See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/326736677

Flowfield Measurements in a Mach 2 Fin-Generated Shock/Boundary-Layer

Interaction

Article in AIAA Journal · July 2018

DOI: 10.2514/1.J056500

CITATIONS READS

31 295

4 authors:

Nishul Arora M. Y. Ali

Hanwha Q CELLS Syracuse University
24 PUBLICATIONS 334 CITATIONS 26 PUBLICATIONS 552 CITATIONS

SEE PROFILE SEE PROFILE

Yang Zhang F. S. Alvi

Florida State University Florida State University
47 PUBLICATIONS 320 CITATIONS 249 PUBLICATIONS 5,037 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Nishul Arora on 25 May 2019.

The user has requested enhancement of the downloaded file.

 AIAA JOURNAL

Flowfield Measurements in a Mach 2 Fin-Generated

Shock/Boundary-Layer Interaction

Nishul Arora,∗ Mohd Y. Ali,† Yang Zhang,‡ and Farrukh S. Alvi§

Florida State University, Tallahassee, Florida
DOI: 10.2514/1.J056500
An experimental study is conducted on the interaction of a swept shock wave with a turbulent boundary layer. The
shock wave is generated by a sharp unswept fin in a Mach 2 flow, where the strength of the interaction is varied from
weak to moderate by changing the angle of attack α of the fin from 10 to 15 deg, which corresponds to a normal Mach
number Mn of 1.3 and 1.4, respectively. Surface oil-flow visualization is used to study the topography of the
interaction where surface flow features such as the upstream influence line and the separation lines are identified. By
taking advantage of the quasi-conical symmetry of the flowfield, velocity field measurements are acquired in the
conical reference frame for the interaction at the moderate interaction strength of Mn
1.4 at two locations from
the fin apex. Flowfield features such as the λ-shock structure, slip line, and the separation bubble with the reverse flow
are clearly visible in the velocity field. These results are also examined in the spherical coordinate frame, and good
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

agreement in the spatial location of the critical features is found, providing direct quantitative, experimental evidence
of quasi-conical symmetry of the flowfield above the surface. An examination of the velocity field downstream of the
rear-foot of the λ-shock shows a region (a “streamtube,” bounded on one side by the slip line emanating from the triple
point) where the flow accelerates to transonic and supersonic speeds. This flow eventually turns toward and impinges
upon the flat plate, a phenomenon referred to as an “impinging jet” in the literature. It is believed to be the principal
cause of the high mean, unsteady pressures, very high heating, and skin friction coefficients that the present
measurements confirm. An examination of the turbulence kinetic energy reveals a significant increase in amplitude in
the separated flow region that is bounded by the flattened vortex under the λ-shock.

Nomenclature δ = boundary-layer thickness, mm

H = height of the fin, mm τp = relaxation time, μs
k = turbulence kinetic energy, m2 ∕s2 τf = characteristic flow time scale, μs
L = length of the fin, mm ωz = z vorticity, s−1
Li = inception length, mm
Mn = normal Mach number I. Introduction
M∞ = freestream Mach number
Po
To
=
=
stagnation pressure, psi
stagnation temperature, K I NTERACTIONS between shock waves and boundary layers are
commonly occurring phenomena in transonic/supersonic flows.
The ubiquitous nature of these interactions makes this a fluid dynamics
U = in-plane velocity, m∕s
U∞ = mean in-plane velocity, m∕s problem of both fundamental and practical importance in internal
V = wall-normal velocity, m∕s and external flows. A shock-wave/boundary-layer interaction (SBLI)
α = angle of attack is generated in a large number of applications, such as at the
βUI = azimuthal angle of upstream influence line wing–fuselage and tail–fuselage junctions of an aircraft as well as inside
βo = azimuthal angle of inviscid shock the supersonic inlets and diffusers. A shock wave of sufficient strength
βSS = azimuthal angle of separation shock interacting with the boundary layer can lead to regions of small
βRS = azimuthal angle of rear shock to massively separated flows that can degrade the aerodynamic
performance, sometimes severely. To understand the underlying flow
physics of such interactions, several extensive studies have been
Presented as Paper 2015-1517 at the 53rd AIAA Aerospace Sciences conducted. The SBLI can be classified as two- and three-dimensional
Meeting, AIAA SciTech Forum, Kissimmee, FL, 5–9 January 2015; received interactions. Typically, in two-dimensional (2-D) interactions, the
5 July 2017; revision received 14 April 2018; accepted for publication 14 boundary layer reattaches and forms a closed separation bubble.
April 2018; published online 30 July 2018. Copyright © 2018 by Authors, However, in three-dimensional (3-D) interactions, the separated
FCAAP. Published by the American Institute of Aeronautics and boundary layer forms an open separation bubble, with reverse flow
Astronautics, Inc., with permission. All requests for copying and permission velocities in the in-plane direction and significant downwash of the
to reprint should be submitted to CCC at www.copyright.com; employ the fluid in the out-of-plane direction [1]. Some notable reviews of the two-
ISSN 0001-1452 (print) or 1533-385X (online) to initiate your request. See
dimensional interactions have been compiled by Delery and Marvin [2],
also AIAA Rights and Permissions www.aiaa.org/randp.
*Graduate Research Assistant, Department of Mechanical Engineering,
Viswanath [3], Clemens and Narayanaswamy [4], and Gaitonde [5].
FAMU-FSU College of Engineering Florida Center for Advanced The present fin generated interaction is considered semi-infinite,
Aero-Propulsion (FCAAP). Student Member AIAA. which implies that changes in the shock generator size/dimensions do
† not affect the interaction. Furthermore, because the sharp fin imposes
Post-Doctoral Research Associate, Department of Mechanical Engineering,
FAMU-FSU College of Engineering Florida Center for Advanced Aero- no additional length scales in the interaction (e.g., as opposed to a
Propulsion (FCAAP); currently Assistant Professor of Practice, Department of blunt fin where the fin leading edge thickness may impose an
Mechanical and Aerospace Engineering, Syracuse University, Syracuse, NY additional length scale), this interaction is dimensionless. The only
13244. Professional Member AIAA. dimension imposed on the present class of interactions is due to the
‡
Graduate Research Assistant, Department of Mechanical Engineering, incoming boundary-layer thickness. As discussed next, the influence
FAMU-FSU College of Engineering Florida Center for Advanced
Aero-Propulsion (FCAAP). Student Member AIAA. of this length scale is confined to a small region. A comprehensive
§
Professor, Department of Mechanical Engineering, FAMU-FSU College review of the classification and earlier research in shock/boundary-
of Engineering Florida Center for Advanced Aero-Propulsion (FCAAP); layer interaction has been presented by Settles and Dolling [6,7].
currently Don Fuqua Eminent Scholar and Associate Dean for Research. Sharp fin-generated swept-shock/boundary-layer interactions, the
Associate Fellow AIAA. focus of the present study, are widely accepted to be quasi-conical in
Article in Advance / 1
 2 Article in Advance / ARORA ET AL.

nature, where the flowfield displays conical symmetry with respect to measurements, surface oil-flow visualization has also been used for
an origin [7–10]. The “footprint” of such an interaction is illustrated initial characterization and identification of the topology of the
in Fig. 1, where the key features are shown such as the upstream interaction. This paper presents selected results of a larger ongoing
influence line, separation line, inviscid shock trace, and the investigation that aims to gain further insight into the mean and
attachment line. The extrapolation of these features appears to unsteady characteristics of three-dimensional SBLIs.
converge at a point upstream of the fin leading edge, referred to as the
virtual conical origin (VCO) [11]. Another important aspect of this
flow is the deviation from the conical behavior in the vicinity of the II. Experimental Setup
fin leading edge. This region is formed in response to the initial A. Wind-Tunnel Facility Description
condition imposed by the boundary-layer thickness δ. The flow The supersonic wind-tunnel facility at the Florida Center for
inside this region does not follow the conical symmetry and has a Advanced Aero-Propulsion of Florida State University is used for
characteristic length called the inception length Li . This region is conducting the experiments. It is a blowdown-type facility, where
known as the “inception zone” [11,12]. Past experimental studies dry, high-pressure air is supplied from six reservoirs. The total
[7–12] have demonstrated that, at least as far as the surface properties capacity of the reservoirs is 114 m3 with a maximum pressure of
are concerned, the flow possesses conical symmetry with respect to 500 psi, and they are charged with three Belliss & Morcom 156 kW
the VCO. (209 hp) reciprocating air compressors. The dew point of air is
Flow unsteadiness is another inherent feature in SBLI, where it is reduced to a nominal value of 228 K (−50°F) via Parker Airtek
characterized by a very large range of frequencies from 0.01 U∞ ∕δ to TX100 desiccant driers before storing. A run time of approximately
U∞ ∕δ. At present, there is still a lack of consensus regarding the 10 min can be easily obtained. The control of the facility is achieved
source of this unsteadiness, whether from the incoming boundary using a LabView-based program. The program actuates a Leslie
layer [4] or the separation bubble [13,14]. It has also been proposed controls D series valve with a Moore 760E positioner for regulating
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

that the upstream influence is more likely to be dominant in weaker the air supply to the tunnel. It is also used to measure and record the
2-D interactions, and the influence of upstream forcing decreases run conditions of the tunnel. It is possible to vary the Mach number of
with increasing interaction strength [4]. Overall, although there is no the facility by the use of interchangeable nozzle blocks. The test
consensus about the unsteady nature of the 2-D interaction, an section of the tunnel is 76.2 by 101.6 mm in height and width,
extensive knowledge base exists. However, the same is not true for respectively. The tunnel has optical accessibility from three sides.
3-D interactions, where quantitative measurements are much more The current experiments are conducted at a nominal freestream Mach
limited. A study by Schmisseur and Dolling [15] has shown that 3-D number M∞ of 2, stagnation pressure Po of 51 psi, and stagnation
swept-shock interactions, similar to 2-D interactions, are also temperature T o of 296 K. During a run, the stagnation pressure was
intermittent. A later study by Garg and Settles [16] analyzed the aft maintained within
0.5 psi of its nominal value, and the nominal
regions of the swept-shock interactions, where strong, low-frequency stagnation temperature decayed by approximately 0.5 K at the end of
oscillations are observed. The author associated them with the the run. It is estimated that these changes in flow conditions caused a
low-frequency random motion of the primary attachment line. variation of 0.9% in the nominal unit Reynolds number value.
The focus of the earlier examinations of this flowfield has been Finally, the layout of the tunnel is shown in Fig. 2 [19].
mainly on understanding the surface topology using surface oil-flow
visualization and surface pressure measurements [15,16], whereas B. Model Details and Test Conditions
flowfield studies conducted above the surface have mostly relied on An unswept fin is designed that is mounted on a flat plate (ceiling
qualitative techniques such as conical shadowgraphy and planar of the test section). The included angle of the fin is 7 deg, and the
laser scattering (PLS) [8,17,18]. The present research aims to length L is 76.2 mm. The height of the fin (H) is 37.5 mm. It is
quantitatively investigate the detailed mean flowfield of this 3-D, determined based on the incoming boundary-layer thickness and is
dimensionless interaction in the conical frame of reference using designed to produce a semi-infinite interaction. Fin angle of attack of
velocity field measurements. The interaction is generated by an 10 and 15 deg are tested in the current experiments. The normal Mach
unswept fin placed at an angle of attack α. The oblique shock number Mn based on the angle of attack is 1.3 and 1.4, respectively.
produced by the fin creates an adverse pressure gradient, which in The incoming boundary layer has a nominal thickness δ of 3.5

turn causes the incoming boundary layer to separate in the vicinity. 0.2 mm as measured from the velocity field results. The boundary-
The separation region thus formed leads to undesirable effects such as layer thickness is also theoretically estimated using the classical
high fluctuating pressure load, high thermal loads, and reduction in approach outlined by Van Driest [20]. It is calculated to be 3.4 mm,
total pressure recovery. The interaction strength of the oblique shock which is in good agreement with the experimentally measured
with the boundary layer can be varied by changing the angle of attack value. Moreover, contamination in particle image velocimetry (PIV)
of the fin or the freestream Mach number. Before the velocity field measurements near the surface due to laser reflections inhibited

Fig. 1 Footprint of the swept-shock interaction with marked surface features.

 Article in Advance / ARORA ET AL. 3

Fig. 2 Layout of the supersonic wind-tunnel facility [19].

detailed boundary-layer profile measurements with reasonable sub-micron-size particles. It is estimated that, for more than 95% of
accuracy. Therefore, boundary-layer integral parameters such as the particles created by the seeder, the particle diameter would be less
shape factor or momentum thickness are not included. Finally, the than 0.5 μm, and the nominal particle diameter size would be at most
unit Reynolds number for a nominal test condition is 4.37 × 106 ∕m. 0.3 μm (see Alkislar [22] for detail). A water base solution of glycol
(Rosco fog fluid) is used as the liquid for seed generation. The seed is
C. Experimental Techniques introduced upstream of the stagnation chamber of the wind-tunnel
1. Surface Oil-Flow Visualization facility to facilitate proper mixing of the seed with the flow.
A LaVsion Pro X camera with a resolution of 1600 × 1200 equipped
Surface oil-flow visualization is an extensively used technique to
with a Nikon 55 mm objective is used to acquire the images. DaVis
study the surface topology of SBLI. This method requires a mixture
8.1.4 software is used to record the images, and a total number of
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

of oil and a dye. The mixture is applied at strategic locations on the

1000 image pairs are obtained for each case. DaVis 8.2.2 is used for
model, which generates a surface streakline/skin-friction line pattern
processing the images. First, the background noise is removed to
when exposed to the flow [21]. For blowdown-type supersonic
enhance the signal-to-noise ratio. Subsequently, a multipass scheme
tunnels, kerosene oil is ideal because it evaporates during the run, and
is applied, starting from two passes of 48 × 48 pixel interrogation
the pattern is not affected by tunnel shutdown. In the current study, a
window to two passes of 24 × 24 pixel interrogation window. For
mixture of kerosene and fluorescent dye was applied upstream of the
each pass, an overlap of 75% and the Gaussian weighting function
fin and at the fin–plate junction. A commercially available UV light are applied. As a result, a vector resolution of 2.5 vectors∕mm is
was used to improve the contrast of the fluorescent dye. Various oil to achieved. Finally, Westerweel and Scarano’s universal outlier
pigment mixture ratios were tried, and a 1∶4 oil-to-pigment ratio was rejection method [23] is used to remove the spurious vectors between
found to work optimally. The wind tunnel was run for 5–6 s. Images each pass. The results are postprocessed for further outlier rejection
of the surface flow patterns were captured using a Nikon D3200 based on the peak value of the correlation, universal outlier detection,
digital single-lens reflex camera with a 18–55 mm lens. and the size of the vector group.
An alternate coordinate system (shown in Fig. 3b) is used to
2. Velocity Field Measurements in a Conical Frame present the velocity field results in this paper. The modified
Two-component PIV is used to obtain the velocity field coordinate system (x 0 , y 0 , z 0 ) is defined such that x 0 is in the in-plane
measurements. The measurements are carried out in the conical direction, y 0 is along the wall-normal direction, and z 0 is along the
reference frame because the flow exhibits quasi-conical flow inviscid shock. A spherical–polar coordinate system (β, ϕ) centered
similarity. The schematic of the PIV setup is shown in Fig. 3a. The at the VCO (shown in Fig. 3b) is also used in presenting the velocity
measurement planes are aligned such that they are perpendicular to field results. From here onward, the following convention is used: the
the oblique shock at a distance of 0.5L and 0.66L from the fin apex. velocity component in the x 0 direction is called the in-plane velocity
The laser sheet superimposed on the surface footprint is also shown in U, and the velocity in y 0 direction is referred to as the wall-normal
Fig. 3a, where the optical axis of the camera is normal to the laser velocity V. To see the effect of using a different coordinate system, the
plane (along with the inviscid shock trace). A New Wave Nd-YAG velocity field is presented in both the orthogonal Cartesian (x 0 , y 0 , z 0 )
laser combined with the light-sheet optics is used to create a thin laser and the spherical–polar (β, ϕ) coordinate systems. The in-plane
sheet (about 1.5 mm). The laser has a repetition rate of 14 Hz, and the component of the freestream velocity U∞  368 m∕s is used to
pulse separation dt is set to 1.2 μs. To reduce the contamination in normalize the velocity field. It should be noted here that the currently
near-wall measurements due to reflection of the laser sheet from the studied flowfield is highly three-dimensional, with a significant out-
surface, the plate surface is covered with 3M orange fluorescent tape, of-plane velocity component (along with the z 0 direction). Thus,
and a band-pass filter (532
10 nm) is mounted on the camera lens. using two-component PIV technique to measure the in-plane velocity
A modified Wright nebulizer is used to seed the flow with field will include a bias due to the unresolved component and yield

Fig. 3 Setup and frames of reference for PIV measurements (not to scale).
 4 Article in Advance / ARORA ET AL.

higher in-plane velocities [24]. Following the analysis suggested by The distributions of uncertainty for the present flowfield at two
Prasad and Adrian, this error is estimated to be close to 10% and has measurement plane locations (0.66L and 0.5L) are shown in Fig. 4,
been accounted for in the corrected in-plane velocity of 368 m∕s. where uncertainties from all the sources are combined in the
Finally, all length scales are normalized by the height of the fin (H). uncertainty field shown. A mean in-plane velocity U∞ of 368 m∕s is
used to normalize the uncertainty values (σ∕U∞ ). Important features
3. Measurement Uncertainties of the interaction are superimposed in white dashed lines. It can be
The static and stagnation pressures of the wind tunnel are sampled observed that the maximum uncertainty of 6.3% occurs in the region
at 1000 Hz using an NI 9205 data acquisition module (16 bit) beneath the separation bubble, where the relative uncertainties
installed in an NI CDAQ 9188 chassis. The uncertainty associated quoted here are with respect to the mean in-plane velocity. However,
normalizing the uncertainty with the local mean velocity is another
with the measurement of static and stagnation pressure of the wind
way to determine the relative accuracy of the measurement. When
tunnel is found to be
0.04 psi. Wind-tunnel Mach number is
using the local mean velocity as the reference, a maximum uncertainty
computed using a ratio of tunnel static and stagnation pressure.
of approximately 25% was observed. This relative error is higher
Uncertainty associated with the Mach number due to the error
because the mean velocity of 98 m∕s is used for normalization, which
propagated from the primary measurements is estimated to be
1% is much lower than the mean freestream in-plane velocity.
of its nominal value of Mach 2.
The uncertainties in PIV measurements are quantified using the
correlation statistics method [25] included in the DaVis 8.1.4 software. III. Results and Discussion
This method is a postprocessing technique based on statistical analysis
of the correlation process using differences in the intensity pattern of Before delving into the results of the current study, a brief review of
the two PIV snapshots. Because this method only uses the information the most relevant results from earlier experimental studies is briefly
discussed to better orient the reader regarding what has been learned
within a correlation map, the computed uncertainty should account for
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

and characterized about this interaction. Mainly, flow visualization

all the combined factors of uncertainty that may affect the correlation
techniques such as the conical shadowgraphy and planar laser
function. Moreover, because the current study is of a high-speed
scattering (PLS) have been used to provide insight into the structure of
compressible flowfield, the uncertainty associated with the particle lag
the interaction [18]. Figure 5a is a shadowgraph image obtained using
can also be important. Modified Stokes’s law is used to estimate the conical shadowgraphy revealing some of the pertinent features of the
particle lag. Additionally, it was also noted that changes in the interaction, such as the separation shock, rear shock, slip line, and
stagnation temperature throughout the wind-tunnel run and small separation bubble (flow in the figure is from left to right). The flow was
discrepancies in the laser sheet alignment with respect to the inviscid also studied using the PLS technique, in which the flow was seeded
shock angle could result in another 3% error in the measured in-plane with acetone and illuminated with a laser sheet. Information from
velocity field. Therefore, to obtain a conservative estimate of the different techniques is combined to make flowfield models; the spatial
uncertainty field, contributions of the uncertainty due to the correlation location of the dominant interaction features are extracted from PLS;
error, particle lag, and other miscellaneous factors are combined to get and small-scale detailed features are derived from shadowgraph
the total uncertainty. images at the same test conditions, which are combined with the
Calculation of uncertainty due to particle lag using the modified surface features extracted from the surface oil-flow method. Such a
Stokes’s law is discussed here. It is known that a particle has some flowfield map is shown in Fig. 5b. The present investigation builds
inertia associated with it; therefore, a lag is introduced in the response upon such earlier studies with the aim of providing more quantitative
of the particle to the flow. Particle response can be quantified using results in the flow above the surface. Besides providing necessary
the relaxation time τp calculated using the modified Stokes’s physical insight, such measurements are also serving as rigorous
law [26]. The value of τp is found to be 1 μs for the current study. The validation database for companion computational studies [28].
selection of characteristic length is based on the maximum velocity
gradient that the particle may experience in the flowfield, and it is thus A. Surface Flow Visualization
used to calculate a conservative error estimate due to particle lag. In An oil and dye mixture is applied to the surface of the flat plate to
our case, the maximum gradient is across the inviscid shock. The obtain the streakline traces shown in Fig. 6. The surface patterns are
characteristic flow time scale τf based on a mean in-plane velocity of obtained at two interaction strengths Mn  1.3 (α  10 deg) and
368 m∕s is computed to be 10 μs. Stokes’s number (the ratio of the Mn  1.4 (α  15 deg), but the surface flow visualization for only
particle relaxation time to the characteristic flow time) is calculated to the higher interaction strength case is shown because it is more
be 0.1. Using the approach described by Samimy and Lele [27], it is pertinent to the current study. Many important features of the
estimated based on the calculated Stokes’s number that the mean interaction are visible in the surface flow and are marked in the figure
velocity field is captured with an estimated uncertainty of 2%. where the flow direction is from left to right. The extrapolated

Fig. 4 Uncertainty field of the in-plane velocity U (critical features of the interaction are shown in white dashed lines).
 Article in Advance / ARORA ET AL. 5

Fig. 5 Interaction of the fin at an angle of attack, α
20 deg with M∞
2.95 (Mn
1.85) flow [18].

Azimuthal angles of the upstream influence line and separation line

Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

are also measured and are shown in Table 1. Additionally, the

azimuthal location of the upstream influence line is also estimated
form the velocity profiles extracted from the velocity fields discussed
in the next section. The β location where a noticeable change is
observed in the boundary-layer edge velocity is noted and found to be
in agreement with the value shown in Table 1.

B. Velocity Field
Results from the PIV measurements conducted in a conical frame
of reference are discussed here. The stronger interaction case
corresponding to Mn  1.4 (M∞  2; α  15 deg) is selected for
the PIV measurements, and the velocity fields are obtained at two
different locations (0.5L and 0.66L) from the fin apex, as shown in
Fig. 6 Kerosene-fluorescent dye traces for the Mach 2, α
15 deg Fig. 7. In the figure, α is the fin angle of attack and βA , βo , and βS
(Mn
1.4) case. The white dashed lines show the upstream influence correspond to the azimuthal locations of the attachment line, inviscid
line, separation line, and shock trace. shock trace and separation line, respectively.

upstream influence line and the separation line intersect upstream of 1. Spherical–Polar Coordinates
the fin edge at the VCO, the origin of the conical symmetry of the It is important to clarify the coordinate systems used in this study
flowfield. Knowing the test conditions (M∞ and α), the inviscid before presenting the velocity field results. As noted earlier, PIV
shock angle is calculated and projected on the surface as a “shock measurements are acquired in a plane at two different distances (0.5L
trace” line. As seen in Fig. 6, the shock trace also intersects with the and 0.66L) away from the VCO of the interaction. One such plane,
upstream influence line and separation line at the VCO. It is well shaded green, is shown in Fig. 8. To present the velocity field results,
known that, for these flows, there exists a small region close to the fin a modified coordinate system with respect to this plane is defined,
tip where the flow deviates from conical symmetry. This region is denoted by x 0, y 0 , and z 0 . It is defined such that the x 0 direction is
known as the inception zone. Inside the inception zone, the surface perpendicular to the inviscid shock trace (in-plane direction), the y 0
traces can be seen deviating from the separation line (shown with a direction is normal to the flat plate, and the z 0 direction is along the
white dashed line) close to the fin apex. The formation of inviscid shock trace. However, the interaction is quasi-conical in
the inception zone is the response of the flowfield to the length scale nature, and the best way to represent it is using a spherical–polar
imposed by the boundary layer δ. As discussed by Lu [29], the coordinate system (β, ϕ) centered at the VCO. Such a coordinate
flowfield inside the inception zone becomes dimensional, and system is also shown in Fig. 8. The spherical–polar coordinates (β, ϕ)
therefore it is possible to define a length scale, inception length li are obtained by converting the Cartesian coordinates (x 0 , y 0 , z 0 ) using
inside the inception zone. The flowfield is dimensionless outside of a linear transformation. This transformation can be performed by the
the inception zone. Finally, it should be noted that the surface trace following two ways. The first method is to assume that the spherical–
downstream of the separation line does not appear to strictly behave polar coordinate of each point is the same as if it were on a spherical
as skin-friction lines. This may in part be due to the consistency of the shell tangent to that location and had the specific radius of that point.
mixture, which may be a bit thick to faithfully follow the surface flow The second method is to assume that all points in the measurement
in the separated region. Moreover, a pattern initially established plane correspond to a spherical shell with the radius equal to the
during startup may indeed contaminate some of the features where
the dye mixture cannot easily reach once the interaction is Table 1 Spatial and azimuthal location of the critical features from
established. However, the upstream influence and the separation lines the surface flow traces at two interaction strengths, Mn
1.3 and 1.4
are well resolved because this part of the flow has more than sufficient
tracer available after the startup transient. α  10 deg; α  15 deg;
Surface feature Mn ≃ 1.3 Mn ≃ 1.4
From the surface traces obtained in the current study, inception
length li and the distance of the VCO from the fin apex (when moving Inception length, mm 45 23
Distance of VCO from fin apex, mm 65 50
along the inviscid shock trace) are measured and shown in Table 1 for
Upstream influence line, deg 45 54
both the interaction strengths. These values are found to be in Separation line, deg 43 51
reasonable agreement with the earlier published results [29].
 6 Article in Advance / ARORA ET AL.

passes through the λ-shock structure. Farther downstream, some of

the fluid at reattachment is directed back into the separation bubble,
resulting in reverse flow in this frame of reference, which is also
clearly seen with the near-wall streamlines. This reverse flow is due to
the presence of a flattened vortex inside the separation bubble. The
fluid above the separation bubble is also deflected upward and
downward by the separation shock and rear shock, respectively. After
passing through the rear shock, this fluid turns and accelerates in the
region bound between the slip line and the separation bubble due to
the presence of expansion waves that are reflected as compression
waves from the slip line. The behavior observed in these direct
measurements of the velocity field supports the observations made
through the conical shadowgraphy images, such as the one shown in
Fig. 7 Top view of the fin geometry with superimposed PIV
measurement planes aligned normal to the inviscid shock (not to scale).
Fig. 5a, where strong compression waves are seen. The accelerated
flow impinges on the plate surface and has been referred to as an
impinging jet by Alvi and Settles [18]. The acceleration and turning
of the flow in the impinging jet region can be observed with the
velocity vectors superimposed on the in-plane velocity field as shown
in Fig. 10a. Upon impingement, this jet is also the primary cause of
adverse effects such as high local thermal and surface pressure loads
in the vicinity of the fin–plate junction.
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

Although most of the interaction features had been qualitatively

observed in prior studies, these velocity field measurements provide
the first direct, quantitative results regarding the flow behavior. For
example, the relative strengths of the shock features can now be
determined by comparing the in-plane velocity jumps across each
feature. The extracted in-plane velocity profiles across the different
interaction features are shown in Fig. 9c, where the locations of the
inviscid shock (IS), separation shock (SS), and rear shock (RS) are
derived from the dilatation field and are shown. The velocity jump
across the inviscid shock can be seen from the extracted velocity
Fig. 8 Schematic of the spherical–polar coordinate system (VCO: profile at y 0 ∕H  0.6, where the velocity after the inviscid shock
virtual conical origin, SL: separation line, inviscid shock trace, AL:
attachment line, and α is the fin angle of attack).
decreases to almost 70% of the freestream velocity. From the velocity
profile derived at y 0 ∕H  0.3, the change in velocity across the
separation and rear shock is also estimated to be 10 and 15%,
respectively. Here, one may observe a very small bump in the in-plane
velocity in the vicinity of the separation shock for y 0 ∕H  0.3
distance measured along the inviscid shock trace on the surface. velocity profile. However, when accounting for the uncertainty in
However, because the solid angle subtended by the interaction in a finding the separation shock location and velocity magnitude (shown
spherical frame of reference is relatively small, the approximation of as horizontal and vertical error bars), this bump lies well within
a plane for a spherical sector of constant radius is a valid assumption. the uncertainty bars. Furthermore, the very high velocity in the
Thus, using the second transformation method projects the velocity impinging jet in the flowfield aft of the rear shock is evident from the
field onto the spherical–polar coordinates, which is shown as the red velocity profile extracted at y 0 ∕H  0.2. In Fig. 9c, after
spherical segment in Fig. 8. The transformed coordinates β and ϕ are x 0 ∕H > 0.8, the maximum velocity of almost 80% of the freestream
in the azimuthal and polar directions, respectively. Finally, the velocity is observed. This rise in velocity is approximately 13%
uncertainty introduced in estimating the spherical–polar coordinates higher than the resulting velocity after the inviscid shock. The jump in
due to the limited spatial resolution of PIV is calculated to be 1.2%. the flow velocity in the impinging jet region provides a quantitative
measure of the reason behind the high surface pressures and
2. Conical Symmetry and Flowfield Features temperatures observed on the surface. Additional flow properties,
The ensemble-averaged in-plane velocity fields measured in the such as vorticity and turbulence characteristics, are further discussed
Cartesian coordinate system at the two locations (0.66L and 0.5L) are in subsequent sections.
shown in Fig. 9. Critical features of the interaction such as the inviscid Another important observation derived from the velocity field is
shock, separation shock, rear shock, and slip line are shown in white the effect of the coordinate reference frame. Because the flowfield is
dashed lines and are superposed on the normalized in-plane velocity quasi-conical with respect to the VCO, when the distance from the
contours. The dilatation is used to identify the mean location of VCO is varied, the spatial extent of the critical features in the inviscid
the shock features (marked by dashed lines), where dilatation is part of the interaction shrink (moving toward the VCO) or expand
the divergence of velocity (dilatation  dU∕dx 0  dV∕dy 0 ) and (moving away from VCO). The velocity fields measured in the
represents compression or expansion of the flow. Also shown in current study at distances of 0.5L and 0.66L from the VCO are
Fig. 9a are black contour lines corresponding to the gradient of the in- projections of the 3-D flowfield on a plane at those two distances. It is
plane velocity in the y 0 direction. The location of the slip line, a shear anticipated for such quasi-conical flows that the flow structure is
layer emanating from the triple point, is extracted from these y 0 smaller at the 0.5L location as compared to the 0.66L location when
gradient of U-velocity contours. From the velocity contours, it can be looking at them in the Cartesian coordinate system. This effect can be
seen that the inviscid shock bifurcates into a λ-shock, where observed in Figs. 9a and 9b at 0.66L and 0.5L, respectively, where it
separation shock is the leading leg of the λ-shock. Because of the can be seen that the spread of the λ-shock structure, the height of the
presence of the separation shock, the incoming boundary layer triple point, and spatial location of the slip line are different for the
separates to form a free shear layer and is deflected away from the two planes, being larger for the 0.66L plane that is further away from
surface. The trailing leg of the λ-shock with respect to the incoming the VCO, as seen from Fig. 9a. The same velocity field results are also
flow is called the rear shock, which deflects this shear layer back presented in the spherical–polar coordinates in Figs. 10a and 10b for
toward the surface, causing it to attach, and an open separation bubble 0.66L and 0.5L planes, respectively. Once transformed into the
is formed. The streamlines superimposed on the in-plane velocity spherical–polar coordinates, the size of the λ-shock structure,
field shown in Fig. 10b clearly show the change in flow direction as it the height of the triple point, and spatial location of the slip line for the
 Article in Advance / ARORA ET AL. 7
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

Fig. 9 a–b) Representations of ensemble-averaged in-plane velocity field in Cartesian coordinates at two measurement planes shown in Fig. 7. Figure 9c
shows the extracted in-plane velocity profiles.

Fig. 10 Ensemble-averaged in-plane velocity field at the two measurement locations in polar coordinates (white dashed lines show the critical features of
the interaction).

two measurement planes are essentially the same in ϕ and β. This is in-plane velocity field. Most of these features are extracted using the
further verified next by extracting the locations of these primary dilatation field. The slip line, however, is better visualized from
features from the dilation field for the two measurement planes and the dU∕dy 0 contour lines because stronger gradients occur across the
overlaying them. slip line in the wall-normal direction. The extracted features from the
Further insight into the similarities and differences between the dilatation and in-plane velocity gradient in the y 0 direction are shown
flowfields measured at two radial locations can be obtained by a for the 0.5L plane in Fig. 11a. In Fig. 11b, an overlay of the primary
closer examination of the principal flow features observed in the features extracted from the two planes is shown. It is evident from
 8 Article in Advance / ARORA ET AL.

Fig. 11 Comparison of the extracted features (λ-shock structure and slip line) from the in-plane velocity fields at two measurement planes.
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

Fig. 11b that, in the spherical–polar coordinates, the location of the [11], Settles and Dolling [7], and Lu [29]), the current study, to the
principal features, especially the inviscid shock, separation shock, authors’ knowledge, provides the first direct quantitative evidence of
and rear shock, is in close agreement. However, some departures the quasi-conical nature of this flowfield above the surface.
from a perfect overlap are expected and indeed observed; hence,
the interaction is referred to possess quasi-conical symmetry. Some 3. Comparison of Flowfield Maps
disparities are due to the uncertainty associated with locating the The next step is to combine the information derived from the
features and the projection of a plane in Cartesian coordinate frame surface flow results and PIV velocity fields to create detailed
onto a spherical segment in spherical–polar coordinates. It can be also flowfield maps at the current interaction strength and compare them
seen that the inviscid shock is tilted, and there is a difference in the tilt to the flowfield maps from the earlier studies at the same interaction
angle between the two measurement planes. It is expected that as one strength (Mn ). An example of such a map from a prior study [18] is
moves up and away from the triple point, the conical zone of shown in Fig. 5b. These flowfield maps consist of the spatial location
influence from the top of the fin leading edge (given by the Mach of the crucial features of the interaction, such as the separation shock,
angle) does begin to intersect with the inviscid shock, which is very rear shock, slip line, triple point, and the region containing the
likely the main reason behind the observed “slant”. The influence of flattened vortex. Figure 12a shows a similar map created for the
this Mach cone will be higher for the 0.66L plane as opposed to the interaction using the present measurements, where the streamlines
0.5L plane. Such tilting/slanting of the inviscid shock above the triple are superposed to highlight the flattened vortex region and reverse
point has also been documented for interactions of similar strength flow. Features derived from the surface flow traces, such as the
(same Mn ) by Alvi and Settles [18] in their PLS images and the upstream influence (UI) line and separation line (S), are also marked
flowfield maps derived from them (see Fig. 12b). Finally, as one in the flowfield map for the present interaction. These flowfield maps
approaches the flat plate surface and the fin, the flowfield by are compared to the similar results from the previous study by Alvi
definition will depart from the conical symmetry (Figs. 3b and 7 and Settles [18], where they used qualitative techniques such as PLS
make it clear that the fin surface cannot lie along a ray emanating from to generate these flowfield models. A flowfield model from the
the VCO, unless the VCO and fin apex are coincident). Despite these previous study at freestream Mach number of 2.95 with a fin at an
small differences, the flowfield measured in the two planes collapses angle of 10 deg is shown in Fig. 12b. Between the two studies, the
very well. Although quasi-conical symmetry of the surface features freestream conditions are different, but they are nominally equivalent
has been clearly demonstrated in past seminal studies (Settles and Lu in strength in terms of Mn , the normal Mach number, which is

Fig. 12 Comparison of the flowfield map with previous results [18] at similar interaction strength, Mn ≃ 1.4.
 Article in Advance / ARORA ET AL. 9

Table 2 Comparison of PLS and PIV extracted from the current study and from Alvi and Settles is within
flowfield maps
0.5 deg. For the present study, this is due to the uncertainty in the
Flowfield map βUI -βo , deg βSS -βRS , deg azimuthal location of features measured using PIV, as noted in section
B-1, as well as the uncertainty associated with the upstream influence
PLS (Alvi and Settles) 7.2 6.5
Velocity field (present study) 7.7 7 line, which was extracted from surface flow visualizations. Thus, it
can be observed that there is a good agreement between the quantities
compared, which further emphasizes the importance of normal Mach
number Mn as a measure of the interaction strength.
approximately equal to 1.4 for both. The interaction is more
sweptback for the Mach 2.95, α  10 deg case when compared to 4. Out-of-Plane Vorticity
the current interaction conditions, which can be observed from the The distribution of out-of-plane vorticity computed for the two
different abscissa (azimuthal angle) of the flow features in two locations is shown superposed on the surface oil-flow visualization in
flowfield models. The different azimuthal angles impede the direct Fig. 13a, where vorticity is nondimensionalized using the boundary-
comparison of the location of the critical features; therefore, the layer height δ and the mean in-plane velocity U∞. In the contour
angular extent of the various features is extracted and compared. plots, the positive vorticity is directed toward positive z 0 direction,
These include the difference in the azimuthal angle of the upstream moving away from the fin apex along the direction of the shock trace
influence line with the inviscid shock angle (βUI -βo ) and spread of the (ST) line. The location of the surface features such as the upstream
λ-shock (βSS -βRS ). The inviscid shock angle βo is 27.8 and 45.3 deg influence line (UI) and separation line (S) is marked on the contour
for Mach 2.95 and 2, respectively. Azimuthal location of the plots. In this plot and subsequent similar contour plots discussed
upstream influence line (βUI ) and rear shock (βRS ) is approximately herein, the spatial location of the shock features and slip line extracted
estimated from the flowfield maps for the past study. However, for the from the in-plane velocity field is shown superposed on the contours
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

current study, they are derived from the surface flow and velocity with dashed black lines. Upon closer investigation of the vorticity
field results, respectively. The computed (βUI -βo ) and (βSS -βRS ) fields, it can be observed that a region of peak vorticity occurs inside
values are tabulated in Table 2. It is estimated that the uncertainty in the incoming boundary layer, before the upstream influence line.
the difference of the azimuthal location of the flowfield features This region of high negative vorticity close to the surface is distinctly

Fig. 13 a) Out-of-plane vorticity fields are shown. b) Vorticity profiles extracted from the 0.5L vorticity field for the following selected azimuthal
locations: β
55, 51, 42, and 35 deg.
 10 Article in Advance / ARORA ET AL.

seen in the vorticity profile extracted at β  55 deg shown in fin–plate junction (β < 34.5 deg) are not shown because they are
Fig. 13b. Between the upstream influence and separation line, this prone to higher noise levels. Alongside Fig. 14a, the profiles
vorticity is diffused, and the peak region continues to diminish until extracted at β locations corresponding to those in Fig. 13 (X1 to X4)
flow separation occurs (β  51 deg), where the boundary layer lifts are presented in Fig. 14b. Turbulence kinetic energy k is computed as
off the surface and forms a free shear layer. The fluid with high k  0.5  u 02  v 02 , and the square of the freestream in-plane
vorticity from the inner layers of the boundary layer feeds this shear velocity U2∞ is used for normalization. The uncertainty in the TKE is
layer, where the vorticity is diffused into a larger band due to estimated using the methods described by Sciacchitano and Wieneke
advection and viscous diffusion. The diffusion of vorticity in the [31] and is found to be within 5%. In the inviscid region of the flow,
wall-normal direction is distinctly observed from the extracted the presence of the separation, rear shock, and slip line has minimal
vorticity profile within the separated flow region (β  42 deg), effect on the turbulence of the incoming fluid, whereas high levels of
where the vorticity close to the surface approaches zero (or small turbulence can be observed in the vicinity of the inviscid shock from
positive values), whereas the peak vorticity now occurs inside the the contour plot. A rise in turbulence levels, however, can be clearly
shear layer. This shear layer with relatively higher vorticity is shown seen in the viscous region of the interaction. The incoming boundary
in the contours in green surrounded by the yellow color. Farther layer has relatively low levels of turbulence, but as the interaction is
downstream, the shear layer with moderately high vorticity can be encountered at UI, an increase in the TKE can be observed. In
observed rolling around the flattened vortex present inside the between UI and S, there is a small increase in the TKE, which
separation bubble. The presence of the flattened conical vortex has is followed by a sharp rise in TKE near the separation line
been shown computationally in the past studies by Knight and (β  51 deg). This high-turbulence region extends up to β 
Horstman [10] and Panaras [30]. An interesting aspect to note here is 40 deg inside the separation bubble. This increase in turbulence
that there exists a region underneath the separation where vorticity is levels can also be distinctly seen in the TKE profiles extracted at X2,
almost zero or positive, which is in contrast to the region around it. where a sharp rise in TKE occurs close to the surface at X2 location
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

This low vorticity zone is more clearly seen in the 0.5L plane location and persists up to and past β  42 deg. Farther downstream,
as red contours. This region is formed because the flow in this zone is moderately high turbulence levels occur inside the separated shear
reversed with respect to the freestream, which generates positive layer due to the advection and diffusion of turbulence because of the
vorticity, leading to cancelation with the vorticity of opposite sign in mixing caused by the flattened vortex. As seen from the TKE profiles
the surrounding region. The presence of this region can also be seen at X2 and X3 locations, the sharp TKE peak that existed at X2
in the vorticity profile at β  42 deg, where the maximum positive evolved into a broad hump inside the separated shear layer at X3.
vorticity value occurs very close to the surface for the 0.5L plane. Underneath the separated shear layer occurs a low-turbulence region
Similar observations about this low vorticity zone have been made in near the surface, which extends up to reattachment (region shown in
a past computational study [30]. Another feature that is clearly blue in the contour plot). Similar observations about this region
observed from the vorticity field is the “slip line” emanating from the extending from β  45 deg to β  35 deg have been previously
triple point and curving toward the plate surface. The fluid bound made from our discussion of the vorticity contour plots, where a zone
between the slip line and upper edge of the separation bubble is of small positive vorticity was seen. An alternative explanation for
accelerated due to the presence of the expansion fans. The this feature can be as follows: the reattachment streamline lies in the
acceleration of this fluid with respect to the flow above the slip line upper half of the incoming boundary layer, where the fluid with
generates positive vorticity, which is visible in the contour plots. In relatively low turbulence exists, as seen from the TKE contours. As
the aft part of the interaction, higher levels of vorticity occur due to the flow reattaches, this fluid with low TKE intensities becomes the
the reattachment of the flow onto the surface. As can be seen from part of the reversed flow and penetrates inside the recirculation
the profile extracted in the vicinity of the attachment line bubble to form this low turbulence zone. In the aft region of the
(β  35 deg), the vorticity is again mostly confined close to interaction, the turbulence levels are again moderately high because
the surface. of the presence of the flow impinging onto the plate surface, where
the flow velocity becomes almost 80% of the freestream velocity as
5. Turbulence Kinetic Energy observed from the in-plane velocity fields (shown in Fig. 10). The
The effect of this interaction on the flowfield turbulence levels is high turbulence levels in the attachment zone are confined to small
discussed here. The turbulence kinetic energy (TKE) contour plot for near-wall region, as seen by the red contours in Fig. 14a and the TKE
the 0.5L plane is shown in Fig. 14a, where the data close to the profile at β  35 deg.

Fig. 14 a) TKE distribution at the 0.5L plane is shown. b) TKE profiles from the 0.5L TKE field for the following selected azimuthal locations: β
55, 51,
42, and 35 deg.
 Article in Advance / ARORA ET AL. 11

IV. Conclusions TKE contours, it was observed that the interaction has minimal effect
The aim of the current study is to further advance the on the flow in the inviscid part of the interaction. However, in the
understanding of three-dimensional shock/boundary-layer interac- viscous region of the flow, a rise of the turbulence levels was seen.
tion using direct quantitative measurements of the flowfield above the A sharp rise in the TKE was observed at the separation line, which was
surface, complemented by surface flow visualizations. An unswept, followed by diffusion of turbulence because of the mixing caused by
the flattened vortex inside the separated shear layer. In the aft region of
sharp edge fin is used to produce a swept oblique shock that interacts
the interaction, turbulence levels increase again inside the impinging
with the incoming supersonic turbulent boundary layer. Initially,
jet region, where the in-plane velocity almost becomes equal to 80% of
surface flow visualization is used to explore the topography of the
the freestream. In summary, this study provided valuable physical
flowfield for two interaction strengths, Mn  1.3 (α  10 deg) and
insight into the flowfield through detailed quantitative measurements.
Mn  1.4 (α  15 deg), where Mn is the normal Mach number.
By acquiring velocity field measurements at locations, the quasi-
Primary surface flow features such as the upstream influence line,
conical nature of the interaction is confirmed and hence the use of Mn
separation line, and VCO are identified within the surface flow traces.
as an appropriate interaction strength parameter. Furthermore, the
The information from the surface flow results was used to guide the direct quantitative measurements obtained herein serve as a much-
velocity field measurements in the conical reference frame, where needed database for the validation of companion simulations of
detailed velocity fields of the stronger interaction case (Mn  1.4) three-dimensional SBLI [32].
were acquired above the surface. To the authors’ knowledge, these
are the first velocity measurements in this interaction and certainly
the first ones in the conical frame of reference. Acknowledgments
The velocity field measurements were acquired at two planes
The authors would like to thank our sponsors, Air Force Office of
located at the distance of 0.5L and 0.66L from the fin apex when
Scientific Research (AFOSR) and Florida Center for Advanced
moving along the inviscid shock trace. In the velocity field contours
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

Aero-Propulsion at Florida State University (FCAAP-FSU), for their

constructed from these measurements, primary interaction features
support. We also, want to thank the machinists R. Avant and
such as the inviscid shock, separation shock, rear shock, separation A. Piotrowski for their help with the model fabrication.
bubble with locally “reverse” flow, and slip line were clearly
identified quantitatively for the first time. The velocity profiles
extracted at various y 0 ∕H locations provided a direct measure of the References
significant changes in the velocity field where the flow shows a [1] Arora, N., Alvi, F. S., and Ali, M. Y., “Flowfield of a 3-D Swept Shock
relative deceleration of 30, 10, and 15% across the inviscid, Boundary Layer Interaction in a Mach 2 Flow,” 46th AIAA Fluid
separation, and rear shock, respectively. The velocity drops across Dynamics Conference, AIAA Paper 2016-3649, 2016.
different shocks render a measure of the strength of these features. doi:10.2514/6.2016-3649
Moreover, because of the interaction, a relative increase of almost [2] Delery, J. M., and Marvin, J. G., “Shock-Wave Boundary-Layer
13% in the in-plane velocity was observed in the impinging jet region. Interactions,” AGARD, Paper 280, Neuilly-sur-Seine, France, Feb. 1986.
This accelerated flow on impingement is responsible for many [3] Viswanath, P., “Shock-Wave-Turbulent-Boundary-Layer Interaction
adverse effects caused by SBLI such as the high surface pressure and and Its Control: A Survey of Recent Developments,” Sadhana, Vol. 12,
Nos. 1–2, 1988, pp. 45–104.
temperatures in the aft region of the interaction. Furthermore, doi:10.1007/BF02745660
Cartesian and spherical–polar coordinate reference systems were [4] Clemens, N. T., and Narayanaswamy, V., “Low-Frequency
used to present the velocity field, and the mean spatial locations of the Unsteadiness of Shock Wave/Turbulent Boundary Layer Interactions,”
critical flowfield features of the interaction were compared between Annual Review of Fluid Mechanics, Vol. 46, 2014, pp. 469–492.
the two planes. As expected, it was found that the extent of the doi:10.1146/annurev-fluid-010313-141346
λ-shock and the locations of other principal features were different in [5] Gaitonde, D. V., “Progress in Shock Wave/Boundary Layer
Cartesian coordinates within the two planes, where the λ-shock Interactions,” Progress in Aerospace Sciences, Vol. 72, Jan. 2015,
pp. 80–99.
structure was physically larger at the 0.66L plane when compared to
doi:10.1016/j.paerosci.2014.09.002
the 0.5L plane. There was, however, a very good overlap between the [6] Settles, G. S., and Dolling, D. S., “Swept Shock/Wave Boundary-Layer
spatial location of the inviscid shock, separation shock, rear shock, Interactions,” Tacticle Missile Aerodynamics, Vol. 104, Progress in
and slip line when the same velocity fields were presented in Astronautics and Aeronautics, AIAA, New York, 1986, pp. 291–319.
spherical–polar coordinates. This agreement of the flowfield features [7] Settles, G. S., and Dolling, D. S., “Swept Shock/Boundary—Layer
outside the inception zone confirms the quasi-conical nature of the Interactions—Tutorial and Update,” 28th Aerospace Sciences Meeting,
interaction, which has been observed in the past studies using surface AIAA Paper 1990-0375, 1990.
and qualitative flowfield techniques [8,12] but is verified herein for doi:10.2514/6.1990-375
[8] Alvi, F. S., “Flowfield Structure of Swept Shock Wave/Turbulent
the first time using direct quantitative measurements. Using the
Boundary-Layer Interactions,” Ph.D. Dissertation, Pennsylvania State
quantitative measurements obtained in this study, detailed physical Univ., State College, PA, 1992.
flowfield maps were created in the conical frame of reference for [9] Inger, G. R., “Spanwise Propagation of Upstream Influence in Conical
direct comparison to the similar flow models created using qualitative Swept Shock Boundary-Layer Interactions,” AIAA Journal, Vol. 25,
planar laser scattering (PLS) technique. Although the two studies had No. 2, 1987, pp. 287–293.
a different freestream Mach number and fin angle of attack, the doi:10.2514/3.9620
agreement was very good in terms of the principal features. This [10] Knight, D. D., and Horstman, C. C., “On the Quasi-Conical Flowfield
further confirms that Mn , the normal Mach number, is the appropriate Structure of the Swept Shock Wave Turbulent Boundary-Layer
Interaction,” 22nd Fluid Dynamics, Plasma Dynamics and Lasers
measure of the interaction.
Conference, AIAA Paper 1991-1759, 1991.
Additional flow properties such as vorticity and turbulence doi:10.2514/6.1991-1759
characteristics were also discussed. From the vorticity contour plots, [11] Settles, G. S., and Lu, F., “Conical Similarity of Shock/Boundary-Layer
it was observed that a region of peak vorticity occurs inside the Interactions Generated Byswept and Unswept Fins,” AIAA Journal,
incoming boundary layer, where the vorticity diffuses further Vol. 23, No. 7, 1985, pp. 1021–1027.
downstream into the shear layer around the flattened vortex due to doi:10.2514/3.9033
advection and viscous diffusion. Outside the separation bubble, [12] Lu, F. K., and Settles, G. S., “Inception Length to a Fully Developed
because of relative acceleration of the flow bound below and above Fin-Generated Shock-Wave Boundary-Layer Interaction,” AIAA
Journal, Vol. 29, No. 5, 1991, pp. 758–762.
the slip line, positive vorticity is generated and is clearly seen in the
doi:10.2514/3.10651
vorticity fields. Farther downstream in the aft region of the flowfield, [13] Dussauge, J.-P., and Piponniau, S., “Shock/Boundary-Layer Inter-
higher levels of vorticity occur at flow reattachment. The effect of the actions: Possible Sources of Unsteadiness,” Journal of Fluids and
presence of the interaction on freestream turbulence was also Structures, Vol. 24, No. 8, 2008, pp. 1166–1175.
analyzed using turbulence kinetic energy (TKE) fields. From the doi:10.1016/j.jfluidstructs.2008.06.003
 12 Article in Advance / ARORA ET AL.

[14] Piponniau, S., Dussauge, J., Debieve, J., and Dupont, P., “A Simple [24] Prasad, A., and Adrian, R., “Stereoscopic Particle Image Velocimetry
Model for Low-Frequency Unsteadiness in Shock-Induced Separation,” Applied to Liquid Flows,” Experiments in Fluids, Vol. 15, No. 1, 1993,
Journal of Fluid Mechanics, Vol. 629, 2009, pp. 87–108. pp. 49–60.
doi:10.1017/S0022112009006417 doi:10.1007/BF00195595
[15] Schmisseur, J., and Dolling, D., “Fluctuating Wall Pressures near [25] Wieneke, B., “Generic A-Posteriori Uncertainty Quantification for PIV
Separation in Highly Swept Turbulent Interactions,” AIAA Journal, Vector Fields by Correlation Statistics,” Proceedings of the 17th
Vol. 32, No. 6, 1994, pp. 1151–1157. International Symposium on Applications of Laser Techniques to Fluid
doi:10.2514/3.12114 Mechanics, Lisbon, Portugal, July 2014, pp. 7–10.
[16] Garg, S., and Settles, G. S., “Unsteady Pressure Loads Generated by [26] Melling, A., “Tracer Particles and Seeding for Particle Image
Swept-Shock-Wave/Boundary-Layer Interactions,” AIAA Journal, Velocimetry,” Measurement Science and Technology, Vol. 8, No. 12,
Vol. 34, No. 6, 1996, pp. 1174–1181. 1997, pp. 1406–1416.
doi:10.2514/3.13209 doi:10.1088/0957-0233/8/12/005
[17] Alvi, F. S., and Settles, G. S., “Structure of Swept Shock Wave/ [27] Samimy, M., and Lele, S., “Motion of Particles with Inertia in a
Boundary-Layer Interactions Using Conical Shadowgraphy,” 21st Fluid Compressible Free Shear Layer,” Physics of Fluids A: Fluid Dynamics,
Dynamics, Plasma Dynamics and Lasers Conference, AIAA Paper Vol. 3, No. 8, 1991, pp. 1915–1923.
1990-1644, 1990. doi:10.1063/1.857921
doi:10.2514/6.1990-1644 [28] Adler, M. C., and Gaitonde, D. V., “Unsteadiness in Swept-
[18] Alvi, F. S., and Settles, G. S., “Physical Model of the Swept Shock Compression-Ramp Shock/Turbulent-Boundary-Layer Interactions,”
Wave/Boundary-Layer Interaction Flowfield,” AIAA Journal, Vol. 30, 55th AIAA Aerospace Sciences Meeting, AIAA Paper 2017-0987, 2017.
No. 9, 1992, pp. 2252–2258. doi:10.2514/6.2017-0987
doi:10.2514/3.11212 [29] Lu, F. K., “Fin-Generated Shock-Wave Boundary-Layer Interactions,”
[19] Zhang, Y., “Three Dimensional Control of High-Speed Cavity Flow Ph.D. Dissertation, Pennsylvania State Univ., State College, PA, 1988.
Oscillations,” Ph.D. Dissertation, Florida State Univ., Tallahassee, FL, [30] Panaras, A. G., “The Effect of the Structure of Swept-Shock-Wave/
2017. Turbulent-Boundary-Layer Interactions on Turbulence Modelling,”
Downloaded by 34.195.88.230 on August 1, 2018 | http://arc.aiaa.org | DOI: 10.2514/1.J056500

[20] Van Driest, E. R., “Turbulent Boundary Layer in Compressible Fluids,” Journal of Fluid Mechanics, Vol. 338, May 1997, pp. 203–230.
Journal of the Aeronautical Sciences, Vol. 18, No. 3, 1951, doi:10.1017/S0022112097004825
pp. 145–160, 216. [31] Sciacchitano, A., and Wieneke, B., “PIV Uncertainty Propagation,”
doi:10.2514/8.1895 Measurement Science and Technology, Vol. 27, No. 8, 2016, Paper
[21] Settles, G. S., and Teng, H.-Y., “Flow Visualization Methods for 084006.
Separated Three-Dimensional Shock Wave/Turbulent Boundary-Layer doi:10.1088/0957-0233/27/8/084006
Interactions,” AIAA Journal, Vol. 21, No. 3, 1983, pp. 390–397. [32] Adler, M. C., and Gaitonde, D. V., “Unsteadiness in Shock/Turbulent-
doi:10.2514/3.8085 Boundary-Layer Interactions with Open Flow Separation,” 2018 AIAA
[22] Alkislar, M. B., “Flowfield Measurement in a Screeching Rectangular Aerospace Sciences Meeting, AIAA Paper 2018-2075, 2018.
Jet,” Ph.D. Dissertation, Florida State Univ., Tallahassee, FL, 2001. doi:10.2514/6.2018-2075
[23] Westerweel, J., and Scarano, F., “Universal Outlier Detection for PIV
Data,” Experiments in Fluids, Vol. 39, No. 6, 2005, pp. 1096–1100. Y. Zhou
doi:10.1007/s00348-005-0016-6 Associate Editor

View publication stats