PIV operation in hydrodynamic facilities, a flat plate benchmark test comparing PIV systems at...

15th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 05-08 July, 2010

- 1 -

PIV operation in hydrodynamic facilities, a flat plate benchmark test comparing

PIV systems at different facilities.

Chittiappa Muthanna1, Fabio Di Felice2, Michiel Verhulst3, Rene Delfos4, Jean-Paul Borleteau5

1: Department of Offshore Hydrodynamics, MARINTEK, Trondheim, Norway, [email protected]

2: INSEAN, ( Italian Ship Model Basin) Rome, Italy, [email protected] 3: Maritime Research Institute Netherlands (MARIN) Wageningen, The Netherlands, [email protected] 4:Laboratory for Aero and Hydrodynamics, 3ME Dept., Delft University of Technology, the Netherlands,

[email protected] 5: SIREHNA, Nantes, France, [email protected]

Abstract As part of the European Network of Excellence Hydro Testing Alliance (HTA) one research activity named as JRP1 has been defined to investigate and develop new and improved methods for PIV operation in hydrodynamic facilities. The group has been working on a review of potential applications and needs from hydrodynamic facilities, then selecting targeted applications for development of solutions (especially reviewing needed adaptations on experimental set up), and ultimately conducting tests in facilities. In particular, a benchmarking program has been established by considering the wake flow of a piercing surface flat plate at incidence.. The objective was to realize a benchmark program to evaluate the performance of the SPIV (Stereo PIV) systems to be used among the JRP-1 partners with the main requirements being a simple and cheap experimental setup that would be representative of typical SPIV problems in commercial hydrodynamic facilities such as towing tanks and cavitation tunnels. Analysis and comparisons of the mean flow fields between three different institutions revealed that the PIV technique is fairly robust and reliable under ideal measurement conditions. One significant outcome from the comparisons is the weakness of the technique with respect to surface light reflections. Reduced laser energy and larger particle could help in increasing the image Signal to Noise Ratio. However when the scattering angle from the model surface is matching one of the camera angle view there is no way to overcome the problem and a measurement is not possible. In such case special model preparation techniques such as a Perspex insert is required to overcome the problem. Despite the fact that the same model was used in three different facilities, there were some differences in the flow field, primarily with respect to the location of various flow features. Whether this is a result of different flow fields being obtained, or a result of how the data is interpolated onto a specified measurement grid is difficult to conclude. 1. Introduction

When studying fluid flow phenomena, particularly time evolving flows, it is desirable to obtain whole field time varying measurements. Particle Image Velocimetry (PIV) is a whole field measurement technique that is able to measure time varying fluid flows, and is thus ideally suited to studying time evolving flows, Raffel et al. (1998), such as those found behind marine structures such as aquaculture cages, pipelines and risers, and the wakes and flow around ship hulls and bluff bodies. Measurements made in traditional towing tanks have mainly concentrated on point based techniques that are robust enough to use i.e. hot film sensors, pitot-probes, and visualizations using video recordings. The use of optical based techniques has proved difficult to accomplish due to the complications that arise with the presence of open water that is found in towing tanks. It goes without saying that PIV has traditionally been a very complicated and time consuming technique that involves accurate adjustments of the optics involved. Extrapolating this technique to be used


- 2 -

in towing tanks presents considerable challenges in designing and operating a system that is to be used underwater, Muthanna et al. (2008).

As part of the European Network of Excellence Hydro Testing Alliance (HTA) one research activity named as JRP1 has been defined to investigate and develop new and improved methods for PIV operation in hydrodynamic facilities. The group has been working on a review of potential applications and needs from hydrodynamic facilities, then selecting targeted applications for development of solutions (especially reviewing needed adaptations on experimental set up), and ultimately conducting tests in facilities. In particular, a benchmarking program has been established by considering the wake flow of a piercing surface flat plate at incidence. The present paper will present a sample of results of the above mentioned benchmarking tests.

The objective was to realize a benchmark program to evaluate the performance of the SPIV (Stereo PIV) systems to be used among the JRP-1 partners. The main requirements were:

- Simple and cheap experimental setup to be used during any test campaign in the measurement facility.

- Detailed specifications to assure a highly repeatable test amongst partners. - Test setup in about 1 hour. - Measurements performed in 1 or 2 carriage runs. - Test case representative of typical SPIV problems in towing tank like surface model

reflections and presence of air bubbles. - Possibility to exchange and compare images and velocity data.

2. Benchmark Flat Plate Model

The test case is a piercing surface flat plate at incidence as seen in Figure 1. It is a simple and significant test case with the presence of high gradients (Tip Vortex), surface effects, bubbles nearby the free surface regions, reflections from the model surface. The plate is a steel rectangular plate of L= 500 mm x H= 800 mm. The thickness is 0.25 inches (6.35 mm). Leading edge and trailing edge (on the 800 mm side) of the plate have a round shape of 0.125 inches radius (3.175 mm) or semi-circular ends of 0.25 inch diameter).

Figure 1: Benchmark Flat Plate Model in the INSEAN towing tank.

The detailed specifications of the benchmarking program are available on the website of HTA Network of Excellence (www.hta-noe.eu) on the page ‘Research/JRP1’. 3. PIV Measurements

The plate was immersed vertically with the tip 300 mm deep with respect to the free surface, and at an incidence of 20° with respect to the flow direction. SPIV measurements were performed on 2 vertical planes near the submerged tip of the rectangular plate and on 1 undisturbed flow vertical plane. The test velocity is U∞=0.4 m/s. The measurement area was rectangular (at least


- 3 -

200 mm high x 300 mm wide) and is located on the suction face of the incident plate. The first plane (P1) is 100 mm upstream respect the trailing edge of the tip of the plate. The second plane (P2) is 200 mm downstream respect the first one. This is illustrated in Figure 2.

The two measurement planes have been chosen because they are representative of two typical situations when using SPIV. The P1 plane is strongly affected by model reflections and large errors are expected. It is the worst case application, representing near proximity model surface measurement applications. Results could depend on a variety of factors i.e. the adopted optical setup, the particle seeding, the surface treatment of the model, etc. The P2 plane is located in the wake of the plate, such that the problems present in P1 are no longer present, and it is representative of an ideal case application of the measurement technique in the free wake of the model. Errors and differences between different contribution are expected to be small

Figure 2: The two test measurement planes. P1 is located on the plate, and P2 is located downstream of the

plate, with the flow from left to right.

4. Discussion of the PIV Data Sets

As detailed in previous sections, stereoscopic PIV measurements of the velocity field behind a flat plate were performed. For all three cases, the same model was used in the measurements. The three data sets that are compared here are from

- INSEAN (Italian Ship Model Basin), - Laboratory for Aero & Hydrodynamics at the Delft University of Technology (TUD). - - Maritime Research Institute Netherlands (MARIN) (cooperation MARIN-

SIREHNA) Among the results discussed here, INSEAN and MARIN performed the benchmark tests in

their respective towing tanks, and TUD performed the tests in their circulating water tunnel. INSEAN have used their PIV system developed in collaboration with TSI. MARIN has used a PIV system developed by Dantec Dynamics, and operated by SIREHNA. TUD has used a custom PIV solution using the DAVIS analysis software. Thus, it can be expected that there will be some differences in the results obtained. The INSEAN configuration was a asymmetrical 3-Component PIV setup as shown in Figure 1. MARIN also used an asymmetric setup in the towing tank as shown in Figure 3. A sketch of the configuration in the water tunnel at Delft is shown in Figure 4.


- 4 -

Figure 3: Sketch of MARIN’s stereoPIV configuration.

Figure 4: Sketch of TUD’s stereoPIV configuration in a circulating water tunnel.

The velocity data was delivered in ASCII format giving the measurement grid in X, Y, Z, and the three velocity components U, V, and W. The data sets being compared here are the average velocity maps as computed by each individual institution's averaging algorithms. The data is presented on the interpolated grid as specified, and again, the interpolation routines were chosen by each institution. It should be noted that the data between the three institutions was not consistent with each other in terms of co-ordinate definitions/directions, but this is not difficult to correct when the flow structure is known ahead of time. The results presented here have been corrected so that comparisons can be made between like data sets.

Of particular interest of this exercise was to see the effectiveness of performing stereoscopic PIV in towing tank facilities. Calibration is particularly difficult in terms of accuracy, and this is especially true when looking at the out-of-plane component of the velocity (U velocity in this situation). While only two towing tank facilities were able to perform the measurements at this time, comparisons are also made with results from a circulating water tunnel where the fundamental concept is similar, but the execution slightly different i.e. rather than a body moving through the fluid, the fluid is moving over the body.

4.1 P1 Measurement Plane

Shown in Figures 5, 6, and 7 are the mean velocity contours of the U, V, and W components respectively for the P1 measurement plane. This plane would be most affected by the presence of the flat plate, due to reflections from the plate surface, and thus impact the overall image quality of the PIV measurement. Depending on the setup, the results should show fairly significant differences and they do between the INSEAN and MARIN data sets. The TUD data was generated in a circulating water tunnel, when compared to the towing tank measurements, placement and adjustment of the PIV hardware, is relatively speaking, easier.


- 5 -

The out-of-plane (i.e. streamwise) component of the velocity, U, is shown in Figure 5. This measurement result is the most sensitive to the setup of the Stereo PIV hardware in terms of making an accurate measurement. While the MARIN data sets show the presence of a large velocity region deficit near the flat plate, the INSEAN data does not. This same region is visible in the TUD data, but it should be noted that the position of the flat plate in the TUD data set seems to be considerably different from that in the two towing tanks (likely relating to coordinate system definition).

Figure 5: U velocity contours at the P1 plane

However, when comparing the in-plane velocity measurement, V (spanwise, or parallel to the free surface, Figure 6), the three data sets are very similar, showing similar values for the measured velocities, as well as the same flow structures with the exception of the INSEAN data, whose results seem to be affected more than the others at this measurement location. However, the general trend of the INSEAN data indicates a similar flow structure as that seen in the other two data sets.

Figure 6: V velocity contours at the P1 plane

The second in-plane velocity measurement, W (normal to the free surface, Figure 7), also shows a similar consistency in values measured in the three data sets, but here there is a difference in the overall flow structure. Again, the INSEAN data seems to be affected the most, and does indicate a slightly different flow structure near the flat plat region.


- 6 -

Figure 7: W velocity contours at the P1 plane

4.2 Raw PIV Images The variation in results at the P1 plane indicates that this plane is very sensitive to the PIV

setup. Shown in Figure 8 are representative raw images from the PIV cameras during the measurements for the INSEAN, MARIN and TUD experiments respectively. It is expected that the measurements would be most affected by the presence of reflections from the flat plate model surface, and the results of the INSEAN measurements in Figures 5, 6, and 7 indicate that this would be the case.

This is supported by the images shown in Figure 8a and 8b, where a very bright area indicative of the laser light reflecting off the surface is apparent and thus affecting the PIV measurement in such a way that it is not possible to obtain good velocity vectors there. Both the MARIN and TUD images (Figure 8c,d and 8e,f) show that the measurements were setup to handle surface reflections a bit better than the INSEAN setup in which the camera 1 is almost matching the reflection angle of light sheet on the plate.. The TUD measurement images show almost no surface reflection.

a. INSEAN Camera 1 b. INSEAN Camera 2


- 7 -

c.MARIN Camera 1 d. MARIN Camera 2

e.TUD Camera 1 f. TUD Camera 2

Figure 8: Sample RAW images from the PIV measurements for each institution at the P1 measurement plane. Note that the INSEAN images are upside down when compared to the other two facility images.

Based on this observation of the raw images, the differences in the results at the P1 measurement plane between INSEAN and the other two laboratories are easier to understand, than just based on differences in the PIV optical setup. 4.3 P2 Measurement Plane

The P2 measurement plane is downstream of the flat plate model (see Figure 2) and the PIV images will not be affected by reflections of the laser light from the model. This would lead us to believe that the results from this measurement plane should be consistent for the three measurements as the flow field will be the same in all three cases, with perhaps a slight difference in the TUD measurement on account of it being done in a relatively narrow cross-section water tunnel.

Sample raw images from one camera are shown in Figure 9 for each of the three institutions. There are no spurious reflections from the models present in the images, with most of the image containing the particles illuminated by the laser light. There is a difference in intensities between the three runs, and what is noticeable is that the MARIN particle images are brighter than the other two due to the larger dimensioned seeding particles. The particle field in the TUD measurement is more consistent, which is to be expected owing to the fact that a nearby laser with optics can be used in their closed water tunnel. However, the results between the three measurements should be fairly consistent both qualitatively and quantitatively.


- 8 -

INSEAN MARIN TUD

Figure 9: Sample RAW images from the PIV measurements for each institution at the P2 measurement plane.

The mean velocity fields, U, V, and W at the P2 measurement plane are shown in Figures 10, 11, and 12 respectively. The results show that at least qualitatively, the results are similar between all the different measurements. The figures all show the presence of the tip vortex in the U velocity contours. The V, and W velocity contours are very similar in their distribution and values between the three measurements.

Figure 10: U velocity contours at the P2 plane


- 9 -

Figure 11: V velocity contours at the P2 plane

Figure 12: W velocity contours at the P2 plane

Based on the results in Figure 10, the vortex centers were identified and the locations as identified are given in the Table 1.

Table 1: Location of Vortex Centers Y, mm Z, mm

INSEAN 65 30 MARIN 75 30

TUD 35 -25 What is immediately obvious is that the vortex centre location for the TUD case is very different from the others when comparing the location, whereas the two towing tank measurements are very close to each other. As stated previously, this is probably due to the way TUD has defined their measurement locations; and possibly from the flow confinement in the circulating tunnel.

From these locations, profiles in the Y, and Z direction can thus be extrapolated from the data set to compare the results quantitatively. Shown in Figure 12 are the U velocity variations with Z (Fig 12a.) and with Y (Fig 12b.) through the vortex centre. The TUD and MARIN measurements


- 10 -

are very similar in magnitude, with the INSEAN measurements again 50% higher in the vortex centre. All the facilities have used a target based calibration system, so it is rather difficult to explain the differences. However, the out-of-plane component is most sensitive to the PIV correlation, so differences can be expected.

a. U velocity variation with Z b. U velocity variation with Y

Figure 13: U velocity variation through the vortex centre at the P2 measurement plane

Comparing the V velocity variation in the Z-direction (Figure 14), the results look much more similar qualitatively. There are slight variations, but in general, the results are very similar near the vortex centre. Here we see that away from the vortex centre at negative displacements there is more of a discrepancy between the results than at positive displacements. At positive displacements, the two towing tank measurements are similar.

Figure 14: V velocity variation with Z through the vortex centre at the P2 measurement plane

Comparing the W-velocity variations in the Y-direction (Figure 15), we again start to see differences in the results. Here the TUD results show a different shape to the profile as can be seen at negative displacements. Around the vortex centre, the TUD and MARIN results are alike, but as the profiles extend away, the MARIN and INSEAN results start to become similar.


- 11 -

Figure 15: W velocity variation with Y through the vortex centre at the P2 measurement plane

The results in the P2 plane show that the results are very similar qualitatively, but when it comes to quantifying the results, there are some major differences. Each of the experiments utilized a different PIV post processing configuration, and had different calibration routines, and to see such differences, one could only hypothesize that the differences are due to calibration and post-processing routines. 5. Concluding Remarks

The results of the mean flow field for the PIV data show that in general, the results obtained from a PIV measurement are consistent qualitatively. The overall flow structure is similar in all the cases, with any differences attributed to a fundamental change in the laboratory or measurement technique. However, quantitatively, there still seems to be some variations in the values being obtained.

Analysis and comparisons of the mean flow fields between three different institutions revealed that the PIV technique is fairly robust and reliable when working under ideal conditions. Despite the fact that the same model was used in three different facilities, there were some differences in the flow field, primarily with respect to the location of various flow features. Whether this is a result of different flow fields being obtained, or a result of how the data is interpolated onto a specified measurement grid is difficult to conclude. The measurement grids are highly dependent on the calibration of the image planes, and thus could contribute to the difference in position. However, from the delivered data sets, it is difficult to accurately pinpoint why this might be the case. A full analysis of the raw images in different PIV processing programs might be able to give some insight as to where the differences come from.

One significant outcome from the comparisons is the weakness of the technique with respect surface light reflections. This is a practical problem in towing tank application when measurements in close proximity to the model surface and standard model preparation techniques are used. Reduced laser energy and larger particle could help in increasing the image Signal to Noise Ratio. However when the scattering angle from the model surface is matching one of the camera angle view as seen in the P1 plane there is no way to overcome the problem and a measurement is not possible. In such case special model preparation techniques such as a Perspex insert is required to overcome the problem.


- 12 -

6. Acknowledgements This research work was supported by the EC project Hydro-Testing Alliance under the Joint

Research Program JRP1 "PIV operation in hydrodynamic facilities". Hydro-Testing Alliance is the European Network of Excellence to facilitate the continuation of world leadership of the European Hydrodynamic testing facilities. HTA is supported with funding from the European Commission's Sixth Framework Program under DG Research, project number 031316. The Network of Excellence started on the 1st of September 2006 and it will have a duration of five years. 7. References [1] Raffel, M., Willert, C., and Kompenhans, J., 1998, “Particle Image Velocimetry, A Practical Guide” corrected 3rd printing, Springer. [2] Muthanna, C., Visscher, J. H., and Pettersen, B., 2008, “Investigating Fluid Flow Phenomena behind Intersecting and Tapered Cylinders using submerged Stereoscopic PIV”, 14th Int Symp on Applications of Laser Techniques to Fluid Mechanics Lisbon, Portugal, 07-10 July, 2008.

Date post:	29-Nov-2023
Category:	Documents
Upload:	independent
View:	0 times
Download:	0 times

PIV operation in hydrodynamic facilities, a flat plate benchmark test comparing PIV systems at...

Documents