of 10
Design and Demonstration of a New Small-Scale Jet
Noise Experiment
Ryan A. Fontaine,
∗
Brock Bobbitt,
†
Gregory S. Elliott,
‡
Joanna M. Austin,
§
and Jonathan B. Freund
University of Illinois at Urbana-Champaign, Urbana, Illin
ois 61801
A facility capable of acoustic and velocity field measuremen
ts of high-speed jets has
recently been built and tested. The anechoic chamber that ho
uses the jet has a 2.1 m
×
2.3 m
×
2.5 m wedge tip to wedge tip working volume. We aim to demonstr
ate that
useful experiments can be performed in such a relatively sma
ll facility for a substantially
lower cost than in larger facility. Rapid prototyping allow
s for quick manufacturing of both
simple and complex geometry nozzles. Sideline and 30
◦
downstream acoustic measurements
between 400 Hz and 100 kHz agree well with accepted results. L
ikewise, nozzle exit-plane
data obtained using particle image velocimetry are in good a
greement with other studies.
I. Introduction
Aviation jet noise is heavily regulated with additional restrictions ant
icipated in the future. In the United
States, for example, the FAA’s FAR 36 Stage 4 noise standards cam
e into effect in January 2006 for vehicles
with a maximum take off weight of over 12,500 lbs, demanding a 10dB EPN
L (effective perceived noise level)
reduction beyond the previous Stage 3 limits.
1
International standards are typically at least as restrictive
and must be satisfied by any commercially viable business jet. Many cu
rrent aircraft operate close to FAA
and other limits. Predicting the impact of jet configuration and place
ment modifications on noise generation
is a significant challenge, because jet noise levels can be sensitive to n
ozzle configurations and because there
is no experimental or numerical substitute for a full-scale experime
ntal engine test. However, relatively
inexpensive, small-scale tests can help avoid full-scale tests to asse
ss the noise impact of design modifications
and study noise mechanisms.
For the experiments reported here we used a new small anechoic ch
amber constructed at the University
of Illinois at Urbana-Champaign. This is the first report on this facility
and its validation. Its design was
guided by the recommendations of Ahuja,
2
as well as consideration of previous facilities whose design and
use have been reported.
3–5
II. Experimental Facility
The facility is a 2.1 m
×
2.3 m
×
2.5 m wedge-tip-to-wedge-tip testing anechoic chamber for aeroa
coustic
and velocity field testing of around 2
.
54 cm nominal inner-diameter nozzles. The cloth covered fiberglass
wedges were manufactured by Eckel Industries to have a low cuto
ff frequency of 250 Hz, tested in accordance
with the Impedance Tube Method-ASTM-C 384-90. Above this freq
uency they are reported to absorb 99%
of the incident energy. We confirm their near-anechoic behavior in S
ection V. Due to the location of the
microphones the actual cutoff is closer to 400 Hz, which is still below t
he range of interest in the facility. This
was determined per the ISO 3745 standard that microphone locatio
ns must be 1/4 wavelength (at cut-off)
∗
Graduate Research Assistant, Department of Mechanical Sci
ence & Engineering, University of Illinois, AIAA Student
Member
†
Undergraduate Research Assistant, Department of Mechanic
al Science & Engineering, University of Illinois
‡
Professor, Department of Aerospace Engineering, Universi
ty of Illinois, Senior Member AIAA
§
Associate Professor, Department of Aerospace Engineering
, University of Illinois, Senior Member AIAA
Professor, Departments of Mechanical Science & Engineerin
g and Aerospace Engineering, University of Illinois, Senio
r
Member AIAA
1 of 10
American Institute of Aeronautics and Astronautics
50th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition
09 - 12 January 2012, Nashville, Tennessee
AIAA 2012-0682
Copyright © 2012 by Ryan A. Fontaine. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
Downloaded by Benjamin Perez on September 25, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-682
from the wedge tips.
6
The walls of the chamber were constructed using standard wood-s
tud framing with
a maximum center to center distance of 40 cm. An image of the comple
ted facility is shown in Figure 1.
To reduce outside noise, five 5.08 cm
×
5.08 cm
×
2.54 cm thick polyurethane rubber feet were installed
between each of the nine supports and the concrete slab floor of t
he lab.
Air is supplied to the experimental facility from a series of pressurize
d tanks (total 140 m
3
of air at 890
kPa). These tanks are charged by a 224 kW Ingersoll-Rand SSK HPE
300 compressor, which both dries and
filters the air. Due to the relatively large amount of air contained in th
e tanks and the small amount of
mass flow required by the jet, 2.54 cm and smaller nozzles can be test
ed continuously. To minimize spurious
acoustic contamination sources, large 20.3 cm piping is used from the
compressed air tanks to the control
valve that regulates the jet flow.
The control valve is a 2.54 cm Fisher 667-ET-DVC6010 Globe Valve Ass
embly with carbon steel body
used in conjunction with manual control from a Honeywell DC1203-
7-7-8-1-1-0-0-0 Model UDC1200 Process
Controller to keep the flow constant despite any slow pressure var
iations in the supply tank pressure. It
has been found that manual control using pressure readings fur
ther upstream, near the nozzle exit, is able
to better set the jet velocity than automated control using the c
omponents delivered with the valve. The
manufacturer specified noise from the valve is less than 83 dB.
Special attention was given to replacement air, to make up for that
entrained by the jet and exhausted
Figure 1: The anechoic jet noise facility.
with it out of the lab. We estimate the amount of air needed based upo
n a reported relationship for turbulent
jet entrainment from the experimental measurements of Ricou an
d Spalding:
7
m
m
o
= 0
.
32
x
D
(1)
where
m
is the mass flow rate of entrained air,
m
o
is the mass flow of the jet,
x
is the distance along the jet
axis measured from the nozzle exit, and
D
is the nozzle diameter. Using this relationship for our 2.54 cm jet
at Mach 0.98, it was determined that there will be roughly 4.8 kg/s of a
ir entrained by the jet before it enters
the exhaust ducting. To provide this make up air, two open sections
have been installed in the upstream
wall of the chamber partially inspired by those in the anechoic facility a
t The Ohio State University.
5
These
sections are rectangular in cross section, 2.1 m in height and 0.28 m in w
idth. The estimated velocity of
air entering through these entrainment sections is 3.5 m/s, which is a
bout one-hundredth the velocity of
the jet. This make-up air is filtered by an aluminum woven-wire screen
with 0.1 cm mesh size. The two
make-up air sections have a 2.54 cm thick acoustic polyurethane foa
m absorber lining with egg carton shape
to reduce any spurious noise from the lab from entering through th
ese sections. They are also designed with
an overhang such that there is no direct line of sight from the lab into
the chamber.
The exhaust consists of a conical 1.2 m diameter bell mouth leading to
a 0.6 m
×
0.6 m exhaust duct. The
exhaust system is made of acoustically absorbing perforated meta
l ducting, and the flow is directed outside
the building smoothly using aerodynamic turning vanes, as shown in Fig
ure 2, to reduce the potential for
downstream disturbances to contaminate the measurements in th
e chamber.
The microphones used for all experimental acoustic measurement
s performed are Br ̈uel & Kjær type
2 of 10
American Institute of Aeronautics and Astronautics
Downloaded by Benjamin Perez on September 25, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-682
Figure 2: Anechoic facility overview schematic.
4939 0.635 cm free-field microphones. These microphones are desig
ned for high-level and high-frequency
measurements with a flat response up to 100 kHz.
III. Facility Characterization
Since its completion in early 2010, the anechoic facility has undergone
successive modifications to improve
the quality of the data acquired. Example preliminary 1
/
3-octave measurements taken in the facility are
shown in Figure 3. (All acoustic measurements for this paper are fo
r uniform stagnation temperature cold
jets.) The multiple spectral peaks are obviously spurious, a clear sig
n of problems with the original facility
design. From this start the facility was improved in a step by step pro
cess, finally resulting in the current
facility, which is able to reproduce accepted jet noise data. Some of
the modifications made are discussed
to illustrate the evolution of the facility to its current capability to ma
tch accepted data. After every
modification the facility was retested and compared to accepted me
asurements to assess improvements. A
few of the more significant improvements are shown here.
Among the first modifications made was the addition of a flexible polyur
ethane foam lining to the
upstream piping in the two horizontal sections before the first con
traction (see Figure 2). This was done to
reduce any noise created in this section as well as attenuate noise f
rom upstream. Care was taken to avoid
introduction of a step or other such points of potential flow separ
ation. This involved the manufacture of
transition pieces of foam that ran smoothly from a 17.8 cm inner diame
ter of the foam lined section back
out to a 20.3 cm inner diameter to match the beginning of the contrac
tion, shown in Figure 4.
The lining improved results, but the noise levels were still unacceptab
ly high, so three baffles made from
Sonex One acoustical foam were inserted into one of the sections o
f the 20.3 cm piping, shown in Figure 5.
The baffles were fashioned entirely out of foam and inserted to block
any direct line of sight from upstream
to the jet exit as shown in Figure 5. Previous researchers have use
d perforated metal shells with fiberglass
baffles,
2
however we wished to avoid any reflections from the metal casing. T
he baffles were spaced to be
incompatible with the obvious standing wave modes of the pipes.
A 2.54 cm thick piece of honeycomb (3.18 mm cell size) was placed downs
tream of the baffles in the
20.3 cm piping section followed by an aluminum screen (1.02 mm cell size).
Both components were placed
upstream of the 20.3 cm to 10.2 cm contraction and were designed to
suppress turbulence. They were
positioned after the baffles to reduce any non-uniformity imposed o
n the flow by the baffles. After this final
screen, the flow path is smooth and therefore is not expected to c
reate significant noise. The screen was
placed downstream of the honeycomb following the recommendation
of Pope.
8
3 of 10
American Institute of Aeronautics and Astronautics
Downloaded by Benjamin Perez on September 25, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-682
10
-2
10
-1
10
0
10
1
60
65
70
75
80
85
90
95
100
M = 0.93
M = 0.73
SPL (dB)
St
D
Figure 3: Sideline
θ
= 90
◦
1/3-octave band spectra showing spurious peaks from preliminary
tests.
(a)
(b)
Figure 4: Upstream acoustic conditioning: (a) staggered baffles in t
he upstream muffler section and (b) foam
lining and tapered transition piece in the upstream piping.
At this point in the facility development, there was still evidence of re
flections within the chamber.
The seemingly largest source was investigated first: the exhaust c
ollector, which was thought to act as
a giant reflector plate downstream of the jet. The effect was so pr
onounced it was discernible with the
human ear while standing inside the chamber. To correct this, the en
tire collector was coated in the same
polyurethane foam material used to line the upstream piping. This re
moved the low-frequency obviously
spurious hump from the spectrum as shown in Figure 6. This effect at
low frequencies was expected given
reported experience with poor anechoic chamber designs leading to
contamination of jet noise spectra at
similarly low frequencies.
9
Once the facility appeared to be free of spurious noise, various tes
ts were performed to confirm that
accurate jet noise data was indeed being acquired. One of these me
thods was comparing the overall sound
level at different flow conditions with jet velocity (
U
j
) to the eighth power. This scaling is followed closely
between Mach numbers of 0.4 and 0.98. Mach number throughout th
is paper is defined with relation to the
speed of sound in the core of the jet. Deviation from the
U
8
j
at low Mach number marks the minimum speed
at which the facility can operate before rig noise becomes dominant.
In our case, this occurs near Mach 0.4.
4 of 10
American Institute of Aeronautics and Astronautics
Downloaded by Benjamin Perez on September 25, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-682
(a)
10
3
10
4
10
5
45
50
55
60
65
70
75
SPL (dB)
Frequency (Hz)
(b)
Figure 5: (a) Foam baffles in the upstream piping and (b) the resulting
improvements in noise measurements
at Mach 0.74:
Red - Without Baffles
; Black - With Baffles. (The high-frequency tone in (b) was due to
microphone grid-cap interference and corrected in subsequent f
acility improvements.)
(a)
10
-2
10
-1
10
0
10
1
50
55
60
65
70
75
80
85
90
M = 0.62
M = 0.74
M = 0.99
SPL (dB)
St
D
(b)
Figure 6: Exhaust collector improvement: (a) foam on the downstr
eam collector with (b) resulting improve-
ments in noise measurements:
◦
- Before;

- After.
IV. Atmospheric Absorption
Atmospheric absorption can play a significant role, especially at highe
r frequencies, and is particularly a
concern in small scale facilities, since so much of the spectrum is high f
requency. As the jet spectra scales
inversely as nozzle size, smaller nozzles require measurements at hig
her frequencies. A correction must be
applied so that all data can be compared in their lossless form. Viswan
athan
10
discusses several methods
for calculating atmospheric absorption coefficients and for small-sc
ale facilities recommends the method of
Shields and Bass,
11
which he deemed superior for high-frequency measurement to the
SAE method,
12
which
is used for full-scale engine corrections. At lower frequencies (suc
h as those of importance in full-scale tests),
these two methods produce similar results, however at higher freq
uencies the deviation can be significant.
After consulting the ANSI standard
13
as well as information from Viswanathan,
10
it was determined that the
5 of 10
American Institute of Aeronautics and Astronautics
Downloaded by Benjamin Perez on September 25, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-682
0.2
0.4
0.6
0.8
1
50
55
60
65
70
75
80
85
90
95
100
105
110
90 degrees
30 degrees
U
j
8
scaling
OASPL (dB)
Ma
Figure 7: Comparison of OASPL with
U
8
j
scaling at both 30
◦
and 90
◦
.
equations given by the ANSI standard produce results that are eff
ectivly indistinguishable from the Shields
and Bass approach.
11
We thus employ the ANSI standard. The attenuation coefficient, in d
ecibels/meter,
is:
α
(
f
) =8
.
686
f
2
[
1
.
84
×
10
−
11
(
p
a
p
r
)
−
1
(
T
T
r
)
1
2
+
(
T
T
r
)
−
5
2
{
0
.
01275 exp
(
−
2239
.
1
T
)
f
rO
f
2
rO
+
f
2
+ 0
.
1068 exp
(
−
3352
.
0
T
)
f
rN
f
2
rN
+
f
2
}
]
(2)
α
- Attenuation Coefficient (dB/m)
f
- Frequency (Hz)
p
a
- Atmospheric Pressure
p
r
- Reference Pressure (101325 Pa)
T
- Atmospheric Temperature (K)
T
r
- Reference Temperature (293.15 K)
f
rO
- Relaxation Frequency for Oxygen (see appendix)
f
rN
- Relaxation Frequency for Nitrogen (see appendix)
V. Anechoic Chamber Characterization
The purpose of an anechoic chamber is, of course, to simulate a fre
e-field environment in a laboratory. To
establish that the chamber was indeed acceptably anechoic, it was t
ested with known sources and without
any airflow. To a good approximation, sound did decay as 1
/R
2
, where
R
is the distance from the noise
source. These tests were performed using a white noise source po
sitioned at the typical jet exit location.
Frequencies up to the 20kHz limit of the noise source were tested. E
xample results are shown in Figure 8.
It is clear that the facility is anechoic down the the expected 400 Hz c
ut-off frequency.
VI. Complex Nozzle Capabilities
An on-campus rapid prototyping shop allows for fast manufacture
of new nozzle designs. These nozzles are
made on a Formiga P 100 Selective Laser Sintering system (SLS) at a c
ost of $0.30 per gram of material. With
such technology, complex shapes that would be difficult (and expens
ive) to manufacture using traditional
metal designs can be made quickly and at roughly the same cost as th
eir baseline counterparts without the
6 of 10
American Institute of Aeronautics and Astronautics
Downloaded by Benjamin Perez on September 25, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-682
50
100
150
60
80
100
120
Wedge
Extents
5000 Hz
8000 Hz
12 kHz
16 kHz
2500 Hz
1250 Hz
315 Hz
Designed
Cut-off
Frequency
SPL (dB)
Distance from Speaker (cm)
(a)
50
100
150
200
100
105
110
115
120
125
130
Wedge
Extents
SPL (dB)
Distance from Speaker (cm)
(b)
Figure 8: Experimental measurements of sound field spherical dive
rgence in chamber

and — 1
/R
2
: (a)
sideline 90
◦
at individual frequencies; (b) downstream 30
◦
OASPL.
complex geometries, all with sub-millimeter resolution.
16
Sample simple designs which have been made and
tested are shown in Figure 9. The chevron nozzle has been designed
so the position of the chevrons can be
rotated in relation to the microphone array which is fixed. When scale
d appropriately, these nozzles were
found to behave consistent with the metal nozzle which was used to
characterize this facility. After testing
various sized nozzles, it was determined that the 1.90 cm diameter no
zzle was the smallest exit area which
could produce scalable data, attributed to Reynolds number effect
s by Viswanathan.
17
In order to validate the facility, measurements were compared with
accepted experimental data
3,17
in
(a)
(b)
Figure 9: Nozzles designed and manufactured with rapid prototypin
g: (a) Standard Contraction; (b) Chevron
Nozzle.
Figure 10. The data are in excellent agreement, except for Mach 0.9
8 measurements taken at 30
◦
. The
disagreement near the peak of the spectrum has not been determ
ined as of this time. Also shown is a
comparison between a 2.68 cm copper nozzle and a 1.90 cm nozzle made
at a cost of about $10 utilizing the
rapid prototyping technology available. The data again are in excellen
t agreement except for the most intense
30
◦
data for the M = 0.98 jet. The cause of the difference is uncertain, a
nd it currently remains unclear why
the present facility is modestly quieter near the peak spectral fre
quencies. Also, nozzles manufactured using
different techniques are able to produce consistent noise spectra
when scaled. These data are presented in
7 of 10
American Institute of Aeronautics and Astronautics
Downloaded by Benjamin Perez on September 25, 2014 | http://arc.aiaa.org | DOI: 10.2514/6.2012-682