Measurements of hypersonic boundary-layer instabilities using a pulsed-laser schlieren technique

Author: Stuart Laurence, Department of Aerospace Engineering, University of Maryland, College Park

1 Introduction

When a hypersonic vehicle travels through the atmosphere, a boundary layer develops in the air close to the vehicle surface. Initially (close to the nose of the vehicle) this boundary layer is laminar, but typically will transition to turbulence at some point downstream. A turbulent boundary layer produces significantly larger heat flux and frictional drag at the vehicle surface than a laminar one, so to be able to accurately predict vehicle performance, knowledge of the laminar-to-turbulent transition process is important. There are a variety of boundary-layer instabilities whose growth and breakdown can lead to transition; for slender planar or axisymmetric bodies and small incidences, a key instability mechanism is the second or Mack mode, which can be thought of physically as acoustic waves that become trapped within the boundary layer. Second-mode waves typically exhibit very high frequencies – around 100 kHz or even higher – which makes their measurement very difficult with conventional techniques. Here I describe measurements of the second-mode instability using a schlieren system incorporating a CAVILUX pulsed-diode laser.

2 Experimental configuration

Experiments were performed at two hypersonic wind tunnels: the High Enthalpy Shock Tunnel Goettingen (HEG) of the German Aerospace Center (DLR), and Hypervelocity Tunnel 9 of the Arnold Engineering Development Center (AEDC) at White Oak, Maryland. HEG is capable of reproducing the extremely high flow velocities typical of atmospheric reentry (up to 7 km/s), though for very short test periods (~1 ms). In the present experiments, the flow velocity and density were 4.4 km/s and 0.0175 kg/m3. Tunnel 9, on the other hand, can produce high Mach numbers with longer test periods (around 1 s), but with lower flow velocities. In the present Tunnel 9 experiments, a variety of flow conditions were used, all having a Mach number of approximately 14 and a flow speed of 2 km/s.

In both cases the test article was a slender, 7° half-angle cone. The flow within the boundary layer over the cone was visualized using a conventional Z-fold schlieren arrangement, as shown in figure 1.

Figure 1: Z-fold schlieren visualization set-up used in the experiments described here.

Schlieren is a technique used to visualize flow features in compressible flows: a density gradient at some location in the imaging plane within the test section (in a direction normal to the knife edge placed in front of the camera) will result in a change in intensity at the corresponding location on the image taken by the imaging device (in this case a high-speed camera). In the HEG experiments the light source was a CAVILUX Smart pulsed-diode laser and the camera was a Vision Research Phantom v1210, recording at 200 kHz. The laser was run in ultra-high-speed mode, with a repeated 4-pulse pattern as shown in figure 2.

Figure 2: Laser pulse pattern used in the HEG experiments: Δt12 is 2 μs and Δt34 is 3 μs.

This pattern was necessary because the characteristic frequency of the second-mode disturbance in this case (~600 kHz) was significantly higher than the recording frequency, so closely spaced pulse pairs were used to unambiguously resolve the wave motion (further details to be provided shortly). In the Tunnel 9 experiments, a CAVILUX HF laser providing, uniformly spaced pulses at approximately 70 kHz, was used together with a Phantom v2512 camera. The laser pulse width in the two experiments varied between 20 ns and 50 ns; such short pulse widths were necessary to freeze the high-speed flow structures in images.

3 Results

The HEG experiments were particularly challenging because of the high flow velocity (meaning high second-mode frequencies) and low density (meaning weak intensity variations in the schlieren images). An example of a visualized second-mode wave packet, visible from its oblique “rope-like” structures close to the surface, is shown as it propagates within a sequence of schlieren images in figure 3. The propagation speed of the wave packet is constant – the apparently uneven motion is a result of the laser pulse pattern. By performing two-dimensional image correlations, it is possible to recover the propagation speed – in this case it is 3.8 km/s. The unequal spacing between the two pulse pairs in figure 2 avoids problems with aliasing in these correlations. By then taking the Fourier transform of rows of pixels parallel to the cone surface, wavenumber spectra can be constructed; these can subsequently be converted into frequency spectra using the propagation speed calculated earlier.

Figure 3: Sequence of reference-subtracted schlieren images showing the propagation of a second-mode wave packet (flow is left to right)

Plots of the averaged power spectral density (PSD) at three locations downstream are shown in the left plot of figure 4. Here we see a strong peak at approximately 600 kHz – this corresponds to the second-mode frequency at these conditions. The peak grows rapidly as we move downstream, showing strong amplification of the second mode. A more detailed picture of this growth is shown in the right plot of figure 4, which is a contour plot of the PSD versus distance downstream. Further details of these measurements can be found in Laurence et al. (2016).

Figure 4: (Left) Plots of the schlieren power spectral density (PSD) near the surface at three locations downstream (s is the distance along the cone from the nose); (right) contour plot of the PSD versus distance downstream.

An example of a propagating second-mode wave packet in one of the Tunnel 9 experiments is shown in figure 5. Again we see the characteristic “rope-like” structures, though now the disturbance energy appears to be less concentrated towards the cone surface than it was in the HEG experiments.

Figure 5: Propagation of a second-mode wave packet in a Tunnel 9 experiment

In the Tunnel 9 experiments, the schlieren system was calibrated by placing a long-focal-length lens in the imaging plane and recording images of it. This enabled a calibration curve relating image intensity to the density gradient to be established. From this calibration curve, one can then quantify the growth rate of the second-mode instability. A contour plot of the integrated growth rate, or N-factor, versus distance downstream and frequency is shown in figure 6.

Figure 6: Contour plot of N-factor versus distance downstream and frequency in Tunnel 9 experiment.

Again we see a strong second-mode contribution, but now at a much lower frequency of approximately 100 kHz. The decrease in this frequency moving downstream is associated with the thickening of the mean boundary layer. Such quantitative measurements are very important as they provide data against which numerical simulations and stability analysis computations can be compared. Further details of the Tunnel 9 experiments can be found in Kennedy et al. (2017).

4 Conclusions

The experiments described here demonstrate that it is possible to use high-speed schlieren techniques to perform quantitative measurements of extremely high-frequency instability waves in hypersonic boundary layers. The capabilities of CAVILUX pulsed-diode laser light sources proved instrumental in enabling these measurements.

5 References

Kennedy, R., Laurence, S., Smith, M., and Marineau, E. (2017), “Hypersonic Boundary-Layer Transition Features from High-Speed Schlieren Images”, 55th AIAA Aerospace Sciences Meeting, AIAA SciTech Forum, AIAA Paper 2017-1683

Laurence S., Wagner, A., and Hannemann, K. (2016), “Experimental study of second-mode instability growth and breakdown in a hypersonic boundary layer using high-speed schlieren visualization”, Journal of Fluid Mechanics, vol. 797, pp. 471-503

About the author

Stuart Laurence (Ph.D) completed his graduate studies at the Graduate Aeronautical Laboratories, California Institute of Technology, in the area of hypersonic flows. He currently is Assistant Professor at the Department of Aerospace Engineering, University of Maryland, College Park

Send me more information

Leave us your contact information and our expert will get back to you.

If you wish to further specify your question you can fill in more detailed information by using our general contact form.

Ignition of Rocket

Combustion of rocket

Visualization of the launch of a combustion rocket with Cavitar’s CAVILUX illumination laser. The video was taken at 1.000 fps.

Send me more information

Leave us your contact information and our expert will get back to you.

If you wish to further specify your question you can fill in more detailed information by using our general contact form.

High-Speed Visualization of Unsteady Processes in a Scramjet Combustion Chamber Using CAVILUX Smart Diode Laser


Author: Stuart Laurence, German Aerospace Center (DLR)

1 Introduction

Scramjets (supersonic combustion ramjets) are a technology for high-speed propulsion, potentially allowing large performance advantages over rockets. Despite decades of development, however, significant obstacles remain to the routine deployment of scramjets for access-to-space or high-speed travel. One critical issue associated with scramjets is that of inlet unstart, defined as the upstream displacement or disgorging of the original inlet shock system. The resulting detached shock that forms in front of the inlet leads to large flow spillage, reducing performance, and can also produce violent, unsteady loading on the engine, potentially resulting in its destruction. One possible cause of inlet unstart is abnormal operating conditions inside the combustion chamber of the engine, leading to the upstream propagation of pressure disturbances which can displace the inlet shock system. Such a process was cited as the reason for the failure of the second X-51 flight, for example. In order to predict, detect, and prevent such unstart events, it is thus necessary to study the responsible mechanisms inside the combustion chamber, and to determine how the transient phenomena that accompany incipient unstart may manifest themselves. This is the focus of the present investigation.

2 Experimental configuration

All experiments were performed in the HEG (High Enthalpy shock tunnel Göttingen) facility of the German Aerospace Center (DLR). The HEG is a reflected-shock wind tunnel, capable of reproducing a wide variety of flow conditions at Mach numbers from approximately 6 to 10. The test time of the facility is limited to a few milliseconds; while this still allows the study of a range of high-speed flow problems, it makes visualization, particularly of unsteady phenomena, especially challenging.

The model configuration for the present study is a full-scale reproduction of the fuelled flow path of the HyShot II flight experiment. HyShot II was the first successful test in the HyShot program, run by the University of Queensland, Australia, designed to provide reference supersonic combustion data at hypersonic Mach numbers. The configuration is thus an academic one, rather than a practical thrustproducing engine; however, the simple design and convenient optical access made the Hyshot II an ideal candidate for the present study. The flowpath is shown in the right part of figure 1. The intake consists of a simple 18º wedge; the boundary layer generated on this surface is swallowed by a boundary-layer bleed channel, rather than being allowed to enter the combustion chamber. The combustion chamber is a constant-area duct of 300 mm length, 9.8 mm height and 75 mm width. Hydrogen fuel is injected 58mm downstream of the intake-side leading edge in the wall-normal direction through four port-holes. The exhaust consists of a simple two-dimensional expansion. A schematic drawing of the model inside the HEG test section is shown in the left part of figure 1.

In the present experiments, the wind tunnel was run in a manner to reproduce the flight conditions of the HyShot II flight experiment at 28 km altitude. The free-stream Mach number was approximately 7.4. Hydrogen fuel was injected at various pressures to induce both steady and unsteady combustion conditions.


Figure 1: (Left) Schematic drawing of the HyShot II scramjet model in the HEG test section: (a) HEG nozzle; (b) valve for hydrogen injection. (Right) Flow path for the HyShot II scramjet (upside-down relative to the left schematic): (c) intake ramp; (d) boundary-layer bleed channel; (e) injection location; (f) cowl-side combustion chamber wall; (g) injector-side wall; (h) exhaust surfaces.

An important component of the present study was visualization of the flow and combustion features; thus, windows were installed in the model, providing optical access to almost the entire combustion chamber. The first type of imaging implemented was Schlieren, for which a conventional Z-type arrangement was employed. The Schlieren technique visualizes the first spatial derivative of the flow density, and is thus useful for imaging features such as shocks, boundary layers, and shear layers in compressible flows. Highspeed Schlieren imaging is challenging in facilities such as HEG, because of both the short test times and the significant amounts of self-luminosity produced by the hot gases in the test flow. This luminosity can potentially overwhelm the light source. If the flow is combusting, as in the present case, the luminosity problem is further aggravated. The use of CAVILUX Smart illumination laser for visualization in the present experiments overcame this problem, as the monochromatic nature of the light produced allowed the insertion of a narrow band-pass filter in the light path, effectively removing the self-luminosity. The incoherent nature of the light also eliminated the speckle and diffraction edges typically associated with laser light sources. A further advantage of this light source was the short pulse duration (here ~30 ns), which effectively freezes the flow structures. The CAVILUX Smart was employed together with a Shimadzu HPV-1 high speed camera, with frame rates of 16 or 32 kfps.

The second type of imaging was OH* chemiluminescence visualization, again using the Shimadzu HPV-1 (without a light source). OH is one of the intermediate products in the combustion of hydrogen; thus, the concentration of the electronically excited radical, OH*, gives a good indication of the onset of the flame. OH* also has the advantage of emitting strongly over a narrow wavelength band (near 310 nm), so by placing a narrow band-pass filter of this wavelength in front of the camera, line-of-sight intensity distributions of this radical can be easily obtained. By combining Schlieren and OH* chemiluminescence visualizations then, we can draw links between the flow and combustion features seen inside the combustion chamber.

3 Results

First we concentrate on results obtained for low hydrogen injection pressures, which resulted in steady combustion conditions developing inside the combustion chamber. In figure 2 are shown Schlieren and OH* visualizations of the flow region close to the injection location (the visualized region is approximately 80 mm long). The injection port-holes are located at the lower left corner of each image. The barrel shock created by the interaction of the injected hydrogen with the incoming test-flow is clearly seen in the Schlieren image, as are several of its reflections down the duct. The injection jets themselves are also visible; the freezing of the turbulent structures by the short laser-pulse duration is shown to good effect Application notes – R&D Combustion Schlieren imaging Cavitar Ltd › here. The penetration depth is approximately one-half of the duct height by the downstream end of the visualization window. In the OH* image, there is no evidence of combustion occurring directly at the point of injection; rather, combustion appears to be initiated close to the injector-side wall by the first reflection of the barrel shock. The temperature and pressure rise across the shock thus seem to be sufficient to bring the hot hydrogen in the boundary layer to the sides of the injection jets to ignition conditions. Combustion remains isolated to the boundary layer until the second reflection of the barrel shock, which “kicks” the flame further out into the duct and increases the intensity of combustion. These images thus show a clear linkage between the flow and combustion features.


Figure 2: (Above) Schlieren image of the flow in the HyShot II combustion chamber near the injection location (seen at the bottom left corner) for steady combustion conditions. (Below) OH* chemiluminescence image of the same region.

As the amount of injected hydrogen was increased, it was noted from pressure transducer measurements that a pressure disturbance would develop and begin to propagate upstream, signaling the onset of unstart. As the nature of this disturbance was not clear, high-speed visualizations were captured of the unsteady development, and sequences from these are shown in figure 3. In the left column is shown a sequence of the flow region close to the injection location (the same region as in figure 2). Initially (1.4 ms) we see that the picture is similar to that for the steady combustion case (note that the exposure time for the OH* image is reduced here, which explains the apparent weakness of the observed combustion). At 3.5 ms, however, we first see the arrival of a shock structure on the cowl-side wall in the Schlieren image, with an accompanying bulging feature in the OH* image. These continue to move upstream, until a quasi-steady configuration arises from around 4.4 ms, with the shock lodged on the cowl-side wall just downstream of the injectors, and the main combustion region located immediately downstream of the impingement point of this shock on the injector-side wall. The appearance of the OH* structure as it moves upstream suggests the development of a separated flow region on the injector-side wall, as the increased residence time in such regions greatly enhances the hydrogen ignition. Unsteady flow structures in the Schlieren images also support this interpretation. In the final image (7.0 ms), the shock structure has continued moving upstream past the injector, but this is after the steady test time and may be a result of the unsteady inflow conditions that subsequently develop.

In the right column of figure 3 are shown corresponding images from the central combustion chamber further downstream. From 2.5 ms, the development of a shock train on the cowl-side wall is visible; this shock train quickly strengthens and moves upstream. Such shock trains in scramjet combustors are typically caused by one of two mechanisms: boundary-layer separation due to the adverse pressure gradient, or thermal choking due to excessive heat release driving the flow to sonic conditions (at which point a steady flow situation is no longer possible). In the OH* images, a gradual intensifying of the combustion is observed, but notably absent is an OH* feature that follows the shock train motion (as observed in the upstream sequence). This suggests that the shock-train development is not linked to boundary-layer separation, and is rather due to thermal choking, which contradicts the findings of previous authors who have investigated this configuration. The boundary-layer separation observed in the upstream sequence must thus develop at some later point in the propagation of the shock train.


Figure 3: Quasi-synchronous Schlieren and OH* chemiluminescence visualizations of the transient flow (near) near the injector and (right) in the central combustion chamber.

4 Conclusions

Through high-speed Schlieren visualizations employing CAVILUX Smart laser and OH* chemiluminescence imaging, we have investigated the unsteady flow and combustion phenomena leading up to unstart in a model scramjet engine. Incipient unstart was seen to take the form of a shock train that developed in the central combustion chamber and subsequently propagated upstream. By comparing Schlieren and OH* images, we were able to rule out boundary-layer separation as the primary mechanism for this unsteady development (although separated regions were observed further upstream); thermal choking due to excessive heat release was thus isolated as the responsible mechanism.

About the author

Stuart Laurence (Ph.D.) completed his graduate studies at the Graduate Aeronautical Laboratories, California Institute of Technology, in the area of hypersonic flows. He currently works at the German Aerospace Center (DLR), Göttingen, on supersonic combustion, boundary-layer transition, and experimental methods for high-speed flows.

Send me more information

Leave us your contact information and our expert will get back to you.

If you wish to further specify your question you can fill in more detailed information by using our general contact form.