squares fit to the scatter plots indicate that particles are sensed within 35 μm of the surface. Given the findings of Petrosky et al. (2015) for inhomogeneous particle intensities and generally low particle image densities from fluorescence when compared with the corresponding Mie image, the current findings indicate a substantial benefit for employing PTV methods in wind tunnel applications of the Kiton red-doped PSL particles.

10 Distance to the plate, mm Particle Velocity, m/s Distance to the plate, mm Particle Velocity, m/s Fig. 10 Scatterplot of vertical distance from the plate and particle velocity for Image A (Left) and Image B (Right). 4. Conclusions Particle tracking velocimetry was applied to measurements using polystyrene latex particles doped with Kiton Red (KR620) for fluorescent properties. The KR620 dye is less toxic than other types of fluorescent paint, making this a safe option for wind tunnel experiments. Centroid interpolation was used to identify particles and their displacements in a low speed laminar flow over a 1 mm thick flat plate. Particles were successfully tracked in regions as close as 33 μm from the plate surface. These results show that this technique allows for an analysis of particles closer to the surface than achieved with PIV, which could only identify particles and calculate their velocities at distances of around 100 μm to the plate surface, due to constraints imposed by the interrogation window size of the cross-correlation analysis. Future work will consist of refining the PTV algorithm for calculating the local velocity of each particle, mainly in the near surface region, and applying this fluorescent particle technique in flows with higher Reynolds numbers and turbulent characteristics. Acknowledgements The authors acknowledge the support of the NASA ARMD Seedling Fund and National Institute of Aerospace Cooperative Agreement NNL09AA00A for the development of this work. We would also like to thank Brian Petrosky for his contributions to this work through data collection and CAPES for support of Tamara Guimarães.

11 References Adrian R. J. (2005) Twenty Years of Particle Image Velocimetry. Experiments in Fluids, 39: Adrian R. J., Westerweel, J. (2010) Particle Image Velocimetry. Cambridge University Press, Cambridge. Brevis W., Niño Y. and Jirka G. H. (2011) Integrating cross-correlation and relaxation algorithms for particle tracking velocimetry. Experiments in Fluids, 50: Cadel D. R., Shin D., Lowe K. T. (2016) A hybrid technique for laser flare reduction. In: Proceedings of 54th AIAA Aerospace Sciences Meeting. Dracos, T. (1996) Particle Tracking Velocimetry (PTV). Springer, Netherlands. Exciton, INC, "Material Safety Data Sheet - Kiton Red 620," [Online]. Available: Kähler, C. J., Scharnowski, S., and Cierpka, C. (2012) On the uncertainty of digital PIV and PTV near walls. Experiments in Fluids, 52: Petrosky, B. J., Lowe, K. T., Danehy, P. M., Wohl, C. J., Tiemsin, P. I. (2015) Improvements in laser flare removal for particle image velocimetry using fluorescent dye-doped particles. Measurement Science and Technology, 26(11): Raffel, M., Willert, C., Wereley, S., Kompenhans, J. (2007) Particle Image Velocimetry: A Practical Guide. Springer Berlin Heidelberg. Scarano, F., Riethmuller, M. (2000) Advances in iterative multigrid PIV image processing. Experiments in Fluids, 29(7):S51 S60. ScienceLab.com, Inc., "Material Safety Data Sheet- Rhodamine B," [Online]. Available: ScienceLab.com, Inc., "Material Safety Data Sheet- Rhodamine 6G," [Online]. Available: Tauro, F., Mocio, G., Rapiti, E., Grimaldi, S., and Porfiri, M. (2012) Assessment of fluorescent particles for surface flow analysis. Sensors, 12: Tiemsin, P. I., and Wohl, C. J. (2012) Refined Synthesis and Characterization of Controlled Diameter, Narrow Size Distribution Microparticles for Aerospace Research Applications. NASA Technical Memorandum TM