You are on page 1of 11

JOURNAL OF SPACECRAFT AND ROCKETS

Preflight Ground Test Analyses of the Boundary


Layer Transition (BOLT) Flight Geometry

Heather E. Kostak∗ and Rodney D. W. Bowersox†


Texas A&M University, College Station, Texas 77845
https://doi.org/10.2514/1.A34858
Wind-tunnel experiments were performed to quantify boundary-layer instabilities both on and off the surface of a
33% scale model of the U.S. Air Force Research Laboratory/U.S. Air Force Office of Scientific Research Boundary
Layer Transition (BOLT) flight geometry. Measurements were obtained in the Mach 6 Quiet Tunnel and Actively
Controlled Expansion (ACE) tunnel located at the Texas A&M University National Aerothermochemistry and
Hypersonics Laboratory. Surface heating was quantified using infrared thermography. Under quiet conditions, heat
flux was characterized by a streak structure with no transition observed. Transition was observed within the
conventional noise ACE facility in two regions: 1) in the vicinity of a mixed-mode structure and 2) in an outboard
traveling crossflow region. The flow structure within a coherent vortex rollup located off centerline was further
characterized under quiet flow conditions with constant-temperature hot-wire anemometry and high-frequency
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

surface-pressure spectra. The measured mean and disturbance flow structure were in qualitative agreement with the
simulations. The hot-wire spectral content showed growth in the 25–40 kHz range, which also agreed with the
simulations. Surface-pressure spectral results showed similar growth in the 25–50 kHz range observed under both
quiet and conventional freestream conditions.

Nomenclature and computations rely on understanding the receptivity process, the


f = frequency linear stability of the problem, with the conclusion of nonlinear break-
K# = Kulite sensor at location number down in hypersonic flow. Early transition and stability work involved
M = Mach number 2-D planar and axisymmetric shapes with insight on the first- and
P# = PCB at location number second-mode instability mechanisms by Mack, Fedorov, and Zhong
pt2 = pitot pressure in test section, in which hpt2 i is equal to and Wang [2–4]. Geometries featuring a 3-D boundary layer gave way
mean and pt20 is equal to rms to a greater understanding of traveling and stationary crossflow, first and
q = heat flux, W∕m2 second modes, and even secondary instabilities [5–10]. The Hypersonic
Re = Reynolds number International Flight Research Experimentation (HIFiRE) program was
T = temperature, K developed to further understand boundary-layer transition for develop-
t = time ment of technology critical for the advancement in hypersonics. Many
uξ0 = freestream velocity perturbation ground tests and simulations have been made with a 38.1% scale
Vb = bridge voltage HIFiRE-5 model before and after the flight test [9–15]. Natural cross-
α = thermal diffusivity, m∕s flow transition was observed on the surface of the elliptic cone during
κ = thermal conductivity, W∕m ⋅ K quiet-tunnel experiments. It is expected that the crossflow instability is
ρU = mass flux, kg∕m2 ⋅ s present in flight. Stationary crossflow exhibits early nonlinear effects
with a strong dependence on surface roughness and receptivity [16–18].
Although it is predicted to have higher growth rates, traveling crossflow
I. Introduction is not observed to be the dominant mechanism in a low-disturbance
environment. Experimental studies by Neel et al. examined freestream
H YPERSONIC flight is of important national interest. Under-
standing the instability mechanisms leading to transition from
laminar to turbulence in hypersonic boundary-layer flow continues to
disturbance effects on the transition process on a 38.1% scale HIFiRE-5
model by comparing results in the Texas A&M University (TAMU)
be a challenge for current state-of-the-art tools. The instability mech- Mach 6 Quiet Tunnel (M6QT) and Actively Controlled Expansion
anisms are dependent on the geometry (e.g., concave/convex surfaces (ACE) tunnel [9,10]. Within the two freestream environments, the
and swept leading edges) and the environment. Current analyses transition process on the elliptic cone was qualitatively and quantita-
involve numerical simulation ranging from stability theory to direct tively different. Under noisy conditions, the crossflow induced transi-
numerical simulation (DNS) and ground test environments limited to tional and turbulent fronts were altered in shape at lower Reynolds
low-temperature quiet flow to high-enthalpy conventional flow. numbers. The near-centerline instabilities were more excited under
noisy conditions than in a lower freestream disturbance environment.
A. Background and Motivation Similar modal content was observed in the region of crossflow on the
Transition is an initial value problem. Disturbances within the free- model in both quiet and noisy environments, although the existence was
stream environment enter the boundary layer, a process known as triggered at a lower Reynolds number in ACE due to the freestream
receptivity, and cause steady and unsteady fluctuations [1]. Experiments content. The current analyses build on this experience in identifying
transition mechanisms for the U.S. Air Force Research Laboratory
(AFRL)/U.S. Air Force Office of Scientific Research (AFOSR) Boun-
Received 13 May 2020; revision received 14 July 2020; accepted for
publication 30 August 2020; published online Open Access 30 September dary Layer Transition (BOLT) flight test experiment.
2020. Copyright © 2020 by the American Institute of Aeronautics and
Astronautics, Inc. All rights reserved. All requests for copying and permission B. BOLT Flight Test Experiment
to reprint should be submitted to CCC at www.copyright.com; employ the To further challenge modern ground simulation and experiments to
eISSN 1533-6794 to initiate your request. See also AIAA Rights and Permis-
predict transition in flight, the AFRL/AFOSR BOLT flight test
sions www.aiaa.org/randp.
*Graduate Research Assistant, Aerospace Engineering. Student Member experiment was established. The BOLT geometry features concave
AIAA. surfaces with swept leading edges, including a 2-D leading edge, as
† shown in Fig. 1 [19]. The geometry contains four surfaces: symmetric
Ford I Professor, Department Head, Aerospace Engineering. Fellow
AIAA. upper and lower surfaces, where experimental measurements are
Article in Advance / 1
2 Article in Advance / KOSTAK AND BOWERSOX

Fig. 1 Rendering of the BOLT flight geometry (adapted from [19]).

Fig. 3 Isosurface of streamwise velocity perturbation for half of the


taken, and two “gutter” surfaces to isolate the flow from the swept 33% scale BOLT geometry, from [27].
leading edges. The idea behind the shape is to excite 3-D crossflow
instabilities [20] and to challenge current tools for transition predic-
tion. The BOLT geometry has been studied by Johns Hopkins in the streamwise direction of BOLT, which was near a row of Kulite
University Applied Physics Laboratory (JHUAPL), University of sensors.
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

Minnesota, AFRL/AFOSR, NASA, TAMU, and Purdue University.


The representative mean flow simulation for a quarter of the geometry C. Objectives and Approach
given in Fig. 2 shows a buildup of flow near the center, which indicates The objective of this paper was to help characterize boundary-layer
that the flow moves inward, creating an opportunity for crossflow instabilities through a recent set of experiments on a 33% scale model
instabilities, which was validated computationally [21–23]. The mush- of the AFRL/AFOSR BOLT flight geometry. The region of an off-
room-like structure contains stationary streamwise vortices, a similar centerline vortical mode identified in the simulations from the Uni-
effect seen on the HIFiRE-5 2:1 elliptic cone [24], which are dependent versity of Minnesota [25,27] was studied. Spectral contents within this
on Mach number and Reynolds number. A large counter-rotating region on and off the surface were quantified by using fast-response
vortex is seen off-centerline caused by a spanwise pressure gradient surface-pressure transducers and hot-wire anemometry, respectively.
effect created by the nose tip, which in turn causes vorticity within the Additionally, surface thermography was used to oversee the global
boundary layer. Previous simulations show that roughness amplifies heat flux on the model. Experiments were performed in the TAMU
the heat flux on the surface of BOLT [25], an effect also verified in M6QT and ACE tunnel.
wind-tunnel testing [26].
The complexity of this off-centerline vortex led to recent studies
focused on capturing instabilities off the surface of BOLT by forcing II. Experimental Overview
disturbances within the flowfield through DNS. A low-dissipation A. Wind-Tunnel Models
shock-capturing method was used to analyze the transition process The BOLT flight test geometry was designed by JHUAPL [19, 28].
[25,27]. Sparsity-promoting dynamic mode decomposition was used Experimental testing of two 33% scale models will be discussed in this
to extract amplified perturbations in the boundary layer from span- paper. Johns Hopkins University Applied Physics Laboratory designed
wise wall-normal momentum forcing, as shown in Fig. 3 [25,27]. and fabricated a 33% scale model that was tested at TAMU, Purdue
Traveling crossflow and brief upstream second-mode instabilities are University, and NASA Langley Research Center. The scale was chosen
present within the flowfield, but two additional structures are present based on testing requirements of the BOLT geometry within the Purdue
within the analysis: a vortical mode due to the rollup of the boundary Boeing/AFOSR M6QT (BAM6QT) [28]. Dimensions for the 33%
layer near the centerline and a mixed mode containing mixed fre- scale model are shown in Fig. 4. The machined model contains a
quency content that is oblique and localized near the boundary-layer polyetheretherketone (PEEK) plastic surface for infrared (IR) thermog-
edge [25,27]. The vortical mode is the primary focus of the off-body raphy with the remaining assembly made of 6061 aluminum. Surface-
measurements presented herein. The mode travels off the centerline pressure spectra and IR thermography results will be discussed for the
JHUAPL model. With spare machined aluminum parts provided
by JHUAPL, TAMU produced a duplicate of the 33% scale geometry
for off-body measurements in the M6QT using constant-temperature
hot-wire anemometry.

B. Facilities
Experimental testing of the BOLT geometry has taken place at
several wind-tunnel facilities, but the focus of this study is at TAMU
National Aerothermochemistry and Hypersonics Laboratory, which
features the M6QT [29] and the ACE tunnel [30]. The ACE tunnel is a
variable-Mach-number conventional facility and has previously
been modified for studies on the BOLT geometry [26]. The M6QT

Fig. 2 Mean flow of BOLT, from [20]. Fig. 4 BOLT geometry with dimensions at 33% scale.
Article in Advance / KOSTAK AND BOWERSOX 3

contains a polished nozzle with an exit diameter of 0.19 m. The


installation of the BOLT model within ACE and the M6QT is similar
to the procedure described by Neel et al. for the HIFiRE-5 model [10].
However, the purely aluminum TAMU model is present during
preheat in the M6QT.
Placement of the BOLT model within ACE and the M6QT is
shown in Fig. 5. For reference, the width of the entry door of ACE
is 0.5 m. The model must be placed inside the quiet core of the M6QT
not only to avoid the nozzle exit shock, but to obtain as much of a low-
freestream disturbance environment over the model [29,31,32]. To
help identify the placement of BOLT within the M6QT, the shock
location was provided by the University of Minnesota in Fig. 5 at
Fig. 6 Freestream disturbance environment in TAMU facilities.
M6QT freestream conditions of Re∕m  9.9 × 106 . The maximum
length of the model extending past the nozzle exit of the M6QT was
2.5 cm for both 33% scale models. The IR camera was angled to allow ∂T ∂2 T
more viewable area of the model within the quiet tunnel. The settling  α 2 ∼ T n1
i − T ni ∕Δt  αT ni1 − 2T ni  T ni−1 ∕Δx2 (1)
∂t ∂x
chamber in each facility is preheated to a stagnation temperature of
430 K before manual placement of the JHUAPL model. Both models The heat flux was then calculated using a second-order approxi-
for each run were mounted at 0 deg angle of attack and yaw. mation of the first spatial derivative using
A comparison of the freestream disturbance environment in ACE
 
and the M6QT is shown in Fig. 6. At a unit Reynolds number of about 3 1
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

3 × 106 , the nozzle sidewalls in ACE begin to transition, elevating the qn  −k − T n0  2T n1 − T n2 Δx (2)
2 2
overall noise field at higher Reynolds numbers. Noise levels reach
their highest in the M6QT around Re∕m  11.4 × 106 . where n is the time coordinate, and i is spatial. Two boundary
conditions were required: a surface condition, which is time depen-
C. IR Thermography dent on the surface temperature from the IR, and a depth condition,
The global surface heating on the JHUAPL model was viewed which was assumed constant within a defined depth of the model.
using an FLIR SC8100 IR camera. The camera has a resolution of Because of the low thermal conductivity of PEEK, the 1-D analysis
1024 × 1024 pixels with a sensitivity to 3–5 μm wavelengths. The was a reasonable approximation for results.
camera outputs 14 bit data, which are converted to temperature
through an in-house calibration by Leidy [34]. A 1-D heat transfer D. Surface-Mounted Pressure Transducers
algorithm code for converting temperature maps of the IR images into Analyses of three miniature Kulite XCE-062-15A surface-
heat flux was developed by Neel [35] with the boundary conditions pressure transducers located within the vortical mode are circled in
modified for the current analyses. The code is inspired by the red in Fig. 7. Their exact locations on the BOLT model are stated in
FORTRAN code QCALC [36] and configured to a more modern Table 1. These transducers have been characterized under a super-
approach for HIFiRE studies [7,37]. All experimental IR data were sonic turbulent boundary layer with a linear frequency response up to
processed using this code, which involves a forward-time central 30–40% of their resonant frequency [38]. Data were acquired using
difference scheme for solving the heat conduction equation, given by an in-house-developed LabVIEW VI with a pair of synchronized

Fig. 5 BOLT in the M6QT (left) from [33], the ACE tunnel (right), and corresponding shock structure at Re∕m  9.9 × 106 (bottom).
4 Article in Advance / KOSTAK AND BOWERSOX

Fig. 8 Photograph of the TSI 1220 probe with slack, from [33]; distance
between the two prongs is 1.27 mm.
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

Fig. 7 Off-centerline Kulite sensor locations analyzed, from [33]; PEEK


surface (left) and aluminum surface (right); length from top to bottom of
each surface is 0.26 m. Table 2 TSI hot-wire probe 1220-PI2.5
Processing Value
Diameter of sensing area, μm 6.3
Table 1 Location of off-centerline Kulite Frequency response, kHz 160
sensors Sample rate, kHz 500
High pass, kHz 1
x Location from the z Location from Low pass, kHz 200
Sensor leading edge, mm the centerline, mm
K1 209.25 6.35
K27 260.05 −6.35
K9 272.75 6.35 BOLT model installed in the M6QT with a hot-wire present is shown
in Fig. 9. Before tunnel start-up and shutdown, the wire is hidden
behind the exit of the nozzle, out of the flow, to limit wire breakage.
Measurements of Cartesian coordinates were converted to cylindrical
National Instruments USB-6366 (16 bit) DAQ systems. The Kulite for off-body measurements of BOLT.
transducers were connected to a custom-built power supply and
conditioning box with the circuit design modified by the one devel-
oped from the S. P. Schneider group at Purdue University [32]. The
box provides a dc and ac signal output for each sensor that is later
converted into pressure. The raw output is obtained by gaining
the input signal by 100 and low-pass filtering with a first-order
resistor–capacitor circuit at 482.5 kHz. The ac-coupled content is
separated by a high-pass filter of 842 Hz with a gain of 28.9. The
Kulite surface pressure transducers were conditioned with a low-pass
200-kHz eight-pole Bessel filter with unity gain. All channels were
acquired at 2 MS∕s per channel for 100 ms per sample. The alumi-
num surface in Fig. 7 has a slight discoloration from previous testing
using temperature-sensitive paint at Purdue.

E. Hot-Wire Anemometry
To quantify off-body content on the BOLT geometry, a TSI 1220
high-temperature hot-wire probe was used. The delicate sensor,
shown in Fig. 8 with a wire diameter of 6.3 microns, responds to
mass flux and total temperature. The slack in the wire alleviates strain
gauging under supersonic operation. The sensor is operated at a high-
temperature loading factor of τ  0.8 to minimize total temperature
sensitivity. The characteristics and processing technique for the probe
used for hot-wire measurements are given in Table 2. An A.A. Lab
Systems AN-1003 constant-temperature anemometer was used and
tuned manually during a wind-tunnel run. The frequency response of
the probe with respect to the 3 dB roll-off is listed in the table. The
sensor was acquired at 500 kHz for 100 ms per measured point
through an in-house LabVIEW code and DAQ system (same system
used by the Kulite sensors). The signals were passed through an
eight-pole 200-kHz low-pass Bessel filter with unity gain and an
eight-pole 1-kHz high-pass Bessel filter with a gain of 10.
The probe is mounted within the M6QT on a three-axis traverse Fig. 9 Hot wire located above the surface of BOLT in the M6QT,
previously shown and described in [33]. A photograph of the TAMU from [33].
Article in Advance / KOSTAK AND BOWERSOX 5

Table 3 PEEK property materials with uncertainty


Material property Value
Thermal conductivity κ, W∕m ⋅ K 0.29  6.9%
Specific heat capacity cp, J∕kg ⋅ K 1026  6.5%
Density ρ, kg∕m3 1300  0.4%

F. Uncertainty Estimates
Facility freestream conditions are computed using the settling-
chamber total temperature and pressure and the nozzle static pres-
sure. The total pressure within the ACE facility is measured using an
Endevco 8540-200 pressure transducer with a factory-stated uncer-
tainty of 2.4%. The total pressure in the M6QT is measured using an
MKS Baratron 615A capacitance manometer with an uncertainty of
0.12%. The static pressure in ACE is measured by a Baratron
631C-10 with a manufacturer-stated uncertainty of 0.5%. The
quiet-tunnel static pressure is measured by an MKS 902 vacuum
transducer with an uncertainty of 1.0%. The Mach number and
Reynolds number are calculated from the measured quantities. At
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

Re∕m  9.9 × 106 , the quiet-tunnel total pressure, total temperature,


and Mach-number standard deviation are 0.8; 0.5, and 0.6%,
respectively. At Re∕m  5 × 106 , the ACE tunnel total pressure,
total temperature, and Mach-number standard deviation are 1.4,
0.57, and 0.13%, respectively.
The model support apparatus used for BOLT in the M6QT was
previously aligned to 0 deg incidence in pitch and yaw using a 7 deg
cone described by Hofferth [32]. The angle was adjusted such that
the second mode frequency was consistent around the azimuth of the
cone. The uncertainty in the angle of the model with respect to the
support structure in the M6QT is 0.1 deg. In ACE, the model is
mounted on the door parallel to the flow. The uncertainty in the angle
of the sting mount located on the door with respect to the bottom of
the test section is 0.22 deg. Fig. 11 Heat flux results on the 33% JHUAPL model in the M6QT with
regions of modes identified from Fig. 3 (modified from [33]).
Uncertainty in the computations of heat flux from IR thermography
involves the properties of the material, camera angle, and calibration
of the camera. Of these parameters, the material properties are the The Kulite XCE-062-15A used for analysis has a factory-rated
main sources of uncertainty. The properties used in this study were uncertainty of 0.1% of its 103.5 kPa range. The Kulite surface
provided by JHUAPL and are listed in Table 3. The uncertainties are pressure transducers used at TAMU were calibrated at room temper-
based on nominal manufacturer documentation. Thermal conductivity ature under vacuum and experienced a ΔT of 50 K during a wind-
and density were seen to vary between 0.25 and 0.29 W∕m ⋅ K and tunnel run. This results in a 1% thermal zero and sensitivity shift.
1300 and 1310 kg∕m3 , respectively, at room temperature. Specific The power spectral density (PSD) plots for the Kulite sensors were
heat variances of 1026–1160 J∕kg ⋅ K were observed within the generated from 781 averages, corresponding to an uncertainty of
temperature range the PEEK surface experienced during testing. 3.6% for a PSD value. The spectral resolution for each point plotted in
Comparisons have been made between the IR thermography and the PSD was 1.95 kHz.
Schmidt Boelter heat flux gauges tested at NASA Langley Research The hot-wire results were acquired at 500 kHz for 100 ms. The
Center, and results were found to agree within 6% by Kostak [39]. PSD plots were generated by 97 averages resulting in an uncertainty

Fig. 10 Heat flux results on the 33% JHUAPL model in ACE (modified from [33]): a) Reynolds number sweep, b) adjusted contour levels to better show
regions of modes from Fig. 3.
6 Article in Advance / KOSTAK AND BOWERSOX

of 10.1% for each value. The spectral resolution for each point plotted The features observed on the BOLT surface were different between
in the PSD was 0.5 kHz. The location of the wire itself within the a quiet and conventional facility. In conventional flow, wedgelike
M6QT test section is dependent on the three-axis traverse hardware structures of higher heating originate at the back of the model shown
that is explained more in-depth by Craig [40] and Craig and Saric in Fig. 10a and travel upstream as the Reynolds number is increased.
[41]. The r axis uses a Faulhaber 2232S024BX4 AES-4096 brushless This effect was seen previously in the BAM6QT on the JHUAPL
dc motor with absolute encoder. The z and θ axes contain a Faulhaber model [26] and at NASA Langley Research Center on a phosphor
2250S024BX4 AES-4096 motor with absolute encoders. The tra- BOLT model [42]. Mode locations are labeled on the model in
verse was designed such that the r and z axes move with a resolution Fig. 10b at an adjusted contour level for comparison to the structures
of 244 nm per encoder line with an accuracy of 1220 nm [40]. The in Fig. 3. Label A represents the location for one of the two faint
θ axis moves with an accuracy of 3 × 10−5 deg per encoder line with streaks located off-centerline within the vortical mode. These two
an accuracy of 3 × 10−4 deg [40]. The relative position of the hot streaks were more prominent at higher Reynolds numbers in Fig. 10a.
wire with respect to the model surface was measured before each Label B in Fig. 10b corresponds to the location of the mixed-mode
tunnel run using a ruler with an uncertainty of 0.4 mm. region in Fig. 3 that is representative of a transition region on BOLT.
At low Reynolds numbers, streaks were observed within this region
and grew fainter with an increase in Reynolds number as the turbulent
III. Results wedges traveled upstream on the model. Label C is located down-
A. Flow Visualization and Heat Transfer stream in an outboard heating wedge that is close to the leading edge
Measurements of surface heating were acquired on the JHUAPL in Fig. 10b. This structure corresponds to the location of the traveling
model in the ACE and M6QT facilities. For each facility, the IR crossflow region indicated in Fig. 3 and is less visible with the current
camera was mounted above looking down into the test section. A color scale.
Reynolds-number sweep was performed during a single run in each The surface flow structure on BOLT in the M6QT in Fig. 11 is
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

facility. The Reynolds-number range for the ACE tunnel and the laminar with no transition to turbulence observed. A more elevated,
M6QT were Re∕m  2.8 × 106 to 8.0 × 106 and Re∕m  7.5 × 106 laminar streak pattern in the mixed-mode region on the model was
to 11.0 × 106 , respectively, for the majority of testing. Within the present under quiet conditions. Within the mixed-mode region, no
Reynolds-number range, snapshots were taken within the run and are modal growth was apparent in the surface-pressure sensors in either
shown in Figs. 10 and 11 for the ACE and M6QT, respectively, with TAMU facility. However, previous results on the model have shown
flow from left to right. Optical access was limited such that the back modal growth at this location under quiet flow at Purdue [27]. The
140 mm of BOLT was observable in ACE and 65 mm in the M6QT. two elevated heat streaks located off-centerline were more prominent

Fig. 12 PSD plots along the vortex structure in ACE (reprinted from [33]): a) Kulite 1, b) Kulite 27, and c) Kulite 9.
Article in Advance / KOSTAK AND BOWERSOX 7

in Fig. 11 under quiet flow than in conventional flow in Fig. 10. The Re∕m  3 × 106 . Around this specific Reynolds number in Fig. 6
Kulite sensor locations discussed in this paper are located near these is when the nozzle sidewalls began to transition in ACE, an
two off-centerline streaks. The higher outboard heating on the model effect observed in the findings from Neel et al. on the HIFiRE-5
in quiet flow in Fig. 11 is believed to be the result of noise radiating off geometry [10].
the nozzle walls. The flow structure on the subscale BOLT model The spectra for the M6QT in Fig. 13 are similar to the results from
under quiet flow conditions was previously compared to DNS and Fig. 12, but for higher Reynolds numbers. An instability in the 25–
computational fluid dynamics [23,25] with similar results between all 40 kHz range located within the off-centerline streak is not apparent in
groups. More results of the JHUAPL model in quiet and conventional the spectra at locations K1 and K27 until around Re∕m  8.5 × 106 .
flow can be found in Refs. [26,28,33,43,44]. The results shown in In Fig. 11, K9 was just shy of the elevated heat streak, which may be a
Fig. 11, especially within the inner 50% of the model, are represen- cause of the lack of growth in the spectra at this location in Fig. 13c. The
tative of these studies. noise interference from the nozzle wall at higher Reynolds numbers
could also have limited the appearance of the instability within
B. Surface-Pressure Spectra the measured spectra. The results in both Figs. 12 and 13 verify the
Surface-pressure spectra were acquired on the JHUAPL BOLT presence of an instability in both freestream environments within the
model to help identify instabilities at various frequencies in ACE vortical mode located off-centerline.
and the M6QT. The three Kulite sensor locations stated in Table 1 are
in the vicinity of the off-centerline vortical structure mentioned in C. Off-Body Measurements
Fig. 3. Power spectral density plots were computed using Welch’s Hot-wire measurements were performed on the TAMU BOLT
method with a hamming window for each of the transducers in each model to qualitatively compare to the off-body flow structures from
facility. The PSD results for the Kulite sensors in ACE are shown in the simulation results by the University of Minnesota [25,27]. The
Fig. 12. The legend in each plot represents the freestream Reynolds focus of the measurements was the vortical mode region indicated in
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

number with respect to the line that is plotted in million per meter. The Fig. 3 to better quantify the spatial distribution of the spectral content
frequency scale is plotted from 10 to 100 kHz. An instability in the within the roll-up structure for simulation validation. The streamwise
20–50 kHz range (centered near 35 kHz) was observed in all three location of interest was x  240 mm from the leading edge of the 33%
plots in Fig. 12 and grew in amplitude as the Reynolds number was scale model. A blown-up view of the vortical roll-up structure studied
increased. This instability is located within the elevated heat streak in in Refs. [25,27] is shown in Figs. 14a and 14b. The mean mass flux
the IR that becomes more prominent at higher Reynolds numbers in contour is shown in Fig. 14a (data provided by John Thome from the
ACE. The mode was first observed at the location of K1 around University of Minnesota), with the dominant frequency at this location
Re∕m  3.5 × 106 , and as it traveled downstream to locations plotted in Fig. 14b (data provided by Anthony Knutson from the
K27 and K9, the mode became apparent and began to grow at University of Minnesota). The x axis represents the z location in meters

Fig. 13 PSD plots along vortex structure in the M6QT (reprinted from [33]): a) Kulite 1, b) Kulite 27, and c) Kulite 9.
8 Article in Advance / KOSTAK AND BOWERSOX
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

Fig. 14 Simulations at M  6.0, Re∕m  9.9 × 106 at x  0.24 m: a) mass flux, and b) magnitude of the uξ0 mode.

from off-centerline. The y axis represents the location from the center The first comparison is the mean flow. At high overheat ratio, the hot-
of the geometry in meters, not the location from off the surface. The wire probe is primarily sensitive to mean mass flow [46], and for the
roll-up structure in Fig. 14a is of interest. This result is consistent with relatively low-density flow, King’s law exponent has been found to be
the mean flow and boundary-layer stability analyses of Mullen et al. close to unity [47]. With this, the mean mass flux is roughly proportional
[23]. A complete description of the numerical results in Fig. 14 is to the square of mean bridge voltage from the anemometry. Hence, the
provided in Refs. [25,27,45]. In summary, these plots show how the square of the bridge voltage from the experiments is shown in Fig. 15.
flow instabilities, indicated by larger fluctuation levels, are coupled to Hot-wire measurement locations are represented by the black dots in
the secondary flow structure. An experimental hot-wire campaign was Fig. 15 with the corresponding mesh of the contour. The x axis
performed to map the vortex region shown in Fig. 14. Listed in Table 4 represents the z location in meters from the centerline of the model,
is a summary of the traverse run measurement locations. The z and y and the y axis is from the center of the geometry in meters. The data
locations correspond to Fig. 14, but in millimeters. All hot-wire results
were acquired at a unit Reynolds number of 9.9 × 106 , which match the
conditions for the DNS data shown in Fig. 14.

Table 4 Hot-wire measurement locations

Run no. Z location, mm Y location, mm


4098 8.7, 9.3, 9.8 32–39
4099 10.3, 10.8, 11.4 32–39
4100 11.9, 12.4, 12.9 32–39
4101 13.5, 14, 14.5 32–39
4103 7.2, 7.7, 8.2 32–39
4104 5.6, 6.1, 6.6 32–39
4105 4, 4.5, 5.1 32–39
4108 15, 15.6, 16.1 32–39
4112 18.2, 18.7, 19.2 32–39
4113 19.8, 20.3, 20.8 32–39

Fig. 15 Visualization of the vortex structure from the hot-wire Fig. 16 Hot-wire results: a) rms∕V with arrow at z  10.8 mm and
measurements. b) PSD spectra at location z  10.8 mm; modal growth circled.
Article in Advance / KOSTAK AND BOWERSOX 9

frequency band, the top of the vortex structure in Fig. 17 has faded, the
content to the left of the structure is now visible, and the structure from
z  15–20 mm is not as prominent. The overall structure from the
experiments resembles that of the computation in Fig. 14b. To help
clarify the peak behavior spatially, line plots of the normalized energy
at 35 kHz and the rms energy within the 25–40 kHz band are plotted in
Figs. 18a and 18b, respectively, for z  10.8 mm, which corresponds
to the high-energy location in Fig. 16b. The peak and rms fluctuation
values increase by a factor of approximately 2.5 within the vortex
region (y  32–37 mm). Overall, the spectral content and location
Fig. 17 Hot-wire rms∕V results between the frequency bands 25– within the mean flow structure agree with the simulations from the
40 kHz with arrow at z  10.8 mm. University of Minnesota.

between z  16.1 and 18.2 mm were influenced by noise, and thus IV. Conclusions
discarded from the linear interpolation. The contour results are com- On-surface and off-body measurements were acquired on a 33%
pared to the simulation in Fig. 14a. The location and structure of the scale BOLT flight geometry and compared with computations and
vortex, as well as the boundary-layer edge and freestream environment simulations. Experiments were conducted in the M6QT and ACE
locations in Fig. 15, are qualitatively similar to those in Fig. 14a. The tunnel at TAMU. The following observations and conclusions were
experimental resolution was insufficient to fully resolve the detailed drawn. Three key regions of instabilities were identified in the
structure, as shown in Fig. 14a. The experiments appeared to have been numerical simulations: the outboard traveling crossflow region, the
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

shifted 1–2 mm as compared to the computation, which may have been near-centerline vortical flow structure region, and a mixed-mode
the result of a slight misalignment of the wind-tunnel model. Nonethe- region. The mixed modes and the off-centerline vortical mode were
less, the vortex structure was present in the experiment. identified as streaks of elevated heating on the surface in the experi-
The second comparison is a 2-D contour map of the fluctuations ments in both facilities. Under conventional freestream noise, the
within the vortex structure. The hot-wire rms of the fluctuating voltage leading transition front appears to coincide with the mixed-mode
normalized by the mean voltage is plotted in Fig. 16a. This ratio is interaction region from the simulation. A second transition front,
approximately proportional to the mass flux fluctuation ratio [48]. The located more outboard, appears to be correlated to the traveling
rms was calculated by numerically integrating the PSD spectra across crossflow instability from the simulation. Streamwise laminar streaks
all frequencies. The fluctuation energy within the vortex structure is were also present on the surface of BOLT in quiet flow, however, with
apparent and resembles the shape in Fig. 14b. Additional content is no transition observed. Spectral growth centered in the 20–50 kHz
seen closer to the surface by the hot wire between 15 and 20 mm that range was observed in the surface-pressure fluctuations within the
is not present in the simulations in Fig. 14b. The data that were vortical mode structure (6.35 mm off the centerline) in both quiet and
discarded may be of influence in amplifying these smaller structures. conventional flow. Additional off-body hot-wire spectra measure-
To quantify the frequency content within the vortex structure, power ments were acquired in this flow structure under quiet conditions to
spectra at various y locations are plotted in Fig. 16b at z  10.8 mm. provide characterization and quantification of the spatial content
The frequency scale is from 10 to 100 kHz with each PSD line within the rollup to help validate numerical simulations. The spectral
corresponding to a specific y location. Modal growth is apparent in content range between 25 and 40 kHz from the hot-wire anemometry
the range of 25–40 kHz in Fig. 16b as the hot wire moves in and out of measurements was consistent with the Kulite surface pressure trans-
the structure shown in Fig. 16a, that is, for y  33–37 mm. Strain ducer results and simulations under quiet flow at a streamwise
gauging is apparent at 20 and 60 kHz, but does not affect the overall location within the vortex structure. Overall, the frequency spectra
results. The PSD spectra were also integrated across the frequency and structure of the vortical mode from the experiments agreed
range of 25–40 kHz. These results are plotted in Fig. 17. Within this qualitatively with the simulations from the University of Minnesota.

Fig. 18 Hot-wire results at location z  10.8 mm: a) V 0 ∕V at 35 kHz and b) rms∕V results between the frequency bands 25–40 kHz.
10 Article in Advance / KOSTAK AND BOWERSOX

Acknowledgments Jan. 1989, pp. 235–284.


https://doi.org/10.1146/annurev.fl.21.010189.001315
The authors gratefully acknowledge Ivett Leyva of the U.S. Air [17] Balakumar, P., and , and Owens, L., “Stability of Hypersonic Boundary
Force Research Laboratory/U.S. Air Force Office of Scientific Layers on a Cone at an Angle of Attack,” AIAA Paper 2010-4718, June
Research for support of this research through grant FA9550-18-1- 2010.
0010. At the National Aerothermochemistry and Hypersonics Labo- https://doi.org/10.2514/6.2010-4718
ratory, the authors acknowledge Farhan Siddiqui, Casey Broslawski, [18] Dinzl, D. J., and , and Candler, G. V., “Direct Numerical Simulation of
and Tyler Dean for their help with experiments and sharing infra- Crossflow Instability Excited by Microscale Roughness on HIFiRE-5,”
structure. The authors thank Rudolph (Rudy) King at NASA Langley AIAA Paper 2016-0353, Jan. 2016.
Research Center for his expertise in hot-wire anemometry. The https://doi.org/10.2514/6.2016-0353
[19] Wheaton, B. M., , Berridge, D. C., , Wolf, T. D., , Stevens, R. T., and , and
authors would like to thank Graham Candler, John Thome, and McGrath, B. E., “Boundary Layer Transition (BOLT) Flight Experiment
Anthony Knutson at the University of Minnesota for their simulation Overview,” AIAA Paper 2018-2892, June 2018.
results and patience. https://doi.org/10.2514/6.2018-2892
[20] Thome, J., , Dwivedi, A., , Nichols, J. W., and , and Candler, G. V.,
“Direct Numerical Simulation of BOLT Hypersonic Flight Vehicle,”
References AIAA Paper 2018-2894, June 2018.
[1] Morkovin, M. V., “Instability, Transition to Turbulence and Predict- https://doi.org/10.2514/6.2018-2894
ability,” AGARD Fluid Dynamics Panel Symposium on ‘Laminar- [21] Moyes, A., , Kocian, T. S., , Mullen, C. D., and , and Reed, H. L., “Pre-
Turbulent Transition’,” Vol. 236, AGARDograph, 1978. Flight Boundary-Layer Stability Analysis of BOLT Geometry,” AIAA
[2] Mack, L. M., “Boundary-Layer Linear Stability Theory,” AGARD Paper 2018-2895, June 2018.
Special Course on Stability and Transition of Laminar Flow, Vol. 709, https://doi.org/10.2514/6.2018-2895
Advisory Group for Aerospace Research and Development, AGARD [22] Moyes, A., and , and Reed, H. L., “Nonlinear Boundary-Layer
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

Rept. 709, 1984, pp. 31–81. Stability Analysis of BOLT and HIFiRE-5,” AIAA Paper 2019-2972,
[3] Fedorov, A., “Transition and Stability of High-Speed Boundary Layers,” June 2019.
Annual Review of Fluid Mechanics, Vol. 43, No. 1, 2011, pp. 79–95. https://doi.org/10.2514/6.2019-2972
https://doi.org/10.1146/annurev-fluid-122109-160750 [23] Mullen, C. D., , Moyes, A., , Kocian, T. S., and , and Reed, H. L., “Heat
[4] Zhong, X., and Wang, X., “Direct Numerical Simulation on the Recep- Transfer and Boundary-Layer Stability Analysis of Subscale BOLT and
tivity, Instability, and Transition of Hypersonic Boundary Layers,” the Fin Cone,” AIAA Paper 2019-3081, June 2019.
Annual Review of Fluid Mechanics, Vol. 44, No. 1, 2012, pp. 527–561. https://doi.org/10.2514/6.2019-3081
https://doi.org/10.1146/annurev-fluid-120710-101208 [24] Li, F., , Choudhari, M., , Chang, C.-L., , White, J., , Kimmel, R., ,
[5] Juliano, T. J., Borg, M. P., and Schneider, S. P., “Quiet Tunnel Mea- Adamczak, D., , Borg, M., , Stanfield, S., and , and Smith, M., “Stability
surements of HIFiRE-5 Boundary-Layer Transition,” AIAA Journal, Analysis for HIFiRE Experiments,” AIAA Paper 2012-2961, June
Vol. 53, No. 4, 2015, pp. 832–846. 2012.
https://doi.org/10.2514/1.J053189 https://doi.org/10.2514/6.2012-2961
[6] Borg, M. P., and , and Kimmel, R. L., “Simultaneous Infrared and [25] Thome, J., , Knutson, A., and , and Candler, G. V., “Boundary Layer
Pressure Measurements of Crossflow Instability Modes for HIFiRE- Instabilities on BoLT Subscale Geometry,” AIAA Paper 2019-0092,
5,” AIAA Paper 2016-0354, Jan. 2016. Jan. 2019.
https://doi.org/10.2514/6.2016-0354 https://doi.org/10.2514/6.2019-0092
[7] Borg, M. P., and , and Kimmel, R. L., “Measurements of Crossflow [26] Kostak, H. E., , Bowersox, R. D., , McKiernan, G. R., , Thome, J.,
Instability Modes for HIFiRE-5 at Angle of Attack,” AIAA Paper 2017- Candler, G. V., and , and King, R. A., “Freestream Disturbance Effects
1681, Jan. 2017. on Boundary Layer Instability and Transition on the AFOSR BOLT
https://doi.org/10.2514/6.2017-1681 Geometry,” AIAA Paper 2019-0088, Jan. 2019.
[8] Borg, M. P., , Kimmel, R. L., , Hofferth, J. W., , Bowersox, R. D., and , https://doi.org/10.2514/6.2019-0088
and Mai, C. L., “Freestream Effects on Boundary Layer Disturbances for [27] Knutson, A. L., Thome, J. S., and Candler, G. V., “Numerical Simu-
HIFiRE-5,” AIAA Paper 2015-0278, Jan. 2015. lation of Instabilities in the Boundary-Layer Transition Experiment
https://doi.org/10.2514/6.2015-0278 Flowfield,” Journal of Spacecraft and Rockets, 2019 (published
[9] Neel, I. T., , Leidy, A., and , and Bowersox, R. D., “Preliminary Study of online).
the Effect of Environmental Disturbances on Hypersonic Crossflow Insta- https://doi.org/10.2514/1.A34599
bility on the HIFiRE-5 Elliptic Cone,” AIAA Paper 2017-0767, Jan. 2017. [28] Berridge, D. C., , McKiernan, G., , Wadhams, T. P., , Holden, M., ,
https://doi.org/10.2514/6.2017-0767 Wheaton, B. M., , Wolf, T. D., and , and Schneider, S. P., “Hypersonic
[10] Neel, I. T., , Leidy, A., , Tichenor, N. R., and , and Bowersox, R. D., Ground Tests in Support of the Boundary Layer Transition (BOLT)
“Influence of Environmental Disturbances on Hypersonic Crossflow Insta- Flight Experiment,” AIAA Paper 2018-2893, June 2018.
bility on the HIFiRE-5 Elliptic Cone,” AIAA Paper 2018-1821, Jan. 2018. https://doi.org/10.2514/6.2018-2893
https://doi.org/10.2514/6.2018-1821 [29] Hofferth, J., , Bowersox, R., and , and Saric, W., “The Mach 6 Quiet
[11] Neel, I. T., , Leidy, A., , Tichenor, N. R., and , and Bowersox, R., Tunnel at Texas A&M: Quiet Flow Performance,” AIAA Paper 2010-
“Characterization of Environmental Disturbances on Hypersonic Cross- 4794, June 2010.
flow Instability on the HIFiRE-5 Elliptic Cone,” AIAA Paper 2018- https://doi.org/10.2514/6.2010-4794
5375, Sept. 2018. [30] Semper, M., , Pruski, B., and , and Bowersox, R., “Freestream Turbu-
https://doi.org/10.2514/6.2018-5375 lence Measurements in a Continuously Variable Hypersonic Wind
[12] Holden, M., , Wadhams, T., , MacLean, M., and , and Mundy, E., Tunnel,” AIAA Paper 2012-732, Jan. 2012.
“Reviews of Studies of Boundary Layer Transition in Hypersonic Flows https://doi.org/10.2514/6.2012-732
over Axisymmetric and Elliptic Cones Conducted in the CUBRC Shock [31] Hofferth, J., and , and Saric, W., “Boundary-Layer Transition on a Flared
Tunnels,” AIAA Paper 2009-782, Jan. 2009. Cone in the Texas A&M Mach 6 Quiet Tunnel,” AIAA Paper 2012-923,
https://doi.org/10.2514/6.2009-782 Jan. 2012.
[13] Berger, K., , Rufer, S., , Kimmel, R., and , and Adamczak, D., “Aero- https://doi.org/10.2514/6.2012-923
thermodynamic Characteristics of Boundary Layer Transition and Trip [32] Hofferth, J. W., “Boundary-Layer Stability and Transition on a Flared
Effectiveness of the HIFiRE Flight 5 Vehicle,” AIAA Paper 2009-4055, Cone in a Mach 6 Quiet Wind Tunnel,” Ph.D. Dissertation, Texas A&M
June 2009. Univ., College Station, TX, 2013.
https://doi.org/10.2514/6.2009-4055 [33] Kostak, H. E., and , and Bowersox, R. D., “Hypersonic Boundary Layer
[14] Juliano, T., and , and Schneider, S., “Instability and Transition on the Off-Body and Surface Measurements on the AFOSR BOLT Geometry,”
HIFiRE-5 in a Mach 6 Quiet Tunnel,” AIAA Paper 2010-5004, June 2010. AIAA Paper 2020-1043, Jan. 2020.
https://doi.org/10.2514/6.2010-5004 https://doi.org/10.2514/6.2020-1043
[15] Kocian, T. S., , Moyes, A., , Mullen, D., and , and Reed, H. L., “PSE and [34] Leidy, A. N., “An Experimental Characterization of 3-D Transitional
Spatial Biglobal Instability Analysis of Reduced Scale and Flight Shock Wave Boundary Layer Interactions at Mach 6,” Ph.D. Disserta-
HIFiRE-5 Geometry,” AIAA Paper 2017-0768, Jan. 2017. tion, Texas A&M Univ., College Station, TX, 2019.
https://doi.org/10.2514/6.2017-0768 [35] Neel, I. T., “Influence of Environmental Disturbances on Hypersonic
[16] Reed, H. L., and Saric, W. S., “Stability of Three-Dimensional Crossflow Instability on the HIFiRE-5 Elliptic Cone,” Ph.D. Disserta-
Boundary Layers,” Annual Review of Fluid Mechanics, Vol. 21, tion, Texas A&M Univ., College Station, TX, 2019.
Article in Advance / KOSTAK AND BOWERSOX 11

[36] Boyd, C. F.,, and Howell, A., “Numerical Investigation of One-Dimen- High-Frequency Instrumentation in Support of the Boundary Layer
sional Heat-Flux Calculations,” U.S. Naval Surface Warfare Center Transition (BOLT) Flight Experiment,” AIAA Paper 2019-0090,
Dahlgren Division Rept. NSWCDD/TR-94/114, Silver Spring, MD, Jan. 2019.
Oct. 1994. https://doi.org/10.2514/6.2019-0090
[37] Juliano, T. J., , Paquin, L., and , and Borg, M. P., “Measurement of [44] Chynoweth, B. C., , Schneider, S. P., and , and Wheaton, B. M.,
HIFiRE-5 Boundary-Layer Transition in a Mach-6 Quiet Tunnel with “Transition Measurements with Forward and Aft Facing Steps on the
Infrared Thermography,” AIAA Paper 2016-0595, Jan. 2016. BOLT Geometry at Mach 6,” AIAA Paper 2020-1560, Jan. 2020.
https://doi.org/10.2514/6.2016-0595 https://doi.org/10.2514/6.2020-1560
[38] Beresh, S. J., Henfling, J. F., Spillers, R. W., and Pruett, B. O. M., [45] Knutson, A., , Gs, S., and , and Candler, G. V., “Instabilities in Mach 6
“Fluctuating Wall Pressures Measured Beneath a Supersonic Turbulent Flow over a Cone with a Swept Fin,” AIAA Paper 2018-3071, June
Boundary Layer,” Physics of Fluids, Vol. 23, No. 7, 2011, Paper 075110. 2018.
https://doi.org/10.1063/1.3609271 https://doi.org/10.2514/6.2018-3071
[39] Kostak, H. E., “Ground Test Analyses of the AFOSR Boundary Layer [46] Smits, A., Hayakawa, K., and Muck, K., “Constant-Temperature Hot-
Transition (BOLT) Flight Geometry,” M.S. Thesis, Texas A&M Univ., Wire Anemometer Practice in Supersonic Flows, Part 1—The Normal
College Station, TX, 2020. Wire,” Experiments in Fluids, Vol. 1, No. 2, 1983, pp. 83–92.
[40] Craig, S. A., “Stability of High-Speed, Three-Dimensional Boundary https://doi.org/10.2514/6.1983-50
Layers,” Ph.D. Dissertation, Texas A&M Univ., College Station, TX, [47] Semper, M. T., and Bowersox, R. D. W., “Tripping of a Hypersonic
2015. Low-Reynolds-Number Boundary Layer,” AIAA Journal, Vol. 55,
[41] Craig, S. A., and , and Saric, W. S., “Experimental Study of Crossflow No. 3, 2017, pp. 808–817.
Instability on a Mach 6 Yawed Cone,” AIAA Paper 2015-2774, June https://doi.org/10.2514/1.J055341
2015. [48] Bowersox, R. D. W., “Combined Laser Doppler Velocimetry and Cross-
https://doi.org/10.2514/6.2015-2774 Wire Anemometry Analysis for Supersonic Turbulent Flow,” AIAA
[42] Berry, S. A., , Mason, M. L., , Greene, F., , King, R., , Rieken, E., and , Journal, Vol. 34, No. 11, 1996, pp. 2269–2275.
and Basore, K., “LaRC Aerothermodynamic Ground Tests in Support of
Downloaded by 212.119.46.126 on October 5, 2020 | http://arc.aiaa.org | DOI: 10.2514/1.A34858

https://doi.org/10.2514/3.13390
BOLT Flight Experiment,” AIAA Paper 2019-0091, Jan. 2019.
https://doi.org/10.2514/6.2019-0091 R. M. Cummings
[43] Berridge, D. C., , Kostak, H. E., , McKiernan, G. R., , Wheaton, B. M., , Associate Editor
Wolf, T. D., and , and Schneider, S. P., “Hypersonic Ground Tests with

You might also like