Published in the proceedings of The Japanese Symposium on Shock Waves,March 19-21, 1999, Aoyama Gakuin University, Shibuya 4-4-25, Shibuya-ku, Tokyo 150-0002, Japan

Two-dimensional velocity-field measurement in a hypersonicgas flow over a cone using planar laser-induced fluorescence

P. MERE, P.M. DANEHY, S. O'BYRNE, P.C. PALMA, M.J. GASTON, and A.F.P. HOUWING

Physics Department, Faculty of Science, The Australian National University, Canberra, ACT, 0200

AbstractPlanar laser-induced fluorescence (PLIF) of nitric oxide is used to measure a component of the velocity field for

the Mach 7 flow around a cone. Velocities are measured in the freestream, shock layer, and the separated regionbehind the cone. The measurements are compared with computational fluid dynamics (CFD) codes. Thecomparison indicates that the CFD correctly predicts the flow over the forebody and in the expansion around thecone shoulder, but incorrectly predicts the location of the re-attachment shock.

1. Introduction

The ability to accurately model hypersonic gas flows in a controllable environment can provide valuable information fordesigning hypersonic vehicles. Consequently, the development of accurate measurement techniques for use in shocktunnels holds considerable interest for the aerospace community.

We are interested in making non-intrusive velocity measurements in shock-layer flows produced in free-piston shocktunnels. Because the flow is supersonic, physical probes such as hot-wire anemometers are inappropriate. Furthermore,imaging methods are preferred to single-point methods such as laser Doppler velocimetry (LDV) because the limited testtime of the facility would make it very expensive to map the flow velocity.

Several laser-based methods have been developed for mapping the velocity in gaseous flows. These include particleimage velocimetry, Rayleigh scattering velocimetry[1], and planar laser-induced fluorescence (PLIF) velocimetry[2].Particle image velocimetry is unattractive in supersonic flows because the particles do not always follow the flow whensteep velocity gradients, such as those resulting from oblique shock waves, are present. Rayleigh scattering velocimetryis a sensitive method that was considered carefully in the present study. However, preliminary measurements in ourfacility showed that Mie scattering (from particles) overwhelmed Rayleigh scattering (from molecules) by over an orderof magnitude. Without a way to separate the Mie and Rayleigh signals in our high-speed flow, we abandon thatapproach in favour of PLIF velocimetry. While PLIF has a lower sensitivity to velocity compared to Rayleighscattering velocimetry, it has been used previously in our facility providing excellent signal-to-noise ratios. In thispaper, we acquire PLIF images at multiple laser frequencies near an isolated spectroscopic transition and then process theimages to determine the flow velocity from the observed Doppler shift.

2. Theory

We have demonstrated this velocimetry method on the supersonic flow over a cone, illustrated schematically in Figure1. The cone simulates an aerospace vehicle carrying a payload, which is represented by the "sting" (the shaft holdingthe cone). Gas approaching the model in the uniform free stream passes through a shock wave which turns the gasaway from the cone, while increasing its temperature and density, and reducing its speed. When the gas reaches theexpansion fan, it is accelerated back towards the flow axis, and its temperature and density are reduced. Due to theinertia of the flow, the gas from the shock layer cannot sharply turn around the shoulder of the corner. Instead, the flowseparates from the model and re-attaches on the sting further downstream at the re-attachment point.

Velocity is an important parameter to measure in this flow. The velocity field provides information about thedistribution of kinetic energy of the flow. We plan to complement this data set with temperature measurements thatwill describe distribution of the thermal energy of the flowfield. Also, velocity measurements provide informationabout the slight divergence of the freestream gas which is important for CFD modelling of the flowfield. An importantdimension of this flow is the distance of the re-attachment point from the base of the cone in relation to the height ofthe cone step. This is an important parameter for designers of re-entry vehicles, as the heat of the flow can damageinsufficiently shielded payloads placed at the re-attachment point.

Published in the proceedings of The Japanese Symposium on Shock Waves,March 19-21, 1999, Aoyama Gakuin University, Shibuya 4-4-25, Shibuya-ku, Tokyo 150-0002, Japan

We use PLIF to measure the velocity. This method is now a well-established flow imaging technique for measuringvelocity as well as temperature and pressure[2,3]. For several years, we have used PLIF in hypersonic shock layer,supersonic combustion and hypersonic mixing studies[4]. The PLIF method uses a pulsed, tuneable laser to generate abeam that is collimated into a thin sheet which illuminates a planar cross-section of the flowfield. This light is absorbedby some of the molecules in the flow and the resulting fluorescence is detected by an intensified digital camera providingvisualisation and quantitative measurements of the flowfield in the cross-section.

In the present experiment, we obtain PLIF images on 30 successive runs of the shock tunnel. Each image isobtained with a slightly different laser frequency near a nitric oxide absorption transition. Through image processing,we obtain a measure of the absorption lineshape at each point in the flow. Locating the centre of the transition allowsus to determine the sum of the pressure and Doppler shifts at each point in the image. As shown in Figure 2, we alignthe laser sheet so that it passes just beneath the sting and continues on through the flow on the opposite side of themodel. Due to the axial symmetry of the model and flow, the flow on one side of the sting is the mirror image of thaton the other side – with the exception of the Doppler shift in respect to the laser. The Doppler shift producesasymmetric PLIF images of an otherwise symmetric flowfield. By assuming flow symmetry during image processing,we extract the Doppler shift from the image (the pressure shift and all other flow parameters are assumed to be identicalacross the axis of symmetry). This Doppler shift is then used to determine the component of the flow velocity in thedirection of the laser.

3. Experiment

The experiments were performed on the T2 free-piston shock tunnel at the Australian National University. The T2nozzle has a 15o full-angle conical geometry with a 6.4-mm diameter throat and an exit-to-throat area ratio of 144. Theshock tube gas was a mixture of 98.9% N2, 1.1% O2. This mixture resulted in 98.1% N2, 1.1% NO, 0.4% O2, and0.3% O in the test section, according to computations.[5] The primary shock speed was 2.4 km/s, which correspondsto a flow enthalpy of 5.3 MJ/kg. The nozzle-reservoir pressure was measured to be 27.9 MPa and the calculatedreservoir temperature was 4219 K[5]. At a distance of 285 mm from the nozzle throat, the calculated freestreamtemperature, pressure, Mach number and velocity were 396K, 4.4 kPa, 7.7, and 3.0 km/s respectively.[5]

Figure 1. Hypersonic gas flow over a cone.

Figure 2. Orientation of the laser sheet as viewed from below.

Published in the proceedings of The Japanese Symposium on Shock Waves,March 19-21, 1999, Aoyama Gakuin University, Shibuya 4-4-25, Shibuya-ku, Tokyo 150-0002, Japan

A schematic of the experiment is shown in Reference 5. We frequency-doubled the output of an excimer-pumpeddye laser to obtain 5-mJ, 25-ns pulses at 225 nm; coinciding with the (0,0) vibrational band of the A–X electronictransition of NO. Most of the laser light was formed into a sheet 8 cm wide and was directed into the test sectionperpendicular to the flow. The laser sheet was measured to be ~0.8 mm thick. Less than 1 mJ was used to form thelaser sheet, resulting in low intensity which reduced power broadening of the transitions. A small portion of the laserbeam was split off and used for wavelength calibration by performing LIF in a gas cell filled with a small quantity ofNO together with a few kPa of N2. Previously, the laser linewidth was measured to be 0.18±0.01 cm-1 based on LIFmeasurements in the same gas cell.[5] This is consistent with specifications from the manufacturer.

We adjusted the laser frequency to a specified tuning near an NO transition prior to each run of the shock tunnel.Immediately before the shot (<2 sec), the tunnel operator stopped the laser via a remote switch next to the firing valve.After the firing valve was opened, the nozzle reservoir pressure transducer detected the shock reflection at the end of theshock tube and the laser fired 350 µs later. This delay was chosen to coincide with the period of steady flow in theshock tube. An intensified CCD camera (Princeton Instruments, 16-bit CCD, 576 by 384 pixels, ~50-ns gating period)captured the fluorescence image at right angles to the laser sheet. An important concern guiding the design of the experiment was the signal-to-noise ratio in the images. By far thelargest source of this noise was the presence of flow luminosity in each image. To reduce this source of noise, a 2-mmthick UG-5 filter was placed in front of the ICCD camera. This filter allowed most of the LIF signal to pass into thecamera, but cut off most of the scattered laser light and some of the flow luminosity. The filter also reduced influenceof radiative trapping.

We probed the R2(20.5) transition of NO. This transition was chosen for its isolation from other transitions and itsappreciable ground-state population for all temperatures expected in the experiment. Absorption of the laser sheet on itspassage through the flow was ignored in the analysis. The maximum level of absorption was around 8% cm-1 in theshock layer when the laser is tuned to transition line centre, and significantly lower elsewhere. Thus, absorption had aminimal impact on the measurement.

4. Analysis and Results

Thirty experimental PLIF images were obtained. The raw images were corrected for camera offset (dark current); residualscattered laser light and flow luminosity; spatial variations in the laser-beam profile; laser intensity; and small cameratranslations. In order to reduce the noise in the calculated velocities and to increase the speed of the analysis, the pixelsof the processed images were binned 3 by 3. Then, the spectral shift at each point in the flow was computed. Becausethe shape of the absorption profile varied between a Gaussian and a Lorentzian throughout the flowfield, both Gaussianand Lorentzian line shapes were fitted to the data to determine the line centre. The final estimation of line centre wasfound by taking the weighted average of the predictions made by the two fits. Figure 3 shows the spectrum acquired at atypical pixel in the shock layer.

Figure 4 shows the resulting velocity map. The uncertainty in the measured velocities is approximately ±150 m/s.The flow is from left to right and the laser sheet enters at the top of the image. The top half of the figure shows theexperimental velocity map; the bottom half shows the corresponding laminar computational fluid dynamics (CFD)velocity map, computed using the commercial program CFD-ACE. In both cases, the velocity shown is thecomponent of the velocity in the direction of the laser sheet. Far from the centreline, this component is nearly equal tothe radial velocity.

-0.6 -0.4 -0.2 0 0.2 0.4Tuning from transition line centre (cm^-1)

GaussianLorentzianData

Figure 3. Spectrum of LIF for a pixel located in the shock layer.

Published in the proceedings of The Japanese Symposium on Shock Waves,March 19-21, 1999, Aoyama Gakuin University, Shibuya 4-4-25, Shibuya-ku, Tokyo 150-0002, Japan

Figure 4. Comparison of the experimental (top) and theoretical (bottom) velocity maps.

The measurements show that the freestream flow is diverging (velocity increasing away from the centre of theimage, which is the axis of symmetry). This flow divergence is caused by using a conical instead of a contoured nozzle.The conical shock wave is clearly visible. As expected, inside the shock, gas near the flow centreline has a lowervelocity, while further away from the centreline the gas velocity increases. The maximum velocity is achieved inoutermost part of the shocklayer, where the velocity has a significant radial component even before it passes through theshock. As the gas expands around the shoulder of the cone, it achieves a negative velocity relative to the laser sheet.

The computed and measured velocity maps agree well regarding the structure of the flow and the overall trends in thevelocity field. The most important discrepancy between the two images is in the location of the re-attachment point.The experiment shows that the flow re-attaches at approximately twice the distance from the base of the cone comparedto calculation. Clearly this CFD analysis has inaccurately computed the size of the separated flow region.

5. Conclusions

We have demonstrated PLIF velocity measurements in a free-piston shock tunnel for the first time. The method used isrobust and reliable but it is also time consuming and requires flow symmetry. A comparison of the experimentalvelocity map with a computation showed good overall agreement, but the computation incorrectly predicted thereattachment point of the separated flow.

References:

[1] Miles, RB and Lempert W (1990). Two-dimensional measurement of density, velocity, and temperature in

turbulent high-speed air flows by UV Rayleigh scattering. Appl. Phys. B. 51: 1-7.[2] Hiller B, Hanson RK (1988), Simultaneous planar measurements of velocity and pressure fields in gas flows using

laser-induced fluorescence, Appl. Opt., Vol. 27, 33-48; Paul PH, Lee MP, Hanson RK (1989), Molecularvelocity imaging of supersonic flows using pulsed planar laser-induced fluorescence of NO, Opt. Lett., Vol. 14,417-419.

[3] McMillin BK, Palmer JL and Hanson RK (1993), Temporally resolved, two-line fluorescence imaging of NO in atransverse jet in a supersonic cross-flow, Appl. Opt., 32, 7532

[4] McIntyre TJ, Houwing AFP, Palma PC, Rabbath P, Fox JS (1997), Imaging of combustion in supersoniccombustion ramjet, Journal of Propulsion and Power, 13(3), 388-394; Palma PC, McIntyre TJ, Houwing AFP(1998), PLIF thermometry in shock tunnel flows using a Raman-shifted tunable excimer laser, Shock WavesJournal Vol. 8, No. 5, pp 275-284.

[5] Palma PC, (1999) Laser-induced fluorescence imaging in free-piston shock tunnels, PhD Thesis, Physics Dept.,The Australian National University. (http://www.anu.edu.au/Physics/aldir/publications/Palma_PhD_1999.pdf)