Proceedings of the Internoise 2012/ASME NCAD meeting
August 19-22, 2012, New York City, NY, USA
ASME/NCAD-1270
A LABORATORY FRAMEWORK FOR SYNCHRONOUS NEAR/FAR-FIELD
ACOUSTICS AND MHZ PIV IN HIGH-TEMPERATURE, SHOCK-CONTAINING, JETS
N. Murray∗ & G. Lyons
C. E. Tinney, B. Donald & W. Baars
Jamie Whitten National Center
for Physical Acoustics
The University of Mississippi
University, Mississippi 38677
Email: nmurray@olemiss.edu
Dept. of Aerospace Engineering
and Engineering Mechanics
The University of Texas at Austin
Austin, Texas 78712
B. Thurow & H. Haynes
P. Panickar
Advanced Laser Diagnostics Lab
Dept. of Aerospace Engineering
Auburn University
Auburn, Alabama 36849
Combustion Research and Flow Technology, Inc.
Pipersville, PA 18947
NOMENCLATURE
Ae Nozzle exit area.
At Nozzle throat area.
AR Exit-to-throat area ratio, AR = Ae /At .
AJL Anechoic Jet Laboratory.
D j Nozzle exit diameter.
Ma Acoustic Mach number.
Md Nozzle design Mach number based on AR.
M j Jet Mach number computed from the NPR.
U j Jet exit velocity.
NPR Nozzle pressure ratio, NPR = P0 /Pa .
P0 Jet stagnation pressure.
Pa Ambient pressure measured inside the AJL.
x Axial coordinate, x = 0 at the nozzle exit, positive in the flow
direction.
r Radial coordinate such that r/D j = 0.5 at the nozzle lip.
θ Polar angle measured from the positive x-axis.
PBL Pulse Burst Laser System
ABSTRACT
This paper describes the experimental study of the noise
generating characteristics of high-temperature, shock-containing
jets emanating from conic-section, converging-diverging (C-D)
nozzles. Conic C-D nozzles consist of two conic sections, one
contracting and the other expanding, joined to form a supersonic
nozzle with a very sharp radius of curvature at the nozzle throat.
An experiment is conducted in which temporally resolved flowfield measurements are acquired simultaneously with near-field
and far-field acoustics to allow investigation of the turbulence associated with noise generation. The MHz rate PIV system and its
synchronization with acoustic measurements is described along
with methods for data analysis. General acoustic results are presented to characterize the spectral content present, and preliminary results on the measured turbulence structures are discussed.
∗ Address
all correspondence to this author.
1
Copyright c 2012 by ASME
FIGURE 1. The mean velocity field (CFD computation) for the Centerbody configuration at fully-expanded conditions illustrating the shock
structure in the plume.
INTRODUCTION
Jet noise generated by supersonic flows from convergingdiverging nozzles that generate internal shocks possess an added
level of complexity in that they include an additional shock system in the jet plume and can involve flow separation inside the
diverging section of the nozzle. Typical variable area nozzles
found on modern high-performance military aircraft are representative of this situation. A renewed interest in studying the
related jet noise has been prompted by the US Navy to specifically address the noise-induced hearing loss and degraded operational awareness resulting from the ever increasing noise levels
of higher thrust engines [1, 2].
The present work focuses specifically on the noise generating characteristics of high-temperature, shock-containing jets
emanating from conic-section, converging-diverging (C-D) nozzles. A conic C-D nozzle here describes the joining of two conic
sections, one contracting and the other expanding, to form a supersonic nozzle with a very sharp radius of curvature at the nozzle throat. The near discontinuity in the slope of the nozzle contour at the throat generates a shock even when the nozzle is operated at its design Mach number, Md . This is illustrated in Fig. 1
which shows the mean streamwise velocity field computed using
a Hybrid RANS/LES approach.
To study the noise generating features of these jets with sufficient temporal fidelity, a synchronized system is here developed
that includes a MHz rate PIV system, and near-field pressure
transducer array, and a far-field microphone array. The MHz
rate flow measurement system offers the opportunity to obtain
time-resolved, high-dynamic range velocity measurements. Synchronizing the PIV with the near-field and far-field arrays provides the means for evaluating the noise generation and propagation.This laboratory framework shares similarities to previous
recent works utilizing either high-frame-rate PIV [3, 4] and/or
near-field/far-field correlation measurements [5–7].
While no results from the fully synchronous system have
yet been obtained, the various components and the synchronized
data acquisition system are here described in detail. A discussion of the high-dynamic range PIV analysis methodology is presented along with preliminary flow-visualization results obtained
with the pulse burst laser system. To characterize the jet itself,
the bulk of the paper focuses on the results of mean flow probe
FIGURE 2. The NCPA Anechoic Jet Laboratory setup for the simultaneous near-field/far-field/MHz-PIV measurements.
measurements and the near-field and far-field acoustic data.
DETAILS OF THE EXPERIMENTAL SETUP
To date, three main experimental entries have been completed. The first measured the mean total pressure distribution
for each jet configuration. The second measured the fluctuating
pressure field using a near-field line array of pressure transducers simultaneously with a far-field arc array of microphones for
each jet configuration. The third entry involved the synchronous
measurement of the near-field and far-field pressure along with
particle image velocimetry (PIV) measurements of the velocity
field use a MHz rate system. A description of the experimental facility and setup for each of these components is discussed
below.
Anechoic Jet Laboratory
The Anechoic Jet Laboratory (AJL) at the University of Mississippi’s National Center for Physical Acoustics is a small facility purpose built for the study of high-temperature, supersonic
jet noise [8, 9]. The facility was designed with upstream and
downstream “stagnation” chambers through which ambient air is
pulled by a 10,000 SCFM fan. The air is allowed to percolate
into the 19-by-20-by-8 foot chamber (wedge tip to wedge tip)
through 50% porosity sliding panels achieving approximately 1
ft/s in the anechoic section (without jet flow). The openings in
the upstream wedge wall allowing the aspiration of the chamber
can be seen in Fig. 2. The aspiration of the chamber results in a
very even temperature distribution throughout the room and allows the jet entrainment to be less effected by the enclosed space.
The jet rig utilizes a propane burner system as shown in
2
Copyright c 2012 by ASME
FIGURE 3. Illustration of the propane burner system and nozzle assembly.
(a) Nozzle Parts
Fig. 3. Air is supplied from an 1100 hp Ingersoll-Rand Centac
compressor through a desiccant dryer system yielding at maximum 5000 SCFM of dry air at 125 psi a enabling continuous operation. The propane combustor is housed upstream of the nozzle
section and is followed by a ceramic flow conditioner and settling
chamber upstream of the main contraction. Figure 3 shows the
nozzle assembly which includes a centerbody housing and the
nozzle section. When the centerbody was not used, a straight extension tube was put in its place so that the nozzle exit remained
in the same location for all test configurations.
Figure 2 shows the AJL with all the data acquisition equipment in place during the experiment discussed herein. To date,
PIV data was only acquired for one of the jet configurations under consideration. All of the acoustic data presented below were
acquired in the second test entry for which the PIV camera and
light sheet optics were removed from the room to have as anechoic an environment as possible.
(b) Faceted Nozzle Assembly
FIGURE 4. The modular nozzle assembly allows for both axisymmetric and segmented inserts to be used for the inner-nozzle contour.
Jet Nozzle Configurations
The jet nozzle assembly is modular offering the ability to
test various inner-nozzle contours as shown in Fig. 4(a). The
inner-contour inserts are locked in place with a threaded locking
ring. This modularity also allows for a faceted inner contour to
be generated by placing a number of identical segments inside
the outer nozzle shell as shown in Fig. 4(b).
Three jet nozzle hardware configurations were considered
during this work. They are referred to in the following discussion
as the Baseline, Centerbody, and Faceted configurations according to the following descriptions.
Centerbody The centerbody configuration used the conic nozzle but included a streamlined centerbody upstream of the
nozzle contraction. The location of the centerbody is illustrated in both Fig. 1 and 3.
Faceted The faceted configuration utilized 12 identical segments to form the C-D nozzle contour as shown in Fig. 4(b).
The exit-to-throat area ratio for the faceted nozzle is the
same as that of the baseline conical nozzle. In the acoustic
data for the Faceted configuration the centerbody was not
used.
Baseline The baseline configuration used the axisymmetric,
conic, converging-diverging (C-D) nozzle. The term ‘conic’
is descriptive of the conic sections that make up the converging and diverging parts of the nozzle. These conic sections
are brought together at a very sharp throat radius causing a
near discontinuity in the wall slope at the throat. The baseline configuration did not include the centerbody section.
The nozzle exit diameter D j = 2 in., and the AR = Ae /At =
1.4.
Jet Operating Conditions
Each of the three hardware configurations were operated at
both over-expanded and fully-expanded M j . The associated operating conditions are tabulated in Tab. 1. In all cases, the jet
temperature was maintained at approximately 1005 K yielding a
temperature ratio of 3.3. Table 1 also includes the β 2 = M 2j − 1
parameter which is known to be a significant parameter for
broadband shock associated noise [10–12].
3
Copyright c 2012 by ASME
NPR
Mj
T j /Ta
β
Over-Expanded
3.92
1.55
3.3
1.18
Fully-Expanded
5.21
1.74
3.3
1.42
tion according to
Pt∗ (x, r) =
TABLE 1. Mean operating conditions for the over-expanded and
fully-expanded jets.
Pt (x, r)
· P0
P0
(1)
where P0 is the ensemble mean of the jet stagnation pressure
while the probe remained in one measurement station and P0 is
the mean of all the ensemble means. This effectively removes
any variation in the probe measurements caused by fluctuations
in the jet stagnation pressure. Finally, the Rayleigh supersonic
pitot formula was applied for measurements obtained within supersonic flow.
During long test runs, the thermal expansion of the nozzle
assembly can be as much as 15% of the nozzle exit diameter. To
account for this, the position of the nozzle exit was continuously
tracked using an expanded He-Ne laser beam and a positionsensitive detector (PSD). The system was calibrated by traversing the nozzle itself through the beam, so that the PSD output
voltage was recorded for a sequence of known positions. During each profile measurement, the PSD voltage was recorded at
regular intervals.This data was used to correct the probe’s axial
position from the nozzle to within 1%.
FIGURE 5. The pitot probe setup for the mean flow measurements included a 3-axis computer-controlled traversing system and a laser-based
position sensitive detector (PSD) to correct axial position for the thermal
expansion of the nozzle.
Acoustic Arrays
Measurements of the near- and far-field pressures were acquired synchronously to develop an intuition for how changes
to the jet flow structure would affect the far-field sound. Signatures registered in the hydrodynamic periphery of the jet produce
a pressure footprint corresponding to the passage of large scale
turbulent structures [13–16].
Mean Flow Profile Traversing System
A traversing total pressure (pitot) probe was used to characterize the mean jet flow for each configuration (see Fig. 5).
The probe was positioned using a three-axis traverse system with
greater than 0.1 mm accuracy. A centerline profile was measured after first determining the actual jet axis through a set
of cross-stream profiles.Detailed cross-stream profiles were then
measured at four distinct axial positions in order to characterize
the mean flow evolution for each configuration.
Near-Field Line Array A line array was constructed for
near-field measurements and comprised thirty-one Kulite XT140-100A transducers as shown in Fig. 6. These are absolute
pressure transducers which have a range of 0-100 psia, a diameter of nominally 2.6 mm, and include a protective grid cap. Signal conditioning and amplification were provided by a National
Instruments PXI system using an NI-PXIe-1073 chassis with
an NI-TB-4330 board and four NI-PXIe-4331 cards (8 channels/card) with built-in anti-aliasing filters. All channels were
sampled synchronously at 100kHz and phase aligned with the
far-field measurements through an external trigger input.
The position of the near-field array relative to the jet axis
is shown in Fig. 7. The most upstream microphone was located
at x/D j = 1.6 and y/D j = 1.87, and the line array was set to
an 8◦ angle relative to the jet axis. The spatial separation between adjacent points was chosen based on the integral scales of
the flow. This was estimated from previous jet data [17] to be
around 1/3 the shear layer thickness. The shear layer thickness is
measured by the distance between the inner (4◦ ) and outer (8◦ )
Anticipating the difficulty in resolving the large gradients
across shocks and the shear layers, an adaptive algorithm was
developed to vary the step size taken by the traverse along the
radial direction.By using this algorithm, without any foreknowledge of the mean flow, more points are taken in regions with
larger total pressure gradients. This algorithm was implemented
on the cross-stream profiles only.
For all total pressure measurements, the probe was traversed
to the desired (x, r) position, allowed to settle, and then pressure
measurements were acquired until the statistical 99% confidence
interval fell below a specified 1% relative error. Once this confidence interval was achieved, the ensemble average was taken
as the total pressure at that position. Each total pressure, Pt (x, r)
was subsequently normalized by the nozzle total pressure varia4
Copyright c 2012 by ASME
◦
θ =105◦ θ =90
◦
θ =120
θ =75◦
θ =60◦
θ =52.5◦
θ =45◦
θ =127.5◦
θ =135◦
58.5 Dj
Nozzle
FIGURE 6. Photo of the upstream end of the near-field line array
showing the small Kulite transducers flush-mounted in the aluminum
housing.
θ =20◦
jet axis
FIGURE 8. The position of the far-field arc array microphones relative to the nozzle exit.
passage of turbulent large scale structures in the jet shear layer.
5
y/Dj
θ
θ =37.5◦
θ =30◦
1
0
−1
0.95
0.5
0.25
8 spread
−4◦ spread
T = 295◦
kulite array
◦
−5
0
5
Far-Field Arc Array The far-field arc array consists of
12 B&K Type 4939 1/4-inch, free-field microphones attached
to Type 2670 pre-amplifiers. The grid caps were removed during the experiments. Microphones were placed between 20◦ and
135◦ with closer spacing near the peak noise directions as shown
in Fig. 8. Signal acquisition was performed using a National
Instruments PXIe-4497 dynamic signal analyzer (16 simultaneously sampled analog input channels, 24 bit resolution) housed in
a PXIe-1082 system. All channels were simultaneously sampled
at 100 kHz and phase aligned with the near-field measurement
through an external trigger input.
0.05
0.025
10
15
20
x/Dj
FIGURE 7. Position of the near-field pressure array relative to a Mach
1.5 heated jet simulated by way of LES.
MHz Rate PIV System
The megahertz rate PIV system used in this work involved
the combined use of a pulse burst laser (PBL) and a high-speed,
gated intensified CCD framing camera. For its ability to acquire
sequences of sixteen images at MHz rates, the system allowed
time-resolved and high dynamic range measurements to be obtained for a heated, supersonic jet. Each component of the system
is explained in detail below along with the experimental setup.
spreading angles of the jet which were determined from preliminary LES computations performed by CRAFT Tech. In doing so,
the length of the array could be designed to capture signatures
registered between the near-nozzle region and the post-potential
core region without compromising spatial resolution.
Tests conducted at UT-Austin compared the output of a
Kulite to a standard 1/4-inch pressure field microphone (PCB)
by placing both transducers within the hydrodynamic periphery
of a perfectly expanded Mach 3.0 jet. Spectra computed from the
Kulite revealed a steeper roll-off at the higher frequencies when
compared to the 1/4-inch PCB microphone. This is caused by the
small vent holes on the Kulite’s protective grid cap which dampens out higher frequencies. Nevertheless, it was determined that
this high-frequency roll-off was above the frequencies of interest
in this case, and the Kulite transducers and 1/4-inch PCB microphones matched well at lower frequencies allowing successful
measurement of the hydrodynamic hump corresponding to the
Pulse Burst Laser System As has been described
in previous publications, [18–20] a PBL system developed at
Auburn University allows a specified number of high-energy,
MHz rate laser pulses to be formed from a given burst of lowenergy, short-duration pulses. It should be noted that several upgrades have been made to this system since these publications: a
new JDSU NPRO 126 continuous-wave (CW) Nd:YAG laser to
enhance the pulse-to-pulse stability of each burst and three supplementary amplification stages (for a total of six amplification
5
Copyright c 2012 by ASME
Pulse Generation
Pulse Energy Amplification
λ/2
Frequency Conversion
in excess of 50 mJ/pulse1 .
The final stage of the PBL is the conversion of the beam’s
wavelength from 1064 nm to 532 nm. This conversion is
achieved via a nonlinear process inside a KTP crystal and results in an unavoidable loss of pulse energy. Nevertheless, the
beam, now in the visible spectrum, can be used for fluid dynamic
measurements including PIV and flow visualization applications.
λ/4
CW Nd:YAG
1064nm
AMP 1
AMP 2
AMP 3
AMP 4
AMP 5
AMP 6
KTP
Pulsed
1064nm
Pulsed
532nm
AOM
λ/2
λ/4
Wave plate
λ/4
Biconvex lens
Mirror
Beam dump
Optical isolator
Plano-concave lens
Cordin 222-4G High-Speed Camera Images were acquired using a Cordin 222-4G gated intensified CCD framing
camera that is capable of recording sixteen images at a maximum, equally-spaced rate of 2,187,500 frames-per-second2 . Image resolution is 2048 × 2048 pixels although the true resolution
is slightly less due to the intensification process. The camera is
able to achieve extremely high acquisition rates because it contains eight independently-controlled optical pathways, each incorporating a micro-channel plate (MCP) for signal intensification and ultimately terminating with a Kodak KAI-4022 CCD
sensor. By allowing each CCD to record two images, sixteen
total images can be acquired over a user-specified time period.
Furthermore, because each pathway is independently operated,
temporal spacing between frames is variable and can be set in an
asynchronous fashion. Such flexibility even allows eight simultaneous exposures to be made. This feature is especially desirable since it enables eight nearly identical velocity fields to be
obtained, with any differences being directly attributable to systematic error. For the experiments of interest, it is sufficient to
note that because the camera can acquire sixteen images over a
user-specified, extremely short time period, temporal-resolution
is possible for all captured fluid motions. Additionally, the ability
to obtain several particle images at varying time intervals relative
to one another provides the means of performing high dynamic
range PIV. Such measurements offer significant improvements
over conventional PIV results since optimal time separations can
be selected for different particle locations depending on the local
velocity.
Optical crystal
Polarizer
FIGURE 9. A top view of schematic of the pulse burst laser (PBL)
system utilized for MHz rate PIV.
stages) to increase the overall energy available for each burst.A
schematic of the upgraded PBL system is shown in Fig. 9.
The design of the PBL can be divided into three fundamental parts as indicated in the schematic: the pulse generation, the
pulse energy amplification, and the frequency conversion. The
pulse generation stage slices the output of the CW laser into a
burst of low-energy, short-duration pulses through the use of an
acousto-optic modulator (AOM) that relies on the principles of
the acousto-optic (AO) effect. In particular, a piezoelectric transducer is used to produce acoustic waves inside an optical crystal such that the traveling waves cause variations in the index of
refraction of the crystal. To an optical beam, these variations
appear as a sinusoidal grating in which the wavelength is equal
to the acoustic wavelength. By controlling when and how frequently acoustic waves are produced in the crystal, the generation of a specified number of short-duration pulses is possible
depending on how often the CW input beam is disturbed. As with
most AO devices, the AOM operates in the Bragg regime where
most of the incident light can be diffracted into the first-order
beam fairly efficiently. Here this diffracted beam constitutes the
desired burst of pulses.
Following the formation of low-energy (nano-Joule order),
short-duration pulses, the remaining stages of the pulse burst
laser consist of pulse energy amplification and frequency conversion. Amplification is provided by six flashlamp-pumped
Nd:YAG rod amplifiers of increasing diameter resulting in sufficient energy for fluid dynamic measurements. The first three
amplifiers are used in a double-pass arrangement whereas the
final three allow only for a single pass. Wave plates and polarizers provide the necessary means for achieving double-passes
through the first three amplifiers. Optical isolators between each
of the first five amplification stages prevent problems associated
with parasitic lasing and amplified spontaneous emission (ASE).
By the end of the amplification chain, pulse energies have increased by a factor of more than 107 and generally reach levels
Setup of the MHz PIV System The MHz PIV system
was only used to acquire data for the Baseline nozzle configuration at the over-expanded condition (M j = 1.55). The field of
view was centered at x/D j = 7 and r/D j = −0.5. This distance
was chosen to coincide with the region around the end of the potential core. The region imaged was slightly less than 4 square
inches and was illuminated by a laser sheet directed vertically upwards and spanning axially along the centerline of the jet. This
1 This
value is measured after the frequency conversion stage and thus accounts for the loss in energy associated with doubling the frequency of the
Nd:YAG laser beam via a KTP crystal.
2 This rate assumes a necessary CCD transfer time of 3.2 µ s (specified by
Cordin) to ensure that the second exposure does not include ghost images from
the first exposure.
6
Copyright c 2012 by ASME
Auburn University
MHz PIV System
NCPA PXIe system
(farfield array measurement)
UTA PXIe system
(nearfield array measurement)
Trigger Signal
Initialize
DAQ
FIGURE 10. Experimental arrangement for the time-resolved, high
dynamic range PIV application (side view). The square region enclosed
by the dashed line indicates the camera’s field of view.
Control Computer
(logs mean flow conditions)
DAQ
Status
particular orientation was chosen for a variety of reasons, including both the need to minimize disruptions in the anechoic environment as well as to ensure the most direct observation of any
shear layer without passing the light sheet through the jet prior
to imaging. This last point was especially important to prevent
problems associated with aero-optical distortions.
A schematic of the experimental setup is shown in Fig. 10.
The 532 nm wavelength beam from the pulse burst laser was
passed into the anechoic room perpendicularly to the jet axis and
opposite the location of the camera. A turning mirror attached to
the stand for the burner system allowed the beam to be directed
downstream of the jet nozzle exit before encountering a 1000 mm
biconvex spherical lens and a second turning mirror. The beam
was then redirected vertically upwards through a cylindrical lens
to form the laser sheet required for light scattering. Extreme care
was taken to ensure that this light sheet was oriented both orthogonally to the axis of the camera lens as well as to the nozzle
exit plane. Additionally, the placement of the spherical lens allowed the thinnest portion of the light sheet to persist across the
camera’s field of view.
Particle seeding of the jet was achieved using aluminum
oxide (Al2 O3 ) particles nominally 0.1 µm in size. A nitrogenpressurized reservoir filled with these particles was connected to
the burner system’s particle seeding tubes shown in Fig. 3. Four
seeding tubes were attached around the burner system symmetrically to provide a uniform seeding density throughout the jet.
PIV measurements were obtained by synchronizing the
framing rate of the camera with the pulse generating rate of
the pulse burst laser system. The chosen rate for all cases was
1 MHz, meaning the 16 images acquired by the camera enclosed
a temporal window spanning 15 µs. To achieve the most consistent pulse-to-pulse intensity within each burst, 60 laser pulses
were generated for a given burst, and the most stable 16 pulses
were selected for synchronization with the 16 camera frames.
Initialize
DAQ
DAQ
Status
FIGURE 11. Schematic of the synchronized data acquisition between
the near-field and far-field microphone arrays and the MHz PIV system.
The duration of each laser pulse was approximately 20 ns such
that no image streaking was observed. A Nikon Nikkor F-mount
70-300 mm objective zoom lens (f/4-5.6G) was used with the
Cordin camera to acquire all image sequences.
Synchronizing the Data Acquisition Systems
The entire data acquisition system was purposely installed in
such a way that time synchronization could be achieved across
all three measurements: near-field pressure, far-field pressure,
and PIV velocity. The goal of this arrangement is to allow for
investigation of the flow-field events related to the generation of
the measured jet noise.
The long transfer time needed to move each 16-image sequence from the camera to the PIV acquisition computer resulted
in approximately 10 seconds between each 16-image acquisition.
This being the slowest of the systems in terms of its data acquisition cycle (initialization, acquisition, saving) it was necessary
utilize it as the trigger for the pressure acquisition systems.
Figure 11 shows a schematic view of the synchronized data
acquisition process; this process was implemented using LabVIEW virtual instrument programs utilizing Data Socket communication and hardware TTL triggers for initiation.
The procedure for obtaining the synchronized data involved
(a) bringing the jet up to predetermined set-point conditions of
pressure and temperature and holding these conditions continuously, (b) the control computer initializing the near-field and
far-field PXIe systems setting data acquisition parameters and
generating a consistent filename on each system, (c) setting the
7
Copyright c 2012 by ASME
pressure acquisition systems in a reference trigger mode wherein
they wait for a TTL trigger to start acquiring data, (d) sending
the TTL trigger from the Cordin camera to the PXI units to initiate an acquisition cycle. The system operates in a continuous
mode such that on each trigger signal from the camera both PXIe
systems acquire and save dynamic data on their respective computers. The status of each PXIe system during data acquisition
is communicated to the control computer which is also recording
and averaging run conditions during the test event.
8
1
9
16
t0
1
PRELIMINARY RESULTS FROM MHZ PIV
High Dynamic Range PIV
The ability to obtain several particle images at varying time
intervals with respect to one another provides the means of performing high dynamic range PIV (HDR-PIV). For such measurements the dynamic velocity range is determined by the maximum and minimum resolvable particle displacements. Unlike
conventional PIV where only one temporal spacing is available
for all velocity determinations, the various combinations of image pairs in HDR-PIV enable velocity measurements to be made
by utilizing optimal time separations for different regions of a
flow field depending on the local velocity. Thus in regions of
a flow field where little or no motion is observed between consecutive images, a larger temporal spacing (i.e., a pair of images separated by a greater time displacement) can be used such
that the particle motions approach the optimal value for accurate cross-correlation analyses. This method offers substantial
improvements over conventional PIV measurements and is especially useful in flows containing a wide velocity range. [21]
A schematic illustrating the capabilities of the HDR-PIV
system is shown in Fig. 12. As mentioned, sequences of 16 images were obtained in which the time displacement, δt, between
subsequent frames was 1 µs. For this arrangement the velocity
field temporally located between frames 8 and 9 can be measured
using different combinations of the 8 image pairs symmetrically
straddling this t0 point. It should be noted that other velocity
fields could also be calculated; however this particular point in
time allows the maximum number of image pairs to be used with
the more accurate central finite-difference scheme. For regions
of the velocity field containing the highest local velocities (i.e.,
the largest particle displacements between consecutive images),
the image pair shown in red can be used to obtain the velocity.
Likewise for regions containing little or no particle motions, an
image pair spanning a larger temporal distance can be chosen.
The image pair shown in blue represents the case of maximum
time displacement and spans the entire sequence window of 15 δt
or 15 µs. By using the information available across all 16 frames
instead of only consecutive images, significant improvements in
mean flow measurements as well as turbulence quantities can be
made. This procedure will be utilized in the near future to analyze the collected data.
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
dt
15 dt
FIGURE 12. Illustration showing the basic principle of HDR-PIV. A
single velocity field at t0 can be determined by combining local velocity measurements acquired from various image pairs. Depending on
the local velocity, image pairs are chosen such that optimal particle displacements occur between them.
Pulse Burst Laser Flow Visualization
To illustrate the general characteristics of the flow field, a
60-pulse burst from the PBL was imaged using a PCO.Edge sCMOS camera. The image shown in Fig. 13 was recorded for the
Centerbody configuration at M j = 1.55. The center of the image
is located near x/D j = 7. Distinct vortical structures of relatively
large size can be clearly identified in the shear layer.
Using the PBL for flow visualization causes the high-speed
particles to appear in the image as dotted streaks. In regions of
the flow where the individual dots can be resolved, it is possible
to determine the direction, velocity, and acceleration along the
streak. As an example of a potential image processing method,
Fig. 13(b) shows a single 128-by-128 pixel interrogation area
extracted from near the centerline of the jet (represented by a
square in Fig. 13(a). A simple auto-correlation of the sample
region reveals multiple peaks as shown in Fig. 13(c). The angle of the dotted line in the auto-correlation gives direction. The
spacing between dots in the auto-correlation gives velocity magnitude when taken together with the time between pulses. Also,
if a change in the distance between dots were evident, it may be
possible to measure acceleration. However, due to the high velocity gradients, the particle streaks in the low-speed portion of
the flow appear as solid and not dotted. This further illustrates
the potential benefits of the HDR-PIV analysis.
FLOW FIELD CHARACTERISTICS
Mean Flow Profiles
The centerline profiles for each nozzle configuration are
shown in Fig. 14. The position of each point has been corrected
for nozzle thermal expansion by a third-order polynomial fit to
8
Copyright c 2012 by ASME
80
Baseline, Fully−Expanded
Baseline, Over−Expanded
Centerbody, Fully−Expanded
Centerbody, Over−Expanded
Faceted, Over−Expanded
Sonic Core Length
Potential Core Length
P0 (psia)
70
60
50
40
30
20
0
2
4
6
8
10
12
14
16
18
20
x/Dj
FIGURE 14. Profiles of the total pressure decay along the get centerline. The end of the potential core is denoted by ▽, and the end of the
supersonic core is denoted by .
(a) 60-Pulse Burst Imaged with a PCO.Edge Camera
(b) 128-Pixel Sample
Faceted data is not shown. The x/D j position of each profile is
shown above each profile and has been corrected for nozzle expansion as discussed above. The evolution of the jet profile shape
from a near top-hat to a self-similar hyperbolic sine function is
evident as is the effect of the probe traversing across shocks in
the near-nozzle profiles. The centerbody itself is supported by
three streamlined pylons equally spaced azimuthally around the
annulus inside the enclosing pipe. The wake from one of these
pylons was in the profile plane and is the cause for the observed
asymmetry in the Centerbody configuration.
(c) Sample Auto-Correlation
Far-Field Acoustic Characteristics
The single-sided auto-spectral density (SPL) of the far-field
pressure signals were generated using an ensemble average of
150 sets of 4096 samples yielding a frequency resolution of
∆ f = 24.4 Hz. The spectra are shown in Fig. 16 for both the
over-expanded (M j = 1.55) and fully-expanded (M j = 1.74) jet
conditions. Beginning with the downstream observer at θ = 20◦ ,
subsequent spectra are shifted by 15 dB for clarity.
As expected, far-field jet noise is dominated by turbulent
mixing noise at shallow angles. And, at the over-expanded M j
(Fig. 16(a)) broadband shock noise (BBSN) in the steeper angles is more dominant compared to the fully-expanded M j . Furthermore, below the Mach wave angle of this jet – where sound
pressure levels form to create the classic heart-shape pattern of
jet noise – the nozzle configuration has little influence on the
spectral levels. However, between the Mach wave angle and the
steeper angles, the effects introduced by the centerbody or the
faceted inner contour are evident and shown to reside in the high
frequency band. This suggests that the predominant effect of
inclusion of a centerbody or a faceted nozzle contour will be noticed in the higher frequencies and in the steeper angles.
The observed directivity of the measured overall sound pressure level (OASPL) for each nozzle configuration is shown in
Fig. 17 for both the over-expanded and fully-expanded M j . There
is an interesting difference between the Baseline and Faceted
FIGURE 13. PBL flow visualization of the M j = 1.55 Centerbody
configuration. The full image (a) is centered near x/D j = 7.
the calibrated positions based on the PSD data. The error implicit
in this method is estimated by the root of the residual sum of
squares to be less than 0.03 jet diameters in all cases, except for
that of the faceted nozzle where unexplained PSD voltage fluctuations yield a 0.5 jet diameter error. The shock cell structure is
evident in all the centerline total-pressure profiles up to near the
end of the potential core. For each case, the end of the potential
core is denoted by a ▽, and the end of the supersonic core is
denoted by a . It is significant that the supersonic core extends
downstream approximately 2 times the length of the potential
core. And, even though the mean profile suggests the influence
of shocks stops at the end of the potential core, the hydrodynamic
pressure field illustrated in Fig. 18 clearly shows that the shock
related oscillations have an effect up to the end of the supersonic
core. Therefore, shocks of weakening strength are likely present
throughout the supersonic region but may be masked by turbulent
fluctuations in these mean pressure measurements.
The mean, cross-stream, total pressure profiles for the Baseline and Centerbody configurations for both M j = 1.74 and
M j = 1.55 are shown in Fig. 15. In the mean, there was no observed difference between the Baseline and Faceted cases, so the
9
Copyright c 2012 by ASME
x/D j = 0.8
x/D j = 3.9
x/D j = 6.9
x/D j = 9.8
Baseline, Mach = 1.74
Baseline, Mach = 1.55
Centerbody, Mach = 1.74
Centerbody, Mach = 1.55
1.5
1
r /D j
0.5
0
−0.5
−1
−1.5
20
40
60
80
20
40
60
80
20
40
60
80
20
40
60
80
Total Pr es s ur e (ps ia)
FIGURE 15.
profile.
r/D j profiles of the total pressure, Pt , for the Baseline and Centerbody configurations. The axial x/D j is shown directly above each
cases, particularly in the upstream, shock-noise dominated direction. At the over-expanded M j , the OASPL near 120◦ is higher
for the Faceted nozzle while at the fully-expanded M j the opposite is true. This suggests that the faceted inner contour is
significant in the organization of the shocks.
The Mach wave angles calculated for the over-expanded and
fully-expanded operating conditions are 47.3◦ and 52.8◦ , respectively. This is based on the convective Mach number of the jet.
Baars et al. [22] has shown this value to range from 0.6Ma for
heated jets to 0.8Ma for cold jets. The acoustic Mach number
(Ma ) is the ratio
√ between the jet exit velocity (U j ) and the ambient
sound speed ( γa Ra Ta ). A convective Mach number of 0.6Ma is
found in the current study and can be confirmed with the spacetime contours in Fig. 18. Figure 17 supports these values in that
the maximum overall sound pressure level for the fully-expanded
condition occurs at a steeper angle than that of the over-expanded
condition, indicating that Mach waves propagate from a location
several diameters downstream of the jet exit.
to as far as fifteen diameters downstream from the nozzle exit.
For the baseline contour the supersonic core is estimated to end
around x/D j = 13; the potential core collapses around x/D j = 7
and 5 for the over-expanded and fully-expanded conditions, respectively. This suggests that the region between the end of the
potential core and the end of the supersonic core may be of significance for the aeroacoustics of supersonic, shock-containing
jets.
Overall sound pressure levels (OASPL) are shown in Fig. 19.
The region between the collapse of the potential core and the end
of the supersonic core contains the most intense turbulence activity and is thought to be the location responsible for the most
intense sound production. Variations in the upstream flow conditions are shown in Fig. 19(a) and 19(b) to have little to no effect on the OASPL. In Fig. 19(c), the fully-expanded nozzle is
shown to comprise a greater abundance of energy than the overexpanded nozzle, as would be expected due to the higher exit
velocity.
Single-sided power spectral densities (SPL) of the near-field
pressure are shown in Fig. 20 for the over-expanded (M j = 1.55)
and fully-expanded (M j = 1.74) jet conditions. These spectra
were generated using an ensemble average of 150 sets of 4096
samples yielding a frequency resolution of ∆ f = 24.4 Hz. In
each figure, six Kulites have been selected to provide a general
overview of the signatures registered in this region of the flow
and the effects caused by upstream flow conditions (center-body
and faceted). Subsequent spectra are shifted by 20 dB. The influence of the internal flow conditions are shown to have a lasting
effect on the lower band of frequencies (below the hydrodynamic
ridge) due to a thicker boundary layer or higher turbulence lev-
Near-Field Acoustic Characteristics
An illustration of the raw time series acquired with the nearfield array is shown in Fig. 18 for the baseline nozzle. The presence of shock cells in the jet plume is evident in the periodic
nature of the contours in the figure. These shock cells are present
even when the nozzle is operated at M j = Md (fully-expanded)
due to internal shock that forms because of the non-ideal innernozzle contour. These signatures are very different than those
found in perfectly expanded (shock-free) flows [22, 23]. The effect of the shock cells on the near-field pressure is evident out
10
Copyright c 2012 by ASME
90◦
120
10
20
60◦
◦
150◦
x/Dj
5
30◦
15
0
10
Baseline
Centerbody
Faceted
180
OASPL [dB ref. 20µPa]
−5
0
115
125
135
0
145
◦
15
20
25
30
35
FIGURE 18. Space-time contours of the raw fluctuating pressure
along the near-field array (in 103 Pa).
(a) Over-Expanded M j
90◦
60
OASPL [dB ref. 20µPa]
170
◦
◦
150◦
30◦
Baseline
Centerbody
Faceted
180◦
OASPL [dB ref. 20µPa]
10
t [ms]
◦
120
5
115
125
135
160
150
140
130
0
0◦
145
Baseline
5
Centerbody
10
Faceted
15
20
25
x/Dj
(a) Over-Expanded M j
(b) Fully-Expanded M j
170
OASPL [dB ref. 20µPa]
90◦
60◦
120◦
150◦
30◦
M = 1.74
M = 1.55
180◦
OASPL [dB ref. 20µPa]
160
150
140
130
0
115
125
135
Baseline
5
Centerbody
10
Faceted
15
20
25
20
25
x/Dj
0◦
145
(b) Fully-Expanded M j
(c) Baseline Nozzle M j Comparison
OASPL [dB ref. 20µPa]
170
FIGURE 17. Overall sound pressure level (OASPL) from the far-field
array. Comparison of inflow conditions for the (a) over-expanded and
(b) fully-expanded nozzles. (c) Comparison of OASPL between overand fully-expanded base inflow condition.
els at the nozzle exit. In particular, the centerbody produces a
low-frequency pulsation that is more pronounced for the overexpanded flow. Periodically shedding vortices that form in the
wake of the centerbody are believed to be responsible for this.
Surprisingly, this wake appears to have little influence on the hydrodynamic ridge formed by the jet column mode. Likewise, the
evanescent signatures produced by these shedding events appear
not to be affected by the jet shear layer as they emanate a recognizable signature several jet diameters downstream.
160
150
140
130
0
Mj = 1.74
5
Mj = 1.55
10
15
x/Dj
(c) Baseline Nozzle M j Comparison
FIGURE 19. Overall sound pressure level (OASPL) from the nearfield array. Comparison of inflow conditions for the (a) over-expanded
and (b) fully-expanded nozzles. (c) Comparison of OASPL between
over- and fully-expanded base inflow condition.
11
Copyright c 2012 by ASME
20
OBSERVATIONS
There are three primary observations of note based on the
data presented herein. First, the flow visualization clearly identifies coherent structures of significant size near the end of the
potential core region. Second, the far-field acoustics suggests
that the effect of the inclusion of a centerbody or a faceted nozzle contour is predominately observed in the higher frequencies
and the steeper angles. Finally, the near-field pressure measurements indicate that the influence of the shocks in the jet extend
all the way to the end of the supersonic core in the jet plume –
suggesting the region between the end of the potential core and
the end of the supersonic core may be of significance for the
aeroacoustics of supersonic, shock-containing jets.
85
18
95
16
105
x/Dj
14
115
12
10
8
115
FUTURE WORK
Future work will focus on the completion of the HDR-PIV
analysis. It is planned that the synchronous data acquisition system will be employed to acquire data for each nozzle configuration. The resulting high-fidelity PIV data will be utilized to
examine the 2-point correlations both within the jet and between
the jet and the near-field and far-field pressure. As additional focus will be to determine the roll of the supersonic core length in
the observed jet noise.
105
6
4
M = 1.55
M = 1.74
2
1.5
2
2.5
3
3.5
4
4.5
log10 (f ) [Hz]
FIGURE 21. Comparison of SPL [dB/Hz] contours for the base nozzle configuration.
ACKNOWLEGDEMENTS
This work is funded through the Office of Naval Research Jet
Noise Reduction program, contract number N00014-11-1-0752,
under the direction of Dr. J. Doychak and Dr. B. Henderson.
Because of the existence of shocks near the nozzle exit, several tones are expected. These tones are more easily captured by
a near-field observer. The tones in the range of several kilohertz
are eventually masked by the broader energy from turbulent mixing toward the downstream locations. These findings are consistent with existing studies of jet noise [24] and include characteristic turbulent mixing noise, broadband shock associated noise,
and tones associated with screech.
In an effort to compare the baseline conditions between the
over-expanded and fully-expanded nozzles, contours of the SPL
are shown in Fig. 21. The frequency axis has been plotted using
a base-10 logarithmic scale to reveal the nearly perfect logarithmic decay in the peak frequency with increasing distance downstream (increasing shear layer growth). Following the work of
Ewing et al. [25], it has been recently shown by Tinney & Jordan [26] how the characteristic axial wavenumbers of the nearfield pressure are homogeneous in this region and that a collapse
of the two-point correlation, obtained between two fixed points,
can be written with a function that depends only on variables obtained from a self similar solution. A logarithmic similarity coordinate of the form ln(x′ ) - ln(x) worked very well for the nearfield study of Tinney & Jordan (see [25] for details concerning
this similarity solution) and would appear to work equally as well
for both base flow conditions here, given the logarithmic shift in
the spectral peak with increasing distance shown in Fig. 21.
REFERENCES
[1] Munday, D., Gutmark, E., Liu, J., and Kailasanath, K.,
2011. “Flow structure and acoustics of supersonic jets from
conical convergent-divergent nozzles”. Physics of Fluids,
23, p. 116102.
[2] Kuo, C.-W., Veltin, J., and Mclaughlin, D. K., 2009.
“Acoustic measurements of models of military style supersonic nozzle jets”. In 47th AIAA Aerospace Sciences Meeting, no. 2009-0018.
[3] Bridges, J., 2006. “Effect of heat on space-time correlations
in jets”. In 12th AIAA Aeroacoustics Conference, no. 20062534.
[4] Wernet, M., 2007. “Time resolved piv for space-time correlations in hot jets”. In 45th AIAA Aerospace Sciences
Meeting and Exhibit, no. 2006-0047, AIAA.
[5] Seiner, J., and Yu, J., 1984. “Acoustic near-field properties
associated with broadband shock noise”. AIAA Journal,
22(9), pp. 1207–1215.
[6] Doty, M., and McLaughlin, D., 2002. “Two-point correlations of density gradient fluctuations in high speed jets
12
Copyright c 2012 by ASME
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
using optical deflectometry”. In 40th AIAA Aerospace Sciences Meeting, no. 2002-0367.
Papamoschou, D., Morris, P., and McLaughlin, D., 2010.
“Beamformed flow-acoustic correlations in a supersonic
jet”. AIAA Journal, 48(10), pp. 2445–2453.
Ponton, M., Seiner, J., Ukeiley, L., and Jansen, B., 2001. “A
new anechoic chamber design for testing high-temperature
jet flows”. In 7th AIAA/CEAS Aeroacoustics Conference
and Exhibit, no. 2001-2190, AIAA.
Ukeiley, L., Tinney, C., Mann, R., and Glauser, M., 2007.
“Spatial correlations in a transonic jet”. AIAA Journal,
45(6), pp. 1357–1369.
Harper-Bourne, M., and Fisher, M. J., 1974. The noise from
shock waves in supersonic jets. Tech. Rep. CP 131, 11.111.13, AGARD.
Norum, T., and Seiner, J., 1982. “Broadband shock noise
from supersonic jets”. AIAA Journal, 20(1), pp. 68–73.
Pao, S., and Seiner, J., 1983. “Shock-associated noise in
supersonic jets”. AIAA Journal, 21(5), pp. 687–693.
Arndt, R., Long, D., and Glauser, M., 1997. “The proper orthogonal decomposition of pressure fluctuations surrounding a turbulent jet”. Journal of Fluid Mechanics, 340,
pp. 1–33.
Lau, J. C., Fisher, M. J., and Fuchs, H. V., 1972. “The
intrinsic structure of turbulent jets”. Journal of Sound and
Vibration, 22, pp. 379–406.
Picard, C., and Delville, J., 2000. “Pressure velocity couping in a subsonic round jet”. International Journal of Heat
and Fluid Flow, 21, pp. 359–364.
Tinney, C. E., Ukeiley, L. S., and Glauser, M. N., 2008.
“Low-dimensional characteristics of a transonic jet. part 2:
Estimate and far-field prediction”. Journal of Fluid Mechanics, 615, pp. 53–92.
Tinney, C. E., Glauser, M. N., and Ukeiley, L. S., 2008.
“Low-dimensional characteristics of a transonic jet. part 1:
Proper orthogonal decomposition”. Journal of Fluid Mechanics, 612, pp. 107–141.
Lempert, W., Wu, P., Zhang, B., Miles, R., Lowrance, J.,
Mastracola, V., and Kosonocky, W. “Pulseburst laser system for high speed flow diagnostics”. AIAA Paper, No.
96-0179, 1996.
Wernet, M., and Opalski, A. “Development and application of a mhz frame rate digital particle image velocimetry
system”. AIAA Paper, No. 2004-2184, 2004.
Thurow, B., Satija, A., and Lynch, K., 2009. “Thirdgeneration megahertz-rate pulse burst laser system”. Applied Optics, 48(11), pp. 2086–2093.
Hain, R., and Kähler, C., 2007. “Fundamentals of multiframe particle image velocimetry (piv)”. Experiments in
Fluids, 42(4), pp. 575–587.
Baars, W. J., Tinney, C. E., Murray, N. E., Jansen, B. J.,
and Panickar, P., 2011. “The effect of heat on turbulent
[23]
[24]
[25]
[26]
13
mixing noise in supersonic jets”. In 49th AIAA Aerospace
Sciences Meeting including the New Horizons Forum and
Aerospace Exposition, no. 2011-1029, AIAA.
Baars, W. J., Tinney, C. E., and Wochner, M. S., 2012.
“Nonlinear noise propagation from a fully expanded mach
3 jet”. In 50th AIAA Aerospace Science Meeting, no. 20121177.
Tam, C., 1995. “Supersonic jet noise”. Annual Review of
Fluid Mechanics, 27, pp. 17–43.
Ewing, D., Frohnapfel, B., George, W. K., Pedersen, J. M.,
and Westerweel, J., 2007. “Two-point similarity in the
round jet”. Journal of Fluid Mechanics, 557, pp. 309–330.
Tinney, C. E., and Jordan, P., 2008. “The near pressure
field of co-axial subsonic jets”. Journal of Fluid Mechanics,
611, pp. 175–204.
Copyright c 2012 by ASME
220
220
135◦
135◦
200
120◦
200
120◦
90◦
90◦
180
75◦
60◦
160
SPL [dB/Hz ref. 20µPa]
SPL [dB/Hz ref. 20µPa]
180
52.5◦
45◦
140
37.5◦
30◦
120
75◦
60◦
52.5◦
160
45◦
140
37.5◦
30◦
120
θ = 20◦
θ = 20◦
100
100
Baseline
Centerbody
Faceted
80
2
10
Baseline
Centerbody
Faceted
80
3
10
10
4
2
10
f [kHz]
3
4
10
10
f [kHz]
(a) Over-Expanded M j = 1.55
(b) Fully-Expanded M j = 1.74
FIGURE 16. Single-sided power spectral densities (SPL) of the far-field pressure.SPLs are shifted cumulatively by 15 dB.
14
Copyright c 2012 by ASME
220
x/Dj = 21.34 (+100dB)
220
x/Dj = 21.34 (+100dB)
200
x/Dj = 15.96 (+80dB)
200
x/Dj = 15.96 (+80dB)
180
180
x/Dj = 9.42 (+60dB)
SPL [dB/Hz ref. 20µPa]
SPL [dB/Hz ref. 20µPa]
x/Dj = 9.42 (+60dB)
160
x/Dj = 5.35 (+40dB)
140
x/Dj = 2.53 (+20dB)
120
160
x/Dj = 5.35 (+40dB)
140
x/Dj = 2.53 (+20dB)
120
100
100
x/Dj = 1.60
80
x/Dj = 1.60
Baseline
Centerbody
Faceted
2
10
80
3
10
4
Baseline
Centerbody
Faceted
2
10
10
f [kHz]
10
3
4
10
f [kHz]
(a) Over-Expanded M j = 1.55
(b) Fully-Expanded M j = 1.74
FIGURE 20. Single-sided power spectral densities (SPL) of the near-field pressure from the over expanded nozzle (Mach 1.55). SPLs are shifted
cumulatively by 20 dB.
15
Copyright c 2012 by ASME