You are on page 1of 16

41st AIAA Fluid Dynamics Conference and Exhibit AIAA 2011-3428

27 - 30 June 2011, Honolulu, Hawaii

Experiments on the Influence of a Microramp Array


on a Hypersonic Shock Turbulent Boundary Layer
Interaction
Anne-Marie Schreyer1 , Dipankar Sahoo2∗ and Alexander J. Smits2†
1
Institute for Aerodynamics and Gas Dynamics, University of Stuttgart, 70569 Stuttgart, Germany.
2
Gas Dynamics Laboratory, Princeton University, Princeton, New Jersey, 08544, United States
Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Experiments were performed to study the effects of an array of microramp sub-boundary


layer vortex generators on a hypersonic shock/turbulent boundary layer interaction. First
the interaction between the turbulent boundary layer over a smooth plate and the oblique
shock wave generated at a 33◦ compression corner was investigated in a Mach 7.2 flow.
Then two staggered rows of microramps were installed upstream of the interaction region,
and the influence of these vortex generators on the flow field and the separation region
in particular was studied. Mean flow surveys and turbulence measurements were per-
formed using Particle Image Velocimetry (PIV). The microramps strongly altered the flow
field, decreasing the mean velocity at the spanwise locations downstream of the microramp
vertices, and increasing it at the spanwise locations in between vertices. The previously
two-dimensional interaction region with a large separated zone in the vicinity of the ramp
corner is broken up into a three-dimensional interaction, where separation is diminished
at the spanwise positions in between the microramp vertices, but increased downstream of
the vertices.

Nomenclature
B Constant on the log-law
M Mach number
Reθ Reynolds number based on momentum thickness
T Temperature, K
U Streamwise velocity, m/s
c Microramp chord length
h Microramp height
l Microramp length
p Pressure, Pa
s Spanwise spacing between the microramps
u Fluctuating component of velocity in the streamwise direction, m/s
uτ Friction velocity, m/s
v Fluctuating component of velocity in the wall normal direction, m/s
x Streamwise distance, mm
y Normal distance from the wall, mm
ζ Streamwise distance in the ramp parallel direction, mm
η Normal distance from the compression ramp surface, mm
δ Boundary layer thickness, mm
δ0 Undisturbed boundary layer thickness measured at the location of the corner, mm
κ Von Karman constant, m/s
ρ Density, kg/m
∗ AIAA Member.
† AIAA Fellow.

1 of 16
Copyright © 2011 by Anne-Marie Schreyer. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
American Institute of Aeronautics and Astronautics
µ Dynamic viscosity, Pa·s2
ν Kinematic viscosity, m/s2
Superscript
+ Inner scaling
Subscript
e Value at boundary layer edge
w Value at wall
0 Value at stagnation conditions
∞ Value in free stream for empty test section
<> Time averaging

I. Introduction
Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

In the flow field around a hypersonic vehicle, shock wave/turbulent boundary interactions (SWTBLI)
occur under numerous circumstances. If the shock wave is strong enough, large-scale separation is induced
causing flow distortion and a significant loss of total pressure. In addition, the unsteadiness associated with
separated SWTBLI can cause large temperature and pressure loads on the structure and it is therefore
important to study means of preventing or at least diminishing shock induced separation.
To accomplish this aim, different methods of flow control have been studied. A common form of flow
control in supersonic engine inlets is boundary layer bleed where low-momentum fluid is removed from the
boundary layer. Because this bleed reduces the mass flow rate into the engine and increases the structural
weight, the efficiency and the overall range of the aircraft are reduced. Therefore, other methods of flow
control have drawn attention, and vortex generators are of particular interest here. Vortex generators with a
height of the order of the boundary layer thickness or larger, as described by Pearcey,1 are in use for several
supersonic and subsonic applications.2 Starting with the work of Anderson et al.,3 smaller sub-boundary layer
vortex generators, particularly the so-called microramps, have been investigated for flow control in supersonic
engine intakes. Microramps extend beyond the viscous sublayer but do not exceed the boundary layer
thickness, and thus cause less disturbance to the outer supersonic flow than conventional vortex generators,
and therefore cause less drag. Devices with heights between 20% and 90% of the uncontrolled boundary
layer thickness have been investigated, either singly or arranged in arrays. For the control of SWTBLI,
experiments and numerical studies have been performed at Mach numbers between 1.84 and 3 to study the
most effective positioning of the microramps upstream of the SWTBLI zone.4–10
In particular, the influence of a single microramp on a shock induced separation region was studied by
Ford and Babinsky4 for a Mach 2.5 impinging shock wave/turbulent boundary interaction and by Babinsky
et al.5 for a Mach 2.5 separated SWTBLI generated by a 7◦ compression corner. The flow field over a
single microramp is shown in Figures 1 and 2 and is similar for devices with different heights.4 A pair of
counter-rotating vortices entraining high momentum fluid is generated in the boundary layer, leading to a
local reduction of the boundary layer displacement thickness. The vortices were observed to remain in the
boundary layer for a significant streamwise distance before lifting off the surface.
The vortices generated by the microramp were found to break up the two-dimensional separation zone
into periodic three-dimensional separation zones. Babinsky et al.5 found that for a single microramp, the
flow remains attached directly downstream of the vertex of the microramp and is separated on both sides of
the attached region. Multiple microramps in single or double row configurations were studied by Blinde et
al.6 who investigated their effects on a separated Mach 1.84 incident shock wave/boundary layer interaction
by means of Particle Image Velocimetry. They found that the flow was decelerated downstream of the vertex
of each device, which increased the streamwise extent of the separated zone at that spanwise location, and the
flow was accelerated between the devices, decreasing the size of the separated zone. A conceptual sketch of
the flowfield is shown in Figure 3. These differences in action were noted by Blinde et al.,6 who ascribed them
to differences in the boundary layer characteristics and resulting differences in vortex migration.6 However,
in both cases the total average length of the interaction zone is reduced by an amount that depends on
the size of the microramps.4, 5 Larger microramps are more effective because they induce larger vortices
entraining more high momentum fluid, but they also increase the associated drag, although the location
where the microramps were installed ahead of the interaction zone did not appear to have a significant
influence on their effect on the separated zone.4 Blinde et al.6 also noted that the staggered array of two

2 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 1. Surface flow visualization of the features of the flow field around a single microramp. Freesteam Mach
number was 2.5, and the height of the ramp was 80% of the undisturbed boundary layer thickness. Image from Ford
and Babinsky.4

Figure 2. Sketch of the flow field around a single microramp. Figure from Babinsky et al.5

rows of microramps showed a stronger effect on the overall flow field than the single row configuration, and
they found that the unsteady motion of the shock was stabilized by both configurations.
Here we extend the study of microramps to hypersonic flow for the first time. We examine the effects
of a staggered array of two rows of microramps on a separated shock/turbulent boundary layer interaction
generated by a 33◦ compression corner at a Mach number (Me ) of 7.2. The undisturbed interaction was
studied in detail by Schreyer et al.11 The Reynolds number based on momentum thickness Reθ = 3500,
which is accessible to DNS.12–14 Detailed mean flow and turbulence measurements were obtained.

II. Experimental facility and measurement techniques


The experiments were performed in the Mach 8 Hypersonic Boundary Layer Facility (HyperBLaF) at
the Princeton Gas Dynamics Laboratory. The HyperBLaF is a blowdown wind tunnel using air as the main
working fluid. The maximum stagnation temperature is 870K at a maximum stagnation pressure of 10 MPa
(1500 psi), and the run time is approximately 2 minutes. The Reynolds number range is such that laminar
flow exists (even on the tunnel walls) at the lowest Reynolds number, and at the highest Reynolds number
fully turbulent boundary layers are generated on a flat plate mounted in the test section.
The circular test section consists of two stainless steel sections of 3 ft (914 mm) length and has an inner
diameter of 9 in (229 mm). One section is equipped with four orthogonal window cavities allowing optical
access to the flow. The cavities are rectangular sections with a size of 127 mm (5 in) x 206 mm (16 in),
beginning 89 mm (3.5 in) from the beginning of the test section. The quartz windows have dimensions
of 225 mm x 137 mm x 12.7 mm and are recessed 38 mm (1.5 in) from the test section wall. To minimize
disturbances detrimental to the starting of the tunnel, ramps were installed at the ends of each of the window
plates. The tunnel is described in detail by Baumgartner.15

3 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 3. Conceptual sketch of the interaction of an incoming turbulent boundary layer with an incident oblique shock
wave. Downstream of the microramp vertices, hairpin vortices (transparent green) are present, and high-speed regions
are found at intermediate positions. Velocity vectors are shown in a convective reference frame moving at 73% of the
mean freestream velocity. Figure taken from Blinde et al.6

Figure 4. Schematic of the wind tunnel flat plate model with the compression corner, the microramp array and the
trip wire.

In the empty working section the Mach number is equal to 8.0 ± 0.1 over the central 80% of the cross-
sectional area.15, 16 When the flat plate model is installed in the the working section, the free-stream Mach
number at the location of measurement is equal to 7.2.
The model used in this study consisted of a 33◦ compression corner and an expansion corner made of
brass which was mounted on a flat plate as shown in the sketch in Figure 4. The ramp corner is located
350 mm downstream of the leading edge. The ramp surface was 30 mm long, and it was followed by an
expansion corner of 33◦ so that the total deflection of the flow was zero. The upstream boundary layer was
tripped with a trip wire of 2.4 mm height mounted 59 mm downstream of the leading edge of the flat plate
to obtain a fully developed turbulent boundary layer at the region of measurement in the vicinity of the
ramp corner. At this point, the undisturbed boundary layer thickness was δ0 = 9.8 mm. The flat plate was
installed in the test section on a support with a diamond-shaped cross-section to minimize blockage of the
flow. The support was fitted to a stainless steel window plate bolted to the side window plate. The side and
bottom windows were used for optical access for the PIV measurements.
A total of seven microramps were arranged in a two row staggered array and positioned 101.6 mm (≈ 10δ)
upstream of the compression corner as shown in Figure 4. The dimensions of the wind tunnel model are
shown in Figure 5. The maximum height of the microramps was chosen to be h = 5 mm (h/δ0 = 0.5). The
other dimensions of the microramps and their spanwise arrangement scaled with the height according to the
specifications described by Anderson et al.3 so that c/h = 7.2 and s/h = 7.5, where c the chord length and s
is the spanwise spacing between microramps. The streamwise spacing between the two rows of microramps
was chosen to equal the microramp length l. The angle of incidence of the microramps was 24◦ . The overall
dimensions are shown in Figure 6.
In the PIV experiments presented here, a New Wave Tempest and Gemini PIV dual head ND:YAG laser

4 of 16

American Institute of Aeronautics and Astronautics


Figure 5. Geometry and dimensions of the wind tunnel model: positioning of the compression corner, trip wire and
microramp array on the flat plate model
Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 6. Dimensions of the microramp sub-boundary layer vortex generators and positioning within the array

system was used as laser source. Each laser delivered 100 mJ energy per pulse at a wavelength of 532 nm.
The pulse width is between 3 − 5 ns and the jitter corresponds to ±0.5 ns. The time delay between the two
lasers was set to 0.4 µs as described by Sahoo et al.14 The images were recorded with a PCO Cooke camera
with a 100 mm lens and then analyzed with standard digital PIV software. The PIV system is described in
detail by Sahoo et al.17
T iO2 particles were used as seeding particles. The particles were introduced upstream of the nozzle
throat in the settling chamber to minimize disturbance. The particles were entrained by high pressure air in
a fluidized bed before being injected into the settling chamber. The manufacturer specified particle diameter
is approximately 50 nm, but the particles tend to agglomerate to clusters with a diameter of approximately
400 nm. Schrijer et al.18 found that the resulting particle frequency response is approximately 400 kHz.
Recently, Nguyen et al.19 used a RANS calculation to estimate the average particle diameter in our previous
8◦ compression ramp experiment,20 and found the particle diameter in that experiment was closer to 700
nm than 400 nm. Also, the particle frequency response will depend on the local viscosity since we assume
Stokes flow. In the 8◦ compression ramp experiment, the frequency response was estimated to be about 38
kHz in the free-stream and 87 kHz in the boundary layer. In the present 33◦ study, special care was taken
to minimize agglomeration, and it seems likely that the particle diameters were between 400 and 700 nm.
The verification of that by means of a particle response assessment is currently in progress.

III. Data reduction and analysis


To obtain meaningful results from the PIV images, some preprocessing was necessary. First, the raw PIV
images showing unsatisfactory seeding density were discarded. Second, the images were shifted and rotated
to correct for camera misalignment and movement of the tunnel during the course of the experiment. Third,
the particles within a distance of 0.3 mm to the edges of the field of view were not taken into account for
the calculation of the velocity vectors to avoid the introduction of erroneous vectors. Then averaged flow
quantities and turbulence statistics were determined from about 1000 image pairs per measurement position.
Due to the limited runtime of the experimental facility and the limited memory of the camera, these images

5 of 16

American Institute of Aeronautics and Astronautics


p0 [MPa] T0 [K] Tw [K] pw [Pa] M∞ Me
7.93 ± 2% 760 ± 1.5% 340 ± 1.5% 1365.6 ± 1.8% 8.0 ± 0.1 7.2 ± 0.1
Table 1. Wind tunnel operating conditions: stagnation pressure (p0 ), stagnation temperature (T0 ), static temperature
at the wall (Tw ), static pressure at the wall (pw ), free stream Mach number (M∞ ) in an empty test section and edge
Mach number (Me ) at the location of measurement

Ue [m/s] uτ [m/s] cf δ [mm] Reθ


1146 57.3 0.00095 9.8 3500
Table 2. Upstream, undisturbed boundary layer parameters.

were collected in several wind tunnel runs. Data from different runs were combined by establishing the
weighted mean x of a nondimenzionalized quantity x, e.g. the streamwise velocity component normalized
with the free-stream velocity. That is,
Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Pn
wi xi
x = Pi=1n
i=1 wi
for n runs and w images obtained for a given run.

IV. Results and discussion


The mean operating conditions are summarized
p in Table 1, and the upstream undisturbed boundary layer
parameters are given in Table 2, where uτ = τw /ρw , τw is the local shear stress, and Cf = 2τw /(ρe Ue2 ).
The flow field was subdivided into several fields of view (FoV) for the PIV measurements, as visualized
in Figure 7, to obtain a minimum resolution of 50 pixels per mm. The field of view was about 32 × 24 mm
for FoV A and B, and about 28 × 24 mm for FoV C.
Profiles of the mean and fluctuating velocities were extracted at twelve locations along the centerline of
the model, as shown in Figure 7. The streamwise position of the ramp corner is at x4 = 0 mm (≡ ζ0 = 0 mm)
and the location of the expansion corner is at x10 = 25.2 mm ≡ ζ10 = 30 mm. Mean velocities and fluctuating
quantities are always given parallel (and normal) to the local wall direction. Upstream of the compression
corner and downstream of the expansion corner, the profiles are given in the (x, y)-coordinate system, the
(ζ,η)-system is oriented parallel and normal to the compression ramp and thus rotated 33◦ in respect to
(x,y), as shown in Figure 8. The distance between the locations at which the profiles are presented, is 5 mm
≈ 0.5δ0 , where δ0 is the boundary layer thickness in the upstream undisturbed turbulent boundary layer.
At the location of the expansion corner, velocity profiles are presented both in the ζ-direction parallel to the
ramp surface and in the x-direction. The profiles are denoted with (10a) and (10b), respectively, for easier
distinction.

Figure 7. Locations of the three fields of view (FoV A, B, and C) and the twelve positions, in respect to the ramp
corner, for which the velocity fluctuations across the boundary layer are presented here (not to scale). The flow is from
left to right.

6 of 16

American Institute of Aeronautics and Astronautics


Figure 8. Coordinate system (x, y) in the directions parallel and normal to the flat plate, and system (ζ,η) in the
directions parallel and normal to the 33◦ compression ramp surface. Both systems originate at the ramp corner.
Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 9. Mean velocity profiles of the incoming boundary layer, as measured by Sahoo et al.17 in an undisturbed
flat plat boundary layer, and in the boundary layer upstream of an 8◦ ramp measured by the present authors.20 On
the left, the profile is shown in physical coordinates. On the right, the profile is shown in inner variables, transformed
according to van Driest.

IV.A. Incoming boundary layer


The undisturbed turbulent boundary layer in the present flow field is identical to that measured by the
authors in a previous study of a 8◦ compression corner.20 The streamwise mean velocity component, non-
dimensionalized with Ue , is shown in Figure 9. The data show good agreement with that taken by Sahoo
et al.17 in the same flow and measurements by Owen et al.21 at a similar Mach number. Also, the profile
in inner variables (transformed according to van Driest) shows the expected agreement with the log-law for
y + > 30. By fitting the log-law region using κ = 0.41 and B = 5.0, the skin friction coefficient Cf was
estimated to be 0.00095 (see Table 2).
The velocity fluctuations and Reynolds shear stresses measured in the upstream undisturbed boundary
layer,20 are shown in Figure 10. We consider the wall normal velocity fluctuation component v 0 in this data
set as preliminary, since by comparison with other data it seems generally too small, especially near the
wall. In contrast, the streamwise component u0 and the shear stress u0 v 0 appear to be captured accurately.
The possible reasons for the low values of v 0 are currently under investigation, but it seems likely due to
inadequate particle displacement between successive images.

IV.B. Mean flow


A useful byproduct of the PIV measurements was that a thin film of PIV particles tended to accumulate
on the surface of the model during a run. These particle deposits help to visualize the surface flow field,
and an example is shown in Figure 11. The undisturbed interaction, shown on the left side of Figure 11, is
approximately two-dimensional within the inner 60% of the model span. Introducing the microramp array
upstream of the interaction breaks up the two-dimensionality of the separation zone, as previously observed
by Ford and Babinsky,4 Babinsky et al.5 and Blinde et al.6 Downstream of the vertices of the microramps,
the separated zone appears to have increased in length so that it extends up to the end of the compression

7 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 10. Streamwise and wall normal velocity fluctuations and Reynolds shear stresses measured in the upstream
undisturbed boundary layer.20

Figure 11. Surface flow visualizations using deposition of PIV seeding particles. Left: undisturbed interaction. The
extent of the separated region onto the ramp surface is marked with a red line. Right: interaction disturbed by two
staggered row of microramps. The yellow lines mark the positions of the laser sheets in respect to the microramps, for
which PIV images were taken: LS 1 downstream of the vertex of a microramp in the first row, LS 2 in between the
vertex of a microramp in the first row and the vertex of a microramp in the second row in the array.

ramp surface (marked with red curves on the right side of Figure 11). Downstream of the locations in between
the vertices of the microramps, the length separated zone appears to have decreased, which agrees with the
observations of Blinde et al.6 The end of the separated zone moves upstream and is located much closer to
the compression corner than in both the undisturbed case and the location downstream of the microramps
vertices.
Based on these observations, the positions of the streamwise laser sheets for the PIV measurements were
chosen. The first position is in the plane directly downstream of the vertex of a microramp in the first row,
the second position is the plane in between the vertex of a microramp in the first row and the vertex of a
microramp in the second row of the array (see Figure 11).
To study the basic features of the interaction in more detail, contour plots of the mean velocity component
in the x-direction with streamlines, and profiles of the mean velocity Umean in the local wall direction are
presented for the region of the compression corner in Figures 12, 13, and 14, for the undisturbed interaction
and the cases with the microramp array, respectively. Umean was normalized with the mean velocity Ue at
the edge of the undisturbed boundary layer. Further details of the unperturbed interaction are given by
Schreyer et al.11
The mean velocity close to the wall in the streamwise direction is considerably altered by the microramp
array. At the spanwise positions downstream of the microramp vertex (LS 1), the mean flow is decelerated,
while at the spanwise positions in between the vertices (LS 2), the mean velocity close to the wall is higher

8 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 12. Undisturbed interaction. Contours of the mean velocity component (given in m/s) in the x-direction with
streamlines in the vicinity of the compression corner, and profiles of the mean velocity Umean in the local wall direction
normalized with Ue at x1 = −1.5δ0 , x2 = −1.0δ0 , x3 = −0.5δ0 , x4 = 0, ζ5 = 0.5δ0 , ζ6 = 1.0δ0 , ζ7 = 1.5δ0 and ζ8 = 2.0δ0 .

compared to the undisturbed interaction. The mean velocity profile at position x1 = −1.5δ0 upstream of
the compression corner reaches the level of the velocity profile in the undisturbed boundary layer20 for a
distance from the wall of y/δ0 ≈ 0.5 for the spanwise position LS 2 in the perturbed case, while the same
level is reached at y/δ0 ≈ 0.8 in the undisturbed interaction and at y/δ0 > 1.2 for the spanwise position
LS 1 in the disturbed case. This agrees well with the conceptual model introduced by Blinde et al.,6 shown
in Figure 3, where the microramps generate longitudinal streamwise vortex pairs on the mean flow level,
inducing low-speed regions downstream of the microramp vertices and high-speed regions at the intermediate
spanwise locations.
In Figure 12 for the undisturbed interaction, a separation bubble extending from approximately 20 mm
upstream of the ramp corner to 5 mm downstream of the corner in the x-direction is clearly visible in the
streamline plot. Also, in the mean velocity profiles extracted at positions x1 to x4 , the separation bubble
is noticeable, although no negative velocities could be observed in the recirculation zone close to the wall,
indicating that the resolution of the PIV measurements in the region very near the wall (≤ 0.1δ0 ) was not
sufficient.
Downstream of the vertices of the microramps, the boundary layer is considerably thickened compared
with the undisturbed interaction, as visible in the mean velocity contour plot and profiles shown in Figure
13. From the surface flow visualizations shown in Figure 11, and the analysis on reversed flow probability
performed by Blinde et al.,6 we expect that the streamwise extent of the separated region is larger downstream
of the microramp vertices than for the undisturbed separation, possibly not reattaching on the surface of
the compression ramp due to its limited length. Unfortunately, the PIV data in the wall near zone in the
compression corner region do not confirm these expectations due to resolution issues. It was found difficult
to get a sufficient seeding density into this zone downstream of the microramp vertex, probably due to the
strong vortices introduced by the microramps. The PIV data close to the wall in the direct vicinity of the
ramp corner can therefore give qualitative trends only.
The data taken at position LS 2 between the microramp vertices, presented in Figure 14, indicate that
the extent of the separated zone was reduced compared with the undisturbed shock/turbulent boundary
layer interaction at the 33◦ ramp. While the peaks of the mean velocity profiles are distorted away from the
wall until ζ < 1.5δ0 in the undisturbed interaction, this is only the case up to ζ < 1.0δ0 for the controlled
case at LS 2.
For the region in proximity to the expansion corner, contour plots of the mean velocity component in
the x-direction with streamlines, and profiles of the mean velocity Umean in the local wall direction are
presented in Figures 15, 16, and 17. The flow downstream of the expansion corner appears to be influenced
by an interaction of several phenomena: the shock wave originating at the compression corner, the expansion
fan forming at the deflection corner itself, and the longitudinal streamwise vortex pairs generated by the
microramps that are lifting off the surface and convecting downstream. This can be seen particularly well in
Figure 16 for the position downstream of the microramp vertex, where the velocity profiles are decelerated.

9 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 13. Perturbed interaction at LS 1. Contours of the mean velocity component (given in m/s) in the x-direction
with streamlines in the vicinity of the compression corner, and profiles of the mean velocity Umean in the local wall
direction normalized with Ue at x1 = −1.5δ0 , x2 = −1.0δ0 , x3 = −0.5δ0 , x4 = 0, ζ5 = 0.5δ0 , ζ6 = 1.0δ0 , ζ7 = 1.5δ0 and
ζ8 = 2.0δ0 .

Figure 14. Perturbed interaction at LS 2. Contours plot of the mean velocity component (given in m/s) in the x-
direction with streamlines in the vicinity of the compression corner, and profiles of the mean velocity Umean in the
local wall direction normalized with Ue at x1 = −1.5δ0 , x2 = −1.0δ0 , x3 = −0.5δ0 , x4 = 0, ζ5 = 0.5δ0 , ζ6 = 1.0δ0 , ζ7 = 1.5δ0
and ζ8 = 2.0δ0 .

For positions (10b) to (12) an intermediate state seems to be reached: towards the boundary layer edge,
the velocity reaches neither the freestream value downstream of the shock wave, nor the freestream value
expected downstream of the expansion fan. Instead, a value in between both is approached. Until x12 = 3.5δ0
downstream of the expansion corner, the boundary layer has not reached the equilibrium state. For the
spanwise position in between the microramp vertices, presented in Figure 17, though, the extent of the
interaction region seems to be diminished, and the equilibrium state seems to be reached even closer to the
expansion corner than in the undisturbed interaction. For position x11 = 3.0δ0 for example, the velocity
reaches the expected freestream value downstream of the expansion fan at y/δ0 ≈ 1.2, while it is not yet
reached in the undisturbed case.

IV.C. Turbulence
The streamwise and wall normal velocity fluctuations throughout the interaction are presented in Figures
18 and 19, respectively. At each of the positions, the turbulent fluctuations for the uncontrolled flow (black
symbols), and the two spanwise measurement positions downstream of the microramp vertex (blue symbols)
and in between the vertices (red symbols), are compared. All variables were non-dimensionalized using the
undisturbed values of the friction velocity uτ and boundary layer thickness δ (= δ0 ).

10 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 15. Undisturbed interaction. Contours of the mean velocity component (given in m/s) in the x-direction with
streamlines in the vicinity of the expansion corner, and profiles of the mean velocity Umean in the local wall direction
normalized with Ue at ζ8 = 2.0δ0 , ζ9 = 2.5δ0 , ζ10a = 3.0δ0 , x10b = 2.5δ0 , x11 = 3.0δ0 and x12 = 3.5δ0 .

Figure 16. Perturbed interaction at LS 1. Contours of the mean velocity component (given in m/s) in the x-direction
with streamlines in the vicinity of the expansion corner, and profiles of the mean velocity Umean in the local wall
direction normalized with Ue at ζ8 = 2.0δ0 , ζ9 = 2.5δ0 , ζ10a = 3.0δ0 , x10b = 2.5δ0 , x11 = 3.0δ0 and x12 = 3.5δ0 .

The turbulent fluctuations measured in the undisturbed turbulent boundary layer were given in Figure
10. In the undisturbed boundary layer, the maximum of the streamwise velocity fluctuation component was
located very near the wall and the fluctuation level decreased monotonically with increasing distance from
the wall.
For the unperturbed interaction at the 33◦ compression ramp corner, the maximum magnitude of the
streamwise component remains at the same level for all upstream positions, positions (1) to (4), while the
wall normal component is amplified as the ramp corner is approached. The separation bubble is evident
in the velocity fluctuations, in that the peak is distorted away from the wall from y/δ0 ≈ 0.25 at x1 to
y/δ0 ≈ 0.55 at x4 . Also, in the expansion further downstream, the streamwise and wall normal velocity
fluctuations, and the Reynolds stresses were further amplified. This is in contrast to the findings by Smith
& Smits,22 who investigated a 20◦ compression corner followed by a 20◦ expansion corner with an incoming
Mach number of 2.9. In the current interaction, the flow downstream of the deflection corner is a result
of the combined influence of the shock wave and the expansion fan, which are very close together, and the
combined effect may explain the second region of increasing turbulence. More details are given by Schreyer
et al.11
Across the shock wave, the streamwise component < u0 > shows no amplification in the near wall region,
but further away from the wall the magnitude increases. The fluctuation level starts to decrease from
ζ6 = 1.0δ0 and remains at a relatively constant level up to ζ9 = 2.5δ0 . The wall normal component < v 0 >

11 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 17. Perturbed interaction at LS 2. Contours of the mean velocity component (given in m/s) in the x-direction
with streamlines in the vicinity of the expansion corner, and profiles of the mean velocity Umean in the local wall
direction normalized with Ue at ζ8 = 2.0δ0 , ζ9 = 2.5δ0 , ζ10a = 3.0δ0 , x10b = 2.5δ0 , x11 = 3.0δ0 and x12 = 3.5δ0 .

is strongly amplified across the interaction, and starts to decrease for 1.0δ0 > ζ > 2.5δ0 as the streamwise
component. The peak for both components is located at around y/δ0 ≈ 0.25 − 0.3 for all positions. At the
expansion corner, both components start to increase again in the wall near region and are then strongly
amplified by the combined influence of the shock wave and the expansion fan.
This behavior is greatly altered when introducing the microramps: the streamwise component still re-
mains at the same magnitude for all upstream positions, but this is also true for the wall normal component.
Compared with the undisturbed interaction, the streamwise component is slightly increased at the spanwise
locations in between the microramp vertices, and the peak is not moving away from the wall, when getting
closer to the ramp corner, but remaining at y/δ0 ≈ 0.25 for all upstream locations. The streamwise compo-
nent at the spanwise location downstream of the microramp vertex, on the other hand, is slightly decreased,
and the peak is located further away from the wall at y/δ0 ≈ 0.75 for all locations. The location of the
peaks is preserved in the wall normal component for both spanwise locations, but the magnitude of both
components is decreased compared with the undisturbed case. This may be due to the fact that < v 0 > is not
captured correctly within the wall near separated region, especially when it is small and the corresponding
particle displacement in the PIV images is small.
Under the influence of the shock wave, < u0 > even decreases slightly for the spanwise position between
the vertices, while it shows no amplification for the position downstream of the microramp vertex. The
< v 0 > component is strongly amplified for both spanwise locations, as in the undisturbed case. Further
downstream of the ramp corner the behavior of the turbulent fluctuations agrees with the undisturbed case:
the peak for both locations is located at y/δ0 ≈ 0.25 and the magnitude decreases, starting already from
ζ ≈ 0.5δ0 for the controlled disturbance. Both < u0 > and < v 0 > remaining at the same, relatively constant
low level as the undisturbed fluctuations up to ζ ≈ 2.5δ0 . The streamwise fluctuation level downstream of
the microramp vertices, starts to increase further upstream of the expansion corner than the other cases,
probably under the influence of the counter-rotating vortex pairs generated at the microramps, that start
to lift off the surface and impinge either at the ramp surface or just above the expansion corner. Further
downstream, influenced by the vortices, the expansion fan and the shock wave, the turbulent fluctuations
are amplified much more strongly than in the undisturbed interaction. As in the undisturbed interaction,
though, the peak moves away from the wall with increasing distance from the expansion corner.
The Reynolds shear stresses throughout the interaction are presented in Figure 20. At each of the
positions, the levels for the uncontrolled flow (black symbols), and the two spanwise measurement positions
downstream of the microramp vertex (blue symbols) and in between the vertices (red symbols), are compared.
In the undisturbed interaction, the shear stresses in the separated zone upstream of the ramp corner
remain at a low level resembling the equilibrium distribution obtained for the undisturbed boundary layer,
as shown in Figure 10, although inside the separation bubble the levels are weakly amplified. Under the
influence of the shock wave, the shear stresses behave like the wall normal fluctuations: the magnitude
in the wall near region increases significantly, and the continued amplification and subsequent relaxation

12 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 18. Streamwise velocity fluctuations in the separated zone upstream of the ramp corner (first row) at positions
x1 = −1.5δ0 , x2 = −1.0δ0 , x3 = −0.5δ0 , and x4 = 0, along the surface of the compression ramp (second row) at ζ5 = 0.5δ0 ,
ζ6 = 1.0δ0 , ζ7 = 1.5δ0 , ζ8 = 2.0δ0 , ζ9 = 2.5δ0 , and ζ10a = 3.0δ0 , and downstream of the expansion corner (third row)
at x10b = 2.5δ0 , x11 = 3.0δ0 , and x12 = 3.5δ0 . At each of the positions, the streamwise velocity components for three
cases are compared: uncontrolled flow without microramps (black symbols), flow controlled with two staggered rows of
microramps, measured downstream of the vertex of a microramp (blue symbols), and measured in between the vertices
of the first and second rows (red symbols).

of the stress levels with increasing distance from the ramp corner is similar. Qualitatively, the Reynolds
stress levels in the controlled interaction show the same behavior, although the magnitudes are dramatically
increased for the region downstream of the expansion corner. At the location of the expansion corner itself
(10b), the magnitudes are increased by a factor of about 4 for the spanwise location downstream of the
microramp vertices, and a factor of 18 for the intermediate location. That amplification is quickly damped,
though, reducing to a factor of only about 2 for the spanwise location downstream of the microramp vertex
at x12 = 3.0δ0 .

V. Conclusions
Experiments were performed to study the effects of an array of microramp sub-boundary layer vortex
generators on a hypersonic shock/turbulent boundary layer interaction. First, the interaction between the
turbulent boundary layer over a smooth plate and the oblique shock wave generated at a 33◦ compression
corner was investigated in a Mach 7.2 flow. Second, two staggered rows of microramps were installed upstream
of the interaction region, and the influence of these vortex generators on the flow field and the separation
region in particular was studied. Mean flow surveys and turbulence measurements were performed using
PIV.
The microramps strongly altered the flow field, decreasing the mean velocity at the spanwise locations

13 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 19. Wall normal velocity fluctuations in the separated zone upstream of the ramp corner (first row) at positions
x1 = −1.5δ0 , x2 = −1.0δ0 , x3 = −0.5δ0 , and x4 = 0, along the surface of the compression ramp (second row) at ζ5 = 0.5δ0 ,
ζ6 = 1.0δ0 , ζ7 = 1.5δ0 , ζ8 = 2.0δ0 , ζ9 = 2.5δ0 , and ζ10a = 3.0δ0 , and downstream of the expansion corner (third row)
at x10b = 2.5δ0 , x11 = 3.0δ0 , and x12 = 3.5δ0 . At each of the positions, the streamwise velocity components for three
cases are compared: uncontrolled flow without microramps (black symbols), flow controlled with two staggered rows of
microramps, measured downstream of the vertex of a microramp (blue symbols), and measured in between the vertices
of the first and second rows (red symbols).

downstream of the microramp vertices, and increasing the mean velocity at the spanwise locations in between
vertices. The previously two-dimensional interaction region with a large separated zone in the vicinity of
the ramp corner is broken up into a three-dimensional interaction, where the separation zone is diminished
at the spanwise positions in between the microramp vertices, but increased downstream of the vertices.
The influence of the shock wave on the boundary layer turbulence appears to be attenuated by the
presence of the microramps, in that the amplification of the turbulent quantities across the shock wave
starts to decrease more quickly than in the uncontrolled case.
In the unperturbed interaction, it was found that, contrary to previous observation in supersonic flows,22
the expansion region downstream of the compression corner serves to amplify the turbulence levels even
further.11 The unexpected second region of amplified turbulence downstream of the expansion corner is
aggravated by the microramps. In the present experiment the length of the ramp was only a little longer
than the length of the separation bubble, whereas in the experiment by Smith and Smits,22 studying a 20◦
compression corner followed by a 20◦ expansion corner with an incoming Mach number of 2.9, the separation
bubble was small, and the ramp face was 6.5δ0 long. Therefore, in the present case, the flow downstream
of the expansion corner was influenced by an interaction of the shock wave, the expansion fan, and the
streamwise vortex pairs introduced to the boundary layer by the microramps.
These preliminary measurements need to be complemented with measurements of the wall pressure
and temperature profiles along the surface of the ramp and downstream of the expansion corner. These

14 of 16

American Institute of Aeronautics and Astronautics


Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Figure 20. Reynolds shear stresses in the separated zone upstream of the ramp corner (first row) at positions x1 =
−1.5δ0 , x2 = −1.0δ0 , x3 = −0.5δ0 , and x4 = 0, along the surface of the compression ramp (second row) at ζ5 = 0.5δ0 ,
ζ6 = 1.0δ0 , ζ7 = 1.5δ0 , ζ8 = 2.0δ0 , ζ9 = 2.5δ0 , and ζ10a = 3.0δ0 , and downstream of the expansion corner (third row)
at x10b = 2.5δ0 , x11 = 3.0δ0 , and x12 = 3.5δ0 . At each of the positions, the streamwise velocity components for three
cases are compared: uncontrolled flow without microramps (black symbols), flow controlled with two staggered rows of
microramps, measured downstream of the vertex of a microramp (blue symbols), and measured in between the vertices
of the first and second rows (red symbols). Note that the scaling of the wall normal coordinate axis is adapted to the
local magnitudes and not constant for all positions.

measurements are in progress, as well as further characterization of the particle lag and its influence on the
data presented here.

Acknowledgments
The support of NASA under Cooperative Agreement No. NNX08AB46A directed by Program Manager
Catherine McGinley is gratefully acknowledged. The German Research Foundation (DFG) is gratefully
acknowledged for their support within the framework of the GRK 1095/1: Aero-Thermodynamic Design
of a Scramjet Propulsion System for Future Space Transportation Systems. Dan Hoffman is gratefully
acknowledged for numerous hours of help with setting up the experiment and running the tunnel.

References
1 Pearcey, H., Boundary Layer Control for Aerofoils and Wings, Boundary Layer and Flow Control: Its Principles and

Application: edited by G.Lachmann, Vol. 2, Pergamon, Oxford, 1961.


2 Lin, J. C., “Review of research on low-profile vortexgenerators to control boundary-layer separation,” Progress in

Aerospace Sciences, Vol. 38, 2002, pp. 389–420.


3 Anderson, B. H., Tinapple, J., and Surber, L., “Optimal Control of Shock Wave Turbulent Boundary Layer Interactions

Using Micro-Array Actuation: AIAA 2006-3197,” 5 - 8 June 2006.

15 of 16

American Institute of Aeronautics and Astronautics


4 Ford, C. W. P. and Babinsky, H., “Micro-Ramp Control for Oblique Shock Wave / Boundary Layer Interactions: AIAA

2007-4115,” 25-28 June 2007.


5 Babinsky, H., Li, Y., and Ford, C. W. P., “Microramp Control of Supersonic Oblique Shock-Wave/Boundary-Layer

Interactions,” AIAA Journal, Vol. 47, No. 3, 2009, pp. 668–675.


6 Blinde, P. L., Humble, R. A., van Oudheusden, B. W., and Scarano, F., “Effects of micro-ramps on a shock-wave/turbulent

boundary layer interaction,” Shock Waves, Vol. 19, 2009, pp. 507–520.
7 Lee, S., Large Eddy Simulation of Shock Boundary Layer Interaction Control Using Micro-Vortex Generators INTER-

ACTION CONTROL USING MICRO-VORTEX GENERATORS , Ph.D. thesis, University of Illinois, Urbana, Illinois, 2009.
8 Galbraith, M. C., Orkwis, P. D., and Benek, J. A., “Multi-Row Micro-Ramp Actuators for Shock Wave Boundary-Layer

Interaction Control: AIAA 2009-321,” 5-8 January 2009.


9 Herges, T., Kroeker, E., Elliott, G., and Dutton, C., “Micro-Ramp Flow Control of Normal Shock/Boundary Layer

Interactions: AIAA 2009-920,” 5-8 January 2009.


10 Hirt, S. M. and Anderson, B. H., “Experimental Investigation of the Application of Microramp Flow Control to an

Oblique Shock Interaction: AIAA 2009-919,” 5-8 January 2009.


11 Schreyer, A.-M., Sahoo, D., and Smits, A. J., “Turbulence Measurements with PIV in a Hypersonic Shock Turbulent

Boundary Layer Interaction,” June 2011.


12 Bookey, P., Wyckham, C., Smits, A. J., and Martin, M. P., “New Experimental Data of STBLI at DNS/LES Accessible
Downloaded by INDIAN INST OF TECHN. BOMBAY on February 4, 2020 | http://arc.aiaa.org | DOI: 10.2514/6.2011-3428

Reynolds Numbers: AIAA 2005-309,” 10 - 13 January 2005.


13 Ringuette, M. J., Martin, M. P., Smits, A. J., and Wu, M., “Characterization of the Turbulence Structure in Supersonic

Boundary Layers using DNS Data: AIAA 2006-3539,” 5-8 June 2006.
14 Sahoo, D., Ringuette, M., and Smits, A. J., “Experimental Investigation of a Hypersonic Turbulent Boundary Layer:

AIAA 2009-780,” 5 - 8 January 2009.


15 Baumgartner, M. L., Turbulence Structure in a Hypersonic Boundary Layer , Ph.D. thesis, Princeton University, Prince-

ton, NJ, June, 1997.


16 Etz, M. R., The effects of transverse sonic gas injection on a hypersonic boundary layer , Ph.D. thesis, Princeton

University, Princeton, NJ, 1998.


17 Sahoo, D., Schultze, M., and Smits, A. J., “Effects of Roughness on a Turbulent Boundary Layer in Hypersonic Flow:

AIAA 2009-3678,” June 22-25 2009.


18 Schrijer, F. F. J., Scarano, F., and van Oudheusden, B. W., “Application of PIV in a Mach 7 double-ramp flow,”

Experiments in Fluids, Vol. 41, No. 2, 2006, pp. 353–363.


19 Tue Nguyen, T., Behr, M., Reinartz, B., Schreyer, A.-M., Williams, O., and Smits, A. J., “Integration of CFD and PIV

Experiment,” .
20 Schreyer, A.-M., Sahoo, D., and Smits, A. J., “Experimental Investigations of a Hypersonic Shock Turbulent Boundary

Layer Interaction,” 4-7 January 2011.


21 Owen, F. K., Horstman, C. C., and Kussoy, M. I., “Mean and fluctuating flow measurements of a fully-developed,

non-adiabatic, hypersonic boundary layer,” Journal of Fluid Mechanics, Vol. 70, No. 2, 1975, pp. 393–413.
22 Smith, D. R. and Smits, A. J., “The effects of successive distortions on the behavior of a turbulent boundary layer in a

supersonic flow,” Journal of Fluid Mechanics, Vol. 351, 1997, pp. 253–288.

16 of 16

American Institute of Aeronautics and Astronautics

You might also like